Research

Alveolar recruitment can be predicted from airway pressure-lung volume loops: an experimental study in a porcine acute lung injury model

Jacob Koefoed-Nielsen¹ , Niels Dahlsgaard Nielsen¹ , Anders J Kjærgaard² and Anders Larsson¹

¹Department of Anesthesia and Intensive Care, Aarhus University Hospital, Aalborg, Hobrovej 18-22, DK-9000 Aalborg, Denmark

²Department of Anesthesia and Intensive Care, Aarhus University Hospital, Århus, Norrebrogade 44, DK-8000 Århus, Denmark

author email corresponding author email

Critical Care 2008, 12:R7doi:10.1186/cc6771

Published:	21 January 2008

See related commentary by Kuhlen, http://ccforum.com/content/12/2/125

Abstract

Introduction

Simple methods to predict the effect of lung recruitment maneuvers (LRMs) in acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) are lacking. It has previously been found that a static pressure–volume (PV) loop could indicate the increase in lung volume induced by positive end-expiratory pressure (PEEP) in ARDS. The purpose of this study was to test the hypothesis that in ALI (1) the difference in lung volume (ΔV) at a specific airway pressure (10 cmH₂O was chosen in this test) obtained from the limbs of a PV loop agree with the increase in end-expiratory lung volume (ΔEELV) by an LRM at a specific PEEP (10 cmH₂O), and (2) the maximal relative vertical (volume) difference between the limbs (maximal hysteresis/total lung capacity (MH/TLC)) could predict the changes in respiratory compliance (Crs), EELV and partial pressures of arterial O₂ and CO₂ (PaO₂ and PaCO₂, respectively) by an LRM.

Methods

In eight ventilated pigs PV loops were obtained (1) before lung injury, (2) after lung injury induced by lung lavage, and (3) after additional injurious ventilation. ΔV and MH/TLC were determined from the PV loops. At all stages Crs, EELV, PaCO₂ and PaO₂ were registered at 0 cmH₂O and at 10 cmH₂O before and after LRM, and ΔEELV was calculated. Statistics: Wilcoxon's signed rank, Pearson's product moment correlation, Bland–Altman plot, and receiver operating characteristics curve. Medians and 25th and 75th centiles are reported.

Results

ΔV was 270 (220, 320) ml and ΔEELV was 227 (177, 306) ml (P < 0.047). The bias was 39 ml and the limits of agreement were – 49 ml to +127 ml. The R² for relative changes in EELV, Crs, PaCO₂ and PaO₂ against MH/TLC were 0.55, 0.57, 0.36 and 0.05, respectively. The sensitivity and specificity for MH/TLC of 0.3 to predict improvement (>75th centile of what was found in uninjured lungs) were for EELV 1.0 and 0.85, Crs 0.88 and 1.0, PaCO₂ 0.78 and 0.60, and PaO₂ 1.0 and 0.69.

Conclusion

A PV-loop-derived parameter, MH/TLC of 0.3, predicted changes in lung mechanics better than changes in gas exchange in this lung injury model.