A post-radical-prostatectomy nomogram incorporating new pathological variables and interaction terms for improved prognosis.

OBJECTIVE

To evaluate new variables in prostate pathology reporting including, the per cent of Gleason patterns 4 and/or 5 (% 4/5), presence or absence of intraductal carcinoma of the prostate (IDCP), tumour volume and the prostatic zone of tumour origin as predictors of post-radical-prostatectomy (RP) biochemical recurrence (BCR). To develop an optimal postoperative nomogram for patients with prostate cancer.

PATIENTS AND METHODS

Our study cohort was 1939 eligible patients from the Abbott West Australian Prostatectomy Database. Multivariate Cox proportional hazard regression models were developed to predict BCR which was defined as prostate-specific antigen (PSA) ≥ 0.2 ng/mL. Our models and the 2009 Kattan postoperative nomogram were compared in terms of discrimination and calibration, with internal validation of our final model performed using bootstrapping methods. Our final model is presented as a nomogram.

RESULTS

The Kattan nomogram was accurate in discriminating our patients according to risk (concordance index: 0.791) but calibration analysis indicated underestimation of patient risk, particularly for high-risk disease. Our nomogram incorporates % 4/5, IDCP and prostate weight plus interaction terms between % 4/5, positive surgical margins and extracapsular extension, giving improved predictive accuracy (concordance index: 0.828) and calibration.

CONCLUSIONS

Nomograms that predict risk of BCR defined as PSA ≥ 0.4 ng/mL may not be optimal for patient cohorts where BCR is defined as PSA ≥ 0.2 ng/mL. If our findings are validated in other populations, current post-RP nomograms may be improved to a modest degree by incorporating the new variables prostate weight, IDCP and % 4/5, and by considering interactions between predictive variables.