- Email: [email protected]
Antibodies to GAD in Japanese diabetic patients: a multicenter study
ELSEVIER
Diabetes Research and Clinical Practice 28 (1995) 191-199
Antibodies to GAD in Japanese diabetic patients: a multicenter study Akira Tsuruokaa, Ikuro Matsubab, Takayoshi Toyotac, Gen Isshikid, Shigenobu Nagatakie, Yoshio Ikeda*f ‘Department of General Internal Medicine, The Jikei Universiry Kashiwa Hospital, Kaskiwa, Japan bDepartment of Internal Medicine (III), The Jikei University School of Medicine, Tokyo, Japan ‘Third Department of Internal Medicine, Tohoku University School of Medicine, Se&i. Japan ‘Department of Pediatrics, Osaka Municipal University School of Medicine, Osaka. Japan ‘firs1 Department of Internal Medicine, Nagasaki University School of Medicine, Nagasaki, Japan ‘Department of Preventive Medicine, The Jikei University School of Medicine, 3-25-8, Nishi-Shlmbashl, Minato-ku. Tokyo 105, Japan
Received 13 January 1995;revision received 7 June 1995;accepted 8 June 1995
Abstract
We determined the prevalence of antibodies to glutamic acid decarboxylase (anti-GAD) in Japanesediabetic patients. Anti-GAD were detected by RIP Anti-GAD Hoechst, which is a new sensitive radioimmunoassay (RIA) kit using purified pig brain GAD as the antigen. One thousand nine hundred Japanesepatients were collected by the Study Group for Antibodies to GAD. The prevalence of anti-GAD in the subjects of this study was: 35.4% (326/921)in all patients with IDDM, 50.3%(96/191)in patients with IDDM lessthan I-year duration, 4.3% (29/680)in NIDDM, 37.9% (39/103)in slowly progressive IDDM, 10.5%(4138)in gestational diabetes mellitus, 0% (O/27)in impaired glucose tolerance, 4.8% (61124)in the school children with glycosuria, 2.1% (l/47) in the relatives of IDDM and 5.0% (l/20) in neurological diseaseswithout diabetes. The prevalence in normal subjectswas 2.2% (7/323). Anti-GAD are frequently detected by the RIA kit in patients with IDDM of short duration and this assaymay be useful for population screening for IDDM and for better understanding of its pathogenesis. Keyworcts:Antibodies to glutamic acid decarboxylase; Radioimmunoassay; Insulin-dependent diabetes mellitus; Noninsulindependent diabetes mellitus
1. Introduction Insulin-dependent diabetes mellitus (IDDM) is caused by the destruction of insulin-secreting flcells of the pancreatic islets [I]. Many groups have * Corresponding author.
already reported the detection of some autoantibodies, such as islet cell antibodies (ICA), islet cell surface antibodies (ICSA), insulin autoantibodies (IAA), and auto-antibodies to 64-kD pancreatic islet @-cells protein (64-kD) in patients with new onset IDDM 121,and that ICA and 64kD are good predictive markers in high risk
0168-8227/95RO9.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0168-8227(95)OllOl-I
192
A. Tsuruoka et al. /Diabetes Research and Clinical Practice 28 (1995) 191-199
groups with IDDM [2-41. Although these antibodies play an important role in the pathogenesis of IDDM, these screening tests have not been widely used because technical difficulty makes large-scale studies impractical [5,6]. Recently, Baekkeskov et al. identified the autoantigen of 64-kD as glutamic acid decarboxylase (GAD), the biosynthesizing enzyme of the inhibitory neurotransmitter GABA (y-aminobutyric acid) [7]. The identification of one of the IDDM autoantigens can provide uniform and reliable antigens for the detection of predictive autoantibodies [5, 81. We developed a new sensitive radioimmunoassay (RIA) kit for the detection of autoantibdies to GAD (anti-GAD), RIP AntiGAD Hoechst (Tokyo, Japan) [9,10]. We have already reported that this assay is very simple and easy enough to allow population screening for anti-GAD [l 11.In the present study, we measured anti-GAD in patients with diabetes or associated diseasesat 14 laboratories in Japan in order to clarify the clinical significance and usefulness of anti-GAD in IDDM. 2. Subjects and methods 2.1. Study group for antibodies to GAD
The Study Group for Antibodies to GAD (coordinators: Ikeda, Y., Toyota, T., Isshiki, G. and Nagataki, S.) consisted of 14 laboratories in Japan, which were chosen because of previous publications reporting study of autoantibodies in IDDM, or interest in participating in this group. All clinical data related to the sampleswere reserved at the secretary office of the Study Group, Department of Internal Medicine (III), the Jikei University School of Medicine, so that the technicians were unaware of the source of the serum until the final results were obtained. 2.2. Subjects
All of the subjects in this study were selected randomly from Japanese patients who were followed and treated at each member center of the study group for antibodies to GAD. The prevalence of anti-GAD was calculated from the results of sera which were obtained at the subject’s first visit to the laboratory to avoid overlapping.
The control samples were obtained from 323 normal individuals without liver damage nor known past history of diabetes (147 males; 176 females, average: 41.6 years of age, distribution: l-89 years of age). We considered sera within normal limits if the antibody titers were within 3 standard deviations above the mean of the controls. Details of clinical characteristics of the subjects in this study are shown in Table 1. The diagnosis of diabetes was made by each laboratory according to the criteria in the technical report of World Health Organization (WHO) [12]. We considered the patients with known past history of overweight (> 24 body mass indices) as obese ones. There were 921 patients with IDDM, including 191patients (84 males; 107females)with IDDM of short duration (< 1 year). There were 680 patients with non-insulin dependent diabetes (NIDDM), including 322 patients with non-obese NIDDM (189 males; 133 females) and 214 patients with obese NIDDM (116 males; 98 females). There were 103 patients with slowly progressive IDDM (SPIDDM) [13, 141. We also tested 38 patients with gestational diabetes mellitus (GDM), 27 patients with impaired glucose tolerance (IGT), 124 non-diabetic school children with glucosuria at entry, 47 non-diabetic first-degree relatives of patients with IDDM, and 20 non-diabetic patients with neurological diseases. 2.3. Anti-GAD assay
Anti-GAD were measured using the anti-GAD immunoassay kit, RIP Anti-GAD Hoechst (Tokyo, Japan), which has been reported in detail elsewhere [lo]. In brief, GAD was eluted and purified from pig brain by affinity chromatography and iodinated with ‘25I to make ‘25I labeled GAD as the antigen. A serum sample (20 ~1)was added to the 100~1 ‘25I labeled GAD and 100 ~1 RIA assay buffer. The mixtures were incubated for 2 h at room temperature, added to anti-human IgG serum as a precipitation regent, incubated for 30 min at room temperature to make “‘I-GAD-antibody complexes, centrifuged for 30 min at 4OC,and precipitated. After the supernant was aspirated, the complexes were counted for radioactivity. To quantify the titer of anti-GAD, we added a positive serum sample to each assay,
A. Tsuruoka et al. /Diabetes Table 1 Clinical characteristics
193
of patients in this study No. of Patients
IDDM Short duration (< I year) Long duration (> I year) NIDDM Non-obese Obese Slowly progressive IDDM GDM IGT School children with glucosuria IDDM Relatives Neurological diseases Total
Research and Clinical Practice 28 (1995) 191-199
(male/female)
Age (Y) Average
(SD.)
921 I91 730 680 322 214
(3961525) (841107) (3W418) (3871293) (189/133) (I 16/98)
23.9 16.6 25.8 46.9 49.5 43.7
(I 5.29) (11.55) (15.58) (19.49) (17.31) (20.97)
I03 38 27 124 47 20
(57146) (-/38) (1918) (56/68) (20/27) (S/12)
40.2 32.2 32.9 13.3 28.3 46.9
(17.77) (5.00) (19.22) (2.74) (19.51) (18.27)
1960
(943/18.27)
Standard deviations of ages are shown in parentheses. IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin dependent diabetes mellitus; SPIDDM, slowly progressive insulin-dependent diabetes mellitus; GDM, gestational diabetes Standard deviations of ages are shown in parentheses. IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin dependent diabetes mellitus; SPIDDM, slowly progressive insulin-dependent diabetes mellitus; GDM, gestational diabetes mellitus; IGT, impaired glucose tolerance. mellitus; IGT, impaired glucose tolerance.
and expressedthe titer in units based on the percentage of radioactivity of the reference serum. The intra-assay coefficient of variation ranged from 2.4 to 5.1%, and the inter-assay coefficient of variation ranged from 3.5 to 8.9%. The assay shows a good reproducibility. 3. Results 3.1. Results in normal healthy controls
The mean f S.D. of titers in normal controls was 2.25 f 0.82 units. The titers of anti-GAD were normally distributed, as shown in Fig. l-a and l-b. Serum sampleswere considered positive if they contained over 5 units, which exceeded3 SD. above the control mean. By this criterion, 7 of the 323 controls (2.2%) had anti-GAD. The comparison of the distribution of titers between males and females(Fig. 1-a), and betweenjuveniles (< 20 years old) and adults (1 20 years old) (Fig. l-b) demonstrated no significant age- or sex-related difference in the distribution of titers.
3.2. Prevalence of anti-GAD in patients with IDDM (Table 2, Fig. 2)
The prevalence of anti-GAD in patients with IDDM was 35.5% (326 of 921). 96 (50.3%) of 191 patients with IDDM of short duration (< 1 year) were positive for anti-GAD, while 230 (3 1.5%) of 730 patients of long duration (~1 year) were positive (Fig. 3-a). The prevalence of anti-GAD gradually decreasedwith increasing durations of diseases, but it remained approximately 30% despite longer duration. The median anti-GAD titer in patients with IDDM of short duration was higher than that of long duration. However, the medians of long duration were stable at approximately 20 units (Fig. 3-b). In patients with IDDM of short duration (< l-year), its prevalence in younger onset ages was higher than that in older ages(Fig. 3-c). 3.3. Prevalence of anti-GAD NIDDM (Table 2, Fig. 2)
in patients
with
The prevalence of anti-GAD in patients with
194
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
a
Titers of antibodies
b
to GAD (U/ml)
200 1
Titers of antibodies
to GAD (U/ml)
Fig. la. Histogram of anti-GAD titers in normal subjects comparing males (@ and females ( q. Fig. lb. Histogram of anti-GAD titers in normal subjects comparing younger (below 20 years old; q ) and older subjects (over 20 years old; @.
A. Tsuruoka er al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
195
Table 2 Prevalence of anti-GAD antibodies in the patients for this study
IDDM Short duration( c 1 year) Long duraion ( > 1 year) NIDDM Non-obese 0beS.Z Slowly progressive IDDM GDM IGT School children with glucosuria IDDM Relatives Neurological diseases
Prevalence of anti-GAD
Positives
Total patients
Median anti-GAD titer
35.4% 50.3% 31.5% 4.3% 5.3% 1.4%
326 96 230 29 17 3
921 191 730 680 322 214
21 u 32 U 20 u 17 u 16 U 7u
37.9% 10.5% 0% 4.8% 2.1% 5.0%
39 4 0 6 1 I
103 38 27 124 47 20
66 u 6U <4u 19 u 18 U 7u
Fig. 2. Distributions of anti-GAD titers. The abbreviations are shown in Table 1. Dotted line (5 U/ml) shows 3 S.D.above the mean for normal subjects..; x 10 males, 0; x 1 male,O;x 10 females, 0;x 1 female.
196
A. Tsuruoka et al. /Diabetes Research and Ciinicai Practice 28 (1995) 191-199
NIDDM was 4.3% (29 of 680). After classifying the types of 536 patients with NIDDM, 3 (1.4 %) of 214 patients with obese NIDDM were antiGAD positive, while 17 (5.3%) of 322 with nonobese NIDDM were positive (P~0.05; x2 test, Fig. 4). 3.4. Prevalence of anti-GAD SPIDDM (Table 2, Fig. 2)
in patients
with
The prevalence of anti-GAD in patients with SPIDDM was 37.9%(39 of 103).The prevalence of anti-GAD in SPIDDM was as high as that in IDDM of long duration. The distribution of titers in SPIDDM was very similar to that of IDDM. b
60
3.5. Prevalence of anti-GAD inpatients with associated diseases (Table 2, Fig. 2)
70 60 z
50
8 5
50
1
40
H
30
30
20
20
6 f g
10
10
5
0
%
40
!z E P
E
a cl
1-2
3-5 Duration
c
C-10 of IDDM
>I5
Ii-15
(Years)
The prevalence of anti-GAD in patients with associateddiseaseswas: 10.5%(4 of 38) in GDM, 0% (none of 27) in IGT, 4.8% (6 of 124) in school children with glucosuria, 2.1% (1 of 47) in nondiabetic relatives of patients with IDDM and 5.0% (1 in 20) in non-diabetic neurological diseases. Although diabetes was not recognized in school children with glucosuria at entry, a follow-up after the study identified IDDM in some children with anti-GAD. Of the 6 children with anti-GAD, we
80 1
" E 3
6-10
11.15
16.25
26-35
10000
i
235
Onset age (Years) Fig. 3a. Prevalence of anti-GAD in total IDDM. Comparison between the prevalence of short duration (less than I year), and long duration (more than I year). Fig. 3b. Prevalence and median titers of anti-GAD positives on duration of IDDM. EJ ; prevalence of male, Q prevalence of female. -0-; medians of titers in male, ---A---; medians of titers in female. Fig. 3c. Prevalence of anti-GAD for onset ages of IDDM with duration less than I year. W, total, Cl ; male, l& female.
, Obese
Non-obese
NIDDM Fig. 4. Distributions of anti-GAD titers comparing non-obese and obese types of NIDDM. Dotted line (5 U/ml) shows 3 S.D.above the mean of normal subjects. 0; x IO males, l ;x I male, 0;x IO females, 0;x I female.
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
found 2 patients with IDDM, one with NIDDM, and one with SPIDDM. The titers of patients with IDDM were 891 units and 492 units, respectively. Of 118 children without anti-GAD, we found no patients with IDDM, but only one patient with NIDDM. At present, we have not yet identified diabetes from any anti-GAD positives in relatives of patient with IDDM, patients with GDM, those with IGT, or those with neurological diseases. 4. Discussion We measured anti-GAD in this study using an RIA kit, RIP Anti-GAD Hoechst (Hoechst Japan Ltd.), which is simple enough to allow screening of a large population [lo]. At the first GAD antibody workshop held by the Immunology and Diabetes Workshop, the result using this kit received a perfect score on lab validity, specificity, and sensitivity [15,16]. Using this kit, the prevalence of anti-GAD in patients with IDDM was 35.5%. The prevalence (50.3%) and median of anti-GAD titers were higher in patients with young onset IDDM of short duration than in ones with older onset of long duration. Using a RIA kit, which seemedvery similar to our method, Rowley et al. [ 171reported that the prevalence of anti-GAD was 63% in Europeans with IDDM of short duration, but that it was much lower in Asian populations including Japanese patients with IDDM (5-33%) than in Europeans [18, 191. However, the prevalence in Japanesepatients with IDDM in our study was higher than their result. At present, any ethnic difference in prevalence remains unresolved because of limited reports, and further studies should be needed. While the titers and prevalence of ICA were reported to fall with increasing duration of IDDM [2, 61,those of anti-GAD in our study persisted even with long duration. Rowley et al. also noted that anti-GAD did not have a transient reactivity [17]. There were some anti-GAD positives with NIDDM. The frequency in non-obese NIDDM (5.3%) tended to be higher than in the obese (1.4%). In some non-Europeans, the classification into IDDM or NIDDM categories may be dif-
191
ticult. Similar classification difficulties are sometimes found in Europeans [ 12, 20, 211. The WHO technical report [12] uses a term of ‘questionable insulin dependency’ for these cases,but it has not been included in the classification of diabetes mellitus because it cannot be clearly defined. We estimate that someanti-GAD positive patients with non-obese NIDDM may be included in this ‘questionable insulin dependency’ group. The pathogenesisin some anti-GAD positive nonobeseNIDDM could be similar to that in IDDM: autoimmune destruction of the pancreatic P-cells. SPIDDM was proposed by Kobayashi et al. as a new category of diabetes [13, 141.In our study, the prevalence of anti-GAD in patients with SPIDDM (37.9%) was almost equal to that in patients with IDDM. This may confirm that SPIDDM could be included in a category of IDDM. Because a high frequency of ICA is reported in SPIDDM, the study of concordance between anti-GAD and ICA is needed. Until now, 64 kD were the most important predictive markers for IDDM [4,22]. Therefore, antiGAD have been reported as additional predictive markers for IDDM [2, 231. Our study also indicated that anti-GAD were associatedwith a very high risk for the development of IDDM. According to the result of school children with glucosuria, we noted that 2 of 6 children with anti-GAD developed IDDM, one developed SPIDDM, and one developed NIDDM. Only one anti-GAD negative child developed NIDDM. All patients with IDDM came from anti-GAD positive group with high titers. These results indicate that anti-GAD are valuable predictive markers for IDDM. We should study the predictive value of anti-GAD for IDDM in a larger number of children. Other groups reported that anti-GAD in some relatives of patients with IDDM should be of value in ascertaining their risk for developing IDDM [25, 261. Becauseof a small number of casesin our study, anti-GAD could not be shown to be valuable predictive markers for IDDM in relatives. In conclusion, compared with the cumbersome traditional ICA method, however, our assay for anti-GAD is simple enough to allow screening of a large population. Anti-GAD are frequently detected using this method in Japanese patients
198
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
with young onset IDDM of short duration. We believe that large-scale studies are needed to confirm the clinical significance and usefulnessof this RIA for anti-GAD by comparison with other methods, and to approve it for international standardization. This assay may be useful for population screening, and aid prediction and prevention of IDDM. Acknowledgments We thank Pharma R & D Division, Drug Discovery Research Laboratories, Hoechst Japan Ltd. for their excellent technical assistance. Members of the Study Group for Anti-GAD Participating members: Third Department of Internal Medicine, Tohoku University; Department of Endocrinology and Metabolism, Jichi Medical College; SecondDepartment of Internal Medicine, Gumma University; Department of Pediatrics, Chiba University; Department of Internal Medicine (III), the Jikei University School of Medicine; Diabetes Center, Tokyo Women’s Medical College; Department of Endocrinology and Metabolism, Toranomon Hospital; Department of Pediatrics, Nihon University; Second Department of Internal Medicine, Osaka University; Department of Pediatrics, Osaka Municipal University; Second Department of Internal Medicine, Kobe University; Third Department of Internal Medicine, Hiroshima University; Fourth Department of Internal Medicine, Kurume University; First Department of Internal Medicine, Nagasaki University. References [I] Leslie, R.D.G. and Elliott, R.B. (1994) Early environmental events as a cause of IDDM. Diabetes 43, 843-850. [2] Atkinson, M.A. and Maclaren, N.K. (1993) Islet cell autoantigens in insulin-dependent diabetes. J. Clin. Invest. 92, 1608-1616. [3] Maclaren, N. (1992) Immunology of diabetes mellitus. Ann. Allergy. 68, 5-9. [41 Atkinson, M.A., MaClaren, N., Scharp, D.W., Lacy, PE. and Riley, W.J. (1990) 64 000 Mr autoantibodies as predictors of insulin-dependent diabetes. Lancet 335, 1357-1367. 151 Glare-Salzler, M.J., Tobin, A.J. and Kaufman, D.L.
(1992) Glutamate decarboxylase: an autoantigen in IDDM. Diabetes Care 15, 132-135. 161Eisenbarth, G.S., Ziegler, A.G. and Colman, P.A. (1994) Pathogenesis of insulin-dependent (Type I) diabetes mellitus. In: CR. Kahn and G.C. Weir (Eds.), Joslin’s Diabetes Mellitus. 13th Edit, Lea and Febiger, Boston, pp. 216-239. 171 Baekkeskov, S., Aanstoot, H-J., Christgau, S. et al (1990) Identification of the 64K autoantigen in insulindependent diabetes as the GABA-synthesizing enzyme glutamic acid decarboxylase. Nature 347, I51- 156. 181DeAixpurua, H.J. and Harrison, L.C. (1992) Glutamic acid decarboxylase in insulin-dependent diabetes mellitus. Diabetes/ Met. Rev. 8, 133-147. [91 Tsuruoka, A., Matsuba, I., Ogata, K. and Ikeda, Y. (1993)Relationship between glutamic acid decarboxylase antibodies (GAD-Ab) and islet cell antibodies in insulindependent diabetes mellitus. J. Jpn. Diab. Sot. 36, 903-908. (in Japanesewith English abstract) 1101Yamaguchi, A., Ogata, K., Kubo, H. et al (1994) Performance and characteristic study of ‘RIP Anti-GAD Hoechst’ RIA kit in detection of anti-GAD antibodies. Med. Pharm. 31, 419-431. (in Japanese) 1111Matsuba, I., Tsuruoka, A., Kawasaki, E. et al (1994) Assay of antibodies to GAD in IDDM, NIDDM, and normal healthy subjects - multi-center study - . Practice I I, 178-182. (in Japanese) 1121WHO Study Group (1985) Definition, diagnosis, and classification: Diabetes Mellitus. In: World Health Organization Technical Report Series727, Geneva, pp. 9-20. iI31 Kobayashi, T. (1994) Natural history of insulindependent diabetes mellitus - slowly progressive IDDM. J. Jpn Diab. Sot. 37, 103-105. (in Japanese) (141 Kobayashi, T., Itoh, T., Kosaka, K., Sato, K. and Tsuji, K. (1987) Time course of islet cell antibodies and &cell function in non-insulin-dependent stage of Type I diabetes. Diabetes 36, 510-517. Maclaren, N. and Lafferty, K. (1993) The 12th international immunology and diabetes workshop. Diabetes 42, 1099-1104. 1161Schmidli, R.S., Colman, P.G., Bonifacio, E., Bottaxxo, G.F., Harrison, L.C. and participating laboratories (1994) High level of concordance between assays for glutamic acid decarboxylase antibodies - the first international glutamic acid decarboxylase antibody workshop. Diabetes 44, 1005-1009. 1171 Rowley, M., Mackay, I.R., Chen, Q., Knowles, W.J. and Zimmet, P. (1992)Antibodies to glutamic acid decarboxylase discriminate major types of diabetes mellitus. Diabetes 41, 548-551. WI Serjeantson, S.W., Kohonen-Corish, M.R.J., Rowley, M.J., Knowlers, W. and Zimmet, P. (1992)Antibodies to glutamic acid decarboxylase are associatedthe HLA-DR genotypes in both Australians and Asians with Type I (insulin-dependent) diabetes mellitus. Diabetologia 35, 996-1001. iI91 Zimmet, P.Z., Rowley, M.J., Mackay, I.R. et al. (1993)
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
The ethnic distribution of antibodies to glutamic acid decarboxylase: presence and levels in insulin-dependent diabetes mellitus in Europid and Asian subjects. J. Diab. Comp. 7, l-7. 1201 Tuomi, T., Groop, L.C., Zimmet, P.Z., Rowley, M.J., Knowles, W. and Mackay, I.R. (1993) Antibodies to glutamic acid decarboxylase reveal latent autoimmune diabetes mellitus in adults with a non-insulin-dependent onset of disease. Diabetes 42, 359-362. [21] Zimmet, P.Z., Tuomi, T., Mackay, 1.R. et al. (1993) Latent autoimmune diabetes mellitus in adults (LADA): the role of antibodies to glutamic acid decarboxylase in diagnosis and prediction of insulin dependency. Diabetic Med. II, 299-303. [22] Christie, M., Landin-Olsson, M., Sundkvist, G., Dahlquist, G., Lernmark, A. and Baekkeskov, S. (1988) Antibodies to a Mr-64 000 islet cell protein in Swedish children with newly diagnosed Type 1 (insulindependent) diabetes. Diabetologia 31, 597-602.
199
[23] DeAizpurua, H.J., Wilson, Y.M. and Harrison, L.C. (1992) Glutamic acid decarboxylase autoantibodies in preclinical insulin-dependent diabetes. Proc. Natl. Acad. Sci. USA 89, 9841-9845. [24] Tuomilehto, J., Zimmet, P., Mackay, I.R. et al. (1994) Antibodies to glutamic acid decarboxylase as predictors of insulin-dependent diabetes mellitus before clinical onset of disease. Lancet 343, 1383-1385. [25] Chen, Q-Y., Rowley, M.J., Byrne, G.C. et al. (1993) Antibodies to glutamic acid decarboxylase in Australian children with insulin-dependent diabetes mellitus and their first-degree relatives. Pediatr. Res. 34, 785-790. [26] Thivolet, C.H., Tappaz, M., Durand, A. et al. (1992) Glutamic acid decarboxylase (GAD) autoantibodies are additional predictive markers of Type I (insulindependent) diabetes mellitus in high risk individuals. Diabetologia 35, 570-576.
Diabetes Research and Clinical Practice 28 (1995) 191-199
Antibodies to GAD in Japanese diabetic patients: a multicenter study Akira Tsuruokaa, Ikuro Matsubab, Takayoshi Toyotac, Gen Isshikid, Shigenobu Nagatakie, Yoshio Ikeda*f ‘Department of General Internal Medicine, The Jikei Universiry Kashiwa Hospital, Kaskiwa, Japan bDepartment of Internal Medicine (III), The Jikei University School of Medicine, Tokyo, Japan ‘Third Department of Internal Medicine, Tohoku University School of Medicine, Se&i. Japan ‘Department of Pediatrics, Osaka Municipal University School of Medicine, Osaka. Japan ‘firs1 Department of Internal Medicine, Nagasaki University School of Medicine, Nagasaki, Japan ‘Department of Preventive Medicine, The Jikei University School of Medicine, 3-25-8, Nishi-Shlmbashl, Minato-ku. Tokyo 105, Japan
Received 13 January 1995;revision received 7 June 1995;accepted 8 June 1995
Abstract
We determined the prevalence of antibodies to glutamic acid decarboxylase (anti-GAD) in Japanesediabetic patients. Anti-GAD were detected by RIP Anti-GAD Hoechst, which is a new sensitive radioimmunoassay (RIA) kit using purified pig brain GAD as the antigen. One thousand nine hundred Japanesepatients were collected by the Study Group for Antibodies to GAD. The prevalence of anti-GAD in the subjects of this study was: 35.4% (326/921)in all patients with IDDM, 50.3%(96/191)in patients with IDDM lessthan I-year duration, 4.3% (29/680)in NIDDM, 37.9% (39/103)in slowly progressive IDDM, 10.5%(4138)in gestational diabetes mellitus, 0% (O/27)in impaired glucose tolerance, 4.8% (61124)in the school children with glycosuria, 2.1% (l/47) in the relatives of IDDM and 5.0% (l/20) in neurological diseaseswithout diabetes. The prevalence in normal subjectswas 2.2% (7/323). Anti-GAD are frequently detected by the RIA kit in patients with IDDM of short duration and this assaymay be useful for population screening for IDDM and for better understanding of its pathogenesis. Keyworcts:Antibodies to glutamic acid decarboxylase; Radioimmunoassay; Insulin-dependent diabetes mellitus; Noninsulindependent diabetes mellitus
1. Introduction Insulin-dependent diabetes mellitus (IDDM) is caused by the destruction of insulin-secreting flcells of the pancreatic islets [I]. Many groups have * Corresponding author.
already reported the detection of some autoantibodies, such as islet cell antibodies (ICA), islet cell surface antibodies (ICSA), insulin autoantibodies (IAA), and auto-antibodies to 64-kD pancreatic islet @-cells protein (64-kD) in patients with new onset IDDM 121,and that ICA and 64kD are good predictive markers in high risk
0168-8227/95RO9.50 0 1995 Elsevier Science Ireland Ltd. All rights reserved SSDI 0168-8227(95)OllOl-I
192
A. Tsuruoka et al. /Diabetes Research and Clinical Practice 28 (1995) 191-199
groups with IDDM [2-41. Although these antibodies play an important role in the pathogenesis of IDDM, these screening tests have not been widely used because technical difficulty makes large-scale studies impractical [5,6]. Recently, Baekkeskov et al. identified the autoantigen of 64-kD as glutamic acid decarboxylase (GAD), the biosynthesizing enzyme of the inhibitory neurotransmitter GABA (y-aminobutyric acid) [7]. The identification of one of the IDDM autoantigens can provide uniform and reliable antigens for the detection of predictive autoantibodies [5, 81. We developed a new sensitive radioimmunoassay (RIA) kit for the detection of autoantibdies to GAD (anti-GAD), RIP AntiGAD Hoechst (Tokyo, Japan) [9,10]. We have already reported that this assay is very simple and easy enough to allow population screening for anti-GAD [l 11.In the present study, we measured anti-GAD in patients with diabetes or associated diseasesat 14 laboratories in Japan in order to clarify the clinical significance and usefulness of anti-GAD in IDDM. 2. Subjects and methods 2.1. Study group for antibodies to GAD
The Study Group for Antibodies to GAD (coordinators: Ikeda, Y., Toyota, T., Isshiki, G. and Nagataki, S.) consisted of 14 laboratories in Japan, which were chosen because of previous publications reporting study of autoantibodies in IDDM, or interest in participating in this group. All clinical data related to the sampleswere reserved at the secretary office of the Study Group, Department of Internal Medicine (III), the Jikei University School of Medicine, so that the technicians were unaware of the source of the serum until the final results were obtained. 2.2. Subjects
All of the subjects in this study were selected randomly from Japanese patients who were followed and treated at each member center of the study group for antibodies to GAD. The prevalence of anti-GAD was calculated from the results of sera which were obtained at the subject’s first visit to the laboratory to avoid overlapping.
The control samples were obtained from 323 normal individuals without liver damage nor known past history of diabetes (147 males; 176 females, average: 41.6 years of age, distribution: l-89 years of age). We considered sera within normal limits if the antibody titers were within 3 standard deviations above the mean of the controls. Details of clinical characteristics of the subjects in this study are shown in Table 1. The diagnosis of diabetes was made by each laboratory according to the criteria in the technical report of World Health Organization (WHO) [12]. We considered the patients with known past history of overweight (> 24 body mass indices) as obese ones. There were 921 patients with IDDM, including 191patients (84 males; 107females)with IDDM of short duration (< 1 year). There were 680 patients with non-insulin dependent diabetes (NIDDM), including 322 patients with non-obese NIDDM (189 males; 133 females) and 214 patients with obese NIDDM (116 males; 98 females). There were 103 patients with slowly progressive IDDM (SPIDDM) [13, 141. We also tested 38 patients with gestational diabetes mellitus (GDM), 27 patients with impaired glucose tolerance (IGT), 124 non-diabetic school children with glucosuria at entry, 47 non-diabetic first-degree relatives of patients with IDDM, and 20 non-diabetic patients with neurological diseases. 2.3. Anti-GAD assay
Anti-GAD were measured using the anti-GAD immunoassay kit, RIP Anti-GAD Hoechst (Tokyo, Japan), which has been reported in detail elsewhere [lo]. In brief, GAD was eluted and purified from pig brain by affinity chromatography and iodinated with ‘25I to make ‘25I labeled GAD as the antigen. A serum sample (20 ~1)was added to the 100~1 ‘25I labeled GAD and 100 ~1 RIA assay buffer. The mixtures were incubated for 2 h at room temperature, added to anti-human IgG serum as a precipitation regent, incubated for 30 min at room temperature to make “‘I-GAD-antibody complexes, centrifuged for 30 min at 4OC,and precipitated. After the supernant was aspirated, the complexes were counted for radioactivity. To quantify the titer of anti-GAD, we added a positive serum sample to each assay,
A. Tsuruoka et al. /Diabetes Table 1 Clinical characteristics
193
of patients in this study No. of Patients
IDDM Short duration (< I year) Long duration (> I year) NIDDM Non-obese Obese Slowly progressive IDDM GDM IGT School children with glucosuria IDDM Relatives Neurological diseases Total
Research and Clinical Practice 28 (1995) 191-199
(male/female)
Age (Y) Average
(SD.)
921 I91 730 680 322 214
(3961525) (841107) (3W418) (3871293) (189/133) (I 16/98)
23.9 16.6 25.8 46.9 49.5 43.7
(I 5.29) (11.55) (15.58) (19.49) (17.31) (20.97)
I03 38 27 124 47 20
(57146) (-/38) (1918) (56/68) (20/27) (S/12)
40.2 32.2 32.9 13.3 28.3 46.9
(17.77) (5.00) (19.22) (2.74) (19.51) (18.27)
1960
(943/18.27)
Standard deviations of ages are shown in parentheses. IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin dependent diabetes mellitus; SPIDDM, slowly progressive insulin-dependent diabetes mellitus; GDM, gestational diabetes Standard deviations of ages are shown in parentheses. IDDM, insulin-dependent diabetes mellitus; NIDDM, non-insulin dependent diabetes mellitus; SPIDDM, slowly progressive insulin-dependent diabetes mellitus; GDM, gestational diabetes mellitus; IGT, impaired glucose tolerance. mellitus; IGT, impaired glucose tolerance.
and expressedthe titer in units based on the percentage of radioactivity of the reference serum. The intra-assay coefficient of variation ranged from 2.4 to 5.1%, and the inter-assay coefficient of variation ranged from 3.5 to 8.9%. The assay shows a good reproducibility. 3. Results 3.1. Results in normal healthy controls
The mean f S.D. of titers in normal controls was 2.25 f 0.82 units. The titers of anti-GAD were normally distributed, as shown in Fig. l-a and l-b. Serum sampleswere considered positive if they contained over 5 units, which exceeded3 SD. above the control mean. By this criterion, 7 of the 323 controls (2.2%) had anti-GAD. The comparison of the distribution of titers between males and females(Fig. 1-a), and betweenjuveniles (< 20 years old) and adults (1 20 years old) (Fig. l-b) demonstrated no significant age- or sex-related difference in the distribution of titers.
3.2. Prevalence of anti-GAD in patients with IDDM (Table 2, Fig. 2)
The prevalence of anti-GAD in patients with IDDM was 35.5% (326 of 921). 96 (50.3%) of 191 patients with IDDM of short duration (< 1 year) were positive for anti-GAD, while 230 (3 1.5%) of 730 patients of long duration (~1 year) were positive (Fig. 3-a). The prevalence of anti-GAD gradually decreasedwith increasing durations of diseases, but it remained approximately 30% despite longer duration. The median anti-GAD titer in patients with IDDM of short duration was higher than that of long duration. However, the medians of long duration were stable at approximately 20 units (Fig. 3-b). In patients with IDDM of short duration (< l-year), its prevalence in younger onset ages was higher than that in older ages(Fig. 3-c). 3.3. Prevalence of anti-GAD NIDDM (Table 2, Fig. 2)
in patients
with
The prevalence of anti-GAD in patients with
194
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
a
Titers of antibodies
b
to GAD (U/ml)
200 1
Titers of antibodies
to GAD (U/ml)
Fig. la. Histogram of anti-GAD titers in normal subjects comparing males (@ and females ( q. Fig. lb. Histogram of anti-GAD titers in normal subjects comparing younger (below 20 years old; q ) and older subjects (over 20 years old; @.
A. Tsuruoka er al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
195
Table 2 Prevalence of anti-GAD antibodies in the patients for this study
IDDM Short duration( c 1 year) Long duraion ( > 1 year) NIDDM Non-obese 0beS.Z Slowly progressive IDDM GDM IGT School children with glucosuria IDDM Relatives Neurological diseases
Prevalence of anti-GAD
Positives
Total patients
Median anti-GAD titer
35.4% 50.3% 31.5% 4.3% 5.3% 1.4%
326 96 230 29 17 3
921 191 730 680 322 214
21 u 32 U 20 u 17 u 16 U 7u
37.9% 10.5% 0% 4.8% 2.1% 5.0%
39 4 0 6 1 I
103 38 27 124 47 20
66 u 6U <4u 19 u 18 U 7u
Fig. 2. Distributions of anti-GAD titers. The abbreviations are shown in Table 1. Dotted line (5 U/ml) shows 3 S.D.above the mean for normal subjects..; x 10 males, 0; x 1 male,O;x 10 females, 0;x 1 female.
196
A. Tsuruoka et al. /Diabetes Research and Ciinicai Practice 28 (1995) 191-199
NIDDM was 4.3% (29 of 680). After classifying the types of 536 patients with NIDDM, 3 (1.4 %) of 214 patients with obese NIDDM were antiGAD positive, while 17 (5.3%) of 322 with nonobese NIDDM were positive (P~0.05; x2 test, Fig. 4). 3.4. Prevalence of anti-GAD SPIDDM (Table 2, Fig. 2)
in patients
with
The prevalence of anti-GAD in patients with SPIDDM was 37.9%(39 of 103).The prevalence of anti-GAD in SPIDDM was as high as that in IDDM of long duration. The distribution of titers in SPIDDM was very similar to that of IDDM. b
60
3.5. Prevalence of anti-GAD inpatients with associated diseases (Table 2, Fig. 2)
70 60 z
50
8 5
50
1
40
H
30
30
20
20
6 f g
10
10
5
0
%
40
!z E P
E
a cl
1-2
3-5 Duration
c
C-10 of IDDM
>I5
Ii-15
(Years)
The prevalence of anti-GAD in patients with associateddiseaseswas: 10.5%(4 of 38) in GDM, 0% (none of 27) in IGT, 4.8% (6 of 124) in school children with glucosuria, 2.1% (1 of 47) in nondiabetic relatives of patients with IDDM and 5.0% (1 in 20) in non-diabetic neurological diseases. Although diabetes was not recognized in school children with glucosuria at entry, a follow-up after the study identified IDDM in some children with anti-GAD. Of the 6 children with anti-GAD, we
80 1
" E 3
6-10
11.15
16.25
26-35
10000
i
235
Onset age (Years) Fig. 3a. Prevalence of anti-GAD in total IDDM. Comparison between the prevalence of short duration (less than I year), and long duration (more than I year). Fig. 3b. Prevalence and median titers of anti-GAD positives on duration of IDDM. EJ ; prevalence of male, Q prevalence of female. -0-; medians of titers in male, ---A---; medians of titers in female. Fig. 3c. Prevalence of anti-GAD for onset ages of IDDM with duration less than I year. W, total, Cl ; male, l& female.
, Obese
Non-obese
NIDDM Fig. 4. Distributions of anti-GAD titers comparing non-obese and obese types of NIDDM. Dotted line (5 U/ml) shows 3 S.D.above the mean of normal subjects. 0; x IO males, l ;x I male, 0;x IO females, 0;x I female.
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
found 2 patients with IDDM, one with NIDDM, and one with SPIDDM. The titers of patients with IDDM were 891 units and 492 units, respectively. Of 118 children without anti-GAD, we found no patients with IDDM, but only one patient with NIDDM. At present, we have not yet identified diabetes from any anti-GAD positives in relatives of patient with IDDM, patients with GDM, those with IGT, or those with neurological diseases. 4. Discussion We measured anti-GAD in this study using an RIA kit, RIP Anti-GAD Hoechst (Hoechst Japan Ltd.), which is simple enough to allow screening of a large population [lo]. At the first GAD antibody workshop held by the Immunology and Diabetes Workshop, the result using this kit received a perfect score on lab validity, specificity, and sensitivity [15,16]. Using this kit, the prevalence of anti-GAD in patients with IDDM was 35.5%. The prevalence (50.3%) and median of anti-GAD titers were higher in patients with young onset IDDM of short duration than in ones with older onset of long duration. Using a RIA kit, which seemedvery similar to our method, Rowley et al. [ 171reported that the prevalence of anti-GAD was 63% in Europeans with IDDM of short duration, but that it was much lower in Asian populations including Japanese patients with IDDM (5-33%) than in Europeans [18, 191. However, the prevalence in Japanesepatients with IDDM in our study was higher than their result. At present, any ethnic difference in prevalence remains unresolved because of limited reports, and further studies should be needed. While the titers and prevalence of ICA were reported to fall with increasing duration of IDDM [2, 61,those of anti-GAD in our study persisted even with long duration. Rowley et al. also noted that anti-GAD did not have a transient reactivity [17]. There were some anti-GAD positives with NIDDM. The frequency in non-obese NIDDM (5.3%) tended to be higher than in the obese (1.4%). In some non-Europeans, the classification into IDDM or NIDDM categories may be dif-
191
ticult. Similar classification difficulties are sometimes found in Europeans [ 12, 20, 211. The WHO technical report [12] uses a term of ‘questionable insulin dependency’ for these cases,but it has not been included in the classification of diabetes mellitus because it cannot be clearly defined. We estimate that someanti-GAD positive patients with non-obese NIDDM may be included in this ‘questionable insulin dependency’ group. The pathogenesisin some anti-GAD positive nonobeseNIDDM could be similar to that in IDDM: autoimmune destruction of the pancreatic P-cells. SPIDDM was proposed by Kobayashi et al. as a new category of diabetes [13, 141.In our study, the prevalence of anti-GAD in patients with SPIDDM (37.9%) was almost equal to that in patients with IDDM. This may confirm that SPIDDM could be included in a category of IDDM. Because a high frequency of ICA is reported in SPIDDM, the study of concordance between anti-GAD and ICA is needed. Until now, 64 kD were the most important predictive markers for IDDM [4,22]. Therefore, antiGAD have been reported as additional predictive markers for IDDM [2, 231. Our study also indicated that anti-GAD were associatedwith a very high risk for the development of IDDM. According to the result of school children with glucosuria, we noted that 2 of 6 children with anti-GAD developed IDDM, one developed SPIDDM, and one developed NIDDM. Only one anti-GAD negative child developed NIDDM. All patients with IDDM came from anti-GAD positive group with high titers. These results indicate that anti-GAD are valuable predictive markers for IDDM. We should study the predictive value of anti-GAD for IDDM in a larger number of children. Other groups reported that anti-GAD in some relatives of patients with IDDM should be of value in ascertaining their risk for developing IDDM [25, 261. Becauseof a small number of casesin our study, anti-GAD could not be shown to be valuable predictive markers for IDDM in relatives. In conclusion, compared with the cumbersome traditional ICA method, however, our assay for anti-GAD is simple enough to allow screening of a large population. Anti-GAD are frequently detected using this method in Japanese patients
198
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
with young onset IDDM of short duration. We believe that large-scale studies are needed to confirm the clinical significance and usefulnessof this RIA for anti-GAD by comparison with other methods, and to approve it for international standardization. This assay may be useful for population screening, and aid prediction and prevention of IDDM. Acknowledgments We thank Pharma R & D Division, Drug Discovery Research Laboratories, Hoechst Japan Ltd. for their excellent technical assistance. Members of the Study Group for Anti-GAD Participating members: Third Department of Internal Medicine, Tohoku University; Department of Endocrinology and Metabolism, Jichi Medical College; SecondDepartment of Internal Medicine, Gumma University; Department of Pediatrics, Chiba University; Department of Internal Medicine (III), the Jikei University School of Medicine; Diabetes Center, Tokyo Women’s Medical College; Department of Endocrinology and Metabolism, Toranomon Hospital; Department of Pediatrics, Nihon University; Second Department of Internal Medicine, Osaka University; Department of Pediatrics, Osaka Municipal University; Second Department of Internal Medicine, Kobe University; Third Department of Internal Medicine, Hiroshima University; Fourth Department of Internal Medicine, Kurume University; First Department of Internal Medicine, Nagasaki University. References [I] Leslie, R.D.G. and Elliott, R.B. (1994) Early environmental events as a cause of IDDM. Diabetes 43, 843-850. [2] Atkinson, M.A. and Maclaren, N.K. (1993) Islet cell autoantigens in insulin-dependent diabetes. J. Clin. Invest. 92, 1608-1616. [3] Maclaren, N. (1992) Immunology of diabetes mellitus. Ann. Allergy. 68, 5-9. [41 Atkinson, M.A., MaClaren, N., Scharp, D.W., Lacy, PE. and Riley, W.J. (1990) 64 000 Mr autoantibodies as predictors of insulin-dependent diabetes. Lancet 335, 1357-1367. 151 Glare-Salzler, M.J., Tobin, A.J. and Kaufman, D.L.
(1992) Glutamate decarboxylase: an autoantigen in IDDM. Diabetes Care 15, 132-135. 161Eisenbarth, G.S., Ziegler, A.G. and Colman, P.A. (1994) Pathogenesis of insulin-dependent (Type I) diabetes mellitus. In: CR. Kahn and G.C. Weir (Eds.), Joslin’s Diabetes Mellitus. 13th Edit, Lea and Febiger, Boston, pp. 216-239. 171 Baekkeskov, S., Aanstoot, H-J., Christgau, S. et al (1990) Identification of the 64K autoantigen in insulindependent diabetes as the GABA-synthesizing enzyme glutamic acid decarboxylase. Nature 347, I51- 156. 181DeAixpurua, H.J. and Harrison, L.C. (1992) Glutamic acid decarboxylase in insulin-dependent diabetes mellitus. Diabetes/ Met. Rev. 8, 133-147. [91 Tsuruoka, A., Matsuba, I., Ogata, K. and Ikeda, Y. (1993)Relationship between glutamic acid decarboxylase antibodies (GAD-Ab) and islet cell antibodies in insulindependent diabetes mellitus. J. Jpn. Diab. Sot. 36, 903-908. (in Japanesewith English abstract) 1101Yamaguchi, A., Ogata, K., Kubo, H. et al (1994) Performance and characteristic study of ‘RIP Anti-GAD Hoechst’ RIA kit in detection of anti-GAD antibodies. Med. Pharm. 31, 419-431. (in Japanese) 1111Matsuba, I., Tsuruoka, A., Kawasaki, E. et al (1994) Assay of antibodies to GAD in IDDM, NIDDM, and normal healthy subjects - multi-center study - . Practice I I, 178-182. (in Japanese) 1121WHO Study Group (1985) Definition, diagnosis, and classification: Diabetes Mellitus. In: World Health Organization Technical Report Series727, Geneva, pp. 9-20. iI31 Kobayashi, T. (1994) Natural history of insulindependent diabetes mellitus - slowly progressive IDDM. J. Jpn Diab. Sot. 37, 103-105. (in Japanese) (141 Kobayashi, T., Itoh, T., Kosaka, K., Sato, K. and Tsuji, K. (1987) Time course of islet cell antibodies and &cell function in non-insulin-dependent stage of Type I diabetes. Diabetes 36, 510-517. Maclaren, N. and Lafferty, K. (1993) The 12th international immunology and diabetes workshop. Diabetes 42, 1099-1104. 1161Schmidli, R.S., Colman, P.G., Bonifacio, E., Bottaxxo, G.F., Harrison, L.C. and participating laboratories (1994) High level of concordance between assays for glutamic acid decarboxylase antibodies - the first international glutamic acid decarboxylase antibody workshop. Diabetes 44, 1005-1009. 1171 Rowley, M., Mackay, I.R., Chen, Q., Knowles, W.J. and Zimmet, P. (1992)Antibodies to glutamic acid decarboxylase discriminate major types of diabetes mellitus. Diabetes 41, 548-551. WI Serjeantson, S.W., Kohonen-Corish, M.R.J., Rowley, M.J., Knowlers, W. and Zimmet, P. (1992)Antibodies to glutamic acid decarboxylase are associatedthe HLA-DR genotypes in both Australians and Asians with Type I (insulin-dependent) diabetes mellitus. Diabetologia 35, 996-1001. iI91 Zimmet, P.Z., Rowley, M.J., Mackay, I.R. et al. (1993)
A. Tsuruoka et al. /Diabetes
Research and Clinical Practice 28 (1995) 191-199
The ethnic distribution of antibodies to glutamic acid decarboxylase: presence and levels in insulin-dependent diabetes mellitus in Europid and Asian subjects. J. Diab. Comp. 7, l-7. 1201 Tuomi, T., Groop, L.C., Zimmet, P.Z., Rowley, M.J., Knowles, W. and Mackay, I.R. (1993) Antibodies to glutamic acid decarboxylase reveal latent autoimmune diabetes mellitus in adults with a non-insulin-dependent onset of disease. Diabetes 42, 359-362. [21] Zimmet, P.Z., Tuomi, T., Mackay, 1.R. et al. (1993) Latent autoimmune diabetes mellitus in adults (LADA): the role of antibodies to glutamic acid decarboxylase in diagnosis and prediction of insulin dependency. Diabetic Med. II, 299-303. [22] Christie, M., Landin-Olsson, M., Sundkvist, G., Dahlquist, G., Lernmark, A. and Baekkeskov, S. (1988) Antibodies to a Mr-64 000 islet cell protein in Swedish children with newly diagnosed Type 1 (insulindependent) diabetes. Diabetologia 31, 597-602.
199
[23] DeAizpurua, H.J., Wilson, Y.M. and Harrison, L.C. (1992) Glutamic acid decarboxylase autoantibodies in preclinical insulin-dependent diabetes. Proc. Natl. Acad. Sci. USA 89, 9841-9845. [24] Tuomilehto, J., Zimmet, P., Mackay, I.R. et al. (1994) Antibodies to glutamic acid decarboxylase as predictors of insulin-dependent diabetes mellitus before clinical onset of disease. Lancet 343, 1383-1385. [25] Chen, Q-Y., Rowley, M.J., Byrne, G.C. et al. (1993) Antibodies to glutamic acid decarboxylase in Australian children with insulin-dependent diabetes mellitus and their first-degree relatives. Pediatr. Res. 34, 785-790. [26] Thivolet, C.H., Tappaz, M., Durand, A. et al. (1992) Glutamic acid decarboxylase (GAD) autoantibodies are additional predictive markers of Type I (insulindependent) diabetes mellitus in high risk individuals. Diabetologia 35, 570-576.