ABSTRACT
INTRODUCTION: Inflammation associated with rheumatic heart disease (RHD) is influenced by gene polymorphisms and inflammatory cytokines. There are currently no immunologic and genetic markers to discriminate latent versus clinical patients, critical to predict disease evolution. Employing machine-learning, we searched for predictors that could discriminate latent versus clinical RHD, and eventually identify latent patients that may progress to clinical disease. METHODS: A total of 212 individuals were included, 77 with latent, 100 with clinical RHD, and 35 healthy controls. Circulating levels of 27 soluble factors were evaluated using Bio-Plex ProTM® Human Cytokine Standard 27-plex assay. Gene polymorphism analyses were performed using RT-PCR for the following genes: IL2, IL4, IL6, IL10, IL17A, TNF and IL23. RESULTS: Serum levels of all cytokines were higher in clinical as compared to latent RHD patients, and in those groups than in controls. IL-4, IL-8, IL-1RA, IL-9, CCL5 and PDGF emerged in the final multivariate model as predictive factors for clinical, compared with latent RHD. IL-4, IL-8 and IL1RA had the greater power to predict clinical RHD. In univariate analysis, polymorphisms in IL2 and IL4 were associated with clinical RHD and in the logistic analysis, IL6 (GG + CG), IL10 (CT + TT), IL2 (CA + AA) and IL4 (CC) genotypes were associated with RHD. CONCLUSION: Despite higher levels of all cytokines in clinical RHD patients, IL-4, IL-8 and IL-1RA were the best predictors of clinical disease. An association of polymorphisms in IL2, IL4, IL6 and IL10 genes and clinical RHD was observed. Gene polymorphism and phenotypic expression of IL-4 accurately discriminate latent versus clinical RHD, potentially instructing clinical management.
