Monoclonal antibodies (mAbs) against immune checkpoints (checkpoint inhibitors, CPIs) have revolutionized the treatment of solid organ tumors and dramatically improved the prognosis of patients with common cancers such as melanoma, lung and bladder tumors. These agents block inhibitory ligands that prevent killing of tumors by immune cells, but may cause immunologic adverse events including autoimmune diseases of the endocrine glands. We originally identified elderly individuals who presented with diabetic ketoacidosis and rapidly destroyed insulin producing ? cells after treatment with mAbs against PD-1 or PD-L1. Several other groups subsequently reported the occurrence of insulin dependent diabetes following CPIs for a variety of tumors. Interestingly, this event, unlike other endocrinopathies, appears to occur only after treatments that block PD-1/PD-L1 and not after treatment with anti-CTLA-4 mAbs. The reasons why particular individuals develop this adverse event and why it is seen with anti-PD-1/PDL1 but not anti-CTLA-4 mAb are not known and are the focus of this proposal. Our studies from preclinical models and the clinical data has suggested a ?two-hit? hypothesis in which inflammatory responses induce PD-L1 expression on ? cells, which may prevent immune attack as a normal response, but this protective mechanism is blocked when CPIs are administered. This model suggests that an autoreactive repertoire may distinguish those who do and do not develop diabetes. We propose to test this hypothesis by enumerating and phenotyping autoreactive CD8+ T cells in patients treated with CPIs who do and do not develop diabetes as well as the general CD4+ and CD8+ T cell populations. We will analyze these cells by CyTOF with Class I MHC tetramers and by creating CD8+ T cells libraries of autoantigen specific T cells. We have found diabetes associated autoantibodies, to two or more diabetes autoantigens in 43% of patients with diabetes suggesting that there may also be failures in B cell tolerance. We will analyze the frequency of autoreactive B cells by analyzing autoreactivity of antibodies expressed by immature and mature nave B cells. In addition, by determining the timing of ? cell killing in patients who develop diabetes, we should be able to determine when the pathologic process begins and identify clinical features that precipitate the onset in some but not all patients. We will use a novel assay to measure ? cell death in vivo in serum samples from a biobank of samples. These studies will identify the autoimmune mechanisms that are associated with diabetes in CPI treated patients and may establish biomarkers that can be used to predict patients who are likely to develop this complication and to monitor disease progression. These studies represent a first step in designing therapies that can inhibit the autoimmunity without affecting the therapeutic action of the drugs. In addition, the mechanisms that we uncover in these patients will shed light on the mechanisms of spontaneous Type 1 diabetes as well as other autoimmune events that occur in patients treated with these life-saving drugs.
We plan to analyze the immune repertoire and timing of ? cell killing in patients who develop autoimmune diabetes following treatment with checkpoint inhibitor therapies. Our studies will identify biomarkers that may distinguish individuals most likely to develop adverse events and shed light on the mechanisms that lead to spontaneous autoimmune diabetes.
