THE HUMAN LEUKOCYTE antigen (HLA) haplotype

0021-972X/00/$03.00/0 Vol. 85, No. 3 The Journal of Clinical Endocrinology & Metabolism Printed in U.S.A. Copyright 2000 by The Endocrine Society Islet Cell Antibody-Positive Relatives with Human Leukocyte Antigen DQA1*0102, DQB1*0602: Identification by the Diabetes Prevention Trial-Type 1* CARLA J. GREENBAUM, DESMOND A. SCHATZ, DAVID CUTHBERTSON, ADINA ZEIDLER, GEORGE S. EISENBARTH, AND JEFFREY P. KRISCHER FOR THE DIABETES PREVENTION TRIAL-TYPE 1 STUDY GROUP Department of Veterans Affairs, Puget Sound Health Care System, and the Department of Medicine, University of Washington (C.J.G.), Seattle, Washington 98108; Department of Pediatrics, University of Florida (D.A.S.), Gainesville, Florida 32610; H. Lee Moffitt Cancer Center and Research Institute, University of South Florida (D.C., J.P.K.), Tampa, Florida 33612; Barbara Davis Center for Childhood Diabetes, University of Colorado (G.S.E.), Denver, Colorado 80262; and Department of Medicine, Endocrinology, and Diabetes, University of Southern California (A.Z.), Los Angeles, California 90033. ABSTRACT The presence of human leukocyte antigen (HLA) haplotype DQA1*0102, DQB1*0602 is associated with protection from type 1 diabetes. The Diabetes Prevention Trial-type 1 has identified 100 islet cell antibody (ICA)-positive relatives with this protective haplotype, far exceeding the number of such subjects reported in other studies worldwide. Comparisons between ICA relatives with and without DQB1*0602 demonstrated no differences in gender or age; however, among racial groups, African-American ICA relatives were more likely to carry this haplotype than others. The ICA DQB1*0602 individuals were less likely to have additional risk factors for diabetes [insulin autoantibody (IAA) positive or low first phase insulin release THE HUMAN LEUKOCYTE antigen (HLA) haplotype DQA1*0102, DQB1*0602 (hereafter referred to as DQB1*0602) is thought to protect people from diabetes, as it is present in less than 1% of Caucasian children and approximately 3% of Caucasian adults with type 1 diabetes (1 6). In contrast, this haplotype occurs in more than 20% of the general Caucasian population. This protective effect of DQB1*0602 has also been reported in some studies of Asian and African-American populations (7, 8), but not in others (9, 10). Importantly, although only a few individuals have been studied, it appears that this protection from overt disease occurs even in DQB1*0602-positive relatives with islet cell antibodies (ICA) (6, 11, 12), a well known marker for diabetes risk. In addition, these subjects reportedly have few other autoantibodies and usually have normal -cell function as Received March 24, 1999. Revision received September 3, 1999. Accepted November 28, 1999. Address all correspondence and requests for reprints to: Dr. Carla J. Greenbaum, Department of Veterans Affairs, Puget Sound Health Care System, and the Department of Medicine, University of Washington, Seattle, Washington 98108. * This work was supported through cooperative agreements by the Division of Diabetes, Endocrinology, and Metabolic Diseases, NIDDK, NIH, and by the NIAID, NICHHD, and National Center for Research Resources, NIH; the American Diabetes Association; the Juvenile Diabetes Foundation International; and various corporate sponsors. (FPIR)] than ICA relatives without DQB1*0602. However, 29% of the ICA DQB1*0602 relatives did have IAA or low FPIR. Although half of the ICA DQB1*0602 relatives had a high risk second haplotype, this was not associated with the additional risk factors for diabetes. Hispanic ICA individuals with DQB1*0602 were more likely to be IAA positive or to have low FPIR than other racial groups. In conclusion, the presence of ICA in the relatives described here suggests that whatever the mechanism that protects DQB1*0602 individuals from diabetes, it is likely to occur after the diabetes disease process has begun. In addition, there may be different effects of DQB1*0602 between ethnic groups. (J Clin Endocrinol Metab 85: 1255 1260, 2000) determined by first phase insulin release (FPIR) in response to iv glucose (6, 13). From 1994 through 1997, the Diabetes Prevention Trialtype 1 (DPT-1) has screened more than 52,000 first and second degree relatives of patients with type 1 diabetes, identifying 3.6% of these relatives to be ICA positive. Antibody-positive subjects who carry the HLA DQA1*0102, DQB1*0602 haplotype are excluded from further participation in the DPT-1 because their risk of disease is low (14). This screening process has now identified 100 ICA relatives with this protective haplotype, far exceeding the number of such subjects reported in other studies. This current study reports the frequency of DQB1*0602 among ICA relatives and the association of DQB1*0602 with demographic variables (age, gender, race, and relationship to proband). In addition, we report the results of other autoantibody measurements, tests of -cell function, and second DQ haplotype in this group. Subjects Subjects and Methods The DPT-1 began in 1994 with the goal of determining whether antigen therapy can prevent or delay the onset of clinical type 1 diabetes. The study is divided into three parts; screening, staging, and intervention. After informed consent, first and second degree relatives of probands with type 1 diabetes (below age 45 yr) are screened for the presence of ICA at 10 or more Juvenile Diabetes Foundation (JDF) units. 1255

1256 GREENBAUM ET AL. JCE&M 2000 Vol 85 No 3 Subjects that are ICA positive are then eligible for staging. This consists of repeat measurement of ICA and determination of insulin autoantibody (IAA) status. In addition, -cell function is assessed by measurement of FPIR, and the presence or absence of HLA DQA1*0102/ DQB1*0602 is determined. If this haplotype is present, the subject is no longer eligible for the DPT-1 study. Relatives who are ICA positive and have low FPIR (in siblings and offspring, below the 10th percentile of normal controls; in parents, below 1st percentile of normal controls) on two occasions are defined as being at high risk for developing type 1 diabetes ( 50% risk over 5 yr). Relatives who are ICA positive, have IAA present, and do not have a low FPIR are defined as being at intermediate risk for developing diabetes (25 50% risk over 5 ys). Relatives confirmed to be ICA positive, but without either low FPIR or IAA are defined as low risk ( 25% risk over 5 yr). Low risk relatives undergo repeat staging every 6 months. Intermediate risk relatives are eligible for random assignment to intervention with either oral insulin or placebo. High risk relatives are eligible for random assignment either to intervention with parenteral insulin or to a closely monitored group. Autoantibody assays ICA ICA are determined by immunofluorescence assay on frozen sections of human pancreas in the DPT-1 ICA Core Laboratory (Gainesville, FL; February 1994 to September 1997; New Orleans, LA, September 1997 to present). Values 10 or more JDF units are considered positive. In the 1995 Immunology of Diabetes Society (IDS) workshop, this ICA assay had a specificity of 100% and a sensitivity of 74.4% for new-onset patients less than age 30 yr. IAA. IAA are determined in the DPT-1 IAA Core Laboratory (Boston, MA) by a fluid phase radioassay employing 600 L serum with duplicate determinations with and without unlabeled insulin for competition. The interassay coefficient of variation is 10.3% at low positive values. In the 1995 IDS workshop, this IAA assay had a specificity of 91% and a sensitivity of 49% for new-onset patients less than age 30 yr. GAD65ab and ICA512ab assay. GAD65 and ICA512 antibodies were determined in Denver, CO, on DPT-1 samples as part of an ancillary study. As previously described, a combined GAA and ICA512AA radioassay was performed (15). Labeled recombinant GAD65 and ICA512 were produced by in vitro transcription/translation with differential labeling ([ 3 H]GAD65 and [ 35 S]ICA512). The levels of both antibodies were expressed as an index. The interassay coefficients of variation are 6% and 9.6% for GAD65ab and ICA512ab, respectively. The upper limits of normal (0.032 for GAD65ab; 0.071 for ICA512ab) were established as the 99th percentile for GAD65 and the 100th percentile for ICA512ab from receiver operating characteristics curves in 198 healthy control subjects and 50 patients with new-onset diabetes. In the 1995 IDS workshop, sensitivity for the GAD65ab assay was 82%, and specificity was 99%. Sensitivity for the ICA512ab assay was 73%, and specificity was 100%. -Cell function Intravenous glucose tolerance tests are performed according to the ICARUS protocol (16, 17). After an overnight fast and insertion of an iv catheter in an antecubital vein, samples are drawn at 10 and 4 min. A solution of 25% glucose (0.5 g/kg; maximum, 35 g) is then administered iv over a 3-min period. After the infusion of glucose, samples are drawn at 1, 3, 5, 7, and 10 min. Insulin and glucose are measured in the DPT-1 Beta Cell Function Core Laboratory (Seattle, WA). The insulin values at 1 and 3 min are added to calculate a measure of FPIR. The 10th percentile of normal controls for siblings and offspring over age 8 yr is 100 U/mL. The 10th percentile for siblings and offspring under age 8 yr and the 1st percentile for parents is 60 U/mL. Both of these values were determined from iv glucose tolerance test studies of 224 nondiabetic individuals from 4 45 yr of age 1 and are used as the thresholds for eligibility for the parenteral insulin intervention trial. 1 Cuthbertson, D., J. Krischer, D. Schatz, S. Johnson, and N. Maclaren, submitted for publication. HLA typing Samples for HLA typing are obtained during the first staging visit for ICA relatives. HLA typing is performed at the DPT-1 HLA Core Laboratory (Denver, CO). HLA-DQA1 and DQB1 alleles are typed using PCR and sequence-specific oligonucleotide probes (19, 20). HLA DQA1*0102, DQB1*0602-positive and -negative control samples are included in all assays. Statistical analysis All data analysis was performed using an individual s first ICA, IAA, and FPIR measurements. Characteristics that may be related to the presence of DQA1*0102, DQB1*0602 (e.g. age at screening, race, gender, relationship to type 1 diabetic proband, and results of IAA and iv glucose tolerance testing), were assessed by univariate analysis. The Kolmogorow-Smirnov test was used to compare the distribution of age between ICA subjects with and without DQA1*0102, DQB1*0602. Categorical variables were compared between those with and without DQA1*0102, DQB1*0602 by the Pearson s 2 test or Fisher s exact test (depending on the number of individuals per cell). Continuous variables were compared by the Wilcoxon rank-sum test. Multivariate statistical models were developed using logistic regression. All tests of significance were two tailed. Statistical analyses were performed using SAS software (SAS Institute, Cary, NC). Results More than 52,000 relatives were screened by December 31, 1997. At that time, 100 of 1376 (7.3%) ICA DQ-typed relatives were found to have the haplotype DQA1*0102, DQB1*0602. The age distribution of these DQB1*0602 ICA relatives was similar to that of DQB1*0602- ICA relatives (P 0.36), and there were no differences in gender (P 0.51). The DQB1*0602 individuals were more likely to be a second degree relative, less likely to be a sibling, and more likely to be African-American compared with DQB1*0602 ICA individuals (Table 1). We also compared the ICA DQB1*0602 individuals with the ICA subjects. There were no significant differences between these groups with regard to age, gender, race, relationship to proband, or the presence of multiplexity. Differences between ICA-positive relatives with DQB1*0602 and those without DQB1*0602 were evidenced TABLE 1. Comparison of gender, proband relationship, and race among DQB1*0602-positive and DQB1*0602-negative relatives DQB1*0602 (n 100) DQB1*0602 (n 1276) P value Relationship to proband 0.01 Sibling 40 (40) 655 (51.3) Offspring 27 (27) 352 (27.6) Parent 17 (17) 188 (14.7) Second degree 16 (16) 81 (6.4) Gender 0.51 Male 49 (49) 669 (47.6) Female 51 (51) 607 (52.4) Race 0.03 Caucasian 83 (85.6) 1130 (90.9) African-American 6 (6.2) 21 (1.7) Hispanic 7 (7.2) 69 (5.6) Other 1 (1.0) 23 (1.9) Unknown 3 33 No. of people with diabetes 0.163 within the family 1 family member 15 (15) 266 (20.6) 1 family member 85 (85) 1010 (79.2) Values are the number of subjects; percentages are in parentheses.

ICA-POSITIVE HLA DQB1*0602 1257 by both a lower median titer [40 JDF units (25th percentile, 20 JDF units;75th percentile, 80 JDF units) vs. 80 JDF units (25th percentile, 20 JDF units; 75th percentile, 160 JDF units)] and a higher frequency of lower ICA titers (by 2 test, P 0.001; Fig. 1). In addition, relatives with DQB1*0602 were less likely to be confirmed ICA positive on a second sample (59% vs. 78%; P 0.001). This does not imply loss of autoimmunity, but, rather, that a second sample was below 10 JDF units. The DQB1*0602 relatives were also less likely to be IAA positive or to have low FPIR than ICA-positive relatives without DQB1*0602 (P 0.01 for both; Table 2). Similarly, comparison of the 71 ICA DQB1*0602-positive and 434 ICA DQB1*0602-negative relatives tested for GAD65 and ICA512 autoantibodies in addition to IAA demonstrated that the DQB1*0602-positive relatives were less likely to have additional autoantibodies than those without DQB1*0602 (P 0.001). Nevertheless, 17 (24%) of these ICApositive DQB1*0602 relatives were positive for at least 1 antibody in addition to ICA (Table 3). Multivariate analysis of these demographic characteristics (race and relationship of relative to proband with diabetes) as well as markers of the diabetes disease process (FPIR, IAA status, and ICA titer) demonstrated that African-American ICA individuals were 4.4 [95% confidence interval (CI), 1.6 11.8] times more likely to be DQB1*0602 positive than Caucasian individuals (P 0.01), and second degree relatives were 2.3 (CI, 1.2 4.8) times more likely to be DQB1*0602 than siblings of patients with diabetes (P 0.02). Those with ICA titers of 160 or more JDF units or with low FPIR were half (CI, 0.3 0.99) as likely to be DQB1*0602 positive compared with those having lower ICA titers (P 0.05) or FPIR above threshold (P 0.05). IAA subjects were 70% (CI, 0.42 0.83) less likely to be DQB1*0602 than those who were IAA negative (P 0.01; Table 4). Forty-nine percent of ICA DQB1*0602 relatives had a high risk second haplotype. Of these, 27 had DQA1*0501/ DQB1*0201, and 22 had DQA1*0301/DQB1*0302. Only 3 subjects were homozygous for DQA1*0102/DQB1*0602. Twenty-nine percent of ICA individuals with DQB1*0602 had IAA present and/or a low FPIR. Three of these individuals were IAA positive and had a low FPIR. There was no TABLE 2. Comparison of IAA positivity and low FPIR among DQB1*0602-positive and DQB1*0602-negative relatives DQB1*0602 DQB1*0602 P value % IAA positive 20/99 (20.2) 604/1270 (47.6) 0.01 % FPIR below threshold 12/92 (13.0) 340/651 (28.3) 0.01 Values are the number of subjects; percentages are in parentheses. TABLE 3. Presence of additional autoantibodies (GAD65ab, IAA, ICA512ab) among ICA DQB1*0602-positive and DQB1*0602- negative relatives No. of antibodies in addition to ICA DQB1*0602 positive relatives DQB1*0602 positive relatives One antibody 10 (14.1) 72 (16.6) GAD65 4 27 IAA 6 42 ICA512 0 3 Two antibodies 5 (7.0) 101 (23.3) GAD65 IAA 3 62 GAD65 ICA512 1 19 IAA ICA512 1 20 Three antibodies 2 (2.8) 72 (16.6) GAD65 IAA ICA512 2 72 P 0.001. Values are the number of subjects; percentages are in parentheses. TABLE 4. Multivariate analysis for risk of being DQB1*0602 in ICA relatives P value Odds ratio 95% confidence interval Race (vs. Caucasian) African-American 0.01 4.4 1.63 11.82 Hispanic 0.76 1.1 0.47 2.81 Other race 0.76 0.7 0.09 5.65 Relationship to proband (vs. sibling) Offspring 0.89 1.0 0.61 1.77 Parent 0.05 0.4 0.18 1.00 Second degree 0.02 2.3 1.15 4.75 IAA positive (vs. negative) 0.01 0.3 0.17 0.58 Low FPIR (vs. FPIR above 0.05 0.5 0.27 0.99 threshold) ICA TITER 160 JDF units (vs. 160 JDF units) 0.05 0.6 0.31 0.99 TABLE 5. Second DQ haplotypes in ICA relatives with DQA1*0102/DQB1*0602: relationship to IAA positivity and low FPIR IAA and above threshold FPIR IAA and/or low FPIR Total Total high risk 32 17 49 DQA1*0501/DQB1*0201 16 11 27 DQA1*0301/DQB1*0302 16 6 22 Nonhigh risk 39 12 51 Total 71 29 P 0.22, by 2 test. FIG. 1. Percentage of subjects with ICA titers of 10, 20, 40, 80, or more than 80 JDF units with ( ) and without (u) HLA DQA1*0102/ DQB1*0602. relationship between the presence of high risk second haplotypes and these additional immune and/or metabolic markers, as 17 of 29 (58.6%) individuals with IAA and/or low FPIR had a high risk second haplotype compared with 32 of 71 (45.1%) DQB1*0602 subjects with ICA alone (P 0.22; Table 5). Examination of demographic characteristics and the presence of autoantibodies demonstrated that median ICA titers were not different between those DQB1*0602 relatives

1258 GREENBAUM ET AL. JCE&M 2000 Vol 85 No 3 with IAA and/or low FPIR and those with ICA alone (P 0.23). There were also no differences in gender or relationship to proband between the groups (Table 6). Further, although overall differences in racial groups was only marginally significant (P 0.09), examination of multiple comparisons determined that Hispanics with DQB1*0602 were 5.9 (CI, 1.1 31.3) times more likely than Caucasians to have low FPIR or to be IAA positive (Table 7). Follow-up information was obtained for 24 of these 29 individuals with IAA and/or low FPIR. Two of these individuals have developed diabetes. One, a 36-yr-old African- American male with a high ICA titer; negative IAA, GAD65ab, and ICA512ab; but low FPIR developed diabetes several weeks after undergoing FPIR testing. The second, a Caucasian female who, at age 9 yr was GAD65ab, IAA, and ICA512ab positive but without a low FPIR, developed diabetes at age 14 yr. The remaining 22 subjects were reportedly free from diabetes after a mean follow-up time of 22.4 months (range, 6 40 months) after IAA or FPIR testing. TABLE 6. Comparison of gender, proband relationship, and race among DQB1*0602 positive relatives IAA and/or low FPIR (n 29) IAA and FPIR above threshold (n 71) P value Relationship to 0.20 proband Sibling 10 (34.5) 30 (42.3) Offspring 12 (41.4) 15 (21.1) Second degree 3 (10.3) 13 (18.3) Parent 4 (13.8) 13 (18.) Gender 0.43 Male 16 (55.2) 33 (46.5) Female 13 (44.8) 38 (53.5) Race 0.09 Caucasian 20 (74.1) 63 (90.0) African-American 3 (11.1) 3 (4.3) Hispanic 4 (14.8) 3 (4.3) Values are the number of subjects; percentages are in parentheses. TABLE 7. Multivariate analysis for risk of being IAA and/or have low FPIR in DQB1*0602 ICA relatives Race (vs. Caucasian) P value Odds ratio 95% confidence interval African-American 0.40 2.1 0.36 12.62 Hispanic 0.04 5.9 1.10 31.28 Discussion The presence of HLA DQA1*0102, DQB1*0602 is unusual in people with type 1 diabetes and their relatives. However, the DPT-1 has identified 100 ICA-positive relatives with this protective allele. As the presence of ICA indicates autoimmunity against islet antigens, examination of this large group of DQB1*0602 individuals may provide important information about how this haplotype impacts the development of clinical disease. As expected, because the risk of diabetes among siblings of diabetic probands is greater than that for other relatives, the ICA DQB1*0602 subjects reported here were more likely to be second degree relatives and less likely to be siblings than the individuals without DQB1*0602. Similarly, relatives with DQB1*0602 had lower median ICA titers and were less likely to be IAA positive or to have low FPIR than ICA DQB1*0602-negative relatives. This suggests that DQB1*0602 ICA relatives are more likely to be nonprogressors; that is, despite being ICA, as a group they are less likely to develop diabetes within the next 5 yr. Given the premise that DQB1*0602 is a protective gene, these data are not unexpected and are consistent with other reports using smaller numbers of individuals (11, 13, 21). For example, the Childhood Diabetes in Finland study group reported that 13 (16%) of ICA- or IAA-positive siblings had either DQB1*0602 or 0603 alleles, and compared with those having high risk DQB1 genotypes (0201, 0302/x, 0201/x; n 46), the DQB1*0602- positive siblings had lower ICA and GAD antibody titers and higher FPIR (21). An interesting question is whether the presence of DQB1*0602 affords relative protection from clinical disease over a lifetime or whether this protection only exists within younger populations. Two studies demonstrating a lessening of the degree of protection with increasing age at diagnosis of diabetes suggest the latter may be true (22, 23). In this regard, our data demonstrating that there was no difference in the age distribution among ICA relatives with and without DQB1*0602 indicates that such delayed development of clinical disease must occur after the initiation of autoimmunity as manifested by ICA. We have identified 29 DQB1*0602 relatives with ICA who had other factors associated with risk of progression to diabetes. Twenty of these had IAA in addition to ICA. Such an antibody profile (in relatives not classified according to HLA type) is associated with a 25 50% risk of diabetes over 5 yr. Similarly, the combination of ICA with low insulin secretion seen in 12 of the DQB1*0602 relatives is associated with a more than 50% risk of disease during that time frame (see Footnote 1). Other studies have noted the presence of low FPIR in 8% of relatives with protective DQ alleles (21). Follow-up information is available for 24 of the 29 DQB1*0602- positive ICA individuals with IAA and/or low FPIR. Two of these 29 subjects have developed diabetes over a brief duration of follow-up. A possible explanation for the additional risk factors in these ICA-positive DQB1*0602 relatives would be the presence of a high risk second haplotype that somehow modulates the protective effect of the DQB1*0602 gene. Indeed, more than half of the ICA-positive DQA1*0102/DQB1*0602 individuals had the high risk second haplotypes DQA1*0501/DQB1*0201 or DQA1*0301/DQB1*0302. However, IAA and/or low insulin secretion was not associated with these high risk second haplotypes. Although these data may indirectly support previous reports indicating that the protective effect of HLA DQA1*0102/DQB1*0602 is dominant (2), these data alternatively imply that some other gene is sufficient for the development of IAA or low FPIR. A polymorphism in the insulin gene region has been reported to be associated with increased diabetes risk in subjects without high risk HLA genes (24). Further, it is possible that the individuals identified in our population as DQB1*0602 have variant sequences. Hoover et al. reported that of five diabetic subjects who were typed as

ICA-POSITIVE HLA DQB1*0602 1259 DQB1*0602, one was DQB1*0603 with the expected sequence of that gene, but four had unique sequences (one associated with DQB1*0602 and three with DQB1*0603) (25). Although such variation was not observed in a recent report of ICA relatives typed for DQB1*0602 (26), sequencing has not been performed in our population. The role of ethnicity in ICA DQB1*0602-positive individuals is of interest, as there are great differences between races with respect to incidence of type 1 diabetes. Within the U.S., the reported age-adjusted incidence ranges from 3.3 17.6/100,000/yr among African Americans, 4.1 15.2/ 100,000 among Hispanics, and 13.1 20.6/100,000 among Caucasians (27, 28). It is thus possible that the presence of DQB1*0602 may have different effects in different racial groups. The apparent protective effect of DQB1*0602 for diabetes, which has been consistently demonstrated in studies of Caucasian populations with diabetes (1 6) and in a population study of Mexican-Americans (29), is less clear among African-Americans; one report indicated a protective effect (8), and another indicated that DQB1*0602 had no effect on diabetes incidence (9). In DPT-1, 81.6% of relatives screened and 85.9% of relatives found to be ICA were Caucasian. However, African-Americans who were ICA positive were more likely to be DQB1*0602 than others. Additionally, among the ICA DQB1*0602 subjects, Hispanic individuals were more likely to have IAA and/or low FPIR than non- Hispanics, and 1 of the 2 subjects who progressed to diabetes was African-American. These data suggest that there are variations in the type 1 diabetes disease process between races, as previously reported (7, 9, 10, 30 38). The mechanism by which HLA DQB1*0602 protects from diabetes is unknown. One possibility is that this class II molecule is unable to bind the initiating autoantigen with enough affinity to allow initiation of the autoimmune response. A similar hypothesis proposes the converse, that this class II molecule preferentially binds the autoantigen but does not initiate an immune response, effectively blocking other class II molecules from propagating autoimmunity. Others have proposed that DQB1*0602 binds a T cell receptor that recognizes a primary autoantigen, but then deletes this T cell clone rather than causing activation. Alternatively, HLA DQA1*0102/DQB1*0602 class II molecules may prevent propagation of the immune response by an inability to respond to activation by superantigens. Another proposal suggests that binding of antigen and T cell by this class II molecule somehow activates a protective, instead of a destructive, immune response (39). In conclusion, the DPT has identified 100 ICA relatives with the HLA DQA1*0102/DQB1*0602 haplotype. The presence of ICA in the individuals described here suggests that whatever the mechanism that protects DQB1*0602 subjects from diabetes, it occurs after the diabetes disease process has been initiated. Thus, rather than preventing the disease process, the presence of HLA DQA1*0102/DQB1*0602 may subsequently arrest or delay the onset of overt disease. In addition, our data showing that 7% of ICA relatives had DQB1*0602, that 29% of these individuals additionally had IAA and/or low insulin secretion, and that 2 subjects have already progressed to diabetes emphasize that the protection afforded by this allele is relative and not absolute. Prospective follow-up of the ICA DQB1*0602 relatives is underway to determine whether our subgroup of subjects with high risk characteristics will develop diabetes over a longer period of time. Further characterization and follow-up of these individuals may lead to an understanding of how this class II molecule protects from disease and whether the effects vary among ethnic groups. References 1. Baisch JM, Weeks T, Giles R, Hoover M, Stastny P, Capra JD. 1990 Analysis of HLA-DQ genotypes and susceptibility in insulin-dependent diabetes mellitus. N Engl J Med. 322:1836 1841. 2. Erlich HA, Griffith RL, Bugawan TL, Ziegler R, Alper C, Eisenbarth GS. 1991 Implication of specific DQB1 alleles in genetic susceptibility and resistance by identification of IDDM siblings with novel HLA-DQB1 allele and unusual DR2 and DR1 haplotypes. Diabetes. 40:478 481. 3. Nepom GT. 1993 Immunogenetics and IDDM. Diabetes Rev. 1:93 103. 4. Kockum I, Lernmark A, Dahlquist G, et al. 1996 Genetic and immunological findings in patients with newly diagnosed insulin-dependent diabetes mellitus. The Swedish Childhood Diabetes Study Group and The Diabetes Incidence in Sweden Study (DISS) Group. Horm Metab Res. 28:344 347. 5. Reijonen H, Ilonen J, Akerblom HK, Knip M, Dosch HM. 1994 Multi-locus analysis of HLA class II genes in DR2-positive IDDM haplotypes in Finland. The Childhood Diabetes in Finland (DiMe) Study Group. Tissue Antigens. 43:1 6. 6. Huang W, She JX, Muir A, et al. 1994 High risk HLA-DR/DQ genotypes for IDD confer susceptibility to autoantibodies but DQB1*0602 does not prevent them. J Autoimmun. 7:889 897. 7. Awata T, Kanazawa Y. 1994 Genetic markers for insulin-dependent diabetes mellitus in Japanese. Diabetes Res Clin Pract. 24:S83 S87. 8. Leech NJ, Kitabchi AE, Gaur LK, et al. 1995 Genetic and immunological markers of insulin dependent diabetes in black Americans. Autoimmunity. 22:27 32. 9. Fernandez-Vina M, Ramirez LC, Raskin P, Stastny P. 1993 Genes for insulindependent diabetes mellitus (IDDM) in the major histocompatibility complex (MHC) of African-Americans. Tissue Antigens. 41:57 64. 10. Huang HS, Peng JT, She JY, et al. 1995 HLA-encoded susceptibility to insulindependent diabetes mellitus is determined by DR and DQ genes as well as their linkage disequilibria in a Chinese population. Hum Immunol. 44:210 219. 11. Pugliese A, Gianani R, Moromisato R. 1995 HLA DQB1*0602 is associated with dominant protection from diabetes even among islet cell antibody positive first degree relatives of patients with insulin-dependent diabetes. Diabetes. 44:608 613. 12. Veijola R, Knip M, Reijonen H, Vahasalo P, Puukka R, Iionen J. 1995 Effect of genetic risk load defined by HLA-DQB1 polymorphism on clinical characteristics of IDDM in children. Eur J Clin Invest. 25:106 112. 13. Gianani R, Verge CF, Moromisato-Gianani RI, et al. 1996 Limited loss of tolerance to islet autoantigens in ICA first degree relatives of patients with type I diabetes expressing the HLA DQB1*0602 allele. J Autoimmun. 9:723 725. 14. DPT-1 Study Group. 1995 The Diabetes Prevention Trial (DPT-1): implementation of screening and staging of relatives. Transplant Proc. 27:3377. 15. Yu L, Rewers M, Gianani R, et al. 1996 Antiislet autoantibodies usually develop sequentially rather than simultaneously. J Clin Endocrinol Metab. 81:4264 4267. 16. McCulloch D, Bingley P, Colman P, Jackson R, Gale E. 1993 Comparison of bolus and infusion protocols for determining acute insulin response to intravenous glucose in normal humans. Diabetes Care. 16:911 915. 17. Bingley PJ, ICARUS Group. 1996 Interactions of age, islet cell antibodies, insulin autoantibodies, and first-phase insulin response in predicting risk of progression to IDDM in ICA relatives: the ICARUS data set. Diabetes. 45:1720 1728. 18. Deleted in proof. See Footnote 1. 19. Gyllensten UB, Erlich HA. 1988 Generation of single-stranded DNA by the polymerase chain reaction and its application to direct sequencing of the HLA-DQA locus. Proc Natl Acad Sci USA. 85:7652 7656. 20. Bugawan L, Erlich HA. 1991 Rapid typing of HLA-DQB1 DNA polymorphism using nonradioactive oligonucleotide probes and amplified DNA. Immunogenetics. 33:163 170. 21. Veijola R, Vahasalo P, Tuomilehto-Wolf E, et al. 1995 Human leukocyte antigen identity and DQ risk alleles in autoantibody-positive siblings of children with IDDM are associated with reduced early insulin response. Diabetes. 44:1021 1028. 22. Caillat-Zucman S, Garchon HJ, Timsit J, et al. 1992 Age-dependent HLA genetic heterogeneity of type 1 insulin-dependent diabetes mellitus. J Clin Invest. 90:2242 2250. 23. Graham J, Kockum I, Sanjeevi CB, et al. 1999 Negative association between

1260 GREENBAUM ET AL. JCE&M 2000 Vol 85 No 3 type 1 diabetes and HLA DQB1 *0602-DQA1* 0102 is attenuated with age at onset. Eur J Immunogenet. 26:117 127. 24. Halminen M, Veijola R, Reijonen H, Ilonen J, Akerblom HK, Knip M. 1996 Effect of polymorphism in the insulin gene region on IDDM susceptibility and insulin secretion. The Childhood Diabetes in Finland (DiMe) Study Group. Eur J Clin Invest. 26:847 852. 25. Hoover ML, Marta RT. 1997 Molecular modeling of HLA-DQ suggests a mechanism of resistance in type 1 diabetes. Scand J Immunol. 45:193 202. 26. Pugliese A, Kawasaki E, Zeller M, et al. 1999 Sequence analysis of the diabetes-protective human leukocyte antigen-dqb1*0602 allele in unaffected, islet cell antibody-positive first degree relatives and in rare patients with type 1diabetes. J Clin Endocrinol Metab. 84:1722 1728. 27. LaPorte RE, Matsushima M, Chang Y-F. 1995 Prevalence and incidence of insulin-dependent diabetes. In: National Diabetes Data Group, ed. Diabetes in America. Bethesda: NIH; 37 46. 28. Libman IM, LaPorte RE, Becker D, Dorman JS, Drash AL, Kuller L. 1998 Was there an epidemic of diabetes in nonwhite adolescents in Allegheny County, Pennsylvania? Diabetes Care. 21:1278 1281. 29. Erlich HA, Zeidler A, Chang J, et al. 1993 HLA class II alleles and susceptibility and resistance to insulin dependent diabetes mellitus in Mexican-American families. Nat Genet. 3:358 364. 30. Balducci-Silano PL, Layrisse Z, Dominguez E, et al. 1994 HLA-DQA1 and DQB1 allele and genotype contribution to IDDM susceptibility in an ethnically mixed population. Eur J Immunogenet. 21:405 414. 31. Balducci-Silano PL, Layrisse ZE. 1995 HLA-DP and susceptibility to insulindependent diabetes mellitus in an ethnically mixed population. Associations with other HLA-alleles. J Autoimmun. 8:425 437. 32. Cruickshanks KJ, Jobim LF, Lawler-Heavner J, et al. 1994 Ethnic differences in human leukocyte antigen markers of susceptibility to IDDM. Diabetes Care. 17:132 137. 33. Erlich HA, Rotter JI, Chang JD, et al. 1996 Association of HLA-DPB1*0301 with IDDM in Mexican-Americans. Diabetes. 45:610 614. 34. Gaber SA, Mazzola G, Berrino M, et al. 1994 Human leukocyte antigen class II polymorphisms and genetic susceptibility of IDDM in Egyptian children. Diabetes Care. 17:1341 1344. 35. Nishimura Y, Oiso M, Fujisao S, et al. 1998 Peptide-based molecular analyses of HLA class II-associated susceptibility to autoimmune diseases. Int Rev Immunol. 17:229 262. 36. Sanjeevi CB, Zeidler A, Shaw S, et al. 1993 Analysis of HLA-DQA1 and -DQB1 genes in Mexican Americans with insulin-dependent diabetes mellitus. Tissue Antigens. 42:72 77. 37. Undlien DE, Hamaguchi K, Kimura A, et al. 1994 IDDM susceptibility associated with polymorphisms in the insulin gene region. A study of blacks, Caucasians and orientals. Diabetologia. 37:745 749. 38. Zamani M, Cassiman JJ. 1998 Reevaluation of the importance of polymorphic HLA class II alleles and amino acids in the susceptibility of individuals of different populations to type I diabetes. Am J Med Genet. 76:183 194. 39. Nepom GT, Kwok WW. 1998 Molecular basis for HLA-DQ associations with IDDM. Diabetes. 47:1177 1184.