SINGLE-CELL RNA SEQUENCING ANALYSIS OF PANCREATIC DUCTAL ADENOCARCINOMA REVEALS TRANSCRIPTOMIC CHANGES IN FIBROBLASTS FOLLOWING FOLFIRINOX-BASED NEOADJUVANT CHEMORADIATION THERAPY
Jon Harrison*1, Jonathan E. Chang2, Michelle Piquet2, Slavica Dimitrieva2, Millicent Gabriel2, Keith D. Lillemoe1, Andrew L. Warshaw1, Mari Mino-Kenudson1, Carlos Férnandez-Del Castillo1, Viviana Cremasco2, Dave Ruddy2, Andrew S. Liss1
1Surgery, Massachusetts General Hospital, Boston, MA; 2Novartis Institutes for BioMedical Research Inc, Cambridge, MA
Background: Understanding the constituents of the pancreatic ductal adenocarcinoma (PDAC) microenvironment may illuminate more efficacious targets for treating this aggressive malignancy. However, the compositional changes occurring in untreated and treated PDAC relative to both normal and inflamed pancreas have not been well characterized. Single-cell gene expression profiling can be employed to differentiate cell populations across complex cellular milieus. With this technique, we aimed to provide a more comprehensive understanding of the PDAC tumor microenvironment relative to the non-cancerous pancreas and in the context of neoadjuvant therapy (NAT).
Methods: Upfront resectable PDAC, FOLFIRINOX-based treated PDAC, chronic pancreatitis, and normal pancreas specimens were obtained for scRNA-seq from patients undergoing pancreatectomy. Following histologic examination, samples were disassociated, loaded onto the 10X Genomics scRNA-seq platform, and sequenced with Next-Generation Illumina sequencing technology. Subsequent transcriptomic analysis was conducted using the Seurat workflow as well as the ClusterProfiler package in R.
Results: scRNA-seq analysis was performed on 12 PDAC tumors (six untreated and six NAT) as well as four pancreatitis and three normal pancreas samples. In total, 156,072 cells were sequenced, which included 57,605 (36%) untreated tumor cells, 47,632 (31%) NAT tumor cells, 24,633 (16%) cells from chronic pancreatitis, and 26,202 (17%) normal pancreatic cells. Dimensionality reduction of this dataset revealed 16 cell clusters, which were annotated for canonical markers of pancreatic ductal and cancer cells, components of the non-immune stroma such as fibroblasts, and constituents of the immune stroma including myeloid-derived cells and lymphocytes. Cluster subsetting revealed cell subtype heterogeneity between normal, inflammatory, and malignant specimens. In terms of pancreatic fibroblasts, two pancreatic fibroblast subtypes were identified in non-cancerous specimens, one of which expanded in the setting of chronic inflammation. This reactive-type fibroblast (RPF) was also seen in tumor samples as well as a smaller population of tumor-specific cancer-associated fibroblasts (CAFs). Notably, the RPF subtype was the dominant fibroblast in 67% of tumors. Analyzing the transcriptomes of fibroblasts in NAT tumors revealed upregulation of DNA damage response pathways in the RPF subtype that were not enriched for in CAFs.
Conclusions: scRNA-seq reveals cell subtype variation between benign and malignant pancreatic tissues. Malignant sample fibroblasts included both RPFs and CAFs, and RPFs were the dominant subtype in most tumors. Transcriptomic differences between these subtypes post-NAT suggest CAFs are more resistant to combination chemotherapy. These findings identify CAFs as a potential cell type for targeted therapy.
Back to 2021 Abstracts