Dissecting the single cell profile of pancreatic ductal adenocarcinoma

Abstract

Background: The interplay of the tumor, stroma and immune system is becoming increasingly recognized as crucial in the pathogenesis of pancreatic ductal adenocarcinoma (PDAC). Traditional bulk RNA sequencing relies on utilizing RNA from a tumor specimen, and subjecting it to modern RNA sequencing techniques. Although this technique has allowed us to make great strides in understanding the heterogeneity and profile of pancreatic adenocarcinoma, we are unable to track RNA expression on the single cell level—which would allow us to pinpoint how treatments or lack thereof are contributing to phenotype and resultant survival. Methods: Immediately following resection of a pancreatic cancer specimen or alternatively after biopsy of a metastatic lesion, a tumor suspension is made. The tumor is broken down with a combination of mechanical and enzymatic digestion. This is followed by isolation of live single cells, and removal of RBCs and debris (Figure 1A). We immediately run a cell suspension of 10,000 live cells on the 10× Genomics single cell platform (Figure 1B), whereby each individual cell receives a unique barcode that it carries throughout the sequencing process. We then perform RNA sequencing. After sequencing, all cells with mitochondrial DNA content >10%, consistent with a dead cell, are removed from further analysis. We are left with a single cell RNA profile of live cells from the tumor. Results: Thus far we have processed 4 primary pancreatic ductal adenocarcinoma specimens, and 1 metastatic lesion, from a total of 5 patients, with 3 samples analyzed thus far. On average, we are left with 2470 live cells per patient sample. For each patient, we were able to identify 12 unique cell populations. Patient 1174, underwent a Whipple without preoperative chemoradiotherapy for a resectable PDAC head mass (T1cN2 on final path). Single cell analysis of the tumor specimen, identified 42.7% of the tumor mass to be PDAC cells (Figure 1C). In this case 4 distinct tumor subpopulations can be identified (Figure 1C). T and B lymphocytes comprised 35.6% of infiltrating cells, macrophages comprised 13.2% of the tumor population, with mast cells, pancreatic stellate cells, acinar cells and endothelial cells comprising the remainder of the tumor mass. Each cell population is identified by multiple characteristic markers, with 1 identifying marker displayed in Figure 1C. Example characteristic marker being KRT19 for pancreatic ductal cancer cells. In contrast to patient 1174, patient 1105, who received upfront neoadjuvant chemotherapy, but was found to be unresectable in the operating room, had a liver tumor with only 23.9% PDAC cells, and ∼50% of infiltrating cells comprising T and B lymphocytes, as well as NK cells. Finally, patient 1160, who had a resectable pancreatic head mass (T2N1 on final path) and did not receive preoperative therapy, had 20.6% PDAC cells, ∼33% infiltrating lymphocytes, and 20% macrophages. Conclusion: Single cell RNA sequencing is a viable strategy to better dissect the makeup of PDACs. It allows for RNA profiling on a single cell level, and provides a landscape of the tumor and tumor microenvironment in a non-biased fashion. Our group is enrolling multiple patients treated and not treated prior to pancreatic cancer resection, as well as consecutive specimens (pre treatment and post surgical) from several patients in order to better assess both the PDAC landscape and how it changes based on different factors such as chemotherapy, stage or location.