Feasibility Study of Noninvasive Ventilation With Helium-Oxygen Gas Flow for Chronic Obstructive Pulmonary Disease During Exercise

BACKGROUND: Individually, noninvasive ventilation (NIV) and helium-oxygen gas mixtures (heliox) diminish ventilatory workload and improve exercise tolerance in patients with chronic obstructive pulmonary disease (COPD). NIV in combination with heliox may have additive effects on exercise tolerance in severe COPD. METHODS: We assessed the safety, tolerability, and efficacy of heliox and NIV during exercise in patients with severe COPD. SETTING: Pulmonary rehabilitation facility in an academic tertiary-care medical center. PROTOCOL: Twelve patients with severe COPI) were enrolled. Using a sequential randomized placebo-controlled crossover study design, the patients performed 4 separate constant-work stationary bicycle cardiopulmonary exercise studies at 80% of maximal workload during application of sham NIV, NIV, 60:40 heliox with sham NIV, and 60:40 heliox with NIV. Tolerability, safety, and exercise duration as determined by constant-work cardiopulmonary exercise test were the primary outcome measures. Secondary outcome measures at peak exercise and iso-time included rate of perceived exertion, dyspnea, leg pain, heart rate, respiratory rate, systolic and diastolic blood pressure, tympanic temperature, and oxyhemoglobin saturation. RESULTS: No adverse effects occurred during or after application of NIV, heliox, or NIV with heliox. Exercise duration using heliox with NIV was significantly longer than both heliox (P = .01) and NIV (P = .007), but not placebo (P = .09). Relative to placebo, all treatment arms permitted lower respiratory rates at peak exercise. Heliox, with or without NIV, was associated with significant improvements in oxyhemoglobin saturation at peak exercise, relative to placebo or NIV alone. CONCLUSIONS: The adjunctive use of NIV with heliox during exercise proved both safe and tolerable in patients with severe COPD. The lack of demonstrable efficacy to any of the treatment arms relative to placebo (P = .09) may be the result of the small sample size (ie, type 2 error)-a conclusion emphasized by the large standard deviations and differences in treatment group variances in exercise duration alone.