Use of the MyProstateScore Test to Rule Out Clinically Significant Cancer: Validation of a Straightforward Clinical Testing Approach

Abstract

Purpose: The MyProstateScore test was validated for improved detection of clinically significant (grade group ≥2) prostate cancer relative to prostate specific antigen based risk calculators. We sought to validate an optimal MyProstateScore threshold for clinical use in ruling out grade group ≥2 cancer in men referred for biopsy.

Materials and methods: Biopsy naïve men provided post-digital rectal examination urine prior to biopsy. MyProstateScore was calculated using the validated, locked multivariable model including only serum prostate specific antigen, urinary prostate cancer antigen 3 and urinary TMPRSS2:ERG. The MyProstateScore threshold approximating 95% sensitivity for grade group ≥2 cancer was identified in a training cohort, and performance was measured in 2 external validation cohorts. We assessed the 1) overall biopsy referral population and 2) population meeting guideline based testing criteria (ie, prostate specific antigen 3-10, or <3 with suspicious digital rectal examination).

Results: Validation cohorts were prospectively enrolled from academic (977 patients, median prostate specific antigen 4.5, IQR 3.1-6.0) and community (548, median prostate specific antigen 4.9, IQR 3.7-6.8) settings. In the overall validation population (1,525 patients), 338 men (22%) had grade group ≥2 cancer on biopsy. The MyProstateScore threshold of 10 provided 97% sensitivity and 98% negative predictive value for grade group ≥2 cancer. MyProstateScore testing would have prevented 387 unnecessary biopsies (33%), while missing only 10 grade group ≥2 cancers (3.0%). In 1,242 patients meeting guideline based criteria, MyProstateScore ≤10 provided 96% sensitivity and 97% negative predictive value, and would have prevented 32% of unnecessary biopsies, missing 3.7% of grade group ≥2 cancers.

Conclusions: In a large, clinically pertinent biopsy referral population, MyProstateScore ≤10 provided exceptional sensitivity and negative predictive value for ruling out grade group ≥2 cancer. This straightforward secondary testing approach would reduce the use of more costly and invasive procedures after screening with prostate specific antigen.

Keywords: biomarkers; biopsy; early detection of cancer; prostate-specific antigen; prostatic neoplasms; tumor.

Figure 1.

Flow diagram of patients providing urine specimens prior to initial prostate biopsy in the training cohort (blue) and validation cohorts (green). MyProstateScore was assessed using the validated, locked model including serum PSA, urinary PCA3 score, and urinary T2:ERG score. The number of patients meeting National Comprehensive Cancer Network (NCCN) indications for biopsy (i.e. PSA 3–10 ng/ml or PSA <3 ng/ml and suspicious DRE) are listed for the validation cohorts (i.e. guideline-directed). EDRN: Early Detection Research Network; PSA: prostate-specific antigen; US: United States.

Figure 2.

Density plots illustrating serum PSA (upper panel) and MPS (lower panel) in patients with negative (−) or Grade Group 1 (GG1) cancer on biopsy (blue) versus GG≥2 cancer on biopsy (yellow). Data are shown for A) the overall biopsy referral population and B) the guideline-directed population (i.e. PSA 3–10 ng/ml or PSA <3 ng/ml and suspicious DRE). The dashed vertical line represents a MPS value of 10.