Development and validation of longitudinal c-reactive protein as dynamic response predictor for PD-L1 blockade in advanced NSCLC: Findings from four atezolizumab clinical trials.

Abstract

e21113 Background: Numerous studies suggested that pre-treatment C-reactive protein (CRP) is a survival predictive biomarker for cancer patients treated with anti-PD-(L)1 blockade. However, the application of longitudinal CRP levels as dynamic biomarker for advanced non-small cell lung cancer (NSCLC) patients treated with anti-PD-(L)1 remains unexplored. Methods: This study analyzed four international, multi-center clinical trials (OAK, BIRCH, POPLAR and FIR trials) to conduct post-hoc analyses of NSCLC patients undergoing atezolizumab (anti-PD-L1) single-agent treatment. CRP test was performed ≤ 96 hours before Day 1 of each cycle. 1438 patients with more than 2 times CRP results (a total of 11,253 records) in all 3 collection timepoints. The primary endpoint was overall survival (OS). First, multivariate survival analysis was conducted to include clinical factors such as gender, ECOG-PS, race and metastases as contributing risk factors. Then, multivariate joint modelling of longitudinal and time-to-event data was used to establish the relationship between longitudinal CRP and OS in the OAK study, whereas external validation was performed by time-dependent AUC analysis on the BIRCH, POPLAR and FIR studies. Results: Log-transformed longitudinal CRP levels (log(CRP)) in combination with clinical factors in absence of PD-L1 expression emerged as independent predictors of worse OS in the OAK study (HR of jointed model = 2.08, 95% CI: 1.84-2.35, p < 0.001) with a high accuracy (AUC = 0.74) for predicting OS. The predictive value of longitudinal CRP was validated in the external validation cohorts with good performance in OS (BIRCH: HR = 2.77 (95% CI: 2.31-3.44), AUC = 0.76; POPLAR: HR = 1.95 (95% CI: 1.50-2.64), AUC = 0.76; and FIR: HR = 1.83 (95% CI: 1.46 -2.39), AUC = 0.72). Conclusions: The Bayesian multivariate joint model demonstrated that longitudinal CRP is a strong survival dynamic predictor for atezolizumab-treated NSCLC patients.