Noninvasive fragmentomic profiling of plasma cell-free DNA (cfDNA) in metastatic prostate cancer (mPC): Associations with transcriptomic phenotypes and disease states.

260 Background: Current liquid biopsy assays yield few clinically actionable insights, because they measure discrete genomic alterations rather than the actual gene expression pathways that drive cancer resistance and progression. This is because cfDNA is stable whereas circulating RNA is degradable and difficult to analyze. To address this challenge, fragmentomic analysis has been developed, which measures the variability of cfDNA fragments at promoter regions (promoter fragment entropy, PFE) to infer tumor gene expression. Here, we developed PC-FragSeq, a PC-specific fragmentomic sequencing panel targeting 590 genes linked to mPC progression, resistance, and clinical outcome. Methods: We assembled the panel through meta-analysis of publicly available gene expression datasets of aggressive, treatment-resistant mPC. Active promoter regions of dominant transcript isoforms were identified, and a custom hybrid capture panel was designed for optimal on-target enrichment at >500X coverage. Corresponding exonic regions were included for parallel genomic analysis. With IRB approval, we analyzed plasma from 48 patients enrolled in a prospective, clinically annotated institutional biorepository. Of these, 20 had metastatic hormone sensitive PC (mHSPC) and 28 had metastatic castration resistant PC (mCRPC), some at multiple time points. Extracted cfDNA was analyzed by low-pass whole genome sequencing (LP-WGS) followed by PC-FragSeq to generate PFEs. These inferred gene expression values were assessed for associations with clinical states and genomic profiles. Results: PC-FragSeq achieved >85% probe coverage at >500X depth across targeted regions. PFE-based analysis robustly distinguished mHSPC from mCRPC, revealing differential gene expression signatures consistent with known transcriptional reprogramming during progression to castration resistance. Additional expression phenotypes were identified in patients with mCRPC progressing through later lines of therapy, including androgen receptor pathway inhibitors, taxane chemotherapy, poly (ADP-ribose) polymerase inhibitors, and radioligand therapy. PC-FragSeq exonic sequence data and LP-WGS from the same cfDNA samples yielded PC-relevant genomic alterations, including AR amplification and TMPRSS2:ERG fusion, which correlated with matched transcriptional shifts. Conclusions: PC-FragSeq enables high-resolution, repeatable assessment of PC gene expression profiles from a simple blood draw. This minimally invasive approach has the potential to enable longitudinal monitoring of disease progression and resistance pathways for discovery of actionable biomarkers and novel therapeutic targets.