External validation of the Briganti nomogram to estimate the probability of specimen-confined disease in patients with high-risk prostate cancer

Objective: To establish an external validation of the updated nomogram from Briganti et al., which provides estimates of the probability of specimen-confined disease using the variables age, prostate-specific antigen (PSA), clinical stage and biopsy Gleason score in preoperatively defined high-risk prostate cancer (PCa).

Patients and methods: The study included 523 patients with high-risk PCa, as defined by d'Amico classification, undergoing radical prostatectomy (RP) and bilateral lymph node dissection in one of two academic centres between 1990 and 2013. Specimen-confined disease was defined as pT2-pT3a node-negative PCa with negative surgical margins. The receiver-operator characteristic (ROC) curve was obtained to quantify the overall accuracy (area under the curve [AUC]) of the model in predicting specimen-confined disease. A calibration curve was then constructed to illustrate the relationship between the risk estimates obtained by the model (x-axis) and the observed proportion of specimen-confined disease (y-axis). The Kaplan-Meier method was used to assess biochemical recurrence (BCR)-free survival.

Results: Patients' median age and PSA level were 64 years and 21 ng/mL, respectively. The definition of high-risk PCa was based on PSA level only in 38.3%, a biopsy Gleason score >7 in 34.5%, a clinical stage >T2b in 6.9%, or a combination of these two or three factors in 20.3% of patients. Positive surgical margins were observed in 43.6%, with a rate of 14.8% in pT2 cancers and lymph node metastasis in 12.1% of patients. pT stage was pT0 in 0.9%, pT2 in 28.9%, pT3a in 37.5% and pT3b-4 in 32.7% of patients. Overall, 44.4% of patients (N = 232) had specimen-confined disease. PSA and cT stage were independently predictive of specimen-confined disease. The median (range) 2-, 5-, and 8-year BCR-free survival rates were significantly higher in specimen-confined disease as compared with non-specimen-confined disease: 80.87 (73.67-86.29) vs 37.55 (30.64-44.44)%, 63.53 (52.37-72.74) vs 26.93 (19.97-34.36)% and 55.08 (41.49-66.74) vs 19.52 (12.50-27.70)%, respectively (P < 0.001). The ROC curve analysis showed relevant accuracy of the model (AUC 0.6470, 95% CI 0.60-0.69) although the calibration plot suggested that, for risks ranging from 0.3 to 0.5, the odds of extracapsular extension were underestimated.

Conclusions: This external validation of the Briganti nomogram shows relevant accuracy, although the relative imprecision for intermediate risk may limit its clinical relevance. Our follow-up findings confirm the large proportion of specimen-confined PCa with good oncological outcomes in this heterogeneous subgroup of patients with high-risk PCa.