Molecular profiling as predictor of outcomes in a Brazilian cohort of stage IV lung cancer.

Abstract

e20668 Background: Lung cancer is the leading cause of cancer deaths globally. In stage IV non-small cell lung cancer (NSCLC), identification of a tumor driver mutation is critical, since it tailors treatment. We aimed to analyze clinical outcomes according to molecular profiling in patients (pts) with stage IV NSCLC. Methods: In this retrospective cohort study, we enrolled 57 pts with stage IV NSCLC treated at a public hospital in Southern Brazil between 2016 and 2018. Results: Median follow-up was 20.3 months, 53% were men, mean age was 65 years, 86% had smoked, 84% had de novo metastatic NSCLC and 96% had non-squamous carcinoma. Regarding molecular profiling, somatic mutations in KRAS, EGFR and BRAF were identified in 33%, 16% and 4% of pts, respectively. None of the pts had ALK mutations, and in 47% no identifiable driver mutation was found. In the EGFR-mutated subgroup, 67% had a deletion of exon 19, 22% had the exon 21 L858R mutation and 11% had exon 20 mutations. Palliative systemic therapy (PST) was delivered to 60% of the pts. Two or more lines of PST were delivered to 23%. The main reason for upfront best supportive care was ECOG PS 3-4 (poor). In the subgroup of pts with sensitizing EGFR mutations (8 pts), 75% received Gefitinib, the only anti-EGFR drug available in our public health system; the other pts died before recognition of the mutation. Median Progression Free-Survival was 6.3 months overall, 10.3 months for EGFR-mutated pts, 7.6 months for KRAS-mutated, 7.5 months for BRAF-mutated, and 5.2 months for pts with no mutation identified. Median Overall Survival (OS) was 7.7 months overall, 13.2 months for EGFR-mutated pts, 7.4 months for KRAS-mutated, 12.9 months for BRAF-mutated, and 5.3 months for pts with no mutation identified. In the Cox regression multivariate analysis, poor PS (HR 3.80, P < 0.0001) and second-line PST (HR 0.23, P = 0.002) were independent predictors of OS. No driver mutations were predictors of OS, although we did find a tendency towards better OS in pts with EGFR mutations and worse OS in pts with KRAS mutations or no identifiable mutation. Conclusions: In this cohort, genotyping results did not predict survival outcomes. This is probably due to the small number of pts studied, and data should be reanalyzed in larger cohorts.