Plasma proteomics based sub-phenotypes in coronary artery disease are related to plaque characteristics and clinical outcome

Le résumé fourni par la source

Abstract Background Relating plasma proteomics to coronary plaque characteristics as assessed by intravascular imaging may provide insights into biological mechanisms as well as prognostication of coronary artery disease (CAD). Objective We investigated whether a proteomics-driven clustering approach can identify distinct CAD sub-phenotypes linked to clinical outcomes and plaque features. Methods We examined 581 CAD patients from the ATHEROREMO study undergoing coronary angiography or percutaneous coronary intervention. Using a targeted high-throughput proteomics platform, we quantified 369 plasma proteins related to cardiometabolic processes. All patients underwent virtual histology intravascular ultrasound (VH-IVUS), and a subset of 191 patients underwent near-infrared spectroscopy (NIRS), in a non-culprit coronary artery. We first assessed associations between individual proteins and plaque features assessed by IVUS and NIRS, and clinical outcomes (all-cause mortality and major adverse cardiac events (MACE)). We then constructed a protein–protein interaction network and applied a Markov clustering algorithm to group proteins by functional processes. We selected the five largest functional groups and performed K-means clustering using the proteins representing each of these groups. For each functional group, patients were assigned to sub-phenotypes based on their proteomic profiles, and these sub-phenotypes were then related to clinical outcomes and plaque characteristics. Results The mean age of the cohort was 61.5 years (IQR:53.5–61.5), and 75% were male. Over a median follow-up of 12.8 years (IQR:10.1–13.4), 179 patients died; while 151 MACEs occurred (follow-up 4.3 years (IQR:1.37- 4.89)). After multiple testing correction, plaque burden was significantly associated with KIT (β=-4.67[95%CI: -7.14 to -2.02],p=0.042) and AZU1 (β=-1.34[95%CI: -2.03 to -0.65],p=0.042). Dense calcium percentage was associated with 25 proteins, including KIT (β=-0.51[95%CI: -0.77 to -0.25],p=0.005) and AZU1 (β=-0.166[95%CI: -0.24 to -0.09], p=0.001). The region-of-interest lipid core burden index(ROI LCBI) was associated with UMOD (β=-0.98[95%CI:-1.4 to -0.47],p=0.05). After biological dimensionality reduction we focused on the 5 biggest functional groups. Associations of the five functional groups with plaque characteristics and clinical outcome are shown in Figure 1. In particular, cluster 2 of functional group 4—encompassing coagulation and vascular processes (i.e. hemostasis and vascular stability)—was significantly associated with MACE, all-cause mortality, and plaque characteristics (lesions with plaque burden ≥70%(PB70) and thin-cap fibroatheroma(TCFA) lesions with PB≥70%). Conclusion This study is the first to apply a proteomics-driven, biologically informed dimensionality reduction with subsequent clustering, in the context of coronary plaque features. We identified distinct CAD sub-phenotypes associated with plaque features and clinical outcome.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.