Clinical relevance of DNA damage repair deficiency in metastatic prostatic adenocarcinoma treated with chemotherapy and hormone therapy, and role of liquid biopsy.

ABSTRACT Background: DNA-damage-repair (DDR) gene alterations, particularly BRCA1/2 and ATM, are clinically relevant in mPC for prognosis and therapy. Tumor tissue remains the standard for molecular profiling, but sample availability and quality are limiting factors. Liquid biopsy (LB), based on circulating free DNA (cfDNA), offers a minimally invasive alternative for real-time monitoring of tumor heterogeneity. Methods: We conducted a retrospective study at the Veneto Institute of Oncology including 50 patients (pts) with mPC (2010–2022) with prior tumor tissue and/or LB testing. Mutational status was already known, with 2 equal different cohorts (DDRwild type (wt) and DDRmutated (mut) patients). cfDNA was extracted from plasma and analysed with amplicon-based NGS (BRCA1/2) and capture-based NGS (16 DDR genes). All mutation classes (pathogenic, likely pathogenic, variants of uncertain significance, and non-reported alterations) were considered. Clinical variables, genomic findings, and outcomes were compared between DDRmut and wt patients. Results: Median age at diagnosis was 68 years; 64% had Gleason score (GS) 8 to 10. DDR variants were present in 25/50 patients, associated with higher PSA at LB (0.76 vs. 0.08 ng/mL; p=0.004), longer time to castration resistance (29 vs. 7 months; p=0.002), and family history of cancer (56% vs. 30%; p=0.044). Germline testing confirmed pathogenic variants in 7/16 tested DDRmut pts. LB confirmed DDR variants previously detected in tumor tissue in 20/25 cases. In the DDRwt cohort, 6/25 pts acquired new DDR alterations at LB (2/6 pts carried 2 variants each): 3 BRCA2, 2 ATM, 2 RAD51B, and 1 BRIP1 (VAF 0.5–3.8%). All had positive family history for DDR-related tumors and were on ARPi (androgen receptor pathway inhibitor) therapy with undetectable or declining PSA, indicating disease control by conventional markers,while cfDNA uncovered emerging resistant clones. This discordance highlights the value of LB in capturing therapy-driven evolution, potentially anticipating disease progression. Twelve DDRwt pts retained a wt profile, and 7 were non-evaluable due to cfDNA quality or quantity. Overall concordance with archival tissue was 79% (κ=0.41). Median cfDNA concentration was 5.27 ng/mL, higher in GS 8 to 10 vs. 6 to 7 tumors (5.57 vs. 4.51 ng/mL; p<0.001) and in high volume disease (ρ=0.32; p=0.041). Median cfDNA did not correlate with PSA, but its dynamics reflected tumor evolution, with both loss and acquisition of variants during therapy. Conclusions: LB showed moderate concordance with tissue but revealed dynamic DDR alterations, including new mutations in DDRwt pts under ARPi despite low PSA, supporting its role in longitudinal monitoring and early detection of resistant subclones. cfDNA levels correlated with GS rather than PSA. These findings support LB as a promising tool in precision oncology for mPC, with the need for prospective validation and standardization before routine clinical implementation.