Documente Academic
Documente Profesional
Documente Cultură
www.elsevier.com/locate/medengphy
Received 8 October 2003; received in revised form 27 February 2004; accepted 19 March 2004
Abstract
Background: Refractometry is the determination of the optical refractive index of a substance or a mixture of substances. It is a
very sensitive method for the detection and quantication of dissolved analytes, but it is incapable of distinguishing between dif-
ferent analytes. The aim of this investigation was to determine the principle suitability of refractometry for the quantication of
glucose (blood sugar) in blood and various blood uids which can readily be obtained for medical diagnosis, in particular blood
plasma, blood serum, and their ultraltrates.
Methods and results: After the oral intake of freshly dissolved a-glucose, the in vivo blood contents of the a and b anomers of
glucose were found to be in an at least approximate equilibrium at all times. This observation is a prerequisite for a refractome-
trical determination of glucose due to the fact that both molecule forms have dierent refractive index increments. An assessment
of the glucose content in untreated blood uids was not possible, since no suitable relationship to the refractive index was found,
most probably due to the inuence of the many other substances present in blood on this parameter. However, after removal of
certain macromolecules by ultraltration, value pairs showed a high level of correlation, providing the nominal molecular weight
limit (cut-o) of the ultralter used possessed a maximum of 300 kDa. Besides macromolecules, the osmolality of the uids
undergoing measurement also proved to be a considerable interfering factor, particularly when values were outside the normal
physiological range between 285 and 293 mmol/L.
Conclusions: If a clinical application of this method is to be contemplated it is imperative (1) that blood cells are separated and
removed, (2) that macromolecules present in plasma or serum are removed, e.g. by ultraltration, and (3) that beyond the results
presented the inuence of all small molecules other than glucose on the overall refractive index be determined and included in the
calculation of analysis results.
# 2004 IPEM. Published by Elsevier Ltd. All rights reserved.
Keywords: Blood glucose; Glucose analysis; Refractometry; Blood uids; In vivo specicity
preparations of detergents and antifreeze, essences and examination of the possibility of the determination of
aromatic substances [2,3,5]. the overall content of dissolved glucose in blood using
Providing a sucient specicity for D-glucose (or refractometry. A refractometrical determination relies
dextrose, or, within the body, blood sugar, hereafter on an equilibrium of the glucose anomers since the
more briey referred to as glucose) could be obtained, total refractive index is determined, which includes the
refractometry as a purely physical measurement contributions of both glucose anomers with their dier-
method could be clinically applicable both, for the in ent respective refractive indices, 147:6 106 L=g for
vitro determination of blood glucose levels in blood a-glucose, and 149:0 106 L=g for b-glucose [11].
samples, and within the frame-work of an insertable or Moreover, the question of whether a sucient equi-
implantable technical in vivo probe for glucose. The librium of the anomers exists at all times in the frame-
subcutaneous connective and fatty tissue appears to be work of this work was linked to the necessity to nd a
a suitable candidate for probe insertion or implan-
simple, but nevertheless reliable reference method for
tation, whereby its interstitial uid (which corresponds
the determination of glucose contents. There are a num-
to the molecular composition of blood uid [6]) is the
ber of analytical methods available for the analysis of
substance undergoing investigation [7,8].
glucose based on an enzymatically catalysed chemical
The aim of this study was, in principle, to determine
reaction. Very often, those analytical methods speci-
the suitability of refractometry for quantication of the
cally use only one anomer of the dissolved glucose as a
glucose content in several blood uids which are easily
obtainable for medical diagnosis (full blood (especially substrate. With technical and clinical analytical meth-
venous and capillary), as well as its derivates blood ods, the enzymes glucose oxidase (b-D-glucose: oxygen
plasma, blood serum, and ultraltrates of plasma and 1-oxidoreductase) and glucose dehydrogenase (b-D-
serum). Of special interest was the possible extent of glucose: NAD(P)+ 1-oxidoreductase), which are highly
the (real or in vivo) specicity for glucose in these specic for b-glucose, are frequently used [12,13].
uids. This paper contains a description of basic Glucose oxidase catalyses the reaction
experiments on the suitability of the refractive indices b-D-glucose O2 H2 O D-gluconic acid H2 O2 ;
of body uids for the analysis of glucose contained in
such uids, as well as a consideration of the disturb- and glucose dehydrogenase catalyses the reaction
ance of the measurement due to other components in
b-D-glucose NAD D-gluconolactone NADH
the investigated samples.
H :
2. Problems and solving attempts In several commercially available analytical glucose tests
the only real analyte is b-glucose, because the analysis
A prerequisite for a refractometrical determination reaction time (often much less than one minute) is far
of glucose in body uids is the technical ability to too short to encompass the analysis of a succeeding com-
exactly account for changes in the measuring quan- plete mutarotation of the a-glucose present in the
titythe refractive index ncaused by changes in the material to be analysed. So, only if the anomers of glu-
glucose concentration. Currently available precision cose are in a steady state equilibrium during calibration
refractometers are able to determine the expected and during measurement, the measuring results obtained
refractive index changes (about 14 106 per 0.1 g/L with those enzymatic methods are correct.
(10 mg/dL) change in glucose concentration) su- The prerequisite of a glucose anomers equilibrium for
ciently [9].
analysis is most certainly fullled in the body uids
In aqueous solutions glucose predominantly exists in
during phases where a stable glucose level and a constant
two isomeric (anomeric) forms: a- and b-glucose (both
body temperature prevail. However, at times when
are pyranoses: a-glucopyranose and b-glucopyranose).
glucose levels change, such an equilibrium may not be
Each anomer can be transformed into the other through
present, and could thus present a measurement problem.
molecular conversionthe so-called mutarotation [2]:
No information could be found in the available
a-D-glucose D-glucose aldose b-D-glucose literature to clarify the existence of an equilibrium
A variety of catalytic mutarotase activities accelerate between a- and b-glucose in the human body at all
the attainment of a temperature-dependent, steady state times. Imbalances may occur, particularly during pha-
equilibrium between these anomeric forms, e.g. at body ses of postprandial absorption (uptake of substances in
temperature the equilibrium is between 36.4% the gastrointestinal tract following a meal) or during
a-glucose and 63.6% b-glucose, with only traces of the hormonally or neurally stimulated mobilisation of
intermediate ring-opened aldehyde form [10]. glucose from glycogen, the storage form of glucose in
The physiological co-existence of two dierent the body, consisting of polymers of covalently com-
molecular forms of glucose in the body prompts an bined a-glucose [10].
K. Zirk, H. Poetzschke / Medical Engineering & Physics 26 (2004) 473481 475
v
An attempt was therefore made to shift the equilib- of the plasma for 1 h at 37 C (in order to obtain a
rium between the two anomers through oral ingestion complete coagulation) and removal of the brin clots
of a substantial amount of a-glucose, and to prove this by centrifugation (15; 000 9:81 m=s2 for 10 min).
shifting with the help of a detection method specic for Ultraltrates of blood plasma and serum were pro-
one anomer. duced with ultraltration stirring cells of dierent
Refractometry is not substance-specic, and many Nominal Molecular Weight Limits (NMWL, referring
substances present in the blood contribute to the to the separation capabilities, the values describe the
bloods refractive index. The contents of these sub- molecular masses (in Dalton or g/mol) of globular pro-
stances therefore need to be considered as disturbing teins which are withheld in a standardised ultraltra-
quantities if the blood glucose level is to be determined tion process to approximately 90%). Technical grade
using refractometry. In order to identify interfering nitrogen served as the propellant (alternatively, oxygen
substances or groups of such substances a number of gave the same results). The stated in vivo values for the
approaches are possible. glucose contents and the refractive indices are stable
One is to remove supposed disturbing substances by plateau values obtained in the course of an ultraltra-
means of a substance-separating process. Under the tion process.
assumption that these components (present in varying
concentrations) are of high molecular size (lipopro-
3.2. Analytic methods and measuring devices
teins, proteins, etc.), a suitable method for their
removal is the ltration of the material to be analysed Two identical commercially available test strip analy-
through ultralters [14,15]. sers served for the determination of glucose content in
A further approach to the identication of disturbing body uids (Accutrend Sensor, Boehringer Mannheim,
substances is a global evaluation of the inuence of all Mannheim, D). They are based on the oxidation of
dissolved components through an in vivo dilution or b-D-glucose (enzymatically catalysed by means of
concentration of the body uids. When a large amount glucose dehydrogenase), and the reaction rate is quan-
of water with a low salt content is ingested, an acute tied amperometrically.
dilution of the extracellular uid can be achieved. Cor- A calibration of the measurement results from the
respondingly, a restriction of water intake should lead two devices was carried out by the construction of cali-
to a concentration of these uids. During both phases bration curves for all used media. Gravimetrically mea-
the blood content of glucose is normally held constant sured amounts of the analyte glucose (as a-glucose)
by hormone-mediated endogenous regulation. served as references and were added to the matrix of
full blood, plasma, serum, and their ultraltrates, as
well as to water, and mixed suciently (in accordance
3. Experimental with our results concerning the kinetics of all media in
terms of reaching a steady state equilibrium of mutaro-
3.1. Materials and preparative methods
tation between a- and b-glucose). The calibration
Crystalline a-D-(+)-glucose (dextrose) was obtained curves were generally linear, and readings showed a
from Roth (Karlsruhe, D; product numbers: M 198, relative imprecision of (usually) less than 5% (regarding
17 or 6780 (a-D-(+)-glucose monohydrate), purity: ve single measurements at each glucose content), cor-
99.5%, Ph.Eur., DAB). responding well with the values of Wahl [16].
Blood samples were obtained from healthy (non- Two dierent commercially available devices were
diabetic) women and men of between 23 and 42 years used for the determination of optical refractive indices.
of age, who participated voluntarily. Prior to each The rst was a simple manually held refractometer
measurement, venous blood samples were obtained by (Eclipse from Bellingham & Stanley, Turnbridge
aspiration puncture of a slightly congested cubital vein, Wells, UK; reading accuracy: 5 104 ) of which the
and capillary blood by piercing a nger tip with a lan- display was calibrated to the value of distilled water
v
cet following hyperaemerisation. 20 IU (international (n 1:333) at room temperature (22 C). The second
units) of heparin sodium salt (4 lL Heparin-Natrium- was a precision refractometer DUR by Schmidt &
25000-Ratiopharm from Ratiopharm, Ulm, D) were Haensch (Berlin, D), with which all measurements were
v
added to each millilitre of venous blood to prevent carried out at a temperature of 30.0 C (set with a cir-
coagulation. culating thermostat), with the display being accordingly
Blood plasma was obtained from this heparinised adjusted to the value of the refractive index of distilled
full blood by centrifugation (15; 000 9:81 m=s2 for 10 v
water at 30.0 C (1.33192). The imprecision of this
min), and blood serum by separating the plasma from refractometer is 105.
the cellular components of native full blood using cen- Osmolalities (the sum of the particular molal con-
trifugation (10; 000 9:81 m=s2 for 2 min), incubation tents of all dissolved components) were determined
476 K. Zirk, H. Poetzschke / Medical Engineering & Physics 26 (2004) 473481
with an osmometer (Halbmikro-Osmometer Typ M were determined (using the manually held refrac-
from Knauer, Berlin, D), the measuring principle of tometer Eclipse).
which is the determination of the decrease in the freez- To explore possible improvements in the correlation
ing point. The device was calibrated with a NaCl cali- of measured refractive indices and glucose contents, the
bration solution (12.687 g NaCl/kg H2O, Knauer). refractive indices of the ultraltrates of plasma samples
According to the manufacturers operating instructions were obtained, using a number of ultralters with
the inaccuracy of the measurements is
2%. decreasing pore diameter, and a comparison was made
Multiple measurements of each sample were per- between ultraltrated and non-ltrated plasmas from
formed with both the precision refractometer and the dierent volunteers: Heparinised plasma samples,
osmometer. Measurement values are presented as arith- obtained from the blood of 6 of the above-mentioned,
metic means, and the standard deviation (SD) serves as non-fasting volunteers (three female, three male) were
a measure of scatter. ltered through ultraltration membranes of dierent
NMWLs (10, 30, 100, 300, and 1000 kDa). The refrac-
3.3. Protocols tive indices (determined using the manually held refrac-
tometer Eclipse) and the glucose concentrations were
The kinetics of anomeric conversion of freshly
determined in both the plasma samples and the corre-
dissolved a-glucose until a substance equilibrium with
b-glucose is reached were determined in preliminary sponding ultraltrates.
experiments. Venous blood of a fasting non-diabetic For the analogue but more precise determination of
male (H.P.) was used for this purpose, from which the inter-individual dierences of the dependence of the
heparinised full blood, plasma and serum were refractive index on the glucose concentration in ultra-
obtained. After addition of a-glucose to these derivates ltrates of blood serum, blood sera (from blood
in vitro, as well as to water, an incubation of the uids obtained from four non-fasting, non-diabetic volun-
v v
(with stirring) at 22 or 37 C ( 1 C in both cases) fol- teers; three male, one female) were ltered through
lowed, and glucose content was determined over time. ultraltration membranes with a NMWL of 3 kDa.
The existence of an equilibrium of the mutarotation The ultraltrated serum of one male volunteer was div-
of a- and b-glucose in the human body (in vivo) was ided into three parts and dierent amounts of a-glucose
examined, especially during a phase where a large were added to two of them. These two aliquots were
v
inux of a-glucose into the blood was occurring (i.e. an then incubated for 15 minutes at 37 C in order to
increase in the blood level). This inow was attained achieve a mutarotation equilibrium between a- and
by oral ingestion of a-glucose. Within 2 min, the same b-glucose. Two of the male volunteers (H.P. and K.Z.)
volunteer drank a freshly prepared solution of 75 g additionally drank a freshly prepared solution of 75 g
a-glucose in 0.2 L water (this is the glucose quantity a-glucose in 0.2 L water within 2 min. At the approxi-
used for a diagnostic oral glucose tolerance (OGT) test mate time of the maximum value for the venous blood
in Germany). glucose content (see Fig. 2), a further blood sample
At once, and after 10, 15, 25, 30, 37 and 40 min, was taken, and the refractive indices (using the precise
capillary blood was used to obtain an in vivo blood refractometer DUR) and glucose concentrations of the
sugar prole, in which the glucose contents (cmGluAS) ultraltrates of these sera were determined.
were determined with the help of the blood sugar mea- The inuence of the osmolality of the serum solutes
suring device Accutrend Sensor (AS), which exclusively
and its possible in vivo extent on the refractive index of
detects b-glucose. In order to detect a possible imbal-
a serum ultraltrate was examined, as well as the
ance in the mutarotation of the glucose anomers in cir-
(refractometrically) indirectly determinable (apparent)
culating blood, at 3, 7 and 16 min after intake of the
glucose concentration. A non-fasting, non-diabetic
glucose solution venous blood samples were obtained,
v
heparinised and incubated at 37 C for up to 20 min, male (H.P.) drank 2.5 L water within 15 min (in order
and were used to determine time-related changes (in to decrease serum osmolality), and venous blood sam-
vitro) in the measurable glucose content (cmGluAS). In ples were taken 12 and 60 min thereafter in order to
the chosen time period the inuence of the glucose con- obtain serum. The refractive indices (precise refrac-
sumption by the erythrocytes is smaller than the scat- tometer DUR), the glucose concentrations, as well as
tering interval of the values of the analysers, and can the osmolalities were determined from 3 kDa ultra-
therefore be ignored [17]. ltrates of the respective serum.
For an assessment of the inter-individual dependence Afterwards, the volunteer underwent a phase of
of the refractive index of blood plasma on the mass approximately 26 h uid starvation (in order to
concentration of glucose, glucose contents and refrac- increase the osmolality) followed by analogue determi-
tive indices of heparinised plasma samples from 11 nation of the mentioned quantities in the serum ultra-
non-diabetics (ve female and six male, all non-fasting) ltrate.
K. Zirk, H. Poetzschke / Medical Engineering & Physics 26 (2004) 473481 477
4. Results
7, and 16 min) showed constant levels for glucose con- For all further tests only blood serum was used. The
tent. Fig. 2a depicts the time course of the (measur- use of blood serum instead of blood plasma eliminates
able) glucose content in the volunteers blood (in vivo) the possible inuence of variable amounts of the non-
and Fig. 2b presents the course of the measured values endogenous macromolecular decoagulant heparin sul-
(in vitro) for the three blood samples obtained. phate as well as of brinogen. Since the size of the lter
For orientation, Fig. 3 shows a graph in which the pores has only a slight inuence on the processing time
refractive index of native blood plasma (the refractive (ltration time), an ultralter with the smallest pore
index of full blood is not determinable with the refrac- size can be used, if an ultraltration is required at all.
tometer used) is plotted against the glucose contents of Ultraltrates of the serum were therefore obtained only
these plasma samples. A useful, recognisable corre- using ultralters with a nominal molecular weight limit
lation is not evident, with the measurement value of 3 kDa. In addition, ultralters with a small pore
pairsrepresented by dots in the diagramappearing size are able to hold back further disturbing serum
as a cloud. components. Because of the expected very small chan-
Fig. 4 shows a comparable diagram of the refractive ges in the determining quantity, the refractive index n,
indices obtained in blood plasma ultraltrates plotted a precise refractometer was applied.
against glucose contents. It is evident that ultraltrates The refractive index of aqueous solutions of glucose
gained using lters with a nominal molecular weight in the examined glucose concentration range are well
cut-o lower than 1000 kDa (300 kDa and smaller) known, exhibiting an approximately linear relation [5].
possess a much smaller refractive index than non-ultra- Fig. 5 shows the refractive index of ultraltrates of
ltrated plasmas and ultraltrates gained using 1000 blood serum obtained from four volunteers together
kDa ultralters. with their respective glucose contents. In some cases,
these contents were increased by the ingestion of
glucose solutions (in vivo), or by mixing crystalline
glucose into these serum samples (in vitro).
The values correlate very well, with a correlation
coecient R of 0.973. The greatest single dierence
between the measured glucose content and the value of
the regression line is 0.1 g/L, which corresponds to a
relative error of about 9% in this particular case.
The dependence of the refractive index of the ultra-
ltrates (obtained using lters with a nominal molecu-
lar weight limit of 3 kDa) of human blood serum on
their glucose contents, as shown in Fig. 5, can serve as
Fig. 3. The refractive index n of the blood plasma of 11 non- a calibration curve to determine glucose content by
diabetics (ve females and six males) as a function of the glucose refractive index measurement.
concentration cmGlu. Each point represents a mean of ve To test the extent of the inuence of glucose content
measurements. on the accuracy of such determinations, global changes
blood cells and 58% blood plasma. Plasma in turn con- tration of the dissolved components has a clear inu-
tains 90% water, approximately 7% proteins, 0.40.7% ence on the true accuracy of the refractometrical
lipids, 0.070.1% glucose, 0.030.07% amino acids, determination of the glucose content. The relative
0.020.05% carbamide, 0.0080.02% lactic acid and deviations already amount to more than 50%, both in
approximately 0.5% inorganic salts [6,15]. the diluted (at a osmolality of 273 mmol/kg) and the
Besides water, protein dominates in the liquid phase slightly more concentrated (at a osmolality of 296
of blood and thus a refractometrical protein determi- mmol/kg) extracellular uid.
nation is possible, as was already shown at the begin- In summary, refractometry is a suitable method for
ning of the 20th century [9,18]. Additionally, as shown the determination of the glucose content of blood
here the glucose content of untreated blood plasma is (using blood plasma or serum ultraltrates as the
not determinable refractometrically. The individual measurement object) of non-diabetics within a small
variation of the protein level and its contribution to the physiological hydration range. However, when this
total refractive index exceeds the contribution of glu- hydration range limit is exceeded, there are distinct
cose many times. deviations of determined glucose concentrations from
After removing high molecular substances by means the true values, so that a refractometrical method alone
of ultraltration, the inuence of these substances on appears unsuitable for a clinical application.
the refractive index was reduced clearly. The necessary Even so, the latter problem can presumably be
separation limit (NMWL) of the ultralters of approxi- solved since the hydration state of a person can be
mately 300 kDa found seems to be indicative of plasma determined inexpensively by a technically simple
proteins rather than large lipoprotein complexes (chy- measurement of the electrical conductivity of extra-
lomicrons, chylomicron-remains, and the other lipopro- cellular body uids, including blood-derived uids. If a
teins VLDL, IDL, LDL and HDL) [13], since the latter universal interindividual dependency (with sucient
can be eliminated through ultraltraters with separ- accuracy) exists between the body uids osmolalities
ation limits far greater than 300 kDa. (at constant glucose levels) and these uids optical
Besides high molecular substances, low molecular refractive index values, then the glucose analysis cali-
solutes can also have an inuence on the total refractive bration functions for various hydration states can be
index and are thus able to reduce the specicity of a determined. By means of mathematical interpolations it
refractometrical glucose determination. The smallest should then be possible to calculate a calibration func-
ultralter used had a nominal molecular weight limit of tion valid for any actual hydration state. Further
3 kDa. The remaining matrix components were there- research on this possibility of eliminating the disturbing
fore salts (in particular sodium and chloride ions, etc.), and biasing eects of the salts and other solutes con-
amino acids and other low molecular organic substances centrations is planned.
(carbamide, lactate, etc.) [6,15], all of which generally
show very similar specic refractive index increments
(Dn/Dc) of approximately 140 200 106 L=g (Ref. [4],
Acknowledgements
or determined by ourselves). However, the absolute con-
centration (or content) of a disturbing substances is not We express our sincere gratitude to Dr. Debra
the decisive factor, but rather its uctuations and there- Kelleher for her careful and critical assistance in pre-
fore its variable contribution to the total refractive paring this manuscript.
index.
By dilution or concentration of the body uids,
many possible interfering quantities may change simul-
References
taneouslywhereas the content of the analyte glucose
should be held constant by endogenous mechanisms. [1] Born M, Wolf E. Principles of opticselectromagnetic theory of
Such changes in the bodys hydration state are propagation, interference and diraction of light. 7th expanded
assumed to occur frequently, especially dehydration ed. Cambridge: University Press; 1999.
caused by physical training, in warmer climates, and in [2] Falbe J, Regitz M, editors. ROMPPLexikon Chemie, 10.
vollig uberarb. Auage. Stuttgart: Thieme; 1999.
elderly persons. We found that an acute experimental
[3] Paul H, editor. Lexikon der Optik. Heidelberg: Spektrum Aka-
dilution of the extracellular uids was easily possible demischer Verlag; 1999.
through ingestion of a large amount of water (with a [4] Weast RC, Astle MJ, Beyer WH. Handbook of chemistry and
low salt content), while restriction of water input physics. 67th ed. Boca Raton: CRC Press; 19861987.
hardly resulted in a concentration of these uids, pre- [5] Homann J, editor. Handbuch der Messtechnik. Munich:
sumably because the kidneys act to markedly reduce Hanser; 1999.
[6] West JB. Best and Taylors Physiological basis of medical prac-
water excretion (anti-diuresis) [19]. tice. 11th ed. Baltimore,: Williams & Wilkins; 1985.
Nevertheless, through the achieved variation of the [7] Zirk K, Poetzschke H, Barnikol WKR. Ein miniaturisierbares,
plasma osmolality it could be shown that the concen- sehr empndliches Polarimeter als Detektor einer implantierba-
K. Zirk, H. Poetzschke / Medical Engineering & Physics 26 (2004) 473481 481
ren Glukosesonde. I. Optische Verstarkung der Messgroe. Bio- [13] Boehm BO, Palitzsch K-D, Rosak C, Spinas GA, editors.
medizinische Technik/Biomedical Engineering 1989;46:15867. Klinische Diabetologie. Berlin: Springer; 2001.
[8] Koschinsky T, Heinemann L. Sensors for glucose monitoring: [14] Nunes SP, Peinemann K-V, editors. Membrane technology in
technical and clinical aspects. Diabetes/Metabolism Research the chemical industry. Weinheim: Wiley-VCH; 2001.
and Reviews 2001;17:11323. [15] Ciba-Geigy, editor. Wissenschaftliche Tabellen Geigy, 8.
[9] Frank H. HenningKlinische Laboratoriumsdiagnostik. 2. Auage. Auage. Basle: Geigy; 1979, 4. Nachdruck 1985.
Munich: Urban & Schwarzenberg; 1960. [16] Wahl HG. Blutzuckerbestimmung, http://www.labor-medizin.
[10] Nelson DL, Cox MM. LehningerPrinciples of biochemistry. de/bz/bz2.htm, July 2002.
3rd ed. New York: Worth Publishers; 2000. [17] Harvey JW. Erythrocyte metabolism. In: Kaneko JJ, editor.
[11] Browne CA, Zerban FW. Physical and chemical methods of Clinical biochemistry of domestic animals. San Diego: Academic
sugar analysisa practical and descriptive treatise for use in Press; 1989, p. 185234.
research, technical and control laboratories. New York: John [18] Reiss E. Ergebnisse innere Medizin und Kinderheilkunde
Wiley & Sons; 1941. 1913;10:531634.
[12] Berger M, editor. Diabetes mellitus. Munich: Urban & Schwar- [19] Guyton AC, Hall JE. Textbook of medical physiology. 9th ed.
zenberg; 1995. Philadelphia: Saunders; 1996.