Advances in Research and Treatment For Alzheimer's Disease

Title:
dvancesinResearchandTreatmentfor A Alzheimersdisease 978-1475095425 1475095422
ISBN-13: ISBN-10:
C overdesignandLayout:
E lizabethLog elizbethlogg@gmail.com
Publisher:
Internet Medical Publishing info@imedpub.com http://imedpub.com/
First edition:
2012
isclaimer:ThisbookcontainsarticlespublishedunderCreativeCommonsAttributionLicense. D CreativeCommonsAttributionLicenseallowscommercialre-useofallcontent.
Advances in Research and Treatment for Alzheimers disease
Foreword PLoS
Alzheimers disease is for many reasons one of the disorders capturing more interest and support for research worldwide. There are many areas within the disease needing solutions urgently. From pathogenesis, to management and treatment. One area of clinical research is focused on treating the underlying disease pathology. Reduction of amyloid beta levels is a common target of compounds under investigation. Immunotherapy or vaccination for the amyloid protein is one treatment modality under study. Unlike preventative vaccination, the putative therapy would be used to treat people already diagnosed. It is based upon the concept of training the immune system to recognise, attack, and reverse deposition of amyloid, thereby altering the course of the disease. An example of such a vaccine under investigation was ACC-001, although the trials were suspended in 2008. Another similar agent is bapineuzumab, an antibody designed as identical to the naturally induced anti-amyloid antibody. Other approaches are neuroprotective agents, such as AL-108 and metalprotein interaction attenuation agents, such as PBT2. A TNF receptor fusion protein, etanercept has showed encouraging results. As of 2012, the safety and efficacy of more than 400 pharmaceutical treatments had been or were being investigated in 1012 clinical trials worldwide, and approximately a quarter of these compounds are in Phase III trials; the last step prior to review by regulatory agencies. In 2008, two separate clinical trials showed positive results in modifying the course of disease in mild to moderate AD with methylthioninium chloride (trade name rember), a drug that inhibits tau aggregation, and dimebon, an antihistamine. The consecutive Phase-III trial of Dimebon failed to show positive effects in the primary and secondary endpoints. This book compiles some of the most interesting articles on Alzheimers disease published by PLoS journals lately: from epidemiology and prevention to management and treatment.
Contents
Epidemiological Pathology of Dementia: Attributable-Risks at Death in the Medical Research Council Cognitive Function and Ageing Study Packages of Care for Dementia in Low- and Middle-Income Countries Operationalisation of Mild Cognitive Impairment: A Graphical Approach Can We Prevent, Delay, or Shorten the Course of Dementia? Educating the Brain to Avoid Dementia: Can Mental Exercise Prevent Alzheimer Disease? Dementia before Death in Ageing SocietiesThe Promise of Prevention and the Reality Cholinesterase Inhibitors: Drugs Looking for a Disease? Cholinesterase Inhibitors in Mild Cognitive Impairment: A Systematic Review of Randomised TrialsCholinesterase Inhibitors in Mild Cognitive Impairment: A Systematic Review of Randomised Trials Cholinesterase Inhibitors and Hospitalization for Bradycardia: A Population-Based Study Reducing Amyloid Plaque Burden via Ex Vivo Gene Delivery of an Ab-Degrading Protease: A Novel Therapeutic Approach to Alzheimer Disease Persistent Amyloidosis following Suppression of Ab Production in a Transgenic Model of Alzheimer Disease A Randomised, Blinded, Placebo-Controlled Trial in Dementia Patients Continuing or Stopping Neuroleptics (The DART-AD Trial) 7 22 31 36 38 41 50
55 66
75 87 103
Epidemiological Pathology of Dementia: AttributableRisks at Death in the Medical Research Council Cognitive Function and Ageing Study
Fiona E. Matthews1., Carol Brayne2., James Lowe3, Ian McKeith4, Stephen B. Wharton5, Paul Ince5*
1 Medical Research Council (MRC) Biostatistics, University of Cambridge, Cambridge, United Kingdom, 2 Institute of Public Health, University of Cambridge, Cambridge, United Kingdom, 3 Department of Neuropathology, University of Nottingham, Nottingham, United Kingdom, 4 Institute for Ageing and Health, University of Newcastle upon Tyne, Newcastle upon Tyne, United Kingdom, 5 Department of Neuroscience, University of Sheffield, Sheffield, United Kingdom
Abstract
Background: Dementia drug development aims to modulate pathological processes that cause clinical syndromes. Population data (epidemiological neuropathology) will help to model and predict the potential impact of such therapies on dementia burden in older people. Presently this can only be explored through post mortem findings. We report the attributable risks (ARs) for dementia at death for common age-related degenerative and vascular pathologies, and other factors, in the MRC Cognitive Function and Ageing Study (MRC CFAS). Methods and Findings: A multicentre, prospective, longitudinal study of older people in the UK was linked to a brain donation programme. Neuropathology of 456 consecutive brain donations assessed degenerative and vascular pathologies. Logistic regression modelling, with bootstrapping and sensitivity analyses, was used to estimate AR at death for dementia for specific pathologies and other factors. The main contributors to AR at death for dementia in MRC CFAS were age (18%), small brain (12%), neocortical neuritic plaques (8%) and neurofibrillary tangles (11%), small vessel disease (12%), multiple vascular pathologies (9%), and hippocampal atrophy (10%). Other significant factors include cerebral amyloid angiopathy (7%) and Lewy bodies (3%). Conclusions: Such AR estimates cannot be derived from the living population; rather they estimate the relative contribution of specific pathologies to dementia at death. We found that multiple pathologies determine the overall burden of dementia. The impact of therapy targeted to a specific pathology may be profound when the dementia is relatively pure, but may be less impressive for the majority with mixed disease, and in terms of the population. These data justify a range of strategies, and combination therapies, to combat the degenerative and vascular determinants of cognitive decline and dementia.
Please see later in the article for the Editors Summary.
Citation: Matthews FE, Brayne C, Lowe J, McKeith I, Wharton SB, et al. (2009) Epidemiological Pathology of Dementia: Attributable-Risks at Death in the Medical Research Council Cognitive Function and Ageing Study. PLoS Med 6(11): e1000180. doi:10.1371/journal.pmed.1000180 Academic Editor: Sam Gandy, Mount Sinai School of Medicine, United States of America Received December 29, 2008; Accepted October 2, 2009; Published November 10, 2009 Copyright: 2009 Matthews et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: Funding was provided by Medical Research Council UK: G9901400. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Abbreviations: AR, attributable risk; CAA, cerebral amyloid angiopathy; CI, confidence interval; DWML, deep white matter lesion; GMS, geriatric mental state; NFT, neurofibrillary tangles; OR, odds ratio; RINI, retrospective informant interview; SVD, small vessel disease; WML, white matter lesion. * E-mail: p.g.ince@shef.ac.uk . These authors contributed equally to this work.
7
PLoS Medicine | www.plosmedicine.org
AgeingStudy
November 2009 | Volume 6 | Issue 11 | e1000180
EpidemiologicalPathologyofDementia:Attributable-RisksatDeathintheMedicalResearchCouncilCognitiveFunctionand
Epidemiological Pathology of Dementia
Introduction
Assessment of brain pathology in the consensus protocols for pathological diagnosis of dementia has been based on semiquantitative methods [13]. These protocols aspire to distinguish demented and nondemented individuals using thresholds of plaques, neurofibrillary tangles (NFTs), infarcts, and Lewy bodies, so that pathology becomes the gold standard for diagnosis. This approach has progressed understanding of clinical phenotypes, genetics, biochemistry, and molecular pathogenesis associated with cognitive decline in older people. Trials of disease modifying therapies are already in progress and proponents of a vascular basis for cognitive dysfunction propose secondary prevention strategies in older people [4]. The scale of the clinical and social problem presented by dementia in ageing populations presents an urgent need to assess the likely impact and cost effectiveness of new, potentially expensive, therapies, and to develop robust biomarkers for diagnosis and progression. Understanding the population impact of therapies that modify the pathobiology of dementia requires an understanding of the burden of cognitive dysfunction directly attributable to a particular pathology. Recently reported trials in Alzheimer
disease (AD), alleging divergent outcomes for inhibition of amyloid gamma-secretase and tau aggregation, exemplify this need [5,6]. These are issues about which conventional casecontrol cohorts and studies in secondary referral populations, memory clinics, or community volunteers are less informative than the population approach used here. Selection biases are associated with nonpopulation-based studies of older people and lead to unknown effects so that their conclusions may not generalise to the whole population. Dementia is associated with a high prevalence of mixed Alzheimer, vascular, and other pathologies, and the thresholds of severity that clearly distinguish between an AD brain and the brain of a nondemented individual only capture around 20% of demented people [79]. True population-based studies of dementia combining longitudinal assessment of clinical state with post mortem brain donation are rare but offer the only means at present of investigating the population-level impact of pathology on cognition [10]. The Cognitive Function and Ageing Study (CFAS) autopsy donor cohort is now of sufficient size to facilitate true epidemiological neuropathology. Here we present estimates of the attributable risk of dementia at death associated with specific neuropathological features in this cohort.
Figure 1. MRC CFAS design. Numbers of interviews and donations from interview waves. doi:10.1371/journal.pmed.1000180.g001

Methods Ethics Statement

All procedures received approval from a multicentre Research Ethics Committee. MRC Cognitive Function and Ageing Study (MRC CFAS) is a population-based longitudinal study of people, in their 65th year and over, enrolled from the population-based registers of primary care physicians in six sites in England and Wales [11]. In 1990 these registers provided full geographical population coverage including people living in institutional settings. In each of five centres random samples generated a recruited cohort of 2,500 individuals per centre (82% response rate) with equal numbers below and above 75 y. Trained interviewers conducted interviews with participants, including basic sociodemographic questions, cognitive examination, and items from the Geriatric Mental State (GMS) organicity scale, activities of daily living, physical health, and medication (see www.cfas.ac.uk). A 20% stratified sample underwent detailed assessment (GMS [12], augmented CAMCOG [13]) repeated after 2 y. This assessment group included those individuals with cognitive impairment and a random sample from the same centre. There were two re-interview cycles of all survivors and several follow-ups in the assessment group only. Study flow is shown in Figure 1. Diagnoses were made using the validated AGECAT algorithm [14]. In the sixth centre (Liverpool) 5,200 people aged 65 y and over, in equal numbers across 5-y bands, were recruited with a population sampling base. This study (ALPHA) started before the other sites but its design and methods enable it to be integrated into CFAS. Interviews in ALPHA were based on GMS and study flow is shown in Figure 2. Individuals, families, and carers in the assessment group were approached by trained liaison officers and invited to participate in counselling around brain donation. Those who agreed to brain donation were provided with information to allow staff or family involved in the final illness to notify the death and initiate brain
donation. Donations still proceeded, wherever possible, for cases coming to autopsy under the coroner. There were 456 individuals in this analysis, representing all completed brain donations before 1st August 2004. The sample includes 207 individuals previously described (two families contributing to that previous cohort subsequently revoked consent and the data were removed) [7]. Dementia status was established using multiple information sources including AGECAT, notification of dementia in death certificates, a retrospective informant interview (RINI; www.cfas. ac.uk) with relatives and carers after death, and the probability of being demented before death from a Bayesian analysis of all individuals modelling the prevalence and incidence of dementia in CFAS. We could not assign dementia status in 30 individuals in whom the study diagnosis was not dementia. These respondents were not included in the analysis because their last interview was more than 6 mo before death, no RINI was available, and dementia was not mentioned on the death certificate. The neuropathology protocol used standardised assessment of paraffin-embedded tissues to record data using the Consortium to Establish a Registry of Alzheimers Disease (CERAD) protocol [2]. CERAD data were augmented by a strategy for evaluating white matter lesions (WML) in the post mortem brain previously validated against histopathology [15]. Neuropathology was assessed without knowledge of clinical, interview, or RINI data. Acceptable inter-rater reliability (,5% with scores more than 1 grade difference) was achieved for cerebral cortical atrophy, NFT, amyloid plaques (diffuse and neuritic), Lewy bodies, and cerebral amyloid angiopathy by circulation of macroscopic brain photographs and microscopic slides.
Statistical Methods
Sample characteristics were compared with the overall population using Chi-squared for association and Mann-Whitney for the median age of death. Unconditional logistic regression analysis examined the
Figure 2. MRC Alpha Study (Liverpool) design. Numbers of interviews and donations from each interview wave. doi:10.1371/journal.pmed.1000180.g002
Table 1. Demographic characteristics of baseline population, deaths, and donors (six centres).
Characteristics Centre Cambridge Gwynedd Liverpool Newcastle Nottingham Oxford Sex Men Women Age group at baseline (y) ,74 7584 8594 .94 Median age (y) Education Missing 09 y .9 y Social class Missing Nonmanual Manual Age group at death (y) ,74 7584 8594 .94 Median age at death
Baseline n = 18,248
Percent
IQR
Died but Not Donors n = 11,465 Percent
IQR
Donors n = 456
Percent
IQR
Donors Versus All Died but Not Donors
2,601 2,625 5,244 2,524 2,514 2,740
14 14 29 14 14 15
1,465 1,387 4,065 1,549 1,423 1,526
13 12 35 14 13 13
96 9 101 55 127 68
21 2 22 12 28 15
7,625 10,623
42 58
5,164 6,301
45 55
188 268
41 59
p = 0.11
8,231 6,982 2,849 186 70
45 38 16 1
6774
3,557 5,159 2,573 176 79
31 45 22 2
73,84
100 215 135 6 81
22 47 29 2
P,0.001
7686
566 11,985 5,697
3 66 32
483 7,812 3,170
4 68 28
32 302 122
6 67 27
p = 0.97
1,371 6,574 10,303
8 36 56
1,035 3,717 6,713
9 32 59
57 165 234
13 36 51
p = 0.02
1,187 4,601 4,782 895 84
10 40 42 8
79,90
43 125 229 59 87
27 50 13
8292 P,0.001
IQR, interquartile range. doi:10.1371/journal.pmed.1000180.t001
10
effect of neuropathological variables and age on dementia at death. Backwards stepwise logistic regression (p = 0.1) assisted model selection. Interactions between pathologies were tested using likelihood ratio tests. The chosen final model best predicted dementia status at death for the least number of factors within the model. The method maximised the sensitivity and specificity using the predicted probabilities from the model, and minimised the number of individuals for whom prediction was not possible. Using the model most likely to correctly predict dementia we estimated the partial population attributable risk (AR) at death, conditional on all other factors, from the adjusted logistic regression [16]. These AR estimates represent the amount of dementia at death in the sample determined by each factor in the model, relative to the reference category, using both the risk and the prevalence of that factor. The AR represents the proportion of dementia at death accounted for after elimination of that risk factor if the remaining distribution of the risk factors remains unchanged. Bootstrap confidence intervals were calculated [17]. The amount of missing
PLoS Medicine | www.plosmedicine.org 4
data was small (,5%) except for brain weight (10%), macroscopic vascular disease (7%) and diffuse plaques in the entorhinal campus/ hippocampus (6%). A sensitivity analysis of the model and PAR estimates was undertaken using multiple imputations using chain equations to impute the data for all individuals with unknown factors in the model process [18]. Ten imputation datasets were used. Analysis was undertaken using Stata software, version 9.2 (Stata Corp.) and the R software package (www.r-project.org).
Results Representativeness of the Brain Donor Cohort

There were no significant differences between the donor cohort and all respondents who died with respect to sex and length in fulltime education (Table 1). Individuals who donated brain tissue were significantly older than all deaths in the population sample, and also than all individuals still alive on 1st August 2004. More donors were in the manual social class compared with all who

Table 2. Number of individuals with neuropathology findings in medial temporal and neocortical regions.
Neuropathology Neuritic plaques
Severity None Mild Moderate Severe Missing
Hippocampus 191 97 123 39 6 203 128 84 16 25 51 118 134 147 6 287 67 39 37 26 432 6 0 0 18
Percent 42 21 27 9 1 45 28 18 4 5 11 26 30 32 1 63 15 9 8 6 95 1 0 0 4
Entorhinal 178 122 106 41 9 142 110 124 51 29 39 94 185 128 10 287 53 46 34 36 409 13 4 1 29
Percent 39 27 23 9 2 31 24 27 11 6 9 21 41 28 2 63 12 10 7 8 90 3 1 0 6
Frontal 185 140 91 37 3 134 106 94 103 19 298 101 40 11 6 432 10 0 0 14 433 8 0 0 15
Temporal 171 111 123 50 1 130 95 122 92 17 230 124 62 37 7 430 12 0 0 14 424 16 0 0 16
Parietal 194 114 108 38 2 151 107 92 88 18 295 94 47 11 9 430 7 0 0 19 428 4 0 0 24
Occipital 200 74 80 23 79 143 94 100 37 82 299 38 26 9 84 384 3 0 0 69 380 2 0 0 74
Overall 143 104 139 70 0 115 95 112 120 14 216 129 70 39 2 425 19 0 0 12 425 18 0 0 13
Percent 31 23 31 15 0 25 21 25 26 3 47 28 15 9 1 93 4 0 0 3 93 4 0 0 3
Diffuse plaques
None Mild Moderate Severe Missing
Tangles
Neuronal loss
Lewy bodies
doi:10.1371/journal.pmed.1000180.t002
died (p = 0.03). There was no significant difference between the brain donor cohort and the CFAS baseline sample with respect to sex, social class, and length in fulltime of education.
Prevalence of Neuropathologies in Old People

Neurodegenerative pathology in the cohort is shown in Tables 2 and 3. NFT were the most prevalent degenerative pathology in the hippocampus and entorhinal cortex (92%). In the neocortex NFT were less (52%), and neuritic plaques more (68%), prevalent. Neuronal loss in hippocampus and subcortical nuclei was reported in 40% and 60% of the sample, respectively. Lewy bodies were found in less than 10% of brains (mainly substantia nigra) reflecting the use of older techniques (hematoxylineosin or ubiquitin staining) rather than synuclein staining. These factors are shown in relation to dementia status in Tables 4 and 5. Neocortical, hippocampal, and entorhinal cortex pathology was more common in individuals with dementia. Vascular pathology was frequently observed (not demented 71%, demented 84%), most frequently as small vessel disease (SVD; not demented 60%, demented 77%; odds ratio [OR] for dementia 1.6, 95% confidence interval [CI] 0.92.8). This diagnosis of SVD is based exclusively on histological criteria. Periventricular white matter lesions (PVL) were common (not demented 87%, demented 95%, OR for dementia with severe PVL 4.3, 95% CI 1.99.8), though deep white matter lesions (DWMLs) were less common (not demented 60%, demented 73%, OR for dementia with severe DWML 3.3, 95% 1.66.8). A combined diagnosis including both
WML and histological SVD was overrepresented among the demented group (32% versus 24%; OR for dementia 2.9, 95% CI 1.65.5). Cerebral amyloid angiopathy (CAA) was overrepresented in the demented group (34% versus 10%; OR for dementia 4.3, 95% CI 2.47.6). Neocortical and hippocampal atrophy were both common in individuals with dementia (75% and 67%), though less so in the nondemented (43% and 31%), and therefore have a very large association with dementia (severe atrophy OR.10). The multivariable logistic regression (Table 5) showed independent association with dementia status for: moderate and severe NFT in the neocortex; severe neuritic plaques in the neocortex; cerebral angiopathy; combined vascular disease; SVD; PVL; Lewy bodies; hippocampal atrophy; brain weight and age. All respondents with severe NFT in the neocortex (39 individuals, 16% of the demented group) were demented giving an infinite odds ratio for dementia risk and representing a perfect predictor (sufficient but not necessary). Seventy individuals had some missing neuropathology measures and were initially excluded from the modelling. Individuals whose missing data did not affect their predicted dementia status were included in the summary (48 of 70). The final prediction was therefore from a model using 404 brains (89%), which represent 4% of all the deceased respondents in CFAS. The multivariable model (Table 5) correctly predicted dementia status in 80% (404 individuals) of the 426 with known dementia status (sensitivity, 71%, 95% CI 6476; specificity, 92%, 95% CI 8796; receiver operating characteristic [ROC] 0.86). Substantial neuropathology without a dementia diagnosis was found in 13
11
Table 3. Number of individuals with neuropathology findings in subcortical nuclei.
Neuropathology Plaques
Severity None Mild Moderate Severe Missing/not measured
Substantia Nigra 389 2 0 0 65 314 89 13 20 20 217 175 36 16 12 409 19 11 8 9
Nucleus Basalis 202 30 13 0 211 93 104 49 47 163 205 61 23 6 161 286 10 1 0 159
Raphe 271 8 0 0 177 157 78 52 36 133 298 23 2
Locus Ceruleus 283 1 0 0 172 166 105 42 14 129 216 84 27 11
Dorsal Vagus 194 0 0 0 262 207 11 0 0 238 230 27 14 3 182 258 7 10 0 181
Overall 353 36 13 0 54 177 122 77 71 9 183 175 67 25 6 406 20 18 9 3
Percent 77 8 3 0 12 39 26 17 15 2 40 38 15 6 2 89 4 4 2 1
Tangles
Neuronal loss
133 327 2 0 0 127
118 313 15 9 1 118
Lewy bodies
individuals (M:F 6:7; age at death, 81102 y). This group includes two individuals with life-long low cognition confirmed by RINI interview. The remaining 11 had an Mini-Mental Status Examination (MMSE) .18 (three with MMSE .26) when they were last measured, including five (45%) who had a RINI. Median time from interview to death in those without a RINI interview was 12 mo. Only two had symptomatic cognitive impairment, but not consistently, and one had depression. The pathologies exhibited by these individuals were SVD (n = 11), low brain weight (n = 8), atrophy (n = 7), severe plaques (n = 5), and moderate NFT (n = 4). Conversely 68 individuals (M:F 24:44; age at death, 71103 y) had a dementia diagnosis before death but showed only modest neuropathology. Sixty (88%) showed moderate or severe cognitive impairment before death. Of the other eight all had died at least 15 mo after the last interview and dementia was confirmed by RINI (n = 6) or death certificate (n = 3). The neuropathology was generally mild and included Lewy bodies (n = 4). Two individuals had severe atrophy of the hippocampus. Neuropathology in the brainstem, not included in the model, was present in 36 individuals (NFT, n = 24; plaques, n = 4; neuronal loss, n = 15). These factors did not improve the overall model when tested across all individuals. Other neuropathological findings in these individuals include Progressive Supranuclear Palsy, hippocampal hypoxic injury, head injury, and mesial temporal sclerosis. The outcome of interest in this analysis was dementia and it therefore does not address cognitive impairment short of dementia in which the factors reported here would also be expected to play a role.
for all others such that 96% of the overall risk is explained. Nearly 20% of this risk is due to the effect of age. Factors conveying more than 8% each of the dementia risk were: NFT in the neocortex; age; neuritic plaques; SVD; moderate/severe atrophy; low brain weight. Alzheimer pathologies together (plaques, tangles, and CAA) account for ,25% of dementia risk, and vascular pathologies ,21%. Other neuropathological factors each convey between 2%5% of the risk.
Neuropathology in the Nondemented

Many nondemented individuals manifest high risk pathologies. A moderate NFT score in the neocortex is rare (4%), and a severe NFT score absent, but multiple vascular disease (24%) and SVD (47%) are common. Neuropathological factors, age, and brain weight only account for 34% of the variability within the model, despite high estimates of AR. This apparent anomaly underscores their relatively poor predictive value in making a diagnosis of dementia.
Prediction from Neuropathology Alone

Univariate modelling of the relationship between dementia and neuropathological findings, excluding age and brain weight, showed a large additional risk associated with having NFT in the hippocampus. However this adjusted out in multivariable analysis. The model based on neuropathology data alone has higher sensitivity (83%), but lower specificity (76%). From the 399 in this model, 80% were correctly classified as either demented or not. The AR at death for neuropathological features was modified only slightly by excluding age and brain weight. The major contributor to dementia risk remained NFT (28%; neocortical, 14%; hippocampal, 14%). Atrophy (20%) and CAA (11%) were more important. Vascular factors (17%), neocortical plaques (7%), and Lewy bodies (4%) remained the same. No interactions were detected.
6 November 2009 | Volume 6 | Issue 11 | e1000180
AR of Dementia for Pathological Features
12
The risk of dementia associated with specific thresholds of pathology is shown in Table 5. Each estimated AR at death adjusts

Table 4. Number of individuals by neuropathology and dementia status at death.
Neuropathological Findings Neocortex: neuritic plaques
Severity None Mild Moderate Severe
No Dementia n = 183 83 51 43 6 54 43 56 28 12 114 59 8 0 101 51 24 1 6 106 37 33 6 1 101 47 26 5 4 34 68 53 27 1 117 33 17 2 14 100 49 28 5 1 81 42 44 11 5 27 55
Percent 45 28 24 3 30 24 31 15 63 33 4 0 57 29 14 1 58 20 18 3 56 26 15 3 19 37 29 15 69 19 11 1 55 27 15 3 46 24 25 6 15 30
Dementia n = 243 47 44 89 63 38 42 60 91 12 81 63 60 39 57 52 94 28 12 67 57 83 31 5 89 70 53 10 21 12 39 72 115 5 71 47 78 18 29 61 64 75 35 8 51 58 73 37 24 7 30
Percent 19 18 37 26 16 18 26 39 33 26 25 16 25 23 41 12 28 24 35 13 40 32 24 5 9 5 16 30 48 33 22 36 8 26 27 32 15 23 26 33 17 3 13
Uncertain n = 30 13 9 7 1 9 4 9 8 0 21 7 2 0 15 12 2 0 1 18 3 7 2 0 13 11 5 1 0 5 11 9 5 0 16 9 2 1 2 17 9 3 1 0 10 10 7 3 0 5 9
Percent 43 30 23 3 30 13 30 27 67 22 11 0 50 40 7 0 60 10 23 7 43 37 17 3 17 37 30 3 0 57 32 7 4 2 57 30 10 3 33 33 23 10 17 30
Neocortex: diffuse plaques
Neocortex: NFT
None Mild Moderate Severe
Neocortex: atrophy
Hippocampus: neuritic plaques
Hippocampus: diffuse plaques
Hippocampus: NFT
Hippocampus: atrophy
Entorhinal cortex: neuritic plaques
Entorhinal cortex: diffuse plaques None Mild Moderate Severe Missing Entorhinal cortex: NFT None Mild
13
Table 4. Cont.
Neuropathological Findings
Severity Moderate Severe Missing
No Dementia n = 183 77 22 2 10 1.24 57 79 47 131 31 17 0 5 17
Percent 43 12 5 32 43 26 73 17 10 0 10
Dementia n = 243 98 100 8 35 1.11 25 110 108 102 54 64 17 5 81
Percent 42 43 14 10 45 44 43 23 27 7 34
Uncertain n = 30 10 6 0 3 1.15 7 15 8 23 4 2 1 0 3
Percent 33 20 10 23 50 27 77 13 7 3 13
Lewy bodies Brain weight kg median Age at death (y)
,80 8089 $90
CAA
Moderate/severe CAA Vascular disease Any vascular disease Haemorrhage Infarct Lacune SVD Overall vascular pathology None One of infarct/lacune/ haemorrhage SVD only Multiple Missing Periventricular WML
None Mild Moderate/ severe Missing
122 9 43 30 104 18 50 14 56 50 18 21 95 49 19 71 67 24 14 9
71 5 24 17 60 29 8 33 29 13 58 29 40 38 14 8
196 9 80 62 178 38 9 76 102 13 11 100 114 17 62 97 36 39 4
84 4 34 26 77 17 4 34 45 5 44 51 27 41 15 17
24 1 14 7 18 5 3 8 12 2 2 19 8 1 12 12 4 1 1
83 4 48 24 64 18 11 29 43 7 66 28 41 41 14 3
Deep subcortical WML
Histological/imaging vascular disease None Infarct or haemorrhage Lacunes/SVD/DWML Both Missing doi:10.1371/journal.pmed.1000180.t004 44 7 84 43 5 25 4 47 24 26 5 130 77 5 11 2 55 32 5 1 11 13 0 17 3 37 43
Sensitivity Analysis
Imputation of variables with missing data was used to test the robustness of the model against both missing outcome variables (30 individuals whose dementia status was not coded) and pathology variables. The multivariable modelling after imputation showed few differences from the original model in Table 5. The
14
neuropathology factors chosen to be represented in the model were checked using ten imputation datasets. Factors associated with dementia were remarkably stable within each imputation dataset. The only factors that appeared to differ were whether Lewy bodies (excluded from five datasets), severe plaques (excluded from four datasets), age (excluded from two datasets),

Table 5. Unconditional logistic regression adjusted for age and multivariable analyses and estimated partial AR at death for dementia.
Age Adjusted Analysis OR 95% CI
Multivariable Analysis
AR Percent 95% CI
p-Value
OR
95% CI
Age at death (y) ,80 8089 $90 Time since last interview (y) ,1 .1 Brain weight for sex (g) Low Average High Neuritic plaques in neocortex None Mild Moderate Severe Diffuse plaques in neocortex None Mild Moderate Severe NFT in neocortex None Mild Moderate Severe Neuritic plaques in hippocampus None Mild Moderate Severe Diffuse plaques in hippocampus None Mild Moderate Severe Tangles in hippocampus None Mild Moderate Severe 1.0 1.3 8.9 1.0 1.9 3.5 8.5 1.0 1.5 2.0 2.2 1.0 1.3 2.7 8.4 0.82.1 4.020 1.13.3 2.15.9 3.222 0.92.5 1.13.5 0.77.2 0.62.8 1.25.9 3.719 1.13.1 2.16.5 4.030 ,0.001 ,0.001 0.06 ,0.001 ,0.001 1.0 1.0 7.1 Not included Not included Not included Not included Not included 0.51.8 2.322 11 Not included Not included Not included Not included Not included 519 1.0 1.2 2.9 18.5 1.0 1.4 1.6 4.2 0.72.2 1.75.0 7.347 0.72.5 0.92.9 2.37.9 ,0.001 ,0.001 1.0 1.0 1.0 9.7 Not included 2.143 8 Not included 314 5.7 2.0 1.0 3.210 1.234 ,0.001 4.1 2.1 1.0 1.99.2 1.04.2 12 5 519 011 1.0 3.2 5.2 1.0 1.2 1.85.5 2.99.4 0.81.7 ,0.001 0.5 1.0 2.5 3.4 Not included 1.15.8 1.48.3 8 10 Not included 016 316
Neuritic plaques in the entorhinal cortex None Mild Moderate Severe 1.0 1.9 3.7 11.1
Diffuse plaques in the entorhinal cortex None 1.0
15
Table 5. Cont.
Age Adjusted Analysis OR 95% CI 1.33.8 1.54.4 2.512 0.64.5 1.48.6 4.834
Multivariable Analysis
AR Percent Not included 95% CI
p-Value
,0.001 ,0.001
OR Not included
95% CI
Mild Moderate Severe Tangles in entorhinal cortex None Mild Moderate Severe CAA None Mild Moderate Severe Moderate/severe CAA No Yes Lewy bodies No Yes Vascular disease (VD) None One of infarct/lacune/haemorrhage SVD only Multiple Overall vascular pathology None Infarcts/haemorrhage SVD/WML/lacunes Both Periventricular WML None Mild Moderate/severe Deep subcortical WML None Mild Moderate Severe Hippocampal atrophy None Mild Moderate Severe Neocortical atrophy None Mild Moderate Severe
2.2 2.6 5.5 1.0 1.7 3.4 12.7
1.0 1.9 4.0 1.0 4.3
1.13.3 2.27.4 2.47.6
,0.001 ,0.001
1.0 1.8 2.9 2.9 Not included
1.0 0.83.8 1.26.8 1.26.8
2 5 5 Not included
06 110 110
1.0 3.2 1.0 0.7 1.6 2.5
1.0 1.56.9 0.31.9 0.92.8 1.44.3
,0.003 ,0.004
1.0 3.5 Not included
1.39.3
3 Not included
17
1.0 1.0 2.5 2.9 1.0 2.0 4.3 1.0 1.6 1.6 3.3
0.33.8 1.44.4 1.65.5 0.94.5 1.99.8 1.02.6 0.93.1 1.66.8
,0.003 ,0.001 ,0.01
1.0 2.4 3.7 4.8 Not included Not included
0.412 1.59.6 1.912
12 9 Not included Not included
319 315
1.0 2.2 6.9 11.1 1.0 1.6 5.9 37.8
1.0 1.33.9 3.713 2.550 1.0 1.02.8 3.310 5.029
,0.001 ,0.01
1.0 1.8 3.4 3.4 Not included
1.0 0.93.7 1.57.5 1.57.5
2 8 8 Not included
06 215 215
16
10

and hippocampal atrophy (excluded from one dataset) should be included in the model. Two factors that were not previously in the models became important: NFT in the hippocampus and neuronal loss in the brainstem. Neuronal loss in the brainstem appeared important in individuals previously misclassified, but did not improve the model using the original data where there was missing data in the covariates and outcome variable. The full model with all these factors is shown in Table 6. The estimations of AR at death were very similar for the imputation datasets. The inclusion of brain stem neuronal loss (AR 13%) and NFT in the hippocampus (AR 5%) emerged from small reductions (1%) in the majority of factors though hippocampal atrophy (8% to 4%) and old age (11% to 8%) were more affected. Analysis adjusting for demographic differences between the brain donor cohort and the rest of the population that died showed only slight change in AR at death for factors most associated with older age (old age, atrophy, and neocortical NFT), whilst vascular disease, low brain weight, plaques, and CAA all showed small increases (1%2%). Only low brain weight (from 11% to 18%) and atrophy (from 8% to 2%) were affected by the age difference between the donor cohort and all those who died in the population. A further sensitivity analysis only in those assessed less than 1 y prior to death was undertaken and all associations increased in strength, suggesting any bias is conservative, and all AR estimates were consistent with the confidence intervals presented.
may include synaptic integrity and the concentrations of peptide oligomers [21,22] but also interindividual variation in diverse factors that determine the neurobiological basis of brain reserve, both innate (synaptic and neuronal density achieved into adult life, potential for neurogenesis, synaptic plasticity) and acquired (educational attainment, sustained intellectual, social, or physical activity in mid-life and old age) [23].
Limitations of the Study

In these six population samples from England and Wales the baseline response rate was good and unlikely to have been severely biased. Considerable attrition over time determines that those who remain in the study tend to have been younger and fitter at enrolment [24]. The cohort reported here is based on individuals who were selected for more detailed assessment at the baseline and year 2 waves. Selection to this group is weighted towards the cognitively impaired, but with random selection from the full population, and created an older sample than the remainder of the baseline sample who died within the study period. Causes of death were similar in the two groups. Because the characteristics of both samples are known, a sensitivity analysis backweighting for this process (and for biases arising from selection into the neuropathology cohort) adequately adjust for these sampling effects, though not for unknown biases. The number of interviews achieved for each individual during the study varied (96% had at least two and 30% had five or more). The AGECAT algorithm, applied to the data at last interview, has been validated against clinical diagnoses and shown to be comparable to Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised (DSM-IIIR) [25]. Fieldwork interviews were rarely started in acutely ill individuals so that diagnoses are unlikely to be influenced by confusional states. Although dementia on death certificates, insensitive but highly specific, was used to find incident dementia between last interview and death it was not used to indicate that an individual was not demented [26,27]. Clinical judgements from informant reports were based on DSM-III-R, consistent with the previous validation studies using the GMS instrument. The extent of misclassification that would be necessary to create the findings observed here would need to be extreme and, when including only those with most recent interview data, the results are not affected. Our previous report, clinically allocating differential diagnoses with a predominance of mixed pathology, remains robust [7]. The factors identified in this analysis coexist and may interact mechanistically. Our analysis does not allow us to elucidate causal directions, which are better investigated using longitudinal analysis and in experimental work. A formal analysis of interactions between vascular and degenerative pathologies will be reported separately but the models here did not reveal interactions. The low prevalence of Lewy bodies in this sample (,10%), reflects methods that were not optimised for the detection and screening of asynucleinopathy because the neuropathology protocol predates the recognition of dementia with Lewy bodies, and the discovery of asynuclein in Lewy bodies. Recent data on a subset of this cohort show synucleinopathy in 37% but no strong association with dementia [28]. Our data on brain weight are based on comparisons within sex but this measure does not distinguish atrophy from innate smallness, an issue that can only be addressed by systematic measurement of total cranial volume. Nor does it distinguish the contribution of vascular and neurodegenerative processes, or other correlates such as synaptic and dendritic loss that were not routinely measured in the pathology protocol. Standardised and validated assessment of vascular pathology is also needed in studies of the pathological correlates of dementia
Discussion
MRC CFAS shows that it is possible to set up and sustain a brain donation programme from a geographically dispersed, population-based study, which is not biased in terms of gender, social class, education, institutionalisation, or access to health care. The resulting brain donor sample is of sufficient size to generate meaningful estimates of AR at death associated with specific pathologies and contributes significantly to understanding the pathobiology of dementia on the basis of epidemiological neuropathology. It also allows the separation of factors that might be amenable to modification from others that may not. The main contributors to AR at death for dementia in MRC CFAS were age (18%), small brain (12%), neocortical neuritic plaques (8%) and neurofibrillary tangles (11%), small vessel disease (12%), multiple vascular pathologies (9%), and hippocampal atrophy (10%). Other significant factors include cerebral amyloid angiopathy (7%) and Lewy bodies (3%). Earlier CFAS analysis showed that Alzheimer pathology and vascular disease are frequently found in both demented and nondemented people [7]. In the present more detailed analysis, with larger numbers, a moderate or severe neocortical NFT score emerged as the best pathological discriminator between the demented and nondemented groups. SVD emerged as an independent contributor to dementia risk in keeping with the evidence that SVD is the substrate for subcortical vascular dementia and contributes a major part of the burden of vascular cognitive impairment in the population [19]. We included deep subcortical WMLs within the vascular disease variable on the basis of evidence that they arise through vascular mechanisms [20]. The estimates of AR at death reveal the relative importance of conventional pathological measures at the population level and show a range of pathological features contributing independently to dementia. The major independent effects of age and relative low brain weight are interesting. The findings imply that other factors, not captured in this standardised approach to pathological analysis, are determinants of cognitive trajectory in older people. These
17
Table 6. Sensitivity analysis: Imputation models.
Multivariable Model Original OR 95%CI
Imputed OR 95% CI
Imputation Model OR 95%CI AR 95% CI
Age at death ,80 y 8089 y $90 y Brain weight for sex Low Average High Neuritic plaques in neocortex None or mild Moderate or severe Tangles in neocortex None Mild Moderate or severe Tangles in hippocampus None or mild Moderate or severe CAA None Mild Moderate or severe Lewy bodies No Yes Overall vascular pathology None Infarcts/haemorrhage SVD/DWML/lacunes Both Hippocampal atrophy None Mild Moderate Severe Brainstem neuronal loss None Mild Moderate Severe doi:10.1371/journal.pmed.1000180.t006 Not included Not included 1.0 2.7 3.3 9.9 1.0 1.54.9 1.48.0 1.854 6 7 7 112 013 013 1.0 1.8 3.4 0.9,3.7 1.57.5 1.0 1.5 2.8 1.0 0.82.9 1.45.6 1.0 1.3 1.9 0.62.6 0.94.2 4 010 1.0 2.4 3.7 4.8 0.412 1.5,9.6 1.912 1.0 1.9 3.1 4.0 0.311 1.46.9 1.89.2 1.0 1.4 3.3 4.2 0.29.4 1.47.5 1.89.9 20 833 1.0 3.5 1.39.3 1.0 3.7 1.68.9 1.0 2.2 0.95.7 2 05 1.0 1.8 2.9 0.83.8 1.26.8 1.0 1.7 3.5 0.93.3 1.77.4 1.0 1.5 3.8 0.73.0 1.78.2 4 09 Not included Not included 1.0 1.8 1.03.3 5 012 1.0 1.0 7.1 0.51.8 2.322 1.0 1.0 6.3 0.61.8 2.615 1.0 0.7 4.6 0.41.4 1.812 10 517 1.0 9.7 2.143 1.0 4.4 1.612 1.0 3.9 1.311 7 317 4.1 2.1 1.0 1.99.2 1.04.2 4.3 1.8 1.0 2.09.2 0.93.4 4.3 1.7 1.0 1.99.6 0.93.4 12 4 519 09 1.0 2.5 3.4 1.15.8 1.48.3 1.0 2.3 4.3 1.14.8 2.09.5 1.0 2.1 4.2 1.04.5 1.89.6 7 8 114 216
18
[29]. Perhaps the greatest difficulty in interpreting these data is that they derive from individuals who have died. People with dementia live for a variable length of time during which burdens of neuropathology are assumed to change. To extrapolate from this sample to an equivalent cross section of the living older population is problematic but, in the absence of methods to achieve in vivo measurement of all pathologies, this is the closest estimate it is currently possible to produce. In due course these data can be
combined with modelling of in vivo population pathology derived from techniques to assess vascular and neuropathological changes (e.g., amyloid positron emission tomography [PET] scans). The pathological features that are associated with dementia in this analysis are well supported by data from other large community-based and population-based studies. There is general agreement from studies of older people in the UK and the US that dementia is predominantly associated with mixed vascular and

Alzheimer lesions together with other contributions of lesser degree (e.g., synucleinopathy) [3032]. Those studies also contribute important insights into the potential interactions of vascular and degenerative pathologies that are not dealt with in the present analysis [3035]. Some studies have emphasised the significance of microscopic infarcts compared to macroscopic infarcts in explaining the relationship between pathology and dementia [30,32,34], whereas others have not demonstrated an independent association of dementia risk with microscopic infarction [31,35]. In the present analysis we did not treat microinfarcts as a single pathological variable. Rather we chose to incorporate them into a global assessment of significant intrinsic SVD that also included microscopic evidence of severe arteriolar sclerosis and the presence of severe white matter attenuation. The Adult Changes in Thought study (ACT) has estimated the OR for dementia associated with Lewy bodies to be 5.1 (95% CI 1.37 18.96) on the basis of a-synuclein immunocytochemistry compared to 3.5 (95% CI 1.39.3) in this study using a less reliable method of detection. The estimates of AR at death for Lewy body pathology are 10% in ACT and 3% in MRC CFAS as reported here. In a subgroup of this CFAS cohort we demonstrated synucleinopathy in 37% of donated brains [28]. Other large cohorts have reported no clear predictive relationship between Lewy body pathology and dementia [36]. Interpretation of data on Lewy body pathology in published multivariable analyses is further complicated by the recent recognition of amygdala predominant disease that may not be reliably detected using some screening protocols. Another pathology recently emphasised in older people is hippocampal sclerosis (HS), which has been shown to contribute a relative risk for dementia of 2.43 (95% CI 1.015.85). This is a microscopic diagnosis that was not included as a separate variable in our data. While we did include macroscopic hippocampal atrophy, and found that it contributes 10% of the AR at death for dementia, it is important for future studies to determine the correlation between macroscopic changes and the microscopic
features of HS, which are also not yet the subject of diagnostic consensus or interlaboratory validation. The present study supports the view that interventions that modify neuropathology related to dysmetabolism of specific proteins (bA4, tau) have the potential to impact on the population burden of dementia. In the context of presymptomatic treatment many individuals without risk of developing dementia would also be treated unless the predictive ability of clinical tests improves dramatically. However the estimates in this analysis indicate that individual pathologies contribute only modestly to the overall risk of dementia and emphasise the need to develop a range of protective strategies. Other factors, potentially less amenable to intervention play a role including age, and underlying innate or acquired factors relating to brain reserve, which, along with the effects of multiple pathological comorbidities, all play a part in the manifestation of dementia at the level of the population as a whole.
Acknowledgments
We thank the CFAS population, their families, and their carers for their participation and the generous gift of brain donation. Local staff in each CFAS centre undertook the enrolment, brain donations, and neuropathology assessments: J. Xuereb (Cambridge); J. MacKenzie, J. Broome (Liverpool); T. Polvikoski (Newcastle upon Tyne); M.M. Esiri (Oxford).
Author Contributions
ICMJE criteria for authorship read and met: FEM CB JL IM SBW PI. Agree with the manuscripts results and conclusions: FEM CB JL IM SBW PI. Designed the experiments/the study: FEM CB JL IM SBW PI. Analyzed the data: FEM CB JL SBW. Collected data/did experiments for the study: FEM JL SBW PI. Enrolled patients: CB JL. Wrote the first draft of the paper: FEM CB PI. Contributed to the writing of the paper: FEM CB IM SBW. Primary statistical analyses and modeling: FEM. Coordinated the study and supervised the enrollment of respondents: CB PI IM. Pathological validation of cases in series: JL. Neuropathological analysis of cases for the study: SBW.
References
1. McKeith IG, Galasko D, Kosaka K, Perry EK, Dickson DW, et al. (1996) Consensus guidelines for the clinical and pathological diagnosis of dementia with Lewy bodies (DLB): report of the consortium on DLB International Workshop. Neurology 47: 11131124. 2. Mirra SS, Heyman A, McKeel D, Sumi SM, Crain BJ, et al. (1991) The consortium to establish a registry of Alzheimers disease (CERAD) Part II. Standardization of the neuropathologic assessment of Alzheimers disease. Neurology 41: 479486. 3. Nation, Reagan Institute Working Group on Diagnostic Criteria for the Neuropathological Assessment of Alzheimers Disease (1997) Consensus recommendations for the postmortem diagnosis of Alzheimers disease. Neurobiol Aging 18(Suppl 1): S1S2. 4. Cechetto D, Hachinski V, Whitehead S (2007) Vascular risk factors and Alzheimers disease. Expert Rev Neurotherapeutics 8: 743750. 5. Green R, Schneider L, Hendrix S, Zavitz K, Swabb E (2008) Safety and efficacy of tarenflurbil in subjects with mild Alzheimers disease: results from an 18month multi-center phase 3 trial. Alzheimers and Dementia 4(Suppl1): T165. 6. Wischik C, Bentham P, Wischik D, Seng K (2008) Tau aggregation inhibitor (TAI) therapy with remberTM arrests disease progression in mild and moderate Alzheimers disease over 50 weeks. Alzheimers and Dementia 4(Suppl1): T167. 7. Neuropathology Group of MRC CFAS (2001) Pathological correlates of lateonset dementia in a multicentre, community-based population in England and Wales. Lancet 357: 169175. 8. Petrovitch H, Ross G, Steinhorn S, Abbott RD, Markesbery W, et al. (2005) AD lesions and infarcts in demented and non-demented Japanese-American men. Ann Neurol 57: 98103. 9. Halliday G, Ng T, Rodriguez M, Harding A, Blumbergs P, et al. (2002) Consensus neuropathological diagnosis of common dementia syndromes: testing and standardising the use of multiple diagnostic criteria. Acta Neuropathol 104: 7278. 10. Zaccai J, Ince P, Brayne C (2006) Population-based neuropathological studies of dementia: design, methods and areas of investigation - a systematic review. BMC Neurology 6: 2. 11. MRC CFAS (1998) Cognitive function and dementia in six areas of England and Wales: the distribution of MMSE and prevalence of GMS organicity level in the MRC CFA study. Psychol Med 28: 319335. 12. Copeland J, Kelleher M, Kellet J, Gourlay AJ, Gurland BJ, et al. (1976) A semistructured clinical interview for the assessment of diagnosis and mental state in the elderly: the Geriatric Mental State Schedule: I. Development and reliability. Psychol Med 6: 439449. 13. Huppert F, Brayne C, Gill C, Paykel E, Beardsall L (1995) CAMCOGa concise neuropsychological test to assist dementia diagnosis: socio-demographic determinants in an elderly population sample. Br J Clin Psychol 34: 529541. 14. Copeland J, Dewey M, Griffiths-Jones H (1986) Computerised psychiatric diagnostic system and case nomenclature for elderly subjects. Psychol Med 16: 8999. 15. Fernando M, OBrien J, Perry R, McMeekin W, Jaros E, et al. (2004) Comparison of pathology of cerebral white matter with postmortem MRI in the elderly. Neuropathol Appl Neurobiol 30: 385395. 16. Land M, Vogel C, Gefeller O (2001) Partitioning methods for multifactorial risk attribution. Stat Methods Med Res 10: 217230. 17. Efron B, Tibshirani R (1993) An introduction to the bootstrap. New York: Chapman and Hall. 18. Rubin D (1996) Multiple imputation after 18+ years (with discussion). J Am Stat Assoc 91: 473489. 19. Roman G, Erkinjuntti T, Wallin A, Pantoni L, Chui H (2002) Subcortical ischaemic vascular dementia. Lancet Neurol 1: 426436. 20. Fernando M, Simpson J, Matthews F, Brayne C, Lewis C, et al. (2006) White matter lesions in an unselected cohort of the elderly: molecular pathology suggests origin from chronic hypoperfusion injury. Stroke 37: 13911398. 21. Kayed R, Head E, Thompson J, McIntire TM, Milton SC, et al. (2003) Common structure of soluble amyloid oligomers implies common mechanism of pathogenesis. Science 300: 486489. 22. Terry R, Masliah E, Salmon D, Butters M, DeTeresa R, et al. (1991) Physical basis of cognitive alterations in Alzheimer disease: synaptic loss is the major correlate of cognitive decline. Ann Neurol 30: 572580.
19
13
23. Valenzuela M (2008) Brain reserve and the prevention of dementia. Curr Opin Psychiatry 21: 296302. 24. Matthews F, Brayne C, MRC CFAS Investigators (2005) The incidence of dementia in England and Wales: findings from the five identical sites of the MRC CFA Study. PLoS Med 2: e193. doi:10.1371/journal.pmed.0020193. 25. Kay D, Dewey M, McKeith I, OCuill M, McCracken C, et al. (1998) Do experienced diagnosticians agree about the diagnosis of dementia from survey data? The effects of informants reports and interviewers vignettes. Int J Geriatric Psychiat 13: 852862. 26. Bjertness E, Torvik A, Ince P, Edwardson J (1998) Validation of Norwegian death certificates on dementia in residents of nursing homes. Epidemiology 9: 584586. 27. Macera C, Sun R, Yeager K, Brandes D (1992) Sensitivity and specificity of death certificate diagnoses for dementing illnesses, 1988-1990. J Am Geriatr Soc 40: 479481. 28. Zaccai J, Brayne C, McKeith I, Matthews F, Ince P (2008) Patterns and stages of alpha-synucleinopathy: relevance in a population-based cohort. Neurology 70: 10421048. 29. Jellinger K, Attems J (2007) Neuropathological evaluation of mixed dementia. J Neurol Sci 257: 8087.
30. Sonnen JA, Larson EB, Crane PK, Haneuse S, Li G, et al. (2007) Pathological correlates of dementia in a longitudinal population-based sample. Ann Neurol 62: 406413. 31. Schneider JA, Arvanitakis Z, Bang W, Bennett DA (2007) Mixed brain pathologies account for most dementia in community dwelling older persons. Neurology 69: 21972204. 32. White L, Small BJ, Petrovich H, Ross GW, Masaki K, et al. (2005) Recent clinical-pathologic research on the causes of dementia in late life: update on from the Honolulu Asia Aging study. J Geriatr Psychiatry Neurol 18: 224227. 33. Schneider JA, Wilson RS, Bienias JL, Evans A, Bennett DA (2004) Cerebral infarctions and the likelihood of dementia from Alzheimer disease pathology. Neurology 62: 11481155. 34. Troncoso JC, Zonderman AB, Resnick SM, Crain B, Pletnikova O, et al. (2008) Effects of infarcts on dementia in the Baltimore Longitudinal study of Ageing. Ann Neurol 64: 168176. 35. Chui HC, Zarow C, Mack WJ, Ellis WG, Zheng L, et al. (2006) Cognitive impact of subcortical vascular and Alzheimers disease pathology. Ann Neurol 60: 677667. 36. Parkkinen L, Kauppinen T, Pirttila T, Autere JM, Alafuzoff I (2005) Alphasynuclein pathology does not predict extrapyramidal symptoms or dementia. Ann Neurol 57: 8291.
20
PLoS Medicine | www.plosmedicine.org 14 November 2009 | Volume 6 | Issue 11 | e1000180

Editors Summary
Background. Losing ones belongings and forgetting peoples names is often a normal part of aging. But increasing forgetfulness can also be a sign of dementia, a group of symptoms caused by several disorders that affect the structure of the brain. The commonest form of dementia is Alzheimer disease. In this, protein clumps called plaques and neurofibrillary tangles form in the brain and cause its degeneration. Vascular dementia, in which problems with blood circulation deprive parts of the brain of oxygen, is also common. People with dementia have problems with two or more cognitive functionsthinking, language, memory, understanding, and judgment. As the disease progresses, they gradually lose their ability to deal with normal daily activities until they need total care, their personality often changes, and they may become agitated or aggressive. Dementia is rare before the age of 65 years but about a quarter of people over 85 years old have dementia. Because more people live to a ripe old age these days, the number of people with dementia is increasing. According to the latest estimates, about 35 million people now have dementia and by 2050, 115 million may have the disorder. Why Was This Study Done? There is no cure for dementia but many drugs designed to modulate specific abnormal (pathological) changes in the brain that can cause the symptoms of dementia are being developed. To assess the likely impact of these potentially expensive new therapies, experts need to know what proportion of dementia is associated with each type of brain pathology. Although some brain changes can be detected in living brains with techniques such as computed tomography brain scans, most brain changes can only be studied in brains taken from people after death (post mortem brains). In this study, which is part of the UK Medical Research Council Cognitive Function and Ageing Study (MRC CFAS), the researchers look for associations between dementia in elderly people and pathological changes in their post mortem brains and estimate the attributable-risk (AR) for dementia at death associated with specific pathological features in the brain. That is, they estimate the proportion of dementia directly attributable to each type of pathology. What Did the Researchers Do and Find? Nearly 20 years ago, the MRC CFAS interviewed more than 18,000 people aged 65 years or older recruited at six sites in England and Wales to determine their cognitive function and their ability to deal with daily activities. 20% of the participants, which included people with and without cognitive impairment, were then assessed in more detail and invited to donate their brains for post mortem examination. As of 2004, 456 individuals had donated their brains. The dementia status of these donors was established using data from their assessment interviews and death certificates, and from interviews with relatives and carers, and their brains were carefully examined for abnormal changes. The researchers then used statistical methods to estimate the AR for dementia at death associated with various abnormal brain changes. The main contributors to AR for dementia at death included age (18% of dementia at death was attributable to this factor), plaques (8%), and neurofibrillary tangles (11%) in a brain region called the neocortex, small blood vessel disease (12%), and multiple abnormal changes in blood vessels (9%). What Do These Findings Mean? These findings suggest that multiple abnormal brain changes determine the overall burden of dementia. Importantly, they also suggest that dementia is often associated with mixed pathological changesmany people with dementia had brain changes consistent with both Alzheimer disease and vascular dementia. Because people with dementia live for variable lengths of time during which the abnormal changes in their brain are likely to alter, it may be difficult to extrapolate these findings to living populations of elderly people. Furthermore, only a small percentage of the MRC CFAS participants have donated their brains so the findings of this study may not apply to the general population. Nevertheless, these findings suggest that the new therapies currently under development may do little to reduce the overall burden of dementia because most peoples dementia involves multiple pathologies. Consequently, it may be necessary to develop a range of strategies and combination therapies to deal with the ongoing dementia epidemic. Additional Information. Please access these Web sites via the online version of this summary at http://dx.doi.org/10. 1371/journal.pmed.1000180.
N N N N N
The US National Institute on Aging provides information for patients and carers about forgetfulness and about Alzheimer disease (in English and Spanish) The US National Institute of Neurological Disorders and Stroke provides information about dementia (in English and Spanish) The UK National Health Service Choices Web site also provides detailed information for patients and their carers about dementia and about Alzheimer disease MedlinePlus provides links to additional resources about dementia and Alzheimer disease (in English and Spanish) More information about the UK Medical Research Council Cognitive Function and Ageing Study (MRC CFAS) is available
21
Neglected Diseases
1
Packages of Care for Dementia in Low- and MiddleIncome Countries

Martin J. Prince1*, Daisy Acosta2, Erico Castro-Costa1,3, Jim Jackson4, K. S. Shaji5
1 Health Service and Population Research Department, Institute of Psychiatry, Kings College London, London, United Kingdom, 2 Universidad Nacional Pedro Henriquez Urena (UNPHU), Santo Domingo, Dominican Republic, 3 Centro de Pesquisa Rene Rachou/Fiocruz, Belo Horizonte, Minos Gerais, Brazil, 4 Alzheimer Scotland, Edinburgh, United Kingdom, 5 Medical College, Thrissur, India
This is the fifth in a series of articles highlighting the delivery of packages of care for mental health disorders in low- and middleincome countries. Packages of care are combinations of treatments aimed at improving the recognition and management of conditions to achieve optimal outcomes.
Introduction
Dementia is a chronic organic brain syndrome, characterised by progressive impairment of multiple cortical functions, including memory, learning, orientation, language, comprehension, and judgement. Diagnosis requires decline in cognitive function and independent living skills (Box 1) [1]. However, for carers and people with dementia, the behavioural and psychological symptoms of dementia (BPSD) affect most quality of life, are an important cause of carer strain [2], and a common reason for institutionalisation [3]. Alzheimers disease, vascular dementia, dementia with Lewy bodies, and frontotemporal dementia are the most common dementia subtypes, but mixed pathologies may be the norm [4]. Some rare causes (subdural haematoma, normal pressure hydrocephalus, hypercalcaemia, and deficiencies of thyroid hormone, vitamin B12, and folic acid) can be treated. Otherwise, the progressive course of dementia cannot be altered, but symptomatic treatments and support can be helpful. Dementia mainly affects older people. Few cases start before the age of 65 y, after which prevalence doubles with every 5-y increase in age [5]. Globally, 24.3 million people are affected by dementia and 4.6 million new cases occur annually [6]. The prevalence of dementia is expected to double every 20 y, reaching 81.1 million by 2040, an increase of 100% in developed countries and of more than 300% in India, China, and their neighbours. Prevalence is lower in low- and middle-income countries (LMICs) than in high income countries (HICs) [6], perhaps because of underdetection of mild cases [7]. Nevertheless, most people with dementia live in LMICs60% in 2001 rising to 71% by 2040 [6]. Dementia contributes 11.2% of years lived with disability among people aged 60 y and over, a higher proportion than stroke (9.5%), musculoskeletal disorders (8.9%), cardiovascular disease (5.0%), and cancer (2.4%) [8]. Its global cost is estimated to be US$317 billion, 77% of this total arising in HICs where formal sector care costs increase with disease progression, and institutionalization is the main cost driver [9]. Family care is more important in resourcepoor countries, accounting for 56% of costs in low-income countries, 42% in middle-income countries, and 31% in HICs [9]. In a pilot study in 26 LMIC centers, carers were economically disadvantaged [10]. A fifth of carers had cut back on paid work, and paid carers were common, which added to the economic strain [10]. Compensatory benefits were practically nonexistent [10,11]. In three qualitative studies in India, features of dementia were widely recognized and named [1214]. However, dementia was perceived as normal ageing rather than as a medical condition. The consequences
were limited help seeking [13] despite disability and carer strain [15], no structured training on the recognition and management of dementia, and no constituency to advocate for more responsive care services [14]. People with dementia were excluded from residential care [13]. Carers misinterpreted BPSD as deliberate misbehavior [14]. BPSD can also lead to stigma and blame attaching to the carers [2]. In India, likely causes of dementia were cited as neglect by family members, abuse, tension and lack of love [13]. In this article, we focus on the effective management of dementia in LMICs, reviewing the evidence on efficacy of interventions and their delivery derived from LMICs where possible. Given the paucity of relevant evidence from LMICs, we also cite systematic reviews and meta-analyses based on trials from HICs. On the basis of our review we propose a package of carea combination of treatments aimed at improving the recognition and management of conditions to achieve optimal outcomesfor dementia.
The Evidence on the Management of Dementia

The principal goals of management of dementia are: early diagnosis; optimization of physical health, cognition, activity, and wellbeing; detection and treatment of BPSD; and the provision of information and long-term support to carers. The evidence base for dementia care comes, overwhelmingly, from HICs (Table 1). All the studies discussed below refer to HICs, unless otherwise specified.
Detection and Diagnosis of Dementia

Many cases of dementia, particularly in LMICs, go undetected,in part because of lack of awareness. Awareness of this disorder can be boosted by dissemination of information from governments, health care providers, and media. Help-seeking can be encouraged by improved case-finding. In India and Brazil, for example, community
Citation: Prince MJ, Acosta D, Castro-Costa E, Jackson J, Shaji KS (2009) Packages of Care for Dementia in Low- and Middle-Income Countries. PLoS Med 6(11): e1000176. doi:10.1371/journal.pmed.1000176 Academic Editor: Vikram Patel, London School of Hygiene and Tropical Medicine, United Kingdom Published November 3, 2009 Copyright: 2009 Prince et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: No specific funding was received for this piece. Competing Interests: DA is Chair of Alzheimers Disease International. JJ is the ex-Chief Executive of Alzheimer Scotland. Abbreviations: BPSD, behavioural and psychological symptoms; ChEI, cholinesterase inhibitor; HIC, high income country; LMIC, low- and middle-income country; RCT, randomised controlled trial. * E-mail: m.prince@iop.kcl.ac.uk Provenance: Commissioned; externally peer reviewed.
22

PackagesofCareforDementiainLow-andMiddle-IncomeCountries
Summary Points
N N N N N
Two-thirds of people with dementia live in low- and middleincome countries (LMICs), where there are few services available and levels of awareness and help-seeking are low. After early diagnosis, the principal goals for management of dementia are optimising physical health, cognition, activity, and wellbeing; detecting and treating behavioural and psychological symptoms (BPSD); and providing information and long-term support to carers. Routine packages of continuing care should comprise diagnosis coupled with information, regular needs assessments, physical health checks, and carer support, and where necessary carer training, respite care, and assessment and treatment of BPSD. Care can be delivered by trained primary care teams working in a collaborative care framework. Continuing care with practice-based care coordination, and community outreach are essential components of this model. Efficient care delivery in LMICs involves integrating dementia care with that of other chronic diseases and communitysupport programs for the elderly and disabled.
Box 1. International Classification of Diseases 10 Criteria for Dementia and Commonly Occurring Features
Diagnostic criteria (ICD-10 Diagnostic Guidelines) The primary requirement for diagnosis is evidence of a decline in both memory and thinking which is sufficient to impair personal activities of daily living. The impairment of memory typically affects the registration, storage, and retrieval of new information, but previously learned and familiar material may also be lost, particularly in the later stages. There is also impairment of thinking and of reasoning capacity, and a reduction in the flow of ideas. The processing of incoming information is impaired, in that the individual finds it increasingly difficult to attend to more than one stimulus at a time, such as taking part in a conversation with several persons, and to shift the focus of attention from one topic to another. If dementia is the sole diagnosis, evidence of clear consciousness is required. The above symptoms and impairments should have been evident for at least 6 months for a confident clinical diagnosis of dementia to be made [1]. The diagnostic criteria do not convey a sense of the typical progression of the disorder: A person with mild dementia has noticed deterioration in their memory for recent events. For example, they may forget that their daughter had visited the previous day. They also find it difficult to concentrate, think flexibly, plan, and take decisions. They are likely to feel bewildered, anxious and sad. They may become angry and defensive when others point out errors. A person with moderate dementia has severe memory problems. Only early memories are retained. Recent events are not remembered, or rapidly forgotten. They may not know the day, date or time of day. They often do not know where they are. They cannot communicate clearly, having problems finding the right word and using the wrong words. They may hear voices or see things that are not there, and can develop false beliefs, for example that children are entering their house and stealing things. They are likely to be anxious, sad, bewildered, and can become agitated or aggressive. A person with severe dementia has complete memory loss. They may no longer recognise their close family. They have severe speech difficulties or are unable to communicate. They may be apathetic and totally inactive, but at times can be agitated and verbally and physically aggressive. They cannot coordinate their physical movements; may have lost the ability to walk and feed themselves and have difficulty swallowing. They are likely to be incontinent.
health care workers could, with a few hours training, identify dementia in the community with a positive predictive value of 66% [16,17]. Population screening for dementia is not considered cost-effective even in HICs [18]. Selective screening can be accomplished by cognitive testing or by informant report of cognitive and functional decline. The Mini-Mental State Examination [19] is widely used in HICs, and adapted versions have been developed for use in many LMICs [2022]. However, this assessment takes 10 min to administer and is prone to educational and cultural bias [23,24]. A systematic review of brief screening assessments identified three tests (the General Practitioner Assessment of Cognition, Mini-Cog, and Memory Impairment Screen) that took less than 5 min to administer and were considered suitable and valid for routine use in general practice [25]; none of these tests has been validated in LMICs. Dementia diagnosis requires cognitive testing, clinical interview, informant interview, and physical examination. The 10/66 Dementia Research Groups culture- and education-fair diagnostic protocol is validated in many LMICs [26,27], but requires adaptation for clinical use. Practice guidelines in HICs [18] advocate a routine dementia screen to exclude treatable causes, which includes haematology, biochemistry, thyroid function tests, vitamin B12 and folate levels. The feasibility and cost-effectiveness of this approach needs to be tested in LMICs. Finally, a recent systematic review based on evidence collected in HICs identified good practice for disclosing dementia diagnosis. This practice should include: preparation; integrating family members; exploring the patients perspective; disclosing the diagnosis; responding to patient reactions; focusing on quality of life and wellbeing; planning for the future; and communicating effectively [28]. The person assessed should be asked if they, and/ or others wish to be told the diagnosis. If so, they should be given information about the signs, symptoms, and course of dementia, available treatments, care and support services.
Physical Assessment
To avoid missing underlying conditions, a physical assessment is recommended before specific treatments for BPSD are initiated [29], and should be a regular part of care for all patients. BPSD may sometimes be caused by pain, constipation, and urinary tract infection,
although such associations are not always observed [30]. Pain is common and poorly controlled in severe dementia [31]. Hearing and visual impairment impede communication and exacerbate disorientation, and deafness predicts rapid cognitive decline [32,33]. Visual and auditory impairment can be associated with hallucinations and delusions [34,35]. Studies indicate that all these impairments are overrepresented in people with cognitive impairment [3638]. There have been very few trials of the effects of physical assessments and interventions on the course of dementia. In a randomised controlled trial (RCT), pain assessment among nursing home residents with dementia was associated with increased analgesic use, reduced pain, and improvements in staff morale [39]. Uncontrolled studies show that audiological assessment is feasible, that hearing aids can be beneficial [40], and that referral to an optician to improve visual acuity may reduce visual hallucinosis [34]. Nutrition is often impaired because of apathy, aversive feeding
23
Table 1. The evidence in support of dementia management.
Outcome Identifying dementia Assessing needs
Intervention
Evidence from HICs
Evidence from LMICs Adapted versions of the Mini Mental State Examination in India, China, Brazil [2022]; Community Screening Instrument for Dementia (CSID) [102,103] Feasibility and reliability in Brazil, community study [94] Small RCT of donepezil in Brazil [52]; Open label trials of galantamine in Brazil [53] and China [54] Small pilot trial of cognitive stimulation in Brazil [45]
Validity of screening Systematic review of screening properties of brief cognitive test measures and informant screening assessments, validated in general practice or community settings [25] Development, reliability, and validity in UK, US, and Sweden [104] Camberwell Assessment of Needs in the Elderly (CANE) ChEIs Meta-analyses of RCTs of donepezil, rivastigmine, and galantamine in Alzheimers disease [4851]; Meta-analysis of RCTs of donepezil, rivastigmine, and galantamine for vascular cognitive impairment [57] Meta-analysis of RCTs of memantine [56]; Meta-analysis of RCTs of memantine for vascular cognitive impairment [57] Meta-analysis of RCTs of reminiscence therapy [47]; RCT of cognitive stimulation [44]; Meta-analyses of RCTs of cognitive training and rehabilitation [46]a Meta-analysis of RCTs of haloperidol [60] Meta-analyses of RCTs of risperidone, olanzapine, quetiapine, and aripiprazole for the treatment of BPSD, aggression, and agitation [61,62] Meta-analysis of RCTs of ChEIs for the treatment of BPSD [51] Meta-analyses of RCTs of memantine in severe dementia [56,58] Small RCT of citalopram for the treatment of agitation [73]; Meta-analysis of RCTs of trazodone [72]a; RCT of sertraline [74]a Two small RCTs of carbamezepine [76,77] Meta-analysis of RCTs of massage and light touch [81]; Meta-analysis of the efficacy of aroma therapy [105]; Meta-analysis of RCTs of Snoezelen (multisensory stimulation) [79]a; Meta-analysis of RCTs of light therapy [78]a Meta-analysis of RCTs of antidepressants to treat depression in dementia [70]a Meta-analyses of RCTs and nonrandomised trials of carer interventions for improving mood in the care recipient [84,86] Meta-analyses of RCTs of atypical antipsychotic drugs for the treatment of psychosis in Alzheimers disease [61,62] Meta-analyses of controlled studies (including nonrandomised trials) of carer interventions in general [86] and the specific effects of psychoeducational, interventions, CBT, counselling, and respite care [84]; Meta-analysis of RCTs of respite care [87]a Meta-analysis of controlled studies (including nonrandomised trials) of carer interventions in preventing or delaying institutionalisation [84,88]
Cognitive impairment
Memantine Nonpharmacological interventions Behavioural symptoms Conventional antipsychotic drugs Atypical (second-generation) antipsychotic drugs ChEIs Memantine SSRI antidepressants Anticonvulsants Environmental/ sensory interventions Depression Antidepressant pharmacotherapy Carer interventions Psychosis Carer strain, subjective wellbeing, and psychological morbidity Atypical antipsychotic drugs Carer interventions
Two small RCTs of a carer education and training intervention in Goa, India [89] and in Moscow, Russia [90]
Institutionalisation Carer intervention
The evidence from these trials and meta-analyses is inconclusive, usually because of a combination of the following factors: small trials; small number of trials; flawed methodology (see text for details). SSRI, selective serotonin reuptake inhibitor. doi:10.1371/journal.pmed.1000176.t001
behaviours, poor dental health, and dysphagia. Although difficult to sustain, nutritional supplementation improved nutrition among nursing home residents [41]; nutritional education for carers had the same effect in the community [42]. A vascular care secondary preventive intervention for people with dementia and cerebrovascular disease (part of current good practice guidelines [18]) that addressed hypercholesterolemia, hypertension, smoking, obesity, exercise, and micronutrient deficiency had no impact on subsequent cognition, disability, or institutionalisation [43].
that cognitive benefits from this intervention are similar to those for cholinesterase inhibitors (ChEIs). More specific cognitive training produced no benefits [46]. Cognitive rehabilitation, an individualised therapy designed to enhance residual cognitive skills and cope with deficits, showed promise in uncontrolled case series undertaken in HICs [46]. A meta-analysis of four trials of reminiscence therapy (the discussion of past activities, events, and experiences) [47] provides evidence for short-term improvement in cognition, mood, and carer strain, but the quality of these trials was poor.
Psychological Treatments
A well-conducted RCT of cognitive stimulation (reality orientation, games, discussions based on information processing rather than knowledge) conducted in the United Kingdom as a group intervention [44], and a small pilot trial from Brazil [45], suggest
Pharmacological Treatments
Targets for pharmacological treatment include cognitive impairment, behavioural symptoms (agitation and aggression), and psychological symptoms (depression, anxiety, and psychosis).
24

There is a strong evidence base for the efficacy of ChEIs (donepezil [48], rivastigmine [49], and galantamine [50]). The use of each of these drugs is associated with modest and comparable improvements in cognitive function, global clinical state, and activities of daily living [51]. The evidence base for ChEIs from LMICs is limited to one small RCT of donepezil in Brazil [52] and open-label trials of galantamine in Brazil [53] and China [54]. The efficacy of this class of drugs in severe dementia is unclear, although useful cognitive benefits were identified for galantamine [55]. A fourth drug for the treatment of cognitive impairment, memantine, has a different mode of action, and is well tolerated, but evidence for its efficacy is limited to people with moderate to severe dementia [56]. ChEIs and memantine are less efficacious in vascular dementia than in other forms of dementia [57]. Their efficacy for the treatment of disturbed behaviour is not established; manufacturer-sponsored licensing trials [51] and post hoc analyses [58] indicate small improvements that have not been confirmed in independent trials [59] and meta-analyses [56]. Meta-analyses of RCTs of haloperidol [60] and atypical antipsychotic drugs for the treatment of agitation [61] and BPSD [62] indicate small treatment effects, most evident for aggression [62,63]. Atypical antipsychotic drugs have also been widely prescribed for psychosis in dementia, but a meta-analysis of their efficacy indicated that only aripiprazole had a statistically and clinically significant effect [62]. Use of these drugs in dementia is associated with an increased risk of death and cerebrovascular adverse events [62,6468]. The benefits of antidepressant treatment in older people are clear [69], but a meta-analysis of their efficacy in people with dementia was inconclusive [70]. Only two small trials were included in this meta-analysis, one of which suggested a benefit of sertraline for some depression outcomes [71]. Antidepressants have also been proposed for the treatment of BPSD. A metaanalysis of two small RCTs of trazodone was inconclusive [72]. Citalopram showed efficacy over placebo for the treatment of agitation in one small RCT [73], while sertraline showed no benefit on any primary behavioural endpoint [74]. A systematic review of trials of anticonvulsants to treat BPSD found sodium valproate to be ineffective [75]. Carbamezepine may be more promising with large benefits noted for global clinical outcomes and agitation in one small parallel group trial [76] and more marginal effects in a small pilot trial [77]. in these studies were conducted in HICs, and many were nonrandomised [84]. Outcomes studied include carer strain, depression, and subjective wellbeing; behaviour disturbance and mood in the care recipient; and institutionalisation. Most carerfocused interventions seemed to reduce carer strain and depression, CBT having the largest impact on depression [84]. Psychoeducational interventions required the active participation of the carer (for example, in role-playing activities) to be effective [84]. Carer support increased carer wellbeing but no other outcomes [84]. For respite care, three methodologically flawed RCTs showed no benefit on any outcome [87]. However, nonrandomised studies suggest that respite care significantly reduces carer strain and psychological morbidity [84]. Interventions targeting the carer may also have small but significant beneficial effects upon the behaviour of the person with dementia [84]. A systematic review of ten RCTs indicated a 40% reduction in the pooled odds of institutionalisation [88]; the effective interventions were structured, intensive, and multicomponent, offering a choice of services and supports [84,88]. Two small trials in LMICs of a brief carer education and training intervention, one from India [89] and one from Russia [90] indicated much larger treatment effects on carer psychological morbidity [89] and strain [90] than typically seen for such interventions in HICs.
Delivery of Effective Interventions

The mechanisms by which effective dementia care treatments may be delivered in LMIC settings are summarized in Table 2.
Interventions to Increase Demand for Services

Alzheimers Disease International has identified raising awareness of dementia among the public, carers, and health workers as a global priority, with an increase in demands for services as one of the intended benefits [91]. Awareness can be raised in several ways. The establishment of a critical mass of informed carers can assist awareness-raising, provide advice and support to families, and work with Alzheimers disease associations to lobby for more services that better meet the needs of carers. Community solidarity can also effect change through support for health and social welfare policies based on equity and justice. Aware communities can provide support and reduce stigma and exclusion. Policymakers can be held to account by media campaigns and advocacy from committed NGOs. In HICs, awareness is growing rapidly, with the media playing an important role [92]. Media in LMICs can be receptive to these stories, but efforts are required to alert them to the importance of ageing and dementia, and to build their capacity to report the problem [92]. Intergenerational solidarity can be promoted through awareness-raising among children and young adults. In many LMICs, many people with dementia live in multigenerational households with young children, and children or children-in-law are the most frequent carers for people with dementia [10], and the most likely to initiate help-seeking. Finally, in LMICs the provision of disability pensions and carer benefits will inevitably increase requests for diagnostic assessment. Importantly, however, efforts to increase awareness must be accompanied by health system and service reform, so that help-seeking is met with a supply of betterprepared, more responsive services.
Sensory Therapy
Various sensory therapies have been proposed as treatments for BPSD but the evidence base for this approach is small and limited by the poor quality of the trials. Current evidence does not support the use of bright light therapy [78] or multisensory stimulation [79]. One small RCT of aromatherapy suggested considerable benefit across a range of behavioural outcomes [80]. Another small but well-conducted trial suggested that hand massage may be effective in reducing agitation [81,82]. More evidence is required to confirm efficacy, exclude the possibility of harm, and define the optimal content and mode of administration of sensory therapies in both HICs and LMICs, where the approach is untested.
Carer-Focused Interventions
A large literature attests to the benefits of carer interventions in dementia [83]. These include: psychoeducational interventions, often including carer training; psychological therapies such as cognitive behavioural therapy (CBT) and counselling; carer support; and respite care. Many interventions combine several of these elements. There are several systematic reviews and metaanalyses of these interventions [8488]; all of the constituent trials
Interventions to Improve the Capacity of Health Care Teams

Primary health care services in LMICs often fail older people because they are clinic-based and preoccupied with simple curative interventions [1315]. A paradigm shift is needed to encompass continuing care and support as part of a wider chronic
25
Table 2. Delivering dementia care treatments.
Step Increasing demand
How Improve awareness among general population, carers, health professionals Combat stigma Provide age-appropriate, accessible services Provide disability pensions for people with dementia, and carer benefits
By Whom Alzheimers associations, other chronic disease NGOs, health professionals, media, government
In What Settings Schools Public arena General practice, primary care Professional education institutions Primary care
Increasing capacity of health care teams
Training of community, primary, and secondary general health care providers (diagnosis, needs assessment, health checks, care package) Paradigm shift to chronic, continuing care Support and supervision from specialists
Medical, nursing, and rural health schools
Dementia specialist clinicians Community health care workers Nonspecialist clinicians Dementia specialist clinicians Community health care workers to deliver carer interventions and act as care coordinators [100] Nonspecialist clinicians to diagnose dementia, optimize physical health, assess needs, and plan community support Carers and other volunteers Primary care case managers
Secondary general health care Community Primary care Secondary general health care Memory clinics Community
Increasing recognition
Community-based case finding [16,17] Selective screening in primary and secondary care [25] Specialist centers
Adapting treatments to increase acceptability or reduce cost
Integrate dementia prevention and care into new chronic disease programs [93] Provide outreach (home-based assessment and care)
Primary care
Integrate long-term care and support interventions into programs for all dependent elderly Ensure carer interventions are culturally appropriate [89] Provide group interventions for carers [83] Develop and use cheaper generic versions of ChEIs and SSRI antidepressants, where available and indicated Practice-based programs to deliver effective treatments Train primary care staff as dementia and elder care case managers [100] Establish collaborative care [100] with involvement of specialist teams Community-based programs to deliver effective treatments Train carers and establish and support groups
Secondary general health care Primary care
Primary care clinicians Primary care health care workers Dementia care specialist Community health care workers
Community
Home outreach and individualized intervention Respite at home and homecare Addressing impact of the disorder on other health/ social outcomes Carer interventions
Peer (carer-to-carer) support and training Community volunteers Government social welfare agencies
Community
Carer benefits Respite care and homecare programs Disability benefits and social pensions for older people [106] Regular physical health checks Nutritional interventions SSRI, selective serotonin reuptake inhibitor. doi:10.1371/journal.pmed.1000176.t002
Community health care workers Case managers
Primary care
disease strategy. Given the frailty of many older people with dementia, there is also a need for outreach to assess and manage patients in their own homes. The World Health Organization
(WHO) Innovative Care for Chronic Conditions framework [93] proposes that the delivery of care for chronic conditions can be improved through a dialogue to build commitment for change,
26

extended and regular health care contact, a multisectoral approach, care centered on patients and families, support for patients in the community, and an emphasis on prevention. Dementia care should be an essential component of such chronic disease care strategies. Training of nonspecialist health professionals should focus on case-finding and on conveying the diagnosis to patients and carers together with information and support, needs assessment, and carer training and support. Training can be service-based as well as through changes to the medical and nursing school, public health, and rural health curricula. Medical and community care services should be planned and coordinated to respond to the increasing need for support as the disease progresses.
Practice-Based Programs to Deliver Effective Treatment

Most LMICs have insufficient specialists to provide national frontline dementia services. Diagnosis and needs assessment can be conducted in primary care, although increased specialist input to help with advice, inpatient or outpatient review of refractory BPSD, and respite care would be desirable. Routine physical reviews within practice-based programs are important but research is needed into the cost-effectiveness of haematological and biochemical screening, and the feasibility and effectiveness of interventions to improve hearing, vision, nutrition, and continence. A commitment to continuing care is essential, with regular review and physical health and needs assessments. Practice-based case managers can coordinate this process; this collaborative care model reduced BPSD and carer strain in an RCT in the US [100].
Interventions to Improve Recognition

The focus here should be first upon case-finding by community health workers [16,17], and then on selective screening coupled with simplified quasi-diagnostic algorithms and needs assessments [94] by nonspecialist clinicians. More research is needed to identify the most feasible and valid methods.
Community-Based Programs to Deliver Effective Treatments

Carer-support programs can be delivered individually or in groups by community health workers or by experienced carers. Carer strain, whether or not associated with BPSD, should trigger more intensive intervention that includes psychological assessment and depression treatment for the carer [101], respite, and carer education and training. Such interventions could be incorporated into horizontally constructed community-based programs that address the generic needs of frail, dependent older people and their carers, whether these needs arise from cognitive, mental, or physical disorders.
Interventions to Increase the Acceptability and Reduce the Costs of Treatments

At present, the costs of ChEIs are reimbursed in some Latin American countries and Chinese provinces, generic ChEIs are available in India, and Huperzine A, a cheap plant extract with similar properties to ChEIs is used in China [95]. In what is a very active and promising field for drug development [96], any new agent with radical disease-modifying properties will be very expensive, which raises important ethical and practical challenges to securing equitable access. An international effort leveraged by grass roots advocacy (similar to that mounted for antiretroviral treatment for HIV) might be required to secure affordable supplies for people with dementia living in LMICs. More positively, costs of primary care and community interventions can be minimised through task-shifting to nonspecialist and paraprofessional staff and integration with more broadly based chronic disease [93], disability, and elder care programmes [97,98]. In India a community-based palliative care program has been successfully extended to include dementia care [99]. Table 3. Packages of care for dementia.
Packages of Care for Dementia in LMICs

On the basis of our review, we have proposed a package of care for LMICs that considers the availability of resources for the detection, diagnosis, and treatment of dementia in this setting. In Table 3, the main features of this package are compared with typical recommendations for well-resourced HIC settings. To improve care for dementia in LMICs, we propose that the basic package of care should focus upon diagnosis coupled with information, regular needs assessments, physical health checks, and carer support. This package should be extended to include carer training, respite care, and assessment and treatment of BPSD
Low Resourced Settings Public information campaigns Community case-finding; Primary care diagnosis and needs assessment Provision of information for people with dementia and carers Continuity of support, coordinated by primary care case manager Regular case review with reassessment of needs; Specialist consultation and referral where available Cautious use of antipsychotic drugs and other experimental treatments for BPSD, not as first line treatment, and preferably initiated and reviewed by specialists Anticholinesterase drug treatment for cognitive impairment is unlikely to be seen as a cost-effective priority within most resource-poor systems Carer-support groups; Individualised community-based carer interventions and support Promote access to general health care
High Resourced Settings [107] Public information campaigns National network of specialist memory services, for diagnosis and as entry point into care pathway Provision of information for people with dementia and carers Continuity of support, coordinated by specialist case manager Integrated and continuing health and social care by multidisciplinary specialist care team Cautious use of antipsychotic drugs and other experimental treatments for BPSD, not as first line treatment, and initiated and reviewed by specialists Anticholinesterase drugs for memory impairment, when considered cost-effective, with specialist initiation and review Comprehensive carers strategy for people with dementia; Community personal support Effective care in general hospitals, intermediate care and care homes. Effective end of life care
27
where possible. We suggest that that care can best be delivered by trained primary care teams, with a paradigm shift towards chronic continuing care and community outreach. Practice-based care coordinators are an essential component of this package, ideally working within a collaborative-care framework. Finally, we note that care delivery will be more efficient when integrated with that of other chronic diseases, and more broadly based communitysupport programs for the elderly and disabled.
ICMJE criteria for authorship read and met: MJP DA ECC JJ KS. Wrote the first draft of the paper: MJP. Contributed to the writing of the paper: MJP DA ECC JJ KS.
References
1. World Health Organization (1992) International classification of diseases (10th Revision). Geneva: World Health Organization. 2. Ferri CP, Ames D, Prince M (2004) Behavioral and psychological symptoms of dementia in developing countries. Int Psychogeriatr 16: 441459. 3. de Vugt ME, Stevens F, Aalten P, Lousberg R, Jaspers N, et al. (2005) A prospective study of the effects of behavioral symptoms on the institutionalization of patients with dementia. Int Psychogeriatr 17: 577589. 4. Neuropathology Group Medical Research Council Cognitive Function and Aging Study (2001) Pathological correlates of late-onset dementia in a multicentre, community-based population in England and Wales. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS) [comment]. Lancet 357: 169175. 5. Ritchie K, Kildea D, Robine JM (1992) The relationship between age and the prevalence of senile dementia: a meta-analysis of recent data. Int J Epidemiol 21: 763769. 6. Ferri CP, Prince M, Brayne C, Brodaty H, Fratiglioni L, et al. (2005) Global prevalence of dementia: a Delphi consensus study. Lancet 366: 21122117. 7. Llibre Rodriguez JJ, Ferri CP, Acosta D, Guerra M, Huang Y, et al. (2008) Prevalence of dementia in Latin America, India, and China: a populationbased cross-sectional survey. Lancet 372: 464474. 8. (1996) The Global Burden of Disease. A comprehensive assessment of mortality and disability from diseases, injuries and risk factors in 1990 and projected to 2020. Cambridge (Massachusetts): Harvard School of Public Health, Harvard University Press. 9. Wimo A, Winblad B, Jonsson L (2007) An estimate of the total worldwide societal costs of dementia in 2005. Alzheimers and Dementia 8191. 10. 10/66 Dementia Research Group (2004) Care arrangements for people with dementia in developing countries. Int J Geriatr Psychiatry 19: 170177. 11. Prince MJ (2009) The 10/66 dementia research group - 10 years on. Indian Journal of Psychiatry 51: S8S15. 12. Cohen L (1995) Toward an anthropology of senility: anger, weakness, and Alzheimers in Banaras, India. Med Anthropol Q 9: 314334. 13. Patel V, Prince M (2001) Ageing and mental health in a developing country: who cares? Qualitative studies from Goa, India. Psychol Med 31: 2938. 14. Shaji KS, Smitha K, Praveen Lal K, Prince M (2002) Caregivers Of patients with Alzheimers disease: a qualitative study from the Indian 10/66 Dementia Research Network. Int J Geriatr Psych 18: 16. 15. Prince M, Livingston G, Katona C (2007) Mental health care for the elderly in low-income countries: a health systems approach. World Psychiatry 6: 513. 16. Shaji KS, Arun Kishore NR, Lal KP, Prince M (2002) Revealing a hidden problem. An evaluation of a community dementia case-finding program from the Indian 10/66 dementia research network. Int J Geriatr Psych 17: 222225. 17. Ramos-Cerqueira AT, Torres AR, Crepaldi AL, Oliveira NI, Scazufca M, et al. (2005) Identification of dementia cases in the community: a Brazilian experience. J Am Geriatr Soc 53: 17381742. 18. National Collaborating Centre for Mental Health (2007) Dementia. A NICESCIE Guideline on supporting people with dementia and their carers in health and social care. National Clinical Practice Guideline number 42. London: National Institute For Health And Clinical Excellence. 19. Folstein MF, Folstein SE, McHugh PR (1975) Mini-mental State: a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12: 189198. 20. Ganguli M, Ratcliff G, Chandra V, Sharma S, Gilby J, et al. (1995) A Hindi version of the MMSE: the development of a cognitive screening instrument for a largely illiterate rural elderly population in India. Int J Geriatr Psych 10: 367377. 21. Xu G, Meyer JS, Huang Y, Du F, Chowdhury M, et al. (2003) Adapting minimental state examination for dementia screening among illiterate or minimally educated elderly Chinese. Int J Geriatr Psychiatry 18: 609616. 22. Castro-Costa E, Fuzikawa C, Uchoa E, Firmo JO, Lima-Costa MF (2008) Norms for the mini-mental state examination: adjustment of the cut-off point in population-based studies (evidences from the Bambui health aging study). Arq Neuropsiquiatr 66: 524528. 23. Ng TP, Niti M, Chiam PC, Kua EH (2007) Ethnic and educational differences in cognitive test performance on mini-mental state examination in Asians. Am J Geriatr Psychiatry 15: 130139. 24. Black SA, Espino DV, Mahurin R, Lichtenstein MJ, Hazuda HP, et al. (1999) The influence of noncognitive factors on the Mini-Mental State Examination in older Mexican-Americans: findings from the Hispanic EPESE. Established 25. Population for the Epidemiologic Study of the Elderly. J Clin Epidemiol 52: 10951102. Brodaty H, Low LF, Gibson L, Burns K (2006) What is the best dementia screening instrument for general practitioners to use? Am J Geriatr Psychiatry 14: 391400. Prince M, Acosta D, Chiu H, Scazufca M, Varghese M (2003) Dementia diagnosis in developing countries: a cross-cultural validation study. The Lancet 361: 909917. Liu SI, Prince M, Chiu MJ, Chen TF, Sun YW, et al. (2005) Validity and reliability of a Taiwan Chinese version of the community screening instrument for dementia. Am J Geriatr Psychiatry 13: 581588. Lecouturier J, Bamford C, Hughes JC, Francis JJ, Foy R, et al. (2008) Appropriate disclosure of a diagnosis of dementia: identifying the key behaviours of best practice. BMC Health Serv Res 8: 95. Ballard CG, Gauthier S, Cummings JL, Brodaty H, Grossberg GT, et al. (2009) Management of agitation and aggression associated with Alzheimer disease. Nat Rev Neurol 5: 245255. Leonard R, Tinetti ME, Allore HG, Drickamer MA (2006) Potentially modifiable resident characteristics that are associated with physical or verbal aggression among nursing home residents with dementia. Arch Intern Med 166: 12951300. Husebo BS, Strand LI, Moe-Nilssen R, Borgehusebo S, Aarsland D, et al. (2008) Who suffers most? Dementia and pain in nursing home patients: a crosssectional study. J Am Med Dir Assoc 9: 427433. Peters CA, Potter JF, Scholer SG (1988) Hearing impairment as a predictor of cognitive decline in dementia. J Am Geriatr Soc 36: 981986. Uhlmann RF, Larson EB, Koepsell TD (1986) Hearing impairment and cognitive decline in senile dementia of the Alzheimers type. J Am Geriatr Soc 34: 207210. Chapman FM, Dickinson J, McKeith I, Ballard C (1999) Association among visual hallucinations, visual acuity, and specific eye pathologies in Alzheimers disease: treatment implications. Am J Psychiatry 156: 19831985. Ballard C, Bannister C, Graham C, Oyebode F, Wilcock G (1995) Associations of psychotic symptoms in dementia sufferers. Br J Psychiatry 167: 537540. Eriksson I, Gustafson Y, Fagerstrom L, Olofsson B (2009) Prevalence and factors associated with urinary tract infections (UTIs) in very old women. Arch Gerontol Geriatr. In press. Natalwala A, Potluri R, Uppal H, Heun R (2008) Reasons for hospital admissions in dementia patients in Birmingham, UK, during 20022007. Dement Geriatr Cogn Disord 26: 499505. Rait G, Fletcher A, Smeeth L, Brayne C, Stirling S, et al. (2005) Prevalence of cognitive impairment: results from the MRC trial of assessment and management of older people in the community. Age Ageing 34: 242248. Fuchs-Lacelle S, Hadjistavropoulos T, Lix L (2008) Pain assessment as intervention: a study of older adults with severe dementia. Clin J Pain 24: 697707. Allen NH, Burns A, Newton V, Hickson F, Ramsden R, et al. (2003) The effects of improving hearing in dementia. Age Ageing 32: 189193. Gil GP, Ramirez Diaz SP, Ribera Casado JM (2003) Dementia and nutrition. Intervention study in institutionalized patients with Alzheimer disease. J Nutr Health Aging 7: 304308. Riviere S, Gillette-Guyonnet S, Voisin T, Reynish E, Andrieu S, et al. (2001) A nutritional education program could prevent weight loss and slow cognitive decline in Alzheimers disease. J Nutr Health Aging 5: 295299. Richard E, Kuiper R, Dijkgraaf MG, van Gool WA (2009) Vascular care in patients with Alzheimers disease with cerebrovascular lesions-a randomized clinical trial. J Am Geriatr Soc 57: 797805. Spector A, Thorgrimsen L, Woods B, Royan L, Davies S, et al. (2003) Efficacy of an evidence-based cognitive stimulation therapy programme for people with dementia: randomised controlled trial. Br J Psychiatry 183: 248254. Bottino CM, Carvalho IA, Alvarez AM, Avila R, Zukauskas PR, et al. (2005) Cognitive rehabilitation combined with drug treatment in Alzheimers disease patients: a pilot study. Clin Rehabil 19: 861869. Clare L, Woods RT, Moniz Cook ED, Orrell M, Spector A (2003) Cognitive rehabilitation and cognitive training for early-stage Alzheimers disease and vascular dementia. Cochrane Database Syst Rev: CD003260. Woods B, Spector A, Jones C, Orrell M, Davies S (2005) Reminiscence therapy for dementia. Cochrane Database Syst Rev: CD001120. Birks J, Harvey RJ (2006) Donepezil for dementia due to Alzheimers disease. Cochrane Database Syst Rev: CD001190.
26.
27.
28.
29.
30.
31.
32. 33.
34.
35. 36.
37.
38.
39.
40. 41.
42.
43.
44.
45.
46.
47. 48.
28

49. Birks J, Grimley EJ, Iakovidou V, Tsolaki M, Holt FE (2009) Rivastigmine for Alzheimers disease. Cochrane Database Syst Rev: CD001191. 50. Loy C, Schneider L (2006) Galantamine for Alzheimers disease and mild cognitive impairment. Cochrane Database Syst Rev: CD001747. 51. Birks J (2006) Cholinesterase inhibitors for Alzheimers disease. Cochrane Database Syst Rev: CD005593. 52. Moraes W, Poyares D, Sukys-Claudino L, Guilleminault C, Tufik S (2008) Donepezil improves obstructive sleep apnea in Alzheimer disease: a doubleblind, placebo-controlled study. Chest 133: 677683. 53. Caramelli P, Chaves ML, Engelhardt E, Machado JC, Schultz RR, et al. (2004) Effects of galantamine on attention and memory in Alzheimers disease measured by computerized neuropsychological tests: results of the Brazilian Multi-Center Galantamine Study (GAL-BRA-01). Arq Neuropsiquiatr 62: 379384. 54. Chu LW, Yik PY, Mok W, Chung CP (2007) A 2-year open-label study of galantamine therapy in Chinese Alzheimers disease patients in Hong Kong. Int J Clin Pract 61: 403410. 55. Burns A, Bernabei R, Bullock R, Cruz Jentoft AJ, Frolich L, et al. (2009) Safety and efficacy of galantamine (Reminyl) in severe Alzheimers disease (the SERAD study): a randomised, placebo-controlled, double-blind trial. Lancet Neurol 8: 3947. 56. McShane R, Areosa SA, Minakaran N (2006) Memantine for dementia. Cochrane Database Syst Rev: CD003154. 57. Kavirajan H, Schneider LS (2007) Efficacy and adverse effects of cholinesterase inhibitors and memantine in vascular dementia: a meta-analysis of randomised controlled trials. Lancet Neurol 6: 782792. 58. Wilcock GK, Ballard CG, Cooper JA, Loft H (2008) Memantine for agitation/ aggression and psychosis in moderately severe to severe Alzheimers disease: a pooled analysis of 3 studies. J Clin Psychiatry 69: 341348. 59. Howard RJ, Juszczak E, Ballard CG, Bentham P, Brown RG, et al. (2007) Donepezil for the treatment of agitation in Alzheimers disease. N Engl J Med 357: 13821392. 60. Lonergan E, Luxenberg J, Colford J (2002) Haloperidol for agitation in dementia. Cochrane Database Syst Rev: CD002852. 61. Ballard C, Waite J (2006) The effectiveness of atypical antipsychotics for the treatment of aggression and psychosis in Alzheimers disease. Cochrane Database Syst Rev: CD003476. 62. Schneider LS, Dagerman K, Insel PS (2006) Efficacy and adverse effects of atypical antipsychotics for dementia: meta-analysis of randomized, placebocontrolled trials. Am J Geriatr Psychiatry 14: 191210. 63. Ballard C, Howard R (2006) Neuroleptic drugs in dementia: benefits and harm. Nat Rev Neurosci 7: 492500. 64. Schneider LS, Dagerman KS, Insel P (2005) Risk of death with atypical antipsychotic drug treatment for dementia: meta-analysis of randomized placebo-controlled trials. JAMA 294: 19341943. 65. Gill SS, Bronskill SE, Normand SL, Anderson GM, Sykora K, et al. (2007) Antipsychotic drug use and mortality in older adults with dementia. Ann Intern Med 146: 775786. 66. Kales HC, Valenstein M, Kim HM, McCarthy JF, Ganoczy D, et al. (2007) Mortality risk in patients with dementia treated with antipsychotics versus other psychiatric medications. Am J Psychiatry 164: 15681576. 67. Kleijer BC, van Marum RJ, Egberts AC, Jansen PA, Knol W, et al. (2008) Risk of cerebrovascular events in elderly users of antipsychotics. J Psychopharmacol. In press. 68. Douglas IJ, Smeeth L (2008) Exposure to antipsychotics and risk of stroke: self controlled case series study. BMJ 337: a1227. 69. Hunkeler EM, Katon W, Tang L, Williams JW, Jr., Kroenke K, et al. (2006) Long term outcomes from the IMPACT randomised trial for depressed elderly patients in primary care. BMJ 332: 259263. 70. Bains J, Birks JS, Dening TR (2002) The efficacy of antidepressants in the treatment of depression in dementia. Cochrane Database Syst Rev: CD003944. 71. Lyketsos CG, DelCampo L, Steinberg M, Miles Q, Steele CD, et al. (2003) Treating depression in Alzheimer disease: efficacy and safety of sertraline therapy, and the benefits of depression reduction: the DIADS. Arch Gen Psychiatry 60: 737746. 72. Martinon-Torres G, Fioravanti M, Grimley EJ (2004) Trazodone for agitation in dementia. Cochrane Database Syst Rev: CD004990. 73. Pollock BG, Mulsant BH, Rosen J, Sweet RA, Mazumdar S, et al. (2002) Comparison of citalopram, perphenazine, and placebo for the acute treatment of psychosis and behavioral disturbances in hospitalized, demented patients. Am J Psychiatry 159: 460465. 74. Finkel SI, Mintzer JE, Dysken M, Krishnan KR, Burt T, et al. (2004) A randomized, placebo-controlled study of the efficacy and safety of sertraline in the treatment of the behavioral manifestations of Alzheimers disease in outpatients treated with donepezil. Int J Geriatr Psychiatry 19: 918. 75. Lonergan ET, Cameron M, Luxenberg J (2004) Valproic acid for agitation in dementia. Cochrane Database Syst Rev: CD003945. 76. Tariot PN, Erb R, Podgorski CA, Cox C, Patel S, et al. (1998) Efficacy and tolerability of carbamazepine for agitation and aggression in dementia. Am J Psychiatry 155: 5461. 77. Olin JT, Fox LS, Pawluczyk S, Taggart NA, Schneider LS (2001) A pilot randomized trial of carbamazepine for behavioral symptoms in treatmentresistant outpatients with Alzheimer disease. Am J Geriatr Psychiatry 9: 400405. 78. Forbes D, Morgan DG, Bangma J, Peacock S, Pelletier N, et al. (2004) Light therapy for managing sleep, behaviour, and mood disturbances in dementia. Cochrane Database Syst Rev: CD003946. 79. Chung JC, Lai CK, Chung PM, French HP (2002) Snoezelen for dementia. Cochrane Database Syst Rev: CD003152. 80. Ballard CG, OBrien JT, Reichelt K, Perry EK (2002) Aromatherapy as a safe and effective treatment for the management of agitation in severe dementia: the results of a double-blind, placebo-controlled trial with Melissa. J Clin Psychiatry 63: 553558. 81. Viggo HN, Jorgensen T, Ortenblad L (2006) Massage and touch for dementia. Cochrane Database Syst Rev: CD004989. 82. Remington R (2002) Calming music and hand massage with agitated elderly. Nurs Res 51: 317323. 83. Sorensen S, Duberstein P, Gill D, Pinquart M (2006) Dementia care: mental health effects, intervention strategies, and clinical implications. Lancet Neurol 5: 961973. 84. Pinquart M, Sorensen S (2006) Helping caregivers of persons with dementia: which interventions work and how large are their effects? Int Psychogeriatr 18: 577595. 85. Smits CH, de LJ, Droes RM, Meiland F, Vernooij-Dassen M, et al. (2007) Effects of combined intervention programmes for people with dementia living at home and their caregivers: a systematic review. Int J Geriatr Psychiatry 22: 11811193. 86. Brodaty H, Green A, Koschera A (2003) Meta-analysis of psychosocial interventions for caregivers of people with dementia. J Am Geriatr Soc 51: 657664. 87. Lee H, Cameron M (2004) Respite care for people with dementia and their carers. Cochrane Database Syst Rev: CD004396. 88. Spijker A, Vernooij-Dassen M, Vasse E, Adang E, Wollersheim H, et al. (2008) Effectiveness of nonpharmacological interventions in delaying the institutionalization of patients with dementia: a meta-analysis. J Am Geriatr Soc 56: 11161128. 89. Dias A, Dewey ME, DSouza J, Dhume R, Motghare DD, et al. (2008) The effectiveness of a home care program for supporting caregivers of persons with dementia in developing countries: a randomised controlled trial from Goa, India. PLoS ONE 3: e2333. doi:10.1371/journal.pone.0002333. 90. Gavrilova SI, Ferri CP, Mikhaylova N, Sokolova O, Banerjee S, et al. (2008) Helping carers to care-The 10/66 dementia research groups randomized control trial of a caregiver intervention in Russia. Int J Geriatr Psychiatry 24: 347354. 91. Graham N, Brodaty H (1997) Alzheimers Disease International. Int J Geriatr Psych 12: 691692. 92. Prince M, Acosta D, Albanese E, Arizaga R, Ferri CP, et al. (2008) Ageing and dementia in low and middle income countries-Using research to engage with public and policy makers. Int Rev Psychiatry 20: 332343. 93. Epping-Jordan JE, Pruitt SD, Bengoa R, Wagner EH (2004) Improving the quality of health care for chronic conditions. Qual Saf Health Care 13: 299305. 94. Sousa RM, Scazufca M, Menezes PR, Crepaldi AL, Prince MJ (2009) Feasibility and reliability of the elderly version of the Camberwell Assessment of Needs (CANE): results from the Sao Paulo Ageing & Health Study. Rev Bras Psiquiatr 31: 3438. 95. Kalaria RN, Maestre GE, Arizaga R, Friedland RP, Galasko D, et al. (2008) Alzheimers disease and vascular dementia in developing countries: prevalence, management, and risk factors. Lancet Neurol 7: 812826. 96. Rafii MS, Aisen PS (2009) Recent developments in Alzheimers disease therapeutics. BMC Med 7: 7. 97. World Health Organization (2004) Towards age-friendly primary health care. Geneva: World Health Organization. Available: http://whqlibdoc.who.int/ publications/2004/9241592184.pdf. Accessed 5 October 2009. 98. United Nations (2002) In: Report of the Second World Assembly on Ageing; Madrid 812 April 2002. A/CONF.197/9. Available: http://www.un.org/ esa/socdev/ageing/secondworld02.html. Accessed 5 October 2009. 99. Shaji KS (2009) Dementia care in developing countries: the road ahead. Indian Journal of Psychiatry 51: S5S7. 100. Callahan CM, Boustani MA, Unverzagt FW, Austrom MG, Damush TM, et al. (2006) Effectiveness of collaborative care for older adults with Alzheimer disease in primary care: a randomized controlled trial. JAMA 295: 21482157. 101. Patel V, Simon G, Chowdhary N, Kaaya S, Araya R (2009) Packages of care for depression in low and middle income countries. PLoS Med 6: e1000159. doi:10.1371/journal.pmed.journal.pmed.1000159. 102. Hall KS, Hendrie HH, Brittain HM, Norton JA, Rodgers DD, et al. (1993) The development of a dementia screening interview in two distinct languages. Int J Meth Psych Res 3: 128. 103. Prince M, Acosta D, Chiu H, Scazufca M, Varghese M (2003) Dementia diagnosis in developing countries: a cross-cultural validation study. Lancet 361: 909917. 104. Reynolds T, Thornicroft G, Abas M, Woods B, Hoe J, et al. (2000) Camberwell Assessment of Need for the Elderly (CANE). Development, validity and reliability. Br J Psychiatry 176: 444452. 105. Thorgrimsen L, Spector A, Wiles A, Orrell M (2003) Aroma therapy for dementia. Cochrane Database Syst Rev: CD003150. 106. Institute of Development and Policy Management/HelpAge International (2003) Non-contributory pensions and poverty prevention. A comparative
29
study of Brazil and South Africa. Final Report, DFID Project R7897, Pensions and Poverty Prevention. Manchester (United Kingdom): Institute of Development and Policy Management and London: HelpAge International. Available: http://www.helpage.org/Resources/Researchreports/Non-contributorypensions. Accessed 5 October 2009.
107. UK Department of Health (2009) Living well with dementia. A National Dementia Strategy. Available: http://www.dh.gov.uk/en/Publicationsandstatistics/ Publications/PublicationsPolicyAndGuidance/DH_094058. Accessed 5 October 2009.
30
Essay
Operationalisation of Mild Cognitive Impairment: A Graphical Approach

Fiona E. Matthews*, Blossom C. M. Stephan, John Bond, Ian McKeith, Carol Brayne on behalf of the Medical Research Council Cognitive Function and Ageing Study
espite intensive use of the term mild cognitive impairment (MCI) to describe an intermediate stage of cognitive decline between normal and pathological brain ageing, no formally agreed process of characterising this condition exists [13]. Various denitions have been proposed in the literature, each with differences in focus (e.g., age-associated change versus pathological decline) and non-uniform diagnostic criteria [418]. The degree of inconsistency is not trivial: current classications dene heterogeneous populations with different patterns of aetiology, cognitive decline, and clinical outcome [19]. As an opportunity for identifying individuals at risk of developing dementia, MCI is an important concept. Yet lack of consensus criteria has lead to debate about the utility of MCI, resulting in calls for abandoning its diagnosis and adopting an alternative nosology [2022]. Consensus conferences are now being held, even though MCI diagnoses are already used in clinical trials for prevention of Alzheimer disease [23]. The aim of this paper is to develop a framework for mapping the different classications of MCI using retrospective information, assessing variations in dening criteria.
Creating a Framework for Mapping MCI

The rst step to coding MCI is to determine the necessary criteria and thresholds for operationalisation of each denition. We compiled a comprehensive list of those classications which represent different aspects and denitions of MCI. The necessary components for each were abstracted and formulated into a diagnostic algorithm.
The Essay section contains opinion pieces on topics of broad interest to a general medical audience.
The main problem encountered was that while some classications have specic criteria for implementation (e.g., amnestic MCI [A-MCI] [13,14] and age-associated memory impairment [5]), others are vague descriptions that require interpretation as to the exact nature of the decit (e.g., age-related cognitive decline [24]). Further complicating the problem is a lack of specication of screening tools and variability in: (1) the domain of impairment (memory versus nonmemory, single- versus multi-domain decits); (2) cut-off scores; (3) acceptable restriction on activities of daily living; and (4) exclusion criteria. Eighteen current denitions of early cognitive impairment were identied in a systematic review of the literature and mapped using a ow diagram as shown in Figure 1. Mapping is completed in two phases: following exclusion of all individuals with dementia, each classication is then operationalised independently. Each classication could be constructed from a subset of 15 different criteria, with memory impairment required as an essential feature in almost all classications. Surprisingly, no two classication systems map on the same path. Each classication is operationalised in the population and has been previously applied to the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS) [19]. Table 1 outlines how each set of criteria was operationalised in CFAS. To facilitate cross classication comparisons, criteria were consistent in terms of level of impairment unless cut-off thresholds were uniquely specied. In reading the ow diagram, the classication arrived at depends on the direction of decision at each criteria (yes, criteria required are fullled as indicated by a green arrow, or no, criteria are not fullled as indicated by a red arrow). If an individual fails to meet the specied outcome for a
1615
given criterion they are excluded from further mapping.
Two Examples: Benign Senescent Forgetfulness and A-MCI

For example, following the ow diagram to arrive at a classication of benign senescent forgetfulness (BSF) takes just three steps: from the START box you move to the Demented box. If the individual does not have dementia (as indicated by the red arrow), you next move to the LongTerm Memory Problem box, and if the
Funding: This work is supported by a Medical Research Council Project Grant (MRC/number G040077) and was approved by the local ethics committee and multi-centre ethics committee approval. The Medical Research Council Cognitive Function and Ageing Study has been funded by the Medical Research Council and United Kingdom Department of Health. The funding bodies have no role in the study design, data collection, analysis, or decision to publish. Competing Interests: The authors have declared that no competing interests exist. Citation: Matthews FE, Stephan BCM, Bond J, McKeith I, Brayne C on behalf of the Medical Research Council Cognitive Function and Ageing Study (2007) Operationalisation of mild cognitive impairment: A graphical approach. PLoS Med 4(10): e304. doi:10.1371/journal.pmed.0040304 Copyright: 2007 Matthews et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: A-MCI, amnestic mild cognitive impairment; BSF, benign senescent forgetfulness; MCI, mild cognitive impairment; MRC CFAS, Medical Research Council Cognitive Function and Ageing Study Fiona E. Matthews is with the Medical Research Council Biostatistics Unit, Institute of Public Health, Cambridge, United Kingdom. Blossom C. M. Stephan and Carol Brayne are with the Department of Public Health and Primary Care, Institute of Public Health, Cambridge, United Kingdom. John Bond is with the Centre for Health Services Research, University of Newcastle upon Tyne, Newcastle upon Tyne, United Kingdom. Ian McKeith is with the Wolfson Research Centre, Newcastle General Hospital, Newcastle upon Tyne, United Kingdom. * To whom correspondence should be addressed. E-mail: ona.matthews@mrc-bsu.cam.ac.uk

OperationalisationofMildCognitiveImpairment:AGraphicalApproach
October 2007 | Volume 4 | Issue 10 | e304
31
doi:10.1371/journal.pmed.0040304.g001
Figure 1. Schematic Overview of the Approach Used to Classify MCI across Different Denitions in the Medical Research Council Cognitive Function and Ageing Study
AACD, age-associated cognitive decline; AAMI, age-associated memory impairment; ACMI, age-consistent memory impairment; ARCD, age-related cognitive decline; CIND, cognitive impairment no dementia; DEM, dementia; LCD, limited cognitive disturbance; MCD (ICD-10), mild cognitive disorder (International Classication of Diseases 10th Revision); MCDi (GDS3), mild cognitive decline (Global Deterioration Scale Stage 3); MCDo(GDS4), moderate cognitive decline (Global Deterioration Scale Stage 4); MD, minimal dementia; M-MCI, multiple mild cognitive impairment; MMSE, Mini Mental State Examination Cut-off Scores; MNCD, mild neurocognitive disorder; N-MCI, non-amnestic mild cognitive impairment; QD, questionable dementia; SD, standard deviation; SMC, self-reported memory complaint.
individual shows a long-term memory problem with intact recent memory (as indicated by the green arrow), you then arrive at a classication of BSF. (See Figure 2 for the BSF-specic path.) In contrast, to arrive at a classication of A-MCI, seven steps are needed: (1) the individual is not demented (as indicated by the red arrow); (2) there is an objective memory impairment (one standard deviation below age-corrected norms, as indicated by the green arrow); (3) the individual or an informant complains of memory loss (as indicated by the green arrow); (4) there is no impairment in activities of daily living (as indicated by the red arrow); (5) there is no impairment of general cognitive functioning (as indicated by the red arrow); (6) the individual has
no underlying medical or neurological condition (as indicated by the red arrow); and (7) there is no impairment in another non-memory domain (as indicated by the red arrow). (See Figure 3 for the A-MCI-specic path.)
Advantages and Disadvantages

The ow diagram details this process for each of the 18 different classication systems, allowing for reproduction across studies. Each concept has been detailed in turn with a criteria-specic graph and the appropriate references on the CFAS Web site (http://www.cfas. ac.uk/mciprogram/). The clinical utility of the concept of mild cognitive impairment depends on the validity of the diagnosis and the ability to predict higher rates of progression
1616
to dementia. The advantage of retrospective denition of these concepts is that they are not adjusted by current knowledge or changing criteria, an unfortunate downside of consensus criteria being that they can be inuenced by changing knowledge over time. The disadvantage of mapping with retrospective information is that it removes clinical experience from the denition, and the information measured from all individuals must have enough scope to encompass the entire original denition. A fundamental problem with a multiple system approach is the failure in consistency of classication. Prevalence estimates for each classication have been previously calculated in CFAS using the pathways
32
Advances in Research and Treatment for Alzheimers disease Table 1. Operationalisation of Components of Classication Systems in the Medical Research Council Cognitive Function and Ageing Study
Criteria
Dementia Long-term memory intact/short-term memory impaired Subjective memory complaint
Operationalisation
AGECAT organic symptom level 3. This corresponds to a diagnosis of dementia as dened by DSM-III-R [25]. CAMCOG remote memory score below the 16th centile score and recent memory score greater than the 16th centile score. Self or informant report. Combined score created from three questions including: (1) Have you had any difculty with your memory? (self report); (2) Have you tended to forget things recently? (self report); and (3) Has he/she had any difculty with his/her memory? (informant report). Responses dichotomised into noncomplainers or complainers (positive response to one or more questions). Unique to each classication. Could include any combination of the following: history of heart attack, chest pain, angina, stroke, Parkinson disease, intermittent claudication, emotional problems, diabetes mellitus, asthma, arthritis, meningitis, head injury, thyroid problems, pernicious anaemia, depression, anxiety, chronic bronchitis, and high blood pressure. Mini Mental Examination (MMSE) score 21. One or more positive responses to the following memory questions, with decline reported as being gradual: (1) Do you have to make more effort to remember things than you used to? What sort of things?; (2) When did you notice this beginning?; (3) Did it come on suddenly?; (4) Would you say there has been a deterioration of memory over a period of more than two years?; and (5) Did these problems with memory begin rapidly or gradually? CAMCOG orientation subtest score below the 16th centile score. Depression and anxiety both dened as AGECAT symptom level 3. Below threshold on one or more of the following subtests of the CAMCOG: orientation, language, attention/ calculation, praxis, abstract thinking, and perception. Modied Townsend Disability Scale, with an additional three items. Impairment dened as requiring help at least several times per week with washing, cooking, and dressing, or as being housebound. Maximum score of 10 derived from the following questions: (1) What is the name of this place? Where is it located? (2 points); (2) What is the date today? (1 point); (3) What is the month? (1 point); (4) What is the year? (1 point); (5) How old are you? (1 point); (6) What is your date of birth (day/month)? (2 points); (7) What is the name of the prime minister? (1 point); (8) Who was the last prime minister? (1 point). Below threshold on one or more of the following memory subtests of the CAMCOG: learning, recent, or remote memory. Positive response to one or more of the following questions: (1) Do you forget the names of your family and friends?; (2) Do you forget where you have placed things?
Exclusions
General cognitive decline Gradual decline (present for at least six months)
Minor errors in orientation Clinical problem (i.e., depression or anxiety) Other (non-memory) cognitive impairment Impaired activities of daily living Mental status questionnaire
Objective memory impairment Forgetfulness
AGECAT, automatic geriatric examination for computer assisted taxonomy; CAMCOG, Cambridge Cognitive Examination; DSM-III-R, Diagnostic and Statistical Manual of Mental Disorders, Third Edition, Revised; MMSE, Mini Mental State Examination Cut-off Scores. doi:10.1371/journal.pmed.0040304.t001
shown in Figure 1, and were found to be highly variable (range 0.1%42%) [19]. Some of this difference results from the fact that not all criteria explain pathological ageing, but rather normal ageing. Although the distinction between those denitions associated with normal age-related change and those with pathological
ageing is not apparent from prevalence estimates alone, it is seen with lower conversion to dementia in those groups dened by non-pathological classications. Furthermore, the same individual could be classied as impaired on one system and normal on another, even within criteria that are supposedly investigating abnormal
change. This makes the interpretation and comparison of results across studies very difcult, where not only the populations but additionally the criteria chosen to estimate MCI are different. At rst glance, a solution to the complexity of the diagram appears simple: reduce all classication
Figure 2. Operationalisation of Benign Senescent Forgetfulness in the Medical Research Council Cognitive Function and Ageing Study
PLoS Medicine | www.plosmedicine.org 1617 October 2007 | Volume 4 | Issue 10 | e304
33
Figure 3. Operationalisation of Amnestic Mild Cognitive Impairment in the Medical Research Council Cognitive Function and Ageing Study
SD, standard deviation
systems to a single concept through the amalgamation of all dening criteria, particularly the measurement of objective memory and the medical (and disability) exclusion criteria. However, this solution assumes that within these criteria there is one that is the best for identication of at-risk individuals. Furthermore, denitions, particularly those of objective cognitive impairment, depend on arbitrary and varying thresholds, frequently with no reference to specic values, methods, or screening measures. In retrospective studies, the mapping of these thresholds will primarily be constrained by study design, though the use of different thresholds can be used to determine the most optimal threshold value to accurately distinguish those individuals at high risk of dementia from those with low dementia risk.
Conclusion
It is time to re-examine the concept of MCI. The diagnostic disparity and the
lack of consistency in case denition calls into question what exactly is being captured in each classication. This is a fundamental weakness of research on MCI, as highlighted by the complicated nature of Figure 1. Using this ow diagram, MCI systems can be mapped in other population datasets to investigate: (1) what are the best boundaries for impairment; (2) which tests are most sensitive for measuring each criteria; (3) which criteria, if any, can adequately predict individuals at risk of developing dementia; and (4) would adopting multiple systems across different populations (specialist clinic versus population based) and age groups be more appropriate? It is hoped that graphical operationalisation of the criteria will aid in diagnostic consistency and assist in the visualisation of the current problem, with the aim of formulating a gold standard denition for both research and clinical practice.
1618
References
1. Gauthier S, Touchon J (2005) Mild cognitive impairment is not a clinical entity and should not be treated. Arch Neurol 62: 11641166. 2. Kave G, Heinik J (2005) Issues to consider when using the new diagnosis of mild cognitive impairment. Isr Med Assoc J 7: 732735. 3. Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, et al. (2004) Mild cognitive impairmentBeyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256: 240246. 4. Levy R (1994) Aging-associated cognitive decline. Working Party of the International Psychogeriatric Association in collaboration with the World Health Organization. Int Psychogeriatr 6: 6368. 5. Crook T, Bartus RT, Ferris SH, Whitehouse PJ, Cohen GD, et al. (1986) Age-associated memory impairment: Proposed diagnostic criteria and measures of clinical change: Report of a National Institute of Mental Health Work Group. Developmental Neuropsychology 2: 261276. 6. Blackford RC, La Rue A (1989) Criteria for diagnosing age-associated memory impairment: Proposed improvements from the eld. Dev Neuropsychol 5: 295306. 7. Kral VA (1962) Senescent forgetfulness: Benign and malignant. Can Med Assoc J 86: 257260. 8. Kral VA (1958) Senescent memory decline and senile amnestic syndrome. Am J Psychiatry 115: 361362.
34

9. Ebly EM, Hogan DB, Parhad IM (1995) Cognitive impairment in the nondemented elderly. Results from the Canadian Study of Health and Aging. Arch Neurol 52: 612619. 10. Gurland BJ, Dean LL, Copeland J, Gurland R, Golden R (1982) Criteria for the diagnosis of dementia in the community elderly. Gerontologist 22: 180186. 11. Reisberg B, Ferris SH, de Leon MJ, Crook T (1982) The Global Deterioration Scale for assessment of primary degenerative dementia. Am J Psychiatry 139: 11361139. 12. World Health Organization (1992) The ICD-10 international statistical classication of diseases and related health problems. 10th Revision. Available: http://www.who.int/classications/ icd/en/. Accessed 27 September 2007. 13. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, et al. (1999) Mild cognitive impairment: Clinical characterization and outcome. Arch Neurol 56: 303308. 14. Petersen RC, Stevens JC, Ganguli M, Tangalos EG, Cummings JL, et al. (2001) Practice parameter: Early detection of dementia: mild cognitive impairment (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 56: 11331142. 15. Roth M, Tym E, Mountjoy CQ, Huppert FA, Hendrie H, et al. (1986) A standarized instrument for the diagnoses of mental disorder in the elderly with special reference to the early detection of dementia. Brit J Psychiat 149: 698709. 16. Hughes CP, Berg L, Danziger WL, Coben LA, Martin RL (1982) A new clinical scale for the staging of dementia. Br J Psychiatry 140: 566572. 17. American Psychiatric Association (1994) Diagnostic and statistical manual of mental disorders, fourth edition [DSM-IV]. Washington (D. C.): American Psychiatric Association. 18. Morris JC, Edland S, Clark C, Galasko D, Koss E, et al. (1993) The consortium to establish a registry for Alzheimers disease (CERAD). Part IV. Rates of cognitive change in the longitudinal assessment of probable Alzheimers disease. Neurology 43: 24572465. 19. Stephan BCM, Matthews FE, McKeith I, Bond J, Brayne C, et al. (2007) Early cognitive change in the general population: How do different denitions work? J Am Geriatr Soc 55: 15341540. 20. DeCarli C (2003) Mild cognitive impairment: Prevalence, prognosis, aetiology, and treatment. Lancet Neurol 2: 1521. 21. Davis HS, Rockwood K (2004) Conceptualization of mild cognitive impairment: A review. Int J Geriatr Psychiatry 19: 313319. 22. Dierckx E, Engelborghs S, De Raedt R, De Deyn PP, Ponjaert-Kristoffersen I (2006) Mild cognitive impairment: Whats in a name? Gerontology 53: 2835. 23. Jelic V, Kivipelto M, Winblad B (2005) Clinical trials in mild cognitive impairment: Lessons for the future. J Neurol Neurosurg Psychiatry 77: 429438. 24. American Psychiatric Association (2000) Diagnostic and statistical manual of mental disorders, fourth edition, text revision [DSMIV-TR]. Washington (D. C.): American Psychiatric Association. 25. Copeland JR, Dewey ME, Grifths-Jones HM (1986) A computerized psychiatric diagnostic system and case nomenclature for elderly subjects: GMS and AGECAT. Psychol Med 16: 8999.
1619
35
Perspectives
Can We Prevent, Delay, or Shorten the Course of Dementia?

Willem A. Van Gool ou already know how it ends when you are born, my mother said in sorrow after my father died. And she is not even medically qualied. Should we extend this common wisdom on the realities of old age towards the development of dementia? In other words, is dementia a curse that is inescapably linked to old age?
A New Study on Dementia before Death

In a new study in PLoS Medicine, Carol Brayne and colleagues show that in a large population sample, the prevalence of dementia and severe cognitive impairment in the year before death rises steeply with age [1]. The researchers found that the prevalence of dementia in the year before death is up to 40%60% in those over 90 years of age. They also found that nearly 80% of those dying over 95 years of age suffered from moderate or more severe cognitive impairment. The researchers examined whether higher levels of education and higher social class (proxies for a healthy lifestyle, e.g., taking exercise and not smoking) protected people from dementia by the time they died. While they did nd such a protective effect, the size of this effect was small: higher social class reduced the risk of dying with dementia by only 2%, and higher levels of education reduced the risk by only 7%.
DOI: 10.1371/journal.pmed.0030430.g001
Figure 1. Curves Depicting Severity of Dementia Depending on Age

The solid line gives an estimate of a present individual trajectory. Broken lines illustrate the potential benet of (A) postponing and shortening the episode with dementia, (B) delaying dementia while reaching a higher age, or (C) preventing dementia.
Policy and Clinical Implications

These gures are important and sobering for policy makers and for those who believe that dementia is preventable. It is difcult to envisage preventive measures that would effectively counteract the profound effects of increasing age on the risk of dementia. As populations age, the
The Perspectives section is for experts to discuss the clinical practice or public health implications of a published article that is freely available online.
burden of dementia will increase, and societies should be prepared for large numbers of elderly patients with dementia. Wishful thinking on prevention of dementia should not defer from the societal investments that are required for providing high quality care for these patients. That is an important message for policy makers and health-care workers alike. While acknowledging both the methodological rigor of Brayne and colleagues analysis and the sobering nature of their data, it is important to note that their dataset does allow for slightly more hopeful interpretations. First, the authors strong emphasis on the large proportion of patients with dementia at the time of death must not let us forget about the importance of the age of onset or the duration of dementia. Their gures do not preclude the possibility that substantial gains might be made by delaying the onset of dementia, shortening the period with dementia, or both (Figure 1). Such scenarios, which would not change the total number of patients suffering from dementia in the period before death, could yield enormous benets in terms of alleviating suffering
1717
at younger ages and reducing the burden for families and other informal caregivers. Second, the observational nature of the present study limits its interpretation in terms of cause and effects, especially with respect to the small effects of potential protective factors. The important role of cardiovascular risk factors for dementia and more specically for Alzheimer disease has been recognized
Funding: The author received no specic funding for this article. Competing Interests: The author is Principal Investigator of pre-DIVA (Prevention of dementia by intensive vascular care, ISRCTN #29711771), a study of primary prevention of dementia. Citation: Van Gool WA (2006) Can We Prevent, Delay, or Shorten the Course of Dementia? PLoS Med 3(10): e430. DOI: 10.1371/journal.pmed.0030430 DOI: 10.1371/journal.pmed.0030430 Copyright: 2006 Willem A. Van Gool. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Willem A. Van Gool is a professor in the Department of Neurology, Academic Medical Centre, Amsterdam, Netherlands. E-mail: w.a.vangool@amc.uva.nl
36

CanWePrevent,Delay,orShortentheCourseofDementia?
Advances in Research and Treatment for Alzheimers disease for almost ten years [2]. This has fuelled the hope that at least some components of dementia may be amenable to preventive measures. Both observational and trial data suggest that treatment of hypertension does indeed protect against dementia, and the same may hold for interventions directed against obesity, smoking, lack of exercise, and glucose intolerance [310]. As acknowledged by Brayne and colleagues, stratication according to educational levels and social classas proxies for the level of cardiovascular preventionis not very strong. Physical tness, body weight, smoking behaviour, blood pressure, and glucose intolerance might have been better indicators, but even then observational studies like theirs, which can be biased by confounding factors, may fail to dene the potential for vascular prevention of dementia. six to eight years may unmask other diseases in the elderly such as ischemic heart disease or cancer, which would have gone undetected if the dementia had not been delayed. The effects of these competing risks on survival may eventually determine if delaying the onset of dementia may ultimately result in prevention of dementia (Figure 1). Individuals who develop dementia at the age of 75 or 85 years may differ in more respects than just the tendency to develop dementia. Therefore, in the absence of randomisation and without an intervention that protects individuals who would have developed dementia at a given age, observational studies can never precisely dene the effects of risks that may compete with the onset of dementia. Direct studies of the benecial and adverse effects of preventive measures are required irrespective of their target, being either -amyloid accumulation in the brain or vascular brain damage. At least one such study is now under way (Prevention of dementia by intensive vascular care [pre-DIVA], ISRCTN #29711771). Because of the long follow up that is required in studies of primary prevention, it will be several years before we will know if the one-way funnel leading towards nursing home placement can be turned into a freeway leading to a life free from dementia at old age.
References
The Need for Randomised Controlled Trials

Only properly designed randomised controlled trials of sufcient power and sufciently long follow up can convincingly dene the window of opportunity for prevention of dementia. Such studies, which will be a major undertaking, can also document the precise effects of risks that may compete with the onset of dementia. Delaying the onset of dementia by, say,
1. Brayne C, Gao L, Dewey M, Matthews FE, Medical Research Council Cognitive Function and Ageing Study Investigators (2006) Dementia before death in ageing societies The promise of prevention and the reality. PLoS Med 3(9): e397. DOI: 10.1371/journal. pmed.0030397 2. Hofman A, Ott A, Breteler MM, Bots ML, Slooter AJ, et al. (1997) Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimers disease in the Rotterdam Study. Lancet 349: 151154. 3. Forette F, Seux ML, Staessen JA, Thijs L, Babarskiene MR, et al. (2002) The prevention of dementia with antihypertensive treatment: New evidence from the Systolic Hypertension in Europe (Syst-Eur) study. Arch Intern Med 162: 20462052. 4. Khachaturian AS, Zandi PP, Lyketsos CG, Hayden KM, Skoog I, et al. (2006) Antihypertensive medication use and incident Alzheimer disease: The Cache County Study. Arch Neurol 63: 686692. 5. Ott A, Andersen K, Dewey ME, Letenneur L, Brayne C, et al. (2004) Effect of smoking on global cognitive function in nondemented elderly. Neurology 62: 920924. 6. Nicolls MR (2004) The clinical and biological relationship between Type II diabetes mellitus and Alzheimers disease. Curr Alzheimer Res 1: 4754. 7. Whitmer RA, Gunderson EP, Barrett-Connor E, Quesenberry CP Jr, Yaffe K (2005) Obesity in middle age and future risk of dementia: A 27 year longitudinal population based study. BMJ 330: 1360. 8. Kivipelto M, Ngandu T, Fratiglioni L, Viitanen M, Kreholt I, et al. (2005) Obesity and vascular risk factors at midlife and the risk of dementia and Alzheimer disease. Arch Neurol 62: 15561560. 9. Weuve J, Kang JH, Manson JE, Breteler MM, Ware JH, et al. (2004) Physical activity, including walking, and cognitive function in older women. JAMA 292: 14541461. 10. Abbott RD, White LR, Ross GW, Masaki KH, Curb JD, et al. (2004) Walking and dementia in physically capable elderly men. JAMA 292: 14471453.
1718
37
Open access, freely available online
Best Practice
Educating the Brain to Avoid Dementia: Can Mental Exercise Prevent Alzheimer Disease?
Margaret Gatz
hysicians now commonly advise older adults to engage in mentally stimulating activity as a way of reducing their risk of dementia. Indeed, the recommendation is often followed by the acknowledgment that evidence of benet is still lacking, but it cant hurt. What could possibly be the problem with older adults spending their time doing crossword puzzles and anagrams, completing gural logic puzzles, or testing their reaction time on a computer? In certain respects, there is no problem. Patients will probably improve at the targeted skills, and may feel goodparticularly if the activity is both challenging and successfully completed.
But can it hurt? Possibly. There are two ways that encouraging mental activity programs might do more harm than good. First, they may offer false hope. Second, individuals who do develop dementia might be blamed for their condition. When heavy smokers get lung cancer, they are sometimes seen as having contributed to their own fates. People with Alzheimer disease might similarly be viewed as having brought it on themselves through failure to exercise their brains.
prevalence of Alzheimer disease in cross-sectional, population-based studies [1] and to a lower incidence of Alzheimer disease in cohorts followed
Citation: Gatz M (2005) Educating the brain to avoid dementia: Can mental exercise prevent Alzheimer disease? PLoS Med 2(1): e7. Copyright: 2005 Margaret Gatz. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Margaret Gatz is in the Department of Psychology, University of Southern California, Los Angeles, California, United States of America, and the Department of Medical Epidemiology and Biostatistics, The Karolinska Institute, Stockholm, Sweden. E-mail: gatz@usc.edu Competing Interests: The author declares that she has no competing interests. DOI: 10.1371/journal.pmed.0020007
What Does the Evidence Show?

Three types of evidence are cited to support the idea that mental exercise can improve ones chances of escaping Alzheimer disease. Epidemiological studies. Having more years of education has been shown to be related to a lower
The Best Practice section summarizes the current evidence on an important health intervention.
38
0038
January 2005 | Volume 2 | Issue 1 | e12 | e7
EducatingtheBraintoAvoidDementia:CanMentalExercisePreventAlzheimerDisease?
Advances in Research and Treatment for Alzheimers disease longitudinally [2]. Typically, the risk of Alzheimer disease is two to four times higher in those who have fewer years of education, as compared to those who have more years of education. Other epidemiological studies, albeit with less consistency, have suggested that those who engage in more leisure activities, especially activities that are mentally stimulating, have a lower prevalence and incidence of Alzheimer disease [3,4]. Additionally, longitudinal studies have found that older adults without dementia who participate in more intellectually challenging daily activities show less decline over time on various tests of cognitive performance [5]. In epidemiological studies, people cannot be randomly assigned to different levels of education, or to different kinds and levels of participation in leisure activities. Consequently, researchers must try to identify confounders and take them into account analytically. However, uncertainties remain. Both education and leisure activities are imperfect measures of mental exercise. For instance, leisure activities represent a combination of inuences. Not only is there mental activation, but there may also be broader health effects, including stress reduction and improved vascular health both of which may contribute to reducing dementia risk [6]. It could also be that a third factor, such as intelligence, leads to greater levels of education (and more engagement in cognitively stimulating activities), and independently, to lower risk of dementia. Research in Scotland, for example, showed that IQ test scores at age 11 were predictive of future dementia risk [7]. Another problem with these epidemiological studies is that reverse causation could be involvedin other words, that incipient dementia could be causing reduced engagement in leisure activities, although some prospective studies have been particularly attentive to controlling for this possibility [8]. Clinical trials are needed to test the hypotheses that emerge from the best epidemiological research. Moreover, because the onset of Alzheimer disease can be hard to pinpoint, and early changes may occur years before the disease is diagnosed, conclusions must be based on large samples, followed over a long period of time. Randomized clinical trials. Many studies support the possibility of enhancing memory and other cognitive performance, or of slowing cognitive decline in older adults without dementia [9]. The most effective programs teach mnemonic strategies, provide practice, and give supportive feedback. Mnemonic strategies include the organization of items into meaningful groups, the use of imagery, and the method of loci (visualizing items to be remembered in a sequence of specic, well-learned locations). Comprehensive programs can also include: DOI: 10.1371/journal.pmed.0020007.g001 encouraging memory aids How effective is mental exercise in holding back dementia? (Illustration: Sapna Khandwala, Public Library of (such as appointment Science) books), teaching relaxation techniques, and providing instruction about memory participants in ACTIVE or in other changes in normal aging. However, memory training programs, it remains improvements are not found in all unknown whether eventual rates of studies. When improvements are Alzheimer disease will be reduced. Neurobiology studies. The third type found, they are often modest, may of evidence suggesting that mental not be maintained over time, and do exercise may help to prevent Alzheimer not generalize beyond the skill being disease comes from neurobiology trained. Often, the subjective gains studies that show greater brain rival the objective ones; for example, complexity in those with higher levels participants do tend to report fewer of mental activity. Many such studies, complaints about their memory. done with animals, show greater These limitations are evident in one neural complexity after having been of the largest randomized controlled exposed to an enriched environment trials of cognitive training with older that provides lots of stimulation, for adults, a large, multisite study named example by including wheels, tunnels, ACTIVE (Advanced Cognitive Training toys, and gnawing sticks [11]. One for Independent and Vital Elderly) human study with magnetic resonance [10]. Participants were assigned spectroscopy showed changes in the to receive training in one of three hippocampus in elderly memory cognitive skills: memory, reasoning, or training participants compared speed of processing. Tests of cognitive to controls [12]. Another report abilities given immediately after found changes on positron emission training showed large improvements on tomography scanning following two the particular cognitive skill on which weeks of a comprehensive memory the individual had been trained, but program that included memory no transfer to the other two cognitive training, special diet, physical exercise, domains. Additionally, for the control and stress reduction [13]. group that received no training, simply taking the test battery at pre-test led Mental Exercise and Cognitive to improvement on the post-test. The Reserve effects of training were maintained over a two-year follow-up. However, The concept of cognitive reserve is the cognitive training program had often used to explain why education no signicant effect on measures and mental stimulation are benecial. of everyday functioning. Finally, for The term cognitive reserve is
0039
January 2005 | Volume 2 | Issue 1 | e7
39
sometimes taken to refer directly to brain size or to synaptic density in the cortex. At other times, cognitive reserve is dened as the ability to compensate for acquired brain pathology. This denition encompasses coping skills as well as recruitment of other brain areas, with cognitive reserve thus accounting for individual differences in severity of cognitive dysfunction when there are pathological neural changes. People with a higher level of education have greater cognitive reserve. In some studies, education or occupation are even used as proxy measures of cognitive reserve, while others are beginning to measure neural substrates that correspond to reserve [14]. Taken together, the evidence is very suggestive that having greater cognitive reserve is related to a reduced risk of Alzheimer disease. But the evidence that mental exercise per se can increase cognitive reserve and stave off dementia is weaker. Epidemiological studies suggest that individual differences in cognitive reserve may actually be lifelong. In addition, people with greater cognitive reserve may choose mentally stimulating leisure activities and jobs, leading to a chickenand-egg dilemma for the interpretation of the relationship between mentally stimulating activities in adulthood and dementia risk. Cognitive training has demonstrable effects on performance, on views of self, and on brain functionbut the results are very specic to the skills that are trained, and it is as yet entirely unknown whether there is any effect on when or whether an individual develops Alzheimer disease. Further, the types of skills taught by practicing mental puzzles may be less helpful in everyday life than more prosaic tricks, such as concentrating, or taking notes, or
putting objects in the same place each time so that they wont be lost.
Conclusion
So far, we have little evidence that mental practice will help prevent the development of dementia. We have better evidence that good brain health is multiply determined, that brain development early in life matters, and that genetic inuences are of great importance in accounting for individual differences in cognitive reserve and in explaining who develops Alzheimer disease and who does not. At least half of the explanation for individual differences in susceptibility to Alzheimer disease is genetic, although the genes involved have not yet been completely discovered [15]. The balance of the explanation lies in environmental inuences and behavioral health practices, alone or in interaction with genetic factors. For older adults, health practices that could inuence the brain include sound nutrition, sufcient sleep, stress management, treatment of mood or anxiety disorders, good vascular health, physical exercise, and avoidance of head trauma. But there is no convincing evidence that memory practice and other cognitively stimulating activities are sufcient to prevent Alzheimer disease; it is not just a case of use it or lose it.
Acknowledgments
The author is grateful to Hans-Olov Adami for helpful discussions, and for grant support from the National Institutes of Health (R01 AG08724, P30 AG17265, P50 AG05142) and from the Alzheimers Association (ZEN-02-3895). References
1. Ott A, Breteler MM, van Harskamp F, Claus JJ, van der Cammen TJ, et al. (1995) Prevalence of Alzheimers disease and vascular dementia:
Association with education. The Rotterdam study. BMJ 310: 970973. 2. Ott A, van Rossum CT, van Harskamp F, van de Mheen H, Hofman A, et al. (1999) Education and the incidence of dementia in a large population-based study: The Rotterdam Study. Neurology 52: 663666. 3. Scarmeas N, Levy G, Tang MX, Manly J, Stern Y (2001) Inuence of leisure activity on the incidence of Alzheimers disease. Neurology 57: 22362242. 4. Wilson RS, Bennett DA, Bienias JL, Aggarwal NT, Mendes de Leon CF, et al. (2002) Cognitive activity and incident AD in a population-based sample of older persons. Neurology 59: 19101914. 5. Hultsch DF, Hertzog C, Small BJ, Dixon RA (1999) Use it or lose it: Engaged lifestyle as a buffer of cognitive decline in aging? Psychol Aging 14: 245263. 6. Fratiglioni L, Paillard-Borg S, Winblad B (2004) An active and socially integrated lifestyle in late life might protect against dementia. Lancet Neurol 2004: 343353. 7. Whalley LJ, Starr JM, Athawes R, Hunter D, Pattie A, et al. (2000) Childhood mental ability and dementia. Neurology 55: 14551459. 8. Verghese J, Lipton RB, Katz MJ, Hall CB, Derby CA, et al. (2003) Leisure activities and the risk of dementia in the elderly. N Engl J Med 348: 25082516. 9. Mohs RC, Ashman TA, Jantzen K, Albert M, Brandt J, et al. (1998) A study of the efcacy of a comprehensive memory enhancement program in healthy elderly persons. Psychiatry Res 77: 183195. 10. Ball K, Berch DB, Helmers KF, Jobe JB, Leveck MD, et al. (2002) Effects of cognitive training interventions with older adults: A randomized controlled trial. JAMA 288: 22712281. 11. Diamond M (1988) Enriching heredity: The impact of the environment on the anatomy of the brain. London: Collier MacMillan. 191 p. 12. Valenzuela MJ, Jones M, Wen W, Rae C, Graham S, et al. (2003) Memory training alters hippocampal neurochemistry in health elderly. Neuroreport 14: 13331337. 13. Small G (2004) The memory prescription. New York: Hyperion. 368 p. 14. Stern Y (2002) What is cognitive reserve? Theory and research application of the reserve concept. J Int Neuropsychol Soc 8: 448460. 15. Ashford JW, Mortimer JA (2002) Non-familial Alzheimers disease is mainly due to genetic factors. J Alzheimers Dis 4: 169177.
40
0040
January 2005 | Volume 2 | Issue 1 | e7
PLoS MEDICINE
Dementia before Death in Ageing Societies The Promise of Prevention and the Reality
Carol Brayne1*, Lu Gao2, Michael Dewey3, Fiona E. Matthews2, Medical Research Council Cognitive Function and Ageing Study Investigators
1 Department of Public Health and Primary Care, Institute of Public Health, Cambridge University, Cambridge, United Kingdom, 2 Medical Research Council Biostatistics Unit, Institute of Public Health, Cambridge, United Kingdom, 3 Institute of Psychiatry, Kings College London, London, United Kingdom Funding: The Medical Research Council Cognitive Function and Ageing Study is supported by the Medical Research Council, United Kingdom. All research has been undertaken independently of the funding bodies. Competing Interests: The authors have declared that no competing interests exist. Academic Editor: Willem Van Gool, University of Amsterdam, Netherlands Citation: Brayne C, Gao L, Dewey M, Matthews FE, Medical Research Council Cognitive Function and Ageing Study Investigators (2006) Dementia before death in ageing societiesThe promise of prevention and the reality. PLoS Med 3(10): e397. DOI: 10.1371/journal. pmed.0030397 Received: April 20, 2006 Accepted: July 26, 2006 Published: October 31, 2006 DOI: 10.1371/journal.pmed.0030397 Copyright: 2006 Brayne et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: CI, confidence interval; GMS, Geriatric Mental State; MMSE, Mini-Mental State Examination; MRC CFAS, Medical Research Council Cognitive Function and Ageing Study * To whom correspondence should be addressed. E-mail: carol.brayne@ medschl.cam.ac.uk
ABSTRACT
Background
Dementia and severe cognitive impairment are very closely linked to ageing. The longer we live the more likely we are to suffer from these conditions. Given population increases in longevity it is important to understand not only risk and protective factors for dementia and severe cognitive impairment at given ages but also whether protection affects cumulative risk. This can be explored by examining the effect on cumulative risk by time of death of factors found consistently to reduce risk at particular ages, such as education and social status.
Methods and Findings

In this analysis we report the prevalence of dementia and severe cognitive impairment in the year before death in a large population sample. In the Medical Research Council Cognitive Function and Ageing Study (a 10-y population-based cohort study of individuals 65 and over in England and Wales), these prevalences have been estimated by age, sex, social class, and education. Differences have been explored using logistic regression. The overall prevalence of dementia at death was 30%. There was a strong increasing trend for dementia with age from 6% for those aged 6569 y at time of death to 58% for those aged 95 y and above at time of death. Higher prevalences were seen for severe cognitive impairment, with similar patterns. People with higher education and social class had significantly reduced dementia and severe cognitive impairment before death, but the absolute difference was small (under 10%).
Conclusions
Reducing risk for dementia at a given age will lead to further extension of life, thus cumulative risk (even in populations at lower risk for given ages) remains high. Ageing of populations is likely to result in an increase in the number of people dying with dementia and severe cognitive impairment even in the presence of preventative programmes. Policy development and research for dementia must address the needs of individuals who will continue to experience these conditions before death.
The Editors Summary of this article follows the references.
1922
41
DementiabeforeDeathinAgeingSocietiesThePromiseofPreventionandtheReality
Dementia before Death in Ageing Societies
Introduction
Changing global population structures require societies to face major issues raised by increasing numbers of the very old. Western societies have experienced sufcient change to be able to anticipate patterns of health and ill health in their ageing populations. These changes are likely to be echoed in a shorter time frame over the next decades in the less wealthy regions of the world because of their more rapid experience of demographic shift. This continuing change in life expectancy has been attributed to improved health and lower morbidity in early life, and to effective primary, secondary, and tertiary prevention in later life. Reductions in incidence have been seen in vascular disease, including cerebrovascular disease [1,2]. Increased life expectancy and improvement in many areas of health have been demonstrated, but sharp increases of morbidity with age are still observed in all populations [3]. Dementia and severe cognitive impairment are amongst the disorders with greatest increase with age in both incidence and prevalence [4,5]. Preventive action at the population level ideally eradicates risk of disease, but in reality much prevention reduces disease at a given age rather than eradicating its occurrence. The overall contribution of primary and secondary prevention to the reduction in cardiovascular disease mortality has been estimated at around half of the effect, with improved care for established disease resulting in the remaining improvement [6]. There is a global effort aimed at improving health in our older populations, and much of this effort hinges on the hope that extension of life expectancy will not be accompanied by increases in morbidity but by compression of morbidity [7]. Many of the factors that have been observed to be associated with reduced risk for chronic disease such as cancer and heart disease are also associated with lower mortality [810]. More recently such factors have also been studied in relation to Alzheimer disease and cerebrovascular disease and have been reported as associated with reduction in risk within the periods of follow-up [11]. These variables are often associated with social class and educational level, whose effects may be attributed to healthier lifestyles and health practices. Social class and education can be seen as proxies for these factors, and higher social class and more education have been shown in many studies to be associated with reduction in dementia outcomes [12,13]. At the same time as this attention to healthy ageing there is increasing attention to the end of life. Many people express a fear of dementia before death as well as a desire to remain independent up to the time of death rather than have prolonged periods of dependency. Quality of life in the period leading up to death is a relatively neglected area in the major emphases of most countries health services, which tend, understandably, to focus on prevention and treatment. There is already clear evidence that individuals who have dementia and cognitive impairment have an excess mortality [14] and a shorter life expectancy [15] than individuals without. However, there is little information about the proportion of the population at any given age who die with dementia or severe cognitive impairment. Individuals at the end of life are likely to be underrepresented in studies of risk with infrequent follow-up as attrition is well known to be associated with cognitive decline and age [16]. Information on
the dementia status of individuals from death certicates is inaccurate and cannot be used as a substitute [17]. More knowledge about the status of individuals in the period before and at death is important to assess the full implications of global population ageing and the potential for prevention of dementia and cognitive decline in the future. If we change risk in the population, can we expect to see different proles in our very old population, including the period close to death? What is the association of dementia and cognitive decline with death (terminal decline), and how much can we estimate that we would avoid if all the preventive actions in place were able to effect a reduction in health inequalities? Longitudinal studies with sufcient follow-up and data on the mortality of all individuals are required to investigate dementia at death. Repeated interviews are necessary to correctly classify dementia at death. Notication of death for the whole cohort irrespective of continuing participation is also required. The analysis presented in this paper investigates the prevalence of dementia and severe cognitive impairment in the period before death in the population-based Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). It also investigates whether those who, on the basis of the literature to date, would be expected to have a relative advantage in risk for dementia at any given age are markedly less likely to die with dementia or severe cognitive impairment than those at greater risk.
Methods Study Design and Population

MRC CFAS is a longitudinal population-based cohort study that involves six different study centres. The six centres were chosen because they represent the main national variation with regards to urbanrural differences, the northsouth and eastwest gradients, and variation in socio-economic levels and in known rates of chronic disease. All centres had existing researchers who were interested in population-based studies of the elderly. Urban sites included Liverpool, Newcastle, Nottingham, and Oxford. Rural sites included Cambridgeshire and Gwynedd, in North Wales. All centres have been used for this analysis. The Liverpool study is described in full elsewhere [18]; a population stratied by 5-y bands over the age of 65 y was interviewed from 1989 onwards. Three full waves of follow-up (each 2 y apart) have been undertaken and used in the following analysis. The full study design of the other ve centres is described in detail elsewhere and is explained briey here [4]. For these ve centres, a total of 13,004 individuals aged 65 y and over were recruited, and interviewed at baseline (19901992) with a schedule including socio-demographic items, general health items, and cognitive items. Participants have been reinterviewed at various times, either the whole cohort or sub-samples stratied by centre, age, and cognitive state. Three types of interviews were used during the study. Baseline examination was conducted via a screening interview. A median of 3 mo after a screening interview, 20% of individuals had a diagnostic interview (assessment interview). The assessment interview consisted of a participant interview and an informant interview. The former consisted of the Geriatric Mental State (GMS), version B3, from which
42

AGECAT [19] could be derived. The latter consisted of the History and Aetiology Schedule [20]. These are standardised psychiatrically based interviews to establish an algorithmic clinic diagnosis described below. Incidence screen and assessments and combined screen and assessment interviews were used during the follow-ups at 2 y, and combined screen and assessment interviews on all participants at 10 y. All interviews contained the Mini-Mental State Examination (MMSE). The Cambridge centre additionally interviewed the complete sample with a combined screen and assessment interview at 6 y. In Liverpool the whole population was interviewed with the full GMS AGECAT algorithm and the clinical version of the MMSE [21] on all these occasions. A study diagnosis of dementia and cognitive impairment can be derived from successful interviews. All participants were agged for mortality information. Death information was obtained from the Ofce of National Statistics up to the end of 2004. Version 8.0 of the ve-centre identical data and version c4 for the Liverpool data have been used.
more deaths, but which underestimates the true amount of dementia/cognitive impairment.
Statistical Methods
Prevalence of dementia at death was calculated assuming binomial proportions with exact condence intervals (CIs). Logistic regression analysis was used to investigate the relationships between confounders and prevalence of dementia at death. All analysis was undertaken using STATA version 8.0 (Stata Corporation, College Station, Texas, United States).
Results Deaths
A total of 4,319 individuals from Liverpool (from an original sample 5,244) and 8,068 individuals from the other centres (from 13,004)altogether 12,387 individualsdied by the time of this analysis. Of these, 12,286 deaths (99%) have information available from their death certicate. A total of 2,566 individuals who had a known dementia status at last interview died within 1 y of this interview, and 5,053 died within 2 y of this interview. A total of 2,555 individuals are included in the analysis of severe cognitive impairment, and 2,573 individuals in the analysis of moderate/severe cognitive impairment.
Definition of Dementia
The study diagnosis of dementia is based on the GMS AGECAT algorithm. This algorithm has been validated against clinicians and DSM-III-R [22]. In addition, for the analysis presented here, organicity level and memory defect rating at the last available interview before death were combined with ICD-10 diagnosis [23] (F00F03, F29, F051, or G30) on the death certicate. Misclassication with organicity due to transient delirium is rare in the population studies [4].
Dementia at Death
Table 1 shows a summary of dementia status at death classied by centre, sex, social class, education level, and age at death. Figure 1 shows a sustained increase in the proportion of individuals receiving a study diagnosis of dementia in the period before death with increased age at death (log). People who died between the ages of 65 and 69 y had a 6% risk of dying with dementia, but people who died above age 95 y had an over 58% risk of dying with dementia. There are wide condence intervals around the estimate for the 31 individuals aged over 100 y at death, compatible with continued increase and stabilisation (42% demented, 95% CI 25%61%). There was a gender difference in risk, with women demonstrating an overall prevalence of dementia before death of 38%, almost double that of men. However, women have a longer life expectancy than men [24], and dementia prevalence increase with age [4]. Adjusting for age showed that women still had a higher absolute proportion of individuals with dementia than men (Table 1; Figure 2); however, there was no evidence that the effect of age differed in men and women (likelihood ratio test for interaction, v2 7.6, p 0.27). Dementia before death was less common in the non-manual-labour social classes than in the manual-labour group at all ages, but not after adjusting for other factors (p 0.09) (Table 1; Figure 2). The absolute difference at death was 2%. A higher level of education was associated with slightly lower prevalence of dementia before death, even after adjusting for the other factors (p 0.02) (Table 1; Figure 2). The individuals who had 10 y or more of full-time education had slightly lower dementia prevalence than individuals with less than 10 y of full-time education. The absolute difference between these educational levels was 6%. There was no evidence of any interaction between sex, education, and social class (likelihood ratio tests all had p . 0.2). An
1924 October 2006 | Volume 3 | Issue 10 | e397
Definition of Cognitive Impairment

MMSE scores were coded within each centre using identical methodology (ignoring backwards WORLD and coding not answered responses to incorrect). Some differences existed between the implementation of the questions (for example, apple, table, and penny in the ve identical centres was any three items in Liverpool); however, examination of the distribution of the scores indicates that they have similar distributional properties at the lower end. The scores were grouped into severe cognitive impairment (MMSE , 18) or not, and moderate/severe cognitive impairment (MMSE , 22). Where the MMSE was missing at the last interview before death, a previous interview was used if cognitive impairment had already been demonstrated. In the analysis presenting cognitive impairment, the classication of dementia status on death certicates was ignored. Individuals with cognitive impairment were classied only from interview responses, and hence some individuals with dementia on death certicates were not classied as cognitively impaired.
Interview Waves
Only study waves where the complete population was interviewed were used in this analysis. Hence, information for Liverpool years 0, 2, and 4 and for the other ve centres years 0, 2, 6 (Cambridge only), and 10 were used. The dementia or cognitive impairment status required is the status at death for an individual, hence only those individuals whose last interview was within 1 y of their death were considered. A sensitivity analysis expanding this to deaths within 2 y has been undertaken, which is based on
43
Table 1. Prevalence of Dementia at Death and Relationship to Potential Risk Factors for All Individuals Who Died within 1 y of Last Interview
Variable
Centre
Level
Cambridge Gwynedd Liverpool Newcastle Nottingham Oxford 6569 7074 7579 8084 8589 9094 95 Male Female Non-manual labour Manual labour Missing ,10 y 10 y Missing
Not Demented (%)

251 205 728 207 196 211 102 231 338 442 368 241 76 966 832 861 801 136 1,272 495 31 1,798 (74) (68) (69) (73) (67) (72) (94) (90) (85) (76) (61) (56) (42) (78) (62) (76) (74) (39) (71) (77) (22) (70)
Demented (%)
90 (26) 97 (32) 323 (31) 77 (27) 98 (33) 83 (28) 7 (6) 27 (10) 62 (16) 140 (24) 239 (39) 187 (44) 106 (58) 267 (22) 501 (38) 271 (24) 284 (26) 213 (61) 510 (29) 145 (23) 113 (78) 768 (30)
Logistic Regression Odds Ratioa (95% CI)
Age
Sex Social Class
1.0 1.8 2.7 4.2 9.6 10.4 19.2 1.0 1.5 0.8 1.0
(0.74.9) (1.17.1) (1.610.5) (3.824.2) (4.126.3) (7.350.5) (1.21.8) (0.71.0)
Education
1.0 0.8 (0.61.0)
Total
a Model for individuals with complete data adjusted for all other factors in the table (excluding centre and individuals with missing data). DOI: 10.1371/journal.pmed.0030397.t001
individual analysis investigated whether the age effect was partially explained by different patterns of dementia seen in different birth cohorts. There was no evidence that suggested there has been a change in the age and sex pattern by birth cohort after adjusting for all other covariates (likelihood ratio test, p . 0.2). Those participants who had an unknown social class and/or educational level (378, 15%) were more likely to have been demented at both baseline and follow-up (62% demented). A sensitivity analysis including these missing values as either the highest or lowest categories did not alter the ndings.
Centenarians
There were 31 centenarians in the study who died within 1 y of interview. Of these, 25 (81%) were female, 13 (42%) had dementia, and 13 (46%) had severe cognitive impairment (with three missing information).
Centre Effects
Although not the primary purpose of this analysis, centre was included as it may indicate unmeasured confounders. For dementia there were no differences between the centres once age and sex were accounted for. There were some small differences between centres for cognition before death, but no consistent pattern emerged.
Cognitive Impairment
The proportion of individuals with severe cognitive impairment before death was very similar to the proportion of individuals with dementia at death, with the same marked increase in cognitive impairment before death with age (Figure 1; Table 2). When individuals with moderate or severe cognitive impairment were included together, the rate was around 10% higher at all ages, with nearly 80% of individuals dying over 95 y of age having this level of cognitive impairment. The relationship between cognitive impairment and demographic risk factors was almost identical between the less and more inclusive levels of impairment and similar to the relationships with dementia at death, though the effects were all slightly stronger. The patterns were also similar if individuals with dementia were removed from the analysis (unpublished data). The absolute difference in severe cognitive impairment at death between social class groups was 5% and for educational levels was 7%. The absolute differences for moderate/severe cognitive impairment were 7% and 10%, respectively.
Sensitivity Analysis
Including all individuals with an interview within 2 y of death did lower the proportions of individuals with dementia before death slightly (27% versus 30%). The proportion with dementia within 1 y of death (30%) was higher than that estimated for individuals seen between 1 and 2 y before death (23%; t-test for difference, p , 0.001), showing the proximity of much dementia to death. The ndings in relation to age, sex, social class, and education were similar. Excluding individuals with limited information (such as screen-only data) did not indicate introduction of bias.
Population Impact
These results were combined with population projections for mortality. The number of deaths per year by age for England and Wales and the US [25,26] was combined with the prevalence of dementia at death. The prevalence rate of dementia was assumed to stay constant with time as there has been no apparent change with each successive 5-y birth cohort studied (year of birth ranges from 1888 to 1928). The
44

Figure 1. Dementia or Cognitive Impairment by Age at Death with 95% CIs DOI: 10.1371/journal.pmed.0030397.g001
results indicate that currently there are 114,000 individuals (95% CI 97,000134,000) in England and Wales and 487,000 individuals (95% CI 362,000503,000) in US who die with dementia each year. In 20 years time, these numbers will have increased to 138,000 (95% CI 118,000160,000) in England and Wales and 528,000 (95% CI 443,000627,000) in the US.
Discussion Summary of Findings

This 10-y longitudinal population study of people aged 65 y and over at baseline examines dementia and severe cognitive impairment status in the period before death and shows that the prevalence of dementia and severe cognitive impairment in the period before death rises steeply with age. Thus, by the time an individual aged 90 y dies, the risk of being demented or severely cognitively impaired is around 60%. Investigating whether higher education and social classproxies for healthy exposures and lifestylesprotected individuals from dementia by the time they died showed that individuals with higher education and social class were at signicantly reduced risk of dying with dementia or cognitive impairment. However, this reduction was limited: the reduction in dementia had an absolute value of only 2% and 7%, respectively, for higher social class and more highly educated groups; the reduction in cognitive impairment was 7% and 10%, respectively; and both high education and social class together conferred a reduction of 7% for dementia and 10% for cognitive impairment. Thus, the inequalities in healthy lifestyles and social advantage observed at given ages appear to be attenuated by the longer life expectancy of those individuals.
Critique of Findings
There are features of the study that need to be taken into account in interpreting the ndings. The response rate at
baseline and at each of the follow-up interviews was around 80% (see http://www-cfas.medschl.cam.ac.uk for details of audit trail). Some bias could be introduced into the analysis through this drop out due to death, refusal, and moving away. We know from earlier analysis of the dataset that individuals with cognitive impairment are at greater risk of death, so there is some potential to underestimate the prevalence of dementia at death given that we do not have cognitive status close to death in all respondents. We have taken this into account through the analysis of only those individuals who died within a relatively short period of being interviewed. If the analysis is conducted on the total sample for whom the last interview was more distant from death, a lower prevalence of dementia and cognitive impairment is found, as expected. However, individuals who dropped out from the study in the year preceding death were no more likely to be demented than those who undertook an interview in this time, indicating that the individuals analysed here are unlikely to be biased. The measure for dementia in this study was GMS diagnostic algorithm organicity status, and for cognitive impairment was MMSE. Both of these will suffer from some misclassication bias. We have found severe MMSE and dementia states to be relatively stable, with only small proportions moving out of these categories in the extensive longitudinal data available. Limiting the analysis to subtypes of dementia such as Alzheimer disease or vascular dementia might indicate preventive potential for specic subtypes by death but would not reveal the true population picture, nor does cognitive impairment and dementia in the oldest old tend to t into pure diagnostic categories [27]. Moreover, the concern many older people express is not about specic forms of dementia but that they will lose their independence and become cognitively frail. We have not addressed the mildest stages of impairment in this analysis.
45
46
Figure 2. Rate of Dementia at Death by Sex, Education, and Social Class DOI: 10.1371/journal.pmed.0030397.g002 PLoS Medicine | www.plosmedicine.org 1927 October 2006 | Volume 3 | Issue 10 | e397

Table 2. Prevalence of Moderate and Severe Cognitive Impairment at Death and the Relationship with Potential Risk Factors
Variable Level Severe Cognitive Impairment Total Sample Size
Centre Cambridge Gwynedd Liverpool Newcastle Nottingham Oxford 6569 7074 7579 8084 8589 9094 95 Male Female Non-manual labour Manual labour Missing ,10 y 10 y Missing 323 329 997 311 305 293 109 268 407 573 615 416 170 1,244 1,314 1,175 1,119 264 1,807 648 103 2,558
Moderate/Severe Cognitive Impairment OR (95% CI)

a
Number with Impairment (%)
Total Sample Size

325 330 1,001 314 308 299 109 269 408 573 620 425 173 1,249 1,328 1,127 1,128 322 1,821 652 104 2,577
Number with Impairment (%)
ORa (95% CI)
Age
Sex Social Class
Education
Total
(26) (35) (28) (25) (34) (31) 5 (5) 33 (12) 53 (13) 129 (23) 230 (37) 191 (46) 104 (61) 251 (20) 494 (38) 298 (25) 298 (27) 149 (56) 512 (28) 137 (21) 96 (93) 745 (29)
84 115 276 77 103 90
1.0 3.7 3.7 6.5 15.0 20.2 38.2 1.0 1.7 0.7 1.0
(1.112.5) (1.112.4) (2.020.9) (4.748.0) (6.365.4) (11.4127.8) (1.42.1) (0.60.9)
1.0 0.7 (0.60.9)
137 163 414 128 150 135 14 58 96 204 334 285 136 419 708 488 405 234 804 224 99 1,127
(42) (49) (41) (41) (49) (45) (13) (22) (24) (36) (54) (67) (79) (34) (53) (43) (36) (73) (44) (34) (95) (44)
1.0 2.1 2.1 3.8 8.0 13.6 22.8 1.0 1.7 0.7 1.0
(1.04.2) (1.14.2) (2.07.4) (4.215.4) (6.926.6) (10.848.2) (1.42.0) (0.50.8)
1.0 0.7 (0.50.8)
a Model for individuals with complete data adjusted for all other factors in the table (excluding centre and individuals with missing data). OR, odds ratio. DOI: 10.1371/journal.pmed.0030397.t002
Examining education and social class as proxies for healthy lifestyles could be seen as a crude approach, but these have been shown in their own right, in some studies, to be independently associated with lower rates of dementia. Certainly it would be desirable to expand this type of analysis to population studies with better lifestyle and health indicatorspeak physical tness, no smoking, moderate alcohol, good nutrition, and no hypertension, diabetes, or heart disease. There are very few studies with which to compare these ndings. One US study examined patterns of change in functional disability in the period before death [28]. Its goal was more to examine the predictive value of such patterns. It is already well established that cognitive impairment and dementia are associated with increased risk of death. We set out to examine how common dementia is by the time of death and whether those groups who are held to be at reduced risk of dementia and chronic disease during life appear to escape the risk before death. We have found that this is not the case, although there is a small diminution in risk before death, consistent with the absolute risk reductions reported for given ages in cohort studies. These ndings conrm the nding of an earlier study by Bickel, who conducted a retrospective study of deaths by asking relatives about the status of the deceased before death and reported a prevalence of dementia-related conditions at death even higher than our estimates [29]. Our data also agree with ndings on the health state of the very old, including centenarians. MRC CFAS reports odds ratios of incidence of dementia of over 25 for the 85-y-and-over age group compared to the 65- to 69-y age
group [30]. In a study of 34 centenarians by Silver and colleagues [31], most (64%) were demented. Systematic reviews have shown that prevalence of dementia at this age is between 50% and 64% [32], with only 15%25% estimated to have functionally intact cognition [33,34].
Implications and Conclusions

These data have serious implications for societies that are ageing. All public health services internationally will and should continue to promote healthy ageing through healthy lifestyles and optimal health care. It is important to continue to research treatments and preventive approaches for specic dementias aimed at lessening the devastation that dementia and cognitive impairment can cause. Substantial claims have been made for the prevention of dementia, and these need to be backed up by robust evidence of risk reduction through preventive strategiesnot yet available as the evidence is based on observational studies. However, our ndings suggest that preventive efforts are unlikely to be able to counteract the profound effects of age and proximity to death altogether. We will continue to examine early risk and track cognitive change, but the analysis presented here is about quality of life at the end of life. It may be that, although there will be a preventable component to dementia giving us a small and important absolute reduction in expectation of dementia at given ages, there is also a component that is not amenable to such types of prevention [35]. Researchers may be doing those who are ageing now and themselves a disservice in the future if they assume, and project to the public, that dementia and cognitive impairment can be prevented altogether during increasingly long lives. Given
47
that the population of all deaths accounted for by people aged 85 y and over has increased from 8% in 1960 to 21% in 2003 in men and 16% to 41% in women in England and Wales, this is a matter of considerable importance [36]. Our data show that in the UK (and potentially in the US) the burden of dementia and cognitive impairment at time of death is substantial and is likely to become an increasing problem as the population ages. It is essential and urgent that societies invest in planning for (and researching) quality at the end of life for those with dementia and severe cognitive impairment [3739], with sensible estimation of the likely increase in the numbers affected in the decades to come.
Acknowledgments
The MRC CFAS is supported by the Medical Research Council, UK. Thanks are due to the MRC CFAS team, our general practices, and most particularly to the respondents and their families. All MRC CFAS interviewing has been covered under local and multi-centre ethical approval. Author contributions. CB conceived this analysis and led the study. LG and FEM designed and conducted the analysis. MD contributed at all stages. All authors discussed the results and commented on the paper. All MRC CFAS (corporate author) principle investigators and team were involved in the study and had an opportunity to comment on the paper.
References 1. Centers for Disease Control and Prevention (1999) Decline in deaths from heart disease and strokeUnited States, 19001999. MMWR Morb Mortal Wkly Rep 48: 649656. 2. Thorvaldsen P, Davidsen M, Bronnum-Hansen H, Schroll M (1999) Stable stroke occurrence despite incidence reduction in an aging population: Stroke trends in the Danish Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) population. Stroke 30: 25292534. 3. Brayne C, Matthews FE, McGee MA, Jagger C (2001) Health and ill-health in the older population in England and Wales. The Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Age Ageing 30: 5362. 4. (1998) Cognitive function and dementia in six areas of England and Wales: The distribution of MMSE and prevalence of GMS organicity level in the MRC CFA Study. The Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Psychol Med 28: 319335. 5. Matthews F, Brayne C, Medical Research Council Cognitive Function and Ageing Study Investigators (2005) The incidence of dementia in England and Wales: Findings from the ve identical sites of the MRC CFA Study. PLoS Med 2: e193. DOI: 10.1371/journal.pmed.0020193 6. Kuulasmaa K, Tunstall-Pedoe H, Dobson A, Fortmann S, Sans S, et al. (2000) Estimation of contribution of changes in classic risk factors to trends in coronary-event rates across the WHO MONICA Project populations. Lancet 355: 675687. 7. World Health Organization (2002) Active ageing: A policy framework. Geneva: World Health Organization. Available: http://whqlibdoc.who.int/ hq/2002/WHO_NMH_NPH_02.8.pdf. Accessed 11 September 2006. 8. Yuyun MF, Khaw KT, Luben R, Welch A, Bingham S, et al. (2004) Microalbuminuria, cardiovascular risk factors and cardiovascular morbidity in a British population: The EPIC-Norfolk population-based study. Eur J Cardiovasc Prev Rehabil 11: 207213. 9. Khaw KT, Wareham N, Luben R, Bingham S, Oakes S, et al. (2001) Glycated haemoglobin, diabetes, and mortality in men in Norfolk cohort of European Prospective Investigation of Cancer and Nutrition (EPICNorfolk). BMJ 322: 1518. 10. Khaw KT, Bingham S, Welch A, Luben R, Wareham N, et al. (2001) Relation between plasma ascorbic acid and mortality in men and women in EPICNorfolk prospective study: A prospective population study. European Prospective Investigation into Cancer and Nutrition. Lancet 357: 657663. 11. Whitmer RA, Gunderson EP, Barrett-Connor E, Quesenberry CP Jr, Yaffe K (2005) Obesity in middle age and future risk of dementia: A 27 year longitudinal population based study. BMJ 330: 1360. 12. Ravaglia G, Forti P, Maioli F, Martelli M, Servadei L, et al. (2005) Incidence and etiology of dementia in a large elderly Italian population. Neurology 64: 15251530.
13. Karp A, Kareholt I, Qiu C, Bellander T, Winblad B, et al. (2004) Relation of education and occupation-based socioeconomic status to incident Alzheimers disease. Am J Epidemiol 159: 175183. 14. Dewey ME, Saz P (2001) Dementia, cognitive impairment and mortality in persons aged 65 and over living in the community: A systematic review of the literature. Int J Geriatr Psychiatry 16: 751761. 15. Hill GB, Forbes WF, Lindsay J (1997) Life expectancy and dementia in Canada: The Canadian study of health and aging. Chronic Dis Can 18: 166 167. 16. Chateld MD, Brayne CE, Matthews FE (2005) A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies. J Clin Epidemiol 58: 1319. 17. Martyn CN, Pippard EC (1988) Usefulness of mortality data in determining the geography and time trends of dementia. J Epidemiol Community Health 42: 134137. 18. Saunders PA, Copeland JR, Dewey ME, Gilmore C, Larkin BA, et al. (1993) The prevalence of dementia, depression and neurosis in later life: The Liverpool MRC-ALPHA Study. Int J Epidemiol 22: 838847. 19. Copeland JR, Dewey ME, Grifths-Jones HM (1986) A computerized psychiatric diagnostic system and case nomenclature for elderly subjects: GMS and AGECAT. Psychol Med 16: 8999. 20. Copeland JRM, Dewey ME (1991) Neuropsychological diagnosis (GMS-HASAGECAT package). Int Psychogeriatr 3: 4349. 21. Folstein M, Folstein S, McHugh P (1975) Mini-mental state: A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12: 189198. 22. American Psychiatric Association (1987) Diagnostic and statistical manual of mental disorders: DSM-III-R, 3rd edition revised. Washington (D. C.): American Psychiatric Association. 567 p. 23. World Health Organization (1993) ICD-10 classication of mental and behavioural disorders diagnostic criteria for research. Geneva: World Health Organization. Available: http://www.who.int/substance_abuse/ terminology/ICD10ResearchDiagnosis.pdf. Accessed 11 September 2006. 24. Neale R, Brayne C, Johnson AL (2001) Cognition and survival: An exploration in a large multicentre study of the population aged 65 years and over. Int J Epidemiol 30: 13831388. 25. National Statistics (2005) Census 2001: England and Wales. Available: http:// www.statistics.gov.uk/census2001/proles/727.asp. Accessed 7 June 2006. 26. US Census Bureau (2006) U. S. interim projections by age, sex, race, and Hispanic origin. Washington (D. C.): US Census Bureau. Available: http:// www.census.gov/ipc/www/usinterimproj. Accessed 3 October 2006. 27. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (2001) Pathological correlates of late-onset dementia in a multicentre, community-based population in England and Wales. Neuropathology Group of the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). Lancet 357: 169175. 28. Lunney JR, Lynn J, Foley DJ, Lipson S, Guralnik JM (2003) Patterns of functional decline at the end of life. JAMA 289: 23872392. 29. Bickel H (1996) [Need for nursing care by the elderly: Results of a population based retrospective longitudinal study]. Gesundheitswesen 58: 5662. 30. Yip AG, Brayne C, Matthews FE, MRC Cognitive Function and Ageing Study (2006) Risk factors for incident dementia in England and Wales: The Medical Research Council Cognitive Function and Ageing Study. A population-based nested case-control study. Age Ageing 35: 154160. 31. Silver MH, Jilinskaia E, Perls TT (2001) Cognitive functional status of ageconrmed centenarians in a population-based study. J Gerontol B Psychol Sci Soc Sci 56: 134140. 32. Hagberg B, Bauer AB, Poon LW, Homma A (2001) Cognitive functioning in centenarians: A coordinated analysis of results from three countries. J Gerontol B Psychol Sci Soc Sci 56: 141151. 33. Perls T (2004) Dementia-free centenarians. Exp Gerontol 39: 15871593. 34. Dewey ME, Copeland JR (2001) Dementia in centenarians. Int J Geriatr Psychiatry 16: 538539. 35. Coggon DI, Martyn CN (2005) Time and chance: The stochastic nature of disease causation. Lancet 365: 14341437. 36. Tomassini C (2005) The demographic characteristics of the oldest old in the United Kingdom. Popul Trends Summer: 1522. 37. Sampson EL, Ritchie CW, Lai R, Raven PW, Blanchard MR (2005) A systematic review of the scientic evidence for the efcacy of a palliative care approach in advanced dementia. Int Psychogeriatr 17: 3140. 38. Murtagh FE, Preston M, Higginson I (2004) Patterns of dying: Palliative care for non-malignant disease. Clin Med 4: 3944. 39. Schulz R, Mendelsohn AB, Haley WE, Mahoney D, Allen RS, et al. (2003) End-of-life care and the effects of bereavement on family caregivers of persons with dementia. N Engl J Med 349: 19361942.
48

Editors Summary
Background. Severe cognitive impairment and its advanced form, dementia, are among the most difficult problems associated with aging in industrialized countries. Age-associated decline in mental functioning is also expected to become more common in developing countries as improvement of conditions that affect health leads to longer life expectancies. Although the risk of cognitive impairment is known to increase with age, the number of people who suffer from loss of mental abilities in the last years of their lives has not been well studied, as such persons are usually reported to have died from other causes. Further, because the very elderly are seldom included in prevention studies, it is not known whether factors found to reduce the risk of developing dementia by a given age will provide protection until the end of life. Why Was This Study Done? This study was designed to follow a representative population of aged people over several years to estimate the risk of developing cognitive impairment or dementia near the end of life and to determine whether factors such as education and social class, which may be protective earlier in life, can ultimately prevent decline in mental functioning. What Did the Researchers Do and Find? Using standardized assessments of cognitive status, the researchers interviewed people age 65 and over at six sites representing rural and urban areas in the United Kingdom. Interviews were conducted at regular intervals over ten years. Of approximately 12,000 study participants who had died by the time of this report, just over 2,500 had an assessment for dementia within one year before dying. Of this group, those who died between ages 65 and 69 had a 6% chance of dying with dementia, and those who died above age 95 had a 58% chance of dying with dementia. When moderate and severe cognitive impairment were considered together, the rate in people above age 95 reached almost 80%. Women were more likely to develop dementia than men, even after taking into account the fact that women tend to live longer than men. A higher level of education was associated with only a slightly lower risk of dementia before death. What Do These Findings Mean? According to these results, as the number of aged persons increases (with improved health care, preventive medicine, and healthier lifestyles), the chances of developing dementia in the last years of life will continue to increase. Factors believed to protect against dementia at earlier times may be of little effect at the end of life. Planning for aging societies must therefore include not only research into treatments and preventive efforts to reduce the impact of dementia at the end of life, but also realistic allocation of resources to support individuals and their caregivers who must deal with the difficulties of cognitive decline. Additional Information. Please access these Web sites via the online version of this summary at http://dx.doi.org/0030397. Web site of the Medical Research Council Cognitive Function and Ageing Study Web site of the Alzheimers Association Wikipedia entry on dementia (note: Wikipedia is a free Internet encyclopedia that anyone can edit)
49
Policy Forum
Cholinesterase Inhibitors: Drugs Looking for a Disease?

Marina Maggini*, Nicola Vanacore, Roberto Raschetti in patients with Alzheimer disease but also in patients with vascular dementia, dementia with Lewy bodies, dementia associated with Parkinson disease, and mild cognitive impairment (MCI). Even when the evidence on the efcacy of these drugs is lacking, or inconclusive, the results are often presented in such a way as to create a false perception of efcacy. For example, about 23 different scales or instruments (on average six per trial) were used, in the trials considered here, as primary or secondary outcome measures. Most of them were not validated for the disease for which the drugs were tested and are not currently used in clinical practice, undermining the translation of these research ndings into clinical practice. Moreover, the treatment effect in the trials is usually expressed through the average change from baseline in test scores, without discussing the clinical importance of the usually small effect size observed. successes by 245% in patients with mild to moderate Alzheimer disease [3]. Donepezil, galantamine, and rivastigmine went on to be approved in many countries for the treatment of Alzheimer disease, even though it was clear that the efcacy, in the short term, was modest, symptomatic, and evident only in a subgroup of patients [48]. In a meta-analysis of randomized, double-blind placebo-controlled trials of cholinesterase inhibitors, Lanctt and colleagues found that the pooled mean proportion of responders to drug treatment in excess of that for placebo treatment was only 10% (95% condence interval, 4%17%) [9]. In this study, response to therapy was dened (according to a denition rst proposed by the US Food and Drug Administration) as an improvement of four or more points on the Alzheimer Disease Assessment Scalecognitive portion (ADAS-cog) [10].
Funding: The authors received no specic funding for this article. Competing Interests: The authors research is often funded by the Italian National Health Service and by the Italian Medicines Agency, which is the authority responsible for drug reimbursement procedures (including costs connected to the prescribing of acetylcholinesterase inhibitors). The authors themselves have no role in decisions concerning such reimbursement. Citation: Maggini M, Vanacore N, Raschetti R (2006) Cholinesterase inhibitors: Drugs looking for a disease? PLoS Med 3(4): e140. DOI: 10.1371/journal.pmed.0030140
andomized controlled trials (RCTs) are generally considered to be a robust form of evidence, free from bias, and the trial results are often used as a powerful tool to promote new drugs [1,2]. However, because the inclusion criteria for many RCTs are often very restrictive (for example, trials generally exclude patients with serious concomitant illnesses) and because patients in trials tend to receive better care than those in standard-care settings, clinicians should be careful about generalizing RCT results to their own patients. Unfortunately, many drug treatments are widely used in clinical practice, sometimes beyond the approved indications, even when doubts remain about whether the results of RCTs of these drugs should be generalized. In this article, we discuss the use of cholinesterase inhibitors in patients with a variety of types of dementia and cognitive impairment, looking critically at the clinical trial evidence on these drugs. If the results of these trials are not carefully evaluated, together with evaluating the methodological quality of the studies, this could lead to inappropriate prescribing of cholinesterase inhibitors. Drug companies have invested heavily in developing treatments for Alzheimer disease, and then were actively involved in expanding the market to other forms of dementia. In the last decade, donepezil, galantamine, and rivastigmine have been tested not only
Alzheimer Disease: Waiting for New Treatments

The cholinesterase inhibitor donepezil was licensed in the US in December 1996, before the full results of clinical trials were available in medical journals [3]. The drug was launched with claims that it had produced highly signicant improvements in cognitive and clinical global assessments in randomized trials lasting 30 weeks and had increased the proportion of treatment
Search Strategy
For this article, we searched the MEDLINE database from 1996 to 2005 using the terms donepezil, galantamine, and rivastigmine to nd randomized controlled clinical trials, systematic reviews, and meta-analyses. Our article is not itself a systematic review, but we discuss all the major RCTs, systematic reviews, and meta-analyses of these drugs as treatments for Alzheimer disease, and we discuss the major RCTs of these drugs for other forms of dementia.
0456
Copyright: 2006 Maggini et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: ADAS-cog, Alzheimer Disease Assessment Scalecognitive portion; MCI, mild cognitive impairment; MMSE, Mini Mental State Examination; NICE, National Institute for Health and Clinical Excellence; RCT, randomized controlled trial Marina Maggini, Nicola Vanacore, and Roberto Raschetti are at the National Center for Epidemiology, National Institute of Health, Rome, Italy. * To whom correspondence should be addressed. E-mail: mmaggini@iss.it
The Policy Forum allows health policy makers around the world to discuss challenges and opportunities for improving health care in their societies.
50

CholinesteraseInhibitors:DrugsLookingforaDisease?
April 2006 | Volume 3 | Issue 4 | e140
Advances in Research and Treatment for Alzheimers disease The most recent systematic review of RCTs, by Hanna Kaduszkiewicz and colleagues, analyzed the scientic evidence for the clinical use of cholinesterase inhibitors in Alzheimer disease, together with the methodological quality of the trials [11]. The authors concluded that the benets are minimal, the methodological quality of the available trials is poor, and the scientic basis for recommendations of these drugs for Alzheimer disease is questionable [11]. A similar conclusion was reported in the preliminary draft of recommendations on the use of cholinesterase inhibitors that is being developed by the United Kingdoms National Institute for Health and Clinical Excellence (NICE), an independent organization responsible for providing national guidance on treating and preventing illness [12,13]. In its preliminary draft appraisal document, the organization stated that the RCT evidence on outcomes of importance to patients and carers, such as quality of life and time to institutionalisation, was limited and largely inconclusive. Moreover, the NICE committee reported that the quality of the reviewed trials was mixed, and that the assessment group suspected selection bias, measurement bias and attrition bias. The preliminary recommendations of the appraisal committee were that donepezil, rivastigmine and galantamine are not recommended for use in the treatment of mild to moderate Alzheimers disease, and that further research is required to identify subgroups of people for whom cholinesterase inhibitors may be effective. The committee recently updated its guidance, as shown in the Sidebar.
NICE Recommendations on Cholinesterase Inhibitors

Revised draft guidance on the use of drugs to treat Alzheimer disease has recently been published (23 January 2006) on the NICE Web site (http://www. nice.org.uk/page.aspx?o=288826). The preceding draft guidance from NICE (http://www.nice.org.uk/page. aspx?o=245908), published 1 March 2005, concluded that there was not enough evidence to support the use of these drugs for all patients. However, responses received from stakeholders during consultation on this rst draft suggested that the drugs may be more effective for certain groups of people. NICE, therefore, asked the pharmaceutical companies involved in the appraisal to look for evidence to support this, from the data in their clinical trials. In conclusion, the Committee considered not just the initial evidence and submissions, but also the comments raised in consultation on the rst Appraisal Consultation Document (notably the improved infrastructure around dementia care) and the evidence factors [14]. The authors concluded that the drug is effective in patients with or without vascular risk factors, and that those with vascular risk factors experience greater clinical benet (cognition, activities of daily living, and disease severity). However, the withdrawal rate was higher for patients given the drug than for patients given placebo, and there was no intention-totreat analysis. The effect of galantamine was examined in a six-month RCT in a mixed population of patients diagnosed as having probable vascular dementia, Alzheimer disease with cerebrovascular disease, or an intermediate diagnosis [15]. Unfortunately, the study was not powered to detect treatment differences in the three subgroups; moreover, as in the study on rivastigmine [14], the primary statistical assessment of efcacy was not based on an intention-to-treat analysis, but only on an observed case analysis. Two trials have been conducted to evaluate the efcacy and tolerability of donepezil in patients diagnosed with vascular dementia; these trials showed modest and inconsistent effects [16,17]. The study design was similar to the design used in trials of cholinesterase inhibitors for Alzheimer disease: the
0457
that was submitted during consultation and the additional analyses undertaken. The Committee concluded that taking all these factors into account, the resulting estimates of cost effectiveness could be considered sufciently acceptable to allow the prescribing of AChE inhibitors, donepezil, galantamine, and rivastigmine, for people with Alzheimers disease of moderate severity only (that is, those with an MMSE score between ten and 20). As in the earlier draft, the committee noted, however, that the evidence available on the long-term effectiveness of the AChE inhibitors on outcomes of importance to people with Alzheimers disease and their carers, such as quality of life and delayed time to nursing home placement, was limited and largely inconclusive. As for memantine, it continued to be not recommended as a treatment option for people with Alzheimers disease except as part of properly constructed clinical studies. vascular dementia trials used similar drug doses and similarly lasted only six months. As with trials of cholinesterase inhibitors for Alzheimer disease, a sixmonth trial period is unjustied for a pathology that develops over decades. Moreover, the assessment scales used in the vascular dementia trials are intended for assessing Alzheimer disease, and are not validated for the evaluation of vascular dementia. The investigators did not nd improvement for all primary and secondary efcacy parameters, and a reverse dose effect was shown: that is, improvement in global function was observed in a greater proportion of patients treated with donepezil than those treated with placebo in the 5-mg group but not in the 10-mg group [16]. The study population was, as reported by the authors, not typical of all patients with vascular dementia (in fact, only patients who were stable with respect to comorbid conditions, hypertension, diabetes, and heart disease were included in these clinical trials) [16]. Even in this highly selected population, an excess of stroke (fatal and nonfatal) was observed among treated patients. The potential implications for clinical practice still remain to be claried. Nevertheless, the drug was presented in
Patients with Alzheimer Disease and Vascular Risk Factors or Patients with Vascular Dementia
The therapeutic potential of cholinesterase inhibitors has been explored in clinical trials of patients with Alzheimer disease with concurrent vascular risk factors, and also in patients with vascular dementia. One 26-week placebo-controlled RCT evaluated the efcacy and safety of rivastigmine for patients with mild to moderately severe Alzheimer disease with or without concurrent vascular risk
51
the trial reports as a safe and effective means of treating vascular dementia. After a pooled analysis of the two trials, the authors wrote that the results ... are somewhat confusing, and further data on donepezils impact on executive functioning would be certainly desirable [18,19]. At the time of writing this article, the data from these vascular dementia trials have not been considered sufcient evidence to license donepezil for treating vascular dementia. However, the positive messages contained in the published RCTs may promote the offlabel use of the drug.
Dementia Associated with Parkinson Disease and Dementia with Lewy Bodies
A Cochrane systematic review identied only one RCT (involving 120 patients) of the efcacy of rivastigmine in patients with probable dementia with Lewy bodies [20,21]. The Cochrane reviewers concluded that the trial showed no statistically signicant difference between the two groups at 20 weeks. A possible benecial effect on neuropsychiatric features was found only in analysis of observed cases, and may therefore be due to bias. Hence the evidence of any benet is currently weak [21]. Two clinical trials have investigated the effect of cholinesterase inhibitors in patients with dementia associated with Parkinson disease. The rst one [22], which found a trend (not statistically signicant) toward better scores on the ADAS-cog is not further discussed here because of its small size (only 22 patients were randomized to receive donepezil or placebo). The second trial, by Emre et al., investigated the effect of rivastigmine in 541 highly selected patients recruited from an unspecied number of centers from 12 countries [23]. Patients included in the trial had received a diagnosis of dementia 6.6 5.2 years (treated arm) and 7.3 5.2 years (placebo arm) after the diagnosis of Parkinson disease. It would be difcult to nd such a population in a clinical setting for a number of reasons. Beyond the diagnostic challenge of differentiating dementia associated with Parkinson disease from dementia of the Lewy body type, there is also evidence that the risk of dementia in Parkinson disease is associated with
age and severity of extrapyramidal signs, and the mean time from onset of Parkinson disease to dementia is estimated to be 10.5 years [2426]. But the exact clinical implications of this RCT are still not clear. The outcome measures used in Emre and colleagues trial were the ADAS-cog and the Alzheimer Disease Cooperative StudyClinicians Global Impression of Change scale. In their trial, the authors considered a mean improvement of 2.25 points in the ADAS-cog score as clinically meaningful, even though this scale has never been used to monitor the progression rate of dementia in Parkinson disease. Among adverse events, Parkinsonian symptoms were reported more frequently in
Clinicians should be careful about generalizing RCT results to their own patients.
the rivastigmine group than in the placebo group. The authors concluded that rivastigmine was associated with moderate but signicant improvements in all symptoms of dementia associated with Parkinson disease, but also with high rates of adverse events, and that the ndings may have implications for clinical practice. But the exact clinical implications of this RCT are still not clear.
Mild Cognitive Impairment: A New Clinical Entity or a New Market Frontier?

Whether MCI can be considered a clinical entity is still a matter of debate (for example, Gauthier and Touchon have argued that there is epidemiological evidence that many subjects labeled as having MCI do not worsen over time and may revert to normal cognitive abilities [27]). Nevertheless, specic drug treatment for MCI has been proposed. Two RCTs have been conducted to investigate whether donepezil delays the onset of dementia in people with MCI. These studies failed to demonstrate any efcacy, while showing a worse safety prole among patients receiving active drug compared with the placebo group. In the rst published trial [28], signicant treatment effects were not seen in
0458
the primary efcacy measures, while more patients treated with donepezil experienced adverse events compared with patients treated with placebo (88% versus 73%). Despite this negative result, a new trial was conducted by Petersen et al., comparing donepezil, vitamin E, and placebo [29]. This study did not show a signicant difference among the three groups in the rate of progression from MCI to Alzheimer disease over a three-year period. Nevertheless, the authors stress some limited effects on secondary measures: a reduced likelihood of progression to Alzheimer disease only during the rst 12 months of treatment, and a benet of donepezil among carriers of one or more apolipoprotein E 4 throughout the three-year follow-up. This latter claim, in particular, was not supported by the data as the study was not statistically powered to evaluate the effect of the treatment in separate groups of apolipoprotein E 4 carriers. Harms-related data were inadequate: the ow of participants through the study phases was not described; the reasons and timing for discontinuation per treatment arm were not reported; only adverse events observed in at least 5% of patients were reported; and the causes of the 23 deaths observed (17 in the double-blind phase and six in the subsequent open-label phase) were not specied. In the double-blind phase, a higher number of deaths was observed in the donepezil arm (n = 7) compared with the vitamin E arm (n = 5) and the placebo arm (n = 5). For the six deaths that occurred during the open-label phase, the original arm (active drug or placebo in the previous double-blind phase) was not reported. (The distribution of these six deaths across the three arms of the trial in the open phase was subsequently reported by Jelic et al. [30]there were three deaths in the donepezil group, one in the vitamin E group, and two in the placebo arm; thus, the total number of deaths per arm in the whole trial was ten in the donepezil group [three from cardiac arrest], six in the vitamin E group, and seven in the placebo group.) Although Petersen et al. conceded that the results do not provide support for a clear recommendation for the use of donepezil in persons with mild cognitive impairment, they did suggest that their ndings could prompt a
52
Advances in Research and Treatment for Alzheimers disease discussion between the clinician and the patient about this possibility [29]. Two trials, each lasting two years and not yet published, evaluated the effect of galantamine on a total of 2,048 patients with MCI randomized to receive galantamine or placebo [31,32]. Overall, the studies did not show that the drug could improve cognition or delay the conversion to dementia. Increased mortality (mostly due to myocardial infarction and stroke) was observed among patients treated with galantamine compared with patients given placebo. On the basis of these results, the US Food and Drug Administration issued a safety warning concerning galantamine [33]. In these trials, the treatment duration (two years) was longer than that of most previous RCTs on Alzheimer disease (typically only six months). The short trials on Alzheimer disease had shown no increased mortality associated with cholinesterase inhibitors compared with placebo. In clinical practice, though, these drugs would likely be prescribed for several years, and the galantamine trials [31,32] have shown that such prolonged use may be associated with increased mortality. A recent review on clinical trials in MCI concluded that none of the reviewed studies met their primary objectives; that is, none of the trials showed a benet of cholinesterase inhibitors in delaying the conversion to dementia or in slowing symptom progression [30]. cohort study of the effectiveness of cholinesterase inhibitors in Alzheimer disease has been conducted in Italy on 5,462 patients [35]. This study showed that the patients most likely to respond to treatment are those without concomitant diseases and those who had demonstrated an early response at three months. Response to treatment did not vary among groups with different Mini Mental State Examination (MMSE) scores at baseline. Based on these results, we suggest that physicians should accurately reevaluate their patients after three months of therapy, and should communicate realistic information to patients and their families about the very modest benets of these drugs. Since 1996, when the rst cholinesterase inhibitor was licensed in the US for the symptomatic treatment of Alzheimer disease, each new published trial on the effect of cholinesterase inhibitors on the various different forms of dementia has raised new questions about the benetrisk prole of these drugs. Reduced cholinergic neurotransmission was the rationale for the use of cholinesterase inhibitors in patients with dementia. Nevertheless, what seemed a biologically plausible intervention has not led to a proven, real improvement in patients well-being.
References
1. Rothwell PM (2005) External validity of randomised controlled trials: To whom do the results of this trial apply? Lancet 365: 8293. 2. McCormack J, Greenhalgh T (2000) Seeing what you want to see in randomised controlled trials: Versions and perversions of UKPDS data. BMJ 320: 17201723. 3. Melzer D (1998) New drug treatment for Alzheimers disease: Lessons for healthcare policy. BMJ 516: 762764. 4. Rogers SL, Fralow MR, Doody RS, Mohs R, Friedhoff LT, et al. (1998) A 24-week, doubleblind, placebo-controlled trial of donepezil in patients with Alzheimers disease. Neurology 50: 136145. 5. Rosler M, Anand R, Cicin-Sain A, Gauthier S, Agid Y, et al. (1999) Efcacy and safety of rivastigmine in patients with Alzheimers disease: International randomised controlled trail. BMJ 318: 633640. 6. Wilcock GK, Lilienfeld S, Gaens E (2000) Efcacy and safety of galantamine in patients with mild to moderate Alzheimers disease: Multicentre randomised controlled trial. BMJ 321: 17. 7. AD2000 Collaborative Group (2004) Longterm donepezil treatment in 565 patients with Alzheimers disease (AD2000): Randomised double-blind trial. Lancet 363: 21052115. 8. Doody RS, Stevens JC, Beck C, Dubinsky RM, Kaye JA, et al. (2001) Practice parameter: Management of dementia (an evidence-based review). Report of the Quality Standards Subcommittee of the American Academy of Neurology. Neurology 56: 11541166. 9. Lanctt KL, Herrmann N, Yau KK, Khan LR, Liu BA, et al. (2003) Efcacy and safety of cholinesterase inhibitors in Alzheimers disease: A meta-analysis. CMAJ 169: 557564. 10. US Food and Drug Administration (1989) Peripheral and Central Nervous System Drugs Advisory Committee Meeting; 1989 July 7; Rockville (Maryland): Department of Health and Human Services, Public Health Service. 227 p. 11. Kaduszkiewicz H, Zimmermann T, BeckBornholdt HP, van den Bussche H (2005) Cholinesterase inhibitors for patients with Alzheimers disease: Systematic review of randomised clinical trials. BMJ 331: 321327. 12. Kmietowicz Z (2005) NICE proposes to withdraw Alzheimers drugs from NHS. BMJ 330: 495. 13. National Institute for Clinical Excellence [NICE] (2006) Appaisal consultation document: Alzheimers diseaseDonepezil, rivastigmine, galantamine and memantine. London: NICE. Available: http:www.nice.org. uk/page.aspx?textsize=10&o=245908. Accessed 16 February 2006. 14. Kumar V, Anand R, Messina J, Hartman R, Veach J (2000) An efcacy and safety analysis of Exelon in Alzheimers disease patients with concurrent vascular risk factors. Eur J Neurol 7: 159169. 15. Erkinjutti T, Kurz A, Gauthier S, Bullock R, Lilienfeld S, et al. (2002) Efcacy of galantamine in probable vascular dementia and Alzheimers disease combined with cerebrovascular disease: A randomized trial. Lancet 359: 12831290. 16. Black S, Roman GC, Geldmacher DS, Salloway S, Hecker J, et al. (2003) Efcacy and tolerability of donepezil in vascular dementia. Positive results of a 24-week, multicenter, international, randomized, placebo-controlled clinical trial. Stroke 34: 23232332. 17. Wilkinson D, Doody R, Helme R, Taubman K, Mintzer J, et al. (2003) Donepezil in vascular dementia. A randomized, placebo-controlled study. Neurology 61: 479486. 18. Geldmacher DS (2003) Donepezil in vascular dementia. A viewpoint. Drugs Aging 20: 1137 1138. 19. Black S (2003) Donepezil in vascular dementia. A viewpoint. Drugs Aging 20: 11371138. 20. McKeith, Del Ser T, Spano P, Emre M, Wesnes K, et al. (2000) Efcacy of rivastigmine in dementia with Lewy bodies: A randomised, double blind, placebocontrolled international study. Lancet 356: 20312036. 21. Wild R, Pettit T, Burns A (2003) Cholinesterase inhibitors for dementia with Lewy bodies. Cochrane Database Syst Rev 2003: CD003672. 22. Ravina B, Putt M, Siderowf A, Farrar JT, Gillespie M, et al. (2005) Donepezil for dementia in Parkinsons disease: A randomised, double blind, placebo controlled, crossover study. J Neurol Neurosurg Psychiatry 76: 934939. 23. Emre M, Aarsland D, Albanese A, Byrne J, Deuschl G, et al. (2004) Rivastigmine for dementia associated with Parkinsons disease. N Engl J Med 351: 25092518. 24. Levy G, Schupf N, Tang MX, Cote LJ, Louis ED, et al. (2002) Combined effect of age and severity on the risk of dementia in Parkinsons disease. Ann Neurol 51: 722729. 25. Apaydin H, Ahlskog E, Parisi JE Boeve BF, Dickson DW (2002) Parkinson Disease neuropathology. Later-developing dementia and loss of the levodopa response. Arch Neurol 59: 102112. 26. Aarsland D, Ballard CG, Halliday G (2004) Are Parkinsons disease with dementia and dementia with Lewy bodies the same entity? J Geriatr Psychiatry Neurol 17: 137145.
Conclusion
At present, donepezil, galantamine, and rivastigmine are licensed only for the treatment of mild to moderate Alzheimer disease. The treatment effect is modest, and there is evidence of wide variability in the outcomes reported: some patients will have improved, others stayed the same, while others will have deteriorated. This variance should be comparative in both the treatment and the placebo groups but care should be taken over the interpretation of the mean scores [34]. However, a minority of people with Alzheimer disease may benet from the cholinesterase inhibitors, and further research is needed to identify these subgroups of people, considering, in particular, long-term and worthwhile improvements such as delay in institutionalization. A
0459
53
27. Gauthier S, Touchon J (2005) Mild cognitive impairment is not a clinical entity and should not be treated. Arch Neurol 62: 11641166. 28. Salloway S, Ferris S, Kluger A, Goldman R, Griesing T, et al. (2004) Efcacy of donepezil in mild cognitive impairment. A randomized placebo-controlled trial. Neurology 63: 651 657. 29. Petersen RC, Thomas RG, Grundman M, Bennett D, Doody R, et al. (2005) Vitamin E and donepezil for the treatment of mild cognitive impairment. N Engl J Med 352: 2379 2388. 30. Jelic V, Kivipelto M, Winblad B (2005) Clinical trials in mild cognitive impairment: Lessons
for the future. J Neurol Neurosurg Psychiatry. Epub ahead of print 23 November 2005. 31. Clinical Study Results Database. Protocol number GAL-INT-11. Available: http:www. clinicalstudyresults.org/documents/ company-study_96_1.pdf. Accessed 28 February 2006. 32. Clinical Study Results Database. Protocol number GAL-INT-18. Available: http:www. clinicalstudyresults.org/documents/ company-study_96_2.pdf Accessed 28 February 2006. 33. US Food and Drug Administration [FDA] (2005) Alert for healthcare professionals on galantamine hydrochloride (marketed
as Reminyl). Rockville (Maryland): FDA. Available: http:www.fda.gov/cder/drug/ InfoSheets/HCP/galantamineHCP.htm. Accessed 16 February 2006. 34. Takeda A, Loveman E, Clegg A, Kirby J, Picot J, et al. (2006) A systematic review of the clinical effectiveness of donepezil, rivastigmine and galantamine on cognition, quality of life and adverse events in Alzheimers disease. Int J Geriatr Psychiatry 21: 1728. 35. Raschetti R, Maggini M, Sorrentino GC, Martini N, Caffari B, et al. (2005) A cohort study of effectiveness of acetylcholinesterase inhibitors in Alzheimers disease. Eur J Clin Pharmacol 61: 361368.
54
0460
PLoS MEDICINE
Cholinesterase Inhibitors in Mild Cognitive Impairment: A Systematic Review of Randomised Trials

Roberto Raschetti1*, Emiliano Albanese1,2, Nicola Vanacore1, Marina Maggini1
1 National Center for Epidemiology, Surveillance and Health Promotion, National Institute of Health, Rome, Italy, 2 Section of Epidemiology, Institute of Psychiatry, Kings College London, London, United Kingdom Funding: The authors received no specific funding for this study. Competing Interests: The authors have declared that no competing interests exist. Academic Editor: Gary Small, University of California Los Angeles Center on Aging, United States of America Citation: Raschetti R, Albanese E, Vanacore N, Maggini M (2007) Cholinesterase inhibitors in mild cognitive impairment: A systematic review of randomised trials. PLoS Med 4(11): e338. doi:10.1371/journal. pmed.0040338 Received: March 21, 2007 Accepted: October 12, 2007 Published: November 27, 2007 Copyright: 2007 Raschetti et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: AD, Alzheimer disease; AE, adverse event; CDR, clinical dementia rating; ChEI, cholinesterase inhibitor; MCI, mild cognitive impairment; MMSE, minimental state examination; RCT, randomized controlled trial * To whom correspondence should be addressed. E-mail: roberto. raschetti@iss.it
ABSTRACT
Background
Mild cognitive impairment (MCI) refers to a transitional zone between normal ageing and dementia. Despite the uncertainty regarding the definition of MCI as a clinical entity, clinical trials have been conducted in the attempt to study the role of cholinesterase inhibitors (ChEIs) currently approved for symptomatic treatment of mild to moderate Alzheimer disease (AD), in preventing progression from MCI to AD. The objective of this review is to assess the effects of ChEIs (donepezil, rivastigmine, and galantamine) in delaying the conversion from MCI to Alzheimer disease or dementia.

The terms donepezil, rivastigmine, galantamine, and mild cognitive impairment and their variants, synonyms, and acronyms were used as search terms in four electronic databases (MEDLINE, EMBASE, Cochrane, PsycINFO) and three registers: the Cochrane Collaboration Trial Register, Current Controlled Trials, and ClinicalTrials.gov. Published and unpublished studies were included if they were randomized clinical trials published (or described) in English and conducted among persons who had received a diagnosis of MCI and/ or abnormal memory function documented by a neuropsychological assessment. A standardized data extraction form was used. The reporting quality was assessed using the Jadad scale. Three published and five unpublished trials met the inclusion criteria (three on donepezil, two on rivastigmine, and three on galantamine). Enrolment criteria differed among the trials, so the study populations were not homogeneous. The duration of the trials ranged from 24 wk to 3 y. No significant differences emerged in the probability of conversion from MCI to AD or dementia between the treated groups and the placebo groups. The rate of conversion ranged from 13% (over 2 y) to 25% (over 3 y) among treated patients, and from 18% (over 2 y) to 28% (over 3 y) among those in the placebo groups. Only for two studies was it possible to derive point estimates of the relative risk of conversion: 0.85 (95% confidence interval 0.641.12), and 0.84 (0.571.25). Statistically significant differences emerged for three secondary end points. However, when adjusting for multiple comparisons, only one difference remained significant (i.e., the rate of atrophy in the whole brain).
Conclusions
The use of ChEIs in MCI was not associated with any delay in the onset of AD or dementia. Moreover, the safety profile showed that the risks associated with ChEIs are not negligible. The uncertainty regarding MCI as a clinical entity raises the question as to the scientific validity of these trials.
CholinesteraseInhibitorsinMildCognitiveImpairment:ASystematicReviewofRandomisedTrialsCholinesteraseInhibitorsin MildCognitiveImpairment:ASystematicReviewofRandomisedTrials
1818
55
Mild Cognitive Impairment: A Review
Introduction
Alzheimer disease (AD) is a neurodegenerative disorder characterized by cognitive and memory deterioration, progressive impairment of activities of daily living, and a multiplicity of behavioural and psychological disturbances. AD is the main cause of dementia syndrome and one of the most burdensome conditions of later life. In a recent Delphi consensus study, based on a systematic review of the literature on the prevalence of dementia, the authors estimated that more than 24 million people worldwide currently have dementia [1]. Dementia is also the leading cause of disability in persons aged 60 y and older, and its direct and indirect costs are very high [2,3]. There exists little evidence of modiable risk factors for AD [1]; disease-modifying therapies are not available [4], and of the symptomatic therapies the efcacy of cholinesterase inhibitors (ChEIsdonepezil, rivastigmine, and galantamine) in mild-to-moderate AD patients is questionable and has been widely debated [5,6]. In recent years, efforts have been made to study individuals believed to be at greater risk of developing dementia and who were considered as having mild cognitive impairment (MCI), which refers to a transitional zone between normal ageing and dementia [79]. However, widely accepted and validated criteria for diagnosing MCI do not exist, and the differences between this term and the other clinical labels given to the cognitive dysfunctions associated with aging are not clear (e.g., benign senescent forgetfulness, age-associated memory impairment, age-associated cognitive decline, mild cognitive decline, mild neurocognitive decline, and cognitive impairment no dementia) [1014]. A number of longitudinal studies have attempted to estimate the rate of conversion from MCI to dementia. When comparing the different studies, the conversion rates vary greatly (from 9% to 40%) because of differences in sampling criteria, the case denition, the length of follow-up, assessment procedures, and the number of persons lost to followup. Moreover, up to 40% of MCI cases reverted to a normal cognitive condition within 23 y [1518]. Despite the uncertainty regarding the denition of MCI as a clinical entity, clinical trials on ChEIs as a preventive treatment have been conducted in the attempt to study the possible role of these agents in slowing the onset of AD, because of the purported pathophysiological relationship between MCI and AD. The rst published trials showed that these agents were not efcacious, yet the authors attributed these ndings mainly to methodological issues, such as the selection of homogeneous samples, the denition of reliable outcomes, and the duration of treatment [19]. The substantial failure of these attempts was recently conrmed by two Cochrane systematic reviews, one on galantamine and the other on donepezil [20,21]. On the basis of the results obtained in the clinical trials on galantamine, in 2005 the US Food and Drug Administration [22] and health authorities all over the world issued a safety warning advising that galantamine should be used only for the approved indications of mild to moderate AD, and that for other possible indications (e.g., MCI) the risks may outweigh the benet. Nevertheless, the suitability of using ChEIs to treat persons labelled with MCI continues to be widely debated [2326].
Most of the trials conducted on the efcacy of donepezil, galantamine, and rivastigmine on MCI remain unpublished even after years from their conclusion. As a consequence, the original data are not available for researchers or physicians who could use these drugs in their clinical practice. In this context, we conducted a systematic review of published and unpublished trials on ChEIs, so as to provide an update on the riskbenet prole for this drug class (donepezil, galantamine, and rivastigmine) in treating MCI.
Methods Search Strategy

In February of 2006, we searched for the terms donepezil, rivastigmine, galantamine and mild cognitive impairment, and their variants, synonyms, and acronyms in the following sources: (1) four electronic databases MEDLINE (http://www.pubmed.gov/; 1990 to February 2006), EMBASE (http://www.embase.com/; 1990 to February 2006), The Cochrane Collaboration (http://www.cochrane.org/index. htm), and PsycINFO (http://www.apa.org/psycinfo/); and (2) three registersin particular, the Cochrane Collaboration Trial Register (http://www.thecochranelibrary.com/), Current Controlled Trials (http://www.controlled-trials.com/), and Clinicaltrials.gov (http://www.clinicaltrials.gov/). We also examined the bibliographies of all of the considered publications so as to identify other studies. We did not consider tacrine in our search, because it is no longer used in clinical practice, because of its high toxicity [27,28].
Inclusion Criteria
Published and unpublished studies were included if they were: randomised controlled trials (RCTs) of cholinesterase inhibitors (donepezil, rivastigmine, galantamine); written in English; and conducted among persons with abnormal memory function documented by a neuropsychological assessment and/or who met diagnostic criteria for MCI. All studies were required to have as an outcome measure the time to development of dementia or of possible or probable AD, or the improvement of measurement concerning cognitive/clinical/neuropsychiatric domains, and/or improvements based on neuroimaging examinations.
Exclusion Criteria
Studies were excluded if the design was not an RCT; the study did not present original data; the study drug was not a ChEI; and participants were cognitively normal for age or had already been classied as having dementia of any type. Ongoing studies were not included in the review.
Data Extraction
We screened the obtained titles, abstracts, and protocols. Data were extracted using a standardized data extraction form that was developed by all authors. The extracted information included doses of medication; duration of the trial; the number, age, and gender of participants; enrolment criteria; funding sources; primary and secondary outcomes; all-cause dropouts; adverse events; and deaths occurring during the study period. The data were extracted and summarized by two investigators (EA and NV) not blinded to the studys authors or to the publication status. To ensure that accurate data were obtained, a third investigator (RR)
56

Figure 1. Trials Identification and Selection Process doi:10.1371/journal.pmed.0040338.g001
checked the extracted data, and discrepancies were resolved through discussions among all investigators.
Data Analysis
When available, we recorded for each outcome the mean difference between baseline and follow-up measures for the individual study arms, and the standard deviation of each difference. When the standard deviation was not given it was estimated from the standard error. For each outcome the effect measure was estimated as the mean difference between treated and placebo groups. A Bonferroni correction for multiple comparisons was done for the individual tests within the same study. We have not included any estimate of pooled effect because of clinical heterogeneity among the populations enrolled in the trials included in the review. Statistical analyses were conducted by using Stata software 8.0 (Stata, College Station, Texas, United States).
because they were not RCTs (n 119) or participants did not have MCI (ve studies). After having evaluated the full text of the 33 remaining studies, we excluded 25 studies for the following reasons. Four were duplicates of other studies; nine had not been conducted among MCI patients (in most cases participants had already received a diagnosis of AD); six were not RCTs (three were comments or editorials on existing studies, and three were observational studies); one was conducted among persons not treated with ChEIs; and ve trials were still ongoing.
Characteristics of the Trials and Participants

The eight trials that investigated the efcacy and safety of ChEIs (three on donepezil, two on rivastigmine, and three on galantamine) in persons with MCI were included in the review (Table 1) [3037]. Of these, three were published in peerreviewed journals [34,36,37], and ve were retrieved from clinical trial registers [3033,35]. Regrettably, extensive synopses were available for only three of these ve trials [3133], whereas for the other two we were able to retrieve only a brief description of the principal characteristics [30,35]. Additional information were sought unsuccesfully from the original investigators and from the investigators institutions. For this reason, these two trials are not always included when discussing the results. Notably, one of them was suspended, yet we were not able to obtain information from the manufacturers on the reasons for suspension [35]. The main characteristics of the eight trials are presented in Table 1. All trials but one were totally or partially sponsored by pharmaceutical companies [30]. The duration of the trials on donepezil was 24 wk, 1 y, and 3 y. Dosage was reported in two of these trials (10 mg/d after a starting dose of 5 mg/d)
Assessment of Methodological Quality

To roughly measure the quality of the study design/ reporting of each trial, we used the validated scale developed by Jadad and colleagues, which assigns a numerical score of 0 5 (5 being the best score) [29].
Results Search Flow

The literature search yielded 157 potentially relevant citations: 109 studies from electronic databases and 48 from clinical trial registers. The selection process is illustrated in Figure 1. Of the 157 citations, 124 were excluded either
57
58
Drug, Dose, Duration Randomized (ChEI; Placebo), n
Rivastigmine, 312 mg/d, 34 y Rivastigmine, , Galantamine, 16 or 24 mg/d, 2 y Galantamine, 16 or 24 mg/d, 2 y Donepezilginkgo biloba, , 1 y Donepezil, 10 mg/d, 24 wk 270 (133; 137) 72 (5589) 42% 40 () (65) 995 (442; 452 )
c
Table 1. Characteristics of the Trials on Cholinesterase Inhibitors Included in the Review

Participants Mean Age (Range), y
70 (5585) (5085) (50) (50) 52% 69 22
PLoS Medicine | www.plosmedicine.org Females, % Main Enrolment Criteriaa Number of Sites Funded or Sponsored by
1,018 (508; 510) 1,062 (498; 511c) Novartis Novartis Johnson & Johnson Johnson & Johnson CDR 0.5, NYDPR,9, HAM-D,13 23,MMSE,27, memory complaint CDR 0.5, memory0.5, NYDPR10 CDR 0.5, memory0.5, NYDPR10 Petersens MCI criteria Donepezil, 10 mg/d, 3 y 73 (5590) 769d (253; 259) 46% 69 Galantamine, 16 or 24 mg/d, 16 wk 19 (8; 11) 71 (5187) 0% MMSE . 24, CDR 0.5, ADL , 1.5, HAM-D,12, Hachinsky scale,4 MMSE.24, CDR 0.5, Logical delayed-recall score 1.52 SD below normal, HAM-D,12, ADCS-ADL-MCI.45 MMSE.26, Petersens MCI criteria 1 Janssen Pharmaceutica
Published or Unpublished
Study
Study Period
Unpublishedb
InDDEX [31]
19992004
NCT00134953 [35]
2003suspended
GAL-INT-18 [33]
20012003
GAL-INT-11 [32]
20012003
NCT00042172 [30]
20022004
1821
Published
Salloway [36]
19992000
National Institute of Mental Health Eisai & Pfizer
Petersen [37]
19992004
National Institute on Aging, Pfizer, and Eisai
Koontz [34]
Unavailable
, not reported a ADL, activities of daily living; CDR, clinical dementia rating (scale); HAM-D, Hamilton Rating Scale for Depression; MMSE, mini mental state examination; NYDPR, New York University Delayed Paragraph Recall. b The year refers to the study start. c Analysed for efficacy. d 257 patients were allocated to the vitamin E arm. doi :10.1371/journal.pmed.0040338.t001

Table 2. Conversion Rates from Mild Cognitive Impairment to Alzheimer Disease or Dementia and Jadad Quality Score
Category Gal-INT-18 [33]: Galantamine
2y 17%; 21% 2
Gal-INT-11 [32]: Galantamine

2y 13%; 18% 2
InDDEX [31]: Rivastigmine

34 y 51%; 63% 17%; 21% 3
Salloway [36]: Donepezil

24 wk 68%; 83% 3
Petersen [37]: Donepezil

3y 64%; 74% 25%; 28% 3
Koontz [34]: Galantamine

16 wk 50%; 36% 3
Duration of the study Subjects completing the study (ChEI; placebo) Conversion rate (ChEI; placebo) Jadad quality score (05)
, not reported. doi:10.1371/journal.pmed.0040338.t002
[36,37]. One of the three trials [34] on galantamine differed from the other two [32,33] in duration (16 wk versus 2 y), whereas dosages were identical in all three (1624 mg/day). One of the two trials on rivastigmine lasted 34 y and applied dosages between 3 and 12 mg/d [31]; for the other trial, we were not able to retrieve information on either duration or dosage [35]. Of the six trials for which dosage was reported, in three, once the maximum dosage was reached, treatment continued with that dosage [34,36,37], whereas in the other three trials, the dosage remained exible (16 or 24 mg for galantamine and 312 mg for rivastigmine) [3133]. One trial was carried out in only one site [34]; the two published studies on donepezil were conducted in 22 and 69 centres, respectively [36,37]. One of the unpublished studies on rivastigmine was conducted in 69 centres. For the remaining four trials, this information was not available. The number of participants ranged from 19 to 1,062 and varied greatly among the trials. In all trials, all participants were greater than 50 y of age. When reported, the percentage of females ranged from 42% to 52%; in one trial, all 19 participants were males [34]. All studies enrolled patients accordingly to the Petersens criteria: not demented, memory complaint, preserved general cognitive function, intact activities of daily living, impaired memory for age and education [9]. However, the operationalisation of the MCI diagnostic criteria differed widely among the trials (Table 1). Cognitive functions were assessed using the mini mental state examination (MMSE) with different cutoffs [3435], or the clinical dementia rating (CDR) scale [31 33], or both [36,37]. Activities of daily living were assessed in two studies using two different scales (ADCL-ADL-MCI and ADL) [36,37]. The assessment of memory was based on the New York University paragraph recall test (immediate and delayed) in three studies [3133], with two different cut-off scores, or on the Wechsler Memory Scale (Revised) Logical Memory delayed recall test [37]; three trials used two unspecied memory tests [32,33,35]. Only three of the trials [31,36,37] ascertained the psychiatric prole of participants, yet poorly, in that they only used the Hamilton Rating Scale for Depression, condition that may inuence the execution of a neuropsychological test on memory. Finally, two studies [30,34] claim generically the use of Petersens MCI criteria without any operational denition. Information on the measurement tools used in the trials is provided in Text S1. The use of this range of diagnostic criteria was an important source of clinical heterogeneity among the
populations enrolled, except for two trials [32,33] that used essentially the same protocol.
Study Quality
The quality assessment was carried out for all of the published trials [34,36,37] and for three of the unpublished ones [3133]. However, it should be kept in mind that scarce or inadequate reporting does not necessarily imply that the methodology was of low quality. The Jadad scores ranged from 2 to 3, indicating medium to low reporting quality (Table 2). The description of the randomization process was adequate in only one trial [37]. In three trials, the description of this process was sufcient for understanding that the allocation of patients could not be predicted [32,33,37]. In all trials, the blinded status of the assessors, care providers, and participants was undermined by the fact that ChEIs have clear side effects. Three studies specied that the placebo was indistinguishable from the drug [32,33,37]. In one trial, an independent panel veried in double-blind conditions the occurrence of the primary outcome, yet not of all the surrogate measures [37]. The baseline characteristics of the participants were available only for the published studies. Four trials used the intention-totreat (ITT) analysis [32,33,36,37]; however, only one carried out a sensitivity analysis [37].
Efficacy and Adverse Events

As mentioned above, we were able to obtain the results for six of the eight studies (Table 2). Conversion from MCI to AD or dementia was considered as the primary end point in four studies. In two studies, AD and dementia were diagnosed according to, respectively, NINCDS-ADRDA (National Institute of Neurological Disorders and StrokeAlzheimers Disease and Related Disorders Association) and DSM-IV (Diagnostic and Statistical Manual of Mental Disorders, 4th Edition) criteria [31,37], whereas in the other two, only a change in CDR score from 0.5 to 1.0 or higher was considered [32,33]. No signicant differences emerged in the probability of conversion between the treated groups and the placebo groups. The rate of conversion ranged from 13% (over 2 y) to 25% (over 3 y) among treated patients, and from 18% (over 2 y) to 28% (over 3 y) among those in the placebo groups (Table 2). Only for two studies was it possible to derive point estimates of the relative risk of conversion: 0.85 (95% condence interval 0.641.12) [31], and 0.84 (0.571.25) [37]. A reduced likelihood of conversion to AD, compared to the placebo group, was reported by one trial during the rst 12 mo of treatment, yet this result was not observed at the end of the 3-y follow-up [37]. As discussed in the Cochrane review on
59
Table 3. Primary (I) and Secondary (II) Efficacy Measures Used in the Trials Included
Category Measure Study Gal-INT18 [33]
Cognitive ADAS-cog ADAS-cog 11 ADAS-cog 13 ADAS-cog MCI BNT Buschke Reminding CANTAB Category Fluency test Clock Drawing test CVLT Delayed word list recall Digit-Backward test Digit Cancellation Task DSST Letter Numbering sequence Maze test Number cancellation test NYU Paragraph Recall, Revised (immediate/delayed) Symbol Digit Modalities task Verbal Fluency Category Warrington Faces Test WMS-R ADCS-CGIC CDR CDR-SB CGIC-MCI GDS MMSE PGA QOL-AD ADCS-ADL ADCS-ADL-MCI FAQ Beck Depression Inventory NPI Benton Judgement MRI hippocampal volume MRI whole brain volume II II I II II II I, II II
Gal-INT11 [32]
II II I II II II I, II II II II
InDDEX [31]
II II II II II II II II II II II II II II II II II II II II II
Salloway [36]
II II II II I II II II I I II
Petersen [37]
II II II II II II II II II II II II II II
Koontz [34]
I II II
Global and Clinical
Activities of daily living
Neuropsychiatric symptoms
Neuroimaging
Efficacy measures are described in Text S1. doi:10.1371/journal.pmed.0040338.t003
60
donepezil, if we accept an effect of donepezil in delaying AD for 12 months we must also accept that it then accelerates the appearance of AD after 18 months [21]. The percentage of the study population that completed the study ranged from 51% to 68% among those allocated to treatment and from 36% to 83% among placebo recipients. The percentage of persons who completed the study was consistently lower for the treatment group, compared to the placebo group, except in the Koontz trial (Table 2). Furthermore, Petersen and colleagues reported that persons with more severe cognitive impairment were more likely to withdraw from the study and that there was a tendency for treated individuals to withdraw from the trial earlier [37]; thus the missing data were not randomly distributed. No study provided information that would allow the reader to assess the potential bias due to differential dropout (e.g., concomitant disorders). The response to treatment was also assessed using a range
of measures derived mainly from AD trials [19]. A total of 36 different scales, tests, and neuropsychological batteries, and two measures of volumetric imaging, were used as either a primary or a secondary end point (Table 3) (detailed information on the measurement tools used in the trials is provided in Text S1). Efcacy was measured on an ITT population in four trials [32,33,36,37], and the last observation carried forward (LOCF) method was used for missing data points in three of them [32,33,36]. Only one trial adjusted the resulting p-values for multiple comparisons [37]. In Figure 2 are reported the point estimates and condence intervals for the outcomes for which specic results from the original studies were available. Data from one trial [34] were not considered as they actually refer to ten of the 19 trial participants, and they were not based on ITT analysis. Statistically signicant differences emerged only for the mean rate of brain volume atrophy, and the CDRSum of

Figure 2. Effect of Treatment on the Efficacy Measures Used in the Included Studies Data points represent change in the treated groups versus the placebo groups. For each measure the range of possible scores are reported, when available, in square brackets. *Z-cognitive is a composite score based on a ten-test neuropsychological test battery in the InDDex study, and on a eight-test battery in the Petersen study. Doi:10.1371/journal.pmed.0040338.g002
the Boxes scores for galantamine [32]; and the cognitive functions evaluated by ADAS-cog 13 for donepezil [36]. When adjusting for multiple comparisons using the Bonferroni method, only the difference in the rate of atrophy in the whole brain remained signicant. The percentages of participants with at least one adverse event (AE), those with severe AEs, and those who discontinued for AEs were reported in only four trials [3133,36]. Three trials described AEs occurring in at least 5% of participants [31,36,37]. One trial did not report AEs at all [34]. The percentage of participants with at least one AE was very high among both treatment recipients (88%96%) and placebo recipients (73%93%) (Table 4). The rate of discontinuation due to AEs was consistently higher for treatment recipients (21%24%) than for placebo recipients (7%13%). The data on causes of death were overall inadequate. Only GAL-INT-11 (the protocol for the galantamine RCT [32]) reported all of the causes of death for each
study arm: one death occurred among placebo recipients (arrhythmia and cardiac arrest) and six deaths occurred among treatment recipients [32].
Discussion
The use of ChEIs in persons with MCI, for periods ranging from less than 4 mo to 3 y, was not associated with any delay in the onset of AD or dementia. Furthermore, according to the 38 surrogate measures used in the trials, and after appropriate adjustment for multiple comparisons, only neuroimaging showed a signicant difference in favour of the drug being studied; the clinical implications of this nding are unclear [32]. Moreover, the safety prole showed that the risks associated with ChEIs are not negligible. These results conrm for the all class of ChEIs those reported by two Cochrane systematic reviews. The rst review was on the effect of galantamine in patient with MCI or AD; the authors concluded that the clinical benet was marginal
61
Table 4. Summary of Data on Adverse Events Extracted from the Included Trials
Published or Unpublished Study Study Drug Any AEs, % Participants, ChEI; Placebo Severe AEs, % Participants, ChEI; Placebo Discontinuation for AEs, % Participants, ChEI; Placebo
24%; 13% 23%; 10% 21%; 9%
Deaths, n, ChEI; Placebo
Causes of Death
Unpublisheda
InDDEX 1999 [31] GAL-INT-18 2001 [33] GAL-INT-11 2001 [32]
Rivastigmine Galantamine Galantamine
96%; 93% 90%; 86% 90%; 88%
28%; 30% 19%; 21% 18%; 17%
7; 13 7; 0 6; 1
Published
Salloway 2004 [36] Petersen 2005 [37] Koontz 2005 [34]
Donepezil Donepezil Galantamine
88%; 73%
4%; 4%
22%; 7%b
0; 1 7; 5
Two sudden deaths, five not reported Galantamine: two suicides, two myocardial infarctions, one bronchial cancer and sudden death, one cerebrovascular disorder and syncope; placebo: one arrhythmia
, not reported. a The year refers to the study start. b Excluding deaths. doi:10.1371/journal.pmed.0040338.t004
62
but galantamine use in MCI is not recommended due to its association with an excess death rate [20]. The second review [21] included two trials on donepezil, one that showed some promise for certain outcomes [36], and the other that showed side effects and no evidence of efcacy [37]. The authors conclusion was there is no evidence to support the use of donepezil for patient with MCI. The putative benets are minor, short lived and associated with signicant side effects [21]. A recent meta-analysis on four trials using progression to dementia as the major parameter of efcacy found an approximate 24% reduction of risk of conversion to dementia and an increase of more than 50% in adverse events [38]. This pooled effect estimate was obtained notwithstanding the substantial heterogeneity of populations enrolled nor the methods of assessment. Moreover, the relative risk of conversion could not be calculated for the two trials on galantamine, as original data were necessary to apply a Cox proportional hazard ratio model. Our revision of all the trials on the three drugs permits an overall comparison across the studies with respect to design, objective, and denition of MCI. The primary end point of prevention trials should be the time to development of dementia or AD; this measure was used in only four of the trials included in this review [31 33,37]. The efcacy of the study drugs was also assessed on cognitive and/or functional domains applying a number of surrogate measures: 38 different instruments were used, considering simultaneously a wide range of hypotheses. Moreover, most of these measures have been developed for AD trials and transposed to MCI trials without rst having been validated. However, it has already been claimed that the validation process is not simple, given that it is subordinate to the denition of MCI as a clinical entity, which itself is controversial [1014,24]. A rst important consequence of this uncertainty is that the trial populations were not homogeneous, even if the same
criteria proposed by Petersen and colleagues were used [9]. Petersen and colleagues, in fact, did not specify which neuropsychological tests or instrument should be used to operazionalise MCI criteria. The predictable consequence of this exibility is that operationalisation of the MCI diagnostic criteria differed widely among the trials, giving rise to quite different populations. This was conrmed in a recent study in which the authors applied the enrolment criteria used in the GAL-INT-11, InDDEX, and Petersen et al. trial to the same cohort of 150 participants sampled in a memory clinic [15]. The study found that MCI was diagnosed in 51.3%, 21.3%, and 16.7% of the participants when applying, respectively, the criteria of GAL-INT-11, InDDEX, and Petersens trial. This wide clinical heterogeneity among the study populations is the main reason for not combining the included studies in a pooled analysis. In general, the uncertainty regarding MCI as a clinical entity raises the question as to the scientic validity and ethical value of these trials [39]. In fact, the requirement of scientic validity regards not only the mere technical domain of the correct design and conduct of a clinical trial. It is also a criterion to apply to the soundness of the clinical question approached by the experimentation. In a recent New England Journal of Medicine editorial, Karlawish used the concept of the logic of clinical purpose in discussing the role of clinical trials in AD [40]. According to this idea, Clinical trials are logically grounded in and ethically justied by the way they reect and contribute to clinical practice. As some participants in a 2006 workshop dedicated to MCI have reported, no agreement emerged during the conference as to the clinical utility of MCI, in that there was no consensus on the nature of MCI (is it a syndrome, a risk state, a new diagnosis?) [13]. Moreover, there is epidemiological evidence that the diagnosis of MCI is often unstable, and many persons labelled as having MCI may revert over time to normal cognition [16

18,23,24]. Since none of the trials provided information on this phenomenon, it cannot be excluded that some participants may have received the treatment despite having reverted to normal cognition. A possible limitation of our review was that there was no information on the results of two trials: one of the trials was suspended for unknown reasons [35], and the other was completed in September 2004, yet the results are not yet available [30]. Nonetheless, this probably did not bias the riskbenet prole of ChEIs for MCI patients, given that publication bias usually works in favour of the study drugs. With regard to the quality of reporting of the trials, a common shortcoming was the inadequate description of the randomization and blinding procedures. Other important weaknesses were the very poor description of dropouts and of harm-related issues (e.g., only one trial reported all of the causes of death for each study arm). Our review was not blinded to the trials. This may have inuenced our assessments of the quality of primary studies, although, given the simplicity of the Jadad scale, distorted judgements are very unlikely. On average, the maximum dosage of ChEIs was used in all trials, which probably contributed to the high frequency of dropout and discontinuation for AEs. Mortality was higher among persons receiving donepezil or galantamine, compared to placebo recipients, and this excess in mortality seems to have been prevalently due to cardio- and cerebrovascular diseases. In GAL-INT-11 [32] and GAL-INT-18 [33], the regulatory authorities considered the increased mortality as noteworthy and stressed that ChEIs are not indicated for MCI patients; they also recommended that ChEIs be used with caution in AD patients with cardiovascular risk factors [22,41,42]. Recently, some authors have suggested the rating of medial temporal atrophy, performed using magnetic resonance imaging, as a routine clinical evaluation to identify individuals with MCI who are destined to progress to dementia within 3 years [43]. MCI seems to be an example of a risk factor conceptualised as a clinical condition, and it is surely still too heterogeneous and unpredictable as a clinical entity to enable researchers to establish its exact role in the progression toward dementia. From a public health point of view, it seems reasonable to afrm that additional research for clearly dening MCI is needed before testing new pharmacological treatments. When there is controversy surrounding the denition of a condition or disease, even inconclusive results from RCTs can be used to suggest treatment for persons tagged with some pre-disease condition. For example, in Italy an extimated 27% of patients diagnosed with MCI are prescribed cholinesterase inhibitors off-label [44]; it is likely that this situation is not limited to Italy. The philosophy of widening the boundaries of treatable illness corresponds to the strategy of expanding the market for new products. This has been recently described as disease mongering [45,46]. This issue was also recently addressed by Britains House of Commons Health Committee: [. . .] There has been a trend towards categorising more and more individuals as abnormal or in need of drug treatment [. . .]. Where disease awareness campaigns end and disease mongering begins is a very indistinct line [47]. This review shows that the diagnosis of MCI is uncertain
and variable, and whatever criteria are adopted, ChEIs are not effective in either preventing AD or improving cognitive functions in persons with MCI. These drugs may even be harmful in some people. Thus the alleged clinical implications of the trials, as claimed by some of the authors, are not justied by the data.
Supporting Information
Text S1. Measurement Tools Used in the Trials Found at doi:10.1371/journal.pmed.0040338.sd001 (57 KB DOC)
Acknowledgments
Author contributions. RR and MM designed the study. RR, EA, NV, and MM analyzed the data and contributed to writing the paper.
References 1. Ferri CP, Prince M, Brayne C, Brodaty H, Fratiglioni L, et al. (2002) Global prevalence of dementia: a Delphi consensus study. Lancet 366: 21122117. 2. WHO (2003) World Health Report 2003-Shaping the future. Geneva: WHO. Available at: http://www.who.int/whr/2003/en/index.html. Accessed 31 October 2007. 3. Langa KM, Chernew ME, Kabeto MU, Herzog AR, Ofstedal MB, et al. (2001) National estimates of the quantity and cost of informal caregiving for the elderly with dementia. J Gen Intern Med 16: 770778. 4. Cummings JL (2004) Alzheimers disease. N Engl J Med 351: 5667 5. Kaduszkiewicz H, Zimmermann T, Beck-Bornholdt HP, van den Bussche H (2005) Cholinesterase inhibitors for patients with Alzheimers disease: systematic review of randomised clinical trials. BMJ 331: 321327. 6. Pelosi AJ, McNulty SV, Jackson GA (2006) Role of cholinesterase inhibitors in dementia care needs rethinking. BMJ 333: 491493. 7. Petersen RC (2003) Mild cognitive impairment clinical trials. Nat Rev Drug Discov 2: 646653. 8. Mount C, Downton C (2006) Alzheimers disease: progress or prot? Nat Med 12: 780784. 9. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, et al. (1999) Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 56: 303308. 10. Ritchie K, Touchon J (2000) Mild cognitive impairment: conceptual basis and current nosological status. Lancet 355: 225228. 11. Gauthier S, Reisberg B, Zaudig M, Petersen RC, Ritchie K, et al. (2006) Mild cognitive impairment. Lancet 367: 12621270. 12. Petersen RC (2006) Mild cognitive impairment. Lancet 367: 1979. 13. Whitehouse P, Brodaty H (2006) Mild cognitive impairment. Lancet 367: 1979. 14. Winblad B, Palmer K, Kivipelto M, Jelic V, Fratiglioni L, et al. (2004) Mild cognitive impairment-beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256: 240246. 15. Visser PJ, Scheltens P, Verhey FR (2005) Do MCI criteria in drug trials accurately identify subjects with predementia Alzheimers disease? J Neurol Neurosurg Psychiatry 76: 13481354. 16. Solfrizzi V, Panza F, Colacicco AM, DIntrono A, Capurso C, et al. (2004) Vascular risk factors, incidence of MCI, and rates of progression to dementia. Neurology 63: 18821891. 17. Larrieu S, Letenneur L, Orgogozo JM, Fabrigoule C, Amieva H, et al. (2002) Incidence and outcome of mild cognitive impairment in a populationbased prospective cohort. Neurology 59: 15941599. 18. Kryscio RJ, Schmitt FA, Salazar JC, Mendiondo MS, Markesbery WR (2006) Risk factors for transition from normal to mild cognitive impairment and dementia. Neurology 66: 828832. 19. Jelic V, Kivipelto M, Winblad B (2006) Clinical trials in mild cognitive impairment: lessons for the future. J Neurol Neurosurg Psychiatry 77: 429 438. 20. Loy C, Schneider L (2006) Galantamine for Alzheimers disease and mild cognitive impairment. Cochrane Database Syst Rev 2006:CD001747. Available: http://mrw.interscience.wiley.com/cochrane/clsysrev/articles/ CD001747/frame.html. Accessed 31 October 2007. 21. Birks J, Flicker L (2006) Donepezil for mild cognitive impairment. Cochrane Database Syst Rev 2006 3:CD006104. Available: http://mrw. interscience.wiley.com/cochrane/clsysrev/articles/CD006104/frame.html. Accessed 31 October 2007. 22. US Food and Drug Administration (2005) Alert for healthcare professionals on galantamine hydrochloride (marketed as Razadyne formerly Reminyl) [FDA Alert]. Available: http://www.fda.gov/cder/drug/InfoSheets/HCP/ galantamineHCP.htm. Accessed 16 February 2006. 23. Gauthier S, Touchon J (2005) Mild cognitive impairment is not a clinical entity and should not be treated. Arch Neurol 62: 11641166.
63

24. Rosenberg PB, Johnston D, Lyketsos C (2006) A clinical approach to mild cognitive impairment. Am J Psychiatry 163: 18841890. 25. Petersen RC, Morris JC (2005) Mild cognitive impairment as a clinical entity and treatment target. Arch Neurol 62: 11601163. 26. Kirshner HS (2005) Mild cognitive impairment: to treat or not to treat. Curr Neurol Neurosci Rep 5: 455457. 27. Watkins PB, Zimmerman HJ, Knapp MJ, Gracon SI, Lewis KW (1994) Hepatotoxic effects of tacrine administration in patients with Alzheimers disease. JAMA 271: 992998. 28. Galisteo M, Rissel M, Sergent O, Chevanne M, Cillard J, et al. (2000) Hepatotoxicity of tacrine: occurrence of membrane uidity alterations without involvement of lipid peroxidation. J Pharmacol Exp Ther 294: 160 167. 29. Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, et al. (1996) Assessing the quality of reports of randomized clinical trials: is blinding necessary? Control Clin Trials 17: 112. 30. National Institute of Mental Health (2005) Treatment for early memory loss. Available: http://clinicaltrials.gov/ct/show/NCT00042172. Accessed 31 October 2007. 31. [No authors listed] (2006) A prospective, randomized, multicenter, doubleblind, placebo-controlled, parallel-group study of the effect of rivastigmine on the time to clinical diagnosis of Alzheimers disease in subjects with mild cognitive impairment (MCI). Available: http://www.novartisclinicaltrials. com/webapp/clinicaltrialrepository/displayFile.do?trialResult1886. Accessed 31 October 2007. 32. De Kosky S (2004) A randomized double-blind, placebo-controlled trial to evaluate the efcacy and safety of galantamine in subjects with mild cognitive impairment (MCI) clinically at risk for development of clinically probable Alzheimers disease. Protocol No. GAL-INT-11. Available: http:// www.clinicalstudyresults.org/documents/company-study_96_1.pdf. Accessed 31 October 2007. 33. Winblad B (2004) A randomized double-blind, placebo-controlled trial to evaluate the efcacy and safety of galantamine in subjects with mild cognitive impairment (MCI) clinically at risk for development of clinically probable Alzheimers disease. Protocol No. GAL-INT-18. Available: http:// www.clinicalstudyresults.org/documents/company-study_96_2.pdf. Accessed 31 October 2007. 34. Koontz J, Baskys A (2005) Effects of galantamine on working memory and global functioning in patients with mild cognitive impairment: a doubleblind placebo-controlled study. Am J Alzheimers Dis Other Demen 20: 295 302. Novartis (2006) Efcacy and safety of rivastigmine in patients with mild cognitive impairment. Available: http://clinicaltrials.gov/ct/show/ NCT00134953. Accessed 31 October 2007. Salloway S, Ferris S, Kluger A, Goldman R, Griesing T, et al. (2004) Efcacy of donepezil in mild cognitive impairment: a randomized placebocontrolled trial. Neurology 63: 651657. Petersen RC, Thomas RG, Grundman M, Bennett D, Doody R, et al. (2005) Vitamin E and donepezil for the treatment of mild cognitive impairment. N Engl J Med 352: 23792388. Sobow T, Ktoszewska I (2006) Cholinesterase inhibitors in mild cognitive impairment: a met-analysis of randomized controlled trials Neurol Neurochirurgia Polska 41: 1321. Emanuel EJ, Wendler D, Grady C (2000) What makes clinical research ethical? JAMA 283: 27012711. Karlawish J (2006) Alzheimers diseaseclinical trials and logic of clinical purpose. N Engl J Med 355: 16041605. Mayor S (2005) Regulatory authorities review use of galantamine in mild cognitive impairment. BMJ 330: 276. ADRAC (2006) Deaths with galantamine in mild cognitive impairment studies. Aust Adv Drug Reactions Bull 25: Available: http://www.tga.gov.au/ adr/aadrb/aadr0602.htm#a1. Accessed 31 October 2007. De Carli C, Frisoni GB, Clark CM, Harvey D, Grundman M, et al. (2007) Qualitative estimates of medial temporal atrophy as a predictor of progression from mild cognitive impairment to dementia. Arch Neurol 64: 108115. Frisoni GB, Canu E, Geroldi C, Zanetti O, Zacchi V (2006) Drug prescription in mild cognitive impairment: the physicians perspective in Italy. Int J Geriatr Psychiatry 21: 10711077. Moynihan R, Heath I, Henry D (2002) Selling sickness: the pharmaceutical industry and disease mongering. BMJ 324: 886890. Moynihan R, Henry D (2006) The ght against disease mongering: generating knowledge for action. PLoS Med 3: e191. doi:10.1371/journal. pmed.0030191 House of Commons Health Committee (2005) The inuence of the pharmaceutical industry. Fourth Report of Session 200405. Volume I. Available: http://www.publications.parliament.uk/pa/cm200405/cmselect/ cmhealth/42/42.pdf. Accessed 31 October 2007.
35. 36. 37. 38. 39. 40. 41. 42. 43.
44. 45. 46. 47.
64

Editors Summary
Background. Worldwide, more than 24 million people have dementia, a group of brain disorders characterized by an irreversible decline in memory, problem solving, communication, and other cognitive functions. The commonest form of dementia is Alzheimer disease (AD). The risk of developing AD increases with ageAD is rare in people younger than 65 but about half of people over 85 years old have it. The earliest symptom of AD is usually difficulty in remembering new information. As the disease progresses, patients may become confused and have problems expressing themselves. Their behavior and personality can also change. In advanced AD, patients need help with daily activities like dressing and eating, and eventually lose their ability to recognize relatives and to communicate. There is no cure for AD but a class of drugs called cholinesterase inhibitors can sometimes temporarily slow the worsening of symptoms. Three cholinesterase inhibitorsdonepezil, rivastigmine, and galantamineare currently approved for use in mild-to-moderate AD. Why Was This Study Done? Some experts have questioned the efficacy of cholinesterase inhibitors in AD, but other experts and patient support groups have called for these drugs to be given to patients with a condition called mild cognitive impairment (MCI) as well as to those with mild AD. People with MCI have memory problems that are more severe than those normally seen in people of their age but no other symptoms of dementia. They are thought to have an increased risk of developing AD, but it is not known whether everyone with MCI eventually develops AD, and there is no standardized way to diagnose MCI. Despite these uncertainties, several clinical trials have investigated whether cholinesterase inhibitors prevent progression from MCI to AD. In this study, the researchers have assessed whether the results of these trials provide any evidence that cholinesterase inhibitors can prevent MCI progressing to AD. What Did the Researchers Do and Find? The researchers conducted a systematic review of the medical literature to find trials that had addressed this issue, which met criteria that they had defined clearly in advance of their search. They identified three published and five unpublished randomized controlled trials (studies in which patients randomly receive the test drug or an inactive placebo) that investigated the effect of cholinesterase inhibitors on the progression of MCI. The researchers obtained the results of six of these trialsfour examined the effect of cholinesterase inhibitors on the conversion of MCI to clinically diagnosed AD or dementia (the primary end point); all six examined the effect of the drugs on several secondary end points (for example, individual aspects of cognitive function). None of the drugs produced a statistically significant difference (a difference that is unlikely to have happened by chance) in the probability of progression from MCI to AD. The only statistically significant secondary end point after adjustment for multiple comparisons (when many outcomes are considered, false positive results can occur unless specific mathematical techniques are used to prevent this problem) was a decrease in the rate of brain shrinkage associated with galantamine treatment. More patients treated with cholinesterase inhibitors dropped out of trials because of adverse effects than patients given placebo. Finally, in the one trial that reported all causes of deaths, one participant who received placebo and six who received galantamine died. What Do These Findings Mean? These findings suggest that the use of cholinesterase inhibitors is not associated with any delay in the onset of clinically diagnosed AD or dementia in people with MCI. They also show that the use of these drugs has no effect on most surrogate (substitute) indicators of AD but that the risks associated with their use are not negligible. However, because MCI has not yet been clearly defined as a clinical condition that precedes dementia, some (even many) of the patients enrolled into the trials that the researchers assessed may not actually have had MCI. Thus, further clinical trials are needed to clarify whether cholinesterase inhibitors can delay the progression of MCI to dementia, but these additional trials should not be done until the diagnosis of MCI has been standardized. Additional Information. Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed. 0040338. An essay by Matthews and colleagues, in the October 2007 issue of PLoS Medicine, discusses how mild cognitive impairment is currently diagnosed The US Alzheimers Association provides information about all aspects of Alzheimer disease, including fact sheets on treatments for Alzheimer disease and on mild cognitive impairment The UK Alzheimers Society provides information for patients and caregivers on all aspects of dementia, including drug treatments and mild cognitive impairment The UK charity DIPEx provides short video clips of personal experiences of care givers of people with dementia
65
Cholinesterase Inhibitors and Hospitalization for Bradycardia: A Population-Based Study

Laura Y. Park-Wyllie1,2,3*, Muhammad M. Mamdani2,3,4,5, Ping Li3, Sudeep S. Gill3,6, Andreas Laupacis2,3,5,7, David N. Juurlink3,5,7,8
1 Department of Family and Community Medicine, St. Michaels Hospital, Toronto, Ontario, Canada, 2 Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michaels Hospital, Toronto, Ontario, Canada, 3 Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada, 4 Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada, 5 Department of Health Policy, Management and Evaluation, University of Toronto, Toronto, Ontario, Canada, 6 Department of Medicine at Queens University, Kingston, Ontario, Canada, 7 Department of Medicine, University of Toronto, Toronto, Ontario, Canada, 8 Department of Medicine, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada
Abstract
Background: Cholinesterase inhibitors are commonly used to treat dementia. These drugs enhance the effects of acetylcholine, and reports suggest they may precipitate bradycardia in some patients. We aimed to examine the association between use of cholinesterase inhibitors and hospitalization for bradycardia. Methods and Findings: We examined the health care records of more than 1.4 million older adults using a case-timecontrol design, allowing each individual to serve as his or her own control. Case patients were residents of Ontario, Canada, aged 67 y or older hospitalized for bradycardia between January 1, 2003 and March 31, 2008. Control patients (3:1) were not hospitalized for bradycardia, and were matched to the corresponding case on age, sex, and a disease risk index. All patients had received cholinesterase inhibitor therapy in the 9 mo preceding the index hospitalization. We identified 1,009 community-dwelling older persons hospitalized for bradycardia within 9 mo of using a cholinesterase inhibitor. Of these, 161 cases informed the matched analysis of discordant pairs. Of these, 17 (11%) required a pacemaker during hospitalization, and six (4%) died prior to discharge. After adjusting for temporal changes in drug utilization, hospitalization for bradycardia was associated with recent initiation of a cholinesterase inhibitor (adjusted odds ratio [OR] 2.13, 95% confidence interval [CI] 1.293.51). The risk was similar among individuals with pre-existing cardiac disease (adjusted OR 2.25, 95% CI 1.184.28) and those receiving negative chronotropic drugs (adjusted OR 2.34, 95% CI 1.164.71). We found no such association when we replicated the analysis using proton pump inhibitors as a neutral exposure. Despite hospitalization for bradycardia, more than half of the patients (78 of 138 cases [57%]) who survived to discharge subsequently resumed cholinesterase inhibitor therapy. Conclusions: Among older patients, initiation of cholinesterase inhibitor therapy was associated with a more than doubling of the risk of hospitalization for bradycardia. Resumption of therapy following discharge was common, suggesting that the cardiovascular toxicity of cholinesterase inhibitors is underappreciated by clinicians.
Please see later in the article for the Editors Summary.
Citation: Park-Wyllie LY, Mamdani MM, Li P, Gill SS, Laupacis A, et al. (2009) Cholinesterase Inhibitors and Hospitalization for Bradycardia: A Population-Based Study. PLoS Med 6(9): e1000157. doi:10.1371/journal.pmed.1000157 Academic Editor: Carol Brayne, University of Cambridge, United Kingdom Received January 14, 2009; Accepted August 21, 2009; Published September 29, 2009 Copyright: 2009 Park-Wyllie et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: LYPW was supported by a fellowship from the Canadian Institutes for Health Research. SSG was supported by a Career Scientist Award from the Ontario Ministry of Health and Long-Term Care. DNJ was supported by a New Investigator Award from the Canadian Institutes for Health Research. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: MMM was employed at Pfizer Global Pharmaceuticals from January 2006 to April 2007. Abbreviations: CI, confidence interval; CIHI, Canadian Institute for Health Information; OR, odds ratio. * E-mail: parkwylliel@smh.toronto.on.ca
66
PLoS Medicine | www.plosmedicine.org 1 September 2009 | Volume 6 | Issue 9 | e1000157
CholinesteraseInhibitorsandHospitalizationforBradycardia:APopulation-BasedStudy

Cholinesterase Inhibitors and Bradycardia
Introduction
Cholinesterase inhibitors such as donepezil, rivastigmine, and galantamine are widely prescribed to improve cognitive function in patients with Alzheimer diseasea condition expected to quadruple in prevalence over the next 50 y [1]. By inhibiting the synaptic metabolism of acetylcholine, these drugs enhance cortical cholinergic neurotransmission [2]. Although cholinesterase inhibitors are generally well tolerated, they may provoke adverse effects in some patients because their cholinergic effects are not confined to the central nervous system [2]. Symptoms of cholinergic excess are often nonspecific and include gastrointestinal upset, diarrhea, hypersalivation, and muscle cramps. In severe instances, these drugs can increase vagal tone and thereby precipitate bradycardia. Anecdotal reports, small observational studies, and post hoc analyses of clinical trials have produced conflicting results, with some suggesting an increased risk of bradycardia during cholinesterase inhibitor therapy and others finding no such association [312]. At the time of our study, no large-scale studies had examined, to our knowledge, whether cholinesterase inhibitor use among older patients predisposes to bradycardia. Frail older adults represent a growing population of cholinesterase inhibitor users. These patients are more prone to the adverse effects of drugs and discontinue cholinesterase inhibitors more often than patients in clinical trials, who are typically healthier than those in clinical practice [10]. We sought to characterize the association between cholinesterase inhibitor therapy and hospitalization for bradycardia in a population of more than 1.4 million older adults.
(the reference interval). Because cases serve as their own controls, fixed patient characteristics are controlled for implicitly under this design [24,25]. However, the case-crossover design can be vulnerable to spurious associations between a drug and an outcome owing to temporal trends in drug utilization. The casetime-control design corrects for this limitation by incorporating a control group of patients who did not experience the outcome of interest [22,2628].
Identification of Case Patients

We included all patients aged 67 y and older hospitalized with a diagnosis of bradycardia between January 1, 2003 and March 31, 2008, and restricted our analysis to those patients who were exposed to a cholinesterase inhibitor in the 9 mo prior to the index date. Because we hypothesized that bradycardia caused by cholinesterase inhibitors would be most likely to manifest during the initial period of therapy, we defined our risk interval as the 3mo period immediately preceding hospitalization, and our reference interval as the months seven through nine prior to the index date (Figure 1). We included a 3-mo wash-out interval between the risk and reference intervals to avoid contamination between the risk and reference intervals, and excluded individuals with pacemaker insertion in the previous 5 y or hospitalization in the year preceding the study entry. Individuals with cholinesterase inhibitor prescription in the wash-out period, or both the risk and reference periods, did not contribute to the analysis. Hospitalizations included emergency department visits and hospital admissions for bradycardia, and were identified using the International Classification of Diseases and Related Health Problems Tenth Revision (ICD-10) [29] code for bradycardia (R001). All hospital visits associated with a diagnosis of bradycardia were included in this study because the CIHI Discharge Abstract Database does not contain direct information on the primary reason for hospital admission. The date of the hospitalization served as the index date, and only the first hospitalization for bradycardia was considered for patients with multiple such admissions during the study period.
Methods Setting and Data Sources

We linked multiple population-based health care databases in an anonymous fashion using unique encrypted health card numbers. This linkage process has been standardized by our research institution (http://www.ices.on.ca), and these methods have been used extensively to study population-based health outcomes, including adverse drug events [1320]. The Ontario Drug Benefit database was used to identify prescription records, and contains comprehensive, high-quality information regarding prescription medications dispensed to Ontario residents aged 65 y and older [21]. The Canadian Institute for Health Information (CIHI) Discharge Abstract Database was used to identify hospital admissions, and contains detailed diagnostic and procedural information for all hospital admissions in Ontario. The National Ambulatory Care Reporting System was used to identify visits to emergency departments. Basic demographic information was obtained from the Ontario Registered Persons Database. Finally, we used the Ontario Health Insurance Plan database to identify claims for inpatient and outpatient physician services. All Ontario seniors receive universal access to hospital care, physicians services, and prescription drug coverage. The study was approved by the Research Ethics Board of Sunnybrook Health Sciences Centre.
Identification of Control Patients

In keeping with the case-time-control design, we corrected for temporal changes in cholinesterase inhibitor use by matching each case with up to three control patients. Control patients did not experience a hospitalization for bradycardia on or before the index date, but did receive at least one prescription for a cholinesterase inhibitor in either of the corresponding risk or reference intervals preceding the index date (Figure 1). To minimize differences between case and control patients, we selected controls matched on age (born within 1 y of case), sex, and their anticipated risk of bradycardia using a disease risk index, as done previously [30,31]. The disease risk index was derived by constructing a multivariable regression model that included multiple potential predictors of bradycardia or death, including socioeconomic status, residence in long-term care facility, overall number of prescription drugs prescribed in the preceding year, beta-blockers, calcium channel blockers, digoxin, antiarrhythmics, nitrates, anticoagulants, antiplatelets, diuretics, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, HMGCoA reductase inhibitors, fibric acid derivatives, ezetimibe, oral hypoglycemic agents, insulin, antipsychotic medications, antidepressants, sedative-hypnotics, chemotherapy, corticosteroids, overall number physician clinic visits, emergency department visits, cardiologist visits, internist visits, neurologist visits, geriatrician visits, psychiatrist visits, coronary artery bypass graft, angiography,
2 September 2009 | Volume 6 | Issue 9 | e1000157
Study Design
We used the case-time-control design to examine the association between cholinesterase inhibitor use and hospitalization for bradycardia among Ontario residents aged 67 y and older. This design is an extension of the case-crossover design first described by Maclure [22,23], which compares within-patient exposure to a potential risk factor in the period immediately preceding a putative adverse event (the risk interval) to exposure during a different time
67
Figure 1. Case-time-control design. doi:10.1371/journal.pmed.1000157.g001
percutaneous transluminal coronary angioplasty, valve surgery, carotid endarterectomy, peripheral vascular disease procedures, dialysis, echocardiography, electrocardiography, holter monitor, nuclear medicine stress test, carotid doppler ultrasonography, charlson comorbidity index score, renal dysfunction, liver dysfunction, heart failure, diabetes, cancer, cerebrovascular disease (strokes, transient ischemic attacks), cardiac dysrhythmias, myocardial infarction, angina and coronary artery disease, peripheral vascular disease, major infections (respiratory, urogenital, abdominal, gastrointestinal, skin, soft tissue), and alcoholism (Text S1). These potential confounders were consequently summarized into a single disease risk index that predicted the probability of hospital admission for bradycardia [30,32]. We selected up to three controls with the disease risk scores (within 0.2 standard deviation) closest to the given case.
Statistical Analysis
We derived the case-time-control odds ratio [OR] by dividing the crossover OR among the cases (i.e., the case-crossover group) by the crossover OR among the controls (i.e., the control-crossover group), thereby producing a case-control OR adjusted for timetrend (Figure 1). Crossover ORs were derived from the ratio of
68
discordant pairs, i.e., the number of individuals exposed exclusively during the risk interval as compared to exclusively during the reference interval [27]. By assuming conditional independence of exposure within each 1:3 matched set, a conditional logistic regression model, adjusting for the length of stay in hospital in the year preceding study entry as an additional measure of comorbidity, was fitted to estimate the overall OR and 95% confidence intervals (CIs) [27]. We hypothesized that individuals with pre-existing cardiovascular disease and individuals co-using negative chronotropic agents might be at particularly high risk for bradycardia. To examine these risk groups, we performed a stratified analysis in those with a history of cardiovascular disease (defined as previous myocardial infarction, congestive heart failure, angina, or arrhythmias), and those coprescribed negative chronotropic medications such as beta-adrenergic antagonists, digoxin, or the nondihydropyridine calcium channel antagonists verapamil and diltiazem. Drug interactions with cholinesterase inhibitors can occur via the cytochrome P450 enzyme 2D6. However, given the relatively few P450 2D6 inhibitors, we felt it was more relevant to focus on the pharmacodynamic interaction between cholinesterase inhibitors and negative chronotropic agents.
September 2009 | Volume 6 | Issue 9 | e1000157

To test the specificity of our findings, we performed a sensitivity analysis in which proton pump inhibitors served as the exposure of interest rather than cholinesterase inhibitors. All analyses used a two-sided type I error rate of 0.05 and were performed using SAS version 9.1 (SAS Institute).
Results
Between January 1, 2003 and March 31, 2008, we identified 27,333 hospitalizations for bradycardia among Ontario residents 67 y and older. Of these, 10,323 were excluded because they were hospitalized in the year prior to the index date, and 15,805 were excluded because they had not used cholinesterase inhibitors in the 9 mo prior to index date (Figure 2). Among the remaining patients, we further excluded 191 individuals exposed to cholinesterase inhibitors during the wash-out interval, and patients who had a pacemaker or could not be matched to at least one control (n#5), leaving 1,009 eligible cases. Among these cases, 848 (84%) received a cholinesterase inhibitor in both the risk and reference periods, leaving 161 cases to inform our matched pairs analysis of individuals who had received a cholinesterase inhibitor in either the risk or reference period, but not both. Of these cases, 148 (92%) were fully matched to three controls and 157 (98%)
were matched to at least two controls. We identified 466 matched controls from 42,833 potential controls. The characteristics of cases and controls were highly similar (Table 1). The mean age of patients was 83 y (standard deviation 5.4 y) and 320 (51%) were female. A higher proportion of controls were in long-term care facilities, although the overall proportion remained low in both groups. Donepezil was the most frequently prescribed cholinesterase inhibitor in these patients accounting for 117 of the 161 (73%) cases and 292 of the 466 (63%) controls. Seventeen patients (11%) received a pacemaker during their hospitalization, and six (4%) individuals died prior to discharge from hospital. In the primary analysis, recent initiation of cholinesterase inhibitors was significantly associated with hospitalization for bradycardia (adjusted OR 2.13, 95% CI 1.293.51, p = 0.003; Table 2). Among cases and controls with previously diagnosed cardiac disease, the association between recent use of cholinesterase inhibitors and bradycardia was similar (adjusted OR 2.25, 95% CI 1.184.28, p = 0.014). The association persisted among the cases and controls receiving negative chronotropic medications (adjusted OR 2.34; 95% CI 1.164.71, p = 0.017). As expected, we found no association between recent initiation of proton pump inhibitors and bradycardia (adjusted OR 1.13, 95% CI 0.931.37, p = 0.228).
Figure 2. Flow diagram of case selection. doi:10.1371/journal.pmed.1000157.g002
69
Table 1. Characteristics of cases and controls.
Table 1. Cont.
Characteristic Age, mean (SD), y Female Low income Charlson Index, mean score (SD) Long-term care, preceding year Drug therapy, preceding year ACE inhibitor Angiotensin receptor antagonist Antiarrhythmic Anticoagulant Antidepressant Antiplatelet Antipsychotic Beta-adrenergic antagonist Calcium channel antagonist Corticosteroid Digoxin Diuretic Fibric acid derivative Oral glucose lowering drug Insulin Nitrate Sedative hypnotic Statin Total n drugs prescribed in preceding year, median (IQR)
Cases (n = 161) 82.8165.41 82 (50.9%) 42 (26.1%) 0.7661.27 8 (5.0%)
Controls (n = 466) 82.7165.37 238 (51.1%) 124 (26.6%) 0.6761.19 46 (9.9%)
Characteristic Medical conditions, preceding 5 y Heart failure Myocardial infarction Peripheral vascular disease Alcoholism Angina
Cases (n = 161)
Controls (n = 466)
28 (17.4%) 74 (46.0%) #5 #5 49 (30.4%) 15 (9.3%) 42 (26.1%) #5 57 (35.4%) 41 (25.5%)
90 (19.3%) 230 (49.4%) #5 19 (4.1%) 160 (34.3%) 31 (6.7%) 109 (23.4%) 11 (2.4%) 146 (31.3%) 112 (24.0%)
58 (36.0%) 18 (11.2%) #5 27 (16.8%) 33 (20.5%) 28 (17.4%) 9 (5.6%) 53 (32.9%) 47 (29.2%) #5 25 (15.5%) 57 (35.4%) #5 21 (13.0%) #5 18 (11.2%) 22 (13.7%) 41 (25.5%) 8 (512)
176 (37.8%) 61 (13.1%) 18 (3.9%) 70 (15.0%) 93 (20.0%) 92 (19.7%) 21 (4.5%) 149 (32.0%) 145 (31.1%) #5 63 (13.5%) 156 (33.5%) 8 (1.7%) 52 (11.2%) 12 (2.6%) 55 (11.8%) 72 (15.5%) 123 (26.4%) 8 (512)
Arrhythmia Diabetes Liver disease Renal dysfunction Stroke Medical procedures, preceding 5 y Coronary artery bypass graft Percutaneous transluminal coronary angioplasty Peripheral vascular disease Angiography/cardiac catheterization Carotid doppler ultrasound Carotid endarterectomy Dialysis Echocardiogram Electrocardiography Holter monitoring Stress and nuclear tests Valve surgery
#5 7 (4.3%) 0 (0.0%) 9 (5.6%) 35 (21.7%) #5 0 (0.0%) 70 (43.5%) 144 (89.4%) 52 (32.3%) 50 (31.1%) 0 (0.0%)
7 (1.5%) 19 (4.1%) #5 23 (4.9%) 107 (23.0%) #5 0 (0.0%) 212 (45.5%) 418 (89.7%) 141 (30.3%) 147 (31.5%) 0 (0.0%)
Total n physician visits in preceding year Cardiologist Mean6SD Median (IQR) Family physician Mean6SD Median (IQR) Geriatrician Mean6SD Median (IQR) Internist Mean6SD Median (IQR) Neurologist Mean6SD Median (IQR) Psychiatrist Mean6SD Median (IQR) Total n emergency department visits, preceding year Total n medical visits, preceding year Total n days stayed in hospital, preceding year 0.3663.19 0 (00) 0.7361.65 22.81619.62 2.0266.27 0.4562.90 0 (00) 0.7761.71 22.19617.81 1.9567.29 0.0860.33 0 (00) 0.0860.36 0 (00) 2.2064.71 1 (02) 1.9664.90 1 (02) 0.1760.84 0 (00) 0.1760.68 0 (00) 13.88613.88 10 (518) 13.75612.07 11 (618) 0.5261.06 0 (01) 0.4461.51 0 (00)
All data presented as number (percentages) except where indicated. Cells #5 are suppressed. ACE, angiotensin-converting enzyme; IQR, interquartile range; SD, standard deviation. doi:10.1371/journal.pmed.1000157.t001
Secondary Analyses
Because bradycardia is a relatively common occurrence in older patients, we hypothesized that the potential contribution of cholinesterase inhibitors to the development of bradycardia might not be recognized, and that therapy might be continued following discharge. After excluding individuals who received pacemakers during their hospital stay, 138 cases survived to discharge. Of these cases, subsequent prescription records indicated that cholinesterase inhibitors were restarted in 78 individuals (57%) within 100 d of hospital discharge. A post hoc examination of the 3-mo period following resumption of therapy in those 78 individuals showed that three (3.8%) individuals were readmitted to hospital or visited the emergency department with a diagnosis of bradycardia.
Discussion
Using the health care records of more than 1.4 million Ontario residents aged 67 y and older, we found that treatment with cholinesterase inhibitors was associated with a doubling in the risk of hospitalization for bradycardia. Importantly, although cholinesterase inhibitors are reversible precipitants of bradycardia, the drugs were resumed following discharge in greater than half the cases, presumably because the potential causative role of these
70

Table 2. Risk of bradycardia-related hospital admissions and recent cholinesterase inhibitor use.
Adjusted ORa (95% CI)
Overall and Subgroup Analyses Full population Overall Cases (n = 161) Control (n = 466) I. Subgroup with cardiac comorbidity Overall Cases (n = 97) Control (n = 274) II. Subgroup using negative chronotropes Overall Cases (n = 80) Control (n = 220)
Exposure in Risk Interval
Exposure in Reference Interval
2.13 (1.293.51) 139 349 22 117 p = 0.003
2.25 (1.184.28) 84 202 13 72 p = 0.014
2.34 (1.164.71) 69 158 11 62 p = 0.017
a Case-time-control OR adjusted for hospitalization length of stay in the preceding year. doi:10.1371/journal.pmed.1000157.t002
drugs in the hospitalization was not appreciated. In many patients, cholinesterase inhibitors are associated with marginal improvement in cognition and global functioning [1,33]. Consequently, recent guidelines suggest that cholinesterase inhibitors should not be standard of care for patients with dementia, but instead urge physicians to weigh each individuals expected risks and benefits before initiating therapy. Our large-scale population-based assessment of the risk of bradycardia with cholinesterase inhibitors in clinical practice should help inform the risk-benefit assessment for clinicians and patients. A recent study used an alternate cohort design to examine a primary outcome of syncope in patients receiving cholinesterase inhibitors. In a secondary analysis, these investigators also observed an elevated risk of bradycardia (adjusted hazard ratio 1.69; 95% CI 1.322.15) associated with cholinesterase inhibitor use, which complements the findings of our study [34]. The risk of bradycardia observed in our study may cause clinicians and patients to reconsider therapy with these drugs, particularly for patients in whom little or no cognitive improvement is observed early in therapy [35,36]. At a minimum, our findings should alert clinicians to the potential role of cholinesterase inhibitors in patients with bradycardia, for whom resumption of treatment may be inadvisable. Our study has several limitations that merit emphasis. Administrative databases contain information on hospitalizations, emergency department visits, diagnoses, procedures, physicians claims, outpatient clinic visits, and drug dispensing data. However, they lack the clinical richness of a medical chart. Accordingly, we were not able to capture the severity of dementia, lifestyle habits such as smoking, diet, alcohol use, exercise, over-the-counter drug use, in-hospital drug use, laboratory values, and the results of diagnostic testing. However, we relied on the unique crossover design feature that controls for fixed patient characteristics by allowing each patient to serve as his or her own control. Consequently, concerns regarding potential confounders, including disease severity, were largely overcome. The crossover design also relies on the temporal association between an exposure and outcome, and this feature enabled us to examine whether bradycardia-related hospitalization was likely secondary to cholinesterase inhibitors as opposed to underlying comorbidities. More specifically, the crossover design examines whether an exposure is more likely to occur immediately
preceding an event or during another period when bradycardia did not occur. If the exposure is not associated with the outcome, then no temporal association would be expected. The presence of underlying comorbidities would not be expected to confound the cholinesterase inhibitorbradycardia relationship unless the comorbidities were associated with both cholinesterase inhibitor use and bradycardia. To further control for confounding, we corrected for temporal changes in cholinesterase inhibitor use by matching each case with up to three control patients on the basis of an extensive disease risk index score. We identified episodes of bradycardia resulting in emergency department visits or hospital admissions, but we were unable to identify individuals in whom bradycardia did not culminate in hospital care, including cases in which bradycardia led to death [37]. Therefore, our analysis likely underestimates the true risk of cardiovascular harm associated with cholinesterase inhibitors. The coding for bradycardia has not been validated and it is possible that some cases of bradycardia were missed if the code for bradycardia has low sensitivity. However, we expect that the positive predictive value for the bradycardia code would be reasonable because the diagnosis for bradycardia is based upon a fairly straightforward medical assessment. The occurrence of random miscoding would only have attenuated our estimates and biased the results towards the null. It is likely that the cases of bradycardia captured in our study were clinically significant because they were severe enough to be documented in the patients chart and coded in the CIHI database. We were unable to assess the absolute risk of bradycardia due to cholinesterase inhibitor therapy, and we could not be certain that resumption of cholinesterase inhibitors following hospital discharge truly reflected a lack of appreciation for the potential negative chronotropic effects of therapy. Since most patients were taking donepezil, we were not able to contrast risks with individual cholinesterase inhibitors given the low prevalence of use with the other agents. Future studies are needed to address the relative harms of the individual drugs in this class. Our post hoc examination of the 78 patients who resumed cholinesterase inhibitor after hospital discharge showed that only 4% were readmitted to hospital or visited the emergency department with a diagnosis of bradycardia in the 3 mo following resumption of therapy. It is hard to know how to interpret this post hoc analysis. We did not evaluate out-of-hospital death, and could
71
not ascertain whether cholinesterase inhibitor therapy was restarted with a reduced dose or more gradual dose titration, whether changes to other medications were made, or whether closer outpatient follow up for bradycardia was undertaken. Finally, because we relied on drug dispensing data as a proxy for drug adherence, there was a potential for exposure misclassification. While bias and confounding can threaten any observational study, the likelihood of these was reduced substantially by the design of our study, in which every case served as his or her own control, thereby lessening between-patient variability, and by the use of a disease risk index, an advanced matching technique, which resulted in considerable similarity between cases and controls (Table 1) [31,32]. In summary, we found that cholinesterase inhibitors are associated with a significantly increased risk of hospitalization for bradycardia among older outpatients, and that the risk is similar in patients with cardiovascular comorbidity and those receiving concurrent therapy with negative chronotropic drugs. Our findings highlight the importance of careful clinical evaluation prior to initiating cholinesterase inhibitor therapy, and vigilant monitoring thereafter in order to prevent adverse cardiac events. Cholinesterase inhibitors should be prescribed judiciously, and because these drugs carry a risk of serious adverse events, they should be continued only if the benefits outweigh the risks. Finally, it is important for clinicians to be aware of the potential negative chronotropic effects of cholinesterase inhibitors, and reassess the merits of continued therapy in patients who develop bradycardia while taking these drugs. The frequent resumption of cholinester-
ase inhibitors following discharge suggests that bradycardia may not be widely recognized as a potential adverse effect of this class of medications.
Text S1 Variables used in the Disease Risk Index. Found at: doi:10.1371/journal.pmed.1000157.s001 (0.06 MB DOC)
Acknowledgments
We thank Harindra Wijeysundera for help with pacemaker code selection. This project was supported by the Ontario Ministry of Health and Long Term Care, which had no role in the design, analysis, writing, or interpretation of the study. The opinions, results and conclusions are those of the authors, and no endorsement by the Ministry of Health and LongTerm Care or by the Institute for Clinical Evaluative Sciences is intended or should be inferred.
ICMJE criteria for authorship read and met: LPW MMM PL SSG AL DNJ. Agree with the manuscripts results and conclusions: LPW MMM PL SSG AL DNJ. Designed the experiments/the study: LPW MMM SSG AL DNJ. Analyzed the data: LPW PL SSG. Wrote the first draft of the paper: LPW. Contributed to the writing of the paper: LPW MMM PL SSG AL DNJ. Helped to design the study, helped to analyze/interpret the data, and contributed to critical revisions of the manuscript: SSG. Supervised LYPW: DNJ.
References
1. Qaseem A, Snow V, Cross JT, Jr, Forciea MA, Hopkins R, Jr, et al. (2008) Current pharmacologic treatment of dementia: a clinical practice guideline from the American College of Physicians and the American Academy of Family Physicians. Ann Intern Med 148: 370378. 2. Masuda Y (2004) Cardiac effect of cholinesterase inhibitors used in Alzheimers diseasefrom basic research to bedside. Curr Alzheimer Res 1: 315321. 3. Calvo-Romero JM, Ramos-Salado JL (1999) [Symptomatic sinus bradycardia associated with donepezil]. Revista de Neurologia 28: 10701072. 4. Suleyman T, Tevfik P, Abdulkadir G, Ozlem S (2006) Complete atrioventricular block and ventricular tachyarrhythmia associated with donepezil. Emerg Med J 23: 641642. 5. Bordier P, Garrigue S, Barold SS, Bressolles N, Lanusse S, et al. (2003) Significance of syncope in patients with Alzheimers disease treated with cholinesterase inhibitors. Europace 5: 429431. 6. Newby VJ, Kenny RA, McKeith IG (2004) Donepezil and cardiac syncope: a case report. Int J Geriatr Psychiatry 19: 11101112. 7. Tavassoli N, Sommet A, Lapeyre-Mestre M, Bagheri H, Montrastruc J (2007) Drug interactions with cholinesterase inhibitors: an analysis of the french pharmacovigilance database and a comparison of two national drug formularies. Drug Safety 30: 10631071. 8. Roisin T, Van Ermen A, Rogiers A (2006) Cardiac adverse effects with cholinesterase inhibitors. Drug Safety 29: 930. 9. Shepherd G, Klein-Schwartz W, Edwards R (1999) Donepezil overdose: a tenfold dosing error. Ann Pharmacother 33: 812815. 10. Gill SS, Bronskill SE, Mamdani M, Sykora K, Li P, et al. (2004) Representation of patients with dementia in clinical trials of donepezil. Can J Clin Pharmacol 11: e274e285. 11. Bordier P, Garrigue S, Lanusse S, Margaine J, Robert F, et al. (2006) Cardiovascular effects and risk of syncope related to donepezil in patients with Alzheimers disease. CNS Drugs 20: 411417. 12. Morganroth J, Graham S, Hartman R, Anand R (2002) Electrocardiographic effects of rivastigmine. J Clin Pharmacol 42: 558568. 13. Juurlink DN, Mamdani MM, Lee DS, Kopp A, Austin PC, et al. (2004) Rates of hyperkalemia after publication of the Randomized Aldactone Evaluation Study. N Engl J Med 351: 543551. 14. Mamdani M, Juurlink DN, Lee DS, Rochon PA, Kopp A, et al. (2004) Cyclooxygenase-2 inhibitors versus non-selective non-steroidal anti-inflammatory drugs and congestive heart failure outcomes in elderly patients: a populationbased cohort study. Lancet 363: 17511756. 15. Ray JG, Mamdani M, Tsuyuki RT, Anderson DR, Yeo EL, et al. (2001) Use of statins and the subsequent development of deep vein thrombosis. Arch Intern Med 161: 14051410. 16. Lee DS, Tu JV, Juurlink DN, Alter DA, Ko DT, et al. (2005) Risk-treatment mismatch in the pharmacotherapy of heart failure. JAMA 294: 12401247. 17. Lipscombe LL, Gomes T, Levesque LE, Hux JE, Juurlink DN, et al. (2007) Thiazolidinediones and cardiovascular outcomes in older patients with diabetes. JAMA 298: 26342643. 18. Juurlink DN, Mamdani M, Kopp A, Laupacis A, Redelmeier DA (2003) Drugdrug interactions among elderly patients hospitalized for drug toxicity. JAMA 289: 16521658. 19. Mamdani MM, Tu K, van Walraven C, Austin PC, Naylor CD (2000) Postmenopausal estrogen replacement therapy and increased rates of cholecystectomy and appendectomy. CMAJ 162: 14211424. 20. Park-Wyllie LY, Juurlink DN, Kopp A, Shah BR, Stukel TA, et al. (2006) Outpatient gatifloxacin therapy and dysglycemia in older adults. N Engl J Med 354: 13521361. 21. Levy AR, OBrien BJ, Sellors C, Grootendorst P, Willison D (2003) Coding accuracy of administrative drug claims in the Ontario Drug Benefit database. Can J Clin Pharmacol 10: 6771. 22. Maclure M (1991) The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 133: 144153. 23. Maclure M, Mittleman MA (2000) Should we use a case-crossover design? Annual Review of Public Health 21: 193221. 24. Donnan PT, Wang J (2001) The case-crossover and case-time-control designs in pharmacoepidemiology. Pharmacoepidemiol Drug Saf 10: 259262. 25. Schneeweiss S, Sturmer T, Maclure M (1997) Case-crossover and case-timecontrol designs as alternatives in pharmacoepidemiologic research. Pharmacoepidemiol Drug Saf 6 (Suppl 3): S51S59. 26. Suissa S (1998) The case-time-control design: further assumptions and conditions. Epidemiology 9: 441445. 27. Suissa S (1995) The case-time-control design. Epidemiology 6: 248253. 28. Greenland S (1996) Confounding and exposure trends in case-crossover and case-time-control designs. Epidemiology 7: 231239. 29. Canadian Institute for Health Information (2003) The International Statistical Classification of Diseases and Related Health Problems - Tenth Revision. Ottawa, Canada: Canadian Institute for Health Information. 30. Juurlink DN, Mamdani MM, Kopp A, Redelmeier DA (2006) The risk of suicide with selective serotonin reuptake inhibitors in the elderly. Am J Psychiatry 163: 813821. 31. Arbogast PG, Ray WA (2009) Use of disease risk scores in pharmacoepidemiologic studies. Stat Methods Med Res 18: 6780. 32. Arbogast PG, Kaltenbach L, Ding H, Ray WA (2008) Adjustment for multiple cardiovascular risk factors using a summary risk score. Epidemiology 19: 30 37.
72

33. Raina P, Santaguida P, Ismaila A, Patterson C, Cowan D, et al. (2008) Effectiveness of cholinesterase inhibitors and memantine for treating dementia: evidence review for a clinical practice guideline. Ann Intern Med 148: 379397. 34. Gill SS, Anderson GM, Fischer HD, Bell CM, Li P, et al. (2009) Syncope and its consequences in patients with dementia receiving cholinesterase inhibitors: a population-based cohort study. Arch Intern Med 169: 867873.
35. Winblad B, Gauthier S, Scinto L, Feldman H, Wilcock GK, et al. (2008) Safety and efficacy of galantamine in subjects with mild cognitive impairment. Neurology 70: 20242035. 36. US FDA (2005) Possible increased mortality in patients with mild cognitive impairment. Dear Healthcare Professional Letter: www.fda.gov/medwatch/ SAFETY/2005/reminylDDLmarch.pdf. 37. Greenberg HM (1984) Bradycardia at onset of sudden death: potential mechanisms. Ann N Y Acad Sci 427: 241252.
73
Editors Summary
Background. Alzheimer disease and other forms of dementia principally affect people aged over 65. These conditions result in confusion, long term memory loss, irritability, and mood swings. As the population of developed countries ages, the prevalence of dementia is expected to increase significantly. It is forecast that the proportion of people with dementia in the US will quadruple by 2045. A common treatment for Alzheimer disease is a class of drug called an acetylcholinesterase inhibitor or cholinesterase inhibitor. These include donepezil (brand name Aricept), rivastigmine (marketed as Exelon and Exelon Patch), and galantamine (branded Razadyne). The benefit of taking cholinesterase inhibitors is generally small and they cannot reverse the effects of dementia. In about 50% of patients they delay the worsening of symptoms for between six months and a year, although a small number of patients may benefit more. They can have unpleasant side effects, which may include diarrhoea and muscle cramps. Why Was This Study Done? Existing evidence is inconclusive on whether cholinesterase inhibitors increase the risk of bradycardia, an abnormally slow resting heart rate of below 60 beats a minute, which can cause fatigue, dizziness, fainting, palpitations, shortness of breath, or death. In this paper, the authors use routinely collected health care data to investigate whether an older person taking a cholinesterase inhibitor is at increased risk of bradycardia. What Did the Researchers Do and Find? They began by supposing that cholinesterase inhibitors might induce bradycardia soon after a patient first began to take them. To investigate this, they obtained health care data on 1.4 million patients aged 67 or over in Ontario, Canada. They identified 161 patients who had visited a hospital for bradycardia and who had previously taken a cholinesterase inhibitor only within specific periods of time. They found that 139 had taken a cholinesterase inhibitor within the previous three months compared with 22 who had stopped taking it at least six months before. They compared these cases with up to three control patients who matched each of the initial case group of 161 patients by age, sex, and risk of bradycardia on the basis of their general health. None of the 466 controls had visited a hospital for bradycardia by the index date, that is, the date of hospitalization of the case patient they matched. The researchers found 349 of the control patients had begun to take a cholinesterase inhibitor in the three months prior to the index date, compared with 117 who had stopped taking it at least six months before. A statistical analysis of these data showed that recent initiation of cholinesterase inhibitors was associated with approximately a doubling of the risk of hospitalization for bradycardia. The authors repeated their procedure to see whether another class of drug, proton pump inhibitors, had a similar effect. As they had expected, it did not. They repeated the analysis for patients taking into account other drugs that slow the heart rate and found that their increased risk of bradycardia when taking a cholinesterase inhibitor persisted. The increase in risk was also similar in patients with preexisting heart problems. The researchers data also showed that, excluding patients who while in the hospital had a pacemaker fitted to control their heart rate, over half of the patients released from hospital started taking a cholinesterase inhibitor again. Of these, a few returned to hospital with bradycardia within 100 days. What Do These Findings Mean? Recent guidelines suggest that doctors should not prescribe cholinesterase inhibitors for dementia patients as a matter of course, but weigh the potential risks and benefits. This paper provides evidence of an additional risk, of which at least some doctors are unaware. It was not possible to compare risk for different cholinesterase inhibitors because most patients took donepezil. A population-based study like this cannot prove that cholinesterase inhibitors cause bradycardia. The authors used routinely collected data and so did not have information on all relevant risk factors, and thus there remains a possibility of bias due to unmeasured factors. In addition the authors had to make assumptions, for instance that patients took the drugs prescribed for them. They also considered only diagnoses of bradycardia made by a hospital doctor and not those made elsewhere, which means the incidence of bradycardia may have been underestimated. A strength of the study is the use of a case-time-control design, which has the advantage of reducing bias due to the different health conditions and lifestyle of individual patients, and also bias due to factors changing over time. Additional Information. Please access these Web sites via the online version of this summary at http://dx.doi.org/10. 1371/journal.pmed.1000157.
N N N N N N N N
Wikipedia contains information on Alzheimer disease (note that Wikipedia is a free online encyclopedia that anyone can edit; available in several languages) Information on bradycardia and its causes can be found in Wikipedia (note that Wikipedia is a free online encyclopedia that anyone can edit; available in several languages) The UKs National Health Service provides information on dementia, including symptoms, causes, diagnosis, treatment, and prevention MedlinePlus provides US-based health information (in English and Spanish) The US National Institute on Aging provides information on health, relevant to older people, including Alzheimer Disease and dementia (in English and Spanish) The US Alzheimers Association contains useful information on the disease, including on medication The Public Health Agency of Canada website provides information on senior health (in English and French) The UK-based Alzheimers Society provides advice on caring for people with dementia
74
PLoS MEDICINE
Reducing Amyloid Plaque Burden via Ex Vivo Gene Delivery of an Ab-Degrading Protease: A Novel Therapeutic Approach to Alzheimer Disease
Matthew L. Hemming1, Michaela Patterson2, Casper Reske-Nielsen2, Ling Lin2, Ole Isacson2, Dennis J. Selkoe1*
1 Center for Neurologic Diseases, Brigham and Womens Hospital and Harvard Medical School, Boston, Massachusetts, United States of America, 2 Neuroregeneration Laboratories, McLean Hospital and Harvard University Udall Parkinsons Disease Research Center of Excellence, Belmont, Massachusetts, United States of America Funding: This work was supported by National Institutes of Health Grants AG12749 (DJS) and NS39793 (OI), the Stern Foundation (OI), the Anti-Aging Foundation (OI), and the Harvard Center for Neurodegeneration and Repair (MLH). DJS is a consultant to Elan. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Academic Editor: Sam Gandy, Farber Institute, United States of America Citation: Hemming ML, Patterson M, Reske-Nielsen C, Lin L, Isacson O, et al. (2007) Reducing amyloid plaque burden via ex vivo gene delivery of an Ab-degrading protease: A novel therapeutic approach to Alzheimer disease. PLoS Med 4(8): e262. doi:10. 1371/journal.pmed.0040262 Received: March 19, 2007 Accepted: July 18, 2007 Published: August 28, 2007 Copyright: 2007 Hemming et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: Ab, amyloid bprotein; ACE, angiotensin-converting enzyme; AD, Alzheimer disease; APP, b-amyloid precursor protein; CHO, Chinese hamster ovary; DAGNPG, 3dansyl-D-Ala-Gly-p-(nitro)-Phe-Gly; GFAP, glial fibrillary acidic protein; HEK, human embryonic kidney; NEP, neprilysin; SEM, standard error of the mean; sNEP, secreted neprilysin * To whom correspondence should be addressed. E-mail: dselkoe@rics. bwh.harvard.edu
ABSTRACT
Background
Understanding the mechanisms of amyloid-b protein (Ab) production and clearance in the brain has been essential to elucidating the etiology of Alzheimer disease (AD). Chronically decreasing brain Ab levels is an emerging therapeutic approach for AD, but no such diseasemodifying agents have achieved clinical validation. Certain proteases are responsible for the catabolism of brain Ab in vivo, and some experimental evidence suggests they could be used as therapeutic tools to reduce Ab levels in AD. The objective of this study was to determine if enhancing the clearance of Ab in the brain by ex vivo gene delivery of an Ab-degrading protease can reduce amyloid plaque burden.

We generated a secreted form of the Ab-degrading protease neprilysin, which significantly lowers the levels of naturally secreted Ab in cell culture. We then used an ex vivo gene delivery approach utilizing primary fibroblasts to introduce this soluble protease into the brains of bamyloid precursor protein (APP) transgenic mice with advanced plaque deposition. Brain examination after cell implantation revealed robust clearance of plaques at the site of engraftment (72% reduction, p 0.0269), as well as significant reductions in plaque burden in both the medial and lateral hippocampus distal to the implantation site (34% reduction, p 0.0020; and 55% reduction, p 0.0081, respectively).
Conclusions
Ex vivo gene delivery of an Ab-degrading protease reduces amyloid plaque burden in transgenic mice expressing human APP. These results support the use of Ab-degrading proteases as a means to therapeutically lower Ab levels and encourage further exploration of ex vivo gene delivery for the treatment of Alzheimer disease.
1405
August 2007 | Volume 4 | Issue 8 | e262
75
ReducingAmyloidPlaqueBurdenviaExVivoGeneDeliveryofanAb-DegradingProtease:ANovelTherapeuticApproachtoAlzheimerDisease
Ab Clearance via Ex Vivo Gene Delivery
Introduction
The pathologic hallmarks of Alzheimer disease (AD) are extracellular plaques of amyloid-b protein (Ab) and intraneuronal neurobrillary tangles of tau protein, both of which accumulate in brain regions mediating memory and cognition [1]. All known early-onset, inherited forms of AD arise from mutations in the b-amyloid precursor protein (APP), from which the Ab peptide is generated, or the presenilins, which are the proteases that effect the cleavage of APP, which liberates Ab. Both amyloid plaques and soluble Ab oligomers are believed to have neurotoxic effects [2,3], and evidence suggests that Ab neurotoxicity can promote neurobrillary tangle formation [4,5]. Though approved treatments for AD ameliorate only the symptoms, potentially disease-modifying interventions under development focus on therapeutically lowering Ab production [6] or enhancing Ab clearance [7,8]. The role of Ab degradation in the clearance of the Ab peptide is becoming more broadly understood and appreciated, with the proteases neprilysin [9], insulin-degrading enzyme [10,11], endothelin-converting enzymes 1 and 2 [12], plasmin [13], and cathepsin B [14] all capable of regulating Ab levels in vivo. Supporting a therapeutic function for Abdegrading proteases, both transgenic overexpression [15] and direct viral vector injection [14,16,17] of these enzymes have been shown to potently lower Ab levels and plaque burden and reduce Ab-associated neuropathology. The applicability of these approaches for AD treatment in humans is uncertain, and alternative methods of gene delivery merit exploration. In ex vivo gene therapy, cells are taken from a patient, genetically modied in vitro, then implanted back into the patient to exert their new salutary effects. This approach has produced therapeutic improvements in experimental models of human diseases and conditions including hemophilia [18], retinal degeneration [19], cancer [20], spinal cord injury [21], myopathy [22], ischemia [23], Parkinson disease [24], Huntington disease [25], amyotrophic lateral sclerosis [26], and Alzheimer disease [27]. Several of these potential treatments have advanced to human trials [2831], with encouraging outcomes for patients. In a recent Phase I clinical trial, autologous broblasts from AD patients were modied to produce nerve growth factor and implanted into the forebrain, where they enhanced cholinergic function [28]. These efforts demonstrate the feasibility of ameliorating cholinergic cell death by providing trophic support using an ex vivo gene delivery approach. However, because Ab-mediated synaptic function and cytotoxicity are likely to be upstream of neurotransmitter abnormalities in AD, enhancing Ab clearance represents a more attractive target for gene therapy in this disorder. Neprilysin (NEP) is a type II transmembrane protein bound to the cell membrane, where it normally degrades Ab intracellularly and on the cell surface [32]. Cerebral NEP levels have been reported to decrease with age and in AD [33,34], which may contribute to disease pathogenesis by compromising Ab catabolism. As an approach to elevate NEP levels and improve the enzymes access to extracellular Ab, we replaced the transmembrane domain of NEP with a signal peptide, which produced a soluble and secreted form of NEP, termed sNEP. We then used ex vivo gene delivery utilizing syngenic primary broblasts to introduce sNEP into the brains of APP transgenic mice that had advanced plaque
deposition and studied the effect of this protein on brain amyloid pathology.
Materials and Methods sNEP Cloning

The sNEP construct was generated by an overlapping PCR method to fuse the 29 amino-acid signal peptide of angiotensin-converting enzyme (ACE) to luminal residues 52750 of neprilysin. Two PCR amplicons were generated that consisted of the ACE signal peptide and the NEP luminal residues, both with a complimentary 23-basepair overlap added to the primer. In a third PCR reaction, the overlapping amplicons were fused using the ACE signal peptide 59 and NEP 39 primers to produce the cDNA encoding sNEP. All constructs were cloned into the pcDNA3.1/Hygro vector (Invitrogen, http://www.invitrogen.com/) for transient and stable expression. The human ACE and human NEP cDNAs have been previously described [15,35]
Cell Culture and Lentiviral Transduction

Chinese hamster ovary (CHO) and human embryonic kidney (HEK) 293 cells were grown in Dulbeccos modied Eagles medium containing 10% fetal bovine serum, 2 mM Lglutamine, 100 lg/ml penicillin, and 100 lg/ml streptomycin. HEK cells stably expressing APP695 bearing the K595N/M596L AD-causing mutation [36] and CHO cells stably transfected with APP751 with the V717F AD-causing mutation [37] were maintained in medium containing 200 lg/ml G418. Transfections were performed with Fugene 6 (Roche, http://www. roche.com/), and stable cell lines were selected using 400 lg/ ml hygromycin B. Cells were conditioned for Ab measurements and protein analysis in Opti-MEM I (Invitrogen) for 1618 h. For cell-free Ab degradation assays, conditioned medium from APP-overexpressing CHO cells was combined with 10-fold concentrated conditioned medium from CHO cells overexpressing empty vector, NEP, or sNEP and incubated at 37 8C for 18 h. Lentiviral vectors were generated by inserting the sNEP and GFP cDNAs into pCDH1 (System Biosciences, http://www. systembio.com/) and were cotransfected with the helper plasmids delta8.9 and VSV-G as previously described [38] into 293-FT cells (Invitrogen). Conditioned media were collected 5060 h later, centrifuged at 500 g for 5 min to remove suspended cells, and stored at 80 8C. Viral multiplicity of infection was estimated by cellular uorescence and Western blotting by transducing 293-FT cells with serial dilutions of viral supernatant. Cells were transduced for experiments at a multiplicity of infection of six to ten viral particles per cell with media containing 6 lg/ml polybrene.
Immunoblotting
Cells and tissue were lysed in 50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% NP-40, protease inhibitor cocktail (Roche), 2 mM 1,10-phenanthroline, and 5 mM EDTA, and the extracts centrifuged at 1,000 g for 10 min to remove nuclei. Protein concentrations were determined using a bicinchoninic acidbased assay (Pierce, http://www.perbiodirect.com/). Samples were then subjected to SDS-PAGE and Western blotting. We detected NEP with monoclonal antibody 56C6 (1:100 dilution; Abcam, http://www.abcam.com/); GFP with polyclonal antibody A11122 (1:1,000 dilution; Invitrogen); Ab with monoAugust 2007 | Volume 4 | Issue 8 | e262
76

clonal antibody 6E10 (1:1,000 dilution; Abcam); and APP with polyclonal antibody C9 (1:1,000 dilution) [35]. Deglycosylation was performed using PNGase F to remove N-linked sugars (Prozyme, http://www.prozyme.com/). Western blots were probed with anti-mouse or anti-rabbit secondary antibodies (1:10,000 dilution) conjugated to Alexa Fluor 680 (Molecular Probes, http://probes.invitrogen.com/) or IRdye 800 (Rockland Immunochemicals, http://www.rockland-inc. com/). Blots were detected using the Odyssey infrared imaging system (LI-COR, http://www.licor.com/).
Mice and Surgical Procedures

The J20 line of APPSwe/Ind heterozygous transgenic mice (C57BL/6 3 DBA2 background) [42] were aged to approximately 2223 mo before surgery to allow for substantial plaque deposition. For surgical procedures, mice received intraperitoneal injections of a preanesthetic mixture (a 1:1 solution of atropine sulfate and acepromazine maleate), followed by a 2:1 solution of ketamine and xylazine. Mice received a unilateral hippocampal injection of 500,000 broblasts expressing sNEP or GFP to allow for comparison to the contralateral side. The stereotaxic coordinates for graft placement were calculated from bregma: AP 3.3 mm; L 3.3 mm; DV 3.12 mm and 1.9 mm. Cells, in a total volume of 4 ll, were injected at two dorsoventral positions to better distribute the graft. Brains were harvested 28 d after surgery and xed in 10% formalin for subsequent immunohistochemical analysis. All animal procedures were approved by the Harvard Standing Committee for Animal Use and by the Animal Care and Use Committee at McLean Hospital (Belmont, Massachusetts, United States).
NEP Activity Assay

NEP proteolytic activity was measured using the substrate 3-dansyl-D-Ala-Gly-p-(nitro)-Phe-Gly (DAGNPG; Sigma, http://www.sigmaco.com.au/) [39,40]. Cell lysate (100 lg) or the same relative amount of concentrated conditioned medium (prepared without protease inhibitors) was incubated with 50 lM DAGNPG and 1 lM captopril (to inhibit any ACE cleavage of DAGNPG) in a volume of 200 ll at 37 8C. Reactions were stopped by heating samples to 100 8C for 5 min, then centrifuging at 5,000 g for 5 min to remove the denatured protein. The whole supernatant was diluted into 400 ll of 50 mM Tris (pH 7.4) and uorescence determined using a Victor2 multilabel plate reader (excitation 342 nm; emission 562 nm).
Immunohistochemistry and Image Analysis

Sagittal brain sections of 18 lm thickness were deparafnized and hydrated through a series of graded alcohol steps and washed in phosphate-buffered saline. Endogenous peroxidase activity was quenched with 0.6% hydrogen peroxide in methanol for 15 min. Sections were blocked in serum for 25 min, then incubated with the anti-Ab antibody R1282 (1:1,000 dilution) [39], anti-NEP antibody 56C6 (1:100), anti-GFP antibody A11122 (1:100), or anti-glial brillary acidic protein (GFAP) antibody (1:1,000; Dako, http://www. dako.com/) for 2 h. After washing, an anti-rabbit or antimouse biotinylated secondary antibody was applied for 30 min, washed, and developed using the avidin/biotin/HRP method (ABC Elite Kit, Vector Labs) and DAB chromogenic reaction (Liquid DAB, BioGenex, http://www.biogenex.com/). Images were captured and quantied from approximately 20 stained sections per brain of each mouse using a 23 or 53 objective. Thioavin-S staining was performed by incubating tissue sections for 8 min in 1% thioavin-S, followed by washing in 80% and 95% ethanol, then water. The brain area covered by Ab, thioavin plaques, and GFAP staining was determined in a blinded fashion using IPLab Spectrum 3.1 Image Analyzer software (Signal Analytics, http://www. scanalytics.com/) and NIH ImageJ software [43]. The ratio between the immunohistochemical and thioavin staining on the ipsilateral versus contralateral hemispheres was calculated for comparisons.
ELISA
ELISAs for Ab were performed as previously described [39,41] with few modications. 96-well ELISA plates (Costar, http://www.corning.com/) were coated with 3.5 lg/ml of the capture antibody. Ab was measured by capturing with monoclonal antibody 266, specic to residues 1328 of Ab. Captured Ab was detected with biotinylated monoclonal antibody 3D6, specic to residues 15 of the Ab N terminus (all ELISA antibodies were gifts of Elan Pharmaceuticals, http://www.elan.com/). ELISAs were developed by incubating the Ab-bound biotinylated 3D6 with avidin-horseradish peroxidase (Vector Labs, http://www.vectorlabs.com/) followed by a 1 h incubation with QuantaBlu uorogenic peroxidase substrate (Pierce), and the resulting uorescence (excitation 340 nm; emission 400 nm) was measured. Plate washing after the antibody and enzyme binding steps was performed three times for 1 min with Tris-buffered saline, 0.05% Tween 20.
Primary Fibroblast Generation and Transduction

Primary broblasts were generated from young wild-type littermates of J20 transgenic mice that expressed human APP. Skin biopsies were swabbed with 70% ethanol to sterilize the tissue, and were then minced with sterile scalpel blades in 0.25% trypsin-EDTA. Minced tissue was incubated in trypsinEDTA at 37 8C for 15 min and triturated to further disrupt the tissue. Cells remaining in suspension were plated in growth medium and allowed to expand to 90% conuency. Fibroblasts were then passaged and transduced with two serial applications of either sNEP (treatment cells) or GFP (control cells) lentiviral vectors to attain strong transgene expression in all cells. Fibroblasts were passaged a maximum of ve times before injections. Conuent cultures of transduced broblasts were maintained to observe cell longevity and transgene expression. After 12 mo in culture, the cells remained metabolically active and continued to express the introduced transgene.
Statistical Analysis
The data were analyzed using a one-way analysis of variance and Tukey post-hoc comparison or a two-tailed Student t-test, where appropriate. Calculated comparisons of p , 0.05 were considered signicant. All reported values represent the means 6 standard error of the mean (SEM).
Results NEP Expression and Generation of sNEP

Several studies point to NEP as one of the prominent Abdegrading proteases within the brain [9,15,44,45]. To determine the tissue distribution of NEP in the brain and several
77
Figure 1. Tissue Distribution of NEP and Generation of Secreted NEP (A) Identical amounts of protein from transfected CHO cells and the indicated mouse tissues were probed by NEP immunoblot. CHO cells were transiently transfected with empty vector (CHO-control) or human NEP (CHO-NEP). The indicated peripheral tissues and brain regions (Bg/Bs, basal ganglia/brainstem; Cb, cerebellum; Ctx, cortex; Hip, hippocampus) were homogenized and probed by Western blot for NEP. The lower blot is a longer exposure of the brain samples, demonstrating low levels of NEP in the brain. Each sample was either left untreated () or deglycosylated with PNGase F () to remove N-linked sugars. (B) Schematic representation of wild-type NEP and sNEP proteins. The sNEP construct was generated by replacing the NEP transmembrane (TM) domain and cytosolic N terminus with a signal peptide (SP) ending at the luminal residue 52 of NEP. Cleavage of the signal peptide produces a soluble, secreted form of NEP. (C) NEP immunoblot of cellular lysates from CHO cells transfected with empty vector, NEP, or sNEP constructs. (D) NEP activity assay in which the fluorogenic NEP substrate DAGNPG was incubated with lysates from CHO cells stably transfected with the indicated constructs. NEP enzymatic activity was inhibited by phosphoramidon. (E) NEP immunoblot of conditioned medium from CHO cells transfected with the indicated constructs. (F) NEP activity assay on conditioned media from cells transfected with the indicated constructs. NEP activity assays of conditioned media (F) were performed using the same relative amounts of material as for the lysates (D). Immunoblots are representative of at least three experiments; NEP activity assays report the mean 6 SEM of six experiments. doi:10.1371/journal.pmed.0040262.g001
78
peripheral tissues, we dissected cortex, hippocampus, cerebellum, and the basal ganglia/brainstem from the murine brain and compared protein expression levels to those in the liver, lung, and kidney. As positive and negative controls,
lysates from CHO cells transfected with NEP or empty vector control were examined in parallel. Equal amounts of protein from each sample were examined by quantitative uorescent Western blotting. Because transfected NEP may run as a

Figure 2. Clearance of the Ab Peptide by NEP and sNEP (A) CHO cells overexpressing APP were cocultured with CHO cells stably expressing either empty vector, NEP, or sNEP constructs. Equal numbers of cells were seeded, grown to confluence, and media were conditioned for 18 h. Conditioned media were analyzed for total Ab levels by ELISA. (B) Cellular lysates from the coculture in (A) were probed by Western blotting for APP (which presents in both mature and immature glycosylated forms), demonstrating equal amounts of APP expression. Each condition is shown in duplicate. (C) Conditioned media from CHO cells stably transfected with APP were incubated in vitro with conditioned media from CHO cells expressing empty vector, NEP, or sNEP. The conditioned media were combined and incubated at 37 8C for 18 h. Ab levels were determined by ELISA. The immunoblot for APP (B) is representative of four experiments; ELISA values for Ab represent the mean 6 SEM of five experiments. For comparisons to the empty vector condition, ** p , 0.01 and *** p , 0.001. doi:10.1371/journal.pmed.0040262.g002
contrast to lung and kidney, brain levels of NEP are very low and detectable only upon longer blot exposure (Figure 1A, lower blot). Thus, although NEP has been characterized as an essential protease for Ab catabolism, its protein expression levels are relatively low in the brain compared to peripheral tissues that express NEP. Moreover, since cerebral NEP expression decreases with age and in AD [33,34], elevating NEP levels could be neuroprotective by enhancing Ab catabolism. In exploring potential therapeutic approaches to reducing brain Ab levels by elevating NEP expression, we evaluated a secreted form of NEP, sNEP. A secreted Ab-degrading protease would possess the advantage of diffusion through the interstitial uid, where it may clear Ab at sites of extracellular deposition. To generate this sNEP construct, we used an overlapping PCR method to fuse the 29-amino acid signal peptide from ACE to the luminal domain of NEP beginning at residue 52 (Figure 1B). This construct, upon being translated and inserted into the endoplasmic reticulum membrane, has the signal peptide removed to liberate a soluble, secreted form of NEP. Transfecting cells with empty vector, NEP, or the sNEP constructs and examining cell lysates by Western blotting revealed a robust immunoreactive NEP band in the wild-type NEP condition and a weaker, incompletely glycosylated, lower molecular weight signal in the sNEP lane (Figure 1C). Measuring NEP enzymatic activity in cell lysates demonstrated NEP activity only in the wild-type NEP condition, and this was inhibitable to baseline signal by the metalloprotease inhibitor phosphoramidon (Figure 1D). The incompletely glycosylated intracellular sNEP showed no enzymatic activity, consistent with reports that glycosylation is required for NEP activity [46]. Concentrating the conditioned media from these cells for Western blot analysis revealed a strong signal for sNEP, consistent with proper sNEP expression and secretion (Figure 1E). Quantifying NEP activity in conditioned media (using the same relative amount of material used for cell lysates in Figure 1D) demonstrated robust NEP activity in the sNEP samples alone, and this was again completely inhibited by phosphoramidon (Figure 1F). These data demonstrate that the sNEP construct is processed and secreted by cells as expected and that sNEP is enzymatically active against a prototypical NEP substrate.
sNEP Reduces Ab Levels in Cell Culture

To determine if the sNEP construct is able to degrade naturally secreted Ab to a similar extent as wild-type NEP, we cocultured CHO cells stably expressing APP at a 1:1 ratio with CHO cells expressing either empty vector as control, wildtype NEP, or sNEP. Cells were allowed to grow to conuency, and media were conditioned for Ab ELISA determinations. Both wild-type NEP and sNEP were capable of promoting Ab catabolism, with the NEP and sNEP conditions reducing Ab levels in the media to 74% and 55% of control, respectively (p , 0.01 for NEP and p , 0.001 for sNEP) (Figure 2A). These reductions in Ab levels were not due to differences in the proliferation rates of the cell lines, as there was equivalent APP expression in each coculture condition (Figure 2B), indicating a similar number of APP expressing cells at conuency. In order to clarify the location of NEP- and sNEP-mediated Ab degradation, we mixed conditioned media from the APPoverexpressing cell line with concentrated media from cells
1409 August 2007 | Volume 4 | Issue 8 | e262
broad band due to its varying degrees of glycosylation, all samples were assessed in the absence or presence of PNGase F to remove N-linked glycosylation and reduce NEP immunoreactivity to a single signal. Mouse lung and kidney express high levels of NEP that comigrates with transfected human NEP, whereas the liver lacks NEP expression (Figure 1A). In
79
Figure 3. Characterization of a sNEP Lentiviral Construct (A) Schematic representation of the lentiviral constructs expressing sNEP and GFP. Blocks indicate lentiviral genetic components. Internal promoters and transgenes are indicated by arrows. (B) HEK cells stably transfected with APP695 K595N/M596L (HEKAPP695) and CHO cells stably transfected with APP751 V717F (CHOAPP751) were transduced with the lentiviral constructs in (A). Lysates and conditioned media (CM) were collected and probed for sNEP, GFP, APP, and Ab. (C) Media from the lentivirally transduced cells were conditioned, and Ab levels were determined by ELISA. Immunoblots are representative of four experiments; Ab ELISAs represent the mean 6 SEM of four experiments, compared to GFP transduced cells: *** p , 0.001 doi:10.1371/journal.pmed.0040262.g003
expressing empty vector, wild-type NEP, or sNEP. Only the sNEP-conditioned media exhibited an increased capacity to degrade Ab compared to control (p , 0.001) (Figure 2C), elevating the amount of Ab degraded 2-fold. Taken together, these data demonstrate that wild-type NEP degrades Ab principally while anchored to the cell membrane, whereas sNEP is efciently secreted into the extracellular environment to mediate Ab catabolism.
Generation of Viral and Cell-Based Ab-Degrading Vectors

To pursue an ex vivo gene delivery approach, we generated lentiviral vectors able to efciently transduce most cell types. Retroviral vectors have been routinely utilized to genetically enhance cells ex vivo prior to engraftment for the experimental treatment of human diseases [4749]. We inserted the cDNA of sNEP or GFP (as control) into a lentiviral vector driven by the cytomegalovirus promotor (Figure 3A). The
80
resulting vectors were able to transduce all cultured cell types tested (HeLa, SH-SY5Y, SK-N-SH, HEK, CHO, and primary and immortalized broblasts), and produced high levels of GFP or sNEP expression. To assess the ability of the sNEP lentiviral vector to reduce Ab levels in APP overexpressing cell lines, we transduced HEK cells stably expressing a familial AD mutant form of APP695 and CHO cells stably expressing a familial AD mutant form of APP751 with the sNEP or GFP vectors. Examining cell lysates and conditioned media by Western blot showed strong transgene expression, with HEK cells attaining higher sNEP and GFP production than did CHO cells (Figure 3B, top three blot pairs). Expression of sNEP did not change APP levels in these cells compared with GFP transduction (Figure 3B, fourth blot pair); however, sNEP expression resulted in a marked decrease in Ab monomer in the conditioned media (Figure 3B, fth blot pair). Quantifying Ab levels in the conditioned media by

Figure 4. Generation of Primary Mouse Fibroblasts Expressing GFP and sNEP Fibroblast cultures were generated from wild-type littermates of J20 APP transgenic mice. (A) Live-cell fluorescent imaging of primary fibroblasts transduced with lentiviral vectors expressing GFP (top) or sNEP (bottom). (B) Conditioned media from the primary fibroblasts were probed for the presence of sNEP. The immunoblot is representative of three experiments. doi:10.1371/journal.pmed.0040262.g004
ELISA demonstrated signicant reductions in Ab due to sNEP expression in both HEK and CHO APP-overexpressing cell lines (p , 0.001) (Figure 3C). Fibroblasts were chosen as the optimal cell type for use in ex vivo gene delivery for a number of reasons. Fibroblasts survive grafting into several organs, including the brain; they survive and can sustain transgene expression for years [28]; they do not form tumors or migrate from the graft site [47]; they cause no detectible toxicity to the host tissue [50]; and they can be noninvasively extracted from patients via skin biopsy. For this study, broblasts were prepared from the wild-type littermates of the J20 APP transgenic mice. Cells were dissociated from skin biopsies and plated in growth medium for brief expansion in culture. Following their rst passage in culture, the cells were transduced with either the sNEP or the GFP lentiviral vector. The broblasts transduced with GFP produce a strong native uorescent signal (Figure 4A), while cells transduced with sNEP robustly secrete the sNEP protein into the extracellular medium (Figure 4B). To minimize the possibility of in vitro cell transformation, the broblasts used in the study were not passaged more the ve times before engraftment into the experimental transgenic mice.
sNEP Cell Grafting Reduces Plaque Burden Locally and Distal to the Implantation Site
A cohort of J20 APP transgenic mice was aged to achieve signicant plaque deposition [42] for use in these studies. To
each of these mice, 500,000 cells expressing sNEP or GFP were stereotaxically injected unilaterally into the hippocampus (stereotaxic coordinates: AP 3.3 mm; L 3.3 mm; DV 3.12 and 1.9 mm), with the uninjected contralateral hippocampus serving as a control. Cells were injected at two dorsoventral positions to better distribute the graft. The brains of the mice were harvested and immunohistochemically assayed 28 d after surgery. Mice implanted with sNEP grafts exhibited NEP immunoreactivity at the graft site, as well as some diffuse NEP signal in nearby areas of the hippocampus that extended approximately 800 lm from the grafted cells (Figure 5A, right photomicrograph), consistent with sNEP secretion. Endogenous NEP immunoreactivity is localized to the cerebral blood vessels and the choroid plexus (Figure 5A, left photomicrograph). GFP broblasts exhibit GFP staining restricted to the grafted cells (Figure 5B, right). To determine the extent of Ab clearance from the brain due to sNEP delivery, we stained sagittal sections for Ab immunoreactivity from the medial to the lateral hippocampus. At the graft site, the ipsilateral hippocampus was largely cleared of plaques near the sNEP-producing broblasts (Figure 5E, right). There was some reduction in plaque load at the GFP graft (Figure 5F, right), which was likely due to the displacement of plaque-laden tissue by the grafted GFP-expressing broblasts. The ipsilateral hippocampus medial to the graft site showed signicant reductions in Ab immunoreactivity in the sNEP-injected mice (compared to
81
Figure 5. Reduction in Hippocampal Plaque Burden following sNEP Cell Engraftment Aged J20 APP transgenic mice were implanted with fibroblasts expressing either sNEP (n 5) (A, C, E, and G) or GFP (n 3) (B, D, F, and H). Cells were stereotaxically placed into the right (ipsilateral, right) hippocampus, with the uninjected left (contralateral, left) hippocampus serving as control. Brains were harvested for analysis 28 d after surgery. Grafted cells were immunoreactive for either NEP (A) or GFP (B), with superior and inferior aspects of the graft, respectively, indicated by arrows. Images (CH) show staining for plaque burden at sites medial to the graft (C and D), at the site of the graft (E and F), and at sites lateral to the graft (G and H). These fields were quantified by calculating the ratio of the area covered by plaques on the ipsilateral versus the contralateral side; quantification was done in a blinded fashion. This ratio was determined separately for the sNEP and GFP groups and compared at the hippocampus medial to the graft (I), the graft site (J), and the hippocampus lateral to the graft (K). Data represent the ratios mean 6 SEM, compared to GFP control: ** p 0.0020 for the medial hippocampus (I); * p 0.0269 for the graft site (J); and ** p 0.0081 for the lateral hippocampus (K). doi:10.1371/journal.pmed.0040262.g005
82
PLoS Medicine | www.plosmedicine.org 1412 August 2007 | Volume 4 | Issue 8 | e262

Figure 6. Reduction in Hippocampal Thioflavin Staining Following sNEP Cell Engraftment Brain sections from mice engrafted with cells expressing sNEP or GFP were stained with thioflavin-S to quantify fibrillar plaques. Photomicrographs (A D) show thioflavin staining at sites medial to the graft (A and B) and at the site of the graft (C and D). Thioflavin-positive plaques were counted, and the ratios of plaques on the ipsilateral versus contralateral hemispheres were calculated for sNEP and GFP groups for the medial hippocampus (E), the graft site (F), and the lateral hippocampus (G). Data represent the mean 6 SEM, compared to GFP control: *p 0.0432 for the medial hippocampus (E), and *p 0.0337 for the graft site (F). The difference at the lateral hippocampus was not significant. doi:10.1371/journal.pmed.0040262.g006
the uninjected contralateral hippocampus), whereas no effect was observed in the GFP-injected mice (Figure 5C versus 5D, respectively). Similarly, in the lateral hippocampus distal to the graft, we observed signicant reductions in plaque burden due to sNEP expression (Figure 5G versus 5H). Changes in plaque burden were measured in blinded fashion with automated image analysis, and the data expressed as a ratio of the area covered by Ab plaque on the grafted ipsilateral side versus the control contralateral side. Comparing the ipsilateral/contralateral ratios of the sNEP and GFP groups, signicant reductions in plaque burden were observed in the ipsilateral hippocampus from mice engrafted with sNEP-secreting broblasts medial to the graft site (34% reduction, p , 0.01), within the graft site (72% reduction, p , 0.05), and lateral to the graft site (55% reduction, p , 0.01) (Figure 5I5K). Thus, ex vivo gene delivery of a soluble, secreted form of human NEP resulted in substantial and
signicant reductions in Ab plaque burden both in the vicinity of the graft and adjacent to the graft site.
Reduction in Thioflavin-Positive Plaques, but Not Astrocytosis, Following sNEP Cell Engraftment
To determine if the engrafted sNEP was able to clear brillar plaques, we examined the number of thioavinpositive plaques in the ipsilateral and contralateral hippocampus. Sagittal sections throughout the hippocampus were stained with thioavin-S, and the ratio between the number of plaques on the ipsilateral versus contralateral hippocampus was determined. Comparing the ipsilateral/contralateral ratios from the sNEP and GFP groups, thioavinpositive plaques were reduced in the medial hippocampus (40% reduction in plaque number, p , 0.05; Figure 6A and 6B) and at the graft site (51% reduction in plaque number, p , 0.05; Figure 6C and 6D), but was not signicantly changed in the lateral hippocampus of the sNEP engrafted brains
83
Figure 7. No Change in Overall Astrocytosis following Cell Implantation Brain sections from engrafted mice were stained for GFAP to probe for changes in astrocytosis. (A and B) Compared to the contralateral hemisphere (A), the graft site (B) demonstrated an absence of GFAP staining within the graft and modestly increased astrocyte staining along the border of the graft. The needle track is indicated by arrows and the graft by an asterisk. This staining pattern was observed for both sNEP and GFP conditions (only sNEP is shown). (CE) The area staining for GFAP on the ipsilateral versus contralateral hippocampus was determined for the medial hippocampus (C), the graft site (D), and the lateral hippocampus (E). Data represent the mean 6 SEM. doi:10.1371/journal.pmed.0040262.g007
84
compared to control (Figure 6E6G). The lateral hippocampus had few or no thioavin-positive plaques in either condition and was thus not a reliable reporter of plaque clearance. These data demonstrate that sNEP delivery promotes the clearance of brillar, thioavin-positive plaques from the brain of APP transgenic mice. The activation of astrocyte growth (astrocytosis) is a marker for brain inammation, and astrocyte hypertrophy and number are increased in AD and may play a role in disease pathogenesis [1]. Transgenic expression of neprilysin has been shown previously to reduce astrocytosis over the lifetime of an APP-transgenic mouse [15], presumably due to reductions in Ab accumulation. However, because we used a surgical approach to exogenously deliver a soluble form of NEP that clears existing plaques, there could be an inammatory response associated with plaque removal or with the soluble protease itself. To examine this possibility, we stained sNEP and GFP brains for the astrocyte GFAP. At the site of cell engraftment, few astrocytes were found to have inltrated the graft, although there was a modest increase in GFAP staining along the edge of the graft and along the needle track (Figure 7A and 7B). However, comparing overall hippocampal GFAP staining on the ipsilateral versus contralateral hemispheres showed that there was no change in astrocytosis due to cell engraftment or sNEP expression in any region of the hippocampus (Figure 7C7E). Thus, sNEP cell delivery did not promote astrocytosis when measured at
28 d postimplantation. Future experiments examining longer time points of sNEP delivery could show that brain inammation decreases after Ab clearance.
Discussion
Various approaches to lowering Ab levels within the brain have come to represent the central focus in AD therapeutic development. In this study, we describe the use of ex vivo gene therapy as a viable approach to reducing Ab levels utilizing a modied Ab-degrading protease. Replacing the transmembrane domain of the normally membrane-anchored NEP with a signal peptide resulted in the robust secretion of an enzymatically active protease. When syngenic broblasts modied to produce sNEP were engrafted into aged APP transgenic mice, the majority of the plaque burden surrounding the graft site was proteolytically cleared. Distal to the graft, in both the medial and lateral hippocampus, signicant reductions in plaque burden were also observed as a result of sNEP gene delivery. These results demonstrate the utility of locally elevating the expression of a secreted Ab-degrading protease as a potential therapeutic approach for AD. The nding that sNEP was able to reduce plaque burden distal to the graft site may be explained by the diffusion of the sNEP enzyme through the parenchyma, by diffusion of soluble forms of Ab toward the graft (which may thus be serving as an Ab sink), or both of these mechanisms. Because

of the low sensitivity of available anti-NEP antibodies, we were unable to detect sNEP diffusion farther than ;800 lm from the graft site, nor could we detect the very low (Figure 1A) endogenous NEP levels expressed by neural tissue [51]. Potential side effects of sNEP expression may arise from the degradation of NEP substrates other than Ab present in the brain, such as enkephalins. Though studies explicitly searching for such side effects were not performed here, no adverse phenotype was reported in NEP transgenic mice that were protected from Ab-induced premature death [15], and in this study we observed no enhanced astrocytosis in response to sNEP expression. A major obstacle to translating the ex vivo methods used in this study into a human disease therapy is the larger dimensions of the human brain, which might require an elevation in Ab-degrading activity throughout a much broader area. However, our experiment involved just a 28 d exposure to the sNEP-secreting cells, and we found these primary cells to maintain transgene expression in vitro for over a year. One approach to the problem of optimal diffusion of the gene product is to implant the cells where they may have the best access to circulating Ab, such as in the cerebrospinal uid or the ventricular wall. Several studies have reported the successful use of encapsulated cell implants that allow for diffusion of therapeutic factors into biological uids, including cerebrospinal uid [29,30]. By analogy to the peripheral sink hypothesis for Ab immunotherapy [52], it may also be feasible and attractive to implant genetically modied cells into peripheral compartments (e.g., the peritoneal cavity or skeletal muscle), where they may substantially increase the Ab-degrading capacity of the periphery and thus increase the rate of efux of Ab across the bloodbrain barrier. Moreover, some of the secreted sNEP provided by such a peripheral implant may reach the brain and cerebrospinal uid to help decrease brain Ab levels directly. The existence of a subset of hematopoietic cells capable of inltrating the brain and surrounding plaques has been described [53,54]. Though these cells may not efciently degrade Ab themselves, if they could be modied to express an Ab-degrading protease in a manner similar to the approach we report here, they may provide an anatomically targeted means for local Ab elimination. In summary, our ndings demonstrate that ex vivo gene delivery of an Ab-degrading protease rapidly lowers hippocampal Ab plaque burden in a mouse model of Alzheimer disease. Ex vivo gene delivery and other strategies to elevate Ab-degrading activity in Alzheimer patients warrant further investigation as potential therapeutic approaches to treat this common and devastating disease.
3.
4. 5. 6. 7. 8. 9. 10.
11. 12. 13.
14. 15.
16. 17. 18. 19.
20.
21. 22.
Acknowledgments
We thank Lennart Mucke for the J20 line of APP transgenic mice, Elan Pharmaceuticals for providing the Ab ELISA antibodies, and Kelly Dakin for critical review and discussion of the manuscript. Author contributions. MLH, OI, and DJS designed the study. MLH, MP, CRN, and LL conducted surgical experiments. MLH collected and analyzed the data. MLH and DJS interpreted the data and drafted the manuscript. All authors have reviewed and approved the manuscript.
References 1. Selkoe DJ (2001) Alzheimers disease: Genes, proteins, and therapy. Physiol Rev 81: 741766. 2. Walsh DM, Klyubin I, Fadeeva JV, Cullen WK, Anwyl R, et al. (2002)
23.
24.
25. 26.
27.
Naturally secreted oligomers of amyloid beta protein potently inhibit hippocampal long-term potentiation in vivo. Nature 416: 535539. DAmore JD, Kajdasz ST, McLellan ME, Bacskai BJ, Stern EA, et al. (2003) In vivo multiphoton imaging of a transgenic mouse model of Alzheimer disease reveals marked thioavine-S-associated alterations in neurite trajectories. J Neuropathol Exp Neurol 62: 137145. Lewis J, Dickson DW, Lin WL, Chisholm L, Corral A, et al. (2001) Enhanced neurobrillary degeneration in transgenic mice expressing mutant tau and APP. Science 293: 14871491. Oddo S, Billings L, Kesslak JP, Cribbs DH, LaFerla FM (2004) Abeta immunotherapy leads to clearance of early, but not late, hyperphosphorylated tau aggregates via the proteasome. Neuron 43: 321332. Siemers ER, Quinn JF, Kaye J, Farlow MR, Porsteinsson A, et al. (2006) Effects of a gamma-secretase inhibitor in a randomized study of patients with Alzheimer disease. Neurology 66: 602604. Gilman S, Koller M, Black RS, Jenkins L, Grifth SG, et al. (2005) Clinical effects of Abeta immunization (AN1792) in patients with AD in an interrupted trial. Neurology 64: 15531562. Hock C, Konietzko U, Streffer JR, Tracy J, Signorell A, et al. (2003) Antibodies against beta-amyloid slow cognitive decline in Alzheimers disease. Neuron 38: 547554. Iwata N, Tsubuki S, Takaki Y, Shirotani K, Lu B, et al. (2001) Metabolic regulation of brain Abeta by neprilysin. Science 292: 15501552. Farris W, Mansourian S, Chang Y, Lindsley L, Eckman EA, et al. (2003) Insulin-degrading enzyme regulates the levels of insulin, amyloid beta protein, and the beta -amyloid precursor protein intracellular domain in vivo. Proc Natl Acad Sci U S A 100: 41624167. Miller BC, Eckman EA, Sambamurti K, Dobbs N, Chow KM, et al. (2003) Amyloid-beta peptide levels in brain are inversely correlated with insulysin activity levels in vivo. Proc Natl Acad Sci U S A 100: 62216226. Eckman EA, Watson M, Marlow L, Sambamurti K, Eckman CB (2003) Alzheimers disease beta-amyloid peptide is increased in mice decient in endothelin-converting enzyme. J Biol Chem 278: 20812084. Melchor JP, Pawlak R, Strickland S (2003) The tissue plasminogen activatorplasminogen proteolytic cascade accelerates amyloid-beta (Abeta) degradation and inhibits Abeta-induced neurodegeneration. J Neurosci 23: 88678871. Mueller-Steiner S, Zhou Y, Arai H, Roberson ED, Sun B, et al. (2006) Antiamyloidogenic and neuroprotective functions of cathepsin B: Implications for Alzheimers disease. Neuron 51: 703714. Leissring MA, Farris W, Chang AY, Walsh DM, Wu X, et al. (2003) Enhanced proteolysis of beta-amyloid in APP transgenic mice prevents plaque formation, secondary pathology, and premature death. Neuron 40: 1087 1093. Iwata N, Mizukami H, Shirotani K, Takaki Y, Muramatsu S, et al. (2004) Presynaptic localization of neprilysin contributes to efcient clearance of amyloid-beta peptide in mouse brain. J Neurosci 24: 991998. Marr RA, Rockenstein E, Mukherjee A, Kindy MS, Hersh LB, et al. (2003) Neprilysin gene transfer reduces human amyloid pathology in transgenic mice. J Neurosci 23: 19921996. Dwarki VJ, Belloni P, Nijjar T, Smith J, Couto L, et al. (1995) Gene therapy for hemophilia A: Production of therapeutic levels of human factor VIII in vivo in mice. Proc Natl Acad Sci U S A 92: 10231027. Uteza Y, Rouillot JS, Kobetz A, Marchant D, Pecqueur S, et al. (1999) Intravitreous transplantation of encapsulated broblasts secreting the human broblast growth factor 2 delays photoreceptor cell degeneration in Royal College of Surgeons rats. Proc Natl Acad Sci U S A 96: 31263131. Peron JM, Bureau C, Gourdy P, Lulka H, Souque A, et al. (2007) Treatment of experimental murine pancreatic peritoneal carcinomatosis with broblasts genetically modied to express IL12: A role for peritoneal innate immunity. Gut 56: 107114. Blesch A, Tuszynski MH (2003) Cellular GDNF delivery promotes growth of motor and dorsal column sensory axons after partial and complete spinal cord transections and induces remyelination. J Comp Neurol 467: 403417. Lattanzi L, Salvatori G, Coletta M, Sonnino C, Cusella De Angelis MG, et al. (1998) High efciency myogenic conversion of human broblasts by adenoviral vector-mediated MyoD gene transfer. An alternative strategy for ex vivo gene therapy of primary myopathies. J Clin Invest 101: 21192128. Ishii S, Koyama H, Miyata T, Nishikage S, Hamada H, et al. (2004) Appropriate control of ex vivo gene therapy delivering basic broblast growth factor promotes successful and safe development of collateral vessels in rabbit model of hind limb ischemia. J Vasc Surg 39: 629638. Frim DM, Uhler TA, Galpern WR, Beal MF, Breakeeld XO, et al. (1994) Implanted broblasts genetically engineered to produce brain-derived neurotrophic factor prevent 1-methyl-4-phenylpyridinium toxicity to dopaminergic neurons in the rat. Proc Natl Acad Sci U S A 91: 51045108. Emerich DF, Winn SR, Hantraye PM, Peschanski M, Chen EY, et al. (1997) Protective effect of encapsulated cells producing neurotrophic factor CNTF in a monkey model of Huntingtons disease. Nature 386: 395399. Sagot Y, Tan SA, Baetge E, Schmalbruch H, Kato AC, et al. (1995) Polymer encapsulated cell lines genetically engineered to release ciliary neurotrophic factor can slow down progressive motor neuronopathy in the mouse. Eur J Neurosci 7: 13131322. Tuszynski MH, Smith DE, Roberts J, McKay H, Mufson E (1998) Targeted intraparenchymal delivery of human NGF by gene transfer to the primate
85
1415

basal forebrain for 3 months does not accelerate beta-amyloid plaque deposition. Exp Neurol 154: 573582. Tuszynski MH, Thal L, Pay M, Salmon DP, U HS, et al. (2005) A phase 1 clinical trial of nerve growth factor gene therapy for Alzheimer disease. Nat Med 11: 551555. Aebischer P, Schluep M, Deglon N, Joseph JM, Hirt L, et al. (1996) Intrathecal delivery of CNTF using encapsulated genetically modied xenogeneic cells in amyotrophic lateral sclerosis patients. Nat Med 2: 696 699. Sieving PA, Caruso RC, Tao W, Coleman HR, Thompson DJ, et al. (2006) Ciliary neurotrophic factor (CNTF) for human retinal degeneration: phase I trial of CNTF delivered by encapsulated cell intraocular implants. Proc Natl Acad Sci U S A 103: 38963901. Simons JW, Mikhak B, Chang JF, DeMarzo AM, Carducci MA, et al. (1999) Induction of immunity to prostate cancer antigens: Results of a clinical trial of vaccination with irradiated autologous prostate tumor cells engineered to secrete granulocyte-macrophage colony-stimulating factor using ex vivo gene transfer. Cancer Res 59: 51605168. Hama E, Shirotani K, Iwata N, Saido TC (2004) Effects of neprilysin chimeric proteins targeted to subcellular compartments on amyloid beta peptide clearance in primary neurons. J Biol Chem 279: 3025930264. Russo R, Borghi R, Markesbery W, Tabaton M, Piccini A (2005) Neprylisin decreases uniformly in Alzheimers disease and in normal aging. FEBS Lett 579: 60276030. Yasojima K, Akiyama H, McGeer EG, McGeer PL (2001) Reduced neprilysin in high plaque areas of Alzheimer brain: A possible relationship to decient degradation of beta-amyloid peptide. Neurosci Lett 297: 97100. Hemming ML, Selkoe DJ (2005) Amyloid beta-protein is degraded by cellular angiotensin-converting enzyme (ACE) and elevated by an ACE inhibitor. J Biol Chem 280: 3764437650. Citron M, Oltersdorf T, Haass C, McConlogue L, Hung AY, et al. (1992) Mutation of the beta-amyloid precursor protein in familial Alzheimers disease increases beta-protein production. Nature 360: 672674. Podlisny MB, Ostaszewski BL, Squazzo SL, Koo EH, Rydell RE, et al. (1995) Aggregation of secreted amyloid beta-protein into sodium dodecyl sulfatestable oligomers in cell culture. J Biol Chem 270: 95649570. Dull T, Zufferey R, Kelly M, Mandel RJ, Nguyen M, et al. (1998) A thirdgeneration lentivirus vector with a conditional packaging system. J Virol 72: 84638471. Hemming ML, Selkoe DJ, Farris W (2007) Effects of prolonged angiotensinconverting enzyme inhibitor treatment on amyloid b-protein metabolism in mouse models of Alzheimer disease. Neurobiol Dis 26: 273281. Florentin D, Sassi A, Roques BP (1984) A highly sensitive uorometric assay for enkephalinase, a neutral metalloendopeptidase that releases tyrosineglycine-glycine from enkephalins. Anal Biochem 141: 6269. 41. Johnson-Wood K, Lee M, Motter R, Hu K, Gordon G, et al. (1997) Amyloid precursor protein processing and A beta42 deposition in a transgenic mouse model of Alzheimer disease. Proc Natl Acad Sci U S A 94: 1550 1555. 42. Mucke L, Masliah E, Yu GQ, Mallory M, Rockenstein EM, et al. (2000) Highlevel neuronal expression of abeta 142 in wild-type human amyloid protein precursor transgenic mice: Synaptotoxicity without plaque formation. J Neurosci 20: 40504058. 43. Abramoff MD, Magelhaes PJ, Ram SJ (2004) Image processing with image J. Biophotonics International 11: 3642. 44. Iwata N, Tsubuki S, Takaki Y, Watanabe K, Sekiguchi M, et al. (2000) Identication of the major Abeta142-degrading catabolic pathway in brain parenchyma: Suppression leads to biochemical and pathological deposition. Nat Med 6: 143150. 45. Huang SM, Mouri A, Kokubo H, Nakajima R, Suemoto T, et al. (2006) Neprilysin-sensitive synapse-associated amyloid-beta peptide oligomers impair neuronal plasticity and cognitive function. J Biol Chem 281: 1794117951. 46. Lafrance MH, Vezina C, Wang Q, Boileau G, Crine P, et al. (1994) Role of glycosylation in transport and enzymic activity of neutral endopeptidase24.11. Biochem J 302: 451454. 47. Tuszynski MH, Senut MC, Ray J, Roberts J (1994) Somatic gene transfer to the adult primate central nervous system: In vitro and in vivo characterization of cells genetically modied to secrete nerve growth factor. Neurobiol Dis 1: 6778. 48. Wen J, Vargas AG, Ofosu FA, Hortelano G (2006) Sustained and therapeutic levels of human factor IX in hemophilia B mice implanted with microcapsules: Key role of encapsulated cells. J Gene Med 8: 362369. 49. Tobias CA, Shumsky JS, Shibata M, Tuszynski MH, Fischer I, et al. (2003) Delayed grafting of BDNF and NT-3 producing broblasts into the injured spinal cord stimulates sprouting, partially rescues axotomized red nucleus neurons from loss and atrophy, and provides limited regeneration. Exp Neurol 184: 97113. 50. Tuszynski MH (2002) Growth-factor gene therapy for neurodegenerative disorders. Lancet Neurol 1: 5157. 51. Li C, Booze RM, Hersh LB (1995) Tissue-specic expression of rat neutral endopeptidase (neprilysin) mRNAs. J Biol Chem 270: 57235728. 52. Citron M (2004) Strategies for disease modication in Alzheimers disease. Nat Rev Neurosci 5: 677685. 53. Simard AR, Soulet D, Gowing G, Julien JP, Rivest S (2006) Bone marrowderived microglia play a critical role in restricting senile plaque formation in Alzheimers disease. Neuron 49: 489502. 54. Stalder AK, Ermini F, Bondol L, Krenger W, Burbach GJ, et al. (2005) Invasion of hematopoietic cells into the brain of amyloid precursor protein transgenic mice. J Neurosci 25: 1112511132.
28. 29.
30.
31.
32. 33. 34. 35. 36. 37. 38. 39. 40.
Editors Summary
Background. Alzheimer disease is the most common form of dementia and is increasingly common as people age. It may affect up to half of the population in people over 85 years of age. The classic pathological features of Alzheimer disease in the brain were described 100 years ago by a German neuropathologist, Alois Alzheimer. They consist of plaques, which are made up of a protein called amyloid-b protein (shortened to Ab); and tangles of another protein, called tau. These proteins accumulate especially in areas of the brain where memory and thought are processed and are believed to be toxic to neurons. There are a number of inherited forms of Alzheimer disease which are caused by mutations either in the protein from which Ab is derived, called betaamyloid precursor protein (shortened to APP) or in other proteins that act to release the Ab protein from APP. Research on these inherited forms of Alzheimer disease has helped in the understanding of how plaques accumulate, which has subsequently led to new potential approaches to the treatment of Alzheimer disease such as lowering the production of Ab from APP or enhancing clearance of the plaques. Why Was This Study Done? The researchers here wanted to investigate the use of a type of gene therapy called ex vivo (which means out of body) gene therapy, in which cells are taken from an individual, genetically altered, then put back into the individual they were taken from. This approach has already been studied in some human diseases and conditions including hemophilia, cancer, and spinal cord injury. The researchers here wanted to investigate whether they could use this approach to deliver to the brain one of the proteases that breaks down amyloid plaques. What Did the Researchers Do and Find? The researchers used a mouse model of Alzheimer disease in which, as the mice age, they develop plaques in the brain made of the same protein found in human Alzheimer disease. The researchers took a particular type of cell from the mice called a fibroblast, used a virus called a lentivirus to insert into the PLoS Medicine | www.plosmedicine.org 1416 cells a protease called neprilysin that can degrade Ab, and then injected these altered cells into the brains of the mice. They then compared results of these experiments with what happened when cells with a control (an inactive virus) were injected. They found that when the active protease was put into the brains of mice there was a substantial clearing of plaques especially in the areas close to the injection site, compared with the control mice. What Do These Findings Mean? These results suggest that this ex vivo gene therapy approach to Alzheimer disease is worth considering further. However, despite showing that the plaques were cleared the researchers did not show whether or not there was any effect on the behavior of the mice, i.e., whether there was any effect on the symptoms that the plaques cause. In addition, before this approach could be used in humans more work would need to be done, including showing that the lentivirus and the protease used were safe in humans, and the injection could be scaled up to the much larger human brain. Additional Information. Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed. 0040262. The slides from the original cases that Alois Alzheimer described are available online Current research is discussed by the Alzheimer Research Forum News and information for researchers, doctors, and patients, including a 24-hour help line, are available from the Alzheimers Association Medline Plus, the health information site for patients from the US National Library of Medicine, has a page of links on Alzheimer disease The US National Institutes of Health National Institute of Aging has a fact sheet on Alzheimer disease, which is also available in Spanish The UKs National Health Service online information site has information on Alzheimer disease August 2007 | Volume 4 | Issue 8 | e262
86

Open access, freely available online
PLoS MEDICINE
Persistent Amyloidosis following Suppression of Ab Production in a Transgenic Model of Alzheimer Disease

Joanna L. Jankowsky1,2*, Hilda H. Slunt1, Victoria Gonzales1, Alena V. Savonenko1, Jason C. Wen3, Nancy A. Jenkins4, Neal G. Copeland4, Linda H. Younkin5, Henry A. Lester2, Steven G. Younkin5, David R. Borchelt1,6*
1 Department of Pathology, The Johns Hopkins University School of Medicine, Baltimore, Maryland, United States of America, 2 Division of Biology, California Institute of Technology, Pasadena, California, United States of America, 3 Department of Cell Biology, The Johns Hopkins University School of Medicine, Baltimore, Maryland, United States of America, 4 Mouse Cancer Genetics Program, National Cancer Institute Frederick Cancer Research and Development Center, Frederick, Maryland, United States of America, 5 Mayo Clinic Jacksonville, Jacksonville, Florida, United States of America, 6 Department of Neuroscience, The Johns Hopkins University School of Medicine, Baltimore, Maryland, United States of America Competing Interests: The authors have declared that no competing interests exist. Author Contributions: JLJ and DRB designed the study and wrote the manuscript, JLJ, HHS, VG, JCW, LHY, SGY and DRB performed experiments, NAJ and NGC generated transgenic founders, HAL assisted with data interpretation, and AVS performed statistical analyses. Academic Editor: Adriano Aguzzi, Zurich University, Switzerland Citation: Jankowsky JL, Slunt HH, Gonzales V, Savonenko AV, Wen JC, et al. (2005) Persistent amyloidosis following suppression of Ab production in a transgenic model of Alzheimer disease. PLoS Med 2(12): e355. Received: June 14, 2005 Accepted: August 22, 2005 Published: November 15, 2005 DOI: 10.1371/journal.pmed.0020355 Copyright: 2005 Jankowsky et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: Ab, amyloid-b; AD, Alzheimer disease; APLP, amyloid precursorlike protein; APP, amyloid precursor protein; CaMKIIa, calciumcalmodulin kinase IIa; dox, doxycycline; FA, formic acid; GFAP, glial fibrillary acidic protein; mo/ huAPP695, mouse APP with a humanized Ab domain; PBS, phosphate-buffered saline; SDS, sodium dodecyl sulfate; swe/ind, Swedish/Indiana; tTA, tetracycline transactivator *To whom correspondence should be addressed. E-mail: jlj2@caltech.edu (JLJ); borchelt@mbi.ufl.edu (DRB) Current address: Department of Neuroscience, McKnight Brain Institute, University of Florida, Gainesville, Florida, United States of America
ABSTRACT
Background
The proteases (secretases) that cleave amyloid-b (Ab) peptide from the amyloid precursor protein (APP) have been the focus of considerable investigation in the development of treatments for Alzheimer disease. The prediction has been that reducing Ab production in the brain, even after the onset of clinical symptoms and the development of associated pathology, will facilitate the repair of damaged tissue and removal of amyloid lesions. However, no longterm studies using animal models of amyloid pathology have yet been performed to test this hypothesis.

We have generated a transgenic mouse model that genetically mimics the arrest of Ab production expected from treatment with secretase inhibitors. These mice overexpress mutant APP from a vector that can be regulated by doxycycline. Under normal conditions, high-level expression of APP quickly induces fulminant amyloid pathology. We show that doxycycline administration inhibits transgenic APP expression by greater than 95% and reduces Ab production to levels found in nontransgenic mice. Suppression of transgenic Ab synthesis in this model abruptly halts the progression of amyloid pathology. However, formation and disaggregation of amyloid deposits appear to be in disequilibrium as the plaques require far longer to disperse than to assemble. Mice in which APP synthesis was suppressed for as long as 6 mo after the formation of Ab deposits retain a considerable amyloid load, with little sign of active clearance.
Conclusion
This study demonstrates that amyloid lesions in transgenic mice are highly stable structures in vivo that are slow to disaggregate. Our findings suggest that arresting Ab production in patients with Alzheimer disease should halt the progression of pathology, but that early treatment may be imperative, as it appears that amyloid deposits, once formed, will require additional intervention to clear.
1318
December 2005 | Volume 2 | Issue 12 | e355
87
PersistentAmyloidosisfollowingSuppressionofAbProductioninaTransgenicModelofAlzheimerDisease
Persistent Amyloid in Tet-Off APP Mice
Introduction
Over a decade ago the amyloid cascade hypothesis predicted that increased levels of amyloid-b (Ab) peptide lead to secondary pathologies that ultimately culminate in the onset of Alzheimer disease (AD) [1]. Early support for this hypothesis came from genetic studies linking early-onset AD to mutations in the amyloid precursor protein (APP), from which Ab is derived, and presenilins 1 and 2, which are interchangeable components of a endoprotease complex that releases Ab from APP (for review see [2,3]). If, as predicted, overproduction of Ab initiates the cascade of events leading to disease, then therapeutic strategies that lower Ab levels should either arrest or reverse the progression from peptide to dementia. Early evidence from clinical trials of antibody-mediated clearance, one of the rst Ab-lowering approaches tested in humans, suggested that treatments designed to reduce amyloid burden may indeed be benecial. Although the trials were halted because of adverse effects in a subset of volunteers [4,5], further analysis of several patients found evidence that amyloid pathology, and to a lesser degree cognitive function, was improved in proportion to the patients titer of Ab-specic antibody [6,7]. While this approach is promising, constant exposure to antibodies that recognize an epitope highly enriched in the brain may have unexpected side effects that will limit its long-term use. An alternative approach that is being actively pursued for future treatment of AD seeks to lower Ab levels by limiting its production from the precursor protein APP. Peptide Ab is released from APP by the action of two enzymes, the bAPP cleaving enzyme 1 (BACE1) and c-secretase, which cleave the holoprotein at the N- and C-termini of Ab, respectively. Several inhibitors of c-secretase have already been produced [8,9], and small molecule inhibitors of b-APP cleaving enzyme 1 are currently being developed [10,11]. The long-term effectiveness of this approach in either humans or model systems, however, has not been reported. Although loss of b-APP cleaving enzyme 1 function can prevent the development of plaques in transgenic mouse models for AD (F. Laird, H. Cai, P. C. Wong, personal communication), it is not known whether the brain can clear pre-existing amyloid deposits once production of Ab has been suppressed. Clearly, the amyloid-lowering approach should be rigorously examined in animal models before these reagents are tested in patients. However, the chemical secretase inhibitors most likely to reach human trials are still in development. Therefore, we developed a mouse model of Alzheimer-type amyloid that expresses a controllable APP transgene. This system, commonly known as the tet-off system, can be regulated by analogs of tetracycline administered in food or water [12,13]. The strong expression levels produced with the tet-off vectors, combined with the ability to reduce this expression by several orders of magnitude with tetracycline [14], allowed for a stringent test of how a highly effective pharmaceutical inhibitor of Ab production would impact the progression of amyloid pathology and whether reversal of these lesions might be possible following such treatment.
Methods Transgene Construction

We created a tetracycline-responsive chimeric mouse/ human APP695Swedish/Indiana (swe/ind) vector by replacing the promoter region of the moPrP.XhoI vector (also known as pPrPpE1/E2,3sal [15]) with the tetracycline-responsive promoter of pTetSplice (Life Technologies, Rockville, Maryland, United States), and then ligating mouse APP with a humanized Ab domain (mo/huAPP695) cDNA into the new vector. We began by cloning the tetracycline-responsive promoter (bp 6481) from pTetSplice by PCR using primers that added external BamHI and NotI sites to the 59 end and a BamHI site to the 39 end, while destroying XhoI and BamHI sites within the promoter (forward: GCC GGA TCC GCG GCC GCC GTC GAG TTT ACC ACT CCC TAT C; reverse: GCC GGA TCC ACT CTA GAA GAT CCC CGG GTA CCG). We then isolated the moPrP.XhoI intron by amplication with primers that added an external BamHI site to the 59 end of exon 1 and ran through the Asp718 site of exon 2 (forward: GCC GGA TCC GAT CAG CAG ACC GAT TCT GG; reverse: GCC GGT ACC ACT AGG AAG GCA GAA TGC). This 2-kb intron fragment was cloned into a TA cloning vector (Invitrogen, Carlsbad, California, United States), then excised by Asp718 digestion and ligated to the 6.8-kb Asp718 fragment of moPrP.XhoI containing exon 2, exon 3, the 39 UTR, and pBluescript to generate an intermediate vector with all three exons and a central intron but no promoter. This vector was then opened at the BamHI site introduced by the intron cloning primer, and ligated to the 0.5-kb BamHIcut tetracycline promoter fragment. This ligation generated a 9.3-kb vector encoding the tetracycline promoter from pTetSplice with two exons, one intron, and the original 39 UTR of the moPrP.XhoI vector, all carried in the pBluescript cloning vector. We incorporated the Swedish (KM570/571NL) and Indiana (V617F) mutations into the mo/huAPP695 cDNA (in BS-KS) by PCR using a four-primer strategy: rst, two partially overlapping products were generated in separate reactions using primers that encode the desired mutations (Swedish forward: GGA GAT CTC TGA AGT GAA TCT GGA TGC AGA ATT CCG/Indiana reverse: GGG TGA TGA AAA TCA CGG TTG C; Indiana forward: CAA CCG TGA TTT TCA TCA CCC TGG/M13 reverse). The two PCR products were ligated, digested with BglII and ApaI and cloned back into the original mo/huAPP695-BS-KS vector. Finally, the new APP695swe/ind was subcloned into the XhoI site of the moPrP-tetP vector from above to complete the construct.
Pronuclear Injection, Screening of Founders, and Maintenance of the Lines

The moPrP-tetP-mo/huAPP695swe/ind vector was linearized and the pBluescript domain excised by digestion with NotI. The puried vector was injected into the pronucleus of fertilized eggs from C57BL/6J 3 C3HeJ F1 matings. Founder animals were screened for the presence of the transgene by three-way PCR using the S36 and PrP-S/PrP-AS primers described below. Transgene-positive founders were bred to animals expressing the tetracycline transactivator (tTA) under control of the calcium-calmodulin kinase IIa (CaMKIIa) promoter obtained from Jackson Laboratory [16]
88

(Bar Harbor, Maine, United States; stock # 3010; B6;CBATgN[Camk2a-tTA]1Mmay). The colony was thereafter maintained by crossing single transgenic tTA and APP offspring for each of the four APP lines. All mice were provided fresh food and water ad libitum. Animal protocols were approved by both the Johns Hopkins University and the California Institute of Technology Institutional Animal Care and Use Committees.
Doxycycline Administration
Doxycycline (dox) was administered through commercially available dox-containing chow (BioServ, Frenchtown, New Jersey, United States). The chow contained 200 mg/kg of antibiotic; based on estimated consumption of 5 g per mouse per day, the expected dose to each animal was 1 mg dox per day. The average 25-g animal therefore received 40 lg of dox per gram body weight per day. Chow was changed 12 times per week to prevent breakdown of the antibiotic.
Genotyping
Offspring were genotyped for the presence of each transgene by PCR amplication of genomic DNA extracted from a 5-mm tail biopsy. Tails were heated to 95 8C for 45 min in 250 ll of 50 mM NaOH, vortexed, then neutralized with an equal volume of 0.5 M Tris-HCl (pH 5.5). Debris was sedimented by centrifugation, and 3 ll of supernatant was used for amplication. Genotyping for APP and tTA transgenes was performed in the same PCR reaction, using ve separate primers. APP was amplied using forward primer S36 located in the 39 end of the APP cDNA (CCG AGA TCT CTG AAG TGA AGA TGG ATG) and reverse primer PrP-AS-J located in the 39 UTR of the vector (CCA AGC CTA GAC CAC GAG AAT GC). The tTA transgene was detected using a primer set that amplied across its two subdomains with tet forward located within the Tn10 tetracycline repressor (CGC TGT GGG GCA TTT TAC TTT AG) and tet reverse within the HSV1 VP16 (CAT GTC CAG ATC GAA ATC GTC). All reactions, whether transgenepositive or not, amplied a segment of the endogenous prion protein gene as a control for DNA quality using a forward primer, PrP-S-J, specic to the mouse PrP open reading frame (GGG ACT ATG TGG ACT GAT GTC GG) and a reverse primer, PrP-AS-J, shared by the 39 UTR of the endogenous PrP gene and the transgene vector. Amplication reactions were run for 37 cycles at 94 8C for 30 s, 64 8C for 1 min, and 72 8C for 1 min. All samples, transgenic and wild-type, gave a 750-bp product from the endogenous PrP gene. The APP transgene yielded an additional band at 400 bp; the tTA product fell in between at 480 bp.
United States) and electrophoresed for several hours in 13 Tris-glycinesodium dodecyl sulfate (13TG-SDS) buffer (6E10 and 22C11; Amresco, Solon, Ohio, United States) or 13 Tristricine-SDS buffer (CT15; Invitrogen, Carlsbad, California, United States). Proteins were transferred overnight to 0.45lm Optitran nitrocellulose (Schleicher and Schuell, Keene, New Hampshire, United States) in 13 TG buffer (Amresco). Blots were blocked in PBS containing 5% nonfat dry milk powder, and incubated for 3 h at room temperature in blocking solution with one of the following antibodies: mouse monoclonal 22C11 (kind gift of Konrad Beyreuther and Andreas Weidemann; [17]) diluted 1:1,000, mouse monoclonal 6E10 (Signet Laboratories, Dedham, Massachusetts, United States) diluted 1:2,500, rabbit polyclonal anti-superoxide dismutase 1 (m/hSOD1) [18] diluted 1:2,500 to 1:4,000, or rabbit polyclonal CT15 (kind gift of Ed Koo; [19]) diluted 1:1,000. Subsequently, the blots were washed with PBS containing 0.1% Tween-20, and then incubated with either goat anti-mouse or goat anti-rabbitHRP conjugated secondary antibody diluted 1:1,000 in blocking solution. After several additional rinses in PBS with 0.1% Tween-20, blots were developed with enhanced chemiluminescence reagent and imaged with the Bio-Rad Molecular Imager FX system. Staining intensity within each lane was quantied using the Quantity One image analysis software (Molecular Imager FX, Bio-Rad Laboratories). Background was calculated from across the image and subtracted from the entire le. The signal intensity for each band (corrected signal intensity 3 pixel number) was then calculated using the Volume report tool.
Slot Blot mRNA Analysis

Five micrograms per sample of total RNA extracted from fresh-frozen brain, liver, kidney, heart, lung, spleen, and skeletal muscle was vacuum-ltered through 0.45-lm Optitran nitrocellulose. After several washes through the manifold with 103 SSC, blots were UV-cross-linked and probed with a radiolabeled ;350-bp BglIIXhoI cDNA fragment of mo/ huAPP695 cDNA. After hybridizing overnight at 65 8C in 1% BSA/1 mM EDTA/0.5 M sodium phosphate buffer (pH 7.2)/7% SDS [20], the blots were washed twice at 65 8C for 30 min each in 0.1% BSA/1 mM EDTA/40 mM sodium phosphate buffer (pH 7.2)/5% SDS before two nal 30-min washes at 65 8C with 1 mM EDTA/40 mM sodium phosphate buffer (pH 7.2)/1% SDS. Blots were wrapped wet and exposed to phosphorscreens overnight at room temperature.
Amyloid Histology
Mice were euthanized by ether inhalation and brains removed for immersion xation in 4% paraformaldehyde/ 13 PBS. After 48 h in xative at 4 8C, brains were transferred to PBS, dehydrated in alcohols, treated with cedarwood oil and methylsalicylate, and embedded in parafn for sectioning. Hirano silver stain. Silver impregnation histology was performed on 10-lm parafn-embedded sections by Hiranos modication of the Bielschowsky method [21]. Briey, sections were deparafnized through xylene and alcohols into tap water before being placed into fresh 20% silver nitrate solution for 20 min. After being washed thoroughly with distilled water, slides were immersed in 20% silver nitrate solution titrated with fresh ammonium hydroxide.
1320 December 2005 | Volume 2 | Issue 12 | e355
Immunoblotting/Quantitation
Frozen cortical or whole forebrain tissue was homogenized by sonication in ve volumes of phosphate-buffered saline (PBS) with 5 mM EDTA and protease inhibitors (Mammalian cell cocktail, Sigma, St. Louis, Missouri, United States), using a probe sonicator set to 50% output (TEKMAR, Cincinnati, Ohio, United States). After dilution with an equal volume of PBS/EDTA/protease inhibitor, the samples were centrifuged briey and the supernatant used for analysis. Fifty micrograms (6E10 and CT15) or 5 lg (22C11) of brain homogenate was loaded per lane onto 7.5%, 10%20%, or 4%20% TrisHCl PAGE gels (Bio-Rad Laboratories, Hercules, California,
89
After 20 min, slides were washed with ammonia water before being individually developed with 100 ll of developer (20 ml of 37% formaldehyde, 100 ml of distilled water, 50 ll of concentrated nitric acid, and 0.5 g of citric acid) added to 50 ml of titrated silver nitrate solution. Slides were then rinsed in tap water, xed in 5% sodium thiosulfate, and dehydrated through alcohols and xylene. Thioavin-S staining. Following deparafnization of sections through xylene and alcohols, amyloid impregnation with thioavin-S was performed according to the Guntern modication of the standard protocol. Slides holding 10-lm parafn sections were washed twice in distilled water, then immersed for 5 min in a 0.25% potassium permanganate solution, followed by 5 min in a 1% potassium metabisulfate/ 1% oxalic acid solution. After this preparation, slides were placed into a ltered aqueous 0.02% thioavin-S solution (Chroma-Gesellschaft Schmid, Kongen, Germany) for 8 min. Excess stain was removed by two brief rinses in 80% ethanol, then two in distilled water, after which slides were nished in aqueous mounting medium for orescence photomicrography. Ubiquitin, glial brillary acidic protein, and Ab immunohistochemistry. Prior to immunostaining, slides were deparafnized by oven heating followed by immersion in xylene. After rehydration through graded alcohols into tap water, endogenous peroxidase activity was quenched by incubation with 3% hydrogen peroxide in methanol. Slides were microwaved for 57 min in water, cooled for 5 min, then washed in TBS. Nonspecic staining was blocked for 1 h with 3% normal goat serum and 0.1% Triton-X 100 in TBS. Slides were then placed into primary antibody (rabbit anti-Ab peptide polyclonal antibody, Zymed Laboratories, South San Francisco, California, United States; rabbit anti-ubiquitin and rabbit antiglial brillary acidic protein (GFAP) polyclonal antibodies, Dako, Carpinteria, California, United States) diluted 1:500 in TBS with 2% normal goat serum and incubated overnight at room temperature. After being washed of excess primary antibody with several changes of TBS, slides were incubated with either the Vectastain Elite anti-rabbit secondary system (anti-Ab; Vector Laboratories, Burlingame, California, United States) or peroxidase/antiperoxidase reagents (anti-ubitquitin and anti-GFAP; Sternberger Monoclonals, Lutherville, Maryland, United States) according to the manufacturers directions. Antibody binding was visualized with diaminobenzidene, and sections were counterstained with hematoxylin.
Tween-20, before antibody binding was detected with enhanced chemiluminescence (PerkinElmer, Boston, Massachusetts, United States). Digital images of each blot were captured with a Molecular Imager FX gel documentation system, and the intensity of Ab staining was quantied using Quantity One image analysis software.
Ab ELISA
An aliquot of cortical homogenate generated for Western analysis described above was subjected to a three-step sequential extraction using PBS, 2% SDS, and 70% formic acid (FA). At each step, the sample was sonicated in appropriate buffer and centrifuged at 100,000g for 30 min (1- to 1.5-mo samples) or 60 min (6- to 12-mo samples) at 4 8C as previously described [2325]. The supernatant was removed for analysis, and the pellet was sonicated in the next solution in sequence. The 2% SDS extracts were diluted in EC buffer, and the FA extracts neutralized with 1M Trisphosphate buffer (pH 11) then diluted with EC buffer prior to testing. Human Ab was measured in each fraction using BAN50 for capture (epitope Ab116) and BA27 and BC05 for detection (Ab40 and Ab42, respectively) (Takeda Chemical Industries, Osaka, Japan). Total Ab (mouse human; 1- to 1.5mo samples only) was measured in each fraction using BNT77 for capture (epitope Ab1128) and BA27 and BC05 for detection. All values were calculated as picomoles per gram based on the initial weight of cortical tissue.
Activity Monitoring
Daily basal activity was studied in 28 CaMKIIa-tTA 3 tetAPPswe/ind line 107 mice between 4 and 5 mo of age. Animals were separated into individual cages immediately before the start of each experiment (n 36 per genotype untreated, 25 per genotype dox-reared). The cages were placed inside activity-monitoring frames designed to count every time the animal passed through one of three photobeams spanning the width of the cage (San Diego Instruments, San Diego, California, United States). Experiments were started midway through the light phase of the day, and data were collected in 1-h bins for the following 48 h. Testing rooms were maintained on the same 13:11 h day:night cycle as the main animal housing areas and were closed to entry during the experiment.
Statistical Analyses
Statistical analyses of protein expression, ELISA data, and lter trap assays were performed by ANOVA with Tukeys honest signicant difference post-hoc test applied to signicant main effects or interactions (Statistica 6.0, StatSoft, Tulsa, Oklahoma, United States). In cases of positively skewed data distribution, log10(x 0.5) transformation was applied to the raw data before submitting them to ANOVA.
Filter Trap Assay

An aliquot of each cortical homogenate used for Western blotting above was partially solubilized by the addition of SDS to a nal concentration of 1%. Serial 1:2 dilutions were made with 13 PBS/1% SDS, and 100 ll of each dilution was then vacuum-ltered through a pre-wet 0.22-lm cellulose acetate membrane (Schleicher and Schuell) [22]. Each well was washed several times with PBS, after which blots were incubated overnight with polyclonal anti-Ab antibody (Zymed Laboratories) diluted 1:600 in a blocking solution of 13 TBS/5% nonfat dry milk powder. After washing the blots three times for 10 min each in 13 TBS/0.1% Tween-20, the membrane was incubated for 1 h with HRP-conjugated protein A (Sigma) diluted 1:5,000 in blocking solution. The membranes were again washed three times with 13 TBS/0.1%
Results
We used the tet-off transgene system to express a double mutant version of chimeric mo/huAPP695 (swe/ind KM570, 571NL, and V617F) from a tetracycline-responsive promoter [12,13]. Transgenic APP expression was activated by crossing the APPswe/ind mice to animals producing tTA under control of the CaMKIIa promoter [16]. After initial screening of
90

founders, we identied four lines of tet-APPswe/ind mice that produced very high levels of transgene product in offspring coexpressing tTA (Figures 1 and S1). Compared to a standard APP transgenic line used for previous amyloid studies by our laboratory (line C33; [15,26,27]), we estimated that the four controllable lines produce transgenic APP protein at 10- to 30-fold over endogenous levels (Figure S1). This estimate was conrmed by direct comparison of APP levels in nontransgenic and tet-off APP mice using an antibody that recognizes both endogenous APP (and amyloid precursor-like protein 2) and the transgenic protein (monoclonal antibody 22C11; Figure 1D). Importantly, all four new lines of tet-off APP mice showed nearly complete suppression of the transgene following dox treatment (Figures 1 and S1). We focused on one of the four lines, line 107, to examine in more detail the time dependence and extent of transgene suppression following either acute or chronic treatment with dox. Two dox-treated groups were compared to two untreated groups: one group of mice was born and raised on dox, a second group was treated with dox for 2 wk starting at 1 mo of age (4 wk 2 wk dox); two untreated groups kept on normal chow were harvested at either 4 or 6 wk of age. Animals born and raised on dox
harbored no transgenic APP (Figure 1A). Following as little as 2 wk of dox treatment, transgenic APP expression was reduced by more than 95% compared to pre-dox levels. The residual expression remaining in acutely treated mice represents less than 4% of the transgenic protein produced in the absence of dox (Figure 1C), and likely results from slight leakage at the level of transcription (data not shown). Importantly, the total amount of APP (endogenous plus transgenic) and related APLPs in both acute and chronically treated animals was statistically indistinguishable from that in nontransgenic mice (Figure 1D; statistical analyses for experiments throughout the study are presented in the accompanying gure legends). To ensure that Ab production was suppressed in concert with the dox-mediated inhibition of its precursor APPswe/ ind, we measured Ab40 and Ab42 levels by ELISA in forebrain homogenates from young tet-off animals. At 1 mo of age, the mice lacked visible amyloid aggregates that might act as an intractable reservoir of peptide remaining in the brain after the transgene had been suppressed. To further ensure we could detect any such insoluble aggregates that might bias our measure of changes in peptide synthesis, we performed a sequential three-step extraction with PBS, 2% SDS, and 70%
Figure 1. Control of Transgenic APP Expression by Dox (A) Western blotting for transgenic APP using the human-specific 6E10 antibody shows expression of full-length protein in forebrain tissue from young predeposit double transgenic animals (line 107) and its suppression following dox treatment. Untreated animals show high levels of transgene expression; protein levels drop dramatically in animals acutely treated with dox for 2 wk. A faint, but detectable band of full-length protein remains in acutely treated animals that can be eliminated in mice born and raised on dox. (B) Immunoblotting with the N-terminal antibody 22C11 to detect both transgenic and endogenous protein shows that dox treatment reduces APP/ APLP to levels found in nontransgenic mice. (C) Measurement of signal intensity from the Western blot in (A) shows transgenic APP levels are decreased more than 95% by dox in both acutely and chronically treated animals (97.2% for 4 wk 2 wk dox, 98.0% for reared on dox versus 4 wk untreated; ANOVA effect of treatment group F4,15 85.55, p , 0.001). APP levels in 4 wk 2 wk dox, reared on dox, and nontransgenic (NTg) were not significantly different (p . 0.9, Tukey post-hoc test). (D) Measurement of signal intensity from the Western blot in (B) shows total APP/APLP levels in dox-treated tTA/APP mice are significantly lower than in 4-wk-old untreated mice (ANOVA effect of treatment group F4,15 84.41, p , 0.001) and indistinguishable from those of nontransgenic animals (p . 0.9, Tukey post-hoc test). *, p , 0.001 versus untreated 4-wk-old mice, Tukey post-hoc test applied to significant effect of group ANOVA. DOI: 10.1371/journal.pmed.0020355.g001 PLoS Medicine | www.plosmedicine.org 1322 December 2005 | Volume 2 | Issue 12 | e355
91
Persistent Amyloid in Tet-Off APP Mice Figure 2. Ab Levels Are Dramatically Reduced by Transgene Suppression Cortical homogenates from young, predeposit tTA/APP mice used for Western blot in Figure 1 (line 107) were fractionated by sequential multistep extraction with PBS, 2% SDS, and 70% FA followed by humanspecific Ab ELISA to measure transgene-derived peptide in each fraction. Ab40 is shown in white, Ab42 in black. (A) PBS-soluble Ab levels are substantially reduced by both acute and chronic dox treatment (ANOVA, effect of treatment group F4,24 137.10 and 386.01, p , 0.001, for Ab40 and Ab42, respectively). Ab levels in treated animals are indistinguishable from nontransgenic (NTg) animals (p . 0.3, Tukey post-hoc test). (B) In the young animals tested here prior to the formation of visible amyloid deposits, most Ab is extracted into the SDS fraction (84% and 76% of all transgene-derived Ab40 and Ab42, respectively). As in the PBSsoluble fraction, Ab levels in the SDS fraction are significantly lowered by dox treatment compared to untreated animals (ANOVA effect of group F4,24 197.57 and 163.48, p , 0.001, for Ab40 and Ab42, respectively). Acutely treated animals retained a small (although significant) amount of residual peptide (p , 0.001 compared to nontransgeinc, Tukey post-hoc test), whereas Ab levels in mice born and raised on dox were reduced to levels indistinguishable from nontransgenic (p . 0.8, Tukey post-hoc test). (C) The FA-soluble fraction already contains a small but significant pool of aggregated Ab42 in untreated animals by 4 wk of age (p , 0.05 versus nontransgenic; Tukey post-hoc test applied to significant effect of group ANOVA F4,24 17.11, p , 0.001). By 6 wk of age, the amount of Ab in the FA fraction is increased significantly preceding the appearance of visible deposits 2 wk later. The FA pool is the only peptide fraction not lowered by acute dox treatment (4 wk untreated 4 wk 2 wk dox, p . 0.9, Tukey post-hoc test), consistent with poor turnover of aggregated Ab species. (D) Measurements of total Ab, including both endogenous and transgene-derived peptides, show that animals born and raised on dox harbor Ab levels identical to nontransgenic animals (p . 0.9, Tukey post-hoc test, effect of group ANOVA F4,24 39.13 and 35.29, p , 0.001, for Ab40 and Ab42, respectively). Whereas chronic transgene suppression fully prevents synthesis of both peptides, acute dox treatment fully suppresses Ab40 levels (p . 0.8, Tukey post-hoc test), but leaves a small amount of nonsuppressed Ab42. The residual Ab42 observed in acutely treated young animals derives from uncleared aggregates extracted in the SDS and FA fractions. *, p , 0.05; **, p , 0.005; ***, p , 0.001 versus 4-wk-old untreated mice, Tukey post-hoc applied to significant effect of group ANOVA. DOI: 10.1371/journal.pmed.0020355.g002
92
FA that would separate peptides by solubility. We compared the levels of human transgene-derived Ab40 and Ab42 in untreated mice at 4 and 6 wk of age to animals that had either been born and raised on dox or that had been left untreated for 4 wk and then placed on dox chow for 2 wk prior to harvest (the same groups described above for immunoblot analysis of APPswe/ind levels, line 107). Consistent with the reduction in full-length APPswe/ind synthesis shown by immunoblot (see Figure 1), we found that transgene-derived Ab levels were completely suppressed in animals born and raised on dox, and were sharply reduced following acute (2 wk) antibiotic treatment. Compared to the levels in untreated 4-wk-old mice, PBS-soluble Ab42 dropped by 95.2% following 2 wk of dox treatment and by 99.2% with chronic treatment (Figure 2A). Similarly, SDS-soluble Ab42 decreased by 75.2% and 94.8% following 2-wk or lifelong dox treatment (Figure 2B). Only the FA fraction revealed a small dox-resistant pool of peptide in acutely treated animals that we believe represents stable predeposit aggregates that have already accumulated by 4 wk of age when treatment was begun (Figure 2C). Indeed, animals that were born and raised on dox did not harbor this reservoir of treatment-resistant peptide, with 96.3% less Ab42 than untreated 4-wk-old mice. Measurement of total Ab in chronically treated mice, including endogenous and transgene-derived peptide, demonstrated that Ab levels in tet-off APP mice were reduced to the level of

endogenous peptide found in nontransgenic animals (Figure 2D). Taken together with the immunoblotting data for fulllength APPswe/ind, the ELISA measurements indicate that dox-mediated suppression of transgenic APPswe/ind synthesis leads to parallel reduction of Ab levels. The ELISA data also conrmed that incorporation of the Swedish and Indiana mutations led to high levels of Ab42, which we predicted would induce rapid plaque formation in untreated animals. Histological characterization of double transgenic (CaMKIIa-tTA 3 tet-APPswe/ind) mice revealed early-onset amyloid formation in all four new lines. Amyloid plaques were seen in mice as young as 8 wk of age (data not shown). Plaques were limited to the forebrain, including the cortex, hippocampus, olfactory bulb, and striatum, where the CaMKIIa promoter is known to be most active [16,28] (Figure S2). By 6 mo of age, amyloid burden became severe, covering large areas of the cortex and hippocampus (Figure S3). No lesions were seen in the cerebellum or brain stem even at late ages, consistent with CaMKIIa-controlled transgene expression. Unlike what is thought to occur in the human disease, the rst visible plaques in the tet-off APP mice are brillarcored deposits. We have noted the same early appearance of cored deposits in other lines of APP transgenic mice that harbor the Swedish mutation [27]. Diffuse plaques were apparent in 6-mo-old tTA/APP mice, and became relatively abundant by 9 mo of age. At older ages (912 mo) amyloid deposits were visible in the thalamus, which has also been observed in mice expressing mutant APP via the Thy-1 promoter. The presence of amyloid pathology in this region has been attributed to axonal transport of APP/Ab to the terminals of cortical neurons in the thalamus [29]. Most importantly, only double transgenic mice, expressing both the tTA and APP transgenes, developed amyloid lesions. Single transgenic mice up to 15 mo of age showed no sign of pathology (Figure S3). Similarly, amyloid pathology can be completely prevented in double transgenic animals born and raised on dox. Animals from our highest expressing line (line 885) maintained on dox for up to 1 y harbored no amyloid pathology (data not shown), indicating that residual leakage of transgene expression in the presence of dox does not provide sufcient Ab peptide to induce amyloid formation even over long periods. To mimic therapeutic intervention with inhibitors of Ab production, we raised a group of 25 double transgenic mice (CaMKIIa-tTA 3 APP line 107) on normal food until 6 mo of age, when we knew amyloid formation was already well underway in the brain. At 6 mo, half of the animals were switched from normal chow to food containing dox at 200 mg/kg until they were sacriced at 9 or 12 mo of age. The remaining control animals were kept on standard chow (untreated). In all, four cohorts were created: 6 mo untreated (n 7), 9 mo untreated (n 5), 6 mo 3 mo treated (n 8), and 6 mo 6 mo treated (n 5). Full suppression (.95%) of transgenic APPswe/ind levels in the dox-treated animals was conrmed by immunoblot (Figure 3). To ensure that the transgene could be suppressed as rapidly in 6-mo-old mice with fulminant pathology as it can in young, predeposit animals, we treated an additional set of 6-mo-old animals with dox for 1 wk prior to harvest. Importantly, both APPswe/ ind and APPC-terminal fragment levels were fully suppressed after only 1 wk of treatment, indicating that the in
vivo half-life of APPswe/ind and its processed C-terminal fragments are relatively short (Figure 3D). Tissue sections from each animal in the four treatment groups were stained for amyloid pathology by Hirano silver, Campbell-Switzer silver, thioavin-S, and Ab immunohistochemistry. As expected, the 6 mo untreated cohort displayed moderate amyloid pathology, and the 9 mo untreated cohort progressed to a severe amyloid burden. In contrast, the extent of amyloid pathology in mice from the 6 mo 3 mo treated or 6 mo 6 mo treated cohorts closely resembled that of the 6 mo untreated cohort, despite the signicant age difference between the treated and untreated groups (Figures 4 and S3). Well-formed plaques remained in the treated animals after 6 mo of transgene suppression, even though as much time was given to clear the lesions as they had taken to form. Moreover, both types of amyloid, diffuse and brillar, remained intact throughout treatment. Using the Campbell-Switzer silver stain to distinguish different forms of amyloid, we found diffuse plaques were as persistent as cored deposits (Figure S4). It was nevertheless clear that dox-induced suppression of transgenic APP had completely halted the progression of pathology. To conrm that the arrest of plaques without any sign of clearance was not unique to the line 107 mice, we repeated the dox-suppression experiment in a second line of tet-off APP mice (CaMKIIa-tTA 3 tet-APPswe/ind line 18; n 22). Again, long-term dox treatment was begun at 6 mo of age, and mice were harvested after 3 mo of transgene suppression (6 mo untreated, n 8; 9 mo untreated, n 6; 6 mo 3 mo treated, n 8). Immunoblotting for APP conrmed full transgene suppression in the treated animals (Figure S5). As in the line 107 mice described above, amyloid burden worsened substantially in the untreated mice between 6 and 9 mo of age. Suppression of transgene expression abruptly arrested progression of pathology (Figure S6), but again without any sign of reduction. Both silver- and thioavin-Spositive plaques could still be found in each of the doxtreated animals. We biochemically measured the amount of aggregated Ab in the brains of our mice before and after transgene suppression using lter trap analysis of cortical tissue from each animal. In this assay, serial dilutions of protein homogenate are passed through a cellulose acetate lter; particles larger than the pore size of the lter become trapped in the membrane and are revealed by immunoblotting [22]. Consistent with our visual analysis of the histological sections, line 107 tTA/APP mice treated with dox for 3 or 6 mo had the same amount of aggregated Ab as when they started treatment at 6 mo of age (Figure 4A and 4B). In contrast, untreated 9-mo-old mice had almost twice as much aggregated Ab as either of the treated groups. Filter trap analysis of line 18 tTA/APP mice yielded similar results: the increase in aggregated Ab observed in untreated animals between 6 and 9 mo of age was completely arrested by transgene suppression (Figure S5C and S5D). We next used ELISA to measure total Ab in the brains of each group to determine whether any change in the amount or solubility of peptide occurred while APPswe/ind expression was suppressed. Cortical homogenates were sequentially extracted to separate peptide into PBS-, SDS-, and FA-soluble fractions, then transgene-derived Ab40 and Ab42 were measured by human-specic ELISA [23]. In all animals
93
Figure 3. Robust Transgene Suppression in Older Mice with Preexisting Amyloid Pathology (A) Cortical homogenates from 6- to 12-mo-old animals used for pathology studies described below (line 107) were immunoblotted with humanspecific antibody 6E10 to examine transgene suppression following 3 or 6 mo of dox treatment. The blot was co-immunostained for endogenous superoxide dismutase 1 (SOD1) as a control for loading. (B) Quantitation of signal intensity from the Western blot shown in (A). Transgenic APP levels are significantly suppressed following 3 or 6 mo of dox treatment (96.9% and 97.6%, respectively). *, p , 0.001 compared to 6-mo-old untreated animals, Tukey post-hoc test applied to significant effect of group ANOVA F3,12 107.22, p , 0.001. These data demonstrate that strong transgene suppression is attained both before and after the onset of amyloid pathology (see Figure 1 for predeposit experiments). (C) Experimental design. To examine the effects of chronic Ab suppression on amyloid pathology after the onset of deposition, we compared untreated controls harvested at 6 and 9 mo of age to animals placed on dox at 6 mo of age and harvested after 3 or 6 mo of treatment. (D) Dox treatment leads to rapid transgene suppression even in 6-mo-old tTA/APP mice. Immunostaining with 6E10 shows APPswe/ind levels are dramatically reduced in 6-mo-old mice treated for 1 wk with dox (upper panel). A separate blot was immunostained for APP C-terminal fragments with CT15 antibody to show that the precursors to Ab cleavage are decreased in parallel with the full-length protein (middle panel). Costaining for superoxide dismutase 1 was used as an internal control for loading (lower panel, taken from bottom half of 6E10 blot). DOI: 10.1371/journal.pmed.0020355.g003
94
harboring amyloid deposits, we found that the vast majority of Ab (.99%) was extracted into the SDS and FA fractions (Figure 5A and 5B). Consistent with the lter trap results presented above, there were no signicant differences in SDS- or FA-soluble Ab between the 6 mo untreated cohort and either the 6 mo 3 mo treated or 6 mo 6 mo treated cohorts. However, brains of both 6 mo 3 mo and 6 mo 6 mo treated cohorts contained roughly twice as much PBSsoluble Ab40 as untreated 6-mo-old mice (Figure 5C). Levels of Ab42 showed a similar trend, but did not reach statistical signicance. In fact, levels of PBS-soluble Ab40 and Ab42 in the 6 mo 3 mo and 6 mo 6 mo treated cohorts were most similar to that of the 9 mo untreated cohort, suggesting that age, as opposed to synthetic rate (which would be negligible in the treated animals), may determine the fraction of PBSsoluble Ab in these animals.
We also assessed neuritic and glial pathology surrounding the plaques to determine whether there were any changes in nearby tissue following long-term transgene suppression. Both Hirano silver stain and ubiquitin immunostaining showed neuritic pathology in all treatment groups (Figure 6). Similarly, activated astrocytes immunostained for GFAP were found near plaques in all animals (Figure 6). Neuritic and glial pathology were more severe in the older untreated mice. In contrast, transgene suppression prevented the growth of individual deposits apparent in untreated mice, and limited the surrounding pathology to what was already present when treatment began. An obvious question we sought to address was whether the deposition of Ab diminished cognitive ability in untreated mice, and what might happen to cognition when the process was interrupted. Unfortunately, efforts to characterize

Figure 4. Suppression of Transgenic APP Arrests Progression of Amyloid Pathology (A) Aggregated Ab was quantified in cortical tissue from dox-treated and control tTA/APP mice (line 107) using a filter trap assay. Serial dilutions of protein homogenate were passed through a cellulose acetate filter; protein aggregates larger than the pore size were trapped and immunostained for Ab. (B) Quantitation of signal intensity in the linear range of each filter trap dilution series (arrow in [A]) was used to compare aggregate load across treatment groups. Aggregated Ab increased significantly between 6 and 9 mo of age in untreated mice (significant effect of group ANOVA F3,18 7.85, p , 0.002). This progression of pathology was completely prevented by transgene suppression. The amount of aggregated Ab was identical in untreated mice at 6 mo of age to that in 9- or 12-mo-old animals treated with dox ( p . 0.9, Tukey post-hoc test). Single transgenic tTA samples were included as negative controls and showed no signal above background. *, p , 0.01; **, p , 0.005 versus 9-mo-old untreated mice, Tukey post-hoc test; ***, p , 0.001 versus 9-mo-old untreated mice, Students t-test. (C) Amyloid pathology in the hippocampus of representative mice from each treatment group: Hirano silver stain (top row), thioflavin-S (middle row), and Ab immunohistochemistry (bottom row). Amyloid burden increases dramatically between 6 and 9 mo of age in untreated animals, but remains stable in transgene-suppressed mice over the same period (6 mo 3 mo dox and 6 mo 6 mo dox). Single transgenic animals (tTA only shown here) show no sign of amyloid pathology at any age tested. DOI: 10.1371/journal.pmed.0020355.g004
cognitive behavior were compromised by severe hyperactivity in untreated double transgenic mice. The tTA/APP animals were often seen running in circles around the perimeter of their cages, and a similar swimming pattern was noted when the mice were tested in the Morris water maze. In the radial water maze, repetitive swim patterns were noted with no evidence of choice-motivated actions. Other studies have dealt with similar problems by excluding animals that do not show adequate attention to the task, retaining only those mice that meet certain performance criteria [30]. In our case, the penetrance of hyperactivity was close to 100%, leaving us with no testable animals. This phenotype has not affected
previous lines of APPswe mice we have produced, such as lines E12 or C33, that express lower levels of transgenic protein. Indeed, in past studies where hyperactivity was not a factor, we established a clear relationship between amyloid load and cognitive ability [31]. However, in the current study, we feel that although the poor performance of the tTA/APP mice in the maze tests could technically be scored as cognitive impairment, the animals severe hyperactivity made interpretation of the cognitive tasks impossible. In order to understand the nature and extent of hyperactivity in the tTA/APP mice, we quantied daily activity levels in double transgenic animals along with their single
95
Persistent Amyloid in Tet-Off APP Mice (C) The PBS fraction represents less than 0.1% of total Ab (note the change in y-axis from [A] and [B]), but only here do Ab levels in the doxtreated mice differ from those in younger untreated mice. Although both peptides appear elevated in the treated groups compared to the untreated 6-mo-old mice, only Ab40 reaches statistical significance (p , 0.05, Tukey post-hoc test applied to significant effect of group ANOVA for Ab40 F3,18 4.60, p , 0.02). A similar trend was seen for Ab42, where ANOVA yielded a significant effect of group for PBS-soluble Ab42 (F3,18 3.75, p , 0.03), however this was due only to differences between the untreated 6- and 9-mo-old groups. , p , 0.05 versus 6-mo-old untreated mice, Tukey post-hoc test; , p , 0.001 versus 6-mo-old untreated mice, Students t-test. DOI: 10.1371/journal.pmed.0020355.g005
transgenic and nontransgenic siblings using four-beam frames designed to monitor ambulation within an enclosed cage. As shown in Figure 7, the double transgenic mice were up to 10-fold more active during the dark phase of the day night cycle than any of the control groups. Activity levels appeared to follow a relatively normal diurnal cycle, decreasing substantially during the daylight hours. However, even during the light phase, the tTA/APP mice remained many-fold more active than normal controls. This behavior was partially, but not consistently, reversed by 1 mo of transgene suppression beginning at 45 mo of age (data not shown). In contrast, hyperactivity was completely abolished by rearing tTA/APP mice on dox. Animals born and raised on dox showed activity levels similar to the untreated controls (Figure 7C). Intriguingly, all of the dox-reared animals, both transgenic and wild-type, showed altered circadian rhythms with far less distinction between their day- and nighttime activity levels.
Discussion
We present a new mouse model for AD that was designed to test the consequences of inhibiting Ab production after the onset of amyloid pathology. New lines of transgenic mice were developed for this study that express high levels of APPswe/ind under the control of a tetracycline-responsive promoter. We demonstrate that treatment with dox suppresses steady-state levels of both APPswe/ind and its Cterminal fragments, indicating that the mutant proteins have a relatively short half-life in vivo. Transgenic expression of APPswe/ind and consequent overproduction of Ab42 cause early-onset amyloid deposition in untreated mice, in which deposits appear as early as 2 mo of age. Amyloid burden worsens signicantly with age, and by 9 mo, the hippocampus and cortex of untreated mice are largely lled with aggregated peptide. We nd that suppression of transgenic APP by more than 95% abruptly halts the progression of amyloid pathology. Importantly, this outcome occurs in animals already harboring considerable amyloid pathology, a situation similar to what might be expected in patients to be treated with secretase inhibitors. Somewhat unexpectedly, we observe no appreciable clearance of deposited amyloid even following periods of transgene suppression equal to the time taken for plaques to form. This latter nding indicates that compared to other disease-associated protein aggregates such as mutant huntingtin, which clear in less than 3 mo [32], the disaggregation of extracellular amyloid is relatively slow. Notably, pharmaceutical c-secretase inhibitors published to date show less strict regulation of Ab production following chronic administration than we have attained here. Two independent compounds tested in Tg2576 [33] and TgCRND8
1327 December 2005 | Volume 2 | Issue 12 | e355
Figure 5. Ab ELISA Confirms Arrest of Progression without Clearance of Peptide in Mice with Preexisting Aggregates Ab levels in untreated 6- and 9-mo-old tTA/APP line 107 mice (shown in Figure 4) were compared to those in 9- and 12-mo-old animals treated with dox from the age of 6 mo. Single transgenic APP samples were included as negative controls. Cortical homogenates were fractionated by sequential multi-step extraction with PBS, 2% SDS, and 70% FA followed by human-specific Ab ELISA to measure transgene-derived peptide in each fraction. Ab40 is shown in white, Ab42 in black. (A and B) Most Ab in the brains of plaque-bearing mice is extracted into the FA and SDS fractions. Consistent with amyloid burden (Figures 4 and S3), SDS- and FA-extracted Ab levels in untreated 9-mo-old mice were significantly higher than in untreated 6-mo-old mice (Tukey post-hoc test applied to significant effect of group ANOVA for SDS and FA fractions F3,18 4.7212.92, p , 0.02). In contrast, 3 or 6 mo of transgene suppression held Ab at levels equivalent to those harbored when treatment was started (p . 0.2 compared to 6 mo untreated mice, Tukey post-hoc test). *, p , 0.05; **, p , 0.005; ***, p , 0.001 versus 9-mo-old untreated mice, Tukey post-hoc test. Significance for APP versus 9-moold untreated mice is based on Students t-test.
96

Figure 6. Neuritic and Glial Pathology Are Unchanged following Transgene Suppression Dystrophic neurites and activated astrocytes surround most compact plaques in tet-off APP mice (line 107). Dark-stained, ubiquitin-filled neurites and reactive astrocytes form a halo around cored, fibrillar deposits by 6 mo of age that worsens with time in untreated mice. Both plaque-associated pathologies are arrested, although not reversed, by transgene suppression. Hirano silver stain (top row); GFAP immunohistochemistry (middle row); ubiquitin immunohistochemistry (bottom row). DOI: 10.1371/journal.pmed.0020355.g006
[34] transgenic mice show no more than 85% suppression of Ab40 levels, and where measured, even less suppression of Ab42 (approximately 60%) [35,36]. Thus, even after accounting for the much higher APP expression levels in our mice than in the Tg2576 and TgCRND8 lines, we have achieved better absolute suppression of Ab production with the tet-off system than is currently possible with published c-secretase inhibitors. Since even the most advanced future pharmaceutical agents are unlikely to attain more complete control of Ab production than achieved here, this system provides a salient test of therapeutic intervention with Ab-lowering compounds. Although the progression of amyloid deposition was sharply arrested by this approach, we found that a substantial amyloid burden remained even after long periods of transgene suppression. We examined a small number of animals after 12 mo of dox treatment (beginning at 6 mo of age), and found that amyloid deposits were still relatively abundant. Longer-term treatments are now in progress. At the latest treatment interval analyzed by ELISA, animals administered dox for 6 mo showed elevated levels of PBS-soluble Ab (see Figure 5) that could be interpreted as an indication that the plaques are slowly releasing peptide (or oligomeric Ab) into the soluble pool and might eventually dissolve. Whether inhibiting Ab production longer than 6 (or 12) mo may ultimately result in clearance of amyloid is under investigation; unfortunately, the life span of the model eventually limits this experiment. Therapeutics used in humans will have
considerably more time to act than is possible within the life span of rodent models. Long-term treatments would certainly be possible and could be a key to effective therapy. Overall, however, we interpret our ndings as evidence that AD therapies that signicantly lower the production of Ab (by either inhibiting secretase activity or inhibiting APP expression) may not quickly reverse preexisting pathology, but should effectively halt further deposition of amyloid. In interpreting our study, it should be remembered that the earliest plaques to appear in these mice, like other APP transgenics harboring the Swedish mutation [27], are predominantly brillar deposits, which may be less tractable than the diffuse aggregates thought to come rst in the course of the human disease. However, our data suggest that once diffuse deposits are formed in these mice, they are no more easily cleared in our system than cored plaques (see Figure S4). An additional consideration we recognize is that a small amount of transgene expression continues in the presence of dox and that endogenous mouse Ab continues to be produced. It is possible that the combined low levels of endogenous mouse Ab and nonsuppressed human peptide are sufcient to maintain existing deposits. However, these low levels of peptide are not sufcient to induce new amyloid formation, as CaMKIIa-tTA 3 tetAPPswe/ind mice raised on dox for up to a year do not develop amyloid lesions (data not shown). It is also clear that in this genetic system, we have raised the production of Ab to levels not found in humans to accelerate pathology into an experimentally feasible time
97
Figure 7. Transgene Suppression Attenuates Hyperactivity in tTA/APP Mice (A) A 48-h measure of ambulation records extreme hyperactivity in untreated double transgenic mice compared to single transgenic and nontransgenic controls (line 107). This phenotype is completely eliminated by rearing the double transgenic mice on dox. (B) The same data shown in (A) are replotted to magnify data from untreated control and dox-treated groups. (C and D) Activity levels in the combined control groups of (A) and (B) are here separated by genotype. None of the single transgenic or nontransgenic control groups display the hyperactivity present in untreated tTA/APP animals. Again, note the y-axes have been enlarged for detail compared to (A). DOI: 10.1371/journal.pmed.0020355.g007
98
frame. This system allowed us to create an approximately 20fold differential between APP/Ab synthetic rates before and after treatment, yet the in vivo equilibrium between aggregated and disaggregated states of Ab still favored the maintenance of amyloid deposits. In our opinion, it seems unlikely that amyloid deposits in human brain would be inherently any less stable than those formed in mouse brain. However, the human brain may harbor clearance mechanisms not shared by mice that would allow more efcient removal of preexisting amyloid. One potential mechanism by which amyloid may be more efciently removed in the human disease than in the mouse models is through microglial phagocytosis. Resident microglia in transgenic mouse models localize to tissue surrounding plaques but show little evidence of amyloid engulfment [37 40]. In contrast, microglia surrounding amyloid plaques in human brain show a much higher state of activation with greater expression of complement receptor [40]. Thus, the role of microglia in amyloid metabolism is minor in transgenic models compared to the human condition. Somewhat paradoxically, several studies further demonstrate that treatment with anti-inammatory drugs to reduce microglial
activation actually lowers amyloid load in APP transgenic mice, suggesting a role for mouse microglia in the formation and maintenance of amyloid aggregates [4143]. However, this outcome may be alternatively explained by direct effects of many anti-inammatory drugs on c-secretase cleavage [44 47]. Nonetheless, the role of microglia in both the human condition and the mouse models is poorly understood, and differences in microglial reactivity between the two could lead to signicantly faster amyloid clearance in the brains of patients with AD than we observe in the tet-off APP mice. Given the relatively minor role played by microglia in other mouse models of amyloidosis, we think it unlikely that these cells have inuenced the rate of amyloid clearance in the tetoff APP mice. Even so, we considered the possibility that chronic dox treatment may have altered the activation state of microglia in our treated mice. Dox is structurally similar to minocycline, a reported anti-inammatory drug and inhibitor of microglial activation [48]. However, if dox does have anti-inammatory activity, then, based on previous studies with other anti-inammatories, we would have expected to nd less amyloid in the dox-treated animals. Clearly, that was not the case. While it is possible that dox acts in some other

way to slow amyloid clearance, data from multiple studies demonstrate that microglial responses are normally weak in the mouse AD models [3740], and thus it is doubtful that dox-mediated microglial inhibition affected the outcome of our study. The persistence and stability of amyloid deposits in our system is unexpected given the speed with which Ab aggregates are cleared in other mouse models of therapeutic intervention. Anti-Ab antibodies injected directly into the brain have been shown to eliminate amyloid deposits in as little as 1 wk after treatment [4951]. Peripheral antibody injection decreases amyloid load more broadly, and although it does not appear to act as quickly as local injection, can signicantly reduce amyloid load within 2 mo of initial treatment [52,53]. More recently, an alternative approach has shown that lentiviral transfer of neprilysin can also reduce the number of aggregates in the area of the injection site [54]. Careful study of the mechanism behind several of the antibody-mediated therapies has suggested that activated microglia play an important role in the removal of brillar plaques after immunization [50,52,55]. However, it has been noted that deletion of the Fc receptor (the primary receptor for microglial opsinization of antibodyantigen complexes) in APP transgenic mouse models has no impact on the effectiveness of antibody-mediated therapy [56,57]. It is, nevertheless, possible that lack of microglia activation is the major difference between the slow clearance described here, where no perturbation of the immune system is expected, and the rapid clearance described in studies involving antibody or viral injection. In isolation, mild activation of microglia by injection damage or opsinization may not be adequate to induce substantial phagocytosis, but when combined with an Ab-lowering agent, such as neprilysin or Ab-targeted antibodies, the two may work in concert to clear peptide deposits. Consistent with this hypothesis, strong activation of microglia through transgenic expression of TGFb [58] or central injection of lipopolysaccharide [59,60] can by itself substantially reduce plaque burden in APP transgenic mice. But in the case of acute antibody- and/or injury-mediated activation, once the inammation has passed, and the antibody and bound peptide have been cleared and degraded, the remaining Ab quickly reaggregates and amyloid pathology is reestablished [49]. This nding reinforces the notion that without continued stimulation, microglia in mouse models do not maintain the same level of sustained activation that may occur in humans. SantaCruz et al. recently published a study of mice that express P301L human tau via a similar vector system [61]. As in our tet-off APP mice, SantaCruz et al. found that tau neurobrillary tangles, like amyloid plaques, are not cleared efciently following transgene suppression. The lack of clearance in both models of AD pathology comes as a stark contrast to the rapid removal of protein aggregates found in similar tet-off mouse models of Huntington [32] and prion disease [62]. In these cases, disrupting the input of new monomer to the system via dox-mediated transgene suppression led to relatively rapid clearance of protein aggregates. By contrast, our study and that of SantaCruz et al. suggest that protein aggregates in AD may be more tenacious than in other neurodegenerative disorders. Perhaps once aggregated, Ab and tau are either inherently more stable than other
protein aggregates or more resistant to intra- and extracellular clearance mechanisms. One question we were not able to address in this study is whether abrogating synthesis of new Ab halts the progression of cognitive decline. Studies from the tet-off tau mice suggest that protein clearance may in fact not be required for cognitive improvement following transgene suppression [61]. At present, because of unexpected noncognitive behavioral abnormalities, it is not clear whether the tet-off APP mice can be used to address the same question in the context of amyloid pathology. Both lines of tTA/APP mice we studied here display extreme hyperactivity visible as cage circling and quantied by activity monitoring (see Figure 7). Many of the double transgenic mice showed similar circular patterns of swimming near the edge of the tank when tested in the Morris water maze. Expression of this phenotype makes standard tests of learning and memory uninterpretable. Hyperactivity nonspecically inhibits choice-driven changes in movement, the key element behind all cognitive behavioral paradigms. We are currently working to determine whether hyperactivity correlates with expression of the APPswe/ind holoprotein or its proteolytic derivatives. Preliminary studies suggest that hyperactivity does not appear quickly when dox-reared mice are shifted to nonmedicated diets (J. L. J., unpublished data). These data may indicate that the neuroactive culprit is not immediately present after transgenic APP synthesis is initiated, but requires additional time to develop. Alternatively, hyperactivity may be caused by neuronal alterations due to transgene expression during early postnatal development. Further experiments are needed to distinguish between these possibilities. In summary, we demonstrate that abrogating Ab production halts the progression of pathologic changes in a transgenic mouse model of Alzheimer-type amyloidosis. However, despite dramatic reductions in Ab synthesis, neuritic plaques are stable structures in vivo that do not quickly disaggregate. It is possible that a combination of therapies to limit Ab production, increase Ab degradation, and enhance phagocytosis of deposited amyloid may be required to reverse damage associated with AD. However, if started early enough in the course of disease, secretase inhibitors alone could provide substantial benet in slowing pathogenic processes linked to amyloid deposition. Even at later stages in the disease, the presence of substantial microglial activation in human AD [40] suggests that simply slowing the formation of new amyloid deposits may allow ongoing phagocytosis to diminish preexisting lesions. However, the development of safe and effective secretase inhibitors will ultimately be required to determine whether the human brain has the capacity to repair amyloidassociated damage of AD once the progression of pathology is arrested.
Figure S1. Transgenic APP Expression and Suppression by Dox in the Four New Tet-Off APP Lines Western blotting with human-specic antibody 6E10 reveals transgene-derived full-length APP in cortical homogenates from untreated animals (left lanes of each panel). The new lines produce exceptionally high levels of transgene expression; an equal amount of brain homogenate from a standard transgenic APP line is shown for comparison (extreme left lane, Standard Tg line C33; [15,63]). After December 2005 | Volume 2 | Issue 12 | e355
99
Persistent Amyloid in Tet-Off APP Mice 1 mo of dox treatment, transgenic protein in all four new tet-off APP lines is reduced to residual levels ( dox; right lanes of each panel). DOI: 10.1371/journal.pmed.0020355.sg001 (474 KB TIF). Figure S2. Transgenic APP mRNA Is Brain-Specic A slot blot of mRNA harvested from various tissues in three of the four new tet-off APP lines and a nontransgenic control was used to examine transgenic mRNA expression. Hybridization is seen only in the brain; no signal above background is seen in any other tissue. DOI: 10.1371/journal.pmed.0020355.sg002 (781 KB PSD). Figure S3. Amyloid Pathology in the Cortex Reiterates That in the Hippocampus Amyloid histology was performed on sections from line 107 double transgenic mice by Hirano silver stain (top row), thioavin-S (middle row), and Ab immunohistochemistry (bottom row) to examine the persistence of pathology following transgene suppression. As in the hippocampus (see Figure 4 and text), the progression of amyloid pathology in the cortex worsens substantially between 6 and 9 mo of age in untreated mice. This progression is completely prevented by suppression of the transgene with dox. For comparison, normal neurohistology is shown in an age-matched single transgenic (tTA only) animal. No amyloid pathology has been detected in either APP or tTA single transgenic animals up to 15 mo of age. DOI: 10.1371/journal.pmed.0020355.sg003 (4.8 MB PSD). Figure S4. Diffuse Deposits Do Not Disperse During Ab Suppression CampbellSwitzer silver stain was used to differentiate cored (brown) from diffuse (black) deposits in line 107 tTA/APP mice. This stain demonstrates that both types of deposit persist throughout long periods of transgene suppression. The lower panels, showing lowpower images (103) of frontal cortex from each condition, reveal little change in the extent of diffuse amyloid following up to 6 mo of Ab suppression. High-power images (403) in the upper panels show that the diffuse halo surrounding individual cored deposits remains relatively unchanged in treated mice. Untreated tTA single transgenic animals are shown as a negative control. Protocol for the Campbell-Switzer Alzheimers Method was kindly shared by Robert Switzer, III (NeuroScience Associates, Knoxville, Tennessee, United States), and can be downloaded at http://www.nsalabs.com/ Documents/publications/campbell-switzer_protocol.htm [64,65]. DOI: 10.1371/journal.pmed.0020355.sg004 (923 KB JPG). Figure S5. Chronic Transgene Suppression and Arrest of Ab Aggregate Formation in an Independent Line of Tet-Off APP Mice (CaMKIIa-tTA 3 tet-APPswe/ind Line 18) (A) The experiment presented in the text for line 107 tet-off APP was repeated with a second tet-off APP line (line 18) to control for integration site artifacts. Cortical homogenates from untreated control and dox-treated double transgenic mice were immunoblotted for full-length APP with the human-specic antibody 6E10 to conrm transgene suppression at the time of harvest. Immunostaining for endogenous superoxide dismutase (SOD1) was included as a loading control. (B) Quantitation of signal intensity from the Western blot in (A) shows transgenic APP levels in line 18 are suppressed by more than 98% following 3 mo of dox treatment (signicant effect of group ANOVA F2,8 1559.7, p , 0.001). This level of suppression was equal to or better than that attained in line 107 (see Figure 3B). (C) Serial dilution lter trap assay was used to quantify aggregated Ab in cortical homogenates. (D) Quantitation of signal intensity in the linear range of the dilution series shown in (C). Consistent with the amyloid histology shown in Figure S5, aggregate formation was signicantly increased between 6 and 9 mo of age in untreated mice (signicant effect of group ANOVA F2,18 12.14, p , 0.001). Aggregate formation was completely arrested by transgene suppression, and is identical in 9-mo-old mice treated with dox for 3 mo as in untreated animals harvested when treatment began (p . 0.5, Tukey post-hoc test). *, p , 0.05; **, p , 0.005; ***, p , 0.001 versus 9-mo-old untreated mice, Tukey post-hoc test. DOI: 10.1371/journal.pmed.0020355.sg005 (962 KB TIF). Figure S6. Arrest of Amyloid Progression by Chronic Transgene Suppression in Line 18 Tet-Off APP Mice Amyloid histology in cortical (rst and third rows) and hippocampal (second and fourth rows) sections from untreated tTA/APP mice shows a dramatic progression of pathology between 6 and 9 mo of age. Suppression of transgenic APP expression arrests this progresPLoS Medicine | www.plosmedicine.org 1331 sion, although without any sign of plaque clearance (6 mo 3 mo dox). Hirano silver stain (top panels); thioavin-S (bottom panels). DOI: 10.1371/journal.pmed.0020355.sg006 (5.8 MB PSD).
Acknowledgments
We thank Patrick Tremblay for helpful advice on the tet system at a critical time in the project, and Mark Mayford for sharing the CaMKIIa-tTA mice through Jackson Laboratory. We also thank Fraser Moss for saving several immunoblots with last-minute shipments, Andy Groves for many thoughtful discussions, Neil Segil for generously sharing his laboratory and equipment, Beth Olson, Natasha Bouey, and Yolanda Jackson for outstanding animal care, Debbie Swing for expert microinjection, and Dave Fromholt for genotyping and dissection. We gratefully acknowledge Takeda Chemical Industries for providing antibodies BAN50, BA27, and BC05, Konrad Beyreuther and Andreas Weidemann for providing 22C11 antibody, and Ed Koo for sharing CT15 antibody. This work was supported by grants from the Johns Hopkins Alzheimers Disease Research Center (JLJ), the National Alliance for Research on Schizophrenia and Depression (Young Investigator Award [JLJ]), the Rose Hills Foundation (JLJ), the Alzheimers Association (Zenith Award [DRB]), the National Institute of Aging (K01 AG2614401 [JLJ], P50 AGO514620 [DRB], R01 AG00665616 [SGY], and P01 AG015453 [SGY]), the National Institute of Neurologic Disease and Stoke (R01 NS 047225 [DRB]), and the National Cancer Institute (NAJ and NGC). The funding agencies generously provided for research supplies, animal care, and salary support; the funders of this work had no role in study design, data collection and analysis, decision to & publish, or preparation of the manuscript.
References 1. Hardy JA, Higgins GA (1992) Alzheimers disease: The amyloid cascade hypothesis. Science 256: 184185. 2. Selkoe DJ, Podlisny MB (2002) Deciphering the genetic basis of Alzheimers disease. Annu Rev Genomics Hum Genet 3: 6799. 3. Rademakers R, Cruts M, Van Broeckhoven C (2003) Genetics of early-onset Alzheimer dementia. ScienticWorldJournal 3: 497519. 4. Orgogozo JM, Gilman S, Dartigues JF, Laurent B, Puel M, et al. (2003) Subacute meningoencephalitis in a subset of patients with AD after Ab42 immunization. Neurology 61: 4654. 5. Check E (2002) Nerve inammation halts trial for Alzheimers drug. Nature 415: 462. 6. Hock C, Konietzko U, Streffer JR, Tracy J, Signorell A, et al. (2003) Antibodies against b-amyloid slow cognitive decline in Alzheimers disease. Neuron 38: 547554. 7. Nicoll JA, Wilkinson D, Holmes C, Steart P, Markham H, et al. (2003) Neuropathology of human Alzheimer disease after immunization with amyloid-b peptide: A case report. Nat Med 9: 448452. 8. Tsai JY, Wolfe MS, Xia W (2002) The search for c-secretase and development of inhibitors. Curr Med Chem 9: 10871106. 9. Wolfe MS (2002) Therapeutic strategies for Alzheimers disease. Nat Rev Drug Discov 1: 859866. 10. Cumming JN, Iserloh U, Kennedy ME (2004) Design and development of BACE-1 inhibitors. Curr Opin Drug Discov Devel 7: 536556. 11. Citron M (2004) b-secretase inhibition for the treatment of Alzheimers diseasePromise and challenge. Trends Pharmacol Sci 25: 9297. 12. Gossen M, Bujard H (1992) Tight control of gene expression in mammalian cells by tetracycline-responsive promoters. Proc Natl Acad Sci U S A 89: 55475551. 13. Furth PA, St Onge L, Boger H, Gruss P, Gossen M, et al. (1994) Temporal control of gene expression in transgenic mice by a tetracycline-responsive promoter. Proc Natl Acad Sci U S A 91: 93029306. 14. Kistner A, Gossen M, Zimmermann F, Jerecic J, Ullmer C, et al. (1996) Doxycycline-mediated quantitative and tissue-specic control of gene expression in transgenic mice. Proc Natl Acad Sci U S A 93: 1093310938. 15. Borchelt DR, Davis J, Fischer M, Lee MK, Slunt HH, et al. (1996) A vector for expressing foreign genes in the brains and hearts of transgenic mice. Genet Anal 13: 159163. 16. Mayford M, Bach ME, Huang YY, Wang L, Hawkins RD, et al. (1996) Control of memory formation through regulated expression of a CaMKII transgene. Science 274: 16781683. 17. Weidemann A, Konig G, Bunke D, Fischer P, Salbaum JM, et al. (1989) Identication, biogenesis, and localization of precursors of Alzheimers disease A4 amyloid protein. Cell 57: 115126. 18. Borchelt DR, Lee MK, Slunt HS, Guarnieri M, Xu ZS, et al. (1994) Superoxide dismutase 1 with mutations linked to familial amyotrophic lateral sclerosis possesses signicant activity. Proc Natl Acad Sci U S A 91: 82928296. 19. Sisodia SS, Koo EH, Hoffman PN, Perry G, Price DL (1993) Identication and transport of full-length amyloid precursor proteins in rat peripheral nervous system. J Neurosci 13: 31363142.
100

20. Church GM, Gilbert W (1984) Genomic sequencing. Proc Natl Acad Sci U S A 81: 19911995. 21. Yamamoto T, Hirano A (1986) A comparative study of modied Bielschowsky, Bodian and thioavin S stains on Alzheimers neurobrillary tangles. Neuropathol Appl Neurobiol 12: 39. 22. Xu G, Gonzales V, Borchelt DR (2002) Rapid detection of protein aggregates in the brains of Alzheimer patients and transgenic mouse models of amyloidosis. Alzheimer Dis Assoc Disord 16: 191195. 23. Kawarabayashi T, Younkin LH, Saido TC, Shoji M, Ashe KH, et al. (2001) Agedependent changes in brain, CSF, and plasma amyloid b protein in the Tg2576 transgenic mouse model of Alzheimers disease. J Neurosci 21: 372381. 24. Suzuki N, Cheung TT, Cai XD, Odaka A, Otvos L Jr, et al. (1994) An increased percentage of long amyloid b protein secreted by familial amyloid b protein precursor (bAPP717) mutants. Science 264: 13361340. 25. Gravina SA, Ho L, Eckman CB, Long KE, Otvos L Jr, et al. (1995) Amyloid b protein (Ab) in Alzheimers disease brain. Biochemical and immunocytochemical analysis with antibodies specic for forms ending at Ab40 or Ab42(43). J Biol Chem 270: 70137016. 26. Jankowsky JL, Fadale DJ, Anderson J, Xu GM, Gonzales V, et al. (2004) Mutant presenilins specically elevate the levels of the 42 residue b-amyloid peptide in vivo: Evidence for augmentation of a 42-specic c-secretase. Hum Mol Genet 13: 159170. 27. Borchelt DR, Ratovitski T, van Lare J, Lee MK, Gonzales V, et al. (1997) Accelerated amyloid deposition in the brains of transgenic mice coexpressing mutant presenilin 1 and amyloid precursor proteins. Neuron 19: 939945. 28. Herms J, Zurmohle U, Brysch W, Schlingensiepen KH (1993) Ca2/ calmodulin protein kinase and protein kinase C expression during development of rat hippocampus. Dev Neurosci 15: 410416. 29. Phinney AL, Deller T, Stalder M, Calhoun ME, Frotscher M, et al. (1999) Cerebral amyloid induces aberrant axonal sprouting and ectopic terminal formation in amyloid precursor protein transgenic mice. J Neurosci 19: 85528559. 30. Kotilinek LA, Bacskai B, Westerman M, Kawarabayashi T, Younkin L, et al. (2002) Reversible memory loss in a mouse transgenic model of Alzheimers disease. J Neurosci 22: 63316335. 31. Savonenko A, Xu GM, Melnikova T, Morton JL, Gonzales V, et al. (2005) Episodic-like memory decits in the APPswe/PS1dE9 mouse model of Alzheimers disease: Relationships to b-amyloid deposition and neurotransmitter abnormalities. Neurobiol Dis 18: 602617. 32. Yamamoto A, Lucas JJ, Hen R (2000) Reversal of neuropathology and motor dysfunction in a conditional model of Huntingtons disease. Cell 101: 5766. 33. Hsiao K, Chapman P, Nilsen S, Eckman C, Harigaya Y, et al. (1996) Correlative memory decits, Ab elevation, and amyloid plaques in transgenic mice. Science 274: 99102. 34. Chishti MA, Yang DS, Janus C, Phinney AL, Horne P, et al. (2001) Earlyonset amyloid deposition and cognitive decits in transgenic mice expressing a double mutant form of amyloid precursor protein 695. J Biol Chem 276: 2156221570. 35. Wong GT, Manfra D, Poulet FM, Zhang Q, Josien H, et al. (2004) Chronic treatment with the gamma-secretase inhibitor LY-411,575 inhibits bamyloid peptide production and alters lymphopoiesis and intestinal cell differentiation. J Biol Chem 279: 1287612882. 36. Barten DM, Guss VL, Corsa JA, Loo A, Hansel SB, et al. (2005) Dynamics of b-amyloid reductions in brain, cerebrospinal uid, and plasma of bamyloid precursor protein transgenic mice treated with a c-secretase inhibitor. J Pharmacol Exp Ther 312: 635643. 37. Sasaki A, Shoji M, Harigaya Y, Kawarabayashi T, Ikeda M, et al. (2002) Amyloid cored plaques in Tg2576 transgenic mice are characterized by giant plaques, slightly activated microglia, and the lack of paired helical lament-typed, dystrophic neurites. Virchows Arch 441: 358367. 38. Wegiel J, Imaki H, Wang KC, Wegiel J, Rubenstein R (2004) Cells of monocyte/microglial lineage are involved in both microvessel amyloidosis and brillar plaque formation in APPsw tg mice. Brain Res 1022: 1929. 39. Wegiel J, Wang KC, Imaki H, Rubenstein R, Wronska A, et al. (2001) The role of microglial cells and astrocytes in brillar plaque evolution in transgenic APPsw mice. Neurobiol Aging 22: 4961. 40. Schwab C, Hosokawa M, McGeer PL (2004) Transgenic mice overexpressing amyloid b protein are an incomplete model of Alzheimer disease. Exp Neurol 188: 5264. 41. Jantzen PT, Connor KE, DiCarlo G, Wenk GL, Wallace JL, et al. (2002) Microglial activation and b-amyloid deposit reduction caused by a nitric oxide-releasing nonsteroidal anti-inammatory drug in amyloid precursor protein plus presenilin-1 transgenic mice. J Neurosci 22: 22462254. 42. Lim GP, Yang F, Chu T, Chen P, Beech W, et al. (2000) Ibuprofen suppresses plaque pathology and inammation in a mouse model for Alzheimers disease. J Neurosci 20: 57095714. 43. Yan Q, Zhang J, Liu H, Babu-Khan S, Vassar R, et al. (2003) Antiinammatory drug therapy alters b-amyloid processing and deposition in an animal model of Alzheimers disease. J Neurosci 23: 75047509. 44. Eriksen JL, Sagi SA, Smith TE, Weggen S, Das P, et al. (2003) NSAIDs and enantiomers of urbiprofen target gamma-secretase and lower Ab42 in vivo. J Clin Invest 112: 440449. 45. Weggen S, Eriksen JL, Das P, Sagi SA, Wang R, et al. (2001) A subset of NSAIDs lower amyloidogenic Ab42 independently of cyclooxygenase activity. Nature 414: 212216. 46. Weggen S, Eriksen JL, Sagi SA, Pietrzik CU, Ozols V, et al. (2003) Evidence that nonsteroidal anti-inammatory drugs decrease amyloid b 42 production by direct modulation of c-secretase activity. J Biol Chem 278: 3183131837. 47. Lleo A, Berezovska O, Herl L, Raju S, Deng A, et al. (2004) Nonsteroidal anti-inammatory drugs lower Ab42 and change presenilin 1 conformation. Nat Med 10: 10651066. 48. Yrjanheikki J, Keinanen R, Pellikka M, Hokfelt T, Koistinaho J (1998) Tetracyclines inhibit microglial activation and are neuroprotective in global brain ischemia. Proc Natl Acad Sci U S A 95: 1576915774. 49. Oddo S, Billings L, Kesslak JP, Cribbs DH, LaFerla FM (2004) Ab immunotherapy leads to clearance of early, but not late, hyperphosphorylated tau aggregates via the proteasome. Neuron 43: 321332. 50. Wilcock DM, DiCarlo G, Henderson D, Jackson J, Clarke K, et al. (2003) Intracranially administered anti-Ab antibodies reduce beta-amyloid deposition by mechanisms both independent of and associated with microglial activation. J Neurosci 23: 37453751. 51. Bacskai BJ, Kajdasz ST, Christie RH, Carter C, Games D, et al. (2001) Imaging of amyloid-b deposits in brains of living mice permits direct observation of clearance of plaques with immunotherapy. Nat Med 7: 369372. 52. Bard F, Cannon C, Barbour R, Burke RL, Games D, et al. (2000) Peripherally administered antibodies against amyloid b-peptide enter the central nervous system and reduce pathology in a mouse model of Alzheimer disease. Nat Med 6: 916919. 53. Wilcock DM, Rojiani A, Rosenthal A, Levkowitz G, Subbarao S, et al. (2004) Passive amyloid immunotherapy clears amyloid and transiently activates microglia in a transgenic mouse model of amyloid deposition. J Neurosci 24: 61446151. 54. Marr RA, Rockenstein E, Mukherjee A, Kindy MS, Hersh LB, et al. (2003) Neprilysin gene transfer reduces human amyloid pathology in transgenic mice. J Neurosci 23: 19921996. 55. Wilcock DM, Munireddy SK, Rosenthal A, Ugen KE, Gordon MN, et al. (2004) Microglial activation facilitates Ab plaque removal following intracranial anti-Ab antibody administration. Neurobiol Dis 15: 1120. 56. Das P, Howard V, Loosbrock N, Dickson D, Murphy MP, et al. (2003) Amyloid-b immunization effectively reduces amyloid deposition in FcRc-/knock-out mice. J Neurosci 23: 85328538. 57. Bacskai BJ, Kajdasz ST, McLellan ME, Games D, Seubert P, et al. (2002) NonFc-mediated mechanisms are involved in clearance of amyloid-b in vivo by immunotherapy. J Neurosci 22: 78737878. 58. Wyss-Coray T, Lin C, Yan F, Yu GQ, Rohde M, et al. (2001) TGFb1 promotes microglial amyloid-b clearance and reduces plaque burden in transgenic mice. Nat Med 7: 612618. 59. DiCarlo G, Wilcock D, Henderson D, Gordon M, Morgan D (2001) Intrahippocampal LPS injections reduce Ab load in APPPS1 transgenic mice. Neurobiol Aging 22: 10071012. 60. Herber DL, Roth LM, Wilson D, Wilson N, Mason JE, et al. (2004) Timedependent reduction in Ab levels after intracranial LPS administration in APP transgenic mice. Exp Neurol 190: 245253. 61. SantaCruz K, Lewis J, Spires T, Paulson J, Kotilinek L, et al. (2005) Tau suppression in a neurodegenerative mouse model improves memory function. Science 309: 476481. 62. Safar JG, DeArmond SJ, Kociuba K, Deering C, Didorenko S, et al. (2005) Prion clearance in bigenic mice. J Gen Virol. In press. 63. Borchelt DR, Thinakaran G, Eckman CB, Lee MK, Davenport F, et al. (1996) Familial Alzheimers disease-linked presenilin 1 variants elevate Ab142/1 40 ratio in vitro and in vivo. Neuron 17: 10051013. 64. Campbell SK, Switzer RC, Martin TL (1987) Alzheimers plaques and tangles: A controlled and enhanced silver staining method. Soc Neurosci Abstr 13:: 189.9. 65. Switzer RC, Campbell SK, Murdock TM, inventors (1993 March 9) A histologic method for staining Alzheimer pathology. United States patent 5,192,688.
101
PLoS Medicine | www.plosmedicine.org 1332 December 2005 | Volume 2 | Issue 12 | e355
Patient Summary
Background Patients with Alzheimer disease (AD) have elevated levels of a small protein called amyloid-b peptide that sticks together to form what are known as amyloid plaques in their brains. This peptide is normally made at low levels in healthy individuals, and is made when a larger protein called amyloid precursor protein (APP) is cut down in size. New treatments are now being developed that will decrease the amount of Ab produced from APP. However, it is not clear whether lowering the production of Ab will allow the brain to heal itself by clearing the amyloid plaques. The answer to this question may be important for deciding when Ab-lowering drugs should be started, and may also determine how effective they are in reversing the mental symptoms of AD. What Did the Researchers Do and Find? Because new drugs designed to lower Ab levels are still in development, they are not available for testing in animal models of the disease. Instead, basic questions about the effectiveness of this type of treatment must be answered using systems that mimic how the drugs work. To do this, the authors created mice that produce too much APP and that develop the same amyloid lesions as do human patients with AD. Unlike normal mice, these mice also carried a switch gene that allowed the researchers to turn off APP by feeding the mice special food. Turning off APP in these mice had the same effect as treating them with Ab-lowering drugs, and so the researchers were able to ask what happened to the amyloid plaques after Ab production was shut down. They showed that lowering Ab production prevents the amyloid lesions from getting worse as the disease progresses. This means that treatment with Ab-lowering drugs may be able to stop the disease from filling the brain with plaques. However, the researchers also found that the amyloid lesions that had formed before treatment was started remained intact throughout the experiment. What Do These Findings Mean? These results indicate that treatments designed to lower the production of Ab may be an important part of future AD treatment, as this approach seems to prevents additional amyloid plaques from forming in the mouse brain. However, by itself, this strategy may not be able to rid the brain of plaques that have already formed in the brain before treatment is started. The findings suggest that early treatment may be important for this approach to succeed. Where Can I Get More Information Online? MedlinePlus has several Web pages of information on Alzheimer disease: http://www.nlm.nih.gov/medlineplus/alzheimersdisease.html The ADEAR Center of the US Governments National Institute on Aging also has information on Alzheimer disease: http://www.alzheimers.org/ The Alzheimers Association Web site contains information on both caregiving and research: http://www.alz.org
102
PLoS Medicine | www.plosmedicine.org 1333 December 2005 | Volume 2 | Issue 12 | e355
PLoS MEDICINE
A Randomised, Blinded, Placebo-Controlled Trial in Dementia Patients Continuing or Stopping Neuroleptics (The DART-AD Trial)
Clive Ballard1*, Marisa Margallo Lana2, Megan Theodoulou3, Simon Douglas4, Rupert McShane5, Robin Jacoby3, Katja Kossakowski1, Ly-Mee Yu6, Edmund Juszczak6, on behalf of the Investigators DART AD
1 Wolfson Centre for Age-Related Diseases, Kings College London, London, United Kingdom, 2 Northgate Hospital, Morpeth, Northumberland, United Kingdom, 3 Department of Psychiatry, University of Oxford, The Warneford Hospital, Oxford, United Kingdom, 4 Department of Psychiatry, Newcastle University, Newcastle upon Tyne, United Kingdom, 5 Oxfordshire and Buckinghamshire Mental Health NHS Trust and University of Oxford, Department of Psychiatry, Fulbrook Centre, Oxford, United Kingdom, 6 Centre for Statistics in Medicine, University of Oxford, Oxford, United Kingdom Funding: See section at end of manuscript. Competing Interests: See section at end of manuscript. Academic Editor: Carol Brayne, University of Cambridge, United Kingdom Citation: Ballard C, Lana MM, Theodoulou M, Douglas S, McShane R, et al. (2008) A randomised, blinded, placebo-controlled trial in dementia patients continuing or stopping neuroleptics (the DART-AD Trial). PLoS Med 5(4): e76. doi:10. 1371/journal.pmed.0050076 Received: May 31, 2007 Accepted: February 15, 2008 Published: April 1, 2008 Copyright: 2008 Ballard et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abbreviations: AD, Alzheimer disease; ANCOVA, analysis of covariance; BADLS, Bristol Activities of Daily Living Scale; CGIC, Clinicians Global Impression of Change; CI, confidence interval; DMC, datamonitoring committee; EPS, extrapyramidal signs and symptoms; FAS, Verbal Fluency Task; FAST, Functional Assessment Staging; IQR, interquartile range; NPI, Neuropsychiatric Inventory; SD, standard deviation; SIB, Severe Impairment Battery; (S)MMSE, (Standardised) Mini Mental State Examination; STALD, Sheffield Test for Acquired Language Disorders; UPDRS, Unified Parkinsons Disease Rating Scale * To whom correspondence should be addressed. E-mail: clive.ballard@ kcl.ac.uk
ABSTRACT
Background
There have been increasing concerns regarding the safety and efficacy of neuroleptics in people with dementia, but there are very few long-term trials to inform clinical practice. The aim of this study was to determine the impact of long-term treatment with neuroleptic agents upon global cognitive decline and neuropsychiatric symptoms in patients with Alzheimer disease.

Design: Randomised, blinded, placebo-controlled parallel two-group treatment discontinuation trial. Setting: Oxfordshire, Newcastle and Gateshead, London and Edinburgh, United Kingdom. Participants: Patients currently prescribed the neuroleptics thioridazine, chlorpromazine, haloperidol trifluoperazine or risperidone for behavioural or psychiatric disturbance in dementia for at least 3 mo. Interventions: Continue neuroleptic treatment for 12 mo or switch to an identical placebo. Outcome measures: Primary outcome was total Severe Impairment Battery (SIB) score. Neuropsychiatric symptoms were evaluated with the Neuropsychiatric Inventory (NPI). Results: 165 patients were randomised (83 to continue treatment and 82 to placebo, i.e., discontinue treatment), of whom 128 (78%) commenced treatment (64 continue/64 placebo). Of those, 26 were lost to follow-up (13 per arm), resulting in 51 patients per arm analysed for the primary outcome. There was no significant difference between the continue treatment and placebo groups in the estimated mean change in SIB scores between baseline and 6 mo; estimated mean difference in deterioration (favouring placebo) 0.4 (95% confidence interval [CI] 6.4 to 5.5), adjusted for baseline value (p 0.9). For neuropsychiatric symptoms, there was no significant difference between the continue treatment and placebo groups (n 56 and 53, respectively) in the estimated mean change in NPI scores between baseline and 6 mo; estimated mean difference in deterioration (favouring continue treatment) 2.4 (95% CI 8.2 to 3.5), adjusted for baseline value (p 0.4). Both results became more pronounced at 12 mo. There was some evidence to suggest that those patients with initial NPI 15 benefited on neuropsychiatric symptoms from continuing treatment.
Conclusions
For most patients with AD, withdrawal of neuroleptics had no overall detrimental effect on functional and cognitive status. Neuroleptics may have some value in the maintenance treatment of more severe neuropsychiatric symptoms, but this benefit must be weighed against the side effects of therapy. Trial registration: Cochrane Central Registry of Controlled Trials/National Research Register (#ISRCTN33368770).
0587
103
ARandomised,Blinded,Placebo-ControlledTrialinDementiaPatientsContinuingorStoppingNeuroleptics(TheDART-ADTrial)
Neuroleptics in Dementia Patients
Introduction
Worldwide, there are 25 million people with dementia [1], the majority of whom have Alzheimer disease (AD). It is a devastating illness that results in a progressive decline in cognitive ability and functional capacity, causes immense distress to patients, their carers, and families, and has an enormous societal impact. Currently the most frequent treatment issue for people with AD presenting to clinical services remains the management of neuropsychiatric symptoms, such as aggression, agitation, and psychosis. Over 90% of people with dementia develop these symptoms at some point during their illness [2]. The symptoms are frequently distressing for the patients who experience them [3] and problematic for their caregivers [4], in whom they are associated with clinically signicant depression [5]. In addition, they are often the precipitant for institutional care [6]. Neuroleptics are widely used as the rst-line pharmacological approach to treat these neuropsychiatric symptoms. Efcacy has been examined in eight randomized, placebocontrolled trials with typical neuroleptics [7,8] and 18 placebo-controlled trials with atypical neuroleptics [911]. The strongest evidence of efcacy is for risperidone, for which there are ve published trials indicating a modest but signicant diminution in aggression compared to placebo, but limited evidence of benet for other neuropsychiatric symptoms [911]. However, given that in the US and Europe, up to 60% of people with dementia residing in care facilities are prescribed neuroleptics (e.g., [12,13]) for median periods of greater than a year [14,15], the pivotal question is whether longer-term therapy with atypical neuroleptics confers any treatment benet. There are only two placebo-controlled trials of a neuroleptic for more than 14 wk, and neither showed signicant efcacy of neuroleptic treatment for neuropsychiatric symptoms [16,17]. Similarly, longitudinal cohort studies [18] and placebo-controlled neuroleptic withdrawal studies [14,15,19] do not indicate benet from neuroleptic therapy. However, all of the withdrawal studies continued for 3 mo or less, leaving some uncertainty regarding long-term symptom outcome. In addition, the largest study did suggest a benet for treatment with atypical neuroleptics in people with scores greater than 14 on the Neuropsychiatric Inventory (NPI) [15]. Any benecial effects of neuroleptics in people with AD must be weighed against the short- and long-term adverse effects which, according to meta-analyses, include parkinsonism, sedation, oedema, chest infections, stroke (odds ratio 2.53) and mortality (odds ratio 1.51.7) [711,2022]. Additional evidence has also highlighted accelerated cognitive decline as an important potential negative consequence of prolonged use of neuroleptics [16,23]. A meta-analysis [17] has conrmed this observation, indicating 0.7 of a Mini Mental State Examination (MMSE) point (95% condence interval [CI] 0.38 to 1.09) greater decline over 612 wk in neuroleptic-treated patients compared to those treated with placebo, which appears modest but represents a doubling in the expected rate of cognitive deterioration over this period. In the US the Food and Drug Administration [22] has warned about the risk of increased mortality and stroke with neuroleptics in people with dementia, and most practice guidelines recommend nonpharmacological approaches as
the rst-line treatment for agitation and other neuropsychiatric symptoms (e.g., [24]). However, there is still considerable debate as to the place of neuroleptics in the management of severe and distressing symptoms that are intractable to other treatment approaches, especially when there is potential risk to the patient or to others. The main aim of the trial was to determine whether ongoing treatment with neuroleptics accelerates cognitive decline in people with AD. We additionally sought to determine whether ongoing treatment with neuroleptics confers any benet for the long-term maintenance treatment of neuropsychiatric symptoms in people with AD.
Methods Participants
Participants were patients in Oxfordshire, South Birmingham, Newcastle and Gateshead, London and Edinburgh prescribed the neuroleptics thioridazine, chlorpromazine, haloperidol, triuoperazine or risperidone for behavioural or psychiatric disturbance in dementia for at least 3 months. Participants had to meet all inclusion criteria. The inclusion criteria were (a) patient lived in a nursing or residential home; (b) patient fullled the NINCDS/ADRDA criteria for possible or probable AD [25]; (c) patient had either a MMSE [26] score . 6 or a Severe Battery Impairment [27] score . 30; and (d) patient was taking at least 10 mg chlorpromazine equivalents (CPZe) of a typical neuroleptic or at least 0.5 mg daily of risperidone. The exclusion criteria were (a) patient was unable to complete primary outcome measures at baseline assessment; (b) clinician responsible for care or study clinician considered that the patient suffered from any physical condition including marked extrapyramidal disorderthat would have made participation in the trial distressing or likely to increase suffering; (c) patient was currently taking thioridazine and showing a prolonged QTc on electrocardiogram [28,29]; (d) the patient was likely to be unable to take capsules.
Ethics
Two caregivers of people with Alzheimers disease were closely involved in the development of the protocol, which was peer reviewed through the auspices of the Alzheimers Research Trust (Cambridge, UK). As this was a multicentre trial, the study was in the rst instance reviewed and approved by a properly constituted Multi-Centre Research Ethics Committee (the North of England MREC). Subsequent site-specic approval was then granted by properly constituted Local Research Ethics Committees at each of the participating centres. All Ethics Committees were conducted under the auspices of the Central Organization of Research Ethics Committees (COREC, now the National Research Ethics Service UK, http://www.nres.npsa.nhs.uk/).
Consent
Potentially suitable individuals residing in care facilities, their next of kin, and staff within the care facilities were provided with comprehensive information about the study. Those wishing to take part were invited to participate. If the potential participant had adequate capacity, the individual him- or herself was asked to complete the written study consent procedures. In these circumstances, the next of kin
104

Table 1. Fixed Dosage Regimens for the Respective Neuroleptics

Neuroleptic Very Low Low High
Risperidone 0.5 mg once daily 0.5 mg twice daily 1 mg twice daily Chlorpromazine 12.5 once daily 12.5 twice daily 25 twice daily Trifluoperazine 0.5 mg once daily 0.5 mg twice daily 1 mg twice daily Haloperidol 0.75 mg once daily 0.75 mg twice daily 1.5 twice daily
(or if no family members were in contact, an appropriate guardian, usually the manager of the care facility) was also asked to provide written assent to participation. If participants did not have adequate capacity, then written assent from the next of kin and agreement as far as could be ascertained from the potential participant were obtained and considered appropriate to enable participation. This procedure was fully approved by the MREC, and was standard procedure in clinical trials involving vulnerable adults in the UK, until subsequent legislative changes (Mental Capacity Act 2005), introduced after the current study, enabled consent to be provided by a caregiver.
Interventions
Participants were randomised in equal numbers either to continue neuroleptic treatment for 12 mo or to switch to placebo. Three xed dosages, named respectively (a) very low; (b) low; and (c) high, were chosen for each of the permitted neuroleptic drugs to correspond as near as possible to the dose the patient was being prescribed prior to trial entry (Table 1). Each of the neuroleptics was overencapsulated to conceal the identity of the contents. Placebo capsules were identical to the overencapsulated neuroleptics, but contained only inert ller. The respective treatments were maintained at the same xed dose throughout the 12 mo treatment period of the trial.
Objectives
The primary aim of this study was to determine whether treatment with neuroleptic agents is associated with an accelerated rate of cognitive decline in dementia. Secondary objectives were: (a) to examine the impact of neuroleptics on function and other cognitive outcomes; (b) to determine whether discontinuing neuroleptics was associated with an exacerbation of neuropsychiatric symptoms, both overall and in people with NPI scores above and below 14 [15]; (c) to examine the impact on parkinsonism; and (d) to determine the impact on global clinician rated outcome.
Outcomes
Primary outcomes. The primary outcome was the total SIB score [27] (change from baseline to 6 mo). This is a wellvalidated instrument designed to evaluate global cognitive functioning in individuals who are too impaired to complete standard neuropsychological tests. There are 40 questions in the SIB, assessing social interaction, memory, orientation, language, attention, praxis, visuospatial ability, construction, and orienting to name. A total score is obtained by summing all questions and ranged from 0 to 100. A higher score indicates higher cognitive ability.
Secondary outcomes. 1. Standardised Mini Mental State Examination (SMMSE) [26]: A widely used instrument for assessing cognitive mental status. It assesses orientation, attention, immediate and shortterm recall, language, and the ability to follow simple verbal and written commands. The maximum total score is 30. A higher total score indicates higher cognitive function. A standardized approach to the administration has been published and was adopted in the current study [26]. 2. FAS test of Verbal Fluency [30]: A verbal uency test for which the total score is presented as the sum of all acceptable words generated. 3. Bristol Activities of Daily Living Scale (BADLS) [31]: A 20item questionnaire to measure daily living abilities, specically in patients with dementia. The maximum score is 60. 4. Shefeld Test for Acquired Language Disorders (STALD) [32]: Developed as a nonspecialist clinical aid to help identify dysphasia. The test assesses receptive and expressive skills. It gives a total score ranging from 0 to 26. 5. NPI [33]: A caregiver-administered questionnaire that assesses 12 behaviours of patients on the basis of frequency and severity: delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, elation/euphoria, apathy/indifference, disinhibition, irritability, aberrant motor behaviour, night-time behaviours, and appetite/eating behaviours. Each behaviour is scored by multiplying the frequency and severity (i.e., frequency 3 severity); the higher the score, the greater the neuropsychiatric impairment. A total score can be calculated by summing the scores of all behaviours (range 1 to 144). Lower scores indicate less frequent/severe. 6. Functional Assessment Staging (FAST) [34]: Has seven main stages that consist of physical and instrumental active daily living, which are intended to project the progression of loss of function in patients with dementia. 7. Modied Unied Parkinsons Disease Rating Scale (MUPDRS): A modication of the full UPDRS, to focus only the items that were independent of cognitive function [35]. A score of 8 or more indicates signicant parkinsonism [35]. 8. Clinicians Global Impression of Change (CGIC): A widely used and validated rating scale [36] based on the health-care providers general clinical impressions with or without the informant input (i.e., family members). It evaluates global function and is scored from 1 (very much improved) to 7 (very much worsened). For scales requiring an informant, the information was provided by a nurse or professional caregiver who had regular contact with the individual, usually the key worker. As far as possible, the same informant provided information for subsequent assessments. The outcome assessment schedule is summarised in Table 2. Although the study was of 12 mo treatment (or discontinuation), our primary focus was the progression of cognitive impairment at the 6 mo assessment. This schedule was predetermined in view of the frailty and predicted high mortality of this patient population.
Sample Size
The planned sample size of this trial was 110 patients per treatment group. In the absence of data from randomized clinical trials to inform power calculations at the time the protocol was developed, this calculation was undertaken using two different approaches. First, based upon the
105
Table 1. Fixed Dosage Regimens for the Respective Neuroleptics

Neuroleptic Very Low Low High
Risperidone 0.5 mg once daily 0.5 mg twice daily 1 mg twice daily Chlorpromazine 12.5 once daily 12.5 twice daily 25 twice daily Trifluoperazine 0.5 mg once daily 0.5 mg twice daily 1 mg twice daily Haloperidol 0.75 mg once daily 0.75 mg twice daily 1.5 twice daily
(or if no family members were in contact, an appropriate guardian, usually the manager of the care facility) was also asked to provide written assent to participation. If participants did not have adequate capacity, then written assent from the next of kin and agreement as far as could be ascertained from the potential participant were obtained and considered appropriate to enable participation. This procedure was fully approved by the MREC, and was standard procedure in clinical trials involving vulnerable adults in the UK, until subsequent legislative changes (Mental Capacity Act 2005), introduced after the current study, enabled consent to be provided by a caregiver.
Interventions
Participants were randomised in equal numbers either to continue neuroleptic treatment for 12 mo or to switch to placebo. Three xed dosages, named respectively (a) very low; (b) low; and (c) high, were chosen for each of the permitted neuroleptic drugs to correspond as near as possible to the dose the patient was being prescribed prior to trial entry (Table 1). Each of the neuroleptics was overencapsulated to conceal the identity of the contents. Placebo capsules were identical to the overencapsulated neuroleptics, but contained only inert ller. The respective treatments were maintained at the same xed dose throughout the 12 mo treatment period of the trial.
Objectives
The primary aim of this study was to determine whether treatment with neuroleptic agents is associated with an accelerated rate of cognitive decline in dementia. Secondary objectives were: (a) to examine the impact of neuroleptics on function and other cognitive outcomes; (b) to determine whether discontinuing neuroleptics was associated with an exacerbation of neuropsychiatric symptoms, both overall and in people with NPI scores above and below 14 [15]; (c) to examine the impact on parkinsonism; and (d) to determine the impact on global clinician rated outcome.
Outcomes
Primary outcomes. The primary outcome was the total SIB score [27] (change from baseline to 6 mo). This is a wellvalidated instrument designed to evaluate global cognitive functioning in individuals who are too impaired to complete standard neuropsychological tests. There are 40 questions in the SIB, assessing social interaction, memory, orientation, language, attention, praxis, visuospatial ability, construction, and orienting to name. A total score is obtained by summing all questions and ranged from 0 to 100. A higher score indicates higher cognitive ability.
Secondary outcomes. 1. Standardised Mini Mental State Examination (SMMSE) [26]: A widely used instrument for assessing cognitive mental status. It assesses orientation, attention, immediate and shortterm recall, language, and the ability to follow simple verbal and written commands. The maximum total score is 30. A higher total score indicates higher cognitive function. A standardized approach to the administration has been published and was adopted in the current study [26]. 2. FAS test of Verbal Fluency [30]: A verbal uency test for which the total score is presented as the sum of all acceptable words generated. 3. Bristol Activities of Daily Living Scale (BADLS) [31]: A 20item questionnaire to measure daily living abilities, specically in patients with dementia. The maximum score is 60. 4. Shefeld Test for Acquired Language Disorders (STALD) [32]: Developed as a nonspecialist clinical aid to help identify dysphasia. The test assesses receptive and expressive skills. It gives a total score ranging from 0 to 26. 5. NPI [33]: A caregiver-administered questionnaire that assesses 12 behaviours of patients on the basis of frequency and severity: delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, elation/euphoria, apathy/indifference, disinhibition, irritability, aberrant motor behaviour, night-time behaviours, and appetite/eating behaviours. Each behaviour is scored by multiplying the frequency and severity (i.e., frequency 3 severity); the higher the score, the greater the neuropsychiatric impairment. A total score can be calculated by summing the scores of all behaviours (range 1 to 144). Lower scores indicate less frequent/severe. 6. Functional Assessment Staging (FAST) [34]: Has seven main stages that consist of physical and instrumental active daily living, which are intended to project the progression of loss of function in patients with dementia. 7. Modied Unied Parkinsons Disease Rating Scale (MUPDRS): A modication of the full UPDRS, to focus only the items that were independent of cognitive function [35]. A score of 8 or more indicates signicant parkinsonism [35]. 8. Clinicians Global Impression of Change (CGIC): A widely used and validated rating scale [36] based on the health-care providers general clinical impressions with or without the informant input (i.e., family members). It evaluates global function and is scored from 1 (very much improved) to 7 (very much worsened). For scales requiring an informant, the information was provided by a nurse or professional caregiver who had regular contact with the individual, usually the key worker. As far as possible, the same informant provided information for subsequent assessments. The outcome assessment schedule is summarised in Table 2. Although the study was of 12 mo treatment (or discontinuation), our primary focus was the progression of cognitive impairment at the 6 mo assessment. This schedule was predetermined in view of the frailty and predicted high mortality of this patient population.
Sample Size
The planned sample size of this trial was 110 patients per treatment group. In the absence of data from randomized clinical trials to inform power calculations at the time the protocol was developed, this calculation was undertaken using two different approaches. First, based upon the
106

impute. The procedure involves lling in missing data m times to generate m complete datasets. Then the m complete datasets are analysed by using standard statistical analyses, and nally, the results from the m complete datasets are combined to produce inferential results. Variables that were included in the imputation model were all assessments of SIB, NPI, FAS, SMMSE, BADLS, STALD-receptive, STALD-expressive, FAST, and CGIC collected at baseline, 6, and 12 mo. Other baseline covariates that were also included in the imputation model were allocated treatment group, sex, centre, presence of signicant EPS, visual hallucinations or delusions, age at randomisation, type of neuroleptic drug at baseline, and whether the participant was taking cholinesterase inhibitors at trial entry. 2. In order to test the robustness of the SIB result, we limited a sensitivity analysis to those patients for whom the risk of possible oor and ceiling effects was smallest, i.e., SIB baseline cut-off values 40 but 90. Prespecied subgroup analyses were carried out on change in SIB and change in NPI. We performed the test of interaction, i.e., examined whether the treatment effects were consistent across subgroups. The subgroups investigated were (a) type of neuroleptic (atypical versus typical)it made clinical sense to examine whether the type of neuroleptic (atypical versus typical) at baseline made a difference since, if one type was more harmful than the other, then discontinuation would benet those patients more; (b) baseline NPI ( 14 versus 15)we intended to look at consistency of treatment effect in relation to baseline NPI ( 14 versus 15 or more) in an attempt to replicate Ballard et al. 2004 [15]; and (c) centre effects (Newcastle versus Oxford versus London/Edinburgh). London and Edinburgh were grouped together because only a small number of participants were recruited from Edinburgh, and all were assessed by the London-based researchers.
between 6 and 12 mo follow-up (Figure 1). Over the full 12 mo of the study 47 people did not complete follow-up. Of these individuals 21 died and a further 26 (55%) stopped the allocated treatment. Baseline demographic and clinical characteristics were evenly balanced across the two groups (Tables 3 and 4). The majority of the participants were taking risperidone/placebo risperidone (n 101) or haloperidol/ placebo haloperidol (n 43) at baseline. A full breakdown of the neuroleptic drugs used in the study is given in Table 5.
Primary OutcomeSevere Impairment Battery

There was no signicant difference between the continue treatment and placebo groups (n 51 in both arms) in the estimated mean change in SIB scores between baseline and 6 mo (Figure 2; Table 6), 5.7 points (SD 14.2) deterioration for the placebo group, and a 6.2 (SD 16.0) deterioration for the continue treatment group; estimated mean difference in deterioration (favouring placebo) 0.4 (95% CI 6.4 to 5.5), adjusted for baseline value (p 0.9).
Secondary Outcomes
Neuropsychiatric symptoms. For the NPI, there was no signicant difference between the continue treatment and placebo groups (n 56 and 53, respectively) in the estimated mean change in NPI scores between baseline and 6 mo, 4.5 points (SD 17.6) deterioration for the placebo group compared to a 1.3 (SD 15.5) deterioration for the continue treatment group; estimated mean difference in deterioration (favouring continue treatment) 2.4 (95% CI 8.2 to 3.5), adjusted for baseline value (p 0.4). In patients with NPI scores 14 (Figure 3) the change in NPI over 6 mo was very similar in the neuroleptic and placebo groups (estimated difference 0.49, 95% CI 5.63 to 6.60), whereas in the people with NPI scores . 14 there was a ve-point, albeit nonsignicant, advantage for people continuing to receive neuroleptics (estimated difference 5.33, 95% CI 15.82 to 5.17). Cognition, language, and function. For the SMMSE, there was no signicant difference between the continue treatment and placebo groups (n 44 and 40, respectively) in the estimated mean change in SMMSE scores between baseline and 6 mo (Table 6), 1.0 points (SD 4.2) deterioration for the placebo group, and a 1.8 (SD 3.6) deterioration for the continue treatment group; estimated mean difference in deterioration (favouring placebo) 1.0 (95% CI 2.7 to 0.7), adjusted for baseline value (p 0.2). For the BADLS, likewise there was no signicant difference between the continue treatment and placebo groups (n 54 and 52, respectively) in the estimated mean change in BADLS scores between baseline and 6 mo, 0.2 points (SD 7.2) worsening for the placebo group, and a 1.8 (SD 8.9) worsening for the continue treatment group; estimated mean difference in improvement (favouring placebo) 1.7 (95% CI 1.2 to 4.6), adjusted for baseline value (p 0.2). For the STALD (receptive skills), once again there was no signicant difference between the continue treatment and placebo groups (n 34 and 39 respectively) in the estimated mean change in STALD (receptive skills) scores between baseline and 6 mo, 0.3 points (SD 2.1) deterioration for the placebo group, and a 0.5 (SD 1.7) deterioration for the continue treatment group; estimated mean difference in deterioration (favouring placebo) 0.2 (95% CI 1.1 to 0.6),
0591 April 2008 | Volume 5 | Issue 4 | e76
Study Governance
A three-person independent data-monitoring committee (DMC) was charged with overseeing patient safety. Its remit included prompt review of serious adverse events and a comprehensive review of all adverse events based upon interim data reports. The remit also required advice regarding any new or emerging information on the safety of the study treatments. If required, the DMC would make recommendations to the trial steering group and the sponsor about the safe continuation of the trial and any issues of concern. These decisions relied upon the independence and expertise of the DMC members and did not follow any strict stopping rules.
Results Participant Flow

The rst patient was randomised in October 2001 and the last in December 2004. 165 patients were randomised (83 to continue treatment and 82 to placebo, i.e., discontinue treatment), of whom 128 (78%) commenced treatment (64 continue/64 placebo) (Figure 1). Of those, 26 were lost to follow-up (13 per arm), resulting in 51 patients per arm analysed for the primary outcome. Of those (102 total), 77 remained on allocated treatment at 6 mo follow-up (40 continue/37 placebo). A further 22 participants withdrew
107
Figure 1. Flow of Participants through the Trial doi:10.1371/journal.pmed.0050076.g001
108
adjusted for baseline value (p 0.6). For the STALD there was no signicant difference between the continue treatment and placebo groups (n 34 and 39, respectively) in the estimated mean change in receptive language scores between baseline and 6 mo. For expressive language there was no signicant difference of placebo compared to the treatment group; estimated mean difference (favouring placebo) 1.0 (95% CI 2.0 to 0.04), adjusted for baseline value (p 0.06). However, for the FAS, there was strong evidence, i.e., a highly signicant difference between the continue treatment and placebo groups (n 34 and 31, respectively) in the estimated mean change in FAS totals between baseline and 6 mo. There was 0.6 points (SD 6.2) improvement for the placebo group compared to a 3.2 (SD 6.6) deterioration for the continue treatment group; estimated mean difference
(favouring placebo) 4.5 (95% CI 7.3 to 1.7), adjusted for baseline value (p 0.002). Parkinsonism. For the M-UPDRS, there was a slight but nonsignicant difference between the continue treatment and placebo groups (n 41 and 43, respectively) in the estimated mean change in M-UPDRS scores between baseline and 6 mo, 0.4 points (SD 3.2) improvement for the placebo group compared to a 0.8 (SD 4.1) deterioration for the continue treatment group; estimated mean difference (favouring placebo) 1.1 (95% CI 0.4 to 2.6), adjusted for baseline value (p 0.1).
Global Outcome
For the change in FAST and CGIC, there was no evidence whatsoever of any differences between the continue treatApril 2008 | Volume 5 | Issue 4 | e76

Table 3. Demographic and Clinical Characteristics and Assessments at Baseline (All Patients Randomized)
Baseline Characteristics or Assessment
Women Age Significant EPS Visual hallucinations Delusions Taking cholinesterase inhibitor Taking atypical neuroleptic prerandomisation SMMSE SIB NPI FAST M-UPDRS BADLS STALD (receptive skill) STALD (expressive skill) FAS
Statistics
No. (%) Mean (SD) [range], y No. (%) No. (%) No. (%) No. (%) No. (%) Mean (SD) (no. of patients) No. (%) scoring below 6 Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) No (%) scoring 8 or more Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR)
Continue Treatment (n 83)

64 (77.1) 84.8 (7.0) [68.3100.2] 39 (47.0) 10 (12.0) 27 (32.5) 2 (2.4) 57 (68.7) 11 (6) (n 83) 16 (19.3) 71.1 (22.7) (n 75) 77 (5891) 17.4 (14.6) (n 75) 15 (524) 5.7 (0.8) (n 72) 6 (66) 2.7 (3.8) (n 64) 6 (9.4) 18.5 (9.0) (n 73) 19 (1224) 3.8 (1.9) (n 59) 4 (35) 6.3 (3.0) (n 59) 7 (49) 8.4 (7.3) (n 48) 7 (313)
Placebo (n 82)
62 (75.6) 84.9 (6.1) [67.0100.6] 39 (47.6) 10 (12.0) 27 (32.9) 3 (3.7) 58 (70.7) 11 (5) (n 82) 14 (17.1) 73.8 (20.7) (n 71) 80 (6392) 15.8 (11.3) (n 70) 14 (624) 5.5 (0.8) (n 72) 6 (56) 2.7 (3.9) (n 64) 10 (15.6) 17.9 (9.5) (n 71) 16 (1223) 4.4 (2.2) (n 58) 5 (56) 7.2 (2.8) (n 58) 8 (510) 10.9 (7.9) (n 45) 10 (514)
ment and placebo groups (Wilcoxon rank-sum test p 0.9 for both).
Prespecified Sensitivity Analyses

Imputed values were included for 12 participants, six in each treatment arm. These individuals had similar characteristics to the overall trial population, but as would be expected for participants unable to complete the trial, at baseline they were more cognitively impaired (SIB 6 SD, 68.1 6 22.3 versus 74.4 6 21.9) and had higher levels of neuropsychiatric symptoms (NPI 6 SD, 22.2 6 14.8 versus 16.0 6 13.9) than patients who subsequently completed the 6 month assessments. The results for SIB (mean difference 2.1, 95% CI 8.8 to 4.6, p 0.5), NPI, SMMSE, UPDRS, BADLS, STALD, and FAS all remained consistent when missing data were replaced using multiple imputation techniques. In addition, the result for sensitivity analysis of SIB excluding participants whose baseline SIB was , 40 or . 90, i.e., excluding patients unlikely to benet from treatment (mean difference 0.2, 95% CI 9.2 to 8.9, p 0.9) was again consistent with the primary result. Prespecied subgroup analysis. There was no evidence of any interaction between treatment group and the various subgroups (Figures 3 and 4).
performed on SMMSE and FAS to assess if results remained consistent when these measures were included. Once again, there were no substantive differences between the primary and sensitivity analyses, i.e., the results are very robust. Although not specied in the original analysis plan, to avoid omitting potentially important clinical differences, additional descriptive data were also obtained regarding emergent delusions and agitation in participants who did not have these symptoms at baseline. The pattern of emergent symptoms appeared similar in the two treatment arms at 1, 3, and 6 mo assessments, respectively, for both delusions (neuroleptics 14%, 16%, 5% versus placebo 5%, 8%, 13%) and agitation (neuroleptics 17%, 23%, 32% versus placebo 23%, 26%, 34%).
12 Month Assessment Data

Analysis at 12 mo was limited to the two main outcomes, cognitive function and neuropsychiatric features, given the large amount of missing data. There was no signicant difference between the continue treatment and placebo groups (n 28 and 27, respectively) in the estimated mean change in SIB scores between baseline and 12 mo, although there was a clinically important but not statistically signicant numerical advantage of 8 points for the placebo group 8.5 points (SD 13.4) deterioration in the placebo group, and a 16.5 (SD 23.1) deterioration in the continue treatment group; estimated mean difference in deterioration (favouring placebo) 8.4 (95% CI 18.6 to 1.7), adjusted for baseline value (p 0.1). For the NPI, there was, however, a signicant difference between the continue treatment and placebo groups (n 28 and 31 respectively) in the estimated mean change in NPI scores between baseline and 12 mo, with 11.4 points (SD 17.7)
0593 April 2008 | Volume 5 | Issue 4 | e76
Post-Hoc Additional Exploratory Sensitivity Analysis

Given that a substantial proportion of patients did not actually start allocated treatment (;22%), we decided to examine the robustness of results by performing an analysis with all data available. In addition, according to the protocol, all patients with SMMSE , 6 at pretrial screening were not required to complete SMMSE and FAS at any subsequent assessments; however, some measures were collected during the follow-up. Therefore, additional sensitivity analysis was
109
Table 4. Demographic and Clinical Characteristics and Assessments at Baseline (Analysis Population)
Baseline Characteristics
Women Age Significant EPS Visual hallucinations Delusions Taking cholinesterase inhibitor Taking atypical neuroleptic prerandomisation SMMSE SIB NPI FAST M-UPDRS BADLS STALD (receptive skill) STALD (expressive skill) FAS
Statistics
No. (%) Mean (SD) [range], y No. (%) No. (%) No. (%) No. (%) No. (%) Mean (SD) (no. of patients) No. (%) scoring below 6 Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) No (%) scoring 8 or more Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR) Mean (SD) (no. of patients) Median (IQR)

38 (74.5) 84.5 (7.7) [68.3100.0] 28 (54.9) 7 (13.7) 21 (41.2) 1 (2.0) 32 (62.8) 12 (5) (n 45) 1 (2.2) 72.4 (22.9) (n 51) 78 (5091) 17.2 (15.9) (n 51) 12 (424) 5.7 (0.8) (n 49) 6 (66) 2.3 (3.0) (n 48) 3 (6.2) 17.7 (8.5) (n 50) 19 (1223) 4.3 (1.7) (n 39) 4 (36) 6.8 (2.8) (n 39) 7 (59) 8.6 (6.8) (n 37) 7 (313)
Placebo (n 51)
38 (74.5) 84.4 (6.4) [67.0100.6] 28 (54.9) 6 (11.8) 18 (35.3) 3 (5.9) 37 (72.6) 13 (4) (n 42) 1 (2.4) 76.3 (20.8) (n 51) 82 (6893) 14.7 (11.5) (n 50) 11 (523) 5.4 (0.9) (n 51) 6 (46) 2.3 (3.5) (n 46) 6 (13.0) 16.1 (8.6) (n 50) 16 (821) 4.5 (2.3) (n 44) 5 (36) 7.2 (2.9) (n 44) 8 (510) 11.3 (8.8) (n 32) 10 (516)
deterioration for the placebo group compared to a 1.4 (SD 22.1) deterioration for the continue treatment group; estimated mean difference in deterioration (favouring continue treatment) 10.9 (95% CI 20.1 to 1.7), adjusted for baseline value (p 0.02). A cut-off 14 on the NPI was again helpful in interpreting the results. For participants with baseline NPI scores below this threshold, there was no signicant difference between treatment groups; estimated mean difference in deterioration (favouring continue treatment) 5.2 (95% CI 15.8 to 5.4), whereas for individuals with
higher NPI scores there was a signicant 16.9 point advantage for the group who continued neuroleptics (95% CI 32.5 to 1.2). However, the test of interaction, although underpowered, was not signicant (p 0.2); therefore there is no evidence of a statistical interaction.
Discussion
Because of difculties in identifying people with Alzheimer disease in nursing homes who were taking neuroleptics and
Table 5. Details of Type and Dose of Treatment before Randomisation by Allocated Treatment Group
Prerandomisation Current TreatmentName of Neuroleptic
Chlorpromazine Haloperidol Olanzapine Promazine Quetiapine Risperidone Thioridazine Trifluoperazine
Randomised Allocation Dose

High Very low/low High Very low/low High Very low/low High Very low/low High Very low/low High Very low/low High Very low/low High Very low/low

0 1 1 22 0 4 0 0 0 1 9 43 0 2 0 0
Placebo (n 82)
1 1 0 20 2 3 0 2 0 3 12 37 0 0 1 0
110
0594

Figure 2. Change in Total SIB Score (Baseline to 6 Months) by Treatment Group doi:10.1371/journal.pmed.0050076.g002
were able to complete the rigorous cognitive assessments, the recruitment target sample size based on the power calculation was not attained. However, despite this drawback, we report the largest and longest-duration randomized placebo controlled trial of neuroleptic discontinuation. To our knowledge this is the rst study of this type to evaluate outcome over 6 mo and beyond. Treatment with neuroleptics was not associated with signicantly greater decline in global cognitive function than placebo, although there were numerical advantages for the placebo-treated group on the SIB and the SMMSE (1 point decrease on SMMSE, 0.4 point decrease on SIB overall, ;3 point decrease on SIB in people with NPI scores 14) at 6 mo, which became more pronounced by month 12, at which point there was an 8 point advantage on the SIB for the placebo-treated group, equivalent to approximately 6 mo of average expected cognitive decline. The failure of these differences to attain statistical signicance may be because of limited statistical power (a type II error), as the magnitude of difference in change in global cognition between neuroleptics and placebo at 6 mo was consistent with the effect size identied in a recent meta-analysis [11], and became more substantial over further follow-up. It would be more usual in a study of this type to hypothesise equivalence or the absence of an advantage to the treatment that is being removed, and there is certainly no evidence at all from these results suggesting any cognitive advantage favouring antipsychotics. On secondary cognitive outcomes, there was a signicant deterioration in verbal uency for patients taking neuroleptics compared to people receiving placebo, and there was a nonsignicant numerical advantage for the placebo-treated group on the BADLS. There were nonsignicant numerical
advantages for the placebo group with respect to the severity of parkinsonism. There was a marginal nonsignicant 2.4 point advantage on the total NPI score for continuing neuroleptic treatment over the rst 6 mo of treatment. Using a baseline NPI threshold 14, previously reported to be predictive of outcome in a 3 mo neuroleptic withdrawal trial [15], the change in NPI did not differ between the treatment groups. Participants with baseline NPI scores . 14 had an almost 5 point advantage (albeit nonsignicant) if they remained on neuroleptics, whereas there was no benet for people with NPI scores below this threshold. For patients with more severe neuropsychiatric symptoms, there were modest benets at 6 mo and more substantial advantages at 12 mo, which have to be weighed against the potential for serious adverse events. Some of the changes in NPI score are likely to be related to natural symptom course, or a Hawthorne effect, or regression to the mean, although there should be no imbalance in these factors between groups. There were no differences between groups for global clinician-rated outcome, and in an additional descriptive evaluation there appeared to be no difference in emergent delusions or agitation between groups. In the post-hoc analysis there was no indication of a difference between people taking typical or atypical neuroleptics. The majority of individuals were taking risperidone or haloperidol, and the number of people taking other drugs was too small to enable any meaningful comparison. In particular, it will be important in further work to determine whether neuroleptics with more prominent antimuscarinic properties have a more potent impact on cognition in patients with dementia. Several studies have demonstrated that psychological management approaches can replace neuroleptic therapy
111
Table 6. Summary of Change from Baseline to 6 Months for Main Outcomes plus Comparative Statistics
Assessment Measure or Category Mean Change (SD)a Continue Treatment
SIB NPI SMMSE M-UPDRS Bristol ADL STALD (receptive skill) STALD (expressive skill) FAS Change in FASTe Cognition Neuropsychiatric symptoms Cognition Parkinsonism Function Receptive language Expressive language Verbal fluency Dementia stage 2 1 0 1 2 Improvement rated by clinician Very much improved Much improved Minimally improved No change Minimally worse Much worse Very much worse (n 51) 6.2 (16.0) (n 56) 1.3 (15.5) (n 44) 1.8 (3.6) (n 41) 0.8 (4.1) (n 54) 1.8 (8.9) (n 34) 0.5 (1.7) (n 34) 0.6 (1.8) (n 34) 3.2 (6.6) (n 53) 0 3 34 12 4 (n 48) 1 (2%) 3 (6%) 7 (15%) 18 (37%) 9 (19%) 7 (15%) 3 (6%)
Placebo
(n 51) 5.7 (14.2) (n 53) 4.5 (17.6) (n 40) 1.0 (4.2) (n 43) 0.4 (3.2) (n 52) 0.2 (7.2) (n 39) 0.3 (2.1) (n 39) 0.2 (2.5) (n 31) 0.6 (6.2) (n 53) 1 4 32 8 8 (n 48) 0 0 14 (29%) 14 (29%) 10 (21%) 10 (21%) 0
Estimated Mean Difference in Changeb (95% CI)

0.4 1.4 2.4 4.3 1.0 0.9 1.1 1.3 1.7 1.0 0.2 0.3 1.0 0.9 4.5 4.6 (6.4 to 5.5)c (5.4 to 8.1)d (8.2 to 3.5)c (10.7 to 2.1)d (2.7 to 0.7)c (2.6 to 0.8)d (0.4 to 2.6)c (0.4 to 3.0)d (1.2 to 4.6)c (2.0 to 3.9)d (1.1 to 0.6)c (1.1 to 0.6)d (2.0 to 0.04)c (2.0 to 0.2)d (7.3 to 1.7)c (7.4 to 1.7)d
p-Value
CGICe
0.9 0.7 0.4 0.2 0.2 0.3 0.1 0.1 0.2 0.5 0.6 0.5 0.06 0.09 0.002 0.002 0.9 0.9
Positive change represents improvement (except for NPI, M-UPDRS, and FAST). Placebo is the reference group. Positive difference favours Continue Treatment group (except for NPI, M-UPDRS, and FAST). c Analysis of covariance (ANCOVA) adjusted for baseline measurement. d ANCOVA adjusted for baseline measurement (continuous) and minimisation factors centre (categories 13, reference Newcastle), significant EPS (reference category absent), visual hallucinations (reference category absent), delusions (reference category absent), taking cholinesterase inhibitors (reference category no), baseline SMMSE (continuous), and type of neuroleptic prerandomisation (reference category atypical). e Wilcoxon rank-sum test. doi:10.1371/journal.pmed.0050076.t006
b
112
without any signicant worsening of neuropsychiatric symptoms [14,39]; and evidence is emerging that cholinesterase inhibitors [40] or memantine [41] may be safer, effective alternatives for some symptoms. The authors of the recent CATIE study [17], a large, pragmatic, 36-wk placebo-controlled trial of atypical neuroleptics in AD, concluded that the modest benets were not sufcient to justify therapy in the presence of the increased risk of serious adverse events. Clinicians should certainly try to replace atypical neuroleptics with safer management approaches. Taking into consideration CATIE, the results of 6- to 12-wk placebocontrolled trials, and our own data, we would suggest that there is, however, a limited place for atypical neuroleptics in the maintenance treatment of severe neuropsychiatric manifestations (particularly aggression) in AD when there is tangible risk or severe distress, and the symptoms have been refractory to other treatment approaches. The magnitude of the impact of neuroleptics upon cognition, although consistent with a recent meta-analysis, was considerably less marked than reported in a recent 6 mo placebo-controlled trial of quetiapine [16]. There are numerous possible explanations for this difference, although
it may be noted that unlike quetiapine, none of the atypical antipsychotics used in the current study had substantial antimuscarinic properties, and it is possible that antimuscarinic properties may exacerbate the impact of neuroleptics upon cognition. Although this is speculative, it is consistent with a study comparing cognition in patients with AD treated with either risperidone or olanzapine, where olanzapine treatment was associated with greater impairment of attentional and executive performance related to anticholinergic activity [42]. Within the current study there was a signicant detrimental impact upon expressive language function, an important skill to enable social communication and maintain quality of life in people with AD residing in care facilities. Further work is needed to examine the effects on different aspects of cognitive function, to clarify the differential impact of individual neuroleptic drugs, and to determine whether the impact upon cognition is sufcient to interfere with everyday activities. The results of the current study must be interpreted within the context of a number of limitations. In particular, the sample size was much smaller than intended, conferring limited statistical power, and the number of deaths and

Figure 3. Subgroup Analysis (Change in NPI from Baseline to 6 Months) doi:10.1371/journal.pmed.0050076.g003
withdrawals precluded meaningful analysis of data beyond the 6 mo follow-up. In addition a sizeable proportion of patients did not start their allocated treatment for a variety of reasons, mainly related to frailty and concurrent illnesses. The sample size achieved was short of the 220 target due mainly to problems identifying eligible patients, which in turn led to slow recruitment, bringing on more centres, and ultimately curtailing recruitment due to a lack of resources. Given the vulnerability of the study population, a substantial number of deaths and withdrawals are an almost inevitable
problem to contend with. Probably the only solution would be to exclude people with profound dementia or with a certain degree of physical frailty, but this would then diminish the validity of drawing more general conclusions from the results. Although it is difcult, therefore, to see how this problem could have been avoided, the high number of drop-outs must be considered when interpreting the results. In addition, the reason for the initial prescription of neuroleptic drugs was unclear in the majority of instances. Most prescriptions had been instigated by primary care
Figure 4. Subgroup Analysis (Change in SIB from Baseline to 6 Months) doi:10.1371/journal.pmed.0050076.g004 PLoS Medicine | www.plosmedicine.org 0597 April 2008 | Volume 5 | Issue 4 | e76
113
physicians and, as a number of the individuals had changed their primary care physician, or been admitted to a care facility, or changed care facility since neuroleptics were rst prescribed, the clinical indication for the original prescription was often lost. Although in many ways this lack of information is unsatisfactory, it does reect real clinical practice, so that the population studied and the information available were representative of what faces clinicians in their routine practice. As few individuals were under specialist care, it is unlikely that treatment-refractory symptoms were a reason for neuroleptic use in most of the participants. The data regarding outcome and the baseline severity of symptoms do, however, provide a useful basis for clinical decisions. It is possible that the prole of the original symptoms for which the neuroleptics were prescribed may have inuenced outcome, but this possibility could practically be investigated only in people with neuroleptic prescriptions of shorter duration, for which the presenting symptoms would be easier to ascertain.
reviewer of the final statistical report. Competing Interests: CB has received honoraria from Novartis, Pfizer, Shire, Lundbeck, Myriad, Janssen, Astra Zeneca, and Servier pharmaceutical companies and research grants from Novartis, Lundbeck, Astra-Zeneca, and Janssen pharmaceuticals. The remaining authors have declared that they have no competing interests. Funding: The DART-AD project was made possible by a grant from The Alzheimers Research Trust, Cambridge, UK (http://www. alzheimers-research.org.uk) to Profs Ballard and Jacoby and to RM. The peer review process undertaken by the funder did result in some modifications to the study design. The funder has no other role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
References 1. Ferri CP, Prince M, Brayne C, Brodaty H, Fratiglioni L, et al. (2005) Alzheimers Disease International. Global prevalence of dementia: a Delphi consensus study. Lancet 366: 21127. 2. Ballard C, Ayre G, Gray A (1999) Psychotic symptoms and behavioral disturbances in dementia: A review. Rev Neurol 155: 4452. 3. Gilley DW, Whalen ME, Wilson RS, Bennett DA (1991) Hallucinations and associated factors in Alzheimers disease. J Neuropsych Clin Neurosci 3: 371376. 4. Rabins PV, Mace NL, Lucas MJ (1982) The impact of dementia on the family. J Am Med Soc 248: 333335. 5. Donaldson C, Tarrier N, Burns A (1997) The impact of the symptoms of dementia on caregivers. Brit J Psychiatr 170: 6268. 6. Steele C, Rovner B, Chase GA (1990) Psychiatric symptoms and nursing home placement of patients with Alzheimers disease. Am J Psych 147: 104951. 7. Schneider LS, Pollock VE, Lyness SA (1990) A metaanalysis of controlled trials of neuroleptic treatment in dementia. J Am Geriatrics Soc 38: 55363. 8. Lonergan E, Luxenberg J, Colford J, Birks J (2005) Haloperidol for agitation in dementia. Cochrane Database Syst Rev 4. Art. No.: CD002852. doi:10. 1002/14651858.CD002852 9. Ballard C, Howard R (2006) Neuroleptic drugs in dementia: benets and harm. Nat Rev Neurosci 7: 492500. 10. Ballard C, Waite J (2006) The effectiveness of atypical antipsychotics for the treatment of aggression and psychosis in Alzheimers disease. Cochrane Database Syst Rev 1: Art. No.: CD003476. doi:10.1002/14651858.CD003476. pub2 11. Schneider LS, Dagerman K, Insel PS (2006) Efcacy and adverse effects of atypical antipsychotics for dementia: meta-analysis of randomized, placebo-controlled trials. Am J Geriatr Psychiatry 14: 191210. 12. McGrath AM, Jackson GA (1996) Survey of prescribing in residents of nursing homes in Glasgow. BMJ 314: 611612. 13. Margallo-Lana M, Swann A, OBrien J, Fairbairn A, Reichelt K, et al. (2001) Prevalence and pharmacological management of behavioral and psychological symptoms amongst dementia sufferers living in care environments. Int J Geriatric Psych 16: 3944. 14. Cohen-Manseld J, Lipson S, Werner P, Billig N, Taylor L, et al. (1999) Withdrawal of haloperidol, thioridazine, and lorazepam in the nursing home: a controlled, double-blind study. Arch Int Med 159: 173340. 15. Ballard CG, Thomas A, Fossey J, Lee L, Jacoby R, et al. (2004) A 3-month, randomized, placebo-controlled, neuroleptic discontinuation study in 100 people with dementia: the neuropsychiatric inventory median cutoff is a predictor of clinical outcome. J Clin Psych 65: 114119. 16. Ballard C, Margallo-Lana M, Juszczak E, Douglas S, Swann A, et al. (2005) Quetiapine and rivastigmine and cognitive decline in Alzheimers disease: randomised double blind placebo controlled trial. BMJ 330: 874. 17. Schneider LS, Tariot PN, Dagerman KS, Davis SM, Hsiao JK, et al. (2006) CATIE-AD Study Group. Effectiveness of atypical antipsychotic drugs in patients with Alzheimers disease. New Engl J Med 355: 152538. 18. Ballard C, Margallo-Lana ML, Fossey J, Reichelt K, Mynt P, et al. (2001) A one year follow-up study of behavioral and psychological symptoms in dementia among people in care environments. J Clin Psych 62: 631636. 19. Bridges-Parlet S, Knopman D, Steffes S (1997) Withdrawal of neuroleptic medications from institutionalised dementia patients, results of a doubleblind baseline-treatment-controlled pilot study. J Geriatric Psych Neurol 10: 119126. 20. Schneider LS, Dagerman KS, Insel P (2005) Risk of death with atypical antipsychotic drug treatment for dementia: Meta-analysis of randomized placebo-controlled trials. JAMA 294: 19341943. 21. Committee for the Safety of Medicines (2004 March 9) Message from Chairman, Committee on Safety of Medicines to medical contacts: Atypical antipsychotic drugs and stroke [memo] Edinburgh (Scotland): The Scottish Government. Available: http://www.scotland.gov.uk/Publications/2004/03/ 19074/34348. Accessed 21 February 2008. 22. FDA (2005) Deaths with antipsychotics in elderly patients with behavioral disturbances. Washington (D.C): U.S. Food and Drug Administration, FDA Public Health Advisory, Centre for Drug Evaluation and Research 13-72005. 23. McShane R, Keene J, Gedling K, Fairburn C, Jacoby R, et al. (1997) Do
Conclusion
For most patients with AD, withdrawal of neuroleptics had no overall detrimental effect on functional and cognitive status and by some measures improved functional and cognitive status. Neuroleptics may have some value in the maintenance treatment of more severe neuropsychiatric symptoms, but this possibility must be weighed against the unwanted effects of therapy. The current study helps to inform a clinical management strategy for current practice, but the considerable risks of maintenance therapy highlight the urgency of further work to nd, develop, and implement safer and more effective treatment approaches for neuropsychiatric symptoms in people with AD.
Text S1. CONSORT Checklist Found at doi:10.1371/journal.pmed.0050076.sd001 (44 KB PDF).
Acknowledgments
Other DART AD investigators: Timothy Smith, Ruth Elvish, Claire Maddison, Carol Bannister. Data Monitoring and Ethics Committee: Clive Holmes (Chair), John Geddes and Robert Clarke. DART-AD Pharmacists: (North East) Vranda Barbieri, Kay Carson, Gavin Mankin; (Oxfordshire) Michael Marven, Kim Gray; (London) Charlotte Gibbs. Author contributions. CB, RMS, RJ, and EJ designed the experiments/the study. MML, MT, SD, and KK collected data or did experiments for the study. RMS, LY, and EJ analyzed the data. MML, MT, SD, and RJ enrolled patients. CB and EJ wrote the first draft of the paper. SD, RMS, RJ, and EJ contributed to the writing of the paper. MT collected all the data for the Oxfordshire patients over the first two years. SD managed the trial in Newcastle for the majority of its running time, collecting data and also setting up the trial in London. SD also entered data and helped to track down missing data prior to analysis, proof read the manuscript and offered suggestions for changes. RJ contributed to the writing of the grant application and was one of the holders of the grant from the Alzheimers Research Trust. RJ was also a principal investigator for the study and supervised the collection of the clinical data in the Oxford (UK) patients. KK identified patients, collected data, and wrote the followup protocol for the study. LY critically reviewed the manuscript. EJ was the trial statistician and part of the trial management group which met every 2 months. EJ presented regular interim analyses to the Data Monitoring Committee, and co-authored the annual report to the funder. EJ co-wrote the statistical analysis plan and was senior PLoS Medicine | www.plosmedicine.org 0598
114

neuroleptic drugs hasten cognitive decline in dementia? Prospective study with necropsy follow up. BMJ 314: 266270. Howard R, Ballard C, OBrien J, Burns A (2001) UK and Ireland Group for Optimization of Management in dementia. Guidelines for the management of agitation in dementia. Int J Geriatric Psych 16: 7147. McKhann G, Drachman D, Folstein M, Katzman R, Price D (1984) Clinical diagnosis of Alzheimers disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimers Disease. Neurol 34: 939944. Molloy DW, Alemayehu E, Roberts R (1991) Reliability of a standardized Mini-Mental State Examination compared with the traditional Mini-Mental State Examination. Am J Psych 148: 1025. Schmitt FA, Ashford W, Ernesto C, Saxton J, Schneider LS, et al. (1997) The severe impairment battery: concurrent validity and the assessment of longitudinal change in Alzheimers disease. The Alzheimers disease cooperative study. Alzheimer Dis Assoc Disord 11: 5156. Committee on Safety of Medicines (2000 December 11) Thioridazine: Restricted indications and new warnings on cardiotoxicity [memo]. London: Department of Health, National Health Service. CEM/CMO/ 2000/18. Available: http://www.info.doh.gov.uk/doh/embroadcast.nsf/ f011981a95f31f4180256c07003d34a0/fc5d1f63a16da86b80256dad00480816? OpenDocument. Accessed: 21 February 2008. Reilly JG, Ayis SA, Ferrier IN, Jones SJ, Thomas SH (2000) QTc-interval abnormalities and psychotropic drug therapy in psychiatric patients. Lancet 355: 104852. Benton AL, Hamsher K, Sivan AB (1983) Multilingual Aphasia Examination. 3rd edition. Iowa City (Iowa): AJA Associates. Byrne LM, Wilson PM, Bucks RS, Hughes AO, Wilcock GK (2000) The sensitivity to change over time of the Bristol Activities of Daily Living Scale in Alzheimers disease. Int J of Ger Psych 15: 65661. Syder D, Levy C (1998) Wanting to talk: Counselling case studies in communication disorders. Philadelphia (Pennsylvania): Whurr Publishers. pp. 256288. Cummings JL, Mega M, Gray K, Rosenburg-Thomas S, Carusi DA (1994) The Neuropsychiatric Inventory: comprehensive assessment of psychopathology in dementia. Neurol 44: 23082314. Sclan SG, Reisberg B (1992) Functional assessment staging (FAST) in Alzheimers disease: reliability, validity, and ordinality. Int Psychogeriatrics 1: 5569. Ballard C, McKeith I, Burn D, Harrison R, OBrien J, et al. (1997) The UPDRS scale as a means of identifying extrapyramidal signs in patients suffering from dementia with Lewy bodies. Acta Neurol Scand 96: 366371. Schneider LS, Olin JT, Doody RS, Clark CM, Morris JC, et al. (1997) Validity and reliability of the Alzheimers Disease Co-operative Study-Clinical Global Impression of Change. The Alzheimers Disease Co-operative Study. Alzheimer Dis Assoc Disord 11: 2232. Cohen J (1992). A power primer. Psychol Bull 112: 155159. [No author listed] (2002) Stata Statistical Software release 7.0. College Station (Texas): Stata Corporation. Fossey J, Ballard C, Juszczak E, James I, Alder N (2006) Effect of enhanced psychosocial care on antipsychotic use in nursing home residents with severe dementia: cluster randomised trial. BMJ 332: 756761. Gauthier S, Feldman H, Hecker J, Vellas B, Ames D, et al. (2002) Donepezil MSAD Study Investigators Group. Efcacy of donepezil on behavioral symptoms in patients with moderate to severe Alzheimers disease. Int Psychogeriatrics 14: 389404. Cummings JL, Schneider E, Tariot PN, Graham SM (2006) Behavioral effects of memantine in Alzheimers disease patients receiving donepezil treatment. Neurology 67: 5763. Mulsant BH, Gharabawi GM, Bossie CA, Mao L, Martinez RA (2004) Correlates of anticholinergic activity in patients with dementia and psychosis treated with risperidone or olanzapine. J Clin Psych 65: 170814.
24. 25.
33. 34. 35. 36.
26. 27.
28.
37. 38. 39. 40.
29. 30. 31. 32.
41. 42.
Editors Summary
Background The number of people with dementia (currently 25 million worldwide) is expected to increase by 5 million each year. The risk of dementia, including Alzheimer disease, increases sharply with age: Alzheimers Disease International estimates that 1.4% of people 6569 have dementia, whereas almost a full quarter of those over the age of 85 years are affected. Almost all older dementia patients will experience, along with the cognitive and functional decline typical of the illness, some neuropsychiatric symptoms. These symptoms can include agitation, aggression, and psychosis, and are often devastating for the older patient and his or her family and caregiver. Managing these symptoms is often a prime concern for health-care providers and families. Neuroleptics (sometimes called antipsychotics) are the class of drugs often used to manage or control neuropsychiatric problems, but there have been questions about their safety and appropriateness. Safety concerns involve risk of stroke, parkinsonism, sedation, edema, and chest infections but also include a worsening of cognitive decline with prolonged use of neuroleptics. Why Was the Study Done? Previous studies on the effectiveness and safety of neuroleptics in older people have been short term. Ballard and colleagues wanted to study over a longer period of time the impact of neuroleptic drugs on elderly patients with dementia. Specifically, they wanted to know if being on a neuroleptic was associated with more cognitive decline than coming off the drug. They also wanted to investigate whether discontinuing the drug exacerbated any neuropsychiatric symptoms, Parkinson disease-like symptoms, or other functional, language, and cognition difficulties frequently associated with dementia. What Did the Researchers Do and Find? The researchers recruited older patients with Alzheimer disease from across England who had been on neuroleptics for at least three months. They randomised patients to one of two groups: the first group continued taking the same neuroleptic at the same dosage level while the second group was switched to an identical-looking placebo. The researchers assessed the patients cognitive status and neuropsychiatric symptoms upon their entry into the study. Six and 12 months later the researchers assessed any cognitive decline and the level of neuropsychiatric and other problems that patients were experiencing. At both 6 and 12 months, the researchers found that there were no differences between the two groups (continued treatment and placebo) in terms of cognitive decline. The placebo group may have had less cognitive decline, but this was not statistically significant. They also found no overall differences between the two groups in the change in the number of neuropsychiatric symptoms over these time periods. Patients with severe neuropsychiatric problems at the outset of the trial did better on continued neuroleptic therapy, but this advantage was not statistically significant. There was a significant decline on the verbal fluency language tests among the patients who continued on their neuroleptic. What Do these Findings Mean? The researchers report perhaps the first trial of this duration on continued versus withdrawn neuroleptic treatment among older dementia patients. The findings do not indicate any benefit of continuing neuroleptic therapies in older patients on either cognitive or neuropsychiatric outcomes. The researchers conclude that neuroleptics, with their known safety issues, should not be used as first-line treatment to manage problems such as agitation or aggression. For older dementia patients whose neuropsychiatric symptoms are not remedied by nonpharmaceutical treatments, the researchers advise caution. More studies are urgently needed to find better solutions to help older patients with dementia who have agitation, aggression, and psychosis. Additional Information Please access these Web sites via the online version of this summary at http://dx.doi.org/10.1371/journal.pmed. 0050076. Alzheimers Disease International is an umbrella organisation of organisations worldwide The Alzheimers Research Trust in the UK is a charity funding research to cure or prevent dementias The US National Institutes of Aging has information on Alzheimer Disease in English and Spanish Two governmental regulatory agenciesthe Medicines and Healthcare Products Regulatory Agency in the UK and the Food and Drug Administration in the USoffer information about antipsychotics in people with dementia
115
PLoS Medicine | www.plosmedicine.org 0599 April 2008 | Volume 5 | Issue 4 | e76
Other titles by iMedPub:

ocial Medicine in the 21st Century bySamuelBarrack. S orld Health Report 2012: No Health Without ResearchbySamuel W Barrack. uality design in Anatomical PathologybyAnilMalleshiBetigeri. Q scherichia coli infectionsbyVirojWiwanitkit. E Atlas of Biomarkers for Alzheimers disease byManuelMenendez.
Follow us:
At Medicalia.org cases Doctors exchange clinical experiences, review their can also access lots of and share clinical knowledge. You icalia.ning. medical publications for free. Join Now! http://med com/
116

Advances in Research and Treatment For Alzheimer's Disease

Încărcat de

Informații document

Titlu original

Drepturi de autor

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Advances in Research and Treatment For Alzheimer's Disease

Încărcat de

Drepturi de autor:

Title:

dvancesinResearchandTreatmentfor A Alzheimersdisease 978-1475095425 1475095422

Internet Medical Publishing info@imedpub.com http://imedpub.com/

Advances in Research and Treatment for Alzheimers disease

Advances in Research and Treatment for Alzheimers disease

November 2009 | Volume 6 | Issue 11 | e1000180

Epidemiological Pathology of Dementia

PLoS Medicine | www.plosmedicine.org

November 2009 | Volume 6 | Issue 11 | e1000180

Advances in Research and Treatment for Alzheimers disease

Methods Ethics Statement

PLoS Medicine | www.plosmedicine.org

November 2009 | Volume 6 | Issue 11 | e1000180

Epidemiological Pathology of Dementia

Died but Not Donors n = 11,465 Percent

Donors Versus All Died but Not Donors

2,601 2,625 5,244 2,524 2,514 2,740

1,465 1,387 4,065 1,549 1,423 1,526

8,231 6,982 2,849 186 70

3,557 5,159 2,573 176 79

100 215 135 6 81

566 11,985 5,697

483 7,812 3,170

1,371 6,574 10,303

1,035 3,717 6,713

1,187 4,601 4,782 895 84

IQR, interquartile range. doi:10.1371/journal.pmed.1000180.t001

Results Representativeness of the Brain Donor Cohort

Advances in Research and Treatment for Alzheimers disease

Neuropathology Neuritic plaques

Severity None Mild Moderate Severe Missing

Parietal 194 114 108 38 2 151 107 92 88 18 295 94 47 11 9 430 7 0 0 19 428 4 0 0 24

Occipital 200 74 80 23 79 143 94 100 37 82 299 38 26 9 84 384 3 0 0 69 380 2 0 0 74

None Mild Moderate Severe Missing

None Mild Moderate Severe Missing

None Mild Moderate Severe Missing

None Mild Moderate Severe Missing

Prevalence of Neuropathologies in Old People

Epidemiological Pathology of Dementia

Table 3. Number of individuals with neuropathology findings in subcortical nuclei.

Severity None Mild Moderate Severe Missing/not measured

Substantia Nigra 389 2 0 0 65 314 89 13 20 20 217 175 36 16 12 409 19 11 8 9

Raphe 271 8 0 0 177 157 78 52 36 133 298 23 2

Locus Ceruleus 283 1 0 0 172 166 105 42 14 129 216 84 27 11

Overall 353 36 13 0 54 177 122 77 71 9 183 175 67 25 6 406 20 18 9 3

None Mild Moderate Severe Missing

None Mild Moderate Severe Missing

133 327 2 0 0 127

118 313 15 9 1 118

None Mild Moderate Severe Missing

Neuropathology in the Nondemented

Prediction from Neuropathology Alone

AR of Dementia for Pathological Features

Advances in Research and Treatment for Alzheimers disease

Table 4. Number of individuals by neuropathology and dementia status at death.

Neuropathological Findings Neocortex: neuritic plaques

Severity None Mild Moderate Severe

No Dementia n = 183 83 51 43 6 54 43 56 28 12 114 59 8 0 101 51 24 1 6 106 37 33 6 1 101 47 26 5 4 34 68 53 27 1 117 33 17 2 14 100 49 28 5 1 81 42 44 11 5 27 55

Dementia n = 243 47 44 89 63 38 42 60 91 12 81 63 60 39 57 52 94 28 12 67 57 83 31 5 89 70 53 10 21 12 39 72 115 5 71 47 78 18 29 61 64 75 35 8 51 58 73 37 24 7 30

Neocortex: diffuse plaques

None Mild Moderate Severe Missing

None Mild Moderate Severe

None Mild Moderate Severe Missing

Hippocampus: neuritic plaques

None Mild Moderate Severe Missing

Hippocampus: diffuse plaques