Project description:Background: Using proteomics, we strove to reveal novel molecular subtypes of human atherosclerotic lesions, study their associations with histology and imaging and relate them to long-term cardiovascular outcomes. Methods: 219 samples were obtained from 120 patients undergoing carotid endarterectomy. Sequential protein extraction was combined with multiplexed, discovery proteomics. Parallel reaction monitoring for 135 proteins was deployed for targeted validation. A combination of statistical, bioinformatics and machine learning methods was used to perform differential expression, network, pathway enrichment analysis and train and evaluate prognostic models. Results: Our extensive proteomics analysis from the core and periphery of plaques doubled the coverage of the plaque proteome compared to the largest proteomics study on atherosclerosis thus far. Plaque inflammation and calcification signatures were inversely correlated and validated with targeted proteomics. The inflammation signature was enriched with neutrophil-derived proteins, including calprotectin (S100A8/9) and myeloperoxidase. The calcification signature contained fetuin-A, osteopontin, and gamma-carboxylated proteins. Sex differences in the proteome of atherosclerosis were explained by a higher proportion of calcified plaques in women. Single-cell RNA sequencing data attributed the inflammation signature predominantly to neutrophils and macrophages and the calcification signature to smooth muscle cells, except for certain plasma proteins that were not expressed but retained in the plaque, i.e., fetuin-A. Echogenic lesions reflect the collagen content and calcification of plaque but carotid Duplex ultrasound fails to capture the extent of inflammatory protein changes in symptomatic plaques. Applying dimensionality reduction and machine learning on the proteomics data defined 4 distinct plaque phenotypes and revealed key protein signatures linked to smooth muscle cell content, plaque calcification and structural extracellular matrix, which improved the 9-year prognostic AUC by 25% compared to ultrasound and histology. A biosignature of four proteins (CNN1, PROC, SERPH, and CSPG2) independently predicted the progression of atherosclerosis and cardiovascular mortality with an AUC of 75% Conclusion: We combined discovery and targeted proteomics with network reconstruction and clustering techniques to provide molecular insights into protein changes in atherosclerotic plaques. The application of proteomics and machine learning techniques revealed distinct clusters of plaques that inform on disease progression and future adverse cardiovascular events.

Dataset Information

Transcription profiling of peripheral blood from patients with carotid artery stenosis

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure