Poor concordance between CA-125 and RECIST at the time of disease progression in patients with platinum-resistant ovarian cancer: analysis of the AURELIA trial

Abstract

Data on CA-125 as a predictor of disease progression (PD) in ovarian cancer come predominantly from patients with platinum-sensitive disease receiving chemotherapy alone. We assessed concordance between CA-125-defined and RECIST-defined PD using data from the Gynecologic Cancer InterGroup (GCIG) randomized phase III AURELIA trial in platinum-resistant ovarian cancer (PROC). Patients with PROC were randomized to receive single-agent chemotherapy with or without bevacizumab. PD by CA-125 was defined according to GCIG criteria (except that confirmatory CA-125 measurement was not required). This exploratory analysis included patients with RECIST PD and a CA-125 reading ≤28 days before and ≤21 days after RECIST-defined PD. Of 218 eligible patients, only 94 (43%, 95% confidence interval 36% to 50%) had concordant RECIST and CA-125 PD status (42% in the chemotherapy-alone arm; 45% in the bevacizumab combination arm, P = 0.6). There was no evidence of CA-125-defined PD in the remaining 124 patients despite PD according to imaging. There were no significant differences in baseline characteristics between patients with PD defined by both RECIST and CA-125 and those with RECIST-only PD. CA-125 was even less sensitive in detecting PD in patients with early (<8 weeks after randomization) compared with later RECIST-defined PD (69% versus 53%, respectively, not meeting CA-125 criteria; P = 0.053). There was no significant difference in survival after PD in patients with concordant PD by RECIST and CA-125 versus those with PD only by RECIST. We validated our findings in an independent study population of PROC. In this platinum-resistant population, PD was typically detected earlier by imaging than by CA-125, irrespective of bevacizumab treatment. Disease status by CA-125 at the time of PD was not prognostic for overall survival. Regular radiologic assessment as well as symptom benefit assessment should be considered during PROC follow-up.