Using real world data to examine outcomes in immunotherapy-treated patients with metastatic non-small cell lung cancer.

Abstract

e21715 Background: Important questions have been raised about whether real world data (RWD) can be relied upon to support clinical and regulatory decision-making. The aims of this study were to assess overall survival (OS), time-to-treatment (TTT), follow-up time, and treatment pathways in metastatic non-small cell lung cancer (mNSCLC) patients treated with programmed cell death protein 1 inhibitors (PD-1i). Methods: Two mNSCLC cohorts were identified in a US based EMR network. To explore treatment pathways, patients had to have an advanced (stage 3/4) diagnosis of NSCLC confirmed by a tumor registry record. A line of treatment (LOT) was defined as a PD-1i or chemotherapy taken within 30 days. OS, time to treatment, and follow-up time were also calculated. In another cohort patients had to have at least one PD-1i (pembrolizumab, nivolumab, durvalumab, or atezolizmab) and an oncology treatment within 3 months of an advanced stage diagnosis. OS was calculated using Kaplan-Meier analysis and stratified by time after nivolumab approval, quarter of PD-1i initiation, and line of therapy of first PD-1i. Patient characteristics were defined by ICD, CPT, LOINC, and RxNorm terminology. Results: Median overall survival was found to be 213 days (IQR 109, 425.25). In the treatment pathways analysis, 58.3% of patients started on a non-PD-1i chemotherapy as an initial line of treatment for mNSCLC. PD-1i was the most common second line treatment and 41.4% of patients who started on a non-PD-1i therapy switched to a PD-1i as their second line therapy. Median time from advanced diagnosis to PD-1i inhibitor initiation (9.6 months (IQR 3.45, 21.45)) and median structured follow-up times from advanced diagnosis (21.87 months (IQR 11.94, 38.97)) and from inhibitor initiation (8.71 months (3.06, 17.26)) were comparable to results found in other RWD. Conclusions: Overall survival, time to treatment, and other outcomes were consistent with comparable RWD sources, (Stewart M, 2019; Sean K 2018) regardless of treatment timeframe. We demonstrated that our real world evidence based approach provides a robust method for analyzing clinical outcomes, supporting the validity of real world data to complement clinical trials and inform clinical decisions.