Serial proteomic characterization of plasma extracellular vesicles (EV) to identify changes that predict outcomes in metastatic castrate-resistant prostate cancer (mCRPC) patients receiving ¹⁷⁷ Lu-PSMA-617.

Abstract

258 Background: Extracellular vesicles (EVs) in plasma offer a minimally invasive window into tumor biology. We hypothesized that deep proteomic profiling of plasma EVs can characterize dynamic molecular changes and predict outcomes in patients with mCRPC treated with 177 Lu-PSMA-617. Methods: Of 100 prospectively enrolled patients receiving 177 Lu-PSMA-617 (Arafa, et al. ASCO 2025), 58 men had serial (baseline and on-treatment) plasma samples. EVs were isolated using differential ultracentrifugation and analyzed by shotgun mass spectrometry. Protein expression patterns (e.g. PSMA, B7-H3) were categorized as undetected at both baseline and follow-up (U->U) or detectable at both timepoints (D->D). Relative protein expression changes were dichotomized as increasing (>10% increase) versus no change/decreasing (not a >10% increase). Surface protein changes were quantified and associations with overall survival (OS) were sought using log-rank tests. Pathway-level analysis was conducted using pre-ranked gene set enrichment analysis (GSEA) to identify pathways associated with outcomes. Results: A total of 6,306 proteins were identified, with about 20% mapping to key cell-surface markers including PSMA, B7-H3, Trop-2, and STEAP1. When patients were stratified by the detection of the key 4 surface proteins (0–1, 2, or 3–4); increasing numbers of detected proteins were associated with progressively shorter OS using both baseline and follow-up samples (p<0.0001 for both). Outcomes based on serial protein detection (D->D) or non-detection (U->U) patterns are shown in the Table. Increases in EV-derived PSMA (HR 2.7, 95% CI 1.3–5.9, p=0.002), Trop-2 (HR 4.8, 95% CI 1.5–15.7, p<0.0001), and STEAP1 (HR 5.7, 95% CI 1.6–20.2, p<0.0001) were associated with worse OS, with a similar trend for increasing B7-H3 (HR 1.7, 95% CI 0.8–3.4, p=0.16). In GSEA analysis, fatty acid metabolism (NES=1.8, q=0.004), Hedgehog signaling (NES=1.7, q=0.02), bile acid metabolism (NES=1.6, q=0.03), and MYC targets (NES=1.5, q=0.04) were enriched in on-treatment samples from progressors, while angiogenesis (NES=1.6, q=0.02) was enriched in on-treatment samples from responders. Conclusions: On-treatment persistent detection or upregulation of EV-derived surface proteins (PSMA, B7-H3, Trop-2, and STEAP1) was associated with inferior overall survival in mCRPC patients treated with 177 Lu-PSMA-617. These findings highlight the potential clinical utility of dynamic plasma EV proteomics for prognostic stratification and clinical trial selection. U->U (days) D->D (days) OS HR (difference) P PSMA 297 141 4.52 0.002 B7-H3 NR 175 7.3 p<0.0001 Trop-2 247 79 5.27 p<0.0001 STEAP1 268 79 10.76 p<0.0001