Insulitis Revisited

Restricted access to the inflammatory lesions in patients with type 1 diabetes (T1D) has slowed insight into human insulitis. The conception of a network collecting pancreases of donors with diabetes (Network for Pancreatic Organ Donors with Diabetes [nPOD]; www.jdrfnpod.org) has focused the international diabetes research community on islet inflammation in new tissue for the first time in decades, leading to an overhaul in insight in human insulitis. In this issue of Diabetes, Campbell-Thompson et al. (1) present the latest status of insulitis in relation to β-cell mass, serving the diabetes research community some unexpected findings. The investigators studied the frequency and composition of insulitis in relation to β-cell mass throughout the natural history of T1D, i.e., from seropositive donors without diabetes to donors with newly diagnosed and long-standing disease. Insulitis in donors with islet autoantibodies was rare (none of the 13 donors with a single antibody, 2 out of 5 with multiple antibodies, with the proportion of inflamed islets ranging from 1.4 to 6.4%), whereas a large proportion of patients with long-lasting disease lacked islet autoantibodies, even if insulitis persisted. The few donors with recent-onset T1D all showed insulitis; in those individuals with established disease, the frequency of insulitis in both individual islets and within patients was less, irrespective of the age of disease onset.

The rate of insulitis remains greatest in islets producing insulin compared with insulin-negative islets (33% vs. 2%), suggesting this inflammatory process is driven by the presence of β-cells. Residual β-cells persisted in all T1D donors with insulitis, and β-cell area and mass were significantly higher in T1D donors with insulitis compared with those without insulitis. The degree of heterogeneity in the numbers of inflamed islets, the rate and composition of insulitis, and the remaining β-cell mass within a tissue and between T1D patients was overwhelming.

In its relatively short existence, nPOD has changed the landscape and paradigms of T1D. Recent lessons from nPOD’s tissue platform include the demonstration of islet-specific autoreactive CD8+ T cells in destructive insulitic lesions, which supports the concept of an autoimmune nature of T1D. Other seminal lessons include the persistence of β-cells, insulin production, and insulitic lesions that endure many years following the clinical manifestation of hyperglycemia. We have learned that there is an apparent disconnect between β-cell mass and function (β-cell quiescence or hibernation), that there is a profound difference in the immunopathology between men and mice, and that there exists an overwhelming heterogeneity in the pathologic lesions (varying from no infiltrate to “pauci”-autoimmune to “fulminant”) of this patient population. nPOD has also led to the demonstration of focal disease activity, similar to vitiligo or alopecia, and the persistence of β-cell mass and function long after manifestation of hyperglycemia (2–4). The latter was recently appreciated again by the findings in Exeter of C-peptide in the urine of the vast majority of T1D patients, in spite of disease duration for decades (3). Although some of the findings presented by Campbell-Thompson et al. (1) have been reported anecdotally before, the mere large sample size in the study alone deserves credit and adds value (2,5–7). Given this is work in progress, corroboration with new pancreas tissue is essential. The term “confirmation” does not do sufficient justice to the studies presented here, and many more studies and reports on many more tissues are warranted for common themes to emerge and be validated.

At the same time, several issues need reconciliation. The study by Campbell-Thompson et al. (1) is descriptive and cross-sectional, precluding firm conclusions on changes in time throughout disease progression. It should also be appreciated that the tissues investigated were derived from autopsy, not biopsy. Indeed, subtle differences in the patterns of infiltration between living and dying donors may emerge, e.g., leukocyte infiltration of the exocrine pancreas increases with the time a donor spent in an intensive care unit (2,8). Although the number of donors and pancreas sections investigated is impressive, the number of newly diagnosed donors or donors without diabetes with multiple islet autoantibodies in particular remains limited. The sample size also impairs interpreting disease variation relative to ethnicity, age, disease duration, and HLA polymorphisms, despite several observations that hint at linked diversity (e.g., low insulitis frequency in African American donors). Finally, the loss of insulin in islets does not necessarily mean the loss of β-cells, as β-cells may be degranulated or progressing toward dedifferentiation, which may confound an accurate determination of β-cell mass.

Collectively, these new insights have set the stage for new therapeutic strategies that may prove effective in protecting β-cells long after the clinical onset of T1D. Decrypting human insulitis may help us to understand the differences in the efficacy of immune intervention strategies between mice and men, as human insulitis differs substantially from that in preclinical models in terms of severity, the apparent lack of regulatory T cells, the frequency of islet infiltration, and the composition of the infiltrate (proportion of CD8+ vs. CD4+ T cells). This awareness should improve the guidance of preclinical disease models into clinical translation (9).

The new observations in human insulitis and the heterogeneity thereof create several opportunities for future studies. All different stages of disease appear present in a single pancreas, from apparently healthy islets through inflamed to the end stage of β-cell–depleted islets without inflammation (Fig. 1). The variation between pancreases also bears implications for therapy. First, given the huge degree of variation in immunopathology, no single bullet is to be expected as a viable immune intervention therapy for the T1D patient community at large. Instead, disease differentiation and fine diagnosis may move immunotherapy to precision and personalized medicine (10,11). The persistence of both β-cells and insulitis may help explain why abatacept may still work at the supposedly late stage of clinical diagnosis or perhaps explain subgroup (in)efficacies (e.g., apparent lack of efficacy of abatacept in African American patients) (12). The disconnect between β-cell mass and function may have consequences for stimulated C-peptide as primary end points of immunotherapeutic efficacy, as any beneficial therapeutic impact on insulitis may remain unnoticed. Insulitis may require different types of therapeutic strategies than reengaging hibernating β-cells.

The journey has been short so far, and beyond doubt, there are many additional secrets waiting to be revealed by investigating diabetic tissue. There is a huge demand for more tissue and for more sharing to guide discoveries of immune correlates to disease progression and to find targets for therapeutic intervention. Joint investigations of diabetic tissue by different laboratories and experts, coupled with this unique data repository, will provide the basis to crack the codes of human insulitis and design immunotherapy accordingly.