Fig. 5

Overview of patient samples and study methods. A Fifty-two plasma samples were used for cfDNA isolation and DNA methylation profiling, 48 of which were further analysed. Raw data were pre-processed and were characterised for cellular origin. B The top 5% most variable probes (MVPs) within each subject over time were filtered for 5′ regulatory regions and used to analyse dynamic methylation patterns, generating congruent clusters. C The effects of drug therapy on cfDNA methylation were analysed by examining changes in methylation of ≥ 0.1 (between drug time-points). D Methylation differences between PC and NPC samples were analysed for the 4 subjects with any number of PC samples. Identified differentially methylated genes (DMGs) were validated using two independent PCa cohorts. BL baseline, cfDNA cell-free DNA, DMGs differentially methylated genes, FU follow-up, mCRPC metastatic castration resistant prostate cancer, mPCa metastatic prostate cancer, MVPs most variable probes, NPC non-prostate content, PC prostate content, SD standard variation