Respiratory cycle-related EEG changes during sleep reflect esophageal pressures

Abstract

Study objectives: Respiratory cycle-related EEG changes (RCREC) have been demonstrated during sleep by digital analysis and hypothesized to represent subtle inspiratory microarousals that may help to explain daytime sleepiness in patients with sleep-disordered breathing. We therefore examined for the first time associations between RCREC and esophageal pressure swings (deltaPes) that reflect work of breathing.

Design: Retrospective analysis.

Setting: Academic sleep laboratory.

Patients: Forty adults referred for suspected sleep disordered breathing.

Interventions: Polysomnography with esophageal pressure monitoring and automatic computation of deltaPes using a novel algorithm.

Results: Computed deltaPes for nearly all respiratory cycles during sleep correlated well with visual scoring of selected respiratory cycle samples (Spearman rho = 0.86, P < 0.0001). The RCREC within the sigma EEG range (12.5-15.5 Hz) rather than that within other frequency ranges most often showed significant within-subject inverse correlations with deltaPes. In contrast, in between-subject comparisons, beta (15.5-30.5 Hz) and to a lesser extent theta (4.5-7.5 Hz) RCREC, rather than sigma RCREC, showed significant inverse associations with mean APes.

Conclusions: Variation within subjects of sigma RCREC with APes supports previous evidence that RCREC within this range may reflect microarousals exacerbated by increased work of breathing. Correlation of beta and theta, but not sigma RCREC with deltaPes in between-subject comparisons is more difficult to explain but suggests that ranges other than sigma also deserve further investigation for clinical utility.

Figures

Figure 1

The mean magnitude of esophageal pressure swings (expressed in cm of water pressure) recorded during sleep, as calculated by a computer, is plotted against results of sampling by human eye (Spearman rho = 0.86, P < 0.0001).

Figure 2

A Bland and Altman plot shows that esophageal pressure changes with each respiratory cycle, as computed automatically by a newly developed algorithm, provided a reasonably accurate measure as compared to human scoring of representative respiratory cycles. Bias was small, with the human measure proving only about 4 cm of water pressure more negative than the automated measure on average.

Figure 3

The respiratory cycle-related EEG change (RCREC) calculation and associated data are shown for a single respiration cycle. The calculation measures the variation of the EEG signal power for a specific frequency band within a single respiratory cycle. The process begins by applying a digital lowpass filter to the measured esophageal pressure (A) to produce a respiration signal with reduced artifacts (B). The minimums and maximums of this signal are used to derive the ΔPes and to define 4 time segments corresponding to different intervals of the respiration cycle. A digital bandpass filter is applied to the measured EEG signal (C) to produce a time series corresponding to a specific frequency band. This signal is squared to produce a time series giving the variation of EEG power with time in the frequency band (D). The mean EEG power in the frequency band (from D) is then computed for each of the four respiration cycle time segments defined by the filtered respiration cycle (B). The mean power for each interval is then normalized by the mean frequency-specific power over the entire respiration cycle. One is subtracted from each result to get the measures shown for the 4 intervals in (E). The RCREC is then computed here as the difference between the maximum and minimum values. In practice, segment-specific EEG powers would be averaged over many respiratory cycles before the difference shown in (E) is computed.

Figure 4

The negative esophageal pressure swings (−ΔPes) are plotted along with the frequency-specific respiratory cycle-related EEG changes (RCREC) computed over an entire night of recording for one subject (21-year-old woman with apnea/hypopnea index = 5.2 and minimum oxygen saturation = 92%). Data were centered and scaled after median filtering. The tendency for the average trend of RCREC to vary with the magnitude of esophageal pressure swings is apparent in this subject.

Figure 5

Average computed esophageal pressure swings (ΔPes) during all sleep epochs is plotted against beta-frequency respiratory cycle-related EEG changes (RCREC) for each of 40 subjects.