Outcome according to residual disease (surgeon's report vs pre-chemotherapy imaging) in patients with bevacizumab-treated ovarian cancer: Analysis of the ROSiA study.

Background and Objectives The single-arm ROSiA study evaluated frontline bevacizumab for advanced ovarian cancer. We explored how discordant surgically and radiologically assessed postoperative residual disease affects outcomes. Methods After debulking surgery, 1021 patients received 4 to 8 cycles of carboplatin-paclitaxel plus bevacizumab until progression or up to 24 months. The primary endpoint was safety; progression-free survival (PFS) was a secondary endpoint. We performed post hoc exploratory PFS analyses in four subgroups: surgeon-reported no visible residuum (NVR) without target lesions; surgeon-reported NVR with target lesions; macroscopic (<= 1 cm) residuum; and >1 cm residuum. Results Surgical and radiological assessments were concordant in 94% of patients; 61 patients (6%; 21% of those with surgeon-reported NVR) had NVR with target lesions. Median PFS was numerically longest in patients with concordant surgically/radiologically assessed NVR (35.5 months), intermediate for surgeon-reported NVR with target lesions (31.8 months), and shortest for visible residuum (27.9 and 20.2 months for visible residuum <= 1 and >1 cm, respectively). One-year and 2-year PFS rates showed the same pattern. Conclusions These analyses suggest that prognosis is potentially worse in patients with radiologically detected target lesions despite surgeon-reported NVR compared with concordant NVR by both assessment methods. Postsurgical imaging may add valuable prognostic information.