These results demonstrate that striking differences in lung morphology, corresponding to different distributions of gas within the lungs, are observed in patients whose respiratory condition fulfills the definition criteria of ARDS.

In patients with ARDS, the cardiorespiratory effects of PEEP are affected by lung morphology rather than by the cause of the lung injury (primary versus secondary ARDS). The regional distribution of the loss of aeration and the type of atelectasis -- "mechanical" with a massive loss of lung volume, or "inflammatory" with a preservation of lung volume-- characterizing the lower lobes are the main determinants of the cardiorespiratory effects of PEEP.

The aim of this study was to assess positive end-expiratory pressure (PEEP)-induced lung overdistension and alveolar recruitment in six patients with acute lung injury (ALI) using a computed tomographic (CT) scan method. Lung overdistension was first determined in six healthy volunteers in whom CT sections were obtained at FRC and at TLC with a positive airway pressure of 30 cm H2O. In patients, lung volumes were quantified by the analysis of the frequency distribution of CT numbers on the entire lung at zero end-expiratory pressure (ZEEP) and PEEP. In healthy volunteers at FRC, the distribution of the density histograms was monophasic with a peak at -791 +/- 12 Hounsfield units (HU). The lowest CT number observed was -912 HU. At TLC, lung volume increased by 79 +/- 35% and the peak CT number decreased to -886 +/- 26 HU. More than 70% of the increase in lung volume was located below -900 HU, suggesting that this value can be considered as the threshold separating normal aeration from overdistension. In patients with ALI, at ZEEP the distribution of density histograms was either monophasic (n = 3) or biphasic (n = 3). The mean CT number was -319 +/- 34 HU. At PEEP 13 +/- 3 cm H2O, lung volume increased by 47 +/- 19% whereas mean CT number decreased to -538 +/- 171 HU. PEEP induced a mean alveolar recruitment of 320 +/- 160 ml and a mean lung overdistension of 238 +/- 320 ml. In conclusion, overdistended lung parenchyma of healthy volunteers is characterized by a CT number below -900 HU. This threshold can be used in patients with ALI for differentiating PEEP-induced alveolar recruitment from lung overdistension.