Development and internal validation of a nomogram predicting extracapsular extension in radical prostatectomy specimens

OBJECTIVES: To present a nomogram predicting the side-specific probability of extracapsular extension (ECE) in radical prostatectomy (RP) specimens. METHODS: Three hundred and fifty-four patients with T1c-T3a prostate cancer undergoing RP were included in the analysis. A receiver operating characteristic (ROC) analysis was carried out to evaluate the predictive values of each clinical and pathological factor, separately and in combination. Based on logistic regression analysis, a nomogram predicting the side-specific probability of ECE was developed. RESULTS: Overall, 146 (40%) of 354 patients and 165 (23%) of 708 lobes had ECE pathologically. The areas under the ROC curve (AUC) of the standard features, such as serum PSA, clinical stage and biopsy Gleason sum on each side, in predicting side-specific probability of ECE were 0.624, 0.627, and 0.747, respectively. When these three features were combined, AUC increased to 0.773 which was not significantly different from 0.791 of maximum percent of cancer alone (P = 0.613) and significantly enhanced by including maximum percent of cancer on each side, 0.799 (P = 0.022). The resulting nomogram was internally validated and had excellent calibration. CONCLUSIONS: The accuracy in predicting ECE is increased by combining standard clinical factors (clinical stage, serum PSA, highest Gleason score) and biopsy features, such as maximum percent of cancer in the cores. The developed nomogram is helpful when deciding whether or not neurovascular bundles can be preserved.