Reply to: Biochemical failure after radical prostatectomy in intermediate-risk group men increases with the number of risk factors.

Dear Editor,

With great interest, I read the recent work by Furubayashi et al. published in the Indian Journal of Urology.[1] The great heterogeneity in oncological outcomes among D'Amico risk groups, particularly in the “intermediate-risk group,” let the authors to further address this issue. Therefore, “intermediate”-risk patients were substratified by the amount of single risk factors exhibited, and the impact of risk factor combinations studied with regard to biochemical recurrence following radical prostatectomy. In part, the authors relied on a multivariate analysis using a combination of pre- and post-operative risk factors to predict the risk of prostate-specific antigen (PSA) failure. Not surprisingly, it was revealed that within the intermediate-risk category, an increasing number of risk factors are significantly associated with the PSA failure-free survival at 5 years of follow-up.

This timely study adds to a growing body of evidence that addresses heterogeneity in outcomes within D'Amico risk strata. Indeed, a recent study used a combined number of risk factors to create five risk categories among 10,139 men with nonmetastatic prostate cancer and showed that that the new classification outperformed not only intermediate-risk category but also the high-risk category.[2] Before that, a similar analysis of 12,821 patients at the Johns Hopkins University by Reese et al. showed significant heterogeneity among the intermediate-risk category.[3] To overcome heterogeneity in biochemical recurrence rates within Gleason score-based risk groups, Epstein et al. have recently proposed a simplified five-grade grading system which has a higher discrimination score than traditional Gleason score.[4]

While the present study confirms these findings, two aspects deserve further consideration. First, overfitting of data might have compromised the accuracy of results. A common rule of thumb suggests 10–20 events per predictor variable in Cox and logistic regression.[5] However, as nine predictor variables were fitted on a very limited number of outcomes (patients having PSA failure in this study was n = 26) in the multivariate Cox regression model, it would be inappropriate to draw conclusions based on these data. Second, the authors conclude that the number of intermediate-risk factors (including stage, Gleason score, and PSA) is predictive of biochemical recurrence. These findings are in contradiction to their multivariate analysis where PSA and Gleason score do not independently predict biochemical recurrence.

Notwithstanding, the study is a step in the right direction. Furthermore, the availability of multiparametric magnetic resonance (MR) imaging and MR-guided biopsy will allow us to advance from the Digital Rectal Examination (DRE)-based D'Amico classification and devise new and more precise prognostic criteria for prostate cancer.