Introduction:

While PSA screening has resulted in decreasing prostate cancer mortality, PSA is also known to be a poorly specific test for the detection of clinically significant cancer and can lead to many unnecessary biopsies, over-detection of low risk prostate cancer or even miss aggressive disease. PSA’s lack of robustness as a screening biomarker has led to the development of new prostate cancer screening tools (urine, blood and imaging based) to add specificity to PSA for the detection of clinically significant prostate cancer. Here we aim to evaluate and compare three screening tools in the clinical setting: serum based 4K score and urine based SelectMDX and ExoDx and their added value when combined with multiparametric parametric MRI (mpMRI).

Methods:

Patients referred to UCSF from 2016 to 2019 for consideration of biopsy were enrolled. Patients with PSA >20 ng/mL or with a prior positive prostate biopsy were excluded. 510 patients met inclusion criteria. All patients underwent DRE followed by urine collection for SelectMDx testing or underwent urine collection for ExoDx without pretesting DRE or had the serum based 4K score. SelectMDxscores were binary: low or high risk for prostate cancer. ExoDx was considered high prostate cancer risk if the IntelliScore was >15.6, and 4K scores >10% were considered high risk. 266 patients had a prostate biopsy. All biopsies were systematic 14 cores with or without targeted ultrasound or mpMRI fusion cores. We defined the sensitivity (SE), specificity (SP), negative and positive predictive values (NPV and PPV) of each biomarker, mpMRI, TRUS and PSAD for the detection of all prostate cancers and specifically high-grade prostate cancers (here defined as Gleason grade of 3+4 or higher). Multivariate logistic regression modeling adjusting for age, PSA, prostate volume, DRE, family history of prostate cancer and history of prior negative biopsy was performed to determine each biomarker’s ability to predict the presence of any cancer on biopsy or high-grade prostate cancer, with or without imaging. Statistical analyses were performed using SAS.

Results:

Mean age was 66.2 years, 71% were Caucasian, 5% African American and 85% of had no family history of prostate cancer. 76% were biopsy naïve and median PSA was 6.4 ng/mL (IQR 4.7-8.6). 151 had the SelectMDx test, 196 ExoDx test and 209 4K score test. SE, SP, PPV and NPV of each biomarker, mpMRI, TRUS and PSAD for the detection of all prostate cancers and high-grade prostate cancers are listed in Table 1. ExoDx had the highest sensitivity overall, but the lowest specificity. SelectMDx had the lowest sensitivity but higher specificity than the other biomarkers for detecting clinically significant cancer. mpMRIs with a PIRADs 4/5 or a positive lesion on TRUS had the highest specificity but sensitivity was considerably lower than 4K or ExoDx sensitivities (Table 1). Multivariate logistic regression modeling showed that a positive result in any one of the biomarkers or imaging (mpMRI PIRAD 4/5 or positive lesion on TRUS) were strongly associated with finding high-grade disease. Furthermore, combining mpMRI with a biomarker increased predictive power for detecting clinically significant cancer as shown in Table 2.  ExoDx had the highest AUC for performance in predicting high-grade disease of all the biomarkers (AUC 0.849 for ExoDx versus 0.835 for 4K and 0.699 for SelectMDx). The addition of PIRADS of 4 or 5 on mpMRI read increased the ExoDx AUC to 0.877 and the 4K AUC to 0.906 and the SelectMDx AUC increased to 0.805 (Table 2).

Conclusion:

4K score >10% or ExoDx >15.6 or PIRAD 4 or 5 on mpMRI all outperformed PSAD, TRUS and SelectMDX for the prediction of high-grade prostate cancer. High 4K scores and high ExoDx scores had high sensitivities while a PIRADS score 4/5 on mpMRI or positive lesion on TRUS had better specificity for high-grade prostate cancer. Combining the biomarkers with mpMRI resulted in the best predictive ability for detecting clinically significant cancer.

Funding: N/A