Plasma proteomics underpins much of today’s biomarker discovery pipeline, yet biological variation and extreme dynamic range continue to challenge depth, reproducibility, and throughput. Recent advances in mass-spectrometry (MS) instrumentation have enabled deeper proteome coverage, but systematic comparisons across biofluids, species, and modern sample-preparation strategies remain limited.
Here, we benchmarked eight sample-preparation workflows spanning neat approaches (SP3, STrap), depletion (perchloric acid, PerCA), corona-enrichment (Enrich-iST, ProteoNano) and extracellular vesicle (EV)-enrichment (MagNet HILIC/SAX). We have also analysed performance across human plasma, human serum and rat plasma, evaluating if methods are biofluid- and species-agnostic. All samples were analysed on Orbitrap Astral (Thermo) using two plasma-optimised data-independent (DIA) methods: one discovery-maximised and one throughput-maximised. We identified 2,726 human and 3,767 rat proteins across workflows and methods, including ~1,000 from neat plasma. As expected, the higher-throughput method reduced proteome depth by ~20–30%, but with workflow- and species-specific effects. Enrichment, via corona-formation or EV-enrichment, strongly shifted the detectable proteome, and redistributed the dynamic range. Resultant subproteomes of circulating EVs achieved excellent depth, but revealed a distinct sub-proteome that was not accessible by neat, depleted, or secreted-protein-enriched approaches. Several workflows performed substantially better in rat plasma, highlighting the need to consider species-specific sensitivity in preclinical studies.
Importantly, each workflow produced a unique distribution of tissue-, organ-, and immune-associated proteins. Depletion and enrichment strongly shifted the detectable proteome, emphasising that workflow choice determines not only depth but biological bias. More than 90% of detected proteins were differentially abundant across workflows, with corona-enrichment revealing the largest methodological effects. Collectively, these results map a rapidly expanding plasma-proteomics landscape and demonstrate that no single workflow is universally optimal.