Real-World Data Validation of the PurIST Pancreatic Ductal Adenocarcinoma Gene Expression Classifier and Its Prognostic Implications

MedRxiv Manuscript
Authors Stephane Wenric, James M. Davison, John Guittar, Gregory M. Mayhew, Kirk D. Beebe, Yun E. Wang, Amrita A. Iyer, Hyunseok P. Kang, Michael V. Milburn, Vincent Chung, Tanios Bekaii-Saab, Charles M. Perou

Background: Pancreatic ductal adenocarcinoma (PDAC) is amongst the deadliest cancers, with few modern tools to inform patient prognosis and help guide treatment options. Transcriptome-based molecular subtyping is one emerging technology that has been employed to help patients optimize available therapeutic approaches. Here we retrospectively demonstrate the clinical validity of PurIST (Purity Independent Subtyping of Tumors), an RNA-based classifier that divides PDAC patients into two subtypes with differential prognoses, as a validated laboratory-developed test (LDT) on the Tempus Labs sequencing platform.

Methods: A cohort comprising 258 late-stage PDAC patients with available transcriptomic and outcomes data was drawn from the Tempus clinicogenomic database and classified using PurIST into one of two subtypes (“Basal” or “Classical”). Differences in patient survival from the date of diagnosis were compared between subtypes, and between two common first-line treatment regimens, FOLFIRINOX, and gemcitabine + nab-paclitaxel.

Results: Of the 258 PDAC patients in the validation cohort, PurIST classified 173 as classical subtype, 59 as basal subtype, and 26 as no-calls. Reinforcing previous findings, patients of the basal subtype had significantly lower overall survival than those of the classical subtype. Notably, differential survival by subtype was significant among the subset of patients on FOLFIRINOX, but not those on gemcitabine + nab-paclitaxel.

Conclusions: The implementation of PurIST on a high-throughput clinical laboratory RNA-Seq platform and the demonstration of the model’s clinical utility in a real-world cohort together show that PurIST can be used at scale to refine PDAC prognosis and thereby inform treatment selection to improve outcomes for advanced-stage PDAC patients.


Related Content

View more