P05.04.B ENRICHMENT AND ANALYTICAL METHODS IN CSF PROTEOMICS: EACH NAIL NEEDS A DIFFERENT HAMMER

Abstract BACKGROUND While proteomic analysis of CSF is emerging as a promising field in neuro-oncology, numerous enrichment and analytical approaches are possible and one size does not fit all. Leveraging the high volume of CSF available through our multi-agent intraventricular therapy program for CNS tumors, we report a systematic technical comparison of various proteomic analytic methods. MATERIAL AND METHODS Samples: Cohort 1: 5 patients (3 CNS lymphoma, 2 CNS leukemia); Cohort 2: 15 CNS lymphoma; Cohort 3: 19 CNS lymphoma.EV isolation: Volume optimization (1.5-9mL ranges) for EV isolation was conducted on Cohort 1. Nanoparticle tracking was used to measure concentration and size distribution of EV fractions. Reproducibility of the fractionation process was evaluated in pooled samples from Cohort 2 (6mL replicates). Evaluation of enrichment methods: Cohort 3 samples were used for evaluation of peptide detection profiles across MStern (50μL), Seer (250μL), N-glycoproteomics (N-Gp, 200μL), P150 and P20 (both EV isolation techniques, 6mL). Bias toward specific biophysical properties based on method was evaluated. RESULTS EV isolation: Based on EV-specific peptides detected, 6mL was established as the lowest feasible input volume. The P20 fractions showed consistently higher protein detection and particle concentration than P150. Evaluation of enrichment methods: Quantitative analysis revealed substantial variation in detection capacity; Seer had the greatest average peptide identification (17,000), followed by P20 (9000), MStern (5,500), P150 (5,000) and N-Gp (1,000). Seer had the greatest number of unique peptides. Principle component analysis showed assay led to greater variance than patient differences. Seer had highest reproducibility, followed by MStern & P20 > P150 > N-Gp. Biophysical properties were similar with all methods except N-Gp, which showed slight preferential enrichment of acidic glycopeptides and preponderance for larger, hydrophilic peptides. Cellular compartment enrichment was complementary: P20 (mitochondrial), N-GP (lysosomal), and Seer (nuclear). Pathway and functional enrichment patterns were complementary across methods as well. CONCLUSION In CSF proteomics, one size does not fit all, and each enrichment method yields complementary information. EV analysis can be conducted on as little as 6mL of CSF. While Seer demonstrates greatest yield and reproducibility, cost is very high. MStern may represent a balance of moderate peptide detection, limited input volume and cost, and high throughput for discovery proteomics. Future optimization will evaluate different mass spectrometry scan modes, such as data intendent acquisition (DIA-MS) to further enhance throughput and quantitative precision.