Purpose: In lung adenocarcinoma, EGFR and KRAS mutations dominate the mutational spectrum and have clear therapeutic implications. We sought to determine whether transcriptional subgroups of clinical relevance exist within EGFR-mutated, KRAS-mutated, or EGFR and KRAS wild-type (EGFRwt/KRASwt) adenocarcinomas.
Experimental design: Gene expression profiles from 1,186 adenocarcinomas, including 215 EGFR-mutated, 84 KRAS-mutated, and 219 EGFRwt/KRASwt tumors, were assembled and divided into four discovery (n = 522) and four validation cohorts (n = 664). Subgroups within the mutation groups were identified by unsupervised consensus clustering, significance analysis of microarrays (SAM) analysis, and centroid classification across discovery cohorts. Genomic alterations in identified mutation subgroups were assessed by integration of genomic profiles for 158 cases with concurrent data. Multicohort expression subgroup predictors were built for each mutation group using the discovery cohorts, and validated in the four validation cohorts.
Results: Consensus clustering within the mutation groups identified reproducible transcriptional subgroups in EGFR-mutated and EGFRwt/KRASwt tumors, but not in KRAS-mutated tumors. Subgroups displayed differences in genomic alterations, clinicopathologic characteristics, and overall survival. Multicohort gene signatures derived from the mutation subgroups added independent prognostic information in their respective mutation group, for adenocarcinoma in general and stage I tumors specifically, irrespective of mutation status, when applied to the validation cohorts. Consistent with their worse clinical outcome, high-risk subgroups showed higher expression of proliferation-related genes, higher frequency of copy number alterations/amplifications, and association with a poorly differentiated tumor phenotype.
Conclusions: We identified transcriptional subgroups in EGFR-mutated and EGFRwt/KRASwt adenocarcinomas with significant differences in clinicopathologic characteristics and patient outcome, not limited to a mutation-specific setting.
©2013 AACR.