Follow us: |
Circulating Protein Biomarkers for Primary Cardiomyopathy: Evidence from Mendelian Randomization and Clinical Validation
1Department of Cardiology, Jintang County Second People’s Hospital, Chengdu, Sichuan, People's Republic of China
2Department of Military Traffic Injury Prevention and Control, Daping Hospital, Army Medical University, Chongqing, China
3Department of Urology, The First Affiliated Hospital of Chongqing Medical University, Chongqing, People's Republic of China
Anatol J Cardiol - PubMed ID: 42552772 DOI: 10.14744/AnatolJCardiol.2026.5927
Full Text PDF

Abstract

Background: Primary (non-ischemic) cardiomyopathy poses diagnostic challenges due to phenotypic heterogeneity and non-specific clinical presentation, and commonly used circulating biomarkers have limited disease specificity. The aim was to prioritize mechanism-informed plasma biomarkers and evaluate their diagnostic discrimination in a clinical cohort.

Methods: pQTL data from 8 proteomic studies covering 5,034 proteins were integrated, and Mendelian randomization was performed across 3 cardiomyopathy GWAS datasets. Top MR candidates were clinically validated by enzyme-linked immunosorbent assays quantification in a prospective case–control cohort of 81 subjects (48 cardiomyopathy patients and 33 healthy controls). Echocardiography evaluated cardiac function and remodeling, correlating with plasma biomarker levels.

Results: Proteome-wide MR prioritized 27 proteins linked to cardiomyopathy risk. Clinical validation confirmed significant differences in plasma levels of 5 selected biomarkers (CD163, LGALS3BP, FABP5, FSTL3, VCAM1). CD163 (odds ratio (OR) = 1.019, P < .0001) was positively associated with CM, while LGALS3BP (OR = 0.390, P < .0001), FABP5, FSTL3, and VCAM1 showed inverse associations. Multivariate analysis revealed CD163 and LGALS3BP as independent CM predictors, strongly correlating with echocardiographic indices of cardiac dysfunction. Receiver operating characteristic analyses showed individual area under the curve (AUC) ranging from 0.632 (VCAM1) to 0.917 (CD163). Given the modest discrimination of VCAM1, 5- versus 4-protein combined models were compared, and a parsimonious 4-protein panel was selected (CD163, LGALS3BP, FABP5, FSTL3). The 4-protein panel achieved an apparent AUC of 0.993, with optimism-corrected and repeated cross-validated AUCs of 0.988 and 0.980.

Conclusion: This integrative genetic-to-clinical approach supports a 4-protein plasma panel with promising diagnostic discrimination for primary cardiomyopathy, warranting external validation. CD163 and LGALS3BP emerged as particularly informative markers for further evaluation.