How often do available preoperative risk factors accurately predict the risk assessed after surgery for localized...

To describe how frequently new information obtained at surgery translates into a substantial change in the risk of recurrence for patients with localized prostate cancer, and to determine what factors contribute to this increase in risk, as the preferred therapy for prostate cancer is often chosen based on available preoperative variables and therefore appropriate decision-making requires an accurate preoperative assessment.

Using the Columbia Comprehensive Clinical Database of Urologic Oncology, we retrospectively analysed 3460 men who had radical prostatectomy (RP) for prostate cancer from 1988 to 2006. Kattan nomograms were used to calculate the 5-year progression-free probabilities before and after RP. The difference between these nomogram scores was used to divide patients into three groups, those with a decrease in the probability of disease-free survival (DFS) of > or =15%, those with an increase in the probability of DFS of > or =15%, and those with an absolute change of <15%.

In all, 1804 men with complete data before and after RP were analysed; 1220 (68.4%) had no significant change in nomogram score, 238 (13.3%) had a significant increase and 327 (18.3%) had a significant decrease in the probability of recurrence. Those patients with an increased probability of recurrence had a greater proportion of patients with pathological Gleason sum of > or =8, higher rates of extraprostatic capsular invasion, positive margins, seminal vesical invasion and lymph node involvement (all P < 0.001).

Accurate risk predictions both before and after RP are central to effective patient counselling and optimal management. Notably, 13.3% of the present patients were faced with a substantial increase of > or =15% in their risk of biochemical failure after pathological variables became available.