Identification of patients with ovarian cancer who are experiencing the highest benefit from bevacizumab in first-line setting based on their tumor intrinsic chemosensitivity (KELIM): GOG-0218 validation study.

Abstract

5553 Background: In patients with high-grade ovarian cancer in first-line setting, predictive factors of bevacizumab efficacy are needed, for selecting patients. In ICON-7 trial, a poor tumor intrisic chemosensitivity (defined by unfavorable modeled CA-125 kinetic ELIMination rate constant KELIM) was a predictive biomarker. Among patients with high-risk diseases, only those with unfavorable KELIM had survival benefit from bevacizumab (mOS: 29.7 vs 20.6 months, HR = 0.78)(Colomban. JNCI CS 2020). The objective was to perform an external validation in GOG-0218 trial (NCT00262847). Methods: In GOG-0218, 1,873 patients were treated with carboplatin-paclitaxel +/- concurrent bevacizumab/placebo followed by a 15 month maintenance. Patient KELIM values were estimated with longitudinal CA-125 kinetics during the first 100 chemotherapy days. The association between KELIM score (categorized as favorable ≥ 1, or unfavorable < 1) and efficacy of bevacizumab (bevacizumab-concurrent + maintenance, vs placebo) for PFS and OS was assessed using univariate/multivariate analyses, in a Training set with 2/3 patients managed the investigators, and then a Validation set with all patients, managed by NGR-GOG. Results: KELIM was assessable in 1,662 patients with ≥ 3 CA-125 available values. In both sets, the patients with unfavorable KELIM derived benefit from bevacizumab compared to placebo (Training: PFS, HR = 0.65 [0.54-0.80]; OS, HR = 0.80 [0.65-0.99]; Validation: PFS, HR = 0.69 [0.59-0.82]; OS, HR = 0.87 [0.73-1.03]), whilst those with favorable KELIM had no benefit from bevacizumab (Training: PFS, HR = 0.96 [0.75-1.23]; OS, HR = 1.05 [0.80-1.37]; Validation, PFS, HR = 0.96 [0.79-1.17]; OS HR = 1.11 [0.89-1.84]). The highest benefit was observed in patients with high-risk diseases (stage IV or sub-optimally resected stage III) characterized by unfavorable KELIM, for PFS (Learning (n = 276): mPFS: 9.0 vs 5.2 months, HR = 0.61 [0.48-0.78]; Validation (n = 433): mPFS: 9.1 vs 5.6 months, HR = 0.64 [0.53-0.78]), and for OS (Learning (n = 278): mOS: 38.9 vs 27.9 months, HR = 0.72 [0.56-0.93], Validation set (n = 438): mOS: 35.1 vs 29.1 months, HR = 0.79 [0.65-0.97]). Conclusions: This validation analysis of GOG-0218 trial confirms the outcomes of ICON-7 trial about the association between poor tumor chemosensitivity and benefit from concurrent + maintenance bevacizumab, suggesting that bevacizumab is mainly effective in patients with poorly chemosensitive diseases. No benefit was found in patients with favorable KELIM. The patients who derived the highest benefit from bevacizumab in PFS and OS (OS absolute benefit ̃ 6 to 9 months) were those with high-risk diseases (stage IV, or incompletely resected stage III) associated with an unfavorable KELIM score (calculator on https://www.biomarker-kinetics.org/CA-125).