A Four-kallikrein Panel Predicts High-grade Cancer on Biopsy: Independent Validation in a Community Cohort

Abstract

Background: A statistical model based on four kallikrein markers (total prostate-specific antigen [tPSA], free PSA [fPSA], intact PSA, and human kallikrein-related peptidase 2) in blood can predict risk of Gleason score ≥7 (high-grade) cancer at prostate biopsy.

Objective: To determine the value of this model in predicting high-grade cancer at biopsy in a community-based setting in which referral criteria included percentage of fPSA to tPSA (%fPSA).

Design, setting, and participants: We evaluated the model, with or without adding blood levels of microseminoprotein-β (MSMB) in a cohort of 749 men referred for prostate biopsy due to elevated PSA (≥3 ng/ml), low %fPSA (<20%), or suspicious digital rectal examination at Skåne University Hospital, Malmö, Sweden.

Outcome measurements and statistical analysis: The kallikrein markers, with or without MSMB levels, measured in cryopreserved anticoagulated blood were combined with age in a published statistical model (Prostate Testing for Cancer and Treatment [ProtecT]) to predict high-grade cancer at biopsy. Predictive accuracy was compared with a base model.

Results and limitations: The %fPSA was low (median: 17; interquartile range: 13-22) in this cohort because this marker was used as a referral criterion. The ProtecT model improved discrimination over age and PSA for high-grade cancer (0.777 vs 0.720; p=0.002). At one illustrative cut point, use of the panel would reduce the number of biopsies by 236 per 1000 and detect 195 of 208 (94%) but delay diagnosis of 13 of 208 high-grade cancers. MSMB levels in blood did not improve the accuracy of the panel (p=0.2).

Conclusions: The kallikrein model is predictive of high-grade cancer if criteria for biopsy referral also include %fPSA, and it can reduce unnecessary biopsies without missing an undue number of tumors.

Patient summary: We evaluated a published model to predict biopsy outcome in men biopsied due to low percentage of free to total prostate-specific antigen. The model helps reduce unnecessary biopsies without missing an undue number of high-grade cancers.

Keywords: Biopsy; Four-kallikrein panel; Prostate cancer; Prostate-specific antigen.

Fig. 1

Calibration plot for the Prostate Testing for Cancer and Treatment model predicts (a) high-grade cancer and (b) any-grade cancer using age, total prostate-specific antigen (PSA), free PSA, intact PSA, and human kallikrein-related peptidase 2 (dashed lines). PSA = prostate-specific antigen.

Fig. 2

Decision curve analysis for the Prostate Testing for Cancer and Treatment model predicts (a) high-grade cancer and (b) any-grade cancer. The dashed gray line is the net benefit of the laboratory model (age and the kallikrein [KLK] panel); the dashed black line is the net benefit of the clinical model (age, digital rectal examination, and KLK panel). As a comparison, the net benefits from a biopsy-all strategy (solid black line) and a biopsy-none strategy (solid gray line) are shown. DRE = digital rectal examination; KLK = kallikrein.

Fig. 3

Clinical implications of using a model developed to predict the risk of Gleason score ≥7 (high-grade) prostate cancer based on four kallikrein markers. The graph illustrates the results of using the model per 1000 men eligible for biopsy, with the x-axis denoting the threshold risk of high-grade cancer and the y-axis indicating the number of men biopsied (black line) or detected with evidence of high-grade cancer (green line).