Autoimmune origin for immune checkpoint inhibitor-diabetes revealed by deep immune phenotyping of the pancreas

Abstract

Immune checkpoint inhibitor-diabetes (CPI-D) is an acute and non-resolving immune-related adverse event (irAE) initiated primarily by disrupting the programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) axis with monoclonal antibodies. A major limitation in understanding CPI-D is the lack of access to pancreatic tissue from patients experiencing this complication. We report a unique patient with no prior history of diabetes or autoimmune disease whose treatment with CPI for metastatic melanoma was complicated by CPI-D requiring insulin therapy. The patient then went on to develop pancreatic cancer. In the setting of the pancreatic cancer treatment, we were able to perform detailed single-cell RNA sequencing and immunophenotyping within the surgically resected pancreas. This revealed substantial lymphocytic infiltration associated with the islets, suggestive of an autoimmune rather than autoinflammatory mechanistic origin for CPI-D.

Keywords: Diabetes; Immune Checkpoint Inhibitor; Immune related adverse event - irAE.

Figure 1. CPI-D is associated with T and B cell pancreatic infiltrate. (A) Schematic of case history relating to cancer diagnosis, cancer treatment, and development of irAEs, particularly CPI-D. (B) Histology of immune infiltrate surrounding an islet is shown by H&E (far left panel) and staining with glucagon (denoted as GCG for alpha cells), insulin (denoted as INS for beta cells), CD3, CD4 or CD8 (T cells), CD20 (B cells). Scale bar 200 µm. (C) Single-cell RNA sequencing of islets, PBMC and pancreatic tumor with (D) cluster identification through the Azimuth reference database. (E) Feature plot of immune markers used to identify cell types of interest is shown with the gene expression in blue (logCPM), red circle denotes GCG-expressing islet cells. CPI, immune checkpoint inhibitor; CPI-D, CPI-diabetes; DC, dendritic cell; DKA, diabetic ketoacidosis; Eryth, erythrocyte; FOXP3, forkhead box protein P3; HLA, human leukocyte antigen; ILC, innate lymphoid cell; INS, insulin; irAE, immune-related adverse event; Mono, monocyte; NCR1, natural cytotoxicity triggering receptor 1; NK, natural killer; PBMC, peripheral blood mononuclear cell; PET, positron emission tomography; Treg, regulatory T cell; T1D, type 1 diabetes.

Figure 2. Prominent T cell activation and lack of antibody-producing plasma cells are characteristics of CPI-D. (A) Overview of pancreatic tissue from the patient with CPI-D with the distribution of islets (in boxes), T/B cell clusters (in circles), and CD8+ T cells (in red) identified by CODEX. Scale bar 1 mm. (B) Mean frequencies of immunomodulatory markers expressed in CD8+ T cells, CD4+FOXP3- T cells, CD4+FOXP3+ T cells, and B cells in the pancreas from the patient with CPI-D. (C) Co-expression of immunomodulatory markers in pancreatic CD8+ T cells is demonstrated by UMAP. (D) Expression of PD-1 in T cells associated with islets from a patient with CPI-D. Scale bar shown in each individual panel is 50 µm. (E) Gene expression of PDCD1 (the PD-1 encoding gene) from PBMCs, islets and pancreatic tumor cells by scRNA-seq; cluster annotation detailed in figure 1D. (F) T/B cell clusters within the pancreas are composed of CD20+ B cells, CD4+ T cells and CD8+ T cells. Representative scale bar 200 µm is consistent across all regions shown. (G) OX40 is expressed in CD4+FOXP3+ T cells in T/B cell clusters. Scale bar shown in each individual panel is 100 µm. (H) Gene expression of TNFRSF4 (the OX40 encoding gene) from PBMCs, islets and pancreatic tumor by scRNA-seq and comparison to the FOXP3 gene expression region outlined in red and further detailed in figure 1E; cluster annotation detailed in figure 1D. (I) CD138 expression in T/B cell clusters in the pancreas of an individual with CPI-D or T1D. Scale bar shown in each individual panel is 50 µm. CODEX, co-detection by indexing; CPI-D, immune checkpoint inhibitor-diabetes; FOXP3, forkhead box protein P3; ICOS, inducible T cell costimulator; Ki67, kiel 67; LAG-3, lymphocyte-activation gene 3; NK, natural killer; PBMC, peripheral blood mononuclear cell; PD-1, programmed death-1; scRNA-seq, single-cell RNA sequencing; T1D, type 1 diabetes; UMAP, Uniform Manifold Approximation and Projection.

Figure 3. CDR3β sequences are expanded, shared between compartments, and matched to spontaneous T1D subjects. (A) Number of T cells with a CDR3β sequence isolated from islets and PBMCs relative to the number of T cells without CDR3β sequences. Number of T cells with CDR3β sequences from (B) PBMCs and (C) islets alongside the frequency (percent, in white) of the features of interest, including sharing across compartments (PBMCs, islet, pancreatic tumor), expansion within compartments (≥2 of the same CDR3β sequence), and identification of matched CDR3β sequences (either exact or within 1 LD) from subjects with spontaneous T1D within the nPOD database and the publicly available McPAS-TCR database. CDR3β, complementarity determining region 3 beta; McPAS-TCR, Manually Curated Pathology-Associated T Cell Receptor database; nPOD, Network for Pancreatic Organ Donors with Diabetes; PBMC, peripheral blood mononuclear cell; T1D, type 1 diabetes.