A Strategy to Enhance Confidence in Respiratory Chemoreflex Characterization by Modified Rebreathing

Abstract

Modelling breath‐by‐breath ventilation (V̇E) in response to carbon dioxide rebreathing in hyperoxic and hypoxic conditions (modified rebreathing) is used to characterize the central and peripheral respiratory chemoreflexes in humans. However, breath‐to‐breath variability produces uncertainty in the estimation of model parameters. We explored whether signal‐averaging of repeated rebreathing trials would improve chemoreflex sensitivity parameter estimation and confidence. Seven healthy males (age: 27±5 years) performed 6 repetitions of modified rebreathing tests in isoxic‐hypoxic (end‐tidal PO2 (PETO2)=50 mmHg) and ‐hyperoxic conditions (PETO2=150 mmHg) over the course of 4 days. End‐tidal PCO2 (PETCO2), PETO2 and V̇E were measured breath‐by‐breath by dual gas analyser and pneumotach. The slope of the V̇E vs PETCO2 relationship above the PETCO2 threshold in the hyperoxic test gave the central chemoreflex sensitivity (L∙min‐1∙mmHg‐1) and the difference in hypoxic versus hyperoxic test slopes provided peripheral chemoreflex sensitivity. Chemoreflex slopes modelled from a single trial versus ensemble‐average of 6 trials were compared by paired t‐test and effect size (Cohen’s d). Breath‐by‐breath data from the 6 trials were aligned to the same starting PETCO2, linearly interpolated to 0.1 mmHg of PETCO2, ensemble‐averaged, and averaged into 0.2 mmHg bins. Central chemoreflex sensitivity did not differ (p=0.94) between single (4.10±2.98 L∙min‐1∙mmHg‐1, 95%CI: 3.82–4.39) versus ensemble‐averaged data (4.13±2.14 L∙min‐1∙mmHg‐1, 95%CI: 3.99–4.27) but the 95%CI was reduced (p=0.031) by 0.29 L∙min‐1∙mmHg‐1 (effect size=1.5) with the ensemble‐averaged approach. Similarly, peripheral chemoreflex sensitivity did not differ (p=0.83) between single (1.69±1.14 L∙min‐1∙mmHg‐1, 95%CI: 1.54–1.84) versus ensemble‐averaged data (1.58±1.32 L∙min‐1∙mmHg‐1, 95%CI: 1.5 –1.66); 95%CI of the ensemble‐averaged data was 0.15 L∙min‐1∙mmHg‐1 lower than a single trial but this was not different (p=0.38; effect size=0.5). Signal averaging of multiple modified rebreathing trials reduces the confidence interval of respiratory chemoreflex sensitivity. This strategy could enhance the ability to detect differences in respiratory chemoreflex control in comparative or interventional studies where modified rebreathing is used.