Prediction of the chemoreflex gain by common clinical variables in heart failure

Background Peripheral and central chemoreflex sensitivity, assessed by the hypoxic or hypercapnic ventilatory response (HVR and HCVR, respectively), is enhanced in heart failure (HF) patients, is involved in the pathophysiology of the disease, and is under investigation as a potential therapeutic target. Chemoreflex sensitivity assessment is however demanding and, therefore, not easily applicable in the clinical setting. We aimed at evaluating whether common clinical variables, broadly obtained by routine clinical and instrumental evaluation, could predict increased HVR and HCVR. Methods and results 191 patients with systolic HF (left ventricular ejection fraction-LVEF-<50%) underwent chemoreflex assessment by rebreathing technique to assess HVR and HCVR. All patients underwent clinical and neurohormonal evaluation, comprising: echocardiogram, cardiopulmonary exercise test (CPET), daytime cardiorespiratory monitoring for breathing pattern evaluation. Regarding HVR, multivariate penalized logistic regression, Bayesian Model Averaging (BMA) logistic regression and random forest analysis identified, as predictors, the presence of periodic breathing and increased slope of the relation between ventilation and carbon dioxide production (VE/VCO2) during exercise. Again, the above-mentioned statistical tools identified as HCVR predictors plasma levels of N-terminal fragment of proBNP and VE/VCO2 slope. Conclusions In HF patients, the simple assessment of breathing pattern, alongside with ventilatory efficiency during exercise and natriuretic peptides levels identifies a subset of patients presenting with increased chemoreflex sensitivity to either hypoxia or hypercapnia.