Understanding why androgen-pathway drugs eventually fail in metastatic prostate cancer is one of oncology's most pressing puzzles. A large-scale genomic analysis tied to a completed Phase III trial now maps — with unusual longitudinal precision — the molecular escape routes tumors exploit under treatment pressure, offering a potential blueprint for predicting which patients will resist therapy early versus late.
Drawing on 327 matched plasma samples from the Phase III Alliance A031201 trial, investigators used the AR-ctDETECT cell-free DNA sequencing assay to profile circulating tumor DNA at both baseline and radiographic progression in men receiving first-line enzalutamide with or without abiraterone. At progression, overall cfDNA levels rose and androgen receptor (AR) gene copy-number gains intensified alongside AR genomic structural rearrangements (AR-GSRs) — alterations predicted to truncate the AR ligand-binding domain, the very pocket targeted by enzalutamide and abiraterone. A distinct extrachromosomal DNA signature co-enriched with these AR-GSRs at progression. Notably, patients with the longest radiographic progression-free survival tended to accumulate AR-centric alterations, whereas early progressors carried a heavier burden of non-AR genomic changes, including copy-number losses in DNA-damage response genes such as FANCA, POLD1, and RAD54L, as well as cell-cycle regulator CDKN2A.
This work is analytically significant for several reasons. First, it confirms and extends prior single-institution observations on AR amplification as a dominant resistance mechanism to ARPI combination therapy in a trial-grade, prospectively collected cohort — lending considerably more weight than retrospective series. Second, the emerging non-AR resistance cluster — centering on homologous recombination and replication-fidelity genes — overlaps mechanistically with PARP inhibitor sensitivity profiles, raising the hypothesis that liquid biopsy at progression could stratify patients toward PARP inhibitor or platinum-based salvage rather than further AR-directed therapy. Key limitations include the assay's AR-focused design, which may undercount alterations in non-AR pathways, and the observational nature of the genomic substudy, which precludes causal inference. Still, for a field moving toward biomarker-guided sequencing of therapies, this dataset represents a meaningful step beyond cross-sectional snapshot genomics.