Prognostic accuracy of Prostate Health Index and urinary Prostate Cancer Antigen 3 in predicting pathologic features after radical prostatectomy

Abstract

Objective

To compare the prognostic accuracy of Prostate Health Index (PHI) and Prostate Cancer Antigen 3 in predicting pathologic features in a cohort of patients who underwent radical prostatectomy (RP) for prostate cancer (PCa).

Methods and materials

We evaluated 156 patients with biopsy-proven, clinically localized PCa who underwent RP between January 2013 and December 2013 at 2 tertiary care institutions. Blood and urinary specimens were collected before initial prostate biopsy for [-2] pro–prostate-specific antigen (PSA), its derivates, and PCA3 measurements. Univariate and multivariate logistic regression analyses were carried out to determine the variables that were potentially predictive of tumor volume >0.5 ml, pathologic Gleason sum≥7, pathologically confirmed significant PCa, extracapsular extension, and seminal vesicles invasions.

Results

On multivariate analyses and after bootstrapping with 1,000 resampled data, the inclusion of PHI significantly increased the accuracy of a baseline multivariate model, which included patient age, total PSA, free PSA, rate of positive cores, clinical stage, prostate volume, body mass index, and biopsy Gleason score (GS), in predicting the study outcomes. Particularly, to predict tumor volume>0.5, the addition of PHI to the baseline model significantly increased predictive accuracy by 7.9% (area under the receiver operating characteristics curve [AUC] = 89.3 vs. 97.2, P>0.05), whereas PCA3 did not lead to a significant increase.

Although both PHI and PCA3 significantly improved predictive accuracy to predict extracapsular extension compared with the baseline model, achieving independent predictor status (all P׳s<0.01), only PHI led to a significant improvement in the prediction of seminal vesicles invasions (AUC = 92.2, P<0.05 with a gain of 3.6%).

In the subset of patients with GS≤6, PHI significantly improved predictive accuracy by 7.6% compared with the baseline model (AUC = 89.7 vs. 97.3) to predict pathologically confirmed significant PCa and by 5.9% compared with the baseline model (AUC = 83.1 vs. 89.0) to predict pathologic GS≥7. For these outcomes, PCA3 did not add incremental predictive value.

Conclusions

In a cohort of patients who underwent RP, PHI is significantly better than PCA3 in the ability to predict the presence of both more aggressive and extended PCa.

Introduction

Most recent data from European Study of Screening for Prostate Cancer reported a 21% relative reduction in the risk of death due to prostate cancer (PCa) at 13 years of follow-up, with a 27% reduction after adjustment for nonparticipation [1] and the Goteborg study, one of the European Study of Screening for Prostate Cancer centers, showed a 44% relative reduction at 14 years of follow-up [2]; however, currently, population screening for PCa remains controversial. The most important reason for controversy is the high percentage of overdiagnosis, calculated as ranging from 1.7% to 67% according to the different designs of the studies (epidemiological, clinical, and autopsy studies) and the consequent overtreatment [3]. However, in the current clinical practice, the increasing use of prostate-specific antigen (PSA) for the detection of PCa, in an “opportunistic” screening scenario, has already led to an important increase in incidence of diagnosed low-risk PCa that may not clinically progress during lifetime [4]. The preoperative tools currently used in this clinical setting, such as PSA, digital rectal examination, and biopsy results fail to accurately predict PCa aggressiveness and distinguish between insignificant PCa, eligible for protocol of active surveillance (AS) or focal therapy, and clinically significant PCa, eligible for radical prostatectomy (RP) or radiation therapy.

Consequently, numerous predictive and prognostic tools have been recently introduced to assist the physicians in the clinical decision-making process. However, these available models are far from perfect in their predictive ability and new biomarkers are required to correctly stratify patient risk before treatment.

In this context, several studies have analyzed the capability of prostate cancer antigen 3 (PCA3) [5], [6], [7], [8], [9], [10], [11] and [-2] proPSA (p2PSA) and its derivative, %[-2] proPSA (%p2PSA), and Prostate Health Index (PHI) [12], [13], [14], [15] in predicting PCa characteristics at final pathology in different and separate study cohorts.

Currently, no evidence is available on the prognostic and pathologic comparison of PCA3 and PHI in a same study cohort at the time of RP.

The aim of this study is to compare the prognostic accuracy of PCA3 and PHI in predicting pathologic features in a cohort of patients who underwent RP for clinically localized PCa.