Clinical and genomic features of classical and basal transcriptional subtypes in pancreatic cancer.

Abstract

Transcriptional profiling of pancreatic cancers (PC) has defined two main transcriptional subtypes, classical and basal. Initial data suggest shorter survival for patients with basal tumors and differing treatment sensitivity to FOLFIRINOX (FFX) and gemcitabine nab-Paclitaxel (GnP) by transcriptional subtype. We examined 8,743 patients with RNA sequencing from PCs performed at Caris Life Sciences (Phoenix, AZ). Classical and basal subtypes were identified using PurIST algorithm on RNA-sequencing and two cohorts were analyzed: (1) Biomarker cohort included patients with complete molecular profiling data (n = 7,250); (2) Outcomes cohort included patients with metastatic disease with available survival outcomes (n=5,335). In the biomarker cohort, 3,063 tumors (42.2%) were strongly classical (SC), and 2,015 tumors (27.8%) were strongly basal (SB). SC and SB tumors showed strong associations with histologic phenotypes and biopsy site. SB tumors had higher rates of KRAS, TP53, and ARID1A mutations, lower rates of SMAD4 mutation, and transcriptional evidence of epithelial mesenchymal transition. Sixty of 77 cases (78%) maintained their transcriptional subtype between temporally and/or spatially disparate lesions. In the outcomes cohort, SB subtype was associated with shorter overall survival time, regardless of whether they received FFX or GnP as first line chemotherapy. Mutant KRAS allele type was prognostic of outcomes, however this impact was restricted to SC tumors, whereas all mutant KRAS alleles had similarly poor outcomes in SB tumors. SB subtype is a strong independent predictor of worse outcomes, irrespective of upfront chemotherapy regimen. Clinical trials should investigate PC transcriptional subtypes as a biomarker.