Telomere-based risk models for the early diagnosis of clinically significant prostate cancer.

Abstract

The objective of this study was to explore telomere-associated variables (TAV) as complementary biomarkers in the early diagnosis of prostate cancer (PCa), analyzing their application in risk models for significant PCa (Gleason score > 6). As part of a larger prospective longitudinal study of patients with suspicion of PCa undergoing prostate biopsy according to clinical practice, a subgroup of patients (n = 401) with PSA 3-10 ng/ml and no prior biopsies was used to evaluate the contribution of TAV to discern non-significant PCa from significant PCa. The cohort was randomly split for training (2/3) and validation (1/3) of the models. High-throughput quantitative fluorescence in-situ hybridization was used to evaluate TAV in peripheral blood mononucleated cells. Models were generated following principal component analysis and random forest and their utility as risk predictors was evaluated by analyzing their predictive capacity and accuracy, summarized by ROC curves, and their clinical benefit with decision curves analysis. The median age of the patients was 63 years, with a median PSA of 5 ng/ml and a percentage of PCa diagnosis of 40.6% and significant PCa of 19.2%. Two TAV-based risk models were selected (TAV models 1 and 2) with an AUC ≥ 0.83 in the full study cohort, and AUC > 0.76 in the internal validation cohort. Both models showed an improvement in decision capacity when compared to the application of the PCPT-RC in the low-risk probabilities range. In the validation cohort, with TAV models 1 and 2, 33% /48% of biopsies would have been avoided losing 0/10.3% of significant PCa, respectively. The models were also tested and validated on an independent, retrospective, non contemporary cohort. Telomere analysis through TAV should be considered as a new risk-score biomarker with potential to increase the prediction capacity of significant PCa in patients with PSA between 3-10 ng/ml.