Introduction: In recurrent ovarian cancer, CA-125 could be the only objective response criteria. This study analyzes response patterns regarding CA-125 in responders versus nonresponders and determines whether a specific cutoff value for CA-125 could predict clinical response, compared with response evaluation criteria in solid tumors, in patients receiving pegylated liposomal doxorubicin (PLD).

Methods: Sixty-eight patients were identified, 78% were platinum resistant. Relative changes in CA-125 values were calculated, and response was defined as higher than 50% reduction in CA-125 from baseline. Receiver operating characteristic (ROC) curves were constructed based on CA-125 value after the first cycle of PLD to evaluate the most precise cutoff point for the decision model predicting response.

Results: Fifty-three patients were assessable for response: 16 patients responded and 37 did not; the median increase of CA-125 was 0.20 (−63; 312) and 52 (−29; 620), respectively. Our ROC curve generated a cutoff value with a sensitivity of 35% (positive test, the proportion of patients who will not respond) and a predictive positive value of 80%. According to the predictive positive value, 20% of the responder patients will be identified as nonresponders; P = 0.025.

Conclusions: Our ROC analysis did not demonstrate any reliable CA-125 cutoff on response. Discontinuation of the therapy before cycle 3 may exclude some patients who will benefit from PLD.