Abstract Significance. Liquid biopsies from patients with metastatic castration resistant prostate cancer (mCRPC) offer a dynamic approach to assess temporal changes in the tumor genome, transcriptome, and proteome. In this study, we deployed a first-in-class extracellular vesicle (EV) proteomic assay (PMID: 39766159) using plasma from 100 prospectively enrolled mCRPC patients treated with 177Lu-PSMA-617 as well as 20 healthy male controls. We interrogated the baseline plasma samples to establish the mCRPC EV proteomic landscape, understand biological mechanisms of response/resistance, and inform patient outcomes based on progression free or overall survival (PFS, OS). Patients and Methods. Clinical variables included baseline PSA, molecular tumor volume (MTV), PSA50 response, metastatic sites, previous therapies, and survival over 12 months. The plasma EV proteome was determined through data-independent acquisition mass-spectrometry (DIA-MS) based on a spectral library we established from prior prostate cancer studies, which accounted for 6000 unique proteins. Statistical analyses (HR, OR) were performed using Cox proportional hazards and logistical regression models. Harrell’s bootstrap optimism method was used to derive optimism-corrected C-indices and AUC for outcomes. Results. In this mCRPC cohort, 91% of patients had bone metastasis, and 72% had received 3 or more systemic therapies. DIA-MS identified 5, 137 proteins in EVs from mCRPC patients and 3, 958 proteins in EVs from healthy controls. Notably, 1500 proteins were exclusively detected in the EVs from mCRPC patients and further evaluated. This included expected prostate cancer proteins such as PSMA, TMPRSS2, KLK2/3, and EPCAM. In addition, we found druggable therapeutic targets that we categorized as cell-surface proteins (HER2, Trop-2, STEAP1), kinases (PIK3CA, CDK6), or immune checkpoint targets (PD-L1, B7-H3, CD47). Surprisingly, HER2 expression exhibited robust positive correlations with PSMA, Trop-2, STEAP1 and B7-H3 in mCRPC patients. Given that suites of EV proteins exhibited prognostic value or were associated with treatment response, we focused on modeling key mCRPC targets. While baseline PSA exhibited modest prognostic discrimination with a C-index of 0. 518 for PFS and 0. 619 for OS, the addition of B7-H3 EVs improved the C-index to 0. 625 for PFS and 0. 694 for OS. Trop-2 similarly increased the C-index to 0. 611 for PFS and 0. 657 for OS. When considering PSA50 response, multivariate analysis indicated that B7-H3 (OR: 0. 292, p=0. 032) and Trop-2 (OR: 0. 146, p=0. 002) were associated with reduced PSA50 response. Baseline PSA demonstrated weak discrimination (AUC: 0. 578) at predicting PSA50 response. the addition of either B7-H3 (AUC: 0. 648) or Trop-2 (AUC: 0. 705) enhanced PSA50 response prediction. Conclusions. Altogether, plasma EV-proteomics can be deployed as a non-invasive assay that provides biological insights into advanced mCRPC patients. Our study supports the integration of our assay and models into biomarker-driven mCRPC trials to improve patient stratification. Citation Format: Justin H. Hwang, Ali T. Arafa, Ella Boytim, Lauren Yu, Nicholas Zorko, Scott M. Dehm, Justin M. Drake, Emmanuel S. Antonarakis. Defining the Extracellular Vesicle Proteome of Metastatic Castration Resistant Prostate Cancer Patients treated with 177Lu-PSMA-617 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 IA005.
Hwang et al. (Tue,) studied this question.