Multiomic insights into gastroenteropancreatic neuroendocrine tumor biology

Abstract

Gastroenteropancreatic neuroendocrine tumors (GEP-NETs) are rare, biologically and clinically heterogeneous malignancies whose incidence has increased significantly over the past few decades in the United States. Therapeutic development for GEP-NETs has lagged due to a paucity of preclinical in vitro and in vivo models, which has also produced an incomplete understanding of the fundamental biology of these tumors. Single-cell genomics permits high-resolution dissection of primary tumor samples with direct patient relevance; combining single-cell and other sequencing modalities therefore promises to yield novel insight into GEP-NET and other solid tumor pathophysiology. In this dissertation, we first perform single-cell RNA-sequencing of over 24,000 cells from eight freshly resected GEP-NET samples, constructing the first-ever single-cell atlas of these rare malignancies. We demonstrate that tumor and lymphoid cells in the GEP-NET microenvironment only sparsely express immune checkpoint targets, explaining the limited efficacy of immunotherapies directed against these targets in previous clinical trials. In contrast, we show that NET-infiltrating myeloid cells robustly express immunosuppressive genes spanning several distinct pathways. The presence of these myeloid cells in both primary and metastatic NETs, and across three different GEP-NET subtypes, represent a shared opportunity for precision medicine development in this otherwise heterogeneous disease context. Next, we investigate the cellular biology of tumor and stromal populations from an archival cohort of 37 pancreatic and small intestinal NET samples using a combination of whole exome sequencing, multimodal single-nuclei RNA and ATAC sequencing, and spatial transcriptomics. We define major axes of transcriptional and epigenetic variation across malignant cells from both small intestinal and pancreatic NETs, their relation to underlying genetic states in these tumors, and characterize site-specific differences in the cellular phenotypes of cancer-associated fibroblasts. Finally, we expand upon inference of relationships between genetic states and cell-type specific transcriptional programs across cancer types. We evaluate cell-type specific enrichment of heritable cancer risk by combining publicly available genome-wide association study data from multiple solid tumor types and single-cell RNA-sequencing datasets from malignant and corresponding normal tissues. Using linkage-disequilibrium score regression, we nominate prostate club epithelium as a potential cellular mediator of prostate adenocarcinoma risk. Together, our findings elucidate the molecular and cellular biology of well-differentiated GEP-NETs and highlight novel opportunities for translational investigation and therapeutic development in this understudied patient population, with potential extension of these concepts across tumor types.