Dr Polascik on an AI-Driven Urine Volatile Compound Profiling Approach for Prostate Cancer Detection

“We were able to detect cancers in men who had MRI-invisible lesions. MRI invisibility is really the new frontier for prostate cancer detection and staging and risk stratification. We’re using mpMRI as a workhorse nowadays, but there are men who have MRI-invisible cancers, some [of which] could be aggressive. We need tests that can detect and risk stratify prostate cancer in the MRI-negative or -invisible cohort of men.”

Thomas J. Polascik, MD, the Lawrence C. Katz Distinguished Professor of Urologic Surgery in the Department of Urology at Duke Cancer Institute, discussed findings from a study of a novel AI-based urine volatile organic compound (VOC) assay for the detection of prostate cancer, which he presented during the 2026 American Urological Association Annual Meeting.

A key finding of the study was the ability of urinary VOC analysis to detect prostate cancer in men with MRI-invisible lesions or tumors that do not produce a visible signal on multiparametric MRI (mpMRI), Polascik began. Although mpMRI has become a central tool in contemporary prostate cancer detection and staging workflows, he noted that a meaningful subset of men harbor cancers that evade MRI detection, some of which may carry aggressive biological features. MRI invisibility has become a defining frontier in prostate cancer diagnostics, underscoring the urgent need for complementary biomarker strategies capable of identifying and risk-stratifying disease in this MRI-negative cohort, he explained.

Additionally, the test demonstrated strong performance independently of serum prostate-specific antigen (PSA) values, the current standard biomarker for prostate cancer detection, Polascik said. This independence is clinically significant, as PSA has well-recognized limitations including poor specificity and inability to distinguish indolent from clinically significant disease. The ability of urinary VOC analysis to detect cancer in men with very low PSA values suggests that the assay captures biological information orthogonal to PSA, potentially expanding its utility to patients who might otherwise be considered low-risk or who fall below conventional thresholds for biopsy referral.