Species-Specific Autoantibodies in Type 1 Diabetes*

4 downloads 0 Views 184KB Size Report
GAD65 autoantibodies (GAD65Ab) are important markers for type. 1 (insulin-dependent) diabetes mellitus. Although most patients have. GAD65Ab at the time of ...
0021-972X/99/$03.00/0 Journal of Clinical Endocrinology and Metabolism Copyright © 1999 by The Endocrine Society

Vol. 84, No. 2 Printed in U.S.A.

Species-Specific Autoantibodies in Type 1 Diabetes* ¨ RTQVIST, O. ROLANDSSON, M. LANDIN-OLSSON, C. TO ¨ RN, C. S. HAMPE, E. O Å. ÅGREN, B. PERSSON, D. B. SCHRANZ†, AND Å. LERNMARK Department of Medicine, University of Washington (C.S.H., D.B.S., A.L.), Seattle, Washington 98195; and the Department of Medicine, University Hospital (M.L.-O., C.T.), Lund; the Department of Family Medicine, Umea University (O.R., A.A.), Umea; and the Department of Woman and Child Health, Karolinska Institute (E.O., B.P.), Stockholm, Sweden ABSTRACT GAD65 autoantibodies (GAD65Ab) are important markers for type 1 (insulin-dependent) diabetes mellitus. Although most patients have GAD65Ab at the time of clinical diagnosis, there are also GAD65Abpositive individuals in the population at low risk of developing type 1 diabetes. The aim of this study was to test the hypothesis that the GAD65Ab reactivity to GAD65 cloned from human, mouse, and rat in newly diagnosed type 1 diabetic patients differ from antibody-positive healthy individuals. Sera from 254 new-onset 0- to 34-yr-old type 1 diabetic patients and 270 controls were assayed for their reactivity to

human, mouse, and rat GAD65. Among the type 1 diabetic patients there was a significant better binding of human GAD65 compared to either mouse (P 5 0.03) or rat GAD65 (P 5 0.0005). The preference for human GAD65 increased with increasing age at onset (P 5 0.0002). This differentiation was not observed in 88 GAD65Ab-positive control subjects. Our data indicate that recognition of epitopes by GAD65Ab in type 1 diabetes is different from that in nontype 1 diabetes, GAD65Ab-positive individuals. (J Clin Endocrinol Metab 84: 643– 648, 1999)

I

to be predominately directed toward epitopes located at the middle (amino acids 240 – 435 5 epitope 1 or E1) (14, 15) and the carboxyl-terminal (amino acids 451–570 5 epitope 2 or E2) regions of GAD65 (14, 15). These epitopes were identified by the use of GAD65/GAD67 chimeric proteins or GAD65, modified by site-directed mutagenesis. The use of these molecules may, however, introduce conformational changes and thus eliminate certain epitopes. GAD65 is a highly conserved protein. It has been isolated and characterized in several mammals (human, murine, and rat) (16 –18). To our knowledge, naturally occurring variants of GAD65 have not been used to determine the presence of epitope-specific antiGAD65 Ig. In this study we have taken advantage of the limited sequence difference, previously viewed as insignificant for the binding by type 1 diabetes-associated antiGAD65 IgG (19), among human, rat, and murine GAD65 to determine their ability to bind autoantibodies in new-onset 0- to 35-yr-old type 1 diabetic patients and in antibodypositive nontype 1 diabetes control subjects.

NSULIN-DEPENDENT (type 1) diabetes mellitus is a chronic autoimmune disease. It is characterized by lymphocytic infiltration of the islet of Langerhans (1) associated with a gradual and specific destruction of pancreatic b-cells (2). This process can last several years and eventually results in complete b-cell destruction, hyperglycemia, and life-long insulin dependency. Most type 1 diabetic patients have circulating autoantibodies directed to islet cell autoantigens (3, 4). The main autoantigens identified are insulin (5, 6), the Mr 65,000 isoform of glutamic acid decarboxylase (GAD65) (5, 7), and tyrosine phosphatase IA-2 (8). These autoantibodies are often detected long before the clinical onset of type 1 diabetes and may therefore predict disease (5, 8, 9). In particular, the presence of all three autoantibodies predicts type 1 diabetes among first degree relatives (8, 9). GAD65 autoantibodies (GAD65Ab) tend to be the first to appear several years before the onset of disease (5, 10), with a diagnostic sensitivity of 75– 85% (5, 11) and a diagnostic specificity of 98 –99% (10, 12). GAD65Ab are found in only 1–2% of healthy subjects (13) and may therefore mark that the type 1 diabetes process is present. GAD65Ab do not bind denatured GAD65, GAD65 protein fragments, or synthetic peptides (14, 15), which implies that they bind protein conformation-dependent epitopes. The identification of type 1 diabetes-specific GAD65 epitopes will be of major importance for the value of GAD65Ab as a major predictive marker. GAD65Ab appear

Subjects and Methods Human sera Three groups of type 1 diabetes sera were used in this study (Table 1). The first group represents 10 children who were diagnosed with diabetes at age 7–12 yr (median, 10 yr) and subjected to plasmapheresis (20). These samples have been used in all Immunology of Diabetes Workshops to standardize islet cell autoantibodies (ICA) (21) and GAD65Ab (22). One sample from this set of 10 samples is serving as the worldwide standard for expression of ICA levels in JDF units (21) and of GAD65Ab as a GAD65 antibody index (7). The second group consists of 2- to 18-yr-old newly diagnosed patients (n 5 126) with type 1 diabetes. All of these patients were from the St. Gorans Children Hospital (Stockholm, Sweden) and represent 90% of all children diagnosed at this clinic during 1993–1995. The third group consists of randomly selected (n 5 118), 15- to 35-yr-old newly diagnosed Swedish insulindependent patients. The subjects were registered between 1992–1993 in the Diabetes Incidence Study in Sweden.

Received September 9, 1998. Revision received November 4, 1998. Accepted November 10, 1998. Address all correspondence and requests for reprints to: Dr. Christiane S. Hampe, Department of Medicine, Box 357710, University of Washington, Seattle, Washington 98195. E-mail: champe@u. washington.edu. * This work was supported by the Juvenile Diabetes Foundation International and the NIH (Grants DK-42654, DK-26190, and DK-53004). † Juvenile Diabetes Foundation International Research Fellow.

643

644

JCE & M • 1999 Vol 84 • No 2

HAMPE ET AL.

TABLE 1. Clinical characteristics of type 1 diabetes patients and control subjects Group

1. 2. 3. 4. 5. 6. 7.

Standards Type 1 diabetes Type 1 diabetes Healthy controls Healthy controls Type 2 diabetes Healthy GAD65Ab-positive individuals

n

Mean age at sampling (yr)

Age range (yr)

Male/female ratio

Mean GAD65Ab index

10 126 117 50 40 132 30

10 9 24 12 29 26 47

7–12 2–18 15–35 6–16 19–35 15–35 40– 60

6/4 78/48 74/43 18/32 20/20 74/57 12/18

0.41 0.42 0.62 0.014 0.031 0.41 0.45

Four groups of sera were used as controls (Table 1). The first group (n 5 132) consists of randomly selected, 15- to 35-yr-old Swedish individuals (mean age, 26 yr). The second group (n 5 50) consists of 2- to 18-yr-old healthy Swedish individuals. The third group consisted of type 2 diabetic patients (n 5 58) who were also part of the Diabetes Incidence Study in Sweden study. The fourth group (n 5 30) consists of healthy controls with a GAD65Ab index above the cut-off detected in a population-based screening of 2276 adults (mean age, 47 yr). Of these 30 GAD65Ab-positive individuals, 83% (n 5 25) were normoglycemic, and 13% (n 5 5) had impaired glucose tolerance. These individuals participated in The Va¨sterbotten Intervention Program, Sweden. All serum samples were kept frozen at 280 C as small aliquots for 1–2 yr before analysis. The study was approved by the ethics committee of the Karolinska Institute (Stockholm, Sweden) and Umea University (Umea, Sweden). All individuals gave their informed consent to participate in the study.

Construction of murine and rat GAD65 and of chimeric molecules Full-length murine and rat GAD65 complementary DNA (cDNA; both were provided by Drs. Daniel Kaufman and Alan Tobin, respectively, University of California, Los Angeles, CA) were inserted into the vector pcDNAII (Invitrogen Corp., San Diego, CA) and coded pcKoM215 and pcKoR91, respectively. An additional 14 bp containing the Kozak sequence GGATCCAATTCACC were inserted directly 59 of the coding sequences. The chimerical molecule consisting of the aminoterminal amino acids (aa) of human GAD65 (aa 1– 83)/GAD67 (aa 89 – 593) was constructed as described previously (23). The amino-terminal portion of human GAD65 (aa 1– 83) was substituted by the aminoterminal portion of rat GAD65, using a native PstI site in both cDNA clones.

GAD65 antibody (GAD65Ab) RIA with recombinant human, murine, and rat GAD65 Recombinant [35S]GAD65 antigens were produced in an in vitro coupled transcription/translation system with SP6 ribonucleic acid polymerase and nuclease-treated rabbit reticulocyte lysate (Promega Corp., Madison, WI) as described previously (24). The in vitro translated [35S]GAD65 was kept at 270 C and used within 2 weeks in RIAs. Equal amounts of GAD65 in all three preparations were verified by densitometric analysis of SDS-PAGE. GAD65Ab were determined by a previously described RIA (7, 24). Human serum samples were tested at a final serum dilution of 1:25 unless indicated otherwise. Recombinant human GAD65 (rhGAD65) expressed in and purified from a baculovirus system (BioSyn, Stockholm, Sweden) was used in the GAD65Ab competitive RIA. The RIA was performed as described above, but in the presence of the indicated concentrations of rhGAD65. The intraassay average coefficient of variations was 5.2; the highest value was 20, and the lowest value was 0.1.

FIG. 1. Alignment of human, rat, and mouse GAD65 (pex9, pKoR91, and pKoM215, respectively). The respective amino acid substitutions are indicated. (index of 0.07) of the normal range was established as the 99th percentile of the levels of 182 healthy control subjects. The Juvenile Diabetes Foundation ICA standard (25), which is also GAD65Ab positive, as verified by immunoprecipitation (26), was used as the GAD65 antibody-positive standard. A randomly selected control serum from a healthy volunteer was used as the negative standard. All samples were tested in duplicate, and the coefficient of variations was determined for each sample. Differences in binding to the three GAD65 antigens were evaluated using the nonparametric Mann-Whitney U test. P , 0.05 was considered statistically significant.

Statistical analysis

Results GAD65 from human, mouse, and rat

Antibody levels were expressed as relative indexes using one positive and two negative standard sera, as previously described (7, 24): GAD65Ab index 5 (cpm of tested sample 2 average cpm of two negative standards)/(cpm of positive standard 2 average cpm of two negative standards). Antibody-positive and -negative samples were included in every assay to correct for interassay variation. The upper limit

GAD65 has been isolated and sequenced from several mammalian species including human, mouse, and rat (16 – 18). The sequence comparison in Fig. 1 shows that human GAD65 differs from rat at 22 residues and from mouse GAD65 at 24 residues. The majority of the amino acid sub-

TYPE 1 DIABETES AUTOANTIBODIES

stitutions are located at the first 100 amino acids [54% (7) human/rat and 62% (19) human/mouse]. The substitutions are radical in 83% (10 of 12 for human/rat) or 86% (13 of 15 for human/mouse), such as proline to serine. The mouse GAD65 differs from rat at 8 residues. Immunoreactivity of the three GAD65 species to standard ICA sera

The immunoreactivity of the different GAD65 species was first tested in RIAs using the ICA standard sera (Fig. 2). The sera were tested at different dilutions (1:25–1:1000) with all three radiolabeled antigens. Sera from two healthy individuals were used as negative controls. GAD65 antibodies were detected in eight of these sera (Table 2). All of the GAD65Abpositive sera recognized all three GAD65 species. Five of them (sera 4, 5, 6, and 8) clearly differentiated between human and rodent GAD65. All of them immunoprecipitated higher amounts of human than of rodent GAD65. The difference in antigen precipitation was observed at all serum dilutions except the very high dilutions, where the frequency of GAD65Ab decreased and gradually overlapped with the range of immunoprecipitation of the healthy control sera. Serum 2 differentiated between human and rodent GAD65 to some extent, whereas sera 1 and 7 did not distinguish among the three antigens. Sera 9 and 10 were both negative for GAD65Ab. Serum 9 is included in the figure as an example of a negative serum. Displacement of human, mouse, and rat GAD65 by rhGAD65

Sera 7 and 8 were analyzed at a dilution of 1:50 for displacement with unlabeled rhGAD65 (Fig. 3). Serum 7 did not show significant differences in the binding of the three GAD65 species (Fig. 2), and the anti-GAD65 IgG was dis-

645

TABLE 2. Binding of eight positive type 1 diabetes standard sera at 1;50 serum dilution Serum no.

1 2 3 4 5 6 7 8

% binding to GAD65 Human

Mouse

Rat

100 100 100 100 100 100 100 100

90 93 78 43 62 50 106 47

90 86 78 43 62 52 118 53

The GAD65Ab indexes were normalized to binding to human GAD65.

FIG. 3. Displacement of human, mouse, and rat GAD65 with rhGAD65. Binding of human (M), mouse (f), and rat (l) GAD65 to serum 8 (1) and serum 7 (2) in Fig. 1 was displaced with rhGAD65. The mean 6 SD are indicated.

placed equally well from all three antigens with increasing concentrations of rhGAD65 (Fig. 3). Serum 8 displayed significant differences in the binding of the three GAD65 species (Fig. 2) and showed different displacement curves (Fig. 3). Although human GAD65 was displaced completely already at 60 ng/mL rhGAD65, both rodent antigens were displaced to a lesser extent at this concentration. Mouse GAD65 was displaced completely by 100 ng/mL rhGAD65, whereas rat GAD65 was still not displaced entirely at this concentration. GAD65Ab reactivity to human, murine, and rat GAD65 in two groups of type 1 diabetic patients, GAD65Ab-positive type 2 diabetic patients, and healthy controls

FIG. 2. Binding of human, mouse, and rat GAD65 by standard sera. Binding of [35S]methionine-labeled human (M), mouse (f), and rat (l) GAD65 to standard sera was measured in RIA. The GAD65Ab index for sera 1–3 is given on the left of panel 1; the GAD65Ab index scale for sera 4 –9 is given on the left of panel 4. Dilution curves of eight GAD65Ab-positive standard sera (1– 8) are shown. Panel 9 shows a GAD65Ab-negative standard serum. The mean 6 SD are indicated.

Sera from two groups of type 1 diabetic patients were tested for their immunoreactivity with the three antigens. The patients had developed diabetes either at 2–18 yr of age or at 15–34 yr of age. Both groups (244 patients) showed a clear preference in binding human GAD65 compared to either mouse (P 5 0.03) or rat GAD65 (P 5 0.002; Fig. 4). In the patients who developed type 1 diabetes at 15–34 yr of age, both preferences were more pronounced (P 5 0.03 and P 5 0.0005) than in the group of patients who developed the disease at a younger age (2–18 yr; both P 5 0.05). There were no significant differences between binding to rat or mouse GAD65 in either patient group (P 5 0.07). The two healthy control groups (total n 5 182) matching the two above patient groups showed no binding to any of the antigens above the cut-off value. In analyzing the third control group (n 5 58) consisting of GAD65Ab-positive type 2 diabetic patients and the fourth group (n 5 30) consisting of healthy individuals with GAD65Ab, we made the surprising observation that

646

JCE & M • 1999 Vol 84 • No 2

HAMPE ET AL.

FIG. 4. Difference between GAD65Ab index of human GAD65 and 1) mouse GAD65 in 2- to 18-yr-old type 1 diabetic patients, 2) rat GAD65 in 2- to 18-yr-old type 1 diabetic patients, and 3) mouse GAD65 in 15to 34-yr-old type 1 diabetic patients, and 4) rat GAD65 in 15- to 34-yr-old type 1 diabetic patients.

there was no preference in binding between human and either rat or mouse GAD65 (P 5 0.38; Fig. 5). Differentiation between human and rat GAD65 is age dependent

The above observation of age-dependent preference for human over rodent GAD65 was further examined by plotting the differences in GAD indexes (GAD index human 2 GAD index rat) to the age at diagnosis. The data (Fig. 6) demonstrate that the difference between human and rat GAD65Ab indexes increases with increasing age at diagnosis (P 5 0.0002). Type 1 diabetic patients do not recognize epitopes in the Nterminus of GAD65

We next studied binding of the GAD65Ab of type 1 diabetic patients to the N-terminus of the GAD65 molecule. Therefore, we employed a chimeric molecule constructed by substituting the first 83 amino acids of rat GAD67 with the respective sequence of human GAD65 only. Thirty of 202 (15%) sera from type 1 diabetic patients bound to the chimera, whereas in type 2 diabetic patients and healthy GAD65Ab-positive individuals, 9 of 29 (33%) and 18 of 39 (46%) sera bound, respectively (P 5 0.024 and 0.0036, respectively). The majority of the binding was due to cross-

FIG. 5. Difference between GAD65Ab index of human GAD65 and 1) mouse GAD65 in healthy GAD65Ab-positive individuals, 2) rat GAD65 in healthy GAD65Ab-positive individuals, 3) mouse GAD65 in GAD65Ab-positive type 2 diabetic patients, and 4) rat GAD65 in GAD65Ab-positive type 2 diabetic patients.

reactivity with GAD65, as only 7 of 155 (4%) sera from the type 1 diabetic patients bound to the chimera and not to GAD67 (data not shown), whereas 4 of 23 (17%) sera from the type 2 diabetic patients and 6 of 28 (21%) sera from the GAD65Ab-positive individuals bound only to the chimera and not to GAD67 (P 5 0.01 and 0.001, respectively), indicating that these sera recognize epitopes in the N-terminus of the molecule. The data confirm that the N-terminal end of GAD65 does not have an important epitope for antibody binding in type 1 diabetic patients. Discussion

The identification of disease-specific epitopes is of major importance for the prediction of type 1 diabetes. In most studies only the presence or absence of GAD65Ab is analyzed (12, 27). Several studies of new-onset patients and controls have attempted to define diagnostic sensitivity and specificity (9, 11, 13, 23). The cut-off for positivity has been arbitrarily estimated by receiver operating characteristics analysis (28), percentiles, or mean 1 3 sd (13). However, less attention has been paid to levels of GAD65Ab as a potential factor for diabetes risk (3, 8). Conformation-specific GAD65Ab to predict type 1 diabetes have been used only in

TYPE 1 DIABETES AUTOANTIBODIES

FIG. 6. Plot of differences between the human and rat GAD65Ab indexes and age at onset. The dotted line indicates the division of the two age groups tested.

one study (23). We here tested the hypothesis that IgG antiGAD65 in type 1 diabetic patients are species specific. Our GAD65Ab RIA with protein A detects primarily IgG (except IgG3) and not antibodies of other isotypes. Furthermore, GAD65-specific IgM was not detected in the sera tested here (Schranz, D. B., personal communication). The use of GAD65 cloned from human, mouse, and rat guarantees preservation of conformational intact GAD65. Our major findings in testing 254 type 1 diabetic sera are that GAD65Ab showed significantly preferred binding to human GAD65 compared to both rodent GAD65 species. This observation may seem obvious, but it has been claimed that GAD65Ab in type 1 diabetic patients do not distinguish GAD65 species differences (19). The preference for human GAD65 is independent of the GAD65Ab index and seems to be age related, as it was more prominent in the patient group who developed type 1 diabetes at an older age (18–35 yr) than in those who developed the disease at a younger age (2–18 yr). It is also noted that 8% (11 of 126) of the young type 1 diabetic patients showed significant preference for rodent GAD65. Studies to determine whether this pattern changes over time or is stable are currently being conducted. GAD65Ab are important markers in the prediction of type 1 diabetes. The GAD65Ab-specific epitopes are believed to be conformational (7, 15). Previous studies with chimeric GAD65/67 have failed to identify epitopes located at the Nterminus (23). These results were confirmed in this study (data not shown). The N-terminal portion of the molecule carries most of the amino acid substitutions among the three species. The remaining molecule is 98% identical between both rodent and human. Four of the amino acid substitutions observed in the three antigens involve proline (amino acid positions 19, 62, 63, and 83). These proline substitutions may have a major influence on conformation and explain the differences in binding.

647

The conformational differences in rodent GAD65 compared to human GAD65 could hinder the accessibility of antibodies to bind to one or both epitopes. The epitopes in type 1 diabetic patients’ sera and sera from healthy controls are identical, but patients’ sera bind significantly better to the C-terminal epitope (23). We speculate that the molecular folding typical for rodent GAD65 may hinder antibody binding to the C-terminal epitope. Studies involving different chimeric molecules are currently underway in our laboratory. Our study shows that sera from type 1 diabetic patients can differentiate between GAD65 species. Only patients’ GAD65Ab differentiate between species, whereas GAD65Ab found in GAD65Ab-positive healthy individuals (n 5 30) and type 2 diabetic patients (n 5 58) do not. Therefore, GAD65Ab in these two control groups are more alike than those found in type 1 diabetic patients. This indicates that GAD65Ab-positive type 1 diabetic patients recognize different epitopes from those recognized in these two GAD65Ab-positive control groups. Although GAD65Ab of both healthy GAD65Ab-positive individuals and type 2 diabetic patients show a broad immune response to GAD65, as shown by equal binding of all three isoforms, GAD65Ab in newly diagnosed type 1 diabetic patients represent a more specific subgroup of antibodies, as they preferentially recognize human GAD65. Epitope analysis as a measure of type 1 diabetes prediction may therefore be critical and useful. The comparison of GAD65Ab binding to human and rodent GAD65 may increase the predictive value for type 1 diabetes and broaden our understanding of the underlying autoimmune process. Acknowledgments We thank Terri Daniels for excellent technical assistance. The samples from the 15- to 34-yr-old new-onset patients were randomly selected from the Diabetes Incidence Study in Sweden, a population-based in¨ stman, Hans J. Arnqvist, Go¨ran vestigation coordinated by Jan O Blohme`, Folke Lithner, Bengt Littorin, Lennarth Nystra`m, Ga`ran Sundkvist, and Lars Wibell.

References 1. Gepts W. 1965 Pathologic anatomy of the pancreas in juvenile diabetes mellitus. Diabetes. 14:619 – 633. 2. Lernmark Å, Kla`ppel G, Stenger D, et al. 1995 Heterogeneity of human islet pathology in newly diagnosed childhood insulin-dependent diabetes mellitus. Macrophage infiltrations and expression of HLA-DQ and glutamic acid decarboxylase. Virchows Arch. 425:631– 640. 3. Bonifacio E, Genovese S, Braghi S, et al. 1995 Islet autoantibody markers in IDDM: risk assessment strategies yielding high sensitivity. Diabetologia. 38:816 – 822. 4. Landin-Olsson M, Karlsson A, Dahlquist G, Blom L, Lernmark Å, Sundkvist G. 1989 Islet cell and other organ-specific autoantibodies in all children developing type 1 (insulin-independent) diabetes mellitus in Sweden during one year and in matched controls. Diabetologia. 32:387–395. 5. Hagopian WA, Sanjeevi CB, Kockum I, et al. 1995 Glutamate decarboxylase-, insulin- and islet cell-antibodies and HLA typing to detect diabetes in a general population-based study of Swedish children. J Clin Invest. 95:1505–1511. 6. Wilkin T, Armitage M, Casey C, et al. 1985 Value of insulin autoantibodies as serum markers for insulin-dependent diabetes mellitus. Lancet. 480 – 482. 7. Grubin CE, Daniels T, Toivola B, et al. 1994 A novel radioligand binding assay to determine diagnostic accuracy of isoform-specific glutamic acid decarboxylase antibodies in childhood IDDM. Diabetologia. 37:344 –350. 8. Verge CF, Gianani R, Kawasaki E, et al. 1996 Prediction of type I diabetes in first-degree relatives using a combination of insulin, GAD, and ICA512bdc/ IA-2 autoantibodies. Diabetes. 45:926 –933. 9. Chaillous L, Delamaire M, Elmansour A, et al. 1994 Combined analysis of islet cell antibodies which cross-react with mouse pancreas, antibodies to the

648

10. 11. 12. 13.

14. 15. 16. 17. 18. 19.

HAMPE ET AL.

Mr64,000 islet protein, and antibodies to glutamate decarboxylase in subjects at risk for IDDM. Diabetologia. 37:491– 499. Seissler J, Amann J, Mauch L, et al. 1993 Prevalence of autoantibodies to the 65- and 67-kD isoforms of glutamate decarboxylase in insulin-dependent diabetes mellitus. J Clin Invest. 92:1394 –1399. Vandewalle CL, Falorni A, Svanholm S, et al. 1995 High diagnostic sensitivity of glutamate decarboxylase autoantibodies in IDDM with clinical onset between age 20 and 40 years. J Clin Endocrinol Metab. 80:846 – 851. Roll U, Christie MR, Standl E, Ziegler A-G. 1994 Association of anti-GAD antibodies with islet cell antibodies and insulin autoantibodies in first-degree relatives of Type 1 diabetic patients. Diabetes. 43:154 –160. Landin-Olsson M, Palmer JP, Lernmark Å, et al. 1992 Predictive value of islet cell and insulin autoantibodies for type 1 (insulin-dependent) diabetes mellitus in a population-based study of newly-diagnosed diabetic and matched control children. Diabetologia. 35:1068 –1073. Daw K, Ujihara N, Powers A. 1996 Glutamic acid decarboxylase autoantibodies in stiff-man syndrome and insulin-dependent diabetes mellitus exhibit similarities and differences in epitope recognition. J Immunol. 156:818 – 825. Ujihara N, Daw K, Gianani R, Boel E, L. Yu L, Powers AC. 1994 Identification of glutamic acid decarboxylase autoantibody heterogeneity and epitope regions in type I diabetes. Diabetes. 43:968 –975. Bu D-F, Erlander MG, Hitz BC, et al. 1992 Two human glutamate decarboxylases, 65-kDa GAD and 67-kDa GAD, are each encoded by a single gene. Proc Natl Acad Sci USA. 89:2115–2119. Erlander MG, Tillakaratne NJK, Feldblum S, Patel N, Tobin AJ. 1991 Two genes encode distinct glutamate decarboxylase. Neuron. 7:91–100. Lee DS, Tian J, Phan T, Kaufman DL. 1993 Cloning and sequence analysis of a murine cDNA encoding glutamate decarboxylase (GAD65). Biochim Biophys Acta. 1216:157–160. Kaufman DJ, Erlander MG, Clare-Salzler M, Atkinson MA, Maclaren NK,

20.

21.

22.

23.

24.

25.

26.

27.

28.

JCE & M • 1999 Vol 84 • No 2

Tobin AJ. 1992 Autoimmunity to two forms of glutamate decarboxylase in insulin-dependent diabetes mellitus. J Clin Invest. 89:283–292. Marner B, Lernmark Å, Ludvigsson J, et al. 1985 Islet cell antibodies in insulin-dependent (type 1) diabetic children treated with plasmapheresis. Diabetes Res. 2:231–236. Bonifacio E, Dawkins RL, Lernmark Å. 1987 Immunology and diabetes workshops: report of the second international workshop on the standardization of cytoplasmic islet cell antibodies. Diabetologia. 30:273. Schmidli RS, Colman PG, Bonifacio E, Bottazzo GF, Harrison LC. 1994 High level of concordance between assays for glutamic acid decarboxylase antibodies. The First International Glutamic Acid Decarboxylase Antibody Workshop. Diabetes. 43:1005–1009. Falorni A, Ackefors M, Carlberg C, et al. 1996 Diagnostic sensitivity of immunodominant epitopes of glutamic acid decarboxylase (GAD65) autoantibodies epitopes in childhood IDDM. Diabetologia. 39:1091–1098. ¨ rtqvist E, Persson B, Lernmark Å. 1995 Radioimmunoassays for Falorni A, O glutamic acid decarboxylase (GAD65) and GAD65 autoantibodies using 35S or 3 H recombinant human ligands. J Immunol Methods. 186:89 –99. Bonifacio E, Lernmark Å, Dawkins RL. 1988 Serum exchange and use of dilutions have improved precision of measurement of islet cell antibodies. J Immunol Methods. 106:83– 88. Hagopian WA, Michelsen B, Karlsen AE, et al. 1993 Autoantibodies in IDDM primarily recognize the 65,000-Mr rather than the 67,000-Mr isoform of glutamic acid decarboxylase. Diabetes. 42:631– 636. Atkinson M, Kaufman D, Newman D, Tobin A, Maclaren N. 1993 Islet cell cytoplasmic autoantibody reactivity to glutamate decarboxylase in insulindependent diabetes. J Clin Invest. 91:350 –356. Zweig M, Campbell G. 1993 Receiver-operating characteristics (ROC) plots: a fundamental tool in clinical medicine. Clin Chem. 39:561–577.