Prospective blinded evaluation predicting efficacy of adjuvant cisplatinum and vinorelbine by a multigene assay after radical surgery in non-small cell lung cancer

Background: This study evaluated the predictive effect of a mRNA-based Drug Response Predictor DRP in the adjuvant setting of non-small cell lung cancer (NSCLC). We apply a mathematical method combining in vitro sensitivity with gene expression patterns in tumors (Knudsen et al PLoS ONE 2014 9(2): e87415.). The cisplatinum DRP has previously been tested in three independent datasets (ASCO Abstract 2015 e18502). Methods: RNA was isolated from formalin-fixed paraffin embedded tumor tissue obtained after radical surgical resection in 95 consecutive stage Ib-IIIb NSCLC patients of different histotypes. Adjuvant chemotherapy was four courses of i.v. cisplatinum and vinorelbine. Affymetrix HG-133_Plus_2 microarray method was applied. Multigene biomarker profiles for cisplatinum and vinorelbine were evaluated blinded on the samples. Primary endpoint was overall survival (OS). Results: Baseline demographics are representative of a NSCLC population. Follow-up time was in median 76.6 months (range 3.5-7 years) after surgery. Statistical analysis suggested that time-dependent analysis of OS was warranted. Using a time dependent Cox model stratifying at 3 years, the hazard ratio (HR) for OS up to 3 years was 0.50 (95% CI: 0.30-0.84, p = 0.008) and the HR for cancer-specific death was 0.46 (95% CI: 0.27-0.78, p = 0.004). Results for multivariable analyses for OS are shown in the table. The DRP remained significant in the multivariable analysis for each endpoint. Conclusions: The DRP for the combined cisplatinum and vinorelbine adjuvant treatment of resectable NSCLC offers definite prediction of effect of the two drugs since it is significant within 36 months after surgery.