Nomogram Predicting Overall Survival in Patients With Locally Advanced Cervical Cancer Treated With Radiochemotherapy Including Image-Guided Brachytherapy: A Retro-EMBRACE Study

Purpose To present a nomogram for prediction of overall survival (OS) in patients with locally advanced cervical cancer (LACC) undergoing definitive radiochemotherapy including image-guided adaptive brachytherapy (IGABT). Methods and Materials Seven hundred twenty patients with LACC treated with radiochemotherapy including IGABT in 12 institutions (median follow-up 56 months) were analyzed; 248 deaths occurred. Thirteen candidate predictors for OS were a priori chosen on the basis of the literature and expert knowledge. Missing data (7.2%) were imputed using multiple imputation and predictive mean matching. Univariate analysis with a multivariable Cox regression model for OS stratified by center was performed. Stepwise selection of predictive factors with the Akaike Information Criterion was used to obtain a predictive model and construct a nomogram for OS predictions 60 months from diagnosis; this was internally validated by concordance probability as a measure of discrimination and a calibration plot. Results Thirteen potential predictive factors were evaluated; 10 factors reached statistical significance in univariate analysis (age, Hemoglobin, FIGO Stage2009, tumor width, corpus involvement, lymph node involvement, concurrent chemotherapy, dose to 90% of the high-risk clinical target volume, volume of CTV at the first brachytherapy [CTVHRVolumeBT], overall treatment time [OTT]). Four factors were confirmed significant within the multivariable Cox regression model (FIGO Stage2009, lymph node involvement, concurrent chemotherapy, CTVHRVolumeBT). The predictive model and corresponding nomogram were based on 7 Akaike Information Criterion–selected factors (age, corpus involvement, FIGO Stage2009, lymph node involvement, concurrent chemotherapy, CTVHRVolumeBT, OTT) and showed promising calibration and discrimination (cross-validated concordance probability c = 0.73). Conclusions This is the first nomogram to predict OS in patients with LACC treated with IGABT. In addition to previously reported factors (age, FIGO2009 stage, corpus involvement, chemotherapy delivery, OTT, lymph node involvement), status of primary tumor at the time of brachytherapy seems to be an essential outcome predictor. These results can facilitate individualized tailoring of treatment and patient counseling during the treatment. To present a nomogram for prediction of overall survival (OS) in patients with locally advanced cervical cancer (LACC) undergoing definitive radiochemotherapy including image-guided adaptive brachytherapy (IGABT). Seven hundred twenty patients with LACC treated with radiochemotherapy including IGABT in 12 institutions (median follow-up 56 months) were analyzed; 248 deaths occurred. Thirteen candidate predictors for OS were a priori chosen on the basis of the literature and expert knowledge. Missing data (7.2%) were imputed using multiple imputation and predictive mean matching. Univariate analysis with a multivariable Cox regression model for OS stratified by center was performed. Stepwise selection of predictive factors with the Akaike Information Criterion was used to obtain a predictive model and construct a nomogram for OS predictions 60 months from diagnosis; this was internally validated by concordance probability as a measure of discrimination and a calibration plot. Thirteen potential predictive factors were evaluated; 10 factors reached statistical significance in univariate analysis (age, Hemoglobin, FIGO Stage2009, tumor width, corpus involvement, lymph node involvement, concurrent chemotherapy, dose to 90% of the high-risk clinical target volume, volume of CTV at the first brachytherapy [CTVHRVolumeBT], overall treatment time [OTT]). Four factors were confirmed significant within the multivariable Cox regression model (FIGO Stage2009, lymph node involvement, concurrent chemotherapy, CTVHRVolumeBT). The predictive model and corresponding nomogram were based on 7 Akaike Information Criterion–selected factors (age, corpus involvement, FIGO Stage2009, lymph node involvement, concurrent chemotherapy, CTVHRVolumeBT, OTT) and showed promising calibration and discrimination (cross-validated concordance probability c = 0.73). This is the first nomogram to predict OS in patients with LACC treated with IGABT. In addition to previously reported factors (age, FIGO2009 stage, corpus involvement, chemotherapy delivery, OTT, lymph node involvement), status of primary tumor at the time of brachytherapy seems to be an essential outcome predictor. These results can facilitate individualized tailoring of treatment and patient counseling during the treatment.