8528 Background: The use of durvalumab after chemoradiation in locally advanced non-small cell lung cancer (NSCLC) patients significantly improves overall survival. However, it is unclear whether this benefit applies to all genetic subtypes of lung cancer. We hypothesize that patients with driver mutation NSCLC may derive less benefit from consolidation durvalumab. Methods: Using the Genomic Marker-Guided Therapy Initiative (GEMINI) database at MD Anderson, we identified 134 patients who were treated with chemoradiation followed by durvalumab for NSCLC. We segregated patients with driver mutations to targetable (EGFR, ALK translocation, ROS1 fusion, MET exon 14 skipping, RET fusion, and/or BRAF) (N = 24) and those driven by canonical KRAS mutations (N = 26). The rest (N = 84) had none of these mutations. We gathered demographic, treatment, and outcome data and compared progression-free survival (PFS) and overall survival (OS) using the Kaplan-Meier method. We used multivariate regression analysis to account for demographic and treatment variables. Results: For our cohort, median age at diagnosis was 64.8, 52% were female (n = 70), and median follow up was 1.5 years. 86% of patients have a history of smoking (n = 115). 21% had squamous cell histology (n = 28). 2 patients had stage IIA disease, 6 had stage IIB, 48 had stage IIIA, 56 had stage IIIB, 13 had stage IIIC, and 9 had stage IV. 73 patients had progression after durvalumab and 37 patients died. Patients with driver mutations had significantly worse median PFS compared to those without driver mutations (8.9 mo vs 26.6 mo; HR 2.62 p < 0.001). Patients with KRAS mutations had particularly poor PFS (Median 7.9 mo, HR 3.34, p < 0.001), while patients with targetable driver mutations trended to worse PFS (Median 14.5 mo, HR 1.96, p = 0.056). The median OS for the cohort was 4.8 yrs with no significant differences based on driver mutation status. On multivariate analysis, only driver mutation status was associated with PFS, but not OS. For patients with first progression, we found the targetable driver group to have significantly improved time to second objective progression (PFS2) compared to the KRAS (HR 0.28, p = 0.011) or non-mutated group (HR 0.38, p = 0.025). All patients in the targetable driver group received targeted therapy after first progression. Conclusions: Our results suggest that patients with driver mutations have worse PFS compared to patients without these mutations after chemoradiation. However, patients with targetable oncogene driver mutations have significantly improved prognosis after initial progression compared to the other groups, likely due to targeted therapy, suggesting that these therapies, including novel approaches towards KRAS mutants, should be further explored in this setting.
Snowball: resampling combined with distance-based regression to discover transcriptional consequences of a driver mutation
