Prognostic role of Ca125 response criteria and RECIST criteria: Analysis of results from the MITO-3 phase III trial of gemcitabine versus pegylated liposomal doxorubicin in recurrent ovarian cancer

Objectives. The aim of the study was to investigate i) the association between the Ca125 based and the RECIST assessed response in recurrent ovarian cancer patients enrolled in a Phase III randomized trial (MITO-3) comparing salvage treatment with pegylated liposomal doxorubicin (PLD) versus gemcitabine (GEM); ii) the correlation between the early modifications of Ca125 levels during treatment and the RECIST assessed response; iii) the prognostic value of response based on Ca125 and the RECIST criteria. Methods. Assessment of response was performed by the RECIST and the GCIG criteria. The prognostic impact of the response by the RECIST criteria and the GCIG criteria was analyzed by the landmark method. Results. Overall, of 30 cases defined as responders on the basis of the GCIG criteria, 20 resulted as responders according to the RECIST criteria (NPV=66.7%); conversely, 93.7% of the cases considered not responders based on the GCIG criteria were defined as unresponsive at RECIST evaluation. Early modifications of Ca125 levels were not completely predictive of the ultimate RECIST defined response. Overall survival (OS) was longer in RECIST defined responders versus non responders, although the statistical significance was not reached (P value=0.092); conversely, median OS was significantly longer in GCIG defined responders than in non responding patients (P value=0.0059). In multivariate analysis, the GCIG assessed response maintained its independent association with OS. Conclusions. GCIG criteria for tumor response could replace the conventional assessment of response in the decision making process relative to the discontinuation or prolongation of the salvage treatment in ovarian cancer. (c) 2008 Elsevier Inc. All rights reserved.