Introduction:
The 18-gene MyProstateScore 2.0 (MPS2) test was developed and validated for detection of Grade Group≥2 (GG≥2) cancer using post-digital rectal examination (DRE) urine specimens. On external validation, MPS2 was shown to reduce unnecessary biopsies by 35-51% while maintaining detection of 95% of GG≥2 cancers. However, DRE is uncomfortable for patients and is now considered optional by clinical guidelines. Additionally, performing DRE is not feasible in the increasing proportion of patients undergoing telehealth consultations. As such, to increase access to and feasibility of testing, we validated the MPS2 assay using first-catch, non-DRE urine in a cohort of patients undergoing prostate biopsy.
Methods:
Patients provided first-catch urine prior to biopsy. RNA extraction was performed from ≤5 mL of urine using a modified extraction method optimized for non-DRE urine (Norgen Biotek Corp). RNA was reverse transcribed to cDNA, amplified, and quantified. All samples were run in triplicate, and mean expression values were normalized to the housekeeping gene KLK3. MPS2 values were calculated using previously validated models differing only by inclusion of clinical factors: biomarkers alone (BA; no clinical data), biomarker expression and clinical factors (BA+CF; age, race, PSA, DRE findings, family history, and prior negative biopsy), and biomarker expression, clinical factors, and prostate volume (BA+CF+PV). The primary outcome was GG≥2 cancer on biopsy; GG≥3 was assessed secondarily. Overall discriminative accuracy was compared to serum PSA and the Prostate Cancer Prevention Trial risk calculator (PCPTrc). Based on a testing threshold providing ≥90% sensitivity for GG≥2 cancer, we calculated performance measures and consequences of MPS2 testing.
Results:
The cohort included 273 consecutive men with median PSA 6.7 ng/mL (IQR 5.0-9.5) presenting for diagnostic biopsy. Forty-seven men (17%) underwent pre-biopsy MRI, of which 25 (53%) had a PI-RADS 3-5 lesion and underwent targeted biopsy in addition to systematic biopsy. Overall, 108 men (40%) were found to have GG≥2 cancer, of which 84 had GG2 (30.8%), 12 had GG3 (4.4%) and 12 had GG4-5 disease (4.4%). The area under the curve for GG≥2 cancer was 58% for PSA, 63% for PCPTrc, and 71% for the MPS2 BA model, 74% for the BA+CF model, and 77% for the BA+CF+PV model (Figure). Negative predictive values (NPV) were 92%-96% for GG≥2 cancer and 96% for GG≥3 cancer. Clinical use of the MPS2 threshold 11.5 to select patients for biopsy would have avoided 38%-43% of unnecessary biopsies (i.e. 38%-43% specificity) while maintaining detection of 91%-94% of GG≥2 cancers (i.e. 91%-94% sensitivity) (Table).
Conclusion:
Using urine obtained without DRE, MPS2 testing provided very high sensitivity and NPV for ruling out GG≥2 cancer. Importantly, the baseline (i.e. biomarker-only) MPS2 model demonstrated clinically significant improvement in diagnostic accuracy relative to PSA-based testing, with further improvements observed with inclusion of optional clinical data. This non-DRE approach provides a convenient, highly accurate testing option to reduce the need for further evaluation with imaging or biopsy in men with elevated PSA.
Funding: This work was supported by the Michigan-Vanderbilt EDRN Biomarker Characterization Center (U2C CA271854) and Prostate Cancer Foundation Young Investigator Awards (Tosoian and Xiao). Other sources of funding not directly involved in design of the study included the Michigan Prostate SPORE (P50 CA186786), NCI Outstanding Investigator Award (Chinnaiyan, R35 CA231996), Howard Hughes Medical Institute (Chinnaiyan), and the American Cancer Society (Chinnaiyan).
Image(s) (click to enlarge):
CLINICAL VALIDATION OF MYPROSTATESCORE 2.0 (MPS2) TESTING USING FIRST-CATCH, NON-DRE URINE
Category
Prostate Cancer > Locally Advanced
Description
Poster #11
Presented By: Keavash Assani
Authors:
Keavash Assani
Yuping Zhang
Jacob I. Meyers
Spencer Heaton
Javed Siddiqui
Lanbo Xiao
Daniel A. Barocas
Janene M. Pierce
Ashley E. Ross
Zoey Chopra
E. Nayiri Ayanian
Peter D. Sienko
Udit Singhal
Simpa S. Salami
Todd M. Morgan
Ganesh S. Palapattu
John T. Wei
Arul M. Chinnaiyan
Jeffrey J. Tosoian