Genomic profiling of advanced cervical cancer to predict response to PD-1 inhibitor combination therapy: a secondary analysis of the CLAP trial

Abstract

<h3>Objectives:</h3> The CLAP trial (NCT03816553) is a multicenter, single-arm, phase II study in patients with metastatic, recurrent, or persistent cervical cancer who were treated with PD-1 inhibitor camrelizumab plus a VEGFR2 inhibitor apatinib. In this study, we performed genomic profiling analysis to identify potential predictive biomarkers for this combination therapy. <h3>Methods:</h3> Genomic profiling was performed on 32 patients with available biopsy or surgical samples by targeted next-generation sequencing of 425 cancer-related genes. Somatic alterations and tumor mutational burden (TMB) were assessed for their predictive value on objective response rate (ORR), progression-free survival (PFS) and overall survival (OS). The Cancer Genome Atlas (TCGA) public dataset was used for validation. <h3>Results:</h3> The ORR of the current cohort was 65.6%, which was consistent with the whole CLAP population. The median PFS and OS were not reached. Frequencies of genetic alterations (<i>PIK3CA,</i> 43.8%; <i>STK11</i>, 25%; <i>FBXW7,</i> 15.6%; <i>PTEN,</i> 15.6%; <i>TP53,</i> 15.6%) were similar to previous reports of cervical cancer. Mutations in PI3K pathway genes were found in 68.75% (22/32) of the patients. <i>PIK3CA</i> mutations (57.1% vs. 18.2%, P=0.03) were significantly enriched in squamous cell carcinoma (SCC), whereas all mutations of <i>TP53, ARID1B, DICER1</i>, ERBB3, <i>NF1</i>, and <i>TEK</i> were detected in adenocarcinoma. Mutations in <i>PIK3CA, PTEN,</i> and <i>ERBB3</i> were significantly associated with survival (univariate, P ≤ 0.05). <i>PIK3CA</i> mutations (PFS hazard ratio [HR], 0.33; P=0.05; OS HR, 0.23; P=0.04) and <i>PTEN</i> mutations (PFS HR, 3.71e-09; P=0.05; OS HR, 3.64e-09; P=0.08) were significantly associated with improved clinical outcome. In contrast, <i>ERBB3</i> mutations (PFS HR, 34.9; P<0.001; OS HR, 19.8; P<0.001) correlated with poor survival. Mutations in the PI3K pathway genes showed improved predictive power compared with <i>PIK3CA</i> or <i>PTEN</i> mutations alone (PFS HR 0.33, P=0.03; OS HR 0.25, P=0.02). The predictive effects of these mutations on PFS, but not on OS, were validated using the TCGA dataset. Furthermore, TMB-high (TMB ≥5 mut/Mb) were associated with prolonged PFS (HR 0.26, P<0.01) and OS (HR 0.31, P=0.05). Consistent with the CLAP trial, patients in this study with positive PD-L1 expression showed better outcome (PFS HR 0.36, P=0.08; OS HR 0.31, P=0.05). TMB combined with PD-L1 expression further stratified clinical benefit for patients, with ORR of 81.3% in TMB-high/PD-L1-positive and 25% in TMB-low/PD-L1-negative subgroup (P<0.05). Multivariate analysis showed that <i>ERBB3</i> mutation (PFS P=0.01; OS P<0.001), positive PD-L1 expression (PFS P=0.01; OS P=0.05), and high TMB (PFS P=0.01; OS P=0.05) remained significantly associated with survival. <h3>Conclusions:</h3> In this study, we uncovered <i>PIK3CA, PTEN, ERBB3,</i> PI3K pathway genes mutations and TMB as novel predictive biomarkers in cervical cancer patients treated with PD-1 inhibitor combination therapy, which might be of great clinical relevance in patient stratification.