Background: A clinical need exists for a biomarker test to accurately delineate aggressive prostate cancer (AgCaP), and thus better assist clinicians and patients decision-making on whether to proceed to prostate biopsy.
Objectives: To develop a blood test for AgCaP and compare to PSA, %free PSA, proPSA, and prostate health index (PHI) tests.
Design, settings and participants: Patient samples from the MiCheck-01 trial were used for development of the MiCheck test.
Methods: Serum analyte concentrations for cellular growth factors were determined using a custom-made Luminex-based R&D Systems multianalyte kit.
Outcome measurements and statistical analysis: Bayesian model averaging and random forest approaches were used to identify clinical factors and growth factors able to distinguish between men with AgCaP (Gleason Score [GS] ≥3+4) from those with non-AgCaP (GS 3+3). Logistic regression and Monte Carlo cross-validation identified variable combinations in order to able to maximize differentiation of AgCaP from non-AgCaP.
Results: The MiCheck logistic regression model was developed and comprises the following variables: serum prostate-specific antigen (PSA), patient age, Digital Rectal Exam (DRE) status, Leptin, IL-7, vascular endothelial growth factor, and Glypican-1. The model differentiated AgCaP from non-AgCaP with an area under the curve of 0.83 and was superior to PSA, %free PSA and PHI in all patient groups, regardless of PSA range. Applying the MiCheck test to all evaluable biopsy patients from the MiCheck-01 study demonstrated that up to 30% of biopsies could be avoided while delaying diagnosis of only 6.8% of GS ≥3+4 cancers, 5% of GS ≥4+3 cancers and no cancers of GS 8 or higher.
Conclusions: The MiCheck test outperforms PSA, %free PSA and PHI tests in differentiating AgCaP vs. non-AgCaP patients. The MiCheck test could result in a significant number of biopsies being avoided with a low number of patients experiencing a delayed diagnosis.
Keywords: Aggressive; Biomarker; Clinical study; Prostate.
Copyright © 2020 Elsevier Inc. All rights reserved.