© 2004 American Thoracic Society
Biotrauma Hypothesis of Ventilator-induced Lung InjuryTo the Editor:In a recent perspective, Dreyfuss and colleagues (1) question the existence and role of cytokines during ventilator-induced lung injury (VILI), in agreement with their only published work in the field of mediators and biotrauma, in which they failed to confirm some of our findings (see Reference 25 in Dreyfuss and colleagues). Our criticism of their perspective is not only that it focused on many trivial so-called inconsistencies within and between studies, but that it did so without a conceptual framework. Provision for the following three concepts would have clarified many misconceptions.
First, Dreyfuss and colleagues (1) do not systematically distinguish between the two independent parts of the biotrauma hypothesis: (1) that ventilation may cause release of mediators, and (2) that these mediators have biological actions. Because the biotrauma hypothesis is rather new, until now most work has been concerned with the first part of this hypothesis and ample evidence (the 2003 American Thoracic Society International Conference was abound in such studies) exists to support that ventilation or stretch can activate cells and promote mediator release. Differences among studies or nonzero baseline levels in control subjects do not invalidate this general notion, and the problem now is not whether, but how ventilation causes mediator release. The arguably more important question concerns the biological role of these mediators. This problem has only recently begun to be addressed, and yet there are already at least three studies in healthy animals where ventilation-induced lung injury was apparently not caused directly by the mechanical forces themselves, but by the mediators produced in response to these forces (2, 3, Reference 50 in Dreyfuss and colleagues). One of these studies, showing the pivotal role of macrophage inflammatory protein (MIP)-2
Second, the assessment that this is "not surprising" is based on the second conceptual deficit of this review, namely that it does not systematically distinguish between one-hit and two-hit models. Almost all of the work before 1996 used two-hit models, typically the lung lavage model followed by different ventilation strategies. Increased cytokine levels caused by injurious ventilatory strategies have been observed in a number of two-hit models, but the interaction between the first (e.g., lavage or acid aspiration) and the second hit (ventilation) is very complex, and the initial stimulus for mediator release and leukocyte activation is difficult to define. Therefore, investigators have started to investigate the effects of ventilation in healthy lungs (one-hit model). And whereas in the one-hit models it has been difficult to always detect tumor necrosis factor (TNF; which is overemphasized by Dreyfuss and colleagues), many other mediators were frequently found such as interleukin (IL)-6, MIP-2 Third, ventilation can cause mediator release by at least four fundamentally different mechanisms: necrosis, decompartmentalization, stretch, or shear stress (5). Unfortunately, Dreyfuss and colleagues (1) confuse these mechanisms. For instance, they mistake the model of ventilation-induced mediator release in uninjured perfused mouse lungs (References 24 and 30 in Dreyfuss and colleagues) for a model of mediator release caused by decompartmentalization (see p. 1,468 in Reference 1). The difference between both models is illustrated by the antithetic effects of surfactant: In the perfused mouse lung model where lung integrity is maintained, surfactant treatment (most likely because it improves stretch) augments mediator release (6); whereas in a model of decompartmentalization surfactant has the opposite effect and reduces mediator release (7). The notion that mediator levels can increase in response to ventilation-induced cell necrosis and decompartmentalization does not exclude the possibility that under less injurious ventilation strategies, stretching of the lung can activate specific mechanotransduction pathways (5).
Thus, most of the so-called inconsistencies could have been reconciled by systematic application of a conceptual framework. In addition, many other arguments in this perspective appear to be misconstrued. For instance, the critique of the work of Stüber and colleagues (Reference 60 in Dreyfuss and colleagues) reads, "It is also difficult to understand why changes in concentration occurred so rapidly ...," thus suggesting—without any further argumentation—that there is something wrong with the data. This sort of argument should not be used in a scientific paper—it betrays the authors' biases. We also fail to see why Dreyfuss and colleagues believe that the additive effect between stretch and lipopolysaccharide (LPS) in the work of Tremblay and colleagues (Reference 23 in Dreyfuss and colleagues) is difficult to explain, and even if so, what this would prove? One possible explanation is that both phenomena act via a common mechanism (e.g., via NF- An unbiased, critical analysis of the biotrauma hypothesis is certainly warranted and could provide novel insights and testable hypotheses. Regrettably, Dreyfuss and colleagues (1) have missed this opportunity. Their perspective finishes with a quote from Molière's comedy The Doctor Despite Himself. Unfortunately, even in this context, the literature is misinterpreted. The citation with respect to the "humeurs peccante" (sinful humors) referred to in their perspective was not, as they implied, the belief of a doctor, but rather the beliefs of a drunken lumberjack who was impersonating a physician.
a Research Center Borstel Borstel, Germany FOOTNOTES Conflict of Interest Statement: S.U. has no declared conflict of interest. M.R. has been reimbursed for attending several meetings organized and financed by various ventilator and pharmaceutical companies (Siemens-Elema, Puritan Bennett, Dräger, GlaxoSmithKline, AstraZeneca, Pharmacia, Eli-Lilly) and received honoraria for presentations, he serves as a consultant to Siemens-Elema, and Dreger, and has received grants from Siemens-Elema, Puritan-Bennett, Dreger, and Sensormedics. A.S.S. is member of advisory boards to Siemens-Elema and to Leo Pharmaceuticals. REFERENCES
From the Authors: The sympathetic enthusiasm of Uhlig, Ranieri, and Slutsky makes them forget basic scientific rules. They believe that they have elaborated a theory linking ventilation, cytokine release, and multiple organ system failure. They do not realize, however, that most of the available data (1) are not suitable to generate what common wisdom calls a scientific theory. Science requires that observations be repeatable and repeated. The best reproducibility of all experiments dealing with cytokine release was achieved by Dr. Uhlig, who published quite similar experiments in two different journals (2, 3). (We wonder why he did not cite the former (2) to support the latter (3) published in AJRCCM.) These two articles demonstrate, among other results, a consistent stimulating finding: that normal tidal volume ventilation leads to considerable release of inflammatory mediators. This breakthrough has now been ignored by its discoverers. Uhlig, Ranieri, and Slutsky are less cautious than Belperio and coworkers (4) who wrote, "in vivo neutralization of murine CXCR2 attenuates VILI"—not "prevented." They dismiss results of a clinical study performed by Stuber and coworkers (5), which contradicts a study by Ranieri, Slutsky, and colleagues (6). Their study (6) suggests that injurious ventilation produces inflammation; in contrast, the former study (5) rather inclines toward antiinflammation (7). They contend that one of our sentences at that point should not be used in a scientific article, and it betrays our supposed bias. We, however, simply quoted the very words of Pugin in the editorial (7) that accompanied the article of Stuber and coworkers (5). They explain that "ventilation can cause mediator release by at least four fundamentally different mechanisms ...," listing in complete confusion primary and secondary mechanisms of cytokine release. Sganarelle (a personage inspired by the Comedia dell'arte) is a churl who makes bundles of firewood to pay his wine and who was the servant of a physician for six years (this is why he knows medical jargon) (8). "Humeurs peccantes" is a medical term that was widely used, both before (9) and during Molière's time. Furetière, in his 1690 Dictionnary (the reference for the time of Louis XIV), wrote "all diseases are caused only by ‘humeurs peccantes.’" Uhlig, Ranieri, and Slutsky can also learn from looking at a fourteenth century picture of the University of Bologna Library that shows a bloodletting performed to evacuate "materia peccante" (http://www.areafit.it/articoli/cuore.htm). We hope that Uhlig, Ranieri, and Slutsky's search and understanding of recent scientific literature will become more accurate than their review of the classics. Last, but not least, they seem to consider that our publishing of only one article on this aspect of ventilation-induced lung injury does not qualify us for writing this Perspective. That criticism is more directed at the editorial process of AJRCCM.
Faculté Xavier Bichat Paris, France Service de Réanimation Hôpital Louis Mourier Assistance Publique-Hôpitaux de Paris Colombes, France FOOTNOTES Conflict of Interest Statement: D.D., J-D.R., and G.S. have no declared conflict of interest. REFERENCES
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
receptors (CXCR2) in a murine VILI model (Reference 50 in Dreyfuss and colleagues) is dismissed by Dreyfuss and colleagues (
B) that becomes saturated at high levels of activation. Furthermore, in contrast to what is insinuated, mediator release in response to overventilation cannot be explained by LPS contamination, because it also occurs in LPS-resistant animals (Reference 30 in Dreyfuss and colleagues). Another putative discrepancy is that an IL-1 receptor antagonist was partially protective, whereas there was no significant increase in IL-1 levels (Reference 43 in Dreyfuss and colleagues). Possible explanations for this finding that are not mentioned include incorrect timing of the IL-1 measurement and the well known fact that the level of cytokines that can be measured in vivo may only represent the tip of the iceberg (
