Abstract
Background
Older studies of diabetes development typically utilized a 7-day incubation polyethylene glycol competitive insulin autoantibody assay (CIAA). Our standard micro-IAA assay (mIAA) utilizes precipitation with proteins A/G and 1-day incubation (1-day mIAA), but is less sensitive compared to the CIAA assay.
Methods
We performed CIAA and mIAA assays in various conditions. We analyzed serum samples from 446 type 1 diabetes patients, from another set of 247 type 1 diabetes patients within 2 weeks of initiation of insulin treatment, from 150 healthy control donors, from 22 healthy participants in the diabetes autoimmunity study in the young (DAISY), and also coded sera from 50 patients with newly diagnosed type 1 diabetes and 50 blood donor control samples.
Results
In the process of our study, we found that the key condition was the incubation time. Therefore, we extended the incubation time to 7 days (7-day mIAA assay). No CIAA-negative control was positive with either 1-day or 7-day mIAA. In a new onset type 1 diabetes and at risk cohorts (DAISY study), the 7-day mIAA identified an additional 18% as being positive along with 16% of those who were initially 1-day mIAA negative and CIAA positive. Most subjects detectable only with the 7-day mIAA assay had intermediate levels of CIAA (80–300 nU/mL) (p = 0.01).
Conclusions
The 7-day mIAA assay identifies a small but significant additional subset of individuals positive on the CIAA assay, while preserving specificity.
Keywords: CIAA, mIAA, DAISY, type 1 diabetes, incubation time, RIA
Introduction
Type 1 diabetes is characterized by autoimmune destruction of the pancreatic β-cells that may begin years before the development of symptoms of diabetes. Although type 1 diabetes is considered to be a T-cell-mediated autoimmune disease, both B-cells and their product, autoantibodies, contribute to the pathogenesis and are markers of the autoimmune process [1,2]. Prior to the development and at the diagnosis of diabetes, subjects may have several different anti-islet autoantibodies. To date, antibodies to insulin (IAA), glutamic acid decarboxylase (GAD-AA), and the protein tyrosine phosphatase insulinoma antigen-2 (IA2-AA) can be detected by biochemical assays. These autoantibodies can be present years before the onset of diabetes, and progression to diabetes is associated with the presence of multiple autoantibodies that persist over time [3], providing crucial information about the natural history of type 1 diabetes and helping in the identification of subjects at risk for development of overt diabetes for intervention trials [4,5].
To detect IAA, there are two common radioimmunoassays (RIA). Early studies of prediabetes utilized a large volume of serum with a 7-day incubation, in competition with unlabeled insulin and polyethylene glycol precipitation (CIAA) competitive insulin autoantibodies [6]. The 99th percentile was 42 nU/mL. A higher cut-off (80 nU/mL) was used for enrollment in the earlier phases of the diabetes prevention trial – type 1 oral insulin trial [7]. Our subsequent studies modifying the assay of Williams and coworkers [8] utilized a small volume of the serum that is termed ‘micro-IAA assays (mIAA)’ and that utilizes precipitation with protein A/protein G (protein A/G) [9] and a 1-day incubation (1-day mIAA). The 1-day mIAA assay in our studies has had an overall lower sensitivity compared to the CIAA assay and with striking, uncommon sera with a high level in the CIAA assay, and lack of signal with the mIAA format. Given the importance of determining insulin autoantibodies with sensitivity and specificity, we have explored potential etiologies of discrepant results in patients with new onset type 1 diabetes or in samples from the diabetes autoimmunity study in the young (DAISY), a prospective study of islet autoimmunity.
Materials and methods
Subjects
Serum samples were collected from 446 type 1 diabetes patients, from another set of 247 type 1 diabetes patients within 2 weeks of initiation of insulin treatment, and from 150 healthy control donors. Serum samples were also collected from 22 healthy DAISY participants. As previously described [3], DAISY is a prospective cohort study that has enrolled and followed young first-degree relatives of patients with type 1 diabetes, and children identified through newborn screening for HLA-DR, DQ genotypes associated with type 1 diabetes. We also used coded sera from 50 patients with newly diagnosed type 1 diabetes and 50 blood donor control samples obtained from the diabetes antibody standardization program (DASP) [10].
The Combined Institutional Review Board at the University of Colorado Health Sciences Center has approved the DAISY project. Informed consent was obtained from all donors, or from their parents. Serum samples were stored at −20°C prior to testing.
CIAA assay
CIAA was measured in a standard RIA [9] incorporating competition with unlabeled insulin with a requirement of 600 μL of serum (150 μL duplicates with and without unlabeled insulin). Human 125I-insulin of 20 000 cpm was incubated with serum for 7days at 4 °C, precipitated with polyethylene glycol, and counted in a γ-counter. The interassay coefficient of variation was 10.3% at low positive levels [11]. The cut point was 42 nU/mL, which represented the 99th percentile of 198 normal control subjects. In the 1995 Immunology of Diabetes Society combinatorial workshop, the assay had a specificity of 98% and a sensitivity of 56%. The IAA testing is not repeated, even if the levels are above the 99th percentile because of limitation of serum volume.
mIAA assay
As previously described [9], the mIAA was performed using a 96-well filtration plate–based RIA. The assay requires 26 μL of serum (6.5 μL duplicates with and without unlabeled insulin). Human 125I-insulin of 20 000 cpm was incubated with serum overnight (1-day mIAA) or for 7 days (7-day mIAA) at 4 °C, precipitated with protein A/G sepharose, and counted in a TopCount β-counter. 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 most recent 2005 DASP workshop, sensitivity was 58% and specificity was 99%. The intra- and interassay coefficient of variation was 12 and 16% respectively.
To test the hypothesis that a unique isotype of IAA is recognized in sera that are CIAA (+) and mIAA (−), two selected sera were incubated with different secondary antibodies (anti-human IgG, IgM, IgA, IgD, and IgE antibodies), after incubating with I125-insulin or cold insulin, and before incubating with protein A/G.
mIAA affinity assay
Affinity was measured by competitive binding experiments. Briefly, 5 μL of serum was incubated in duplicate for 7 days at 4 °C in the presence of human 125I-insulin with or without increasing quantities of unlabeled human insulin (2.2 × 10−5 to 2.7 × 10−11 mol/L) in a final volume of 55 μL. Immune complexes were precipitated using protein A/G-Sepharose, and washed. Bound 125I-insulin was measured using a TopCount β-counter. Kd values (mol/L) were calculated using the GraphPad Prism 3 program (GraphPad Software Inc., San Diego, California, USA), and IAA affinity was expressed as reciprocal Kd values (L/mol). IAA capacity was expressed as max binding (Bmax).
Statistics
The difference between the two groups was tested by Mann–Whitney’s U-test. Fisher’s exact test was used for categorical variables. A p-value of less than 0.05 was regarded as being significant.
Results
We analyzed 446 serum samples from patients with type 1 diabetes (Figure 1A) with the mIAA and CIAA assays. For mIAA, 211/446 (47.3%) of patients with type 1 diabetes were positive, whereas for CIAA 274/446 (61.4%) were positive (p < 0.001), suggesting that the mIAA assay was less sensitive compared to the CIAA assay. The assays gave concordant results for 76.0% of the samples (339/446). Few sera were positive in the mIAA assay and negative in the CIAA assay (mIAA (+), CIAA (−): 22/446 (4.9%)), and these were all low positive in the mIAA assay. More common were sera positive in the CIAA assay and negative with the mIAA format (mIAA (−), CIAA (+): 85/446 (19.1%)). Furthermore, we found some samples with an extremely high level of CIAA that were mIAA (−).
Figure 1.
Serum samples from 446 type 1 diabetes patients (A) and from 247 new onset type 1 diabetes patients (B) were analyzed with mIAA and CIAA assays. The levels of IAA determined with these two methods were very well correlated. However, the mIAA assay was less sensitive compared to the CIAA assay. Horizontal line: cut-off for mIAA (0.01). Vertical line: cut-off for CIAA (40 nU/mL)
Polyethylene glycol in the CIAA assay can precipitate all immunoglobulin (and many proteins), but protein A/G in the mIAA assay can bind only subsets of immunoglobulin. Therefore, we hypothesized that a unique isotype of IAA (not IgG1, e.g. IgE) was underlying the above differences. To test this hypothesis, we first confirmed the existence of uncommon sera CIAA (+) and mIAA (−) using 247 new onset type 1 diabetes patients. As shown in Figure 1B, 194/247 (78.5%) new onset patients with type 1 diabetes were concordant (positive or negative). We found 44 samples, CIAA (+) and mIAA (−) (17.8%), whereas only 9 samples were mIAA (+) and CIAA (−) (3.6%). To confirm CIAA (+) and mIAA (−) in these 44 samples, we repeated the mIAA assay, and 34 samples remained mIAA negative (Figure 2). Samples (N = 10) that became positive in the second mIAA assay had significantly higher CIAA levels compared to those in samples that remained negative (p < 0.0001, Mann–Whitney U-test).
Figure 2.
Forty-four samples detected in Figure 1(B) were repeated by the mIAA assay, and 34 samples remained mIAA negative. Samples (N = 10) that became positive in second mIAA assay had a significantly higher CIAA level compared to samples that remained negative (p < 0.0001, Mann–Whitney U-test). Arrows are selected sera for studying the unique isotype of IAA
To test our hypothesis that a unique isotype of IAA is recognized in sera that are CIAA (+) and mIAA (−), two sera were selected (Figure 2, arrows; CIAA >160 nU/mL) and incubated with different secondary antibodies (antihuman IgG, IgM, IgA, IgD, and IgE antibodies). The use of antihuman IgG as a secondary antibody increased the sensitivity of the assay; and, antihuman IgM, IgA, IgD, and IgE antibodies also increased the sensitivity. One major difference between the original mIAA assay and this modified mIAA assay was the extended incubation period of the serum with labeled insulin. Therefore, this gave rise to a new hypothesis that the extended incubation time improved the sensitivity for mIAA, particularly for CIAA (+), mIAA (−) sera.
Given this hypothesis, we analyzed control sera (n = 150), new onset type 1 diabetic patient sera (n = 247), and sequential prediabetic sera (DAISY) (n = 22). We compared a 1-day mIAA assay and 7 days of incubation with sera and I125-insulin or cold insulin (7-day mIAA). As shown in Figure 3, 3.4 and 3.3% of control samples were mIAA positive in the 1-day mIAA and 7-day mIAA, respectively, with no significant difference between the two assays. In the 247 samples with new onset type 1 diabetes, for samples previously found to be both CIAA and 1-day mIAA positive, 83.0% were positive in 1-day mIAA and 89.4% in 7-day mIAA (not significantly different). For samples CIAA (+) and 1-day mIAA (−), the 7-day assay improved sensitivity (22.7% positive in the 1-day assay and 52.3% positive in the 7-day assay) (p < 0.01). All 147 samples, CIAA (−) and 1-day mIAA (−), were negative on repeat testing with both the 1-day mIAA and 7-day mIAA.
Figure 3.
Comparison with 1-day mIAA assay and 7-day mIAA. Healthy control samples showed 3.4 and 3.3% of mIAA positive in the 1-day mIAA and 7-day mIAA respectively. In the 247 samples with new onset type 1 diabetes, only for samples previously found to be CIAA (+) and 1-day mIAA (−), the 7-day assay improved sensitivity (22.7 and 52.3% positive, respectively) (p < 0.01). Horizontal line: cut-off for mIAA (0.01)
Figure 4 illustrates the relationship between CIAA and remeasurement 1-day mIAA or 7-day mIAA. Of the 247 subjects analyzed, 91 (36.8%) were positive for CIAA. Of these 91 subjects, 42 (46.2%) were negative on the 1-day mIAA; however, the 7-day mIAA identified an additional 15 (16.5% of all CIAA (+)) subjects as being positive. The CIAA levels of these 7-day (+) but 1-day (−) were between 57–308 nU/mL. Even when using higher cut-off of CIAA (80 nU/mL), the 7-day mIAA assay improved the sensitivity.
Figure 4.
The relationship between CIAA (lower (A) and higher (B) cut-off) and remeasurement 1-day mIAA or 7-day mIAA. The 7-day mIAA assay improved the sensitivity both in lower and higher cut-off of CIAA. Horizontal line: cut-off for mIAA (0.01). Vertical line: cut-off for CIAA (40 or 80 nU/mL)
In DAISY, 22 samples were identified as being confirmed CIAA (+) and 1-day mIAA (−). Four out of 22 (18.2%) of these samples were positive with the 7-day assay and negative with the 1-day mIAA assay (CIAA range of those changing to positive was 91–350 nU/mL). In a set of workshop-coded samples provided by DASP, the sensitivity was 58% and the specificity was 99% in 1-day mIAA. Similarly, in 7-day mIAA, the sensitivity was 60% and the specificity was 98%.
Having found a set of samples positive with the 7-day mIAA but negative with the 1-day mIAA, we analyzed the samples by competition with different concentrations of unlabeled insulin. Samples that were positive with the 7-day mIAA but negative with the 1-day mIAA showed significantly higher affinity, and lower capacity and lower [affinity × capacity] than samples positive with both assays. The CIAA level was linearly related to [affinity × capacity] (r= 0.85, p < 0.01).
Discussion
Insulin is an important target of autoimmunity that leads to type 1 diabetes [12,13] and IAA is one of the earliest indicators of anti-islet autoimmunity. In the NOD mouse, IAA can be detected as early as 3 weeks of age. In humans, IAA with GAD-AA and IA2-AA are used for the diagnosis and prediction of type 1 diabetes [14]. However, the concordance of mIAA assays in international workshops is poor compared to that of GAD-AA and IA2-AA assays [10,15,16]. Also, we found sera to be positive in the CIAA assay and negative with the mIAA format, suggesting the opportunity to improve the sensitivity of the mIAA assay. In the present study, we did not find any unique isotypes of IAA, but we found that, for a subset of sera, sensitivity of mIAA assays could be improved by using a lengthened incubation time.
The prevention of type 1 diabetes has been the subject of several large-scale trials including the diabetes prevention trial – type 1. To date, there has been no effective method of preventing diabetes. However, as we learn more about the pathogenesis of type 1 diabetes, strategies to prevent diabetes may be identified and implemented in clinical trials. The ability to accurately identify islet autoimmunity is of paramount importance for the success of these treatment trials. The initial treatment trials have targeted subjects after the onset of islet autoimmunity. Additionally, trials will be implemented prior to the development of islet autoimmunity and follow subjects for islet autoimmunity. In children, insulin autoantibodies are often the initial autoantibodies present. Therefore, assays that more accurately identify subjects with insulin autoantibodies without increasing the false-positive rate are critically needed. Through a fairly simple maneuver of increasing the incubation time for the mIAA assay, the sensitivity improves from 20.6% (51/247) to 26.3% (65/247) without changing the specificity. At this stage, however, 7-day mIAA is still slightly less sensitive compared to CIAA, though more sensitive than standard mIAA. The major advantage of the 7-day mIAA is the lower serum volume.
To study the characteristics of specific samples, namely 7-day mIAA (+) and 1-day mIAA (−), we analyzed autoantibody affinity. Those samples that are only positive with the 7-day assay had a slightly higher affinity and much lower capacity, compared to both 7-day mIAA (+) and 1-day mIAA (+) samples. The IAA level has not been correlated with IAA affinity [17], but has been correlated with [Affinity × Capacity] as shown in this and previous studies [18]. Thus, the levels of anti-insulin autoantibodies reflect both the apparent affinity and capacity of antibody binding.
High CIAA (CIAA ≥80 nU/mL) and low (CIAA ≥42 nU/mL) cut-offs have been used as entry criteria for the oral insulin arm of the diabetes prevention trial – type 1 [7]. As shown in Figure 4, the 7-day mIAA identified 60/65 (92%) of samples with CIAA ≥80 nU/mL as positive and 180/182 (99%) with CIAA <80 nU/mL as mIAA negative. As shown in Table 1, more subjects with an intermediate level of CIAA (80–300) could be detected with the 7-day mIAA (p = 0.01). Samples with high levels of CIAA (>300) were almost always positive on 1-day and 7-day mIAA (94 and 97% mIAA (+)).
Table 1.
Difference in sensitivity between 1-day and 7-day mIAA assays on different CIAA levels
| CIAA level (nU/mL) | mIAA assay | Positive | Negative | p-value |
|---|---|---|---|---|
| >42 and ≤80 | 1-day | 0 | 26 | 0.11 |
| 7-day | 4 | 22 | ||
| >80 and ≤300 | 1-day | 20 | 14 | 0.01 |
| 7-day | 30 | 4 | ||
| >300 | 1-day | 29 | 2 | 1.00 |
| 7-day | 30 | 1 |
Statistical analysis was performed by Fisher’s exact test.
In conclusion, our study suggests that the proportion of potentially false-negative IAA, namely CIAA (+) and regular mIAA (−), samples might not be worth considering, but the 7-day mIAA identifies a small but significant additional subset of individuals without sacrificing specificity.
Acknowledgements
This work was supported by grants from the National Institutes of Health (DK32083, DK55969, DK62718), Diabetes Endocrine Research Center (P30 DK57516), the American Diabetes Association, the Juvenile Diabetes Foundation, and the Children’s Diabetes Foundation. M.N. was supported by a fellowship from the Juvenile Diabetes Foundation.
Footnotes
Conflict of Interest
None declared.
References
- 1.Serreze DV, Fleming SA, Chapman HD, Richard SD, Leiter EH,Tisch RM. B lymphocytes are critical antigen-presenting cells for the initiation of T cell-mediated autoimmune diabetes in nonobese diabetic mice. J Immunol 1998; 161: 3912–3918. [PubMed] [Google Scholar]
- 2.Robles DT, Eisenbarth GS, Dailey NJ, Peterson LB, Wicker LS. Insulin autoantibodies are associated with islet inflammation but not always related to diabetes progression in NOD congenic mice. Diabetes 2003; 52: 882–886. [DOI] [PubMed] [Google Scholar]
- 3.Barker JM, Barriga KJ, Yu L, et al. Prediction of autoantibody positivity and progression to type 1 diabetes: Diabetes Autoimmunity Study in the Young (DAISY). J Clin Endocrinol Metab 2004; 89: 3896–3902. [DOI] [PubMed] [Google Scholar]
- 4.Aly TA, Ide A, Humphrey K, et al. Genetic prediction of autoimmunity: Initial oligogenic prediction of anti-islet autoimmunity amongst DR3/DR4-DQ8 relatives of patients with type 1A diabetes. J Autoimmun 2005; 25(Suppl): 40–45. [DOI] [PubMed] [Google Scholar]
- 5.Babaya N, Nakayama M, Eisenbarth GS. The stages of type 1Adiabetes. Ann N Y Acad Sci 2005; 1051: 194–204. [DOI] [PubMed] [Google Scholar]
- 6.Kurtz AB, Matthews JA, Mustaffa BE, Daggett PR, Nabarro JD. Decrease of antibodies to insulin, proinsulin and contaminating hormones after changing treatment from conventional beef to purified pork insulin. Diabetologia 1980; 18: 147–150. [DOI] [PubMed] [Google Scholar]
- 7.Skyler JS, Krischer JP, Wolfsdorf J, et al. Effects of oral insulin in relatives of patients with type 1 diabetes: The Diabetes Prevention Trial – Type 1. Diabetes Care 2005; 28: 1068–1076. [DOI] [PubMed] [Google Scholar]
- 8.Williams AJ, Bingley PJ, Bonifacio E, Palmer JP, Gale EA. Anovelmicro-assay forinsulin autoantibodies. JAutoimmun 1997; 10: 473–478. [DOI] [PubMed] [Google Scholar]
- 9.Yu L, Robles DT, Abiru N, et al. Early expression of antiinsulin autoantibodies of humans and the NOD mouse: evidence for early determination of subsequent diabetes. Proc Natl Acad Sci U S A 2000; 97: 1701–1706. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Bingley PJ, Bonifacio E, Mueller PW. Diabetes Antibody Standardization Program: first assay proficiency evaluation. Diabetes 2003; 52: 1128–1136. [DOI] [PubMed] [Google Scholar]
- 11.Vardi P, Dib SA, Tuttleman M, et al. Competitive insulin autoantibody assay. Prospective evaluation of subjects at high risk for development of type I diabetes mellitus. Diabetes 1987; 36: 1286–1291. [DOI] [PubMed] [Google Scholar]
- 12.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. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kent SC, Chen Y, Bregoli L, et al. Expanded T cells from pancreatic lymph nodes of type 1 diabetic subjects recognize an insulin epitope. Nature 2005; 435: 224–228. [DOI] [PubMed] [Google Scholar]
- 14.Verge CF, Gianani R, Kawasaki E, et al. Prediction of type I diabetes in first-degree relatives using a combination of insulin, GAD, and ICA512bdc/IA-2 autoantibodies. Diabetes 1996; 45: 926–933. [DOI] [PubMed] [Google Scholar]
- 15.Verge CF, Stenger D, Bonifacio E, et al. Combined use of autoantibodies (IA-2 autoantibody, GAD autoantibody, insulin autoantibody, cytoplasmic islet cell antibodies) in type 1 diabetes: Combinatorial Islet Autoantibody Workshop. Diabetes 1998; 47: 1857–1866. [DOI] [PubMed] [Google Scholar]
- 16.Franke B, Galloway TS, Wilkin TJ. Developments in the prediction of type 1 diabetes mellitus, with special reference to insulin autoantibodies. Diabetes Metab Res Rev 2005; 21: 395–415. [DOI] [PubMed] [Google Scholar]
- 17.Achenbach P, Koczwara K, Knopff A, Naserke H, Ziegler AG,Bonifacio E. Mature high-affinity immune responses to (pro)insulin anticipate the autoimmune cascade that leads to type 1 diabetes. J Clin Invest 2004; 114: 589–597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Castano L, Ziegler AG, Ziegler R, Shoelson S, Eisenbarth GS. Characterization of insulin autoantibodies in relatives of patients with type I diabetes. Diabetes 1993; 42: 1202–1209. [DOI] [PubMed] [Google Scholar]