Jesus Fernandez-Gomez, Jesus Rodriguez-Molina, C. Pertusa, Miguel Blas, M. Unda, Eduardo Solsona, M. Gonzalez, Anabel Gimeno, Luis Martínez-Piñeiro, Antonio Ojea, Mariano Rabadan, Santiago Isorna, Jose Emilio Camacho, Ander Astobieta, J.A. Portillo, P. Muntanola, Martínez-Piñeiro Ja, Rosario Madero, and Manuel Montesinos
Background European Organization for Research and Treatment of Cancer (EORTC) risk tables only included 171 patients treated with bacillus Calmette-Guerin (BCG) for non–muscle-invasive bladder cancer (NMIBC). Objective To evaluate the external validity of the EORTC tables in patients with NMIBC treated with BCG over 5–6 mo. Design, setting, and participants Data on 1062 patients treated with BCG were analyzed. Measurements Discrimination was assessed using the concordance index (c-index) and the prognostic separation index (PSEP). For calibration, probabilities of recurrence and progression obtained with the EORTC risk tables in our series were compared with those reported by the EORTC. Results and limitations With respect to the discriminative ability of the EORTC model, c-index was similar to those reported in the EORTC series for recurrence. However, c-indices for progression in our series were lower than c-indices reported by Sylvester et al. [1]. Although PSEP in our series was lower than in the EORTC series for recurrence at 1 yr, similar results were found at 5 yr. Regarding progression, PSEP in our series was lower than in the EORTC series. Whilst a successful stratification of recurrence and progression probability at 1 and 5 yr was achieved using the EORTC tables in our series, model calibration showed lower risks of recurrence than those reported by Sylvester et al. [1] in all groups. For progression, lower risks were found in higher-risk groups. There are some limitations in the present study. A different distribution of patients was found, with higher proportions of primary grade 3 T1 tumors and tumors in situ than in the EORTC series. An additional limitation is that prior recurrence of the EORTC table was not included in our parameters. Consequently, two separate analyses were performed for recurrence. Conclusions The EORTC model successfully stratified recurrence and progression risks in our cohort. However, the discriminative ability of the EORTC tables decreased in our patients for progression. Moreover, these tables overestimated risks of recurrence and progression after BCG therapy.