Message in a Model

Brett A. Simon

Department of Anesthesiology and Critical Care Medicine, Johns Hopkins UniversityBaltimore, Maryland

	ARTICLE

TOP ARTICLE REFERENCES

In the past several years, research into ventilator-induced lung injury has expanded from primarily mechanical and supportive issues into the cellular and molecular basis of the disease process. At the forefront of this development was a study by Tremblay and colleagues in which high levels of inflammatory cytokines were measured in lavage fluid from isolated, nonperfused rat lungs subjected to high tidal volume (VT), low positive end-expiratory pressure (PEEP) ventilation (1). The prospect that inflammatory mediators produced by injurious ventilation of the lung could enter the systemic circulation provided a potential mechanism that could account for the observation that most patients who die from acute respiratory distress syndrome succumb to multiple system organ failure rather than to respiratory failure (2). Since that time there has been an explosion of studies investigating complex relationships between a multitude of lung injury models, ventilatory strategies, inflammatory and vasoactive mediators, and cellular and molecular responses, which have rekindled hopes for pharmacologic, immunologic, and/or gene therapies to improve outcomes for patients with these deadly syndromes (3).

In this issue of the Journal (pp. 1176-1180), Ricard and coworkers report an animal study in which injurious ventilatory strategies did not cause, what might now reasonably be expected, increases in inflammatory mediators in bronchoalveolar lavage fluid (BALF) or the systemic circulation (4). In the first part of this study, healthy, intact rats were subjected to high- and low-stretch mechanical ventilation such that severe pulmonary edema was produced. To the investigators' surprise, analysis of BALF showed no tumor necrosis factor (TNF)-, low levels of interleukin (IL)-1 that increased with VT, and moderate levels of the chemokine macrophage inflammatory protein-2 (MIP-2) which were independent of ventilatory strategy. The authors then went on to repeat the original isolated, nonperfused rat lung protocol of Tremblay and colleagues (1), both with and without endotoxin (lipopolysaccharide [LPS]) priming before lung harvest. In contrast to Tremblay and colleagues' results, they failed to demonstrate stretch-dependent increases in TNF-. IL-1 was present only in very low concentrations, although there were statistically significant increases in the high-stretch group. MIP-2 levels in BALF increased with stretch and were comparable to those found by Tremblay and colleagues. After LPS priming, BALF TNF- levels increased only slightly, but results for IL-1 and MIP-2 were comparable to those of Tremblay and colleagues. The authors suggest that the isolated, nonperfused lung model is unstable and poorly reproducible, and they conclude that mechanical ventilation with large VT alone (i.e., without some predisposing event such as LPS exposure or mechanical or chemical lung injury) is not sufficient to increase the release of proinflammatory cytokines into the airspaces.

While one could argue whether the differences or similarities in results between these two carefully designed and executed studies are more significant, it is worth first stepping back and considering the use of experimental models and the process by which scientific findings achieve validity and acceptance. By any account, the use of an isolated, nonperfused lung as a physiologic model is limited. In the absence of a prior positive observation, the failure to find a particular response in such a system would hardly be of interest. Certainly, the negative results reported here by Ricard and coworkers (4) would not have been important 5 yr ago because of lack of a context in which to place them. In first approaching a complex system, the use of a highly simplified model, such as the isolated, nonperfused lung, is justified in the tradeoff between control and relevance, eliminating complexity at the expense of external validity. Negative results in this setting have little predictive value. However, a positive finding, such as that of Tremblay and colleagues (1), reveals a world of possibilities. At a minimum, it serves to generate new hypotheses and to stimulate discussion, even if eventually it is disproved. The work ultimately achieves importance through the studies that follow, building on the original in more complex, relevant models.

The concept that high-stretch ventilatory patterns in injured lungs can promote the release of inflammatory mediators into the airspaces and systemic circulation is now strongly supported by evidence from experimental models ranging from mechanically stressed cell systems (5) to isolated lungs (6) and intact animals (7, 8) and is consistent with observations made in patients (9). Whether this holds true for normal lungs subjected to injurious ventilation is less clear. Ricard and colleagues noticed that in the work of Tremblay and colleagues there was little difference in the production of mediators with or without LPS pretreatment. They hypothesized that the untreated, LPS-negative controls of Tremblay and colleagues may have been inadvertently effectively "primed" with the equivalent of low-dose LPS, and thus Ricard's LPS group was pretreated with a much lower dose of LPS in order to simulate this effect. Since Ricard and colleagues' LPS-pretreated results most closely approximate those of Tremblay and colleagues' control animals, they may be correct. However, the question of whether unprimed or uninjured lungs release cytokines with injurious ventilation will not be answered in isolated lung preparations, particularly since, along these same lines, isolated lungs can never be truly considered "uninjured."

Given the subtleties and myriad details of the implementation of even "simple" experimental models, the question of why an expected result is not observed in a particular model may be as important as why it is observed in another. Ricard's study may serve to alert us to the effects of previously ignored contaminant levels of endotoxin in acute experimental preparations. Another potential confounder that is not widely appreciated is the use of heparin. Systemic heparin has been shown to improve oxygenation and to reduce lung edema, neutrophil sequestration, and systemic cytokine levels in several lung injury models (10) by modulating the inflammatory response. Although heparin is routinely used to prepare organs for ex vivo study, and was used in similar manner by Ricard and Tremblay and their coworkers, perhaps it has a role in the poor reproducibility of the isolated lung model. Thus, not only do complex models build on results from more simplified models, but the reverse may also be true. We must continue to scrutinize and reappraise these exciting works in progress to insure that we receive both the message in the results and the message in the model.

	References

TOP ARTICLE REFERENCES

1. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventilatory strategies increase cytokines and c-fos m-RNA expression in an isolated rat lung model. J Clin Invest 1997; 99: 944-952 [Medline].

2. Montgomery AB, Stager MA, Carrico CJ, Hudson LD. Causes of mortality in patients with adult respiratory distress syndrome. Am Rev Respir Dis 1985; 132: 485-489 [Medline].

3. Matthay MA. Conference summary: acute lung injury. Chest 1999;116 (Suppl. 1):119S-126S.

4. Ricard J, Dreyfuss D, Saumon G. Production of inflammatory cytokines in ventilator-induced lung injury: a reappraisal. Am J Respir Crit Care Med 2001; 163: 1176-1180 [Abstract/Free Full Text].

5. Vlahakis NE, Schroeder MA, Limper AH, Hubmayr RD. Stretch induces cytokine release by alveolar epithelial cells in vitro. Am J Physiol 1999;277(1 Pt. 1):L167-L173.

6. von Bethmann AN, Brasch F, Nusing R, Vogt K, Volk HD, Muller KM, Wendel A, Uhlig S. Hyperventilation induces release of cytokines from perfused mouse lung. Am J Respir Crit Care Med 1998; 157: 263-272 [Abstract/Free Full Text].

7. Chiumello D, Pristine G, Slutsky AS. Mechanical ventilation affects local and systemic cytokines in an animal model of acute respiratory distress syndrome. Am J Respir Crit Care Med 1999; 160: 109-116 [Abstract/Free Full Text].

8. Imai Y, Kawano T, Iwamoto S, Nakagawa S, Takata M, Miyasaka K. Intratracheal anti-tumor necrosis factor-alpha antibody attenuates ventilator-induced lung injury in rabbits. J Appl Physiol 1999; 87: 510-515 [Abstract/Free Full Text].

9. Ranieri VM, Suter PM, Tortorella C, De Tullio R, Dayer JM, Brienza A, Bruno F, Slutsky AS. Effect of mechanical ventilation on inflammatory mediators in patients with acute respiratory distress syndrome: a randomized controlled trial. JAMA 1999; 282: 54-61 [Abstract/Free Full Text].

10. Abubakar K, Schmidt B, Monkman S, Webber C, de SD, Roberts R. Heparin improves gas exchange during experimental acute lung injury in newborn piglets. Am J Respir Crit Care Med 1998;158(5, Pt. 1): 1620-1625.

11. Cox CS Jr,, Zwischenberger JB, Traber DL, Traber LD, Haque AK, Herndon DN. Heparin improves oxygenation and minimized barotrauma after severe smoke inhalation in an ovine model. Surg Gynecol Obstet 1993; 176: 339-349 [Medline].

12. Darien BJ, Fareed J, Centgraf KS, Hart AP, MacWilliams PS, Clayton MK, Wolf H, Kruse-Elliott KT. Low molecular weight heparin prevents the pulmonary hemodynamic and pathomorphologic effects of endotoxin in a porcine acute lung injury model. Shock 1998; 9: 274-281 [Medline].