OBJECTIVE

We evaluated a novel electrochemiluminescent assay for insulin/proinsulin autoantibodies (ECL-IAA) as a new marker of the onset of islet autoimmunity and as a predictor of type 1 diabetes.

RESEARCH DESIGN AND METHODS

The Diabetes Autoimmunity Study in the Young (DAISY) prospectively follows children at increased genetic risk for development of islet autoimmunity (defined as presence of autoantibodies to insulin, GAD65, IA-2, or zinc transporter 8 [ZnT8]) and type 1 diabetes (general population of children and first-degree relatives). Serial serum samples from subjects who progressed to type 1 diabetes and who had their first islet autoantibodies measured by age 18 months (N = 47) were tested using ECL-IAA.

RESULTS

Almost all prediabetic children tested positive for ECL-IAA (46 of 47, 98%) during follow-up. ECL-IAA was almost always the first autoantibody to appear (94% total; 21% very first [by itself]; 23% with only mIAA; 19% with another islet autoantibody [GAD or ZnT8]; and 30% with ≥2 other antibodies [mIAA, GAD, IA-2, or ZnT8]). Among the 46 subjects who were ECL-IAA positive, ECL-IAA antedated the onset of other islet autoantibodies by a mean of 2.3 years (range, 0.3–7.2 years). Both the age of appearance of autoantibody and IAA levels (but not GAD65, IA2, or ZnT8 levels) are major determinants of the age of diabetes onset.

CONCLUSIONS

This new ECL-IAA assay defines more precisely the onset of prediabetic autoimmunity and may help identify events triggering islet autoimmunity, as well as allow earlier intervention for type 1 diabetes. Nearly all young children progressing to diabetes are insulin autoantibody positive.

Anti-islet autoimmunity currently detected by measurement of islet autoantibodies almost always precedes by years the development of type 1A diabetes. If the autoimmunity is triggered by time-correlated factors such as acute viral infections, then the discovery of pathogenic viruses may depend on accurate timing of the appearance of islet autoantibodies. Most of the trials to prevent type 1A diabetes target persons in the preclinical phase of the disease marked by the presence of persistent islet autoantibodies (1). Because this preclinical period is quite variable, accurate prediction of the time to progression to overt diabetes is critical for the design and implementation of preventive trials. Although genetic markers can identify varying risk, it is only once autoimmunity has begun (marked by the presence of multiple autoantibodies to pancreatic β-cell antigens) that a high positive predictive value (>90%) can be achieved. Multiple autoantibodies are present in the majority of prediabetic individuals (24). Screening for risk of type 1 diabetes uses “biochemical” autoantibody assays for specific islet autoantigens (1). These include insulin autoantibodies (IAA) (5), GAD65 (6), protein tyrosine IA-2 (ICA512) (7), and, most recently, zinc transporter 8 (ZnT8) (8). Individuals having a single positive autoantibody (insulin, GAD65, IA-2, or ZnT8 autoantibodies) are at low risk for progression to diabetes, whereas individuals expressing two or more positive autoantibodies, especially on multiple tests over time, are at very high risk for progression to diabetes (9,10).

IAA are usually extremely high at the onset of diabetes in young children but usually negative in individuals first presenting with diabetes after age 12 years. There is a log-linear inverse relationship between these levels and the age of onset of diabetes (11), as well as between levels of IAA and time of progression from first appearance of islet autoantibodies to diagnosis of diabetes in prospectively followed Diabetes Autoimmunity Study in the Young (DAISY) children (10). We recently have reported (12) development of an electrochemiluminescence assay for IAA (ECL-IAA) using Meso Scale instrumentation and ruthenium-labeled proinsulin. This assay detects high-affinity IAA and is more sensitive than the micro-IAA (mIAA) radioassays in the last Diabetes Autoantibody Standardization Program (DASP) workshop, yet is equally specific.

In this study we evaluated ECL-IAA as a new marker of the onset of islet autoimmunity and as a predictor of progression to diabetes among antibody-positive subjects. We found that this novel nonradioactive IAA assay is more sensitive and defines the timing of the initial autoantibody appearance earlier than the previously used mIAA radioassay. We report the predictors of progression to diabetes and the determinants of age at diagnosis among children at high risk participating in the prospective DAISY.

Study population

Since 1993, DAISY has followed two cohorts of young children at increased risk for type 1 diabetes, including a cohort of relatives of type 1 diabetic patients (siblings and offspring) and the general population newborn cohort. The latter consists of children with type 1 diabetes–susceptible HLA-DR/DQ genotypes identified through screening of >31,000 newborns at St. Joseph’s Hospital in Denver, Colorado. The details of screening and follow-up have been previously published. Briefly, autoantibodies were tested at 9, 15, and 24 months and annually thereafter or at their first visit and annually thereafter if the child was enrolled after birth. Children who tested autoantibody positive were given an accelerated testing schedule of every 3–6 months. A total of 47 DAISY children who had development of type 1 diabetes and had their first islet autoantibodies measured by age 18 months were included in this study. Subjects were tested for mIAA, GAD65, IA2, and ZnT8 antibodies at each visit. For ECL-IAA, all first autoantibody-positive samples, all previous samples, and yearly follow-up samples until diabetes were tested. The mean age of diabetes onset was 7.7 years (range, 0.9–14.4 years). Informed consent was obtained from the parents of each study subject. The Colorado Multiple Institutional Review Board approved all study protocols.

Islet autoantibodies

Measurement of biochemical islet autoantibodies was performed in the laboratory of Dr. George Eisenbarth at the Barbara Davis Center for Childhood Diabetes in Denver, Colorado. We used radioimmunoassays for autoantibodies to insulin, GAD65, IA-2, and ZnT8 using methods previously described (1,13). The interassay coefficients of variation were 10 and 5% (n = 50) for GAA and IA-2A, respectively. In the 2009 DASP workshop, the sensitivity and specificity were 72 and 99%, respectively, for GAA, and 62 and 99%, respectively, for IA-2A. For ZnT8A, the interassay coefficient of variation was 10.4% (n = 15), and the sensitivity and specificity were 62 and 99%, respectively, in the 2009 DASP workshop. IAA were measured by a mIAA assay. The interassay coefficient of variation was 20% (n = 100) at low positive levels, and the sensitivity and specificity for mIAA were 48 and 100%, respectively, in the 2009 DASP workshop. The ECL-IAA using Meso Scale instrumentation and ruthenium-labeled proinsulin is a nonradioactive IAA assay in which bivalent IAA cross-link two “insulin” moieties in a fluid phase. The IAA in serum will link the Sulfo-tagged proinsulin to the biotinylated proinsulin, which will be captured on the solid phase of the streptavidin-coated plate. Detection of plate-captured Sulfo-tagged proinsulin is accomplished with electrochemiluminescence. IAA levels from ECL-IAA assay and our current radioassay (mIAA) were compared in 150 DASP subjects; the two assays were correlated (P < 0.0001), but ECL-IAA assay had higher sensitivity (68 vs. 56%) with the same specificity of 99% for both assays (12).

Statistical analysis

Statistical analyses were performed using PRISM and SAS software. The IAA, GAA, IA-2, and ZnT8 levels were log-transformed for analyses. Negative or zero values were changed to 0.001. Proportions were compared using χ2 or Fisher exact test. Multiple linear regression was used to evaluate potential predictors of age of diabetes onset, including age at first positive antibody, initial number of positive antibodies, family history, high-risk HLA-DR3/4, ethnicity, and IAA, GAD, IA-2, and ZnT8 levels (both initial and mean levels) in subjects who had their first autoantibody measurements before age 1.5 years and had progression to diabetes (N = 47). A two-tailed P value with an α level for significance was set at 0.05.

All 47 subjects with progression to diabetes were positive for at least one of the four biochemical antibodies (IAA, GAA, IA-2, and ZnT8). All subjects except one were positive for ECL-IAA (46 of 47, 98%) during follow-up compared with those positive for mIAA assay (42 of 47, 89%). The four prediabetic individuals who tested negative for mIAA but positive for ECL-IAA all had negative mIAA levels for multiple samples over time, although they had several positive ECL-IAA titers over time. On their first antibody-positive visit, 28% subjects had one positive antibody, whereas 43% had two, and 29% subjects had three or more positive antibodies, respectively. ECL-IAA was the most frequent antibody (94%) present at the first positive antibody visit, followed by mIAA (47%), GAD (36%), ZnT8 (26%), and IA2 (11%) (P < 0.0001). Table 1 demonstrates initial antibody detection in these 47 prediabetic DAISY subjects; ECL-IAA almost always was the first autoantibody detected: 94% (44 of 47) total; 21% alone (by itself); 23% with only mIAA; 19% with another islet autoantibody (GAD or ZnT8); and 30% with two or more other antibodies (mIAA, GAD, IA-2, or ZnT8). Among the 46 subjects who were ECL-IAA positive, the ECL-IAA detected onset of islet autoimmunity earlier than any of the four traditional biochemical antibodies in 10 (22%) children; i.e., ECL-IAA was the only positive antibody, whereas all four biochemical antibodies measured by standard radioassays were negative. For those with ECL-IAA as the first antibody by itself, ECL-IAA antedated the onset of other islet autoantibodies by a mean of 2.3 years (range, 0.3–7.2 years). Once positive, ECL-IAA titers tend to be relatively stable overall over time; i.e., most subjects found to be positive for ECL-IAA will stay positive for ECL-IAA over time, although there are individual variations.

Table 1

Initial antibody detection in DAISY prediabetic subjects (N = 47)

Initial antibody detection in DAISY prediabetic subjects (N = 47)
Initial antibody detection in DAISY prediabetic subjects (N = 47)

We additionally tested for ECL-IAA in 1,812 antibody-negative DAISY subjects and 14 subjects who had development of diabetes with negative biochemical antibodies (i.e., all four biochemical antibodies negative). ECL-IAA is very specific, and only 5 of 1,812 (0.3%) antibody-negative subjects tested positive for ECL-IAA. Table 2 gives the sensitivity, specificity, and positive and negative predictive values for diabetes outcome for this new ECL-IAA assay in the DAISY cohort. When looking at these results, one should take into account that these values were calculated in cohorts of children at increased risk for type 1 diabetes, a cohort of relatives of type 1 diabetes patients, and a general population cohort selected for type 1 diabetes high-risk HLA-DR/DQ genotypes. Among 14 antibody-negative DAISY subjects with type 1 diabetes, two were positive for ECL-IAA; i.e., ECL-IAA picked up an additional 14% of cases of diabetes. We also tested ECL-IAA among remaining nondiabetic antibody-positive DAISY subjects. Among 32 DAISY subjects positive for mIAA alone, only nine (28%) were positive for ECL-IAA. However, ECL-IAA was positive in 18 of 22 (82%) DAISY subjects who tested positive for mIAA and one or more other antibodies (GAD65, IA2, or ZnT8).

Table 2

Sensitivity, specificity, and positive and negative predictive values for ECL-IAA in the DAISY cohort of children at increased risk for diabetes

Sensitivity, specificity, and positive and negative predictive values for ECL-IAA in the DAISY cohort of children at increased risk for diabetes
Sensitivity, specificity, and positive and negative predictive values for ECL-IAA in the DAISY cohort of children at increased risk for diabetes

As previously published, levels of mIAA, but not GAD or IA-2 autoantibodies levels, can predict age of diagnosis of type 1 diabetes (10). In this study, we confirm that higher mean ECL-IAA levels were correlated with faster progression to diabetes from first appearance of islet autoantibodies (P = 0.02), whereas levels of ZnT8, GAD, or IA-2 autoantibodies did not correlate with time to progression to diabetes (Fig. 1).

Figure 1

Age of onset of diabetes correlates with ECL-IAA levels, but not with ZnT8, GAD, or IA-2 levels. Analysis performed for all subjects who had their first antibody measurement before 1.5 years of age and who had progression to diabetes (N = 47). Antibody levels were log10-transformed for analyses. Negative or zero values were changed to 0.001 (which corresponds to −3 on log scale). Time to diabetes (years) is from first positive antibody visit (any of the 5 antibodies).

Figure 1

Age of onset of diabetes correlates with ECL-IAA levels, but not with ZnT8, GAD, or IA-2 levels. Analysis performed for all subjects who had their first antibody measurement before 1.5 years of age and who had progression to diabetes (N = 47). Antibody levels were log10-transformed for analyses. Negative or zero values were changed to 0.001 (which corresponds to −3 on log scale). Time to diabetes (years) is from first positive antibody visit (any of the 5 antibodies).

Close modal

Multiple linear regression was used to evaluate potential predictors of age of diabetes onset, including age at first positive antibody, initial number of positive antibodies, family history, high-risk HLA-DR3/4, ethnicity, and IAA, GAD, IA-2, and ZnT8 levels (both initial and mean antibody levels). For the first model including initial antibody levels, only initial mIAA levels and age at first positive antibody were significant predictors of diabetes onset age (Fig. 2A). When analyzing the mean levels of autoantibodies over time for each subject, mean ECL-IAA levels and age of the first positive antibody were significant predictors of age of diabetes diagnosis (Fig. 2B).

Figure 2

Predicted age of onset of diabetes. Multiple linear regression was used to evaluate potential predictors of age of diabetes onset, including age at first positive antibody (Ab), initial number of positive antibodies, family history, high-risk HLA-DR3/4, ethnicity, and IAA, GAD, IA-2, and ZnT8 levels (both initial and mean antibody levels). For the first model including initial antibody levels, only initial mIAA levels (loginitialmIAA) and age at first positive antibody (agefirstAb+) were significant predictors of diabetes onset age (A). When analyzing mean antibody levels (B), diabetes onset age was best predicted by mean ECL-IAA levels (logmeanECL-IAA) and age at first positive antibody (agefirstAb+).

Figure 2

Predicted age of onset of diabetes. Multiple linear regression was used to evaluate potential predictors of age of diabetes onset, including age at first positive antibody (Ab), initial number of positive antibodies, family history, high-risk HLA-DR3/4, ethnicity, and IAA, GAD, IA-2, and ZnT8 levels (both initial and mean antibody levels). For the first model including initial antibody levels, only initial mIAA levels (loginitialmIAA) and age at first positive antibody (agefirstAb+) were significant predictors of diabetes onset age (A). When analyzing mean antibody levels (B), diabetes onset age was best predicted by mean ECL-IAA levels (logmeanECL-IAA) and age at first positive antibody (agefirstAb+).

Close modal

In the prospective DAISY cohort, ECL-IAA antedated the onset of islet autoimmunity by a mean of 2.3 years in the 22% children initially positive only for ECL-IAA. Results from this study indicate that this novel ECL-IAA assay not only is more sensitive but also defines the timing of the initial autoantibody appearance earlier than the previously used mIAA radioassay. This earlier detection in timing of onset of islet autoimmunity is of importance in finding potential environmental causes of diabetes and in our understanding of the etiology of type 1 diabetes. However, assays obviating the use of radioactive isotopes also would help to disseminate islet autoantibody testing possibly to point of care. Finally, this novel nonradioactive IAA assay seems to be more disease specific; ECL-IAA is typically positive when two or more autoantibodies are present (i.e., subjects at high risk for development of diabetes), whereas the majority of mIAA-positive-only (single antibody–positive) nondiabetic subjects are ECL-IAA negative and these IAA are demonstrated as low-affinity antibodies. We speculate that the difference between these two assays is antibody affinity binding. These factors are critical in moving from prevention trials in first-degree relatives to the general population, in which 90% of type 1 diabetes cases are found.

Several studies now have shown that once multiple autoantibodies appear (two or more autoantibodies), almost all these young children have progression to diabetes, although some progression will take >10 years (10,14,15). In children expressing two or more autoantibodies, this progression to diabetes is not influenced by family history, and HLA-DR3/4-DQB1*0302 genotype is only an additional predictor of progression to diabetes in children expressing one or two autoantibodies but not among patients expressing three autoantibodies (10). However, factors influencing rate of progression to diabetes are still largely unknown. In the TrialNet study, a Diabetes Prevention Trial–Type 1 Risk Score threshold of 9.00 identifies individuals who are very likely to have progression to diabetes within 2 years (16). However, the Diabetes Prevention Trial–Type 1 Risk Score includes C-peptide and glucose indexes from oral glucose tolerance testing. In the DAISY study, we previously have shown that age of first positive antibody and mean mIAA levels (but not GAD65 or IA-2) are major determinants of age of onset of diabetes (10). In this study, we confirm that ECL-IAA levels (but not ZnT8) are major determinants of age of onset of diabetes. The unique association of age of diagnosis with levels of IAA perhaps is not unexpected given accumulating evidence that targeting of insulin may be a primary determinant of diabetes of the spontaneous NOD mouse model (17) and the association of polymorphisms of the insulin gene with diabetes risk in humans. For instance, knocking out endogenous insulin genes and mutating a single amino acid of insulin prevents all diabetes of NOD mice (18,19), whereas knocking out either GAD65 or IA-2 has no influence on progression to diabetes of NOD mice (20,21). The mechanism underlying the specific association of levels of IAA with rate of progression to diabetes is not defined. Autoantibodies or the B cells that produce them may play a pathogenic role with evidence from the NOD mouse model (22) and human trials of rituximab (23). Alternatively, insulin autoantibody levels simply may reflect activity or number of CD4 T cells targeting insulin.

This new nonradioactive ECL-IAA assay can define earlier onset of prediabetic autoimmunity and thus may help identify time-dependent events triggering islet autoimmunity. Nearly all young children with progression to diabetes are insulin autoantibody positive, and understanding the underlying mechanism of elevated insulin antibodies may contribute to design of future therapies.

This research was supported by National Institutes of Health grants R37 DK32493, DK32083, DK050979, AI050864, and N01 AI15416. A.K.S. was supported by the JDRF Early Career Patient Oriented Research Award 11-2010-206.

No potential conflicts of interest relevant to this article were reported.

L.Y. researched data, contributed to discussion, and reviewed and edited the manuscript. F.D. and J.M.W. researched data and reviewed and edited the manuscript. D.M. and A.R.F. researched data. A.K.S. wrote the manuscript and contributed to discussion. A.K.S. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Parts of this study were presented at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, Pennsylvania, 8–12 June 2012.

The authors thank the late Dr. George S. Eisenbarth (Barbara Davis Center for Childhood Diabetes, University of Colorado Denver, Aurora, Colorado) for inspiration to carry on this work.

1.
Yu L, Cuthbertson DD, Maclaren N, et al. Expression of GAD65 and islet cell antibody (ICA512) autoantibodies among cytoplasmic ICA+ relatives is associated with eligibility for the Diabetes Prevention Trial-Type 1. Diabetes 2001;50:1735–1740
2.
Orban
T
,
Sosenko
JM
,
Cuthbertson
D
, et al
Diabetes Prevention Trial-Type 1 Study Group
.
Pancreatic islet autoantibodies as predictors of type 1 diabetes in the Diabetes Prevention Trial-Type 1
.
Diabetes Care
2009
;
32
:
2269
2274
[PubMed]
3.
Siljander HT, Simell S, Hekkala A, et al. Predictive characteristics of diabetes-associated autoantibodies among children with HLA-conferred disease susceptibility in the general population. Diabetes 2009;58:2835–2842
4.
Bingley PJ, Gale EA. Progression to type 1 diabetes in islet cell antibody-positive relatives in the European Nicotinamide Diabetes Intervention Trial: the role of additional immune, genetic and metabolic markers of risk. Diabetol 2006;49:881–890
5.
Greenbaum
CJ
,
Palmer
JP
,
Kuglin
B
,
Kolb
H
.
Insulin autoantibodies measured by radioimmunoassay methodology are more related to insulin-dependent diabetes mellitus than those measured by enzyme-linked immunosorbent assay: results of the Fourth International Workshop on the Standardization of Insulin Autoantibody Measurement
.
J Clin Endocrinol Metab
1992
;
74
:
1040
1044
[PubMed]
6.
Baekkeskov
S
,
Aanstoot
H-J
,
Christgau
S
, et al
.
Identification of the 64K autoantigen in insulin-dependent diabetes as the GABA-synthesizing enzyme glutamic acid decarboxylase
.
Nature
1990
;
347
:
151
156
[published erratum appears in Nature 1990;347(6295):782]
[PubMed]
7.
Gianani R, Rabin DU, Verge CF, et al. ICA512 autoantibody radioassay. Diabetes 1995;44:1340–1344
8.
Wenzlau
JM
,
Juhl
K
,
Yu
L
, et al
.
The cation efflux transporter ZnT8 (Slc30A8) is a major autoantigen in human type 1 diabetes
.
Proc Natl Acad Sci USA
2007
;
104
:
17040
17045
[PubMed]
9.
Verge
CF
,
Gianani
R
,
Kawasaki
E
, et al
.
Number of autoantibodies (against insulin, GAD or ICA512/IA2) rather than particular autoantibody specificities determines risk of type I diabetes
.
J Autoimmun
1996
;
9
:
379
383
[PubMed]
10.
Steck
AK
,
Johnson
K
,
Barriga
KJ
, et al
.
Age of islet autoantibody appearance and mean levels of insulin, but not GAD or IA-2 autoantibodies, predict age of diagnosis of type 1 diabetes: diabetes autoimmunity study in the young
.
Diabetes Care
2011
;
34
:
1397
1399
[PubMed]
11.
Vardi
P
,
Ziegler
AG
,
Mathews
JH
, et al
.
Concentration of insulin autoantibodies at onset of type I diabetes. Inverse log-linear correlation with age
.
Diabetes Care
1988
;
11
:
736
739
[PubMed]
12.
Yu L, Miao D, Scrimgeour L, Johnson K, Rewers M, Eisenbarth GS. Distinguishing persistent insulin autoantibodies with differential risk: nonradioactive bivalent proinsulin/insulin autoantibody assay. Diabetes 2012;61:179–186
13.
Wenzlau
JM
,
Moua
O
,
Sarkar
SA
, et al
.
SlC30A8 is a major target of humoral autoimmunity in type 1 diabetes and a predictive marker in prediabetes
.
Ann N Y Acad Sci
2008
;
1150
:
256
259
[PubMed]
14.
Siljander HT, Veijola R, Reunanen A, Virtanen SM, Akerblom HK, Knip M. Prediction of type 1 diabetes among siblings of affected children and in the general population. Diabetol 2007;50:2272–2275
15.
Knip
M
,
Korhonen
S
,
Kulmala
P
, et al
.
Prediction of type 1 diabetes in the general population
.
Diabetes Care
2010
;
33
:
1206
1212
[PubMed]
16.
Sosenko
JM
,
Skyler
JS
,
Mahon
J
, et al
Type 1 Diabetes TrialNet Study Group
Diabetes Prevention Trial-Type 1 Study Group
.
The application of the diabetes prevention trial-type 1 risk score for identifying a preclinical state of type 1 diabetes
.
Diabetes Care
2012
;
35
:
1552
1555
[PubMed]
17.
Nakayama
M
,
Abiru
N
,
Moriyama
H
, et al
.
Prime role for an insulin epitope in the development of type 1 diabetes in NOD mice
.
Nature
2005
;
435
:
220
223
[PubMed]
18.
Babaya N, Nakayama M, Moriyama H, et al. A new model of insulin-deficient diabetes: male NOD mice with a single copy of Ins1 and no Ins2. Diabetol 2006;49:1222–1228
19.
Nakayama
M
,
Babaya
N
,
Miao
D
, et al
.
Long-term prevention of diabetes and marked suppression of insulin autoantibodies and insulitis in mice lacking native insulin B9-23 sequence
.
Ann N Y Acad Sci
2006
;
1079
:
122
129
[PubMed]
20.
Kash
SF
,
Condie
BG
,
Baekkeskov
S
.
Glutamate decarboxylase and GABA in pancreatic islets: lessons from knock-out mice
.
Horm Metab Res
1999
;
31
:
340
344
[PubMed]
21.
Kubosaki A, Miura J, Notkins AL. IA-2 is not required for the development of diabetes in NOD mice. Diabetol 2004;47:149–150
22.
Hu
CY
,
Rodriguez-Pinto
D
,
Du
W
, et al
.
Treatment with CD20-specific antibody prevents and reverses autoimmune diabetes in mice
.
J Clin Invest
2007
;
117
:
3857
3867
[PubMed]
23.
Pescovitz
MD
,
Greenbaum
CJ
,
Krause-Steinrauf
H
, et al
Type 1 Diabetes TrialNet Anti-CD20 Study Group
.
Rituximab, B-lymphocyte depletion, and preservation of beta-cell function
.
N Engl J Med
2009
;
361
:
2143
2152
[PubMed]