P78.15 Real-World Prognostic Model of Overall Survival in Patients With Advanced NSCLC Receiving Anti-PD1/PD-L1 as 2L Monotherapy

Abstract

Listen

Translate article

The objective of this retrospective, observational, non-interventional cohort study was to investigate prognostic factors of overall survival and build a model in patients with advanced non-small cell lung cancer (NSCLC) who received anti-PD1/PD-L1 immune checkpoint inhibitors (atezolizumab, nivolumab and pembrolizumab) as a single agent in second-line (2L) treatment. Patients with advanced NSCLC newly diagnosed between January 2011 and February 2020 who received atezolizumab, nivolumab or pembrolizumab as 2L monotherapy were included for analysis from the Flatiron Health electronic health record-derived de-identified database. Patients could not have received first-line treatment with clinical study drug(s) or anti-PD1/PD-L1 nor anti-CTLA4 immune checkpoint inhibitors. A Cox model with lasso regularization was used to build the prognostic model based on baseline clinical characteristics. The model was validated externally in 406 patients receiving 2L atezolizumab monotherapy from the OAK Phase III trial (NCT02008227). A total of 4224 patients were included with a median age of 69 years; 54% were male, 76% white, 67% had Stage IV at initial diagnosis and 69% had non-squamous histology. Top risk prognostic characteristics were ECOG performance status 2 and 3, older age, abnormal albumin and abnormally high calcium levels. Top protective prognostic characteristics were PD-L1 expression, extended time from advanced diagnosis to 2L treatment start date, abnormally high chloride levels, previous/current smoking history and abnormally low white blood cell count. The model was able to predict overall survival and showed a good performance in both the real-world cohort and the OAK trial (Figure). The prognostic model was able to predict overall survival in patients with advanced NSCLC receiving anti-PD1/PD-L1 immune checkpoint inhibitors as 2L monotherapy based on demographics and baseline clinical factors from a real-world setting.