Figure 5.

Exploration of the standardised net benefit (sNB) by decision curve analysis (DCA) for adopting risk models to aid the decision to undertake an initial biopsy for patients presenting with a serum PSA ≥ 4 ng/mL, where current clinical practice is to biopsy all patients. The accepted patient/clinician risk threshold for accepting biopsy is detailed on the x-axis. Different biopsy outcomes are shown in each of the three panels; (A) detection of Gleason ≥ 4 + 3, (B) detection of Gleason ≥ 3 + 4, (C) any cancer; Blue—biopsy all patients with a PSA >4 ng/mL, Orange—biopsy patients according to the SoC model, Green—biopsy patients based on the Engrailed model, Purple—biopsy patients based on the exoRNA model, Red—biopsy patients based on a the ExoGrail model. To assess the benefit of adopting these risk models in a clinically relevant population, we used data available from the control arm of the The Cluster Randomized Trial of PSA Testing for Prostate Cancer (CAP) study [42] for proportionally resampling the ExoGrail cohort. DCA curves were calculated from 1000 bootstrap resamples of the available data to match the distribution of disease reported in the CAP trial population. Mean sNB from these resampled DCA results are plotted here. See Methods for full details.