A liquid biopsy targeted proteomic assay to predict outcomes in castration resistant prostate cancer patients treated with 177Lu-PSMA-617.

Abstract

250 Background: Plasma-based targeted proteomics is an emerging area with great potential to aid in cancer diagnostics via rapid, repeated, and quantifiable measurements of key biomarkers. We developed a liquid biopsy-based proteomic assay to quantify druggable protein targets in plasma that may predict response to 177 Lu-PSMA-617 in metastatic castration resistant prostate cancer (mCRPC) patients. Methods: Blood was serially collected from 37 mCRPC patients (84 samples in total) as part of a prospective clinical trial. Circulating tumor cells (CTC) were isolated from 27 samples, while plasma was obtained from 84 samples, 10 of which overlapped with the CTC samples. Samples were processed using Astrin Bioscience’s AI-empowered label-free holographic imaging and in-flow protein marker expression to enrich CTCs. CTCs and plasma samples were processed to generate peptides for analysis on a FAIMS equipped Stellar instrument. Heavy labeled peptide standards were used for internal normalization. Each protein evaluated provided single digit attomole limit of detection (LOD). Log-rank tests were performed to assess for overall survival (OS). Results: CTCs were successfully isolated from 27 samples, with a mean CTC count of 3.2 cells/mL and a median of 1.7 CTCs/mL. 28 proteins were selected for targeted proteomic analysis, and we assessed 10 key proteins in our preliminary study. When comparing the CTC proteome vs plasma proteome, we found that only 10% of proteins were unique to CTCs, while 76% were distinct to plasma or shared between CTCs and plasma, supporting the shift to plasma for future analyses. Serum PSA levels correlated with PSA protein expression measured by our assay (r=0.7, p = 0.0005). Serum alkaline phosphatase levels correlated with Trop-2, PSA, AR, DLL3, and PD-L1 protein levels (all p<0.05). PET scan-derived molecular tumor burden correlated with PD-L1 expression (r=0.36 p=0.04) but inversely correlated with CHGA levels (r= -0.41, p=0.01). Higher levels of DLL3 [HR=7.62, (CI 2.85 - 20.37) p=0.01], Trop-2 [HR=2.37, (CI 0.90-6.21) p=0.06] and PD-L1 [HR=3.53, (CI:1.40-8.92) p=0.005] were associated with worse OS. Serial (baseline and on-treatment) sampling of a subset of these patients revealed dynamic changes in proteomic markers including PSMA, PD-L1, and Trop-2 that may predict response to 177 Lu-PSMA-617. Conclusions: We have developed a quantitative proteomic assay capable of detecting attomole concentrations of key druggable proteins in mCRPC patients. The assay’s ability to detect and quantify key therapeutic targets in both CTCs and plasma makes it broadly applicable. This first-of-its-kind proteomic assay can guide precision oncology approaches by profiling and quantifying multiple druggable proteins from individual patients, ultimately minimizing harm and improving targeted therapy selection.