Plasma proteins identify increased number of carotid plaques and predict occurring atherosclerotic events

BackgroundThe number of carotid plaques independently predicts incident atherosclerotic cardiovascular disease (ACVD). However, performing vascular imaging in apparently healthy subjects is challenging, owing organizational/economical barriers. Plasma proteomics can offer an alternative approach to identify individuals with carotid plaques, at high risk of eventually developing ACVD. MethodsIn this observational study, we studied by Normalized Protein eXpression (NPX; OlinkTM), the plasma levels of 368 proteins in 664 subjects from the PLIC study, who were screened by ultrasound for the presence of carotid plaques. We clustered, by artificial intelligence, the proteins that more accurately identified subjects, stratifying them according to the number of plaques they presented with. We also study prediction of occurring events over 22 years. Results299/664 subjects had at least 1 carotid plaque. Among those, 77 subjects presented with only one plaque, 101 with 2 plaques and 121 with [&ge;]3 plaques (3+). The remaining 365 subjects with no plaques acted as controls. The proteins differently expressed versus controls increased as a function of the number of plaques. 32 proteins were shared among the groups of subjects with plaques, but 87, significantly associated with the presence of 3+ plaques, improved the AUC of the ROC, together with the ACVD risk factors, to discriminate subjects with 3+ plaques versus the AUC of the ROC considering the ACVD risk factors only (AUC= 0.918 (0.887-0.943) vs AUC= 0.760 (0.716-0.801) respectively, p<0.001). The ACVD risk factors barely predicted the 198 occurring events (AUC= 0.559 (0.521-0.598)), but proteomics associated with plaques improved the prediction (AUC= 0.739 (0.704-0.773), p<0.001). By analyzing the biological processes, we identified that chemotaxis/migration of leukocytes and the signaling of interleukins/cytokines were the top pathways involved. ConclusionsPlasma proteomics helps to identify apparently healthy subjects with higher number of carotid plaques more accurately and to predict occurring ACVDs in those individuals.