Background and purpose: Patients with Gleason (GL) 6 prostate cancer in one or two biopsy cores can be upgraded and/or upstaged at the time of surgery, which may adversely impact long-term outcome. A novel model for prediction of adverse pathologic outcomes was developed using preoperative characteristics.
Patients and methods: Between 2003 and 2007, 1159 patients underwent robot-assisted radical prostatectomy (RARP) at our institution. GL 6 prostate cancer in one or two biopsy cores was identified in 416 (36%) patients. Logistic regression analyses were used to assess the rate of GL ≥7 and/or extraprostatic extension at RARP. Covariates consisted of age, body mass index (BMI), number of positive cores, greatest percent of cancer in a core (GPC), clinical stage, and preoperative prostate-specific antigen (PSA) level. After backward variable selection, the developed model was internally validated using the area under the curve and subjected to methods of calibration.
Results: Respectively, 278 (67%) and 138 (33%) patients had one or two positive biopsy cores. At RARP, 90 (22%) patients were upgraded to GL ≥7 and 37 (9%) had extraprostatic extension. The novel model relied on age, BMI, preoperative PSA level, and GPC for prediction of adverse pathologic outcomes and was 69% accurate. Calibration plot revealed a virtually perfect relationship between predicted and observed probabilities.
Conclusions: In patients with GL 6 prostate cancer in one or two biopsy cores, 25% have more ominous pathology at RARP. The model provides an individual assessment of adverse outcomes at surgery. Consequently, it may be considered when counseling patients regarding their management options.