Do prostate cancer nomograms give accurate information when applied to European patients?

OBJECTIVE

The aim of this study was to validate and compare the performance of preoperative risk assessment tools in a population of men treated with radical prostatectomy at a single European institution.

MATERIAL AND METHODS

Patients were identified from databases of radical prostatectomy between 1996 and 2011 from a single UK centre. Information was obtained on demographics, prostate-specific antigen, staging, biopsy and specimen histopathology, and follow-up. Data were inputted into the Memorial Sloan-Kettering Cancer Center (MSKCC), Partin 1997 and Makarov/Partin 2007 nomograms, and the University of California San Francisco-Cancer of the Prostate Risk Assessment tool (UCSF-CAPRA). The risks of extracapsular extension (ECE), seminal vesicle invasion (SVI) and lymph-node involvement (LNI) were calculated and compared with known outcomes. Nomogram performance was measured using Hosmer-Lemeshow (HL) goodness-of-fit tests, calculating concordance indices (c-indices) and calibration curves.

RESULTS

Data were obtained for 541 patients. Prediction of ECE was relatively poor using all nomograms, with the Makarov/Partin 2007 the most accurate at prediction over the range of risk stratification (HL 9.9, c-index 0.62). Predictions of SVI and LNI were better than for ECE, with the MSKCC nomogram performing best for SVI (HL 10.9, c-index 0.73) and all nomograms performing well for LNI prediction (c-indices 0.8 to 0.815). CAPRA predicted best for SVI (OR 1.49, 95% confidence interval 1.27-1.74).

CONCLUSIONS

To the authors' knowledge, this is the first head-to-head comparison of the accuracy of these commonly used risk calculators in a North European population. Caution should be used when counselling patients using nomograms. Although nomograms may be used as a guide, patients should be warned that they often have not been validated on different European populations and may give misleading information regarding a patient's specific risks.