Preoperative prostate health index is an independent predictor of early biochemical recurrence after radical prostatectomy: Results from a prospective single-center study

Background: The aim of this study was to test the hypothesis that preoperative prostate health index (PHI) levels could help to predict early biochemical recurrence (BCR) in a contemporary population of patients with prostate cancer treated with robot-assisted radical prostatectomy (RARP).

Methods: The study population consisted of 313 patients treated with RARP for clinically localized prostate cancer at a single institution between 2010 and 2011. Patients subjected to neoadjuvant or adjuvant therapies and patients with a follow-up of<2 years were excluded. BCR was defined as a postoperative level of total prostate-specific antigen ≥0.2 ng/ml and elevating after RARP. The minimum P-value method was used to determine the most significant PHI cutoff value to discriminate between patients with and without BCR. The Kaplan-Meier method was used to determine BCR-free survival rates. Finally, Cox regression models were fitted to determine the predictors of BCR, and the predictive accuracy (area under the curve) of each predictor was determined with the Harrell concordance index.

Results: Mean total prostate-specific antigen and mean PHI levels were 5.76 ng/ml (interquartile range: 4.2-8.7) and 46.0 (35-62), respectively. Biopsy Gleason score was 6 in 173 (55.3%), 7 in 121 (38.7%), and ≥8 in 19 (6.1%) patients. At final pathology, extracapsular extension was observed in 59 (18.8%), seminal vesicle invasion in 24 (7.7%), and lymph node invasion in 11 (3.5%) patients, whereas 228 (72.8%) patients had organ-confined disease. The 2-year BCR-free survival rate was 92.5% in the overall population and was 96.7% in patients with organ-confined disease. The most significant PHI cutoff value to discriminate between patients with and without BCR was 82. Specifically, the 2-year BCR-free survival rate was 97.7% in patients with a preoperative PHI level<82 relative to 69.7% in patients with a PHI level ≥82 (log-rank test: P<0.001). Finally, in multivariable Cox regression analyses, PHI level emerged as an independent predictor of BCR in both the preoperative and the postoperative settings and was more accurate than several established BCR predictors were.

Conclusions: Preoperative PHI levels may discriminate between patients who are at a high risk vs. low risk of BCR after RARP. External validation of our findings within a larger population with a longer follow-up time is needed.