Reply to Fisher and Beers: Animal models of acute lung injury

Abstract

reply: We thank Drs. Fisher and Beers (3) for their comments, which support the key statements made in the hyperoxia section of our manuscript (6), namely that there is no conclusive evidence (level A or B) showing that, in humans, exposure to 100% oxygen under normobaric conditions results in acute lung injury/acute respiratory distress syndrome (ALI/ARDS); that, in contrast, in most animal studies, exposure to 100% oxygen consistently results in lethal lung injury; and that, therefore, there are important differences in susceptibility to oxygen in humans and other mammalian species. We did not intend to imply that hyperoxia is harmless in humans or that clinicians should not pay attention to the FiO2 in ventilated patients. We are simply highlighting the fact that although exposure of normal animals to 100% oxygen under normobaric conditions reproducibly causes lung injury, the same is not true in humans, and therefore caution should be used when extrapolating findings from the hyperoxic model to humans, particularly when investigating putative causes of ALI/ARDS in previously uninjured lungs. However, as we discuss in the review, the hyperoxic animal model has many advantages, including reproducibility, and it remains an excellent model in which to study lung repair or the contributions of hyperoxia to preexisting lung injury. Furthermore, the hyperoxia model appears to be more physiological than other ALI models such as intratracheal bleomycin. As Fisher and Beers (3) point out, our statements about hyperoxia and human ALI/ARDS are partly based on a study by Barber and Hamilton (1) in which patients with brain injury were exposed to 100% oxygen but failed to develop ALI. We believe that this negative study is important because it clearly demonstrates that normobaric hyperoxia is not necessarily followed by lung injury in humans. The Kapanci et al. (5) paper that according to Fisher and Beers (3) “demonstrated significant lung injury” in response to oxygen is more difficult to interpret because other factors now known to be associated with lung injury, such as the tidal volume, were not reported. For the same reasons, it is difficult to conclude that Hyde and Rawson's study (4) demonstrated that “exposure of patients with normal lungs to prolonged hyperoxia resulted in clinical findings compatible with oxygen toxicity.” The single patient who died in that study was found disconnected from the ventilator, making it unclear as to whether his death was truly “pulmonary-related.” As mentioned by Fisher and Beers (3), humans exposed to normobaric hyperoxia develop tracheobronchitis and mild changes in permeability but not ALI (2). Thus the existing literature does not show that exposure of humans to 100% oxygen under normobaric conditions results in ALI/ARDS. We agree with Fisher and Beers (3) that the difference in oxygen susceptibility seen in humans and other mammalian species likely reflects inheritable variability at the genetic level, and it remains possible that subgroups of people exist who have increased susceptibility to oxygen-induced lung injury. Thus we do not believe that our published paper has “misleading and potentially dangerous statements concerning the role of hyperoxia in human acute lung injury.” Although we agree that “diligence is necessary when humans are exposed to elevated partial pressures of O2,” we also believe that it is important to be cautious when clinical decisions are based on data generated primarily from animal studies.