| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
H. Lee Moffitt Cancer Center and Research Institute (J.P.K., D.D.C.), University of South Florida, Tampa, Florida 33612-9497; Barbara Davis Center for Childhood Diabetes (L.Y., G.S.E.), University of Colorado, Denver, Colorado 80262; Department of Pediatrics (N.M.), Weill Medical College of Cornell University, New York, New York 10021; Joslin Diabetes Center (T.O., R.J.), Harvard Medical School, Boston, Massachusetts 02115; Departments of Pathology and Laboratory Medicine (W.E.W.) and Pediatrics (D.A.S.), University of Florida, Gainesville, Florida 32611; and Veterans Affairs Puget Sound Health Care System (J.P.P.), University of Washington, Seattle, Washington 98108
Address all correspondence and requests for reprints to: Jeffrey P. Krischer, Ph.D., H. Lee Moffitt Cancer Center and Research Institute, 12902 Magnolia Drive, Tampa, Florida 33612. E-mail: jpkrischer{at}moffitt.usf.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Relatives of subjects with type 1 diabetes were screened for ICA as part of the Diabetes Prevention Trial-Type 1. A total of 71,148 samples were also tested for GAA and ICA512AA. IAA results were available on 17,207 of these samples using a protein A/protein G MIAA assay as well. The study population was defined to be those in which all four antibodies were tested.
There were 1010 (5.9%) relatives with a single autoantibody on initial screening and 394 (2.3%) with two or more autoantibodies. GAA was more sensitive than ICA [GAA, 91% (357 of 394); ICA, 82% (324 of 394)] in the detection of multiple antibody-positive individuals. The addition of ICA512AA to GAA as a screening test increased sensitivity to 97% (381 of 394), whereas adding ICA512AA to ICA as a screening test increased sensitivity to 93% (367 of 394). GAA and ICA identified somewhat nonoverlapping subgroups of multiple antibody-positive subjects. Thus, the substitution of GAA or ICA for the other failed to detect 8–17% of multiple antibody subjects. Higher ICA titers were associated with increased percentages of multiple antibody-positive subjects; 86% of subjects having Juvenile Diabetes Foundation titers of at least 160 were positive for two or more antibodies. A screening strategy combining GAA and ICA512AA resulted in a higher sensitivity than using any marker individually, although statistically it was not significantly higher than using GAA alone.
Screening for any three antibodies guaranteed that all multiple antibody-positive subjects were detected. Screening for two antibodies at one time and testing for the remaining antibodies among those who are positive for one resulted in a sensitivity of 99% for GAA and ICA, 97% for GAA and MIAA or GAA and ICA512AA, 93% for ICA512AA and ICA, 92% for MIAA and ICA, and 73% for ICA512AA and MIAA.
From a laboratory perspective, screenings for GAA, ICA512AA, and MIAA are semiautomated tests with high throughput that, if used as initial screen, would identify at first testing 67% of the 2.3% of multiple antibody-positive relatives (100% if antibody-positive subjects are subsequently tested for ICA) as well as 4.7% of relatives with a single biochemical autoantibody, some of whom may convert to multiple autoantibody positivity on follow-up. Testing for ICA among relatives with one biochemical antibody would identify the remaining 33% of multiple antibody-positive relatives. Further follow-up and analysis of actual progression to diabetes will be essential to define actual diabetes risk in this large cohort.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Despite these observations, the pathogenesis of type 1 diabetes is still not completely understood. Occasionally, individuals lacking the HLA genetic susceptibility haplotypes develop diabetes, and those with the protective HLA go on to develop diabetes nonetheless. Those who develop diabetes usually display islet autoantibodies before disease onset (8). Yet, some individuals fail to present with diabetes-associated immunological markers at the time of diagnosis, suggesting that the markers may be transient or that some individuals never express the autoantibodies reacting with currently defined islet autoantigens.
It has also been recognized that the immunological markers convey different levels of diabetes risk. The presence of a single autoantibody may not appreciably increase risk, but the presence of multiple antibodies adds significantly to the risk of diabetes. As well, the autoantibodies differ in their prevalence, correlation with each other, and the level of risk they confer (2, 4, 9, 10, 11, 12, 13, 14).
Adding to these issues is the different reliability with which each autoantibody can be detected. International efforts have been established to standardize the sensitivity and specificity of assays and laboratories (15, 16). Finally, the different approaches to measuring autoantibodies have resulted in differing requirements for amounts of serum and assay cost. These observations have led to the need to develop the most accurate model of diabetes risk and to develop strategies by which to screen populations to identify individuals at high risk.
This study addresses screening strategies using immunological markers to identify individuals expressing multiple autoantibodies associated with type 1 diabetes. It is based upon the results of the Diabetes Prevention Trial-Type 1 (DPT-1) in which thousands of relatives of probands with type 1 diabetes have been screened for islet cell antibodies (ICA). As part of a related study, as of June 30, 2000, 71,148 of the screening samples were also tested for glutamic acid decarboxylase (GAD65) autoantibodies (GAA), and ICA512 autoantibodies (ICA512AA). Although we have previously reported (17) an association with eligibility for the DPT-1 high risk intervention studies, the risk of diabetes in this population is not known at this time because not all participants have been followed for an adequate length of time and because some of those who are both ICA and insulin autoantibody (IAA) positive are enrolled in a masked prevention trial that is still underway. Nonetheless, the literature supports the conclusion that those who are positive for multiple autoantibodies are at increased risk of developing diabetes compared with those who are antibody negative or who express only one antibody (Fig. 1
; Refs. 2 and 4). This analysis uses the identification of subjects with multiple autoantibodies to compare alternative screening strategies.
|
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The DPT-1 screened 71,148 first- and second-degree relatives of individuals with type 1 diabetes who have been diagnosed before the age of 40 yr. To be eligible for screening for cytoplasmic ICA, a relative must be age 45 or younger. The screening samples on 5,546 of the relatives who participated in the DPT-1 were also tested for IAA using a polyethylene glycol assay and on 17,207 relatives using a micro IAA (MIAA) assay. GAA and ICA512AA were also measured on the same 17,207 screening samples. Here, we report the results of the 17,207 relatives in which all four antibodies (ICA, GAA, MIAA, and ICA512AA) were tested.
ICA assay
Cytoplasmic ICA were measured on frozen sections of human pancreas by the DPT-1 ICA Core Laboratory (University of Florida, Gainesville, FL, February 1994 to September 1997, and October 1999 to present; and Research Institute for Children, New Orleans, LA, September 1997 to October 1999; Ref. 7). Ten or more Juvenile Diabetes Foundation (JDF) units was considered positive. In the Immunology of Diabetes Society (IDS) Combinatorial Autoantibody Workshop, reported in Orvieto, Italy, 1995, this ICA assay had a specificity of 100%, with a sensitivity of 74.4% in new onset patients younger than 30 yr.
GAA and ICA512AA assay
GAA and ICA512 autoantibodies were measured simultaneously by combined GAA and ICA512 radioassay as previously described in the DPT-1 GAA and ICA512AA Core Laboratory at the Barbara Davis Center (Ref. 18 ; full-length GAD65 and ICA512bdc cDNA clones). The assay was performed in 96-well filtration plates with autoantibody-bound 3H-GAD65 and 35S-ICA512AA precipitated with protein A Sepharose. The cut-points were set at indexes of 0.032 (mean ± 2 SD, GAA) and 0.049 (mean ± 6 SD, ICA512AA), the 99th percentile, respectively, of 198 normal controls. The interassay coefficients of variation of the assays are 6.5% and 9.6%, respectively, for GAA and ICA512AA assays. In the IDS Combinatorial Autoantibody Workshop for patients younger than 30 yr, assay specificity was 99% for GAA and 100% for ICA512AA, and sensitivity was 83.7% for GAA and 74.4% for ICA512AA.
IAA assay
IAA was determined with a fluid phase radioassay using polyethylene glycol precipitation in the DPT-1 IAA Core Laboratory at the Joslin Diabetes Center (Boston, MA; Ref. 19). The cut-point was 39 nU/ml (mean ± 2 SD), 99th percentile of 151 normal controls. The interassay coefficient of variation is 10.3% at low positive values. In the IDS Combinatorial Autoantibody Workshop, the assay had a specificity of 91% and sensitivity of 49%.
MIAA assay
Williams et al. (20) and Naserke et al. (21) described IAA assays that use either protein A-Sepharose or protein A/protein G-Sepharose. These assays use smaller volumes of serum and correlate well with large volume IAA assays. The MIAA assay in this study was done by G.S.E. at the Barbara Davis Center and by T.O. at the Joslin Diabetes Center with 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). The assay was tested in a series of samples from individuals with new onset diabetes, individuals at risk for type 1 diabetes, control subjects, and nonobese diabetic mice. As recently reported (22), this high throughput IAA assay, using 25 µl serum, has a sensitivity approaching standard assays using 600 µl serum and has a better specificity avoiding artifacts created by hemolyzed sera and cord blood. The upper limit of normal (0.010) was chosen as the 99th percentile from receiver operating characteristics curves in 106 healthy control subjects and 105 patients with new onset diabetes.
Statistical analysis
Categorical variables were analyzed using 
2 tests or Fisher’s exact tests, depending on cell size. Continuous variables were compared using the Student’s t tests, or the Wilcoxon rank sum test, depending on the distribution of the variable of interest. Statistical analyses were performed using SAS software (SAS Institute, Inc., Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
).
|
).
|
Although both GAA and ICA are present in a high proportion of multiple antibody-positive subjects, they identified somewhat different populations (Table 3). Of the 394 multiple autoantibody-positive subjects, 33 (8.4%) were positive for ICA but not for GAA, and 66 (16.8%) were positive for GAA but not for ICA. The nonoverlapping cases are different, depending upon the autoantibody used for screening. Four additional cases were positive for ICA512AA and MIAA, but not for GAA or ICA.
|
). Screening with GAA and ICA512AA or GAA and MIAA detects an equivalent number of multiple antibody-positive subjects (96.7%), but the GAA and ICA512AA combination results in 24% fewer (769 vs. 1007) subjects that need subsequent testing, because more are multiple antibody-positive with the screening sample alone. Taking into account the subjects who are positive for both of the screening antibodies, then the strategy of screening with GAA and ICA512AA requires subsequent testing on only 574 (769 minus 195) compared with 888 (1007 minus 119) subjects with GAA and MIAA (35% fewer who need subsequent testing).
|
If screening or subsequent testing for ICA is dropped, then the best screening strategy that includes the remaining three autoantibodies can detect at most 66.8% of multiple autoantibody-positive subjects.
Increasing JDF titer is associated with an increased percentage of multiple antibody-positive subjects (Table 5). At JDF values of 160 or greater, 86% of ICA-positive subjects are positive for multiple antibodies, which would obviate the necessity for further antibody testing. However, titers in this range account for only 28% of ICA-positive subjects. Raising the ICA threshold to 20 JDF units would miss 32 of the 394 multiple antibody-positive subjects (8%) and reduce the sensitivity to only 74% for this marker used as a single screening antibody.
|
The polyethylene glycol-based IAA test was evaluated in an initial cohort of 5546 subjects. Although it identified 83% of multiple antibody-positive relatives, it was found to be positive in 16% of subjects as a single autoantibody, suggesting an unacceptable false positive rate in this large multicenter study. The high prevalence of IAA (18.5%) in this population, coupled with the observation that IAA was positive as a single antibody in the vast majority (84.5%) of these cases, raised concerns that the assay was yielding more false positives than expected from studies of comparable populations, perhaps due to the analysis of hemolyzed sera.
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
, ICA and GAA identified nonoverlapping subpopulations of multiple antibody-positive subjects. Subjects who were multiple antibody-positive but not ICA-positive would fail to be detected if GAA were omitted from the screening tests (n = 66), and those that are multiple antibody-positive but GAA-negative would not be detected if ICA were omitted from the screening tests (n = 33). Those who would not be detected by the omission of ICA were more likely to be siblings of diabetic probands (P = 0.01) and were younger (P = 0.001). Hence, they tended to be at higher risk on the basis of the natural history of type 1 diabetes. However, this analysis cannot conclude that one group is more likely to develop diabetes than the other (at higher risk), in part due to the paucity of data on diabetes risk among ICA-negative subjects. The results presented in this study assume testing all subjects for a marker(s) and then testing those who are positive for all the remaining markers. An alternative strategy would be to eliminate ICA entirely. This would not only limit the population to those multiple antibody-positive subjects who express GAA, but also miss subjects who are GAA-negative and ICA-positive. This effectively eliminates 131 subjects, thereby reducing the sensitivity of this strategy to only 67%, which is less than screening with ICA. Of these 131 subjects, 105 have GAA and ICA as their only autoantibodies, 13 have ICA and ICA512AA, and 13 have ICA and MIAA. In that GAA are a major component of ICA, the diabetes risk in subjects who have GAA and ICA as their only autoantibodies is not addressed in this study.
Finally, the sensitivity of GAA and ICA512AA combined (96.7%) is similar to the strategy of MIAA and GAA (96.7%), ICA and ICA512AA (93.2%), GAA and ICA (99.0%), and screening for all three biochemical autoantibodies (100%). Of course, these conclusions are subject to revision when follow-up studies, currently under way by the DPT-1 study group, can estimate the risk of diabetes in multiple antibody-positive subgroups. Until such time, we rely on the studies of others in smaller populations to establish diabetes risk. For future trials in which screening is necessary to identify a population with sufficient diabetes risk, GAA and ICA512AA would augment the population identified by ICA (or vice versa). In addition, there is a clear potential role for all the markers. A strategy to screen all subjects for GAA and ICA512AA and test those positive for one antibody for ICA and IAA is consistent with the results of smaller studies (23, 24). In terms of the ease of testing, a strategy that measures GAA, ICA512AA, and MIAA is attractive, but at a cost of adding to the number of subjects that need subsequent testing for ICA. It is likely that the biochemical autoantibody assays can be improved and in particular, IA-2ic autoantibodies can identify a small subset of autoantibodies distinct from ICA512AA (measured with the ICA512bdc construct, an alternative splice variant joining exon 12 to 14; Ref. 25) and may be required for optimal sensitivity. The MIAA assay, although much more specific compared with the polyethylene glycol based-IAA assay, had a lower sensitivity in identifying multiple autoantibody-positive relatives, suggesting that this assay might also be improved.
We have used the terms specificity and sensitivity in terms of single antibody or multiple autoantibody detection, with comparison to prior studies. We believe sensitivity and specificity will need to be defined in terms of actual progression to diabetes as the DPT-1 screened cohorts are followed for diabetes risk, and the DPT-1 oral trial is concluded.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Abbreviations: DPT-1, Diabetes Prevention Trial-Type 1; GAA, GAD65 autoantibodies; GAD65, glutamic acid decarboxylase; HLA, human leukocyte antigen; IAA, insulin autoantibodies; ICA, islet cell antibodies; ICA512AA, ICA512 autoantibodies; JDF, Juvenile Diabetes Foundation; MIAA, micro IAA.
Received May 15, 2002.
Accepted September 19, 2002.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  K. C. Herold, B. Brooks-Worrell, J. Palmer, H. M. Dosch, M. Peakman, P. Gottlieb, H. Reijonen, S. Arif, L. M. Spain, C. Thompson, et al. Validity and Reproducibility of Measurement of Islet Autoreactivity by T-Cell Assays in Subjects With Early Type 1 Diabetes Diabetes, November 1, 2009; 58(11): 2588 - 2595. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Sosenko, J. P. Palmer, L. Rafkin-Mervis, J. P. Krischer, D. Cuthbertson, J. Mahon, C. J. Greenbaum, C. C. Cowie, J. S. Skyler, and the Diabetes Prevention Trial-Type 1 Study Group Incident Dysglycemia and Progression to Type 1 Diabetes Among Participants in the Diabetes Prevention Trial-Type 1 Diabetes Care, September 1, 2009; 32(9): 1603 - 1607. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Yu, A. K. Steck, S. Babu, L. Yu, D. Miao, K. McFann, J. Hutton, G. S. Eisenbarth, and G. Klingensmith Single Nucleotide Transcription Factor 7-Like 2 (TCF7L2) Gene Polymorphisms in Antiislet Autoantibody-Negative Patients at Onset of Diabetes J. Clin. Endocrinol. Metab., February 1, 2009; 94(2): 504 - 510. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Fineberg, T. T. Kawabata, D. Finco-Kent, R. J. Fountaine, G. L. Finch, and A. S. Krasner Immunological Responses to Exogenous Insulin Endocr. Rev., October 1, 2007; 28(6): 625 - 652. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. Wang, D. Miao, S. Babu, J. Yu, J. Barker, G. Klingensmith, M. Rewers, G. S. Eisenbarth, and L. Yu Prevalence of Autoantibody-Negative Diabetes Is Not Rare at All Ages and Increases with Older Age and Obesity J. Clin. Endocrinol. Metab., January 1, 2007; 92(1): 88 - 92. [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]
|
|
|
|
|
|
J. P. Palmer, C. S. Hampe, H. Chiu, A. Goel, and B. M. Brooks-Worrell Is Latent Autoimmune Diabetes in Adults Distinct From Type 1 Diabetes or Just Type 1 Diabetes at an Older Age? Diabetes, December 1, 2005; 54(suppl_2): S62 - S67. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Lernmark Type 1 Diabetes -- Does Suppressing T Cells Increase Insulin? N. Engl. J. Med., June 23, 2005; 352(25): 2642 - 2644. [Full Text] [PDF]
|
|
|
|
|
|
M. A. Atkinson Thirty Years of Investigating the Autoimmune Basis for Type 1 Diabetes: Why Can't We Prevent or Reverse This Disease? Diabetes, May 1, 2005; 54(5): 1253 - 1263. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. S. Harris Vitamin D in Type 1 Diabetes Prevention J. Nutr., February 1, 2005; 135(2): 323 - 325. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. A. Rossini Autoimmune Diabetes and the Circle of Tolerance Diabetes, February 1, 2004; 53(2): 267 - 275. [Abstract] [Full Text]
|
|
|
|
|
|
P. Achenbach, K. Warncke, J. Reiter, H. E. Naserke, A. J.K. Williams, P. J. Bingley, E. Bonifacio, and A.-G. Ziegler Stratification of Type 1 Diabetes Risk on the Basis of Islet Autoantibody Characteristics Diabetes, February 1, 2004; 53(2): 384 - 392. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. P. Palmer, G. A. Fleming, C. J. Greenbaum, K. C. Herold, L. D. Jansa, H. Kolb, J. M. Lachin, K. S. Polonsky, P. Pozzilli, J. S. Skyler, et al. C-Peptide Is the Appropriate Outcome Measure for Type 1 Diabetes Clinical Trials to Preserve {beta}-Cell Function: Report of an ADA Workshop, 21-22 October 2001 Diabetes, January 1, 2004; 53(1): 250 - 264. [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 |
) of type 1 diabetic probands and among school children (
) with no family history of diabetes (