Low serum irisin levels predicted newly atrial fibrillation in patients with any phenotypes of chronic heart failure

Abstract Background Atrial fibrillation (AF) continues to be a major health problem among patients with known cardiovascular disease including heart failure (HF). AF is not only powerful risk factor for newly HF, but also it contributes to untoward clinical course of the condition. Although, natriuretic peptides are well-validated biomarkers to predict poor clinical outcomes of HF, their accuracy for AF in patients with different phenotypes of HF is not optimal. Recent studies showed that adipocytokines including irisin affect the risk of AF occurrence. We hypothesized that serum irisin can have additional discriminative potency for AF among HF in patients. Methods The study group comprised 541 consecutive patients with chronic HF aged 41 to 76 years recruited for the study from October 2020 to December 2022. We included 153 patients with any phenotypes of HF who were hospitalized due to paroxysmal or persistent AF and a control group of 388 age- and sex-matched individuals admitted to hospitals due to exacerbation of HF. All patients included in the study were examined by conventional general clinical and physical investigations, B-mode echocardiography and Doppler. The plasma levels N-terminal-natriuretic pro-peptide (NT-proBNP), high sensitivity (hs) C-reactive protein, tumor necrosis factor-alpha (TNF-alpha), hs-troponin T, adropin, apelin, leptin, adiponectin, and irisin were measured using ELISA. Results Compared with patients from control groups, patients with AF had greater left atrial volume index (LAVI), higher serum levels of NT-proBNP, hs-CRP, TNF-alpha, and lower left ventricular ejection fraction (LVEF), irisin and apelin levels. There were no significant difference between groups in left ventricular mass index, the serum levels of hs-troponin T, adropin, leptin, and adiponectin. Multivariate linear regression adjusted to LVEF and LAVI revealed that irisin and NT-proBNP were the predictors for AF. We found that the well-balanced cut-off point for serum concentration of irisin (AF versus non-AF) were 10.5 ng/mL (area under curve [AUC] = 0.96, sensitivity = 81.0%, specificity = 88.0%; P = 0.0001). Cutoff point of NT-proBNP that distinguished patients with AF and without AF was 750 pmol/L (AUC = 0.96; sensitivity = 72.7%, specificity 76.5%, р = 0.0001). Multivariate Cox regression model yielded that the serum levels of irisin < 10.5 ng/mL (odds ratio [OR] = 1.40; P = 0.001) and NT-proBNP > 750 pmol/mL (OR = 1.22; P = 0.012), LAVI > 34 mL/m2 (OR = 1.06; P = 0.010) remained to be strong predictors for newly AF. Add-on of irisin < 10.5 ng/mL to the predictive model constructed from NT-proBNP (> 750 pg/mL) significantly improved discriminatory potency of the whole model for HF. Conclusion We found that decreased serum levels of irisin significantly predicted newly AF in patients with any phenotypes of HF.