Single-breath CO2 analysis as a predictor of lung volume change in a model of acute lung injury.

OBJECTIVE: To examine the utility of single-breath CO2 analysis as a measure of lung volume change in a model of acute lung injury.

SETTING: Animal laboratory in a university-affiliated medical center.

DESIGN: Prospective, animal cohort study comparing 21 variables derived from single-breath CO2 analysis with lung volume measurements determined by nitrogen washout.

SUBJECTS: Seven lambs with saline lavage-induced acute lung injury.

METHODS: Animals were treated with repetitive saline lavage to achieve a uniform degree of acute lung injury (PaO2 < 100 torr [13.32 kPa] on FiO2 of 1.0). Twenty-one derived components of the CO2 expirogram were evaluated as predictors of lung volume change. Lung volume was manipulated by 3-cm H2O incremental increases in positive end-expiratory pressure from 0 to 21 cm H2O and ranged between 90 and 765 mL.

MEASUREMENTS AND MAIN RESULTS: Fifty-five measurements of lung volume were available for comparison with derived variables from the CO2 expirogram. Stepwise linear regression identified five variables that were most predictive of lung volume change: a) dynamic lung compliance; b) the slope of phase III; c) the slope of phase II divided by the mixed expired CO2 concentration; d) airway deadspace; and e) PaO2/FIO2 ratio. The multivariate equation was highly statistically significant and explained 94% of the variance (adjusted r2 = .94, p < .0001). The bias and precision of the calculated lung volume were 10.9 and 55.9, respectively. The mean percentage difference for the lung volume estimate derived from the single-breath CO2 analysis station was 3.3%.

CONCLUSIONS: Our data indicate that analysis of the CO2 expirogram can yield accurate information about lung volume in animals with saline lavage-induced acute lung injury. Specifically, five variables derived from a plot of expired CO2 concentration vs. expired volume predict changes in lung volume in healthy lambs with an adjusted coefficient of determination of 0.94. We hope to further define the utility of this technique by prospective application of this methodology in the clinical setting.