Abstract 409: A novel algorithm to redefine blood-based tumor mutational burden for optimized prediction of clinical benefits from immunotherapy

Abstract

Abstract Purpose: Blood-based tumor mutational burden (bTMB) was recently identified as a predictor of progression free survival (PFS) benefit for non-small cell lung cancer (NSCLC) patients treated with atezolizumab over docetaxel. However, the reported bTMB failed to predict overall survival (OS) outcome from atezolizumab over docetaxel, or to predict PFS or OS benefit in the atezolizumab arm upon our analysis. We further revealed that bTMB correlates with maximum somatic allele frequency (MSAF), which is an independently negative prognostic factor and impeded the predictive value of bTMB. Here we aimed to explore a novel algorithm to eliminate the interference from MSAF in bTMB calculation and redefine bTMB to predict clinical benefits in both PFS and OS from immunotherapy. Patients and Methods: Data from POPLAR (NCT01903993, N = 211) and OAK (NCT02008227, N = 642) trials was first adapted for algorithm development in bTMB redefinition. The derived algorithm was validated with another gene panel (NCC-GP150) in two independent NSCLC cohorts with blood samples (N = 184) and patients treated with mono-immunotherapy (N = 50). Results: The previous reported bTMB-H (bTMB ≥ 16) was not associated with favorable PFS (P = 0.09, HR 0.81, 95% CI, 0.64-1.03) or OS (P = 0.68, HR 1.06; 95% CI, 0.81-1.38) in patients treated with immunotherapy, instead, it was associated with poor OS outcomes (HR 1.29, 95% CI, 1.02-1.63). Further analysis revealed that bTMB correlates with MSAF (Pearson r = 0.47), which is a negative prognostic biomarker. As a result, the predictive value of bTMB was impeded by the negative prognostic performance contributed by a high MSAF. After we redefined bTMB by filtering out mutations with AF > 5%, the correlation between redefined bTMB and MSAF was no longer observed (Pearson r = 0.09) and the negative prognostic property of bTMB was eliminated. More importantly, the redefined bTMB-H (cutoff=12) serves as an optimal predictor for both PFS and OS from atezolizumab over docetaxel as tested in POPLAR (PFS HR 0.41, 95% CI, 0.21-0.80; OS HR 0.34, 95% CI, 0.16-0.71) and validated in OAK (PFS HR 0.49, 95% CI, 0.33-0.71; OS HR 0.47, 95% CI, 0.29-0.77), respectively. Meanwhile, the redefined bTMB-H can identify patients with favorable PFS (HR 0.62, 95% CI, 0.47-0.80) and OS (HR 0.70, 95% CI, 0.52-0.95) in the immunotherapy arm. Consistent results were obtained when this algorithm was tested in our independent cohort (Pearson r reduced from 0.36 to 0.07; bTMB-H population with immunotherapy, PFS HR 0.28, 95% CI, 0.09-0.94). Conclusion: MSAF plays a dominant role in previous reported bTMB and should be adjusted in bTMB calculation to remove the prognostic property contributed by MSAF. We have redefined bTMB and demonstrated the feasibility of the redefined bTMB as an actionable predictor of favorable PFS and OS from immunotherapy in future clinical practice. Citation Format: Zhijie Wang, Guoqiang Wang, Jianchun Duan, Jing Zhao, Zhengyi Zhao, Hua Bai, Shuhang Wang, Shangli Cai, Jie Wang. A novel algorithm to redefine blood-based tumor mutational burden for optimized prediction of clinical benefits from immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 409.