Loop gain as a means to predict a positive airway pressure suppression of Cheyne-Stokes respiration in patients with heart failure.

Abstract

Patients with heart failure (HF) and Cheyne-Stokes respiration or periodic breathing (PB) often demonstrate improved cardiac function when treatment with continuous positive airway pressure (CPAP) resolves PB. Unfortunately, CPAP is successful in only 50% of patients, and no known factor predicts responders to treatment. Because PB manifests from a hypersensitive ventilatory feedback loop (elevated loop gain [LG]), we hypothesized that PB persists on CPAP when LG far exceeds the critical threshold for stable ventilation (LG = 1). To derive, validate, and test the clinical utility of a mathematically precise method that quantifies LG from the cyclic pattern of PB, where LG = 2π/(2πDR - sin2πDR) and DR (i.e., duty ratio) = (ventilatory duration)/(cycle duration) of PB. After validation in a mathematical model of HF, we tested whether our estimate of LG changes with CPAP (n = 6) and inspired oxygen (n = 5) as predicted by theory in an animal model of PB. As a first test in patients with HF (n = 14), we examined whether LG predicts the first-night CPAP suppression of PB. In lambs, as predicted by theory, LG fell as lung volume increased with CPAP (slope = 0.9 ± 0.1; R(2) = 0.82; P < 0.001) and as inspired-arterial PO(2) difference declined (slope = 1.05 ± 0.12; R(2) = 0.75; P < 0.001). In patients with HF, LG was markedly greater in 8 CPAP nonresponders versus 6 responders (1.29 ± 0.04 versus 1.10 ± 0.01; P < 0.001); LG predicted CPAP suppression of PB in 13/14 patients. Our novel LG estimate enables quantification of the severity of ventilatory instability underlying PB, making possible a priori selection of patients whose PB is immediately treatable with CPAP therapy.