Abstract PR029: Monitoring tumor evolution and phenotypic diversity in metastatic prostate cancer using liquid biopsy profiling

Abstract Background Tumor heterogeneity and lineage plasticity pose major challenges in the management of metastatic prostate cancer (mPC), particularly as the disease evolves under therapeutic pressure. While resistance is often driven by genomic mutations, transcriptomic adaptations also play a critical role. Liquid biopsy offers a minimally invasive strategy to longitudinally monitor both genomics (via circulating tumor DNA, ctDNA) and transcriptomics (via extracellular vesicle RNA, EV-RNA) in real time, providing insights into disease evolution and treatment response. Methods Longitudinal blood samples were collected from a prospective cohort of 145 mPC patients treated at Vall d’Hebron Hospital, spanning hormone-sensitive (HSPC), castration-resistant (CRPC), and neuroendocrine (NEPC) states, and across treatment lines (ARSI, taxanes, others). Plasma was processed to isolate ctDNA and EV-RNA. Mutational profiling of ctDNA (n=262) was performed using Guardant Infinity. Whole-transcriptome profiling of EV-RNA (n=126) was conducted using a custom RExCuE-based library preparation enabling simultaneous gene expression analysis. Exome-capture RNA-seq was performed in matched tumor and blood samples from a collection of patient-derived xenografts (PDXs) representing HSPC (n=12), CRPC (n=4), and NEPC (n=2). Results CtDNA profiling recapitulated the mutational landscape and phenotypic diversity of mPC. In CRPC, AR alterations were most frequent (51%), whereas NEPC showed enrichment of MYC (70%), RB1 (60%), and PI3K pathway mutations (30–70%). Although no variants were detected in 3/262 samples, tumor fraction (TF) was identifiable through methylation-based scoring, underscoring the platform’s sensitivity for both genomic and epigenetic alterations. Baseline ctDNA TF, estimated from variant allele frequency (SNV-TF) or methylation score (methylation-TF), was significantly associated with shorter progression-free survival, discriminating patients’ outcome even at very low TF (2% SNV-TF; 1% methylation-TF). SNV and methylation-TF were strongly correlated (r=0. 8, p0. 0001), though less so in NEPC. Matched EV-RNA revealed concordant gene expression shifts across ctDNA genotypes with prognostic relevance. Both ctDNA and EV-RNA captured inter- and intrapatient heterogeneity, with ctDNA clonality reflecting therapeutic trajectories. CtDNA also captured homologous recombination deficiency (HRD) status, consistent with RAD51 foci assay. Transcriptomic profiling of EV-RNA identified CRPC-related and NEPC-associated gene signatures in both PDXs and patient samples. Furthermore, longitudinal profiling enabled real-time monitoring of lineage transitions, capturing early transcriptomic adaptations to therapy, including emergence of castration resistance and neuroendocrine features. Conclusions Combined analysis of ctDNA and EV-RNA provides complementary insights into tumor evolution and phenotypic diversity, enabling minimally invasive longitudinal monitoring, with EV-RNA uniquely capturing early adaptive changes in gene expression during lineage transformation. Citation Format: Irene Casanova-Salas, Laura Martínez, Daniel Aguilar, Haitham Alatoom, Manuel J. Ramos, Pablo Cresta, Gisela Mir, Anna Oliveira, Mar Moré, Lara De Llobet, Joaquin Mateo. Monitoring tumor evolution and phenotypic diversity in metastatic prostate cancer using liquid biopsy profiling abstract. In: Proceedings of the AACR Special Conference in Cancer Research: Innovations in Prostate Cancer Research and Treatment; 2026 Jan 20-22; Philadelphia PA. Philadelphia (PA): AACR; Cancer Res 2026;86 (2Suppl): Abstract nr PR029.