LAUSR

105 Background: The use of immune targeted agents for NSCLC has improved outcomes for patients (pts) with metastatic disease. Biomarkers such as the neutrophil-to-lymphocyte ratio (NLR) prior to therapy have been shown to predict outcomes however little is known about dynamic changes in immune subset composition during therapy. Methods: A single institution, retrospective review was performed of 88 NSCLC pts who received anti-PD1 therapy. NLR and relative lymphocyte count (RLC) were recorded at baseline, 4 weeks after treatment initiation, and at every subsequent computed tomography scan until time of progression. Endpoints included overall survival (OS) and durable clinical benefit (DCB), defined as stable disease or response at 6 months after therapy. Non parametric tests were used to compare NLR and RLC between responders and non-responders, cox-proportional hazards regression analysis was used for univariate associations with OS and Kaplan-Meier survival curves were compared by the log-rank test. Results: Our cohort comprised of 60 male pts with a median age of 63 years and a median OS of 38.5 months. Baseline NLR and RLC were not associated with response to therapy (Mann Whitney p=0.25 and p=0.23, respectively). We identified a statistically significant higher RLC and lower NLR at week 4 in pts with DCB (Mann Whitney p=0.001). Continuous NLR and RLC values at 4 weeks were statistically significantly correlated with OS (HR=1.008, 95% CI 1.001-1.015, p=0.03 and HR=0.939, 95% CI 0.9-0.98, p=0.004 respectively). Using median RLC at 4 weeks as a threshold, pts with high RLC at 4 weeks had significantly favorable survival (log-rank p<0.0001). Using the previously reported cut-off point of NLR=4, pts with NLR>4 had significantly worse OS (log-rank p<0.0001). For pts with acquired resistance to therapy, RLC increased early during treatment followed by a decrease at the time of progression. Conclusions: Our findings suggest early immune cell subset dynamics are associated with clinical outcomes for NSCLC pts treated with anti-PD1 therapy. Our observations reflect the differential immune repertoire shifts and if prospectively validated may be used to stratify pts receiving immune targeted agents.