Mass-spectrometry-based plasma proteomics has advanced rapidly through improved enrichment chemistries, automation, and high-resolution instrumentation. However, reproducibility across clinical studies remains limited by the extreme dynamic range of the plasma proteome and by pre-analytical variation in sample collection. As a core proteomics facility serving diverse academic and clinical collaborators, we frequently receive plasma samples collected under variable conditions and must process them reproducibly, cost-effectively, and without introducing artefacts. Achieving a balance between analytical depth, quantitative precision, and robustness to variable sample quality is therefore essential.
To address this, we benchmarked five plasma proteomics workflows—MagNet (SAX beads, ReSyn), ENRICHplus (PreOmics), HP kit (ReSyn), neat plasma (agarose bead precipitation), and perchloric acid precipitation with neutralisation (PCA-N)—using human EDTA plasma from two pre-analytical sources differing in centrifugation protocols. All samples were prepared in duplicate and analysed in data-independent acquisition (DIA) mode on a Bruker timsTOF HT MS using a 30-minute gradient and a 25 cm IonOpticks Aurora column.
Bead-based workflows (HP, PreOmics, SAX) achieved the greatest proteome depth but also showed enrichment of proteins consistent with plasma contamination—an artefact previously described by Korff et al.1. This effect was absent in PCA-N and neat agarose-based workflows, which yielded cleaner quantitative profiles, albeit at reduced depth. The results align with recent benchmarking studies showing that bead-based enrichments enhance sensitivity to low-abundance proteins but are more prone to pre-analytical bias2-4.
Beyond proteome coverage, we also examined quantitative performance, an aspect increasingly critical for translational applications. Across the distribution of protein coefficients of variation (CVs), PCA-N and neat agarose-based workflows achieved the lowest CVs, reflecting strong reproducibility and quantitative stability. Bead-based methods displayed slightly higher variability but covered a broader dynamic range, with the majority of proteins still maintaining CVs below 25%. Importantly, workflows exhibiting the most consistent intensity profiles also maintained optimal chromatographic sampling, achieving approximately ~6 data points across a chromatographic peak, which supports precise area-under-curve integration and improved reproducibility across replicates. Comparative analyses performed on our in-house Orbitrap Astral platform (20-minute gradient, 15 cm IonOpticks Aurora column) yielded higher identification rates with shorter run times, albeit at the expense of reduced chromatographic sampling (~3 data points per peak), illustrating the practical trade-off between throughput, depth, and quantitative precision.
These findings provide practical guidance for selecting plasma workflows in translational proteomics, particularly within shared-resource environments where pre-analytical control is limited but high-quality, reproducible, and affordable data remain paramount.