Validity of Screening for Individuals at Risk for Type I Diabetes by Combined Analysis ot Antibodies to Recombinant Proteins

Transcription

1 E p i d e m i o I o g y H e a h S e r v i c e s P s y c h o s o c i a I N A L A R T I C L E Research Validity of Screening for Individuals at Risk for Type I Diabetes by Combined Analysis ot Antibodies to Recombinant Proteins ICHAEL R. CHRISTIE, PHD URSULA ROLL, PHD ARK A. PAYTON, PHD EA C.I. HATIELD, CCP ANETTE-G. ZIEGLER, D OBJECTIVE To determine whether screening for the presence of multiple antibody markers for DD is effective at identifying individuals with high risk for disease development. RESEARCH DESIGN AND ETHODS Antibodies to GAD and the tyrosine phosphatase-like protein IA- were determined in sequential serum samples from first-degreerelatives of IDD patients, identified as possessing islet cell antibody (ICA) andor insulin autoantibody (), who were followed prospectively for IDD development. ICA,, and antibodies to GAD and IA- were also determined in 93 cases of new-onset nonfamilial IDD. RESULTS The presence of two or more antibodies in addition to ICA or conferred high risk (6%) for development of IDD within years of entry into the study and identified 89% of those who have developed IDD on current follow-up. None of the relatives positive for ICA or alone, in the absence of other antibody markers, have developed IDD. Antibodies to islet antigens could both appear and disappear in follow-up samples obtained after entry into the study. The majority (6%) of young (<6 years), sporadic cases of IDD had multiple antibodies to islet antigens, but this proportion was lower in older patients (37%). CONCLUSIONS A screening strategy based on the analysis of antibodies to multiple islet antigens can predict IDD at high sensitivity and specificity insfamilies, and such a strategy may also be applicable to identify young individuals in the general population with high disease risk. Since appearance of antibodies to different antigens occurs sequentially rather than simultaneously, accurate assessment of diabetes risk based on the presence of multiple antibodies will require follow-up over a number of years after the first evidence of islet autoimmunity. The screening for individuals at risk for IDD has until recently relied almost exclusively on the detection of islet cell antibodies (ICA) measured on frozen sections of pancreas (). Although ICA is a sensitive marker for diabetes in both family studies and in the general population, these antibodies can develop in individuals with low risk for the disease (-). A number of specific molecular targets for autoantibodies in IDD have now been identified (), including GAD and a tyrosine phosphatase-like protein designated IA- (or ICA), the latter protein being originally detected as -kda tryptic fragments of an islet 6-kDa antigen (6,7). Bingley et al. (), by measuring antibodies immunoprecipitating GAD enzyme activity from brain extracts and antibodies to -kda tryptic fragments, showed that secondary screen- rom the Department of edicine (.R.C., AR, E.C.I.H.), King's College School of edicine, London, U.K.; and the Diabetes Research Institute (U.R., A.-G. Z.), Academic Hospital unchen-schwabing, Germany. Address correspondence and reprint requests to ichael R. Christie, King's College School of edicine, Department of edicine, Bessemer Road, London SE 9PJ, U.K. m.christie@kcl.ac.uk. Received for publication July 996 and accepted in revised form January 997. Boehringer annheim has provided consulting fees and funds to.r.c.'s laboratory in connection with development of assays for antibodies to islet antigens in IDD., insulin autoantibody; IA-, tyrosine phosphatase-like protein; IA-A, tyrosine phosphatase-like protein antibody; ICA, islet cell antibody; JD U, Juvenile Diabetes oundation units. ing of ICA-positive individuals for these antibodies together with insulin autoantibodies (s) identified a group of relatives with high diabetes risk (close to 9%) at high sensitivity (>7%). The development of antibody assays based on recombinant proteins may improve the performance of such a screening strategy. The aim of this study was to assess the validity of combined analysis of antibodies to islet antigens by studying their distribution in firstdegree relatives followed prospectively for diabetes development, as well as in sporadic cases of IDD, and to determine the course of antibody development in firstdegree relatives over time. RESEARCH DESIGN AND ETHODS irst-degree relatives of IDD patients irst-degree relatives of IDD patients attending the Academic Hospital unchen- Schwabing, Germany, are routinely invited to participate in a screening program for diabetes-associated antibodies. Since 989,,6 relatives have been screened for the presence of ICA and, and 3 (3.6 %) have been identified to have at least one detectable antibody (ICA or ) confirmed by at least two consecutive serum samples (8). Two of these 3 relatives were identified to already have overt IDD, and 7 were entered into a prophylactic insulin trial (9). In this study, sequential serum samples from the remaining nondiabetic relatives were tested for antibodies to IA- and GAD. All antibody-positive relatives (mean age ± SD,.9 ± 3. years; range years; median.6 years; see Table ) were followed for diabetes onset at 6-month intervals. Diagnosis of diabetes was determined by oral glucose tolerance test (-h blood glucose > mg%). Newly diagnosed IDD patients Serum samples were obtained from 93 patients with newly diagnosed IDD (mean age at onset ± SD of 6.8 ±. DIABETES CARE, VOLUE, NUBER 6, JUNE

2 Antibodies to antigens in prediabetes Table Characteristics ofica- or -positive relatives who developed IDD during follow-up Age at entry (years) Sex Patient with IDD HLA-DR Years pre-idd CA (JD U) Antibodies in first sample (numl) GAD antibody (units) IA-A (units) ather Parents other ather ather ather ather* Twin ather ather other other * () <, ,7 7 78? () 87 9 t (3) 37T (7) T(3) * also has IDD. Positive antibodies are given in bold. Also indicated are cases where relatives gained antibodies (t) or lost antibodies (f) during follow-up at ages (years) indicated in parentheses. HLA-DR serotypes are indicated where known. No association between HLA-DR and antibodies was detected. years; range.-.3 years), who were hospitalized at the Third edical Department of the Academic Hospital unchen- Schwabing. Serum was taken at the day of hospitalization before exogenous insulin therapy was initiated. Antibodies to IA- and GAD Antibodies to IA- (IA-A) and GAD were analyzed by radioligand binding assays. cdna representing the cytosolic domain (amino acids ; IA-ic) of IA- was cloned into the pge-z in vitro transcription vector (7) and cdna for GAD6 cdna in the vector pb88 (gift of Dr. Thomas Dyrberg, Novo Nordisk, Denmark) (). IA-ic and GAD6 were transcribed and translated in vitro using commercial kits (Promega, adison, WI) in the presence of [3S]-methionine (Amersham, Amersham, U.K.). Incorporated radioactivity was determined by precipitation with % trichloroacetic acid and scintillation counting. Aliquots ( pi) containing, cpm of in vitro translated protein in immunoprecipitation buffer ( mmol HEPES, ph 7., mmol NaCl,.% Triton X-, mmol benzamidine,. mgml bovine serum albumin, and mmol methionine) were incubated with pi of test serum for h at C. Immune complexes were isolated on pi of protein A-Sepharose, and immunoprecipitates were washed extensively with immunoprecipitation buffer and once with water. Precipitates were resuspended in pi of water, and radioactivity in pi aliquots was determined by liquid scintillation counting. Radioactivity precipitated was expressed relative to an antibody-positive serum, arbitrarily assigned a value of units. Sera were regarded positive if antibody levels were above the 99th percentile of antibody levels in sera from 7 healthy subjects (aged -3 years) with no family history of IDD, of whom 37 were below the age of years. The thresholds for positivity for antibodies to were units for IA- and 3 units for GAD antibody. The GAD antibody assay was entered into the nd GAD antibody workshop () and achieved a sensitivity of 87.7% and a specificity of 96.8%. The interassay coefficients of variation around the threshold of positivity were 9.9% for GAD antibody and 7% for IA-A. easurement of and ICA Sera were tested for as described previously using a competitive fluid phase radiobinding assay and expressed in numl of insulin precipitated (,3). Sera were regarded as positive if above the 99th percentile of normal control subjects (cutoff at numl). The interassay coefficient of variation around the threshold was.3%. The assay achieved % consistency, % specificity, 86% sensitivity, and 9% validity in the 3rd proficiency test. Undiluted sera were screened for conventional ICA-IgG by means of indirect immunofluorescence on cryostat sections of blood group O human pancreas (). Positive samples were then titered by doubling dilutions in phosphate-buffered saline. End-point titers were converted to Juvenile Diabetes oundation units (JD U), and samples were considered positive for ICA at > JD U. The results of the assay in the 9th ICA proficiency test were 78% consistency, % specificity, 67% sensitivity, and 78% validity Statistical analysis The significance of differences of antibody frequencies between groups was determined by x analysis. The cumulative risk and confidence intervals were evaluated with life-table analysis (statistical package: SPSS, Chicago, IL). Differences were considered significant if P <.. RESULTS Antibodies to islet antigens and development of diabetes in nondiabetic first-degree relatives of IDD patients Since 989, first-degree relatives of IDD patients attending our unich clinic have 966 DIABETES CARE, VOLUE, NUBER 6, JUNE 997

3 Christie and Associates Table Characteristics oflca- or laa-positive relatives without development ofldd during follow-up Age at entry (years) Sex Patient with IDD HLA-DR ollow-up (years) CA (JDU) Antibodies at entry (numl) GAD antibody (units) A-A (units) ather ather ather ather other other other Twin other ather ather ather other other ather Twin ather Daughter Son <* (3) <* (3) <* () t () < T (8) < 8 < < 8 < < 3 8 <* (7) 6 n = 6. Positive antibodies are given in bold. Also indicated are cases where relatives gained antibodies ( (years) in parentheses. 7 8t () * (8) 6t (9) t (7) 8 3 T (7) 7 * () * (3) 6* (7) t (8) * () 3 * () * (6) 9 6t (7) 9 8 ) or lost antibodies (t) during follow-up at ages indicated been screened for the presence of ICA and. To determine whether the additional detection of IA-A or GAD antibody is useful for the identification of relatives most at risk for IDD, relatives with ICA andor (8ICA+", ICA-+, and ICA + + ), who have been followed for later development of diabetes (median follow-up 3.8 years, range.-7. years), were selected for further study Details of these antibody-positive family members, including levels of antibodies at entry, are listed in Tables and. To date, 8 (%) ICAor -positive first-degree relatives of IDD subjects have developed overt IDD. The group of relatives who progressed to IDD did not differ significantly in their age at entry from those who have not developed the disease. The cumulative risk for IDD depending on their initial ICA and status was determined by life-table analysis. or ICA + ~ relatives, the risk for development of IDD within years was 8.% (9% CI ); for ICA" + relatives, it was 9.% (9% CI -6.); and for relatives positive for both ICA and, it was 68.% (9% CI , data not shown). Of the selected relatives, (.%) were positive for IA-A at entry into the study, of whom have developed IDD (Table ). An additional three relatives turned from IA-A-negative to positive on follow-up at 3,, and 6 years of age (marked with an asterisk in Tables and ), and the youngest of these has developed IDD. Thirty-one (7.%) of the relatives were positive for GAD antibody, and of these have developed diabetes. The cumulative risk of IDD for IA-A and GAD antibody in the population was determined by life-table analysis (ig. ). or relatives positive for IA-A, the risk of developing IDD within years was 6.3%, compared with 3.% for those negative for these antibodies (9% CIs % and -6.8%, respectively; P <. for association with IDD). The same figures for GAD antibody-positive and -negative relatives were.8 and 3.9%, respectively (9% CIs % and.3-7.%, respectively; not significant). The cumulative risk by life-table analysis on combining all antibodies is shown in ig.. The -year cumulative risk for diabetes was significantly increased (P <.) in relatives positive for two or more antibody specificities in addition to ICA or (6.%; 9% CI.-8.%) compared with individuals positive for ICA or alone, none of whom have developed IDD. Antibodies to islet antigens in sequential samples from first-degree relatives ultiple serum samples (>) were available from 36 of the relatives. The mean number of samples analyzed per relative was., obtained over a mean period (± SE) of 3. ±. years. In the majority of these cases, antibody levels showed little fluctuation over the period followed. However, individuals, 3 of whom developed IDD, demonstrated conversion of one or more antibody specificities from negative to positive (see Tables and and ig. 3). Of these, four became ICA-positive (e.g., igs. 3A and 3B), four developed (ig. 3D), DIABETES CARE, VOLUE, NUBER 6, JUNE

4 Antibodies to antigens in prediabetes % IDD-free survival ollow up (years) GADA GADA % IDD-free B survival l oo IA-A- IA-A+ ollow up (years) igure Life-table analysis of cumulative risk of type diabetes within years in first-degree relatives selected as positive for ICA andor in relation to the additional presence of antibodies to GAD (GADA) (A) or IA- (B). A significant difference in -year cumulative risk for diabetes was observed between IA- antibody-positive and -negative subjects (P <.). three GAD antibody (ig. 3A), and three IA- antibodies (igs. 3B and 3C); some individuals developed two or more antibodies during follow-up (e.g., igs. 3A and 3B). The majority (7%) of the antibody conversions occurred before the age of 6, but antibody conversions were observed in one relative between the ages of and 7 years. In eight cases, loss of antibody positivity was observed (Tables and and igs. 3C, 3E, and 3). In particular, one relative became negative for LAA and had decreasing levels of GAD antibody and ICA before developing diabetes at the age of (ig. 3E). A second relative lost positivity for, GAD antibody, and A-A between the ages of 6 and 9 years and currently remains nondiabetic (ig. 3). Antibodies to IA- and other antigens in recent-onset IDD patients The frequencies of ICA and antibodies to insulin, IA-, and GAD were determined in a representative sample of 93 newly diagnosed diabetic patients admitted to the Third edical Department of the Academic Hospital unchen-schwabing. Of these patients, 6 (69%) were positive for ICA, (3%) for, (%) for IA-A, and 8 (%) for GAD antibody. Of the 63 ICA-positive patients, (87%) were positive for at least one of the antibodies to specific antigens (insulin, GAD, or IA-), whereas only 6 (3%) of the 3 ICA-negative patients were positive. The distribution of antibodies to specific antigens in young (<6 years) and older patients is illustrated in ig. The frequency of was significantly higher in patients diagnosed before the age of 6 (6%) than in older patients (6%; P =.). urthermore, a greater proportion of the younger patients (6%) were positive for three or more antibodies (including ICA) compared with the older patients (37%; P =.7). CONCLUSIONS Clinical trials to test preventative treatments for IDD in first-degree relatives of patients with IDD currently use ICA as a screen for individuals with high disease risk. The description of novel antibodies to specific islet cell proteins has prompted studies to determine whether one or more of these can supplement or replace ICA in the screening for individuals with high disease risk. By analyzing antibodies to a number of specific islet cell proteins, Bingley et al. (), snowed that antibodies to 37k islet antigens, now identified as tyrosine phosphatase-like proteins related to IA- (7,,), are closely associated with rapid progression to IDD. However, highest sensitivity and specificity for disease prediction was achieved by testing for the presence of two or more antibodies in addition to ICA. Since 989, we have routinely screened relatives of IDD patients for the presence of both ICA and. ICA- or -positive relatives have subsequently been followed closely for diabetes development and all relatives were asked at the time of screening to inform us in case of disease development. As far as we are aware, all cases of IDD have been positive for either ICA or, demonstrating the sensitivity of the screening procedure. However, not all ICA positive individuals develop diabetes (-). In the cohort of relatives screened in this study, ICA was shown to confer a 8% risk of developing IDD within years and a 6% risk (6). To further determine the value of multiple antibody testing in disease prediction, we have now analyzed sequential samples from these ICA- or -positive relatives for the % IDD-free survival -. Ab+ Ab+ 9 Ab+ 3 >Ab ollow up (years) o Ab+ igure Life-table analysis of cumulative risk of IDD within years in first-degree relatives selected as positive for ICA andor in relation to numbers of additional positive antibodies. A significantly higher -year cumulative risk for IDD was observed for relatives positive for 3 or more antibodies compared to relatives positive for ICA or alone (P <.). presence of antibodies to GAD and IA- using sensitive radioligand binding assays. Life-table analysis confirmed the high specificity of the IA- antibody test for disease prediction. Thus, IA- antibody positivity conferred a 67.% cumulative risk for disease development within years and these antibodies were present in 8 (78%) of the relatives who have become diabetic on follow-up. urthermore, the results demonstrate the advantage of screening for multiple antibodies, since all but two of the relatives (89%) who progressed to IDD were positive for 3 or more antibody specificities, and these multiple antibodies conferred a 67% risk of disease progression. Similar results have been reported by Verge etal. (7). Estimates of diabetes risks for possession of individual antibodies, or combinations of antibodies, were calculated according to antibody status in the first available sample for each relative. The determination of antibodies in sequential samples from nondiabetic relatives also allowed an analysis of antibody progression over time. A recent prospective study by our group on offspring of diabetic parents followed from birth (8) showed that, GAD antibody, and IA-A can all appear early in life, commonly between the ages of two and five. These antibodies appeared sequentially during this period, although no consistent order of appearance of the antibodies could be demonstrated. In the present study, even with quite short periods of 968 DIABETES CARE, VOLUE, NUBER 6, JUNE 997

5 Christie and Associates follow-up (mean of 3.8 years), conversion of specific antibodies from negative to positive was observed, further illustrating the sequential nature of the appearance of antibodies to different islet antigens. ost cases of conversions to antibody positivity occurred at a young age (76% before the age of 6), although appearance of antibodies was detected at older ages. In general, once antibodies appeared they were maintained at relatively constant levels during the follow-up period. However, in a number of cases antibody levels decreased or disappeared with time. Loss of antibody positivity did not necessarily indicate cessation of the destructive process against the pancreatic beta cells, since one individual both lost AA and had decreasing levels of ICA and GAD antibody during a 3-year period preceding the onset of clinical IDD (ig. 3E). These observations have important implications for the timing and frequency of antibody screening. The early appearance of antibodies indicates that screening can be initiated at a young age. However, since antibodies to islet antigens can both appear and disappear over time, assessment of diabetes risk according to the possession of specific antibodies, or antibody combinations, may be highly dependent on the time of sampling. Effective screening strategies involving multiple antibody analysis are likely to require determination of antibodies in additional samples after thefirstdetection of diabetes-associated immune markers. The ultimate goal of research in this field is to devise a screening strategy that is applicable to the general population. Preliminary studies have suggested that approaches for screening designed on the basis of observations in family studies are also useful in the general population. Thus, in a study of almost 3, schoolchildren in the U.K. (9), antibodies to 37kD antigen were detected in all 3 schoolchildren who developed diabetes during a 3.- year follow-up, confirming the association of these antibodies with rapid disease progression, and all had or more of the antibodies tested. However, the number and type of antibodies detected in sporadic diabetic patients is influenced by the age of diagnosis. In IDD patients studied here, 63% of IDD patients who developed diabetes at a young age (<6 years) were positive for antibodies to IA-A and a similar number were positive for antibodies to two or more antigens. In patients 6 years or older at onset, while the prevalence of antibodies to GAD was similar to that in younger patients A (nuanl) 6 (nlnl) ' 3 Neg ICA (JD unit*) AOADA IA-A Age(yrs) ICA (JD unite) \ h V' Age(Yre) ICA (JD unit*) 9.3 Age(Yr») \ IDD QADAIA-A 3 <""» > (numi) QADAIA-A (Unite) 8 6 (nilml) 3 ( ' QADA-A (Unite) (numl) \ L 3 Age(yre) ICA (JD units) I \ T^^ r IA-A QADA "So Ago(Yre) QAAIA-A (Unite) QADAIA-A (Unit*) igure 3 Analysis of antibodies in sequential samples from initially nondiabetic first-degree relatives selected as positive for ICA andor. Levels of (circles), GAD antibody (GADA) (triangles), and IA- (squares) are presented graphically for each of six representative relatives and data for ICA are given at the top of each panel. Open symbols indicate that the sample was considered negative for that particular antibody, closed symbols are positive. Note that the scales for antibody levels are not the same for all patients. GADA Age<6 () 3(3) IA-A ICA only: n= No antibodies: n= GADA Age ICA only: n= No antibodies: n= igure A Distribution of antibodies to insulin (), GAD (GADA), IA- QA-A), and ICA in recentonset IDD patients aged < 6 years (h = 7) or ^.6 years (n = 6) at diagnosis. The number of patients positive for antibodies to, GAD, or IA-, or combination of antibodies to these antigens, are given within the circles. The numbers of patients positive for ICA for each combination of antibodies are given in parentheses. DIABETES CARE, VOLUE, NUBER 6, JUNE

6 Antibodies to antigens in prediabetes (approximately %), the prevalence of and A-A was decreased (although not significant for IA-A) and only a minority (37%) had multiple antibodies. Indeed % of the older patients were negative for all antibodies tested. All of these patients had ketone bodies in urine at diagnosis, had BI below, and have been dependent on insulin for at least years, consistent with a diagnosis of IDD rather than NIDD. The finding that a considerable number of older IDD patients lack four major antibody markers associated with autoimmune diabetes may indicate different characteristics of childhood and adult IDD, as suggested by others (). However, since we observed loss of antibody positivity in a number of first degree relatives on follow-up, the antibody status at the time of diabetes onset may not provide an accurate reflection of autoimmune responses during the prediabetic period. It is therefore possible that these older patients may have been positive for IDD-associated antibodies at an earlier age, but had lost these antibodies over time. There is clearly an urgent need for further studies on the time course of antibody appearance and disappearance, both in families and in the general population. In conclusion, the results presented here show that antibodies to GAD and IA- are important in the identification of individuals most at risk for diabetes development. Combined analysis of different antibody specificities clearly has advantages in identification of individuals at risk. With the emergence of better assays for antibodies to individual islet antigens, large scale studies are now possible to evaluate which combinations of antibodies can best be used to identify individuals at risk for IDD, both in family members of IDD patients, and in the general population. Acknowledgments U.R. was supported by the 'Studienstiftung des Deutschen Volkes' and this study is a part of her 'dissertation' at me Ludwig-aximilians-University of unich, Germany..R.C. is a Royal Society University Research ellow and E.C.I.H. a Wellcome Trust Research Training ellow. A.G.Z. is supported by the 'Deutsche orschungsgemeinschaft' (Heisenberg: Zi 3:6-7 and Sta 68-). This work is supported by grants of the BT (KD893), of the German Diabetes Association, the British Diabetic Association, the Juvenile Diabetes oundation and the Joint Research und of King's Healthcare. The excellent technical assistance of U. ollenhauer is greatly appreciated. We thank Dr. E. Keller and Prof. Dr. E. Albert (Institute for Immune Genetics, Ludwig-aximilians-University unich) for HIA-analysis. References. Bonifacio E, Bingley PJ, Shattock, Dean B, Dunger D, Gale EA, Bottazzo G: Quantification of islet cell antibodies and prediction of insulin-dependent diabetes. Lancet 33:7-9, 99. Bingley PJ, Bonifacio E, Shattock, Gillmor HA, Sawtell PA, Dunger DB, Scott R, Bottazzo G, Gale E: Can islet cell antibodies predict IDD in the general population? Diabetes Care 6:-, Landin-Olsson, Palmer JP, Lernmark A, Blom L, Sundkvist G, Nystrom L, Dahlquist G: Predictive value of islet cell and insulin autoantibodies for type (insulin-dependent) diabetes mellitus in a populationbased study of newly-diagnosed diabetic and matched control children. Diabetologia 3:68-73, 99. Bingley PJ, Christie R, Bonifacio E, Bonfanti R, Shattock, onte T, Bottazzo G, Gale E: Combined analysis of autoantibodies improves prediction of IDD in islet cell antibody-positive relatives. Diabetes 3:3-3, 99. Harrison LC: Islet cell autoantigens in insulin-dependent diabetes: Pandoras box revisited. Immunol Today 3:38-3,99 6. Christie R, Vohra G, Champagne P, Daneman D, Delovitch TL: Distinct antibody specificities to a 6-kD islet cell antigen in type diabetes as revealed by trypsin treatment. JExped 7:789-79, Payton, Hawkes C, Christie : Relationship of the 37,- and,-r tryptic fragments of islet antigens in insulindependent diabetes to the protein tyrosine phosphatase-like molecule IA- (ICA). JClin Invest 96:6-, Roll U, Christie R, Standl E, Ziegler AG: Associations of anti-gad antibodies with islet cell antibodies and insulin autoantibodies in first-degree relatives of type I diabetic patients. Diabetes 3:-6,99 9. Ziegler AG, Bachmann W, Rabl W: Prophylactic insulin treatment in relatives at high risk for type diabetes. Diabetes etab Rev 9:89-93,993. Petersen JS, Hejnaes KR, oody A, Karlsen AE, arshall O, Hoieradsen, Boel E, ichelsen BK, Dyrberg T: Detection of GAD(6) antibodies in diabetes and other autoimmune disease using a simple radioligand assay. Diabetes 3:9-67, 99. Schmidli RS, Colman PG, Bonifacio E, participating laboratories: Disease sensitivity and specificity of assays for glutamic acid decarboxylase antibodies. The Second GADAb Workshop. Diabetes :636-6, 99. Ziegler AG, Ziegler R, Vardi P Jackson RA, Soeldner JS, Eisenbarth GS: Life-table analysis of progression to diabetes of antiinsulin autoantibody-positive relatives of individuals with type I diabetes. Diabetes 38:3-3, Greenbaum CJ, Wilkin TJ, Palmer JP: ifth International Serum Exchange Workshop for Insulin Autoantibody () Standardization. DiabetologLa 3:798-8,99. Lu J, Li Q, Xie H, Chen ZJ, Borovitskaya AE, aclaren NK, Notkins AL, Lan : Identification of a second transmembrane protein tyrosine phosphatase, IA- beta, as an autoantigen in insulin-dependent diabetes mellitus: precursor of the 37-kDa tryptic fragment. Proc Natl Acad Sci USA 93:37-3,996. Hawkes CJ, Wasmeier C, Christie R, Hutton JC: Identification of the 37kDa-antigen in IDD as a tyrosine phosphatase-like protein (phogrin) related to IA-. Diabetes :87-9, Seidel D, Ziegler AG: Prediction of type diabetes. Horm Res :36-39, Verge C, Gianani R, Kawasaki E, Yu L, Pietropaolo, Jackson RA, Chase P, Eisenbarth GS: Prediction of type I diabetes in first degree relatives using a combination of insulin, GAD, and ICAbdcIA- autoantibodies. Diabetes :96-933, Roll U, Christie R, uchtenbusch, Payton A, Hawkes CJ, Ziegler AG: Perinatal autoimmunity in offspring of diabetic parents: the German multicentre BABY-DIAB study: detection of humoral immune responses to islet antigens in early childhood. Diabetes : , Genovese S, Bingley PJ, Bonifacio E, Christie R, Shattock, Bonfanti R, oxon R, Gale EA, Bottazzo G: Combined analysis of IDD-related autoantibodies in healthy schoolchildren. Lancet 3:76, 99. Karjalainen J, Salmela P, Ilonen J, Surchel H, Knip : A comparison of childhood and adult type diabetes mellitus. N EnglJ ed 3:88-886, DIABETES CARE, VOLUE, NUBER 6, JUNE 997