Andrew D. Kelly

Validation of comprehensive genomic profiling for prognostic and potential therapeutic molecular classification of endometrial cancer

We sought to validate the prognostic utility of comprehensive genomic profiling (CGP)-based molecular stratification for patients with endometrial carcinoma and to assess co-occurring biomarkers across subtypes. This study included patients from a de-identified nationwide (US-based) endometrial cancer clinicogenomic database who underwent CGP testing as part of routine care. Molecular subtypes were classified as POLE mutated (POLEmut), MSI-H, TP53 mutated (TP53mut), and no specific molecular profile (NSMP). Time to next treatment and overall survival were compared between molecular subtypes, with multivariable Cox models adjusted for relevant covariables. Of 1,139 evaluated patients with advanced or recurrent endometrial carcinoma, the prevalence of the 4 molecular subtypes was 1% POLEmut, 22% high microsatellite instability, 47% TP53mut, and 31% NSMP. Compared with NSMP patients, POLEmut patients had numerically more favorable time to next treatment (HR 0.50, 95% CI 0.21 to 1.21) and overall survival (HR 0.52, 95% CI 0.17 to 1.66). High microsatellite instability patients had similar time to next treatment (HR 1.08, 95% CI 0.89 to 1.30) and overall survival (HR 0.91, 95% CI 0.71 to 1.19) relative to NSMP patients. TP53mut patients had the least favorable outcomes for time to next treatment (compared with NSMP, HR 1.39, 95% CI 1.19 to 1.62) and overall survival (HR 2.15, 95% CI 1.77 to 2.61). In multivariable analysis, TP53mut status was associated with less favorable time to next treatment and overall survival. Frequencies of other biomarkers varied by molecular subtype. The Cancer Genome Atlas (TCGA)/Proactive Molecular Risk Classifier for Endometrial Cancer (ProMisE) molecular classifier for endometrial carcinoma can be recapitulated using CGP and provides prognostic stratification even within an advanced or recurrent disease cohort. CGP for molecular subclassification could support trial design and enrollment and inform treatment escalation or selection.