TRACING PANCREATIC TUMOR EVOLUTION THROUGH SPACE AND TIME

Background: Pancreatic ductal adenocarcinoma (PDAC) is a lethal disease with metastasis and therapy resistance being a driver in poor survival. Unlike other solid tumor types, precision care has not made similar advancements in PDAC, partly due to the fact that the tumor profiling performed on each patient only captures a small window of the disease. An optimal system to study how tumors evolve from the primary site to the metastatic site would be to perform deep profiling such as single cell RNA sequencing (scRNAseq) on matched primary and metastatic tumors from the same patient. Similarly, longitudinal studies profiling tumors before and after therapy would provide the framework to understand how tumors change in response to therapy and develop therapy resistance. We have established a robust pipeline utilizing endoscopic fine needle biopsies for scRNAseq and have leveraged this pipeline to acquire both primary and metastatic PDAC tumors from the same patient to parse the niche-specific biology of PDAC.
Methods: Fresh treatment naive tumor biospecimens for PDAC patients with multifocal disease were acquired via endoscopic biopsy of matched primary and liver metastasis at time of diagnostic biopsy. For resectable or locally advanced patients, treatment naïve tissue was acquired at time of diagnostic biopsy and post-treatment tissue acquired at surgical resection or endoscopy at time of fiducial placement. scRNAseq was performed on all 22 specimens. Inferred copy number variation (CNV) analysis using Numbat pipeline was performed to infer the evolutionary dynamics of profiled tumors.
Results: We derived several tumor subclones showing diversity on an inter-patient and intra-patient level. Within each patient, tumor clones displayed common ancestry, but also clearly showed distinct copy number alterations, for example, between the primary and metastatic niche. Importantly, the inferred CNV landscape of our tumor samples mirrored whole-exome sequencing results of PDAC patients from publicly available datasets. Interestingly, one longitudinal case revealed a soft tissue para-aortic mass that arose over a year after resection which matched the CNV profile of the first diagnostic biopsy, indicating recurrence derived from the original tumor, despite analysis of his surgically resected sample showing no tumor cells captured.
Conclusions: We have leveraged a robust method of inferred CNV analysis to trace PDAC tumors through space and time and have found remarkable consistency between our results and validated external genomic datasets. Our results set the stage for a robust, single-cell correlation of genotypes to phenotypes in PDAC patients.