Characterizing Subgroups of Type 1 Diabetes

The prevailing model of type 1 diabetes (T1D) evolution is based on a now classic 1986 article by the late George Eisenbarth (1). A recent review noted that although the model still holds to a large extent, there are subtleties that suggest it is not as simple as originally proposed (2). The 1997 American Diabetes Association Expert Committee divided T1D into type 1A (immune mediated) and type 1B (idiopathic) (3), although this classification is not commonly used. Earlier, in 1977, Irvine (4) divided the disease into three subtypes: severe autoimmune/nonviral, moderate autoimmune/moderate viral, and severe viral/nonautoimmune. The severe viral subgroup appears to be rare and may have an explosive onset with a fatal outcome (5,6). The Japan Diabetes Society divides T1D into three categories depending on the manner of onset and progression—namely fulminant, acute-onset, or slowly progressive type 1 diabetes (SPID) (7). It is likely that the severe viral subgroup noted by Irvine and the Japanese fulminant subgroup are the same.

It is clear that the rate of decline of β-cell function differs by age (8). It also has been noted that incidence rates of T1D have been increasing, particularly in children under the age of 5 years (9,10). Thus, the progressive linear decline originally outlined by Eisenbarth (1) cannot be simply applied. If nothing else, the T1D disease process evolves at different rates. The number of diabetes autoantibodies appears to predict the development of diabetes—those with more antibodies develop clinical diabetes more rapidly (11). Furthermore, in those with high-risk HLA types identified during newborn screening, development of two or more diabetes autoantibodies is an almost inevitable predictor of diabetes development over the next two decades (12). Yet, this also emphasizes that the rate of diabetes development is heterogeneous, with some individuals developing diabetes soon after the appearance of two antibodies and others having the disease evolve more slowly over time (12). Going beyond antibodies and humoral immunity, in recent years there also has been improvement in the development of assays of cellular immunity (T-lymphocyte assays) in T1D (13,14).

Moreover, it has become apparent, particularly through the use of ultrasensitive assays for C-peptide, that β-cell destruction is not complete in T1D as C-peptide is measurable in some individuals decades after the onset of T1D (15–17). This also serves to raise the question of whether the decreased β-cell function in T1D is a result solely of loss of β-cell mass or whether it could be, at least in part, a loss of β-cell secretory response to glucose—namely a functional defect (18).

Classic pathologic findings in T1D, at least in specimens from individuals whose pancreases were studied after a short duration of diabetes, were noted by Gepts (19) and Gepts and Lecompte (20) as being a combination of insulitis in some islets, normal histology in some islets, and pseudo-atrophic islets (i.e., lacking β-cells but retaining α-cells and δ-cells) in perhaps a majority of islets. These findings were confirmed by Foulis et al. (21) in a large series of patients. More recently, the Network for Pancreatic Organ Donors with Diabetes (nPOD) was established to better characterize the histopathology of T1D (22). Among the findings that have emerged from nPOD are a consensus definition for insulitis (23) and a description of the histopathology of long-standing childhood-onset diabetes, the latter showing two quite different patterns (24).

In the current issue of Diabetes, a distinguished group of investigators from across the U.K. take the concept of heterogeneity of T1D much further by characterizing what appear to be two distinct immunological responses (25). Their report comprises essentially three careful analyses of children and adolescents, including 1) subjects with recent-onset T1D, 2) siblings of individuals with T1D, and 3) pancreatic autopsy specimens of individuals with T1D. Arif et al. (25) carefully characterize diabetes autoantibody levels and diabetes CD4 cellular immune responses in the living subjects and separately examine the degree and characteristics of cellular infiltrates in the postmortem samples.

In the autopsy specimens, examining the pancreata from individuals who died shortly after the time of T1D onset, Arif et al. characterize two distinct patterns of infiltration based on degree of cellular infiltrate and presence of B lymphocytes (which have CD20 on their surface). One pattern has high infiltration and a high frequency of CD20⁺ cells (“hyper-immune CD20Hi”), while the other has low infiltration and a low frequency of CD20⁺ cells (“pauci-immune CD20Lo”). These patterns were consistent for each individual pancreas that was examined. Interestingly, the hyper-immune CD20Hi group had a lower mean age of T1D onset than the pauci-immune CD20Lo group.

Using multiparameter phenotyping in the T1D patients and siblings with two or more islet autoantibodies, the authors were able to identify two patterns of response. One was characterized by multiple autoantibodies and a proinflammatory interferon-γ (IFN-γ) cellular response to islet antigen epitopes (using peptides derived from proinsulin, insulin, GAD, and IA2), and the other was characterized by fewer “pauci” autoantibodies and an apparent regulatory interleukin (IL)-10 cellular response to the same islet antigen epitopes.

As summarized in Table 1, Arif et al. speculate that the pattern of multiple autoantibodies and proinflammatory IFN-γ cellular response may have its histological counterpart in the hyper-immune CD20Hi cellular infiltrates. In turn, the pattern of fewer autoantibodies and regulatory IL-10 cellular response may have its histological counterpart in pauci-immune CD20Lo cellular infiltrates. Given that the antibodies and cellular response data were collected from a different group of subjects than the postmortem histopathology samples, this alignment of patterns must remain speculative. Nonetheless, the data are quite provocative and beg confirmation in subjects who have both blood samples and histopathology available. Ideally, the histopathology might come from pancreatic biopsy specimens around the time of diagnosis (26,27). However, obtaining such specimens is both challenging and not without risk to the patient (28).

Immune intervention strategies are being tested in T1D. Interestingly, one study with an anti-CD3 monoclonal antibody targeting T lymphocytes found that some subjects responded to the treatment with maintenance of β-cell function for 2 years, while others showed essentially no response at all (29). Another study using an anti-CD20 monoclonal antibody targeting B lymphocytes found that younger subjects responded better than older subjects (30). Could it have been that the responders in these studies had the hyper-immune patterns described in the newly published report by Arif et al. (25)? Could further characterization of subjects in a variety of trials help identify subgroups that may respond to one or another intervention? These are critical questions brought to the fore by the important and provocative observations from the U.K. investigators (25).