Methods
We analyzed 305 patients with mCRPC treated with Lu-177-PSMA-617and standardized Guardant Health (GH) liquid-biopsy genomic profiling from ConcertAI’s Prostate Translational360™ (an integrated EHR, claims, and social determinants of health linked with molecular datasets). Baseline variables included prior ARPI/taxane exposure, metastatic pattern, and labs for prostate-specific antigen (PSA), alkaline phosphatase (ALP), and hemoglobin (Hgb). Baseline genomics were grouped by pathway: AR, PTEN, HRR (limited to commonly reported genes across GH assays), and exploratory (FGFR1/EGFR/CTNNB1). Maximum variant allele fraction (mVAF) and tumor mutational burden (TMB) were evaluated as measures of genomic burden. Outcomes evaluated were PSA50 ≤12 weeks, PSA90 ≤12 weeks, progression free survival (PFS), time to next treatment (TTNT) and overall survival (OS).
Results
The cohort was heavily pre-treated (≥2 ARPIs 60%, ≥2 taxanes 56%) and predominantly bone-dominant (80% bone-only, 15% visceral ±bone). Median baseline PSA = 17.1ng/mL; mVAF= 1.0% [IQR 0.3-5.7]; TMB =8.3mut/Mb [IQR 5.8-11.1]. PSA50≤12w was achieved by 33% and PSA90 by 10%. AR amplification (HR: 1.99 PFS, 2.3 OS) and PTEN (HR: 1.33 PFS, 1.3 OS) were associated with shorted PFS and OS compared with wt-tumors. Similarly, exploratory alterations (HR: 2.56 PFS, 3.21 OS) showed adverse outcomes, whereas HRR mutations (HR: 1.15 PFS, 1.46 OS) were not independently associated with response.
Conclusions
Baseline AR activation, PTEN, and alternative pathway alterations identify mCRPC patients less likely to benefit from PSMA-RLT, while bone dominant disease and wt-profiles show more favorable outcomes. Integrating genomic profiling into RLT eligibility and trial design could optimize patient selection and guide early combination strategies. Subsequent analyses will expand to multi-tested patients to validate these predictors across real world platforms.
.svg)
