It has been known for decades that some people with apparent type 2 diabetes have evidence of islet autoantibodies associated with β-cell autoimmunity (1). This population is highly heterogeneous, with varying prevalence and overlapping clinical, biomarker, and genetic features between type 1 diabetes and type 2 diabetes (2). The extent to which this represents an intermediate phenotype of autoimmune diabetes, or a mixture of autoimmune and nonautoimmune diabetes as a result of an interaction between imperfect assay specificity and low prevalence of type 1 (autoimmune) diabetes in this population, is a matter of debate (2, 3).
Islet antibodies represent just one aspect of the immune response, and in autoimmune diabetes the extent to which they are pathogenic or are a surrogate marker of loss of immune tolerance remains unclear (4). In contrast, substantial evidence supports a pathogenic role of β-cell–reactive T cells in type 1 diabetes (5). While islet antibodies have been very extensively studied in those diagnosed with type 2 diabetes, less is known about the role of islet-reactive T cells in this condition. While a number of small studies have suggested that islet-reactive T cells can be identified in antibody-negative type 2 diabetes (6) and are associated with differences in phenotype including β-cell function (7–10), T-cell–mediated cellular autoimmunity has not been generally accepted as a major contributor to type 2 diabetes pathogenesis (11).
In this issue of Diabetes, Brooks-Worrell et al. (12) shed further light on the potential role of islet-reactive T cells in type 2 diabetes. In cross-sectional analysis of participants recruited to Glycemia Reduction Approaches in Diabetes: A Comparative Effectiveness Study (GRADE), they assessed the relationship between T-cell autoreactivity (to islet-derived antigens) and clincial characteristics, islet antibodies, and β-cell function in 322 participants with metformin-treated type 2 diabetes of mean 4 years’ duration. The authors found evidence of T-cell reactivity to antigens derived from one islet preparation, as assessed by an immunoblotting/T-cell proliferation assay, in 41% of the cohort. Positivity on this assay was associated with an 11–19% reduction in two oral glucose tolerance test–based insulin sensitivity–adjusted measures of β-cell function and modest worsening of glycemic control, equating to a 0.17% (2 mmol/mol) and 7.7 mg/dL (0.4 mmol/L) differences in HbA1c and fasting glucose. T-cell autoreactivity was not associated with other participant characteristics such as BMI, other features of the metabolic syndrome, or age, and there was no association between positivity on this measure of cellular autoimmunity and the presence of islet autoantibodies.
This research represents an important step forward in elucidating the contribution of islet-reactive T-cell immunity in populations with type 2 diabetes, with the detailed careful study of a large, well-characterized population. If T-cell autoreactivity is important to the development or progression of antibody-negative type 2 diabetes, then interventions to suppress or prevent islet autoimmunity could have a role in disease prevention and/or treatment in future. However, the study, and its relationship to our current understanding of type 2 diabetes, raises many questions and needs cautious interpretation.
First, we must be cautious in how we interpret the results for assays with imperfect specificity in this setting (3, 13, 14). The quoted specificity of the T-cell assay in this study for islet-specific antigens is 83%, with broadly comparable specificity in studies of children and young adults without diabetes or personal or family history of autoimmune disease (15, 16). An 83% specificity would mean that 17% of people without autoimmune etiology diabetes (or diabetes-induced autoimmunity) would test positive on this assay. This can be seen in other settings; for example, 1% of patients with (nonautoimmune) monogenic diabetes will test positive for islet antibody assays and thresholds robustly shown to have 1% prevalence in those without diabetes (99% specificity), and assays with lower specificity will give comparably higher prevalence (17). This means that if the quoted specificity is applicable to this population, ∼17% of those with type 2 diabetes that is unrelated to autoimmunity would have tested positive. Specificity could potentially be lower given that it has not been derived from a comparable population of older adults with high BMI. We should therefore be cautious in interpreting these findings as showing that 41% of those with type 2 diabetes have cellular autoimmunity contributing to diabetes pathogenesis or, as the authors discuss, autoimmunity induced by a process specific to diabetes. We should also consider that a more modest number of cases of autoimmune diabetes, with more severe features, within those with T-cell autoreactivity in the studied cohort may explain the modest overall differences observed in those with and without these markers. Consistent with caution in ascribing assay positivity to pathological autoimmunity, others working in this field have noted the widespread presence of T cells reactive to specific β-cell epitopes in those without diabetes, sometimes described as “benign” islet autoimmunity, and have shown that frequency of autoreactive CD8+ T cells to β-cell antigens distinguish individuals with type 1 diabetes from control subjects when measured in pancreatic tissue but not in peripheral blood (18, 19). In addition, it is known that many people exhibit blood markers for a wide spectrum of autoimmune diseases without necessarily developing the associated autoimmune condition (13).
How these findings relate to existing literature also raises important questions for further investigation. For example, if cellular autoimmunity and subsequent reduced β-cell function is contributing meaningfully to disease etiology, why do we not see the differences in clinical characteristics (for example, lower BMI and reduced features of the metabolic syndrome) that we usually see in most studies of those diagnosed with type 2 diabetes who have islet autoantibodies? Likewise, if autoimmune etiology is widespread within those with type 2 diabetes, why are weight loss and other nonimmune interventions so effective for both treatment and prevention, and why does the genetic background of robustly defined type 2 diabetes not demonstrate a major contribution of genetic factors related to autoimmunity (20)?
Taken together, these issues may indicate that widespread meaningful contribution of autoimmunity to the development of autoantibody-negative type 2 diabetes is unlikely and that an alternative hypothesis, such as the secondary induction of T-cell autoreactivity by islet inflammation and metabolic stress suggested by the authors, is probable. If this is the case, then a key question is whether the observed T-cell autoreactivity represents “smoke” (T-cell autoreactivity as a feature of metabolic stress/islet inflammation, without necessarily contributing to pathogenesis) or “fire” (a process directly contributing to pathogenesis and β-cell failure). As participants were recruited from the baseline visit of the prospective GRADE, the investigator team is ideally placed to further address this question in future. In addition, further studies should address whether the observed T-cell autoreactivity is mediated by CD4+ and/or CD8+ T cells and whether responses are restricted to certain HLA alleles/haplotypes, and should identify the antigenic drivers (proteins and epitopes) to determine whether these are consistent with those observed in type 1 diabetes.
In summary, these are intriguing findings that could be important to our understanding and, potentially, therapeutic intervention of type 2 diabetes in future. However, we should be cautious in our interpretation until robust prospective studies and assay specificity from clearly comparable control cohorts are available. We look forward to the further research that will shed light on these issues in future.
See accompanying article, p. 1261.
Funding. A.G.J. is supported by National Institute for Health Research (NIHR), U.K., Clinician Scientist award CS-2015-15-018 and NIHR Global Health Group 17/63/131. M.E. is supported by UK Research and Innovation (UKRI) via Research England’s Expanding Excellence in England (E3) fund, and JDRF UK (1-SGA-2021-0003).
The views expressed are those of the authors and not necessarily those of the funding bodies (NIHR or the Department of Health and Social Care, UKRI, or JDRF).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.