Article Text
Abstract
Objectives To present a nomogram for prediction of overall survival (OS) in locally advanced cervical cancer patients (LACC) undergoing definitive radiochemotherapy including image guided adaptive brachytherapy (IGABT).
Methods We retrospectively reviewed 720 patients with LACC acrrued onto the RetroEMBRACE database (12 international institutions). All patients were treated with External Beam Radiotherapy (EBRT) concurrent with Chemotherapy (CHT) and MRI/CT-guided brachytherapy. Missing data (7.2%) were imputed by multiple imputations and predictive mean matching. Stepwise selection of predictive factors with the Akaike information criterion (AIC) was used to obtain a predictive model and construct a nomogram for OS predictions 60 months from diagnosis. The model was internally validated by concordance probability as a measure of discrimination and a calibration plot (both corrected for optimism using bootstrap cross-validation).
Results Through the predictive model (AIC), seven factors were selected to develop the nomogram: FIGO stage (2B vs.1A,1B,2A; 3A,3B,4A,4B vs.1A,1B,2A), age, corpus involvement, lymph node status, concurrent chemotherapy given, overall treatment time, volume of the high risk clinical target volume at the time of first brachytherapy (CTVHR). CTVHR is considered surrogate for response at concurrent EBRT+CHT. This OS nomogram had a satisfactory calibration and useful discrimination (concordance probability c=0.74).
Conclusions Prognostic factors were used to develop the first nomogram for 5-year OS in patients with LACC in the setting of IGABT treatment. Response to concurrent EBRT and CHT (CTVHR) seems to be a new essential outcome predictor for OS. These nomograms can be used to better estimate individual and collective outcomes and may facilitate personalized patient counselling during the treatment.