Developing Prostac, a PVT1-based composite, to score improving prostate cancer diagnostic precision for diverse ancestries.
This describe the development and initial validation of Prostac, a novel composite score designed to improve prostate cancer (PCa) diagnosis by integrating copy number measurements of three PVT1-derived transcripts. The work responds to well-documented limitations of the prostate specific antigen (PSA) assay its high false positive rate, low specificity, and low sensitivity, which lead to unnecessary prostate biopsies and attendant complications. Recognizing the urgent need for better, preferably noninvasive biomarkers that can distinguish indolent from aggressive disease, the authors build on previous findings that specific PVT1 exons (4A, 4B, and 9) are significantly overexpressed in prostate cancer epithelial cells and particularly in men of African ancestry (moAA). PVT1, a long non-coding RNA on chromosome 8q24, has been implicated across multiple cancer types, and three of its exons were previously shown to be promising diagnostic markers in prostate cancer datasets, including evidence that exon 9 is associated with aggressive disease in moAA.
To translate these observations into a clinically useful tool, the team developed a noninvasive copy number-based quantification assay using real-time quantitative polymerase chain reaction (qPCR) to measure PVT1 exons 4A, 4B, and 9 in tissues and biofluids. They treated the copy number outputs as features and applied supervised machine learning, specifically support vector machines (SVM), to identify a classification hyperplane separating benign from malignant prostate epithelial cells. When trained on data from nonmalignant prostate epithelial cells (RWPE1) and prostate cancer cells (MDA PCa 2b), the Prostac composite score, derived from the SVM hyperplane, achieved perfect classification of the training samples in cross-validation, indicating that the three PVT1-derived markers are linearly separable and robustly overexpressed in cancer cells in this dataset.
The creation of Prostac provides a single numeric score that aggregates information from multiple PVT1 transcripts, making interpretation more practical for clinicians than evaluating multiple individual biomarkers. By combining molecular signals into one composite measure, Prostac aims to improve diagnostic accuracy, reduce unnecessary biopsies and facilitate earlier, more appropriate treatment decisions. Importantly, because the underlying biomarkers were identified and validated in samples that include men of African ancestry, who bear a disproportionate burden of prostate cancer incidence and mortality, the score has particular relevance for addressing disparities in diagnosis and outcomes.
The authors position Prostac as a groundwork-laying development: its promising in vitro classification performance motivates further clinical evaluation. The next critical steps implied by the study are clinical trials and broader validation in patient-derived tissue, serum, and urine samples to determine diagnostic sensitivity, specificity and real-world utility across diverse populations. The collaborators propose Prostac as a measurable, machine learning-derived composite biomarker score based on PVT1 exon copy numbers that could potentially enhance prostate cancer diagnosis, particularly among populations disproportionately affected by aggressive disease.
