Objective The aim of this study was to identify predictors of the response to paclitaxel-carboplatin chemotherapy (TC) in recurrent or patients with advanced cervical cancer.

Methods The records of 61 consecutive women with recurrent or advanced cervical cancer who were treated with TC were retrospectively reviewed. Data regarding their primary disease, follow-up, recurrence, and the activity and toxicity of TC were collected. Multivariate analysis was performed using the Cox proportional hazards regression model to identify predictors of the response to TC. Survival was calculated using the Kaplan-Meier method and compared using the log-rank test.

Results Overall, TC was well tolerated and displayed a response rate of 60.7% (19 complete response and 18 partial response). The median progression-free survival was 14 months for all patients and 20 months for the responders. Grade 3 to grade 4 toxicities were observed in 51 patients (83.6%). Multivariate analysis revealed that performance status, symptom status, and prior chemotherapy were independent prognostic predictors of a poor response. Patient survival was inversely correlated with the number of these prognostic factors. When the patients were divided into 2 prognostic groups (low risk: patients with no or one poor prognostic factor; and high-risk: patients with 2 or more poor prognostic factors), the patients in the high-risk group had a significantly shorter progression-free survival than those in the low-risk group (4 vs 16 months, log-rank; P < 0.0001).

Conclusions The combination of paclitaxel and carboplatin is effective in patients with recurrent or advanced cervical cancer. Our prognostic model composed of 3 clinical variables might enable physicians to identify patients who would not derive clinical benefit from TC and offer them the opportunity to receive other types of treatment.