| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Barbara Davis Center for Childhood Diabetes (J.M.B., L.Y., D.M., G.S.E., M.R.) and Department of Preventive Medicine and Biometrics (K.J.B., J.M.N.), University of Colorado Health Sciences Center, Denver, Colorado 80262; and Department of Human Genetics (H.A.E.), Roche Molecular Systems, Inc., Alameda, California 94501
Address all correspondence and requests for reprints to: Jennifer M. Barker, M.D., 4200 East 9th Avenue, Box B-140, Denver, Colorado 80262. E-mail: jennifer.barker{at}uchsc.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The Diabetes Autoimmunity Study in the Young (DAISY) has followed, for the development of T1DM, a cohort of young first-degree relatives of type 1 diabetic patients and infants identified by newborn screening for human leukocyte antigen (HLA)-DR,DQ genotypes associated with diabetes. Through this prospective follow-up, children with antiislet autoantibodies have been discovered. Some of these children have progressed to diabetes, others remain antibody positive without yet developing T1DM, and still others have lost autoantibody expression. We have analyzed factors that are associated with confirmation of autoantibody positivity, the persistence of autoantibodies, and progression to diabetes.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
DAISY is a prospective cohort study that has enrolled and followed, since 1993, 837 young first-degree relatives of patients with T1DM (sibling/offspring cohort) and 1,135 children identified through newborn screening for HLA-DR,DQ genotypes associated with T1DM (newborn cohort). Newborn cohort participants were identified by screening of umbilical cord blood obtained from over 27,000 newborns at St. Joseph’s Hospital in Denver, Colorado. EDTA sample of cord blood was sent for HLA typing, as previously described (4). The high-risk genotype (odds of T1DM by age 20 yr, 1:16) was defined as DRB1*04, DQB1*0302/DRB1*0301, DQB1*0201 (DR3/4 DQ8). Moderate risk genotypes (odds of T1DM by 20 yr, 1:75 in non-Hispanic whites and 1:230 in Hispanics) included DRB1*0301/*0301; DRB1*04, DQB1*0302/DRB1*04, DQB1*0302; and DRB1*04, DQB1*0302/x (where x is not DRB1*04, DQB1*0302, nor DRB1*0301 or DR2). The remainder of the genotypes were classified as average or low risk.
In the DAISY follow-up, all children are tested at 9, 15, and 24 months of age and annually thereafter for autoantibodies to pancreatic islet antigens (as described below). Children who are autoantibody positive are placed on an accelerated clinic visit schedule on which they return every 3–6 months for autoantibodies, hemoglobin A1c, and random blood glucose. Individuals who test negative for the autoantibodies on two consecutive visits return to the annual clinic visit schedule. All newborns with high risk and a subset with moderate or average risk HLA genotypes are invited to participate in the follow-up and have completed 4.5 (range, 1–15) visits on average. In addition, 170 general-population children identified at birth with the HLA-DRB1*0301, DQB1*0201/x genotypes (average risk for diabetes but high risk for celiac disease) were enrolled at the age of 2 yr and followed in identical manner with 2.5 (range, 1–8) visits, on average. Siblings and offspring of patients with T1DM are enrolled before age 4 yr, as soon as identified. The average age at the first visit in this group was 2.5 yr, and the average number of visits was 4.7 (range, 1–20). Of those, 105 of 837 (12.5%) relatives were first tested at 9 months of age and followed with 4.2 (range, 1–14) visits, on average.
The Combined Institutional Review Board at the University of Colorado Health Sciences Center has approved the DAISY project. Informed consent was obtained from the parents of each study subject at genetic screening (for the HLA-screened population) and again at enrollment in the DAISY follow-up.
Measurement of autoantibodies
GAA and IA-2AA were measured simultaneously by combined GAA and IA-2 radioassay (full-length glutamic acid decarboxylase and ICA512bdc cDNA clones). The assay was performed in the 96-well filtration plates with autoantibody-bound 3H-glutamic acid decarboxylase and 35S-IA-2AA precipitated with protein A-Sepharose. The cut-points for positivity were set at indexes of 0.032 (mean ± 2 SD, GAA) and 0.049 (mean ± 6 SD, ICA512AA), the 99th percentile of 198 normal controls. The interassay coefficient of variation of the assays is 6.5 and 9.6% for GAA and ICA512AA assays, respectively. Coefficients of variation are determined at high levels of positivity. In the Diabetes Autoantibody Standardization Program (DASP) 2002 proficiency workshop, assay specificity was 93% for GAA and 99% for ICA512AA, and sensitivity was 90% for GAA and 62% for ICA512AA.
As previously described (5), the micro-IAA assay in this study used 125I-insulin in standard 96-well filtration plates that quantitate precipitated 125I-insulin with a multichannel 96-well ß-counter. Scintillation fluid is directly added to the 96-well plate (Top Counter scintillation counter; Hewlett-Packard, Palo Alto, CA). The upper limit of normal (0.01) was chosen as the 99th percentile from receiver operator curves in 106 healthy control subjects and 105 patients with new-onset diabetes. In the 2002 DASP workshop, sensitivity was 62% and specificity was 98%. The intra- and interassay coefficient of variation is 12 and 16%, respectively. Coefficients of variation are determined at high levels of positivity. Indices for all autoantibodies were calculated using a consensus value for negative and positive control specific for that autoantibody.
DAISY uses an extensive system of quality assurance to correctly classify autoantibodies as positive or negative (Fig. 1
). The autoantibodies are measured by RIA in duplicate. Autoantibodies are classified as positive when average level of the duplicate is greater than the 99th percentile of the normal population. If the initial measurement is negative, the sample is reported as negative; however, a random 5% of these samples is retested in a blinded fashion. If the initial result is positive, the assay is repeated in unblinded fashion on the same aliquot and if two of two or two of three repeat tests are positive, the result is reported as positive. However, if only the initial one of three repeat tests is positive, it is reported as negative. When DAISY staff, independent of the laboratory, receive a positive test report, two additional blinded aliquots of the original sample are sent to the laboratory, and the process described above is repeated independently for both aliquots. The blinded aliquots are repeated in separate assays. To be confirmed as positive, the sample has to be reported as positive on at least two of three blinded aliquots (and with two of two or two of three tests positive for each aliquot). If the autoantibody fails to confirm (only the original aliquot is positive), the final result is classified as a false positive. Therefore, a confirmed positive result is based on at least four positive tests.
|
illustrates autoantibody patterns of selected individuals in the transiently positive, persistent nondiabetic, and prediabetic groups. These groups were compared for HLA genotypes, family history of T1DM, level of autoantibody positivity, and number of positive autoantibodies.
|
Statistical analysis
Statistical analyses were performed using SAS 8.0 (SAS Institute, Cary, NC). Proportions were compared using the 
2 test unless 20% of the cells had an expected value less than five, in which case Fisher’s exact test used. Analysis for levels of autoantibody was performed after log transformation to normalize the distribution. ANOVA with Bonferroni correction to control for multiple comparisons was used to compare transformed autoantibody levels in various study subgroups.
Survival analysis was performed using proportional hazards modeling with progression to diabetes or development of persistent positivity as the outcome variable in the model included positive family history of T1DM, HLA genotype, number of autoantibodies positive, and the highest level of first positive autoantibody. In the survival model, time to diabetes was calculated as the difference between diabetes development and first positive autoantibody. Time to persistent positivity was defined as the difference between last visit at which the individual was still positive and first positive autoantibody. The highest level of first positive autoantibody was determined for each individual at the visit in which the individual was positive for the first time for at least one autoantibody. When more than one autoantibody was positive at the same visit, the autoantibody with the highest level was chosen. Analysis for levels of autoantibody was performed after log transformation to normalize the distribution. Individuals with confirmed positive autoantibodies were included in the proportional hazards modeling.
Positive predictive values (PPVs) were calculated. Individuals included in this analysis were followed from 1.25 yr or less to ensure that they were identified as close to the initiation of autoantibody positivity. Levels were defined as positive at greater than or equal to 2, 5, or 10 SD above the mean. If multiple levels were positive at the first visit, the highest level was used.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
0.5% of GAA, 1.1% of IAA, and 0.1% of IA-2AA). Of those confirmed positive, in 50 children the autoantibodies were transient, in 58 persistently positive, and in four children the autoantibodies were positive once at the last or only visit and therefore could not be classified as transient or persistent. In the transiently positive group, 22 of 50 children were positive at more than one visit (mean, 3.3 visits), but only 2 of the 50 transiently positives were positive for more than one autoantibody. None of the transiently positive children has developed diabetes. In contrast, in the persistently positive group, 24 of 58 children have progressed to diabetes and 34 continue to be persistently positive but nondiabetic. The proportion of individuals with a first-degree T1DM relative was lower: 34% (17 of 50) in the false-positive children compared with 60% (30 of 50) in transiently positive and 76% (44 of 58) of the persistently positive group (P < 0.0001). The proportion of children with the highest risk HLA-DR3/4 DQ8 genotype differed in similar manner, with 20% of false positive, 20% of transiently positive, and 48% of persistently positive children carrying this genotype (P = 0.001).
Figure 3, A–C, shows the flow diagram for the autoantibody assays. One hundred five children were found to be IAA positive (Fig. 3A). Of these, 70 of 102 (69%) were confirmed positive on analysis of blinded aliquots (three samples had insufficient volume). Of children confirmed positive, 40 of 63 (63%) were positive on their subsequent visit. Thus, 43% (69% x 63%) were both confirmed positive and positive on the next visit.
|
), of which 64 of 96 (67%) were confirmed positive. Of those confirmed positive, 38 of 56 (68%) were positive on the next visit. Thus, 46% (67% x 68%) were both confirmed positive and positive on the next visit. Forty-eight children were found to be IA-2AA positive (Fig. 3C), of which 87% (39 of 45) were confirmed positive and 84% (26 of 31) were positive on the next visit. Thus, 73% (87% x 84%) were both confirmed positive and positive on the next visit. Rates of confirmation differed among the three autoantibodies: 87% IA-2AA, 67% GAA, and 69% IAA (P = 0.04). There was also a trend for IA-2AA to be more often positive at the next visit: 84% IA-2AA vs. 68% GAA and 63% IAA (P = 0.13).
For each of the three autoantibodies, presence of other autoantibodies was associated with confirmation. For IAA, 54% (38 of 70) of confirmed positive children had another autoantibody compared with 19% (six of 32) of those who did not confirm (P = 0.001). Similarly, 42% (27 of 64) of confirmed GAA-positive individuals and 85% (33 of 39) of confirmed IA-2AA-positive individuals had another positive autoantibody compared with 19% (six of 32) (P = 0.04) and 50% (three of six) (P = 0.08) of those who did not confirm positive for GAA and IA-2AA, respectively. Presence of the highest-risk HLA-DR3/4 DQ8 genotype was not significantly associated with confirmation for any of the autoantibodies. Those who confirmed for GAA tended to have more often a relative with T1DM compared with those that did not confirm (70% vs. 50%; P = 0.07); a similar pattern was observed for IAA (65% vs. 44%; P = 0.05).
Autoantibody levels on the first positive sample were log-transformed (Fig. 4) and compared, using ANOVA, among four groups: false positive, transiently positive, persistently nondiabetic, and prediabetic. IAA levels differed among these four groups (P < 0.0001), with those in the prediabetic, persistent nondiabetic, and transient groups higher than those in the false-positive group (P < 0.05 for pairwise comparisons). GAA levels also differed among the four groups (P < 0.0001) with prediabetic and persistent nondiabetic levels higher than transiently or false positive (P < 0.05) and no difference between the persistent nondiabetic and prediabetic groups. IA2-AA levels differed among the four groups (P = 0.03), whereas none of the pairwise comparisons reached significance.
|
In the proportional hazards modeling (Table 1), the HLA-DR3/4 DQ8 genotype predicted both persistence of autoantibodies and progression to diabetes. The level of the autoantibody at the initial sample and presence of another autoantibody predicted persistence.
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The expression of ICA and the biochemical autoantibodies GAA, IAA, and IA-2AA identify individuals at a high risk of T1DM. These autoantibodies can be detected during infancy (10, 11) and usually precede T1DM by many years (12, 13). Characteristics of the autoantibody expression have been used to predict T1DM. The persistent production of autoantibodies has been associated with T1DM. In the Finnish study, the Type 1 Diabetes Prediction and Prevention (DIPP), which identifies children at risk for T1DM with HLA screening (14), 15 children progressed to T1DM. None had reverted to negative autoantibodies before diabetes, and 14 of the 15 expressed IAA, which was usually the first autoantibody to develop (11). In our group, all children who developed diabetes were persistently positive for at least one autoantibody at the time of diagnosis with 23 of 24 positive for more than one autoantibody. Twenty-two of 24 were positive for IAA, 18 were positive for GAA, and 18 for IA-2AA. The German BABY-DIAB study (15, 16) has followed children of parents with T1DM for the development of islet autoimmunity and T1DM. Those children who went on to diabetes were positive for both ICA and IAA.
ICA levels have been associated with progression to diabetes (17, 18). Levels of ICA greater than 20 Juvenile Diabetes Foundation units were associated with a 53% 10-yr risk of diabetes, compared with a 43% risk for patients with levels greater than 10 Juvenile Diabetes Foundation units (1). Progression to diabetes has previously been associated with the number of biochemical autoantibodies and their levels (1, 2). A population-based study in school children found that higher autoantibody levels were associated with a higher specificity for diabetes, at the expense of sensitivity (2). Linking the levels of ICA and biochemical islet autoantibodies, levels of ICA were associated with the expression of multiple biochemical islet autoantibodies (19). Adding to the complexity of diabetes prediction, HLA has been associated with autoantibody level. HLA DR4 has previously been associated with insulin autoantibody level in the competitive insulin autoantibody assay (20). In our study, higher levels of each of the autoantibodies were associated with confirmation, persistence of autoantibodies, and development of T1DM. In particular, higher peak levels of IAA were associated with a greater progression to T1DM. In the proportional hazards model, the level of autoantibody was only predictive of persistent positivity, not reaching statistical significance to predict T1DM.
Autoantibody assays are problematic, especially when positive values are defined as above the 99th percentile in low previous probability populations (as in this study). This explains the high proportion of individuals with false-positive autoantibodies in this study (31%). Extensive checks and balances need to be in place to ensure the most accurate assessment of autoantibody level. In particular, before the first serum aliquot in DAISY is reported as positive, it is confirmed in separate assays with two of three positive tests defining positivity.
A group of particular interest is those that transiently express autoantibodies. This group has been identified in other studies, such as DIPP in which almost half of autoantibody-positive children reverted to negative (14), similar to our study in which 45% of confirmed autoantibodies were transiently positive. This group remains a mystery. It is possible that transient expression is due to assay variability. However, when considering the confirmatory process performed in the DAISY study and the fact that 22 of 50 individuals with transient autoantibodies were positive on multiple occasions (with at least four separate assays higher than the 99th percentile to be confirmed on any single visit), it is unlikely that assay variability alone accounts for transient positivity. In contrast, transient expression may represent "transient" islet cell autoimmunity that was somehow reversed. Identifying those that can reverse the autoimmune process may identify factors that may be critical to the development of diabetes. Transient expression of autoantibodies in childhood may be associated with the return of positive antibodies in later childhood and adulthood, or development of diabetes as an adult due primarily to ß-cell defects. Only with continued follow-up will the natural history of transiently positive individuals be clarified.
Of interest, in our group one individual who progressed to T1DM was positive for IAA and then negative at two visits and then returned to positivity before developing diabetes. Two individuals initially persistently positive were positive for IA-2AA, became negative, and then developed GAA in one case and GAA, IA-2AA, and IAA in the other. They still have not progressed to diabetes.
Although our study has been enrolling patients since 1993, allowing for a relatively long follow-up period, the time course of progression from autoimmunity to T1DM is over many years and maybe even decades. Therefore, many of the individuals identified as persistent positives in this study who have not yet developed diabetes may still go on to develop diabetes. The distinction between the persistent positive and diabetic group therefore may not be a qualitative one. The follow-up time of individuals who have developed diabetes is shorter than those who have remained persistently positive and not yet diabetic (3.49 vs. 5.49 yr; P = 0.002). Those who have progressed to clinical diabetes likely experience a more aggressive autoimmune attack of the pancreas.
Taken in aggregate, the use of the level of autoantibody can provide additional predictive information for the persistence of autoantibodies and development of T1DM and may be used to further stratify individuals for epidemiological and preventive studies.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: GAA, Antibodies to glutamic acid decarboxylase; HLA, human leukocyte antigen; IA-2AA, antibodies to protein tyrosine phosphatase IA2; IAA, antibodies to insulin; ICA, islet cell autoantibodies; PPV, positive predictive value; T1DM, type 1 diabetes mellitus.
Received October 30, 2003.
Accepted January 12, 2004.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. K. Steck and G. S. Eisenbarth Genetic Similarities Between Latent Autoimmune Diabetes and Type 1 and Type 2 Diabetes Diabetes, May 1, 2008; 57(5): 1160 - 1162. [Full Text] [PDF]
|
|
|
|
|
|
T. A. Aly, E. E. Baschal, M. M. Jahromi, M. S. Fernando, S. R. Babu, T. E. Fingerlin, A. Kretowski, H. A. Erlich, P. R. Fain, M. J. Rewers, et al. Analysis of Single Nucleotide Polymorphisms Identifies Major Type 1A Diabetes Locus Telomeric of the Major Histocompatibility Complex Diabetes, March 1, 2008; 57(3): 770 - 776. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. E. Baschal, T. A. Aly, S. R. Babu, M. S. Fernando, L. Yu, D. Miao, K. J. Barriga, J. M. Norris, J. A. Noble, H. A. Erlich, et al. HLA-DPB1*0402 Protects Against Type 1A Diabetes Autoimmunity in the Highest Risk DR3-DQB1*0201/DR4-DQB1*0302 DAISY Population Diabetes, September 1, 2007; 56(9): 2405 - 2409. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. F. Green Summary of Workgroup Meeting on Use of Family History Information in Pediatric Primary Care and Public Health Pediatrics, September 1, 2007; 120(SUPPLEMENT_2): S87 - S100. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. R. ROSE Prediction and Prevention of Autoimmune Disease: A Personal Perspective Ann. N.Y. Acad. Sci., August 1, 2007; 1109(1): 117 - 128. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. A. Aly, A. Ide, M. M. Jahromi, J. M. Barker, M. S. Fernando, S. R. Babu, L. Yu, D. Miao, H. A. Erlich, P. R. Fain, et al. Extreme genetic risk for type 1A diabetes PNAS, September 19, 2006; 103(38): 14074 - 14079. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. A. Deeg, D. Pompetzki, A. J. Raith, S. M. Hauck, B. Amann, S. Suppmann, T. W. F. Goebel, U. Olazabal, H. Gerhards, S. Reese, et al. Identification and Functional Validation of Novel Autoantigens in Equine Uveitis Mol. Cell. Proteomics, August 1, 2006; 5(8): 1462 - 1470. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. D. G. Leslie, R. Williams, and P. Pozzilli Type 1 Diabetes and Latent Autoimmune Diabetes in Adults: One End of the Rainbow J. Clin. Endocrinol. Metab., May 1, 2006; 91(5): 1654 - 1659. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. Larizza, V. Calcaterra, M. Martinetti, R. Negrini, A. De Silvestri, M. Cisternino, A. M. Iannone, and E. Solcia Helicobacter pylori Infection and Autoimmune Thyroid Disease in Young Patients: The Disadvantage of Carrying the Human Leukocyte Antigen-DRB1*0301 Allele J. Clin. Endocrinol. Metab., January 1, 2006; 91(1): 176 - 179. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Achenbach, E. Bonifacio, K. Koczwara, and A.-G. Ziegler Natural History of Type 1 Diabetes Diabetes, December 1, 2005; 54(suppl_2): S25 - S31. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Pihoker, L. K. Gilliam, C. S. Hampe, and A. Lernmark Autoantibodies in Diabetes Diabetes, December 1, 2005; 54(suppl_2): S52 - S61. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. J. Woodward and J. W. Thomas Multiple Germline {kappa} Light Chains Generate Anti-Insulin B Cells in Nonobese Diabetic Mice J. Immunol., July 15, 2005; 175(2): 1073 - 1079. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. BABAYA, M. NAKAYAMA, and G. S. EISENBARTH The Stages of Type 1A Diabetes Ann. N.Y. Acad. Sci., June 1, 2005; 1051(1): 194 - 204. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
