PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 4, 2026Journal of Clinical Oncology0 citations

Plasma cell-free DNA methylation as a predictive biomarker for hormonal therapy response in advanced prostate cancer.

View Full Paper
KMKyle S.Y. MiyazakiMHMaha H. A. HussainTWTanye Wen

Key Points

  • This research aims to evaluate cell-free DNA methylation as a predictive biomarker for hormonal therapy response in advanced prostate cancer.
  • Conducted a prospective study with 48 men with advanced prostate cancer receiving ADT+ARPI.
  • Collected plasma samples at baseline and specific intervals for methylation sequencing.
  • Analyzed associations of baseline PSA and CpG methylation with progression-free survival using Kaplan-Meier and Cox regression.
  • Median progression-free survival was 12.3 months for castration-resistant prostate cancer (CRPC) and 48.4 months for metastatic hormone-sensitive disease (mHSPC).
  • Identified over 33,000 differentially methylated CpGs in CRPC versus 1,621 in mHSPC, with significant associations found at both stages.
  • Stage-specific epigenetic changes were linked to therapeutic response and systemic toxicities, particularly highlighting immune-derived cfDNA alterations.

Abstract

235 Background: Androgen deprivation therapy (ADT) plus androgen receptor pathway inhibitors (ARPI) is standard for advanced prostate cancer (PCa), but progression is inevitable and early biomarkers are lacking. Cell-free DNA (cfDNA) methylation offers a potential minimally invasive approach to monitor tumor and systemic responses. Methods: We conducted a prospective study of 48 men with advanced PCa: 10 with castration-resistant (CRPC) and 38 with metastatic hormone-sensitive disease (mHSPC), all receiving ADT+ARPI. Plasma was collected at baseline and 3 mo (CRPC) or 7 mo (mHSPC). cfDNA was profiled by enzymatic methylation sequencing of 3.98M cytosine-phosphate-guanine motifs (CpGs). Kaplan-Meier and Cox regression tested associations of baseline PSA and CpG methylation with progression-free survival (PFS). Landmark analyses assessed change in CpG from baseline. Differentially methylated regions (DMRs) were identified using a fixed-effect model with gene and tissue-of-origin annotation. Results: Median age was 75 y (CRPC) and 67 y (mHSPC); 90% and 71% were white. Median PFS was 12.3 mo (95% CI, 5.3–NR) for CRPC and 48.4 mo (95% CI, 17.3–NR) for mHSPC. Global methylation shifts were greater in CRPC, suggesting more extensive epigenetic remodeling. At the CpG level, 33,958 differentially methylated CpGs (DMCs) were identified in CRPC vs 1,621 in mHSPC (max ∆β ≤0.8 vs ≤0.55); 15% and 12% reached p<0.01. DMR analysis identified 450 regions in CRPC and 601 in mHSPC (p<0.01), with only 29 genes overlapping, underscoring stage-specific programs. Pathway enrichment was largely driven by normal tissue cfDNA: CRPC showed cardiovascular-associated loci (HSPA12B hypomethylation), aligning with ARPI-related CV toxicity, while mHSPC showed neural loci (AOX1 hypomethylation), consistent with docetaxel neuropathy. Notably, 30–40% of DMRs localized to promoters and 30–50% to CpG islands, highlighting many changes occurring in regions critical for transcriptional regulation. Tissue-of-origin deconvolution confirmed negligible prostate cfDNA in CRPC, consistent with tumor apoptosis evasion. Immune-derived cfDNA showed the strongest changes (4,638 DMCTs in CRPC vs 474 in mHSPC, p<0.01), reflecting immune activation under therapy. Normal tissue cfDNA also revealed systemic off-target effects, paralleling known ARPI- and docetaxel-related cardiovascular and neuropathic toxicities. Conclusions: cfDNA methylation profiling revealed stage- and tissue-specific epigenetic changes linked to PFS, integrating tumor, immune, and normal tissue signals. These data support cfDNA methylation as a minimally invasive biomarker to track therapeutic response and systemic toxicities in advanced PCa.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Miyazaki et al. (2026) studied this question.

synapsesocial.com/papers/69a7cd6ed48f933b5eed9b48https://doi.org/10.1200/jco.2026.44.7_suppl.235
Ask AI
Helpful
Bookmark
Share
View Full Paper