Anti-islet autoantibodies serve as key markers in immune-mediated type 1 diabetes (T1D) and slowly progressive T1D (SPIDDM), also known as latent autoimmune diabetes in adults (LADA). Autoantibodies to insulin (IAA), glutamic acid decarboxylase (GADA), tyrosine phosphatase-like protein IA-2 (IA-2A), and zinc transporter 8 (ZnT8A) are currently employed in the diagnosis, pathological analysis, and prediction of T1D. GADA can also be detected in non-diabetic patients with autoimmune diseases other than T1D and may not necessarily reflect insulitis. Conversely, IA-2A and ZnT8A serve as surrogate markers of pancreatic β-cell destruction. A combinatorial analysis of these four anti-islet autoantibodies demonstrated that 93–96% of acute-onset T1D and SPIDDM cases were diagnosed as immune-mediated T1D, while the majority of fulminant T1D cases were autoantibody-negative. Evaluating the epitopes and immunoglobulin subclasses of anti-islet autoantibodies help distinguish between diabetes-associated and non-diabetes-associated autoantibodies and is valuable for predicting future insulin deficiency in SPIDDM (LADA) patients.
Name of Antigen | Localization | Function | Reference |
---|---|---|---|
Insulin | Insulin secretory granules | Regulate glucose levels in the blood and induce glucose storage in the liver, muscles, and adipose tissue | [7] |
GAD65 | Synaptic-like vesicles in the cytoplasm of β-cells | Rate-limiting enzyme engaged in the synthesis of the neurotransmitter γ-aminobutyric acid from L-glutamate | [8] |
GAD67 | Cytosol of β-cells | Rate-limiting enzyme engaged in the synthesis of the neurotransmitter γ-aminobutyric acid from L-glutamate | [11] |
IA-2 | Insulin secretory granule membrane | Regulate insulin secretory granule content and β-cell growth | [9][12][9,12] |
Phogrin/IA-2β | Insulin secretory granule membrane | Regulate insulin secretory granule content and β-cell growth | [13][14][13,14] |
Carboxypeptidase H | Insulin secretory granules and granule membrane | Convert proinsulin into insulin and C-peptide by catalyzing the release of C-terminal arginine or lysine residues from polypeptides | [15] |
ICA69 | Insulin secretory granule membrane | Dense-core vesicles signaling and maturation | [16] |
ZnT8 | Insulin secretory granule membrane | Transport zinc ion from the cytosol into the insulin secretory granules | [17][18][17,18] |
GM2-1 ganglioside | Secretory granules in β-cells and non-β-cells | unknown | [19] |
Heat shock protein 60 | Insulin secretory granules | Assist correct folding of partially folded polypeptides and presentation of antigen to MHC molecules | [20] |
GLUT2 | β-cell surface membrane | Uptake glucose from the blood into β-cells | [21] |
Tetraspanin-7 | Insulin secretory granule membrane | Regulate Ca2+-dependent insulin exocytosis | [22] |
ICA12/SOX13 | Cytoplasm and nucleus in β-cells and non-β-cells | Transcription factor (Function in the islets is unknown) | [23] |
Subject | Prevalence |
---|---|
Healthy control | <1% |
Acute-onset type 1 diabetes (at onset) | 60–80% |
Fulminant type 1 diabetes | 5–9% |
LADA (SPIDDM) | 100% |
Type 2 diabetes (diet/OHA) | 4–5% |
Polyglandular autoimmune syndrome, type 1 | 30–40% |
Polyglandular autoimmune syndrome, type 2 | 30–50% |
Autoimmune thyroid disease | 6–8% |
Stiff-person syndrome | 60–70% |
IA-2A is present in 60-70% of prediabetic relatives and new-onset patients with T1D. We and others have analyzed IA-2A epitopes recognized by diabetic sera using a series of IA-2 fragments or IA-2/phogrin chimeric proteins, and found that the major epitopes are localized in the cytoplasmic domain [51]. Approximately 95% of T1D patients and prediabetic relatives who are IA-2A positive recognize the PTP-like domain (amino acids 687-979), whereas only 5% of sera react with the luminal domain [51, 52]. Furthermore, our binding and competition analysis using multiple IA-2/phogrin chimeric constructs demonstrated that a major unique epitope for IA-2A is localized to amino acids 762-887. A conformational epitope associated with the C-terminal 31 amino acids of IA-2 is recognized by one-third of sera, and a minor epitope is located on amino acids 601-762 of IA-2. Notably, intramolecular epitope spreading was found for relatives of T1D patients who later progressed to T1D. However, relatives who remained nondiabetic exhibited a decrease in the number of recognized epitopes. These studies are consistent with the hypothesis that IA-2 may be recognized as a consequence of β-cell destruction [33].
Another important epitope has been mapped in the juxta-membrane domain of IA-2 (amino acids 601-629; IA-2JM). Our data demonstrated that the age of disease onset in patients with IA-2JMA only was significantly higher than that in patients who reacted with the PTP-like domain, suggesting that autoantibody recognition of IA-2 epitopes in autoimmune diabetes is associated with the age of disease onset, which may reflect the intensity of the β-cell destruction process [53].
7.4. ZnT8 autoantibodies
In 2007, Sladek and coworkers identified four loci containing variants that confer T2D risk through a genome-wide association study, including a non-synonymous polymorphism in the ZnT8 gene (SLC30A8), rs13266634 (C/T), which causes an R325W modification in the protein sequence [54]. In the same year, Hutton and coworkers discovered ZnT8 as a major autoantigen in T1D, and ZnT8A has been recognized as one of the four major anti-islet autoantibodies [23].
ZnT8A are present in 50-60% of prediabetic relatives and new-onset patients with T1D. As shown in Figure 4C, ZnT8 is a 369-amino acid polytopic transmembrane protein with cytoplasmic N- and C- terminal tails. It has been reported that ZnT8A recognizes 101 amino acids localized in the cytoplasmic C-terminal region. In particular, the amino acid residue 325 (R325W) defined by the SLC30A8 polymorphism is critical for humoral autoimmunity to this autoantigen, and binding of ZnT8A against two isotypes (ZnT8-325R, ZnT8-325W) depends on the patient's SLC30A8 genotype [55, 56]. Consequently, heterozygotes with the CT genotype respond to both ZnT8-325R and ZnT8-325W, while CC and TT homozygotes respond exclusively to ZnT8-325R or ZnT8-325W, respectively. Thus, individuals respond to endogenous ZnT8 protein determined by their own genome, and therefore, the current ZnT8A assay, therefore, uses a hybrid protein of two ZnT8 isotypes as antigens.
Furthermore, Wenzlau and coworkers identified that residues 332R, 333E, 336K, and 340K contribute to a conformational ZnT8A epitope independent of residue 325 by comparing human and mouse chimeric ZnT8 proteins [57], suggesting that this epitope may add to the diagnostic utility of measuring ZnT8A.
Figure 4. Illustration of antigenic epitopes recognized by T1D sera in GAD65 (A), IA-2 (B), and ZnT8 (C) proteins.
7.5. Other anti-islet autoantibodies
Other anti-islet autoantibodies include autoantibodies against GM2-1 ganglioside, HSP60, GLUT2, tetraspanin-7, and ICA12/SOX13 (Table 1). Among these, the epitopes of GM2-1 autoantibodies and GLUT2 autoantibodies have not been analyzed so far. It has been reported that HSP60 autoantibodies recognized two epitope regions on HSP60 (amino acids 394-413 and amino acids 435-454). The first region similar to the sequence found in GAD, whereas the second one overlaps with p277 T-cell epitope to a large extent [58]. Using a series of overlapping peptide fragments, Eugster and coworkers mapped map autoepitopes recognized by tetraspsnin-7 autoantibodies and found that autoantibody epitopes lie predominantly within the first and third cytoplasmic domains of the protein. Further characterization of autoantibody binding to mutated constructs revealed that epitopes lie within a relatively short (20-amino acid) region represented by at least two of the three cytoplasmic domains, providing further evidence of the importance of protein conformation in antibody binding [59]. Furthermore, epitope mapping of ICA12/SOX13 autoantibodies using several truncated fragments of SOX13 suggests that autoantibodies are directed to at least two epitopes, one that requires amino acids 66-604, and a second confined within amino acids 327-604 [22].
SPIDDM (also known as LADA) is characterized by the presence of anti-islet autoantibodies and a gradual decline in insulin secretory capacity. The Immunology of Diabetes Society defined LADA as follows; 1) onset of diabetes >35 years, 2) positive test for at least one of the known anti-islet autoantibodies, and 3) requirement of insulin treatment >6 months after the diagnosis of diabetes [60]. LADA encompasses anti-islet autoantibody-positive diabetic patients in both insulin-dependent and non-insulin-dependent states, which is nearly identical to SPIDDM. According to the recently revised diagnostic criteria for SPIDDM, the interval from diabetes diagnosis to the requirement of insulin treatment is >3 months [61]. Additionally, patients with exhausted endogenous insulin secretion (Fasting C-peptide <0.6ng/ml) at the last observed time point are defined as “SPIDDM (definite)”. In contrast, anti-islet autoantibody-positive patients in non-insulin-dependent state are classified as "SPIDDM (probable)" (Table 3). Using this diagnostic criterion, measuring anti-islet autoantibodies other than GADA results in an approximately 3-fold increase in the incidence of SPIDDM among non-insulin-treated diabetic patients compared with measuring GADA alone (2.0-2.4% vs. 7-8%) [62-64]. Indeed, in the Nagasaki Autoimmune Diabetes Intervention/Prevention Study, the prevalence of anti-islet autoantibodies other than GADA in insulin naïve adult-onset diabetes was 8.6%, which is 2.6-fold compared to that of GADA (3.3%) (Figure 5). Since this subtype of T1D is generally indistinguishable from T2D at the time of diagnosis, measuring anti-islet autoantibodies is crucial for early diagnosis and appropriate treatment of SPIDDM (LADA).
Figure 5. Prevalence of GADA, IA-2A, ZnT8A, and IAA in 788 insulin naïve adult-onset patients with diabetes
Anti-islet autoantibody positivity, especially ICA and GADA, is predictive for progression to a future insulin-dependent state after the diagnosis of diabetes. For example, the UKPDS (United Kingdom Prospective Diabetes Study) found that at least 50% of LADA patients required insulin treatment 6 years post-diagnosis [65]. However, not all SPIDDM (LADA) patients required insulin treatment, even after 10 years from diagnosis.
According to a nationwide survey [66] conducted by the Japan Diabetes Society, the predictors of progression to insulin dependent state include (1) age of onset ≤ 47 years, (2) period until GADA positive detection ≤ 5 years, (3) GADA titer (RIA method) ≥ 13.6 U/ml, and (4) fasting C-peptide ≤ 0.65 ng/mL. Additionally, the number of positive anti-islet autoantibodies and GADA epitope recognition are also important for prediction. To identify the predictive markers for early insulin requirement in non-insulin-dependent SPIDDM (probable), we evaluated IAA, IA-2A, and ZnT8A along with GADA-specific epitope recognition in 47 GADA-positive diabetic patients [63]. Among these patients, 38% had one or more of IAA, IA-2A, or ZnT8A and 15% had two or more of these autoantibodies. A high GADA titer (≥ 10U/mL), the presence of GADA-E1, and the presence of one or more among IAA, IA-2A, or ZnT8A at diagnosis marked the risk for early insulin therapy requirement. Furthermore, multiple anti-islet autoantibodies were the most relevant risk factor for the insulin requirement (odds ratio 13.77; 95% CI 2.77-68.45; P<0.001) in a multivariate logistic regression analysis. Therefore, measuring anti-islet autoantibodies other than GADA and ICA is essential for predicting the progression risk of SPIDDM (LADA) patients.
Clinicians should be aware that complications arising from additional autoimmune disorders are more frequently observed in T1D patients [67], with the most common organ-specific autoimmune disease associated with T1D being autoimmune thyroid disease (AITD), such as Graves’ disease and Hashimoto’s thyroiditis, affecting more than 90% of people with T1D and autoimmune disorders [68, 69]. Additionally, it has been reported that children, particularly girls, with T1D at an increased risk for developing other autoimmune diseases, with the prevalence of anti-thyroid autoantibodies at disease onset being about 20% [70]. Furthermore, the prevalence of anti-thyroid antibodies increases with age, and the presence of these antibodies at diagnosis of T1D is predictive of future thyroid disease [70]. Long-term follow-up suggests that up to 30% of patients with T1D develop AITD [71], and patients with thyroid autoimmunity are 18 times more likely to develop AITD than those without [72]. Therefore, to enable early diagnosis of AITD in children with T1D, the International Society for Pediatric and Adolescent Diabetes Consensus Clinical Guidelines recommend screening for thyroid function by analyzing circulating TSH at the time of diabetes diagnosis and every 2nd year thereafter in asymptomatic individuals without goiter, and with more frequent screening in patients with goiter present [73].
To characterize the T1D patients with coexisting AITD (referred to as autoimmune polyendocrine syndrome type 3 variant, APS3v), we analyzed their clinical characteristics compared to those without AITD [74]. Patients with APS3v demonstrated a significant female predominance, a slower and older age of onset for T1D, and a higher prevalence and level of GADA. Furthermore, among patients with Graves’ disease, 60% developed the disease before T1D, 30% had antecedent T1D, and 10% developed T1D and Graves’ disease simultaneously. The interval between the onset of T1D and Graves’ disease was <10 years in most cases but approached or exceeded 20 years in some instances.
The prevalence of GADA in nondiabetic patients with ATID is 8-10% [75-77], and it has been reported that GADA positivity is associated with a decreased insulin secretion capacity [78], suggesting that GADA positivity could be a marker of subclinical insulitis. Furthermore, the incidence of T1D/SPIDDM in anti-islet autoantibody-positive patients with Graves' disease has been reported to be 2.5 times higher than in autoantibody-negative patients [79]. The exact reasons why the level of GADA in T1D patients with AITD is extremely high compared to patients without AITD remain unclear. Previous studies have reported that GAD65 is expressed in the thyroid gland [80, 81], and T1D patients with AITD demonstrated higher polyclonality in GADA epitope recognition and IgG subclasses compared to patients without AITD (Figure 6). Forty-eight percent of T1D patients with AITD recognized GAD65 N-terminal region, middle domain, and C-terminal region epitopes, while the same was true of only 9% of those without AITD (P<0.005). Additionally, the prevalence of patients with two or more IgG subclasses was significantly higher in T1D patients with AITD than those without AITD (57% vs. 17%, P<0.001). Therefore, the overproduction of GADA in T1D patients with AITD might be attributable to the activation of polyclonal B-lymphocyte response by GAD in the thyroid gland [75].
Figure 6. Comparison of GADA epitopes (top) and IgG subclasses (bottom) in T1D patients without and with autoimmune thyroid disease. Left panel: T1D without AITD; Right panel: T1D with AITD
As described in the previous section, ICA is detected by immunohistochemistry using indirect immunofluorescence or immunoenzymatic techniques on frozen sections of human blood group O pancreas. Achieving concordance of results obtained from different laboratories is essential for comparing studies from various centers. To address this issue, a series international workshops on the standardization of ICA was initiated in 1985 under the auspices of the Immunology and Diabetes Workshop (currently Immunology and Diabetes Society). However, wide variations in results persisted despite using the same experimental protocol [82]. Therefore, subsequent workshops employed standard curves constructed from a reference serum (JDRF standard), which led to a decrease in interlaboratory variation [83].
Following the discovery of GAD as a major target antigen against ICA, numerous recombinant anti-islet autoantibody assays, including radioligand binding assays (RBA), RIA, and ELISA, have been developed. With these assays, the reported prevalence of anti-islet autoantibodies such as GADA, in patients with recently diagnosed T1D ranged widely from 25% [84] to 80% [85, 86]. To standardize the recombinant anti-islet autoantibody assays, the first GADA workshop was held in 1993. In contrast to the ICA workshop, considerable concordance in ranking GADA levels in 16 different samples was observed among various assay formats [87]. In the second GADA workshop, which used a large number of T1D and control sera, the RBA format demonstrated greater sensitivity than conventional ELISA [88]. Solid-phase ELISA formats can detect insulin-binding antibodies in patients receiving insulin injections. However, these assays fail to detect IAA associated with disease risk., likely due to the high apparent affinity (1010) and extremely low capacity (10-12) of prediabetes autoantibodies [39]. Furthermore, the widely used fluid-phase RBA has proved challenging for many laboratories to implement. In response to the need for improved anti-islet autoantibody assays, a nonradioactive fluid-phase assay has been developed that offers high sensitivity and specificity. This assay employs a modified ELISA format, which is based on the autoantibodies’ ability to form a bridge between recombinant autoantigens (such as GAD, IA-2, and ZnT8) coated on the ELISA plate and biotin-labeled corresponding autoantigens, allowing for the detection of autoantibody-bound antigens rather than immunoglobulins themselves [89]. Using the same assay format, an anti-islet autoantibody assay has recently been developed that allows for the simultaneous measurement of multiple autoantibodies in a single well (3 Screen ICA ELISA) [90]. Although this assay cannot distinguish which of the three autoantibodies are present, it may serve as a useful screening test for T1D and facilitate the efficient diagnosis of immune-mediated T1D. Other sensitive high-throughput assays effective for large-scale screening include a multiplex ECL assay and a multiplex agglutination-PCR autoantibody assay that combines all four biochemically defined anti-islet autoantibodies [91, 92]. These innovations coming from around the globe will enhance our ability to identify high-risk individuals for T1D more accurately and efficiently at an early stage, promoting the advancement of early interventions for the benefit of public health.
The most important goal of diabetes management is to improve the quality of life and extend the lifespan by preventing micro- and macro-vascular diseases. To achieve this goal, it is crucial to maintain good glycemic control and residual β-cell function. Over the past 30 years, while numerous clinical trials regarding primary prevention have been conducted, to date, no treatment or agent has been proven effective in preventing T1D in susceptible individuals. However, several therapeutic agents have demonstrated usefulness in delaying progression from stage 2 to stage 3 T1D while also preserving β-cell function in stage 3. These agents include abatacept (anti-CTLA-4), alefacept (a fusion protein of soluble lymphocyte function antigen with Fc fragments of IgG1), rituximab (anti-CD20), and anti-thymocyte globulin, which preserve C-peptide production compared with placebo in stage 3 T1D [93-96]. Moreover, teplizumab (anti-CD3) has been shown to delay progression from stage 2 to stage 3 T1D by a median of 3 years [97, 98]. Based on these results, in November 2022, the U.S. Food and Drug Administration approved teplizumab as the first drug to delay progression from stage 2 to stage 3 T1D in adults and children > 8 years. The previous prevention trials targeted the subjects who were positive for at least one or more anti-islet autoantibodies. However, since these studies used the RBA method for measuring anti-islet autoantibodies, there is a possibility that low-risk subjects were also included. Therefore, it is crucial to screen subjects with assays that can exclusively detect high-affinity autoantibodies in order to verify more reliable preventive effects. Additionally, the ability to assess multiple autoantibodies in a single test should prove valuable for future interventional trials.
This article focused on reviewing the current understanding of anti-islet autoantibodies in T1D. The clinical utilities of anti-islet autoantibodies in patients with diabetes include diagnosis (immune-mediated or idiopathic), prediction (progressor or non-progressor) and understanding of pathophysiology (insulitis-specific or nonspecific phenomenon) (Figure 7). Since the autoantibody level of anti-islet autoantibodies decreases with disease duration and can become negative, it is essential to measure them early in the onset of T1D for accurate diagnosis. SPIDDM or LADA is often indistinguishable from T2D; therefore, earlier measurement of anti-islet autoantibodies is of great clinical importance for early diagnosis and appropriate treatment. In addition to the anti-islet autoantibody profiles, age of onset and genetic risk score should also be considered for risk triage. Furthermore, the development of a high-throughput assay to detect epitope-specific or immunoglobulin isotype-specific autoantibodies should warrant accurate diagnosis and prediction of autoimmune disorders. Besides, the new type of autoantibody assays, which can simultaneously measure multiple autoantibodies, have the advantages of high sensitivity and specificity, and the ability to measure a large number of samples, making it suitable for large-scale population screening of T1D.
Figure 7. Clinical utilities of anti-islet autoantibodies in patients with diabetes