Extracellular vesicles (EVs) are nano-sized lipid bilayer structures released by most cells, and their distinct proteomic contents offer significant potential for disease biomarker discovery. A crucial yet frequently overlooked aspect is the robustness of EV proteomic profiling across different liquid chromatography-mass spectrometry (LC-MS) platforms. Determining key reproducibility parameters among LC-MS platforms/workflows can streamline downstream validation and expedite the clinical translation of EV protein biomarkers.

To address this, we evaluated the performance of the high-throughput EvoSep ONE platform and the conventional EASY nLC 1200 platform in profiling the serum EV proteome. Using samples from patients with pulmonary nodules (PNs) patients, we compared the sensitivity and accuracy of these two platforms, as well as their performance in stratifying benign vs. malignant PNs individuals. For qualitative profiling, EASY nLC 1200 had deeper coverage of EV proteins than the high-throughput EvoSep workflow, demonstrating superior sensitivity. Despite this difference, both platforms exhibited robust performance in stratifying malignant and benign PN patients using the label-free quantification (LFQ) abundance of EV proteins. We determined that EV proteins of mid and high LFQ abundance exhibit strong intra- and inter-platform reproducibility. Notably, a common panel of differential EV proteins was identified by both platforms to be significantly upregulated in patients with malignant PNs. The malignant upregulation of ORM1, a high-LFQ differential EV protein, was confirmed through ELISA validation in an independent patient cohort. Conversely, ELISA assays failed to validate the differential expressions of COMP, MVP, and RBP4, which displayed low LFQs in the proteomics data. Collectively, these results indicate that EV proteins of mid-to-high LFQ abundance are more reliable in differential analysis.

Our findings underscore the importance of cross-platform validation in proteomic profiling of EVs and suggest a threshold of EV protein LFQ abundance for developing reliable biomarkers. This study provides valuable insights that could guide future research in the field.