RESUMEN
OBJECTIVE: To determine whether subclassification of positive surgical margins (PSMs) increases predictive ability for biochemical recurrence (BCR) and aids clinical decision-making in patients undergoing radical prostatectomy. PATIENTS AND METHODS: We studied 2147 patients with pT2 and pT3a prostate cancer with detailed surgical margin parameters and BCR status. We compared a base model, a linear predictor calculated from the Memorial Sloan Kettering Cancer Center postoperative nomogram (prostate-specific antigen, pathological tumour grade and stage), with the addition of surgical margin status to five additional models (base model plus surgical margin subclassifications) to evaluate enhancement in predictive accuracy. Decision curve analysis (DCA) was performed to determine the clinical utility of parameters that enhanced predictive accuracy. RESULTS: Among 2147 men, 205 had PSMs, and 231 developed BCR. Discrimination for the base model with addition of surgical margin status was high (c-index = 0.801) and not meaningfully improved by adding surgical margin subclassification in the full cohort. In analyses considering only men with PSMs (N = 55 with BCR), adding surgical margin subclassification to the base model increased discrimination for total length of all PSMs - alone or with maximum Gleason grade at the margin (c-index improvement = 0.717 to 0.752 and 0.753, respectively). DCA demonstrated a modest benefit to clinical utility with the addition of these parameters. CONCLUSIONS: Specific subclassification parameters add predictive accuracy for BCR and may aid clinical utility in decision-making for patients with PSMs. These findings may be useful for patient counselling and future adjuvant therapy trial design.