of the type of interferon-beta used for patients with relapsing-remitting multiple sclerosis N Koch-Henriksen 1 , PS Sorensen 2 , K Bendtzen 3 and EM Flachs 4,* Objective To establish whether the clinical effect of neutralizing antibodies (NAbs) against interferon-beta (IFN) depends on the type of IFN (1a or 1b) used for treatment of patients with relapsing-remitting multiple sclerosis (MS). Introduction NAbs against IFN-1b appear faster and may be more evenly distributed on IgG sub- classes, whereas NAbs against IFN-1a develop more slowly and may be devoid of IgG3. This might cause different clinical responses to NAbs. Design/patients All Danish MS-patients who had started first-time treatment with IFN-1a 22 g s.c tiw (Rebif22) or IFN-1b 250 g s.c. qod (Betaferon) before January 1st 2003 were included. Relapses were recorded at bi-annual visit. Methods We measured NAbs every 12 months using a clinically validated cytopathic effect assay. A blood sample with a neutralizing capacity of 20% or more was considered as NAb-positive. We used a mixed logistic regression analysis in which NAb-status (three levels), IFN-preparation, and time since treatment started were included as explanatory variables, and relapse rate as response variable. Results In 1,309 patients, who were observed for 21,958 months, 32.3% were classified as NAb- positive. The odds-ratio (OR) for relapses in NAb-positive months compared with NAb-negative months was 1.25; P = 0.02. The risk of relapses was higher with Betaferon than with Rebif22 (OR 1.26; P < 0.01). The effect of NAb-level on relapses was independent of whether the patients were treated with Betaferon or Rebif22 (P = 0.89) and of time (P = 0.80). Conclusion NAbs caused by IFN-1a s.c. do not differ from NAbs caused by IFN-1b in their detri- mental clinical effect. Multiple Sclerosis 2009; 15: 601605. http://msj.sagepub.com Key words: interferon-beta; interferon-beta antibodies; multiple sclerosis; neutralizing antibodies; therapy; treatment effect 1 Department of Neurology, Aarhus University Hospital in Aalborg, Denmark and The Danish MS Treatment Register, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark 2 Danish Multiple Sclerosis Research Center, Department of Neurology, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark 3 Institute for Inflammation Research, Copenhagen University Hospital Rigshospitalet, Copenhagen, Denmark 4 National Institute of Public Health, University of Southern Denmark, Copenhagen, Denmark * On behalf of the Danish Multiple Sclerosis Study Group: Per Soelberg Sorensen, Rigshospitalet, Copenhagen; Finn Sellebjerg, Rigshospitalet, Copenhagen; Jette Lautrup Frederiksen, Glostrup University Hospital; Kai Jensen, Hillerod Hospital; Anne Heltberg, Roskilde Hospital; Ole Kristensen, Holbk Hospital; Sven Deth, Naestved Hospital; Claus Madsen, Odense University Hospital; Egon Stenager, Vejle, Esbjerg, and Sonderborg Hospital; Thor Petersen, Aarhus University Hospital; Bjarne Sivertsen, Viborg Hospital; Jesper Torring, Holstebro Hospital; Nils Koch-Henriksen; Aarhus University Hospital in Aalborg/Danish MS Registry. Correspondence to: Nils Koch-Henriksen, Aarhus University Hospital in Aalborg/Danish MS Registry. Email: koch-henriksen@stofanet.dk Received 6 August 2008; accepted 20 November 2008 RESEARCH PAPER Multiple Sclerosis 2009; 15: 601605 The Author(s), 2009. 10.1177/1352458508101946 Reprints and permissions: http://www.sagepub.co.uk/journalsPermissions.nav Introduction A substantial proportion of patients treated with interferon beta (IFN) for relapsing-remitting multi- ple sclerosis (MS) develops neutralizing antibodies (NAbs). Several studies have demonstrated that presence of NAbs hampers the clinical effect of IFN-1b [13] as well of IFN-1a [36]. A possible differential clinical importance of NAbs in patients treated with different types of IFN has recently been suggested in a study claim- ing, on indirect evidence, that NAbs against IFN- 1b are not associated with disease worsening [7]. According to these authors, difference, if existing, might be explained by different antibody responses to IFN-1b and IFN-1a in terms of NAb levels [811] and differences in IgG-subclass distribution of the NAbs [11]. These differences might cause dif- ferent clinical consequences of NAbs in patients treated with IFN-1a and IFN-1b. The aim of this study was, therefore, to compare the effect of NAbs on relapse rates in patients trea- ted with IFN-1a 22 g sc tiw (Rebif22) and IFN-1b 250 g s.c. qod (Betaferon). Patients and methods As a part of a national protocol for disease modify- ing treatment of MS, a national MS treatment regis- ter was established in 1996 [12]. In this register, we have recorded all Danish patients treated with IFN. The use of IFN treatment for MS in Denmark is limitedtodepartments of neurology, andall depart- ments of neurology send record forms with clinical information fromall obligatory visits every 6 months to the database. We included patients with relapsing-remitting MS who started first-time treatment with IFN-1a 22 g s.c tiw (Rebif22) or IFN-1b 250 g s.c. qod (Betaferon) before January 1st 2003. In Denmark, 22 g tiw was the preferred dose of Rebif before Jan- uary 1st 2004. The reasons for not including patients treated with IFN-1a 30 g once a week (Avonex) was that Avonex would be a possible con- founder because 1) we predominantly use Avonex for cases with low disease activity and 2) Avonex is less liable to cause NAbs [13]. This decision was made a priori. At the regular visits, relapses were assessed by his- tory and physical examination and is defined according to Schumacher [14] as the appearance of a new symptom or worsening of an old symptom attributable to MS, accompanied by an appropriate new neurological abnormality or focal neurological dysfunction lasting at least 24 h in the absence of fever and preceded by stability or improvement for at least 30 days. Blood samples for measurements of NAbs against IFN were collected by venipuncture at least 48 h after an injection of IFN, and the sera were isolated and stored at 20C until assayed. We used an antiviral neutralization bioassay for detection of NAbs, essentially as previously described [15]. Briefly, MC-5 cells, a subclone of A549 cells, were seeded in microtrays at a concen- tration of 15,000 cells per well and incubated for 24 h at 37C in a humidified 5% CO 2 atmosphere. IFN (using recombinant IFN homologous to the treatment received by the individual patient) at a concentration of 10 LU per ml was preincubated with diluted serum at 5% concentration in a vol- ume of 240 l. After 1 h, 200 l were transferred to the A549 cells. The cells were challenged with encephalo-myocarditis virus at the lowest concen- tration that by itself caused 100% cell death. After another 24 h, the antiviral effect of IFN was mea- sured by use of a 3-(4,5 dimethyl thiazol-2-yl)- 2,5-difenyl tetrazolium bromide assay. To avoid false-positive and false-negative results, controls for endogenous antiviral activity and serum toxicity were included in each assay. The neutralizing capa- city (NC) of each serum sample was measured; i.e. the percentage of the added IFN that had been neutralized by the NAbs. The investigators in the laboratory had no knowledge of the patients disease characteristics. We defined a NAb-negative sample as a sample with NC < 20%, a low-NAb- positive blood sample as a sample with NC 20 79%, and a medium/high NAb-positive blood sample as a sample with NC of 80% or above. The cut-off value for being Nab positive was chosen as the lowest concentration of NAbs that had signifi- cant clinical importance as shown in a previous study [3]. The MS Treatment Register was notified of the results of all NAb-tests. As clinically relevant NAbs usually do not appear until 912 months after initiation of IFN therapy [13], we used the NAb tests from the 12-, 24-, 36-, 48-, and 60-month visits in this study. The ethical committees were not specifically involved in this study because it was a pure retro- spective observational study made on patients who were treated with approved drugs on the basis of national recommendations. Measuring NAB on reg- ular basis on all IFN-treated patients is obligatory in Denmark, a decision that has been approved by the National Ethics Committee. Statistical methods We used the interval analysis or all switches considered method for defining NAb-positivity [16,17]. For a period of a half-year before and a half-year after a NAb-test was taken, each month 602 N Koch-Henriksen et al. Multiple Sclerosis 2009; 15: 601605 http://msj.sagepub.com was classified either as NAb-negative or NAb- positive according to the result of the NAb-test. In the simple calculations, each individual month of treatment was regarded as a unit, characterized by relapse/no relapse, preparation, and NAb-status (+/). The analyses were performed with 20% NC as cut-off value. We also employed a mixed logistic regression analysis in which the unit was the individual itself (rather than the individual month) with relapse rate as the response variable and NAb status (+/), prepa- ration (Rebif22/Betaferon), and time (period 15, representing the months 718, 1930, 3142, 4354, and 5566, counted from treatment start) as the explanatory variables. Results In all, 1,309 patients with clinically definite MS, 398 men and 911 women, fulfilled the criteria of having started first-time IFN treatment with Rebif22 (n = 892) or Betaferon (n = 417) before Jan- uary 1st 2003. Clinical and demographic character- istics by treatment are shown in Table 1. The patients were observed for a total of 21,963 months of which 7,098 (32.3%) were classified as NAb- positive according to the above given definition. With Rebif22, 3,927 out of 12,494 months of obser- vation (31.4%) were NAb-positive, and with Beta- feron 3,172 out of 9,468 months of observation (33.0%) were NAb-positive; P = 0.001, (tied observa- tions). Among the Nab-positive blood tests, 77% were medium/high positive (NC 80%) with Rebif22 and 67% with Betaferon (P = 0.004, see Figure 1). The cumulative risk of being NAb positive was at 12 months: 0.19 for Rebif22 and 0.35 for Betaferon, at 24 months: 0.35 for Rebif22 and 0.48 for Betaferon, and at 36 months: 0.55 for Rebif22 and 0.56 for Betaferon. The risk of relapses was higher with Betaferon than with Rebif22 (odds- ratio [OR] 1.31, P = 0.0002), which was true for Nab-positive months (OR 1.43; 95% CI 1.141.78; P = 0.002) as well as for NAb-negative months (OR 1.22; 95% CI 1.031.46; P = 0.028). The annualized relapse rate in the NAb-positive months was 0.54 compared with 0.41 in the NAb- negative months (OR 1.32, P < 0.0001). In a mixed logistic regression analysis, we included the three main effects: period, preparation, and NAb, and all possible second-order interaction terms between the three explanatory variables. In this setting, we partitioned NC of NAbs into three levels: (negative: <20%; low positive: 2079%; medium/high positive: 80%). With this, the OR for relapses in all NAb-positive months (low positive as well as medium/high posi- tive) compared with NAb-negative months was 1.25; P = 0.02. The OR for relapses in medium/high NC Nab-positive months compared with Nab-negative months was 1.28 (95% CI: 1.051.55). The OR of having relapses on treatment with Betaferon com- pared with Rebif22 was 1.26; (95% CI 1.061.51; P < 0.01). Period as single factor did influence relapse rate statistically significantly with an OR of 0.85 per step (95% CI 0.790.90), and this effect was linear. We aimed at determining whether preparation influences the effect of NAbs on relapse frequency. The null-hypothesis was tested by estimating the significance of the interaction term NAb*prepara- tion in the mixed logistic regression equation. The effect of NAb-status on relapses did not differ between the two preparations (P = 0.89), meaning, that NAbs have the same detrimental effect with Betaferon as with Rebif22. Moreover, there was no 0 200 400 600 800 1000 Rebif 22 Betaferon N u m b e r
o f
N A b - t e s t s <20% 20-79% 80+% 67.9% 7.3% 24.8% 66.3% 11.2% 22.5% Figure 1 Absolute and percentage distribution of neutra- lizing capacities (NC) on negative (NC < 20%), medium positive (NC 2079%) and highly positive (NC > 80%) NAb-tests with Rebif22 and Betaferon. Table 1 Clinical and demographic characteristics at baseline IFN-1a 22 g s.c. qod n = 417 IFN-1b 250 g s.c. tiw n = 892 Female:male ratio 2.37 2.14 Age at treatment start Mean 37.9 38.2 Minmax 1368 1664 Duration of MS, years Mean 5.5 7.1 Minmax 044 051 Baseline EDSS Mean 2.65 2.86 Minmax 06.5 06.0 #Relapses 24 months prior to treatment Mean 2.56 2.89 Minmax 18 08 IFN, interferon beta; MS, multiple sclerosis; EDSS, expanded disability status scale. IFN preparation and clinical effect of neutralizing antibodies 603 http://msj.sagepub.com Multiple Sclerosis 2009; 15: 601605 interaction between period and the effect of NAbs (P = 0.80). Discussion We found no indications of less harmful consequ- ences of NAbs on clinical relapses in NAb-positive patients treated with Betaferon as hypothesized by others [7]. Differences in NAb-levels are likely to impact the clinical consequences of NAbs. Although moderate and high concentrations of NAbs reduce or abolish IFN bioactivity, patients with low NAb concentrations may have retained a full or partial in-vivo response to IFN. According to one study, NAb-positive patients treated with IFN- 1b had higher NAb titers compared to NAb-positive patients on IFN-1a s.c. [11], but two other studies have shown higher titers with IFN-1a s.c. [10,18], which also is in accordance with the present study. This, if true, would hypothetically favor NAb-positive patients treated with IFN-1b. The dif- ference between the above-mentioned study [11] and the present study could be ascribed to differ- ences in NAb-assay or cut off for NAb positivity. We did not use a quantitative immunochemical method for IFN antibody measurements as these would detect binding anti-IFN antibodies, which might not neutralize the clinical effects of IFN. Rather, we used a semi-quantitative functional assay for ex vivo IFN neutralization to grade blood samples as NAb-negative, low-level NAb-positive, or medium/high-level NAb-positive with a ceiling effect, preventing us from distinguishing very high levels of NAb from medium/high levels. Further titration of Nab concentrations is of limited interest for the risk of having relapses, as all patients with medium/high Nab levels have lost their in vivo bio- availability of IFN and thereby lost the therapeutic benefit of IFN [17]. Another difference between NAbs raised against IFN-1b and NAbs raised against IFN-1a is the per- sistence of NAbs. Nab-positive patients treated with IFN-1b have a higher probability of reverting to the NAb-negative state within a treatment period of 35 years [13], and this would favor IFN-1b if the once positive, always positive method is used for evaluation of the clinical significance of NAbs [16]. However, we used the interval analysis or all switches considered method for defining NAb- positivity in order to eliminate the error caused by patients who regain the therapeutic effect of IFN, as they revert to the NAb-negative state [19]. We found that, even when accounting for NAb- status, the OR for relapses with Betaferon was 1.26 compared with Rebif22 (P < 0.001). This difference is probably due to the circumstance that Betaferon was approved 2 years before Rebif22, and during this interval, we gradually began to treat patients with slightly less active disease. The IgG-subclass distribution in NAb-positive patients seems to vary with IFN preparation and time [11]: In Betaferon-treated NAb-positive patients, all four subclasses of IgG may be induced with IgG1 and IgG3 peaking the first year and IgG4 later on. In Rebif22-treated NAb-positive patients, the IgG- subclass profile is dominated by IgG4 emerging later, and it has been suggested that a distinctive pattern of IgG-subclasses could be of functional sig- nificance [11]. We did not perform IgG-subclass measurements of our blood samples, but our results do not support the idea that any difference in IgG- subclass distribution caused by the IFN-preparation should influence the clinical effect of NAbs, as an interaction term between preparation and the effect of NAbs was far from significant and did not even show a weak trend. Neither did the proposed tempo- ral differentiation in distribution of IgG-subclasses result in a time effect on the clinical consequences of being NAb-positive, as the interaction between time (period) and clinical NAb-effect was nonexist- ing (P = 0.90). In conclusion, we found no indication of less detrimental effects of NAbs on clinical relapse rate in NAb-positive patients treated with IFN-1b compared with NAb-positive patients treated with IFN-1a s.c. Acknowledgements Economic support to the Danish Multiple Sclerosis Research Center was obtained from the Danish Multiple Sclerosis Society, the Warwara Larsen Foundation, the Danish Medical Research Council, the European Union Sixth Framework Programme: Life sciences, Genomics and Biotechnology for health: Contract No. 18926: Neutralizing antibodies in multiple sclerosis (NABINMS). The MS Treatment Register is funded by the Copenhagen Hospital Council and by the Association of Danish Regions. References 1. Interferon beta-1b is effective in relapsing-remitting mul- tiple sclerosis. I. Clinical results of a multicenter, ran- domized, double-blind, placebo-controlled trial. The IFNB Multiple Sclerosis Study Group [see comments]. Neurology 1993; 43: 655661. 2. Neutralizing antibodies during treatment of multiple sclerosis with interferon beta-1b: experience during the first three years. The IFNB Multiple Sclerosis Study Group and the University of British Columbia MS/MRI Analysis Group [see comments]. Neurology 1996; 47: 889894. 3. Sorensen, PS, Ross, C, Clemmesen, KM, et al. Clinical importance of neutralising antibodies against interferon 604 N Koch-Henriksen et al. Multiple Sclerosis 2009; 15: 601605 http://msj.sagepub.com beta in patients with relapsing-remitting multiple sclero- sis. Lancet 2003; 362: 11841191. 4. PRISMS-4. Long-term efficacy of interferon-beta-1a in relapsing MS. Neurology 2001; 56: 16281636. 5. Francis, GS, Rice, GP, Alsop, JC. Interferon beta-1a in MS: results following development of neutralizing antibodies in PRISMS. Neurology 2005; 65: 4855. 6. Kappos, L, Clanet, M, Sandberg-Wollheim, M, et al. Neu- tralizing antibodies and efficacy of interferon beta-1a: a 4-year controlled study. Neurology 2005; 65: 4047. 7. Goodin, DS, Hurwitz, B, Noronha, A. Neutralizing anti- bodies to interferon beta-1b are not associated with disease worsening in multiple sclerosis. J Int Med Res 2007; 35: 173187. 8. Bertolotto, A, Malucchi, S, Milano, E, Castello, A, Capobianco, M, Mutani, R. Interferon beta neutralizing antibodies in multiple sclerosis: neutralizing activity and cross-reactivity with three different preparations. Immu- nopharmacology 2000; 48: 95100. 9. Sominanda, A, Hillert, J, Fogdell-Hahn, A. In vivo bioac- tivity of interferon-beta in multiple sclerosis patients with neutralising antibodies is titre-dependent. J Neurol Neurosurg Psychiatry 2008; 79: 5762. 10. Sominanda, A, Rot, U, Suoniemi, M, Deisenhammer, F, Hillert, J, Fogdell-Hahn, A. Interferon beta preparations for the treatment of multiple sclerosis patients differ in neutralizing antibody seroprevalence and immunogeni- city. Mult Scler 2007; 13: 208214. 11. Gibbs, E, Oger, J. The IgG subclass-specificities of anti- IFNbeta antibodies change with time and differ between the IFNbeta products in relapsing remitting multiple sclerosis patients. J Neuroimmunol 2007; 190: 146150. 12. Koch-Henriksen, N, Sorensen, PS; The Danish Multiple Sclerosis Group. The Danish National Project of interferon-beta treatment in relapsing-remitting multiple sclerosis. Mult Scler 2000; 6: 172175. 13. Sorensen, PS, Koch-Henriksen, N, Ross, C, Clemmesen, KM, Bendtzen, K. Appearance and disappearance of neutralizing antibodies during interferon-beta therapy. Neurology 2005; 65: 3339. 14. Schumacher, GA, Beebe, G, Kibler, RF, et al. Problems of experimental trials of therapy in multiple sclerosis: report by the panel on the evaluation of experimental trials of therapy in multiple sclerosis. Ann N Y Acad Sci 1965; 122: 552568. 15. Bendtzen, K. Anti-IFN BAb and NAb antibodies: a mini- review. Neurology 2003; 61: S6S10. 16. Hesse, D, Sorensen, PS. Using measurements of neutraliz- ing antibodies: the challenge of IFN-beta therapy. Eur J Neurol 2007; 14: 850859. 17. Sorensen, PS, Tscherning, T, Mathiesen, HK, et al. Neu- tralizing antibodies hamper IFNbeta bioactivity and treatment effect on MRI in patients with MS. Neurology 2006; 67: 16811683. 18. Gneiss, C, Tripp, P, Reichartseder, F, et al. Differing immunogenic potentials of interferon beta preparations in multiple sclerosis patients. Mult Scler 2006; 12: 731737. 19. Sorensen, PS, Koch-Henriksen, N, Bendtzen, K. Is treat- ment effect of IFN-beta restored after disappearance of neutralizing antibodies? Mult Scler 2008; 14: 837842. IFN preparation and clinical effect of neutralizing antibodies 605 http://msj.sagepub.com Multiple Sclerosis 2009; 15: 601605