Critical assessment of preoperative urinary prostate cancer antigen 3 on the accuracy of prostate cancer staging

Background

Knowledge about the staging significance of the prostate cancer antigen 3 (PCA3) score to better identify pathologic features after radical prostatectomy (RP) is limited and controversial.

Objective

Our aim was to study the clinical staging significance of PCA3 to identify pathologic favorable and/or unfavorable features in the RP specimen.

Design, setting, and participants

Complete retrospective clinical and pathologic data of consecutive men who had undergone RP from three tertiary referral centers including preoperative PCA3 scores (n = 305) and computer-assisted planimetrically measured tumor volume data (n = 160) were available.

Intervention

All patients were treated with RP.

Measurements

PCA3 scores were assessed using the PROGENSA assay (Gen-Probe, San Diego, CA, USA). Beyond standard risk factors (age, digital rectal examination, prostate-specific antigen, prostate volume, biopsy Gleason score, percentage of positive cores), five different PCA3 codings were used in logistic regression models to identify five distinct pathologic end points: (1) low-volume disease (<0.5 ml), (2) insignificant prostate cancer (PCa) according to the Epstein criteria, (3) extracapsular extension (ECE), (4) seminal vesicle invasion (SVI), and (5) aggressive disease defined as Gleason sum ≥7. Accuracy estimates of each end point were quantified using the area under the curve (AUC) of the receiver operator characteristic analysis in models with and without PCA3.

Results and limitations

PCA3 scores were significantly lower in low-volume disease and insignificant PCa (p ≤ 0.001). AUC of multivariable low-volume disease (+2.4 to +5.5%) and insignificant PCa models (+3 to +3.9%) increased when PCA3 was added to standard clinical risk factors. In contradistinction, regardless of its coding, PCA3 scores were not significantly elevated in pathologically confirmed ECE (p = 0.4) or SVI (p = 0.5), respectively. Higher PCA3 scores were associated with aggressive disease (p < 0.001). Importantly, the addition of PCA3 to multivariable intermediate- and high-grade models did not improve prediction. Despite reporting the largest pathologic PCA3 study, the main limitation resides in its small sample size.

Conclusions

PCA3 was confirmed as a valuable predictor of pathologically confirmed low-volume disease and insignificant PCa. Further exploration of its role as an additional marker to select patients for active surveillance may be warranted. In contradistinction, assessment of pathologically advanced or aggressive PCa is not improved using PCA3.