Maximum somatic allele frequency-adjusted blood-based tumor mutational burden balances the effect of intratumor heterogeneity on response to immune checkpoint Inhibitors in non-small cell lung cancer patients.

Abstract

e21137 Background: ctDNA abundance can influence the accuracy of blood-based tumor mutational burden (bTMB), driving a paradox of bTMB in predicting response and survival benefit of patients with non-small cell lung cancer (NSCLC) receiving immune checkpoint inhibitors (ICIs) therapy. However, the optimal algorithm of ctDNA adjusted bTMB remains largely unestablished. We aimed to develop a novel approach to optimize the calculation of bTMB by integrating maximum somatic allele frequency (MSAF) to better differentiate the ICI-beneficial NSCLC patients, as well as explore the correlated mechanism. Methods: A 1021-gene panel-based targeted capture sequencing was utilized to evaluate the correlation between MSAF and bTMB based on blood samples in 1679 Chinese NSCLC patients (GCGD database, N = 1679). The establishment of MSAF-adjusted bTMB (Ma-bTMB) was derived from three independent NSCLC cohorts treated with programmed death ligand 1 (PD-L1) or the programmed death 1 (PD-1) inhibitors, including POPLAR cohort (N = 231) as training set, and OAK (N = 642) and Wang (N = 81) cohorts as validation sets separately. PyClone was employed to estimate clonal and subclonal populations. Results: MSAF demonstrated a modest positive correlation between bTMB and MSAF (Spearman’s R = 0.41, P < 0.001, GCGD cohort) and was negatively correlated with survival benefit (Spearman’s rank sum test, overall survival (OS): R = −0.39, P < 0.001; progression-free survival (PFS): R = −0.41, P < 0.001, Wang cohort). Patients with high Ma-bTMB (Ma-bTMB-H) demonstrated improved survival outcomes of anti-PD(L)-1 therapy compared to chemotherapy in POPLAR (PFS, hazard ratio (HR) = 0.63, interaction P = 0.030; OS, HR = 0.44, interaction P = 0.020), which was validated in OAK cohorts (PFS, HR = 0.72, interaction P = 0.040; OS, HR = 0.50, interaction P = 0.010). In addition, Ma-bTMB-H was associated with a better clinical benefit compared to low Ma-bTMB (Ma-bTMB-L) in Wang cohort (PFS, HR = 0.55, P = 0.012; OS, HR = 0.46, P = 0.036). Furthermore, there was a positive correlation between intratumor heterogeneity (ITH) and MSAF (R = 0.37, P < 0.001) in GCGD database, and the effect of ITH on Ma-bTMB (subclones, P = 0.57; subclonal mutations, P = 0.8) was substantially reduced compared to traditional bTMB (subclones, P < 0.001; subclonal mutations, P < 0.001) in Wang cohort. Conclusions: We developed and validated Ma-bTMB, a convenient, readily available, non-invasive predictive biomarker, which reduced the confounding effect of ITH and could effectively differentiate beneficiaries of anti-PD(L)1 therapy, warranting investigation in future clinical trials. Clinical trial information: NCC2018-092.