Clinical and molecular characteristics of responders versus non-responders to immune checkpoint inhibitors (ICI) in NSCLC.

114 Background: Evidence from randomized clinical trials and retrospective studies has shown that tumors with specific driver mutations may be less likely to respond to ICI; the optimal timing of ICI is unknown in this population. In this analysis, we evaluated the characteristics of responders to ICI compared to non-responders, with emphasis on the role of driver mutations. Methods: We retrospectively collected clinical characteristics and outcomes for patients who received ICI for NSCLC at Stanford University between April 2015 and August 2016. Patients were classified as responders if time on ICI was 180 days or more, and as non-responders if less than 180 days. Outcomes included radiographic improvement while on ICI, survival from diagnosis, and survival from ICI initiation. Results: Data were available for 84 patients, with median follow-up of 31 months. 18 patients were deemed responders (21%); 66 were deemed non-responders (79%). Of the patients who underwent genomic testing, 13% (2/15) of responders had mutations in EGFR/ALK/ROS1, compared to 32% (17/49) of non-responders (p = 0.195). Within the responders, 89% (16/18) had an associated radiographic improvement, compared to 11% (7/66) of non-responders (p < 0.0001). Patients with mutations in EGFR/ALK/ROS1 had a median overall survival of 46 months from diagnosis, compared to 26 months in those without these mutations (p = 0.08). Patients with mutations in EGFR/ALK/ROS1 had a median survival of 4.9 months from ICI initiation, compared to 7.0 months in those without (p = 0.45). Conclusions: Patients receiving ICI for 180 days or more had a significantly higher rate of radiographic improvement, suggesting that time on therapy appears to be a surrogate endpoint of clinical benefit. While there was not a statistically significant smaller number of responders with EGFR/ALK/ROS1, there appeared to be clinically meaningful lower response rate in these patients. This was likely limited by sample size; expansion of the sample size is planned. There was a trend to overall survival benefit in patients with driver mutations, likely driven by targeted therapy as there were no statistical differences in survival from time of ICI initiation.