Effect of Diabetes Mellitus on Endotoxin-Induced Lung Injury FREE

THE PULMONARY microvasculature is an important target of the systemic inflammatory response in critically ill patients. Increases in microvascular permeability and alterations in local perfusion result in hypoxemia, often necessitating mechanical ventilation and aggressive pulmonary support. Nearly all of the experimental observations regarding the lung's response to local and systemic inflammatory mediators have occurred in young, otherwise healthy animal models. Although this strategy has allowed a more focused examination of the physiological and immunological features of inflammation, it has ignored the fact that many critically ill patients have coexisting chronic illnesses that undoubtedly influence the patient's physiological response to inflammatory mediators.

Since the microvasculature is an important target of diabetes mellitus, we postulated that the pulmonary microvascular responsiveness of diabetic animals to systemic inflammation would be substantially different from that of nondiabetic animals. In preliminary studies that used an isolated, perfused lung model, we found the pulmonary microvasculature of diabetic animals to be significantly less responsive to thromboxane-induced increases in permeability than that of normal animals (unpublished data, 1999). The purpose of this study was to extend our observations regarding the effect of diabetes mellitus on pulmonary microvascular function to a more clinically relevant model of acute inflammatory lung injury. Specifically, this study examines the hypothesis that Zucker diabetic fatty (ZDF) rats are resistant to the effects of intratracheal endotoxin (lipopolysaccharide [LPS])on the extravasation of plasma proteins into the lungs.

Male ZDF rats (21-26 weeks of age; 400-600 g) were obtained from a colony at the Dallas Veterans Affairs Medical Center, Dallas, Tex. These animals possess a mutation of the leptin receptor that is phenotypically expressed as obesity, hypertriglyceridemia, and hyperglycemia by 7 to 9 weeks of age; as such, they have been commonly used as an animal model of type 2 diabetes mellitus.¹ Controls consisted of age-matched, lean, normoglycemic male heterozygotic rats.

These experiments were reviewed and approved by the Subcommittee on the Care and Use of Animals in Research at the University of Texas Southwestern Medical School and the Dallas Veterans Affairs Medical Center.

The ZDF and control animals were anesthetized with pentobarbital sodium (40 mg/kg of lean body weight, intraperitoneally) and the trachea was exposed through a small cervical incision. Escherichia coli LPS (100 or 200 µg in 250 µL of sterile isotonic sodium chloride solution; E coli serotype 0111:B4; Sigma-Aldrich Corp, St Louis, Mo) or an equal volume of sterile isotonic sodium chloride solution was injected rapidly into the trachea during inspiration by means of a 27-gauge needle. The wound was then closed and the animals were allowed to recover. This method induces an acute lung injury manifested by increased epithelial and microvascular permeability, neutrophil sequestration, and interstitial edema.^{2 - 8} Pulmonary microvascular leak (as assessed by Evans blue dye [EBD] extravasation) and neutrophil activation (CD11b/CD18 expression) and sequestration within the lung were assessed 4 hours after the instillation of LPS into the airways. This time point was chosen on the basis of preliminary experiments demonstrating maximal extravasation of plasma proteins into the lung at 4 to 8 hours after exposure to endotoxin. By 24 to 48 hours after injury, the microvascular leak is significantly attenuated, with concentrations of dye within the lungs being no different from that in controls (unpublished data, 1998).

Evans blue dye has been used to quantitate protein extravasation during acute inflammatory events. Previous studies^{9 - 10} have demonstrated that, after intravenous infusion, more than 99% of the dye exists bound to plasma proteins. This method has been shown to compare favorably with the use of radiolabeled albumin as a marker of plasma protein extravasation and, hence, microvascular "leakiness" during acute inflammatory events.¹⁰

Three hours after the administration of LPS or isotonic sodium chloride solution, EBD (20 mg/kg) was infused into the femoral vein of the ZDF and control animals (n = 5-8 per group). Sixty minutes later, the heart and lungs were excised and the pulmonary vasculature was flushed free of blood by gently infusing sterile saline into the beating right ventricle. The left lower lobe of the lung was then excised, weighed, and placed in 5 mL of formamide. The tissue was homogenized and incubated in formamide for 18 hours at 37°C. The tissue suspension was then centrifuged at 12 000 rpm for 20 minutes, after which the supernatant was collected and its optical density determined spectrophotometrically (OD₆₂₀). The amount of dye contained within the sample was extrapolated by means of a standard curve and expressed as milligrams of dye per gram of wet lung weight.

In a separate set of experiments, bronchoalveolar lavage (BAL) was performed 4 hours after exposure to LPS (200 µg) or an equivalent volume of sterile isotonic sodium chloride solution. In these experiments, 7 mL of sterile isotonic sodium chloride solution was instilled into the right lung 3 times. The lavage fluid was collected and pooled, and the total cell count was obtained with an automated cell counter. The number of neutrophils present within the lavage fluid was determined by light microscopy after Wright staining.

An aliquot of BAL fluid was collected to quantitate the expression of the adhesion glycoproteins CD11b and CD18 on neutrophils. In this experiment, neutrophils isolated from the BAL fluid were stained with fluorescein isothiocyanate–conjugated mouse anti–rat monoclonal antibodies against CD11b and CD18 (Pharmigen Inc, San Diego, Calif). An isotypic IgG was used as a control. Cell staining was quantitated by fluorescence-activated cell analysis sorting (FACSTAR; Beckton-Dickinson Co, San Jose, Calif) and expressed as the percentage of cells stained by the fluorescein isothiocyanate–labeled antibodies. A minimum of 10 000 cells were examined per sample.

All data are expressed as mean ± SEM and analyzed by analysis of variance with a Fisher post hoc test. A P value of less than .05 was considered a statistically significant difference between groups.

Extravasation of EBD into the lungs of the ZDF and control animals exposed to 0, 100, or 200 µg of LPS is shown in Figure 1. In the absence of LPS, the amount of EBD within the lungs of diabetic animals was about 33% less than that of nondiabetic controls (P = .002). The injection of LPS into the trachea of the control animals resulted in a dose-dependent increase in the extravasation of EBD into the lungs, with a 32% increase for the 200-µg dose (P = .008). In contrast, the injection of even 200 µg of LPS into the airways of the ZDF rats had no significant effect on the concentration of EBD within the lungs when compared with that of ZDF animals exposed to isotonic sodium chloride solution alone. The concentration of EBD within the lungs of control animals exposed to 200 µg of LPS was more than twice that of similarly treated ZDF animals (P<.001).

The effect of the ZDF phenotype on LPS-induced neutrophil sequestration within the airways is shown in Figure 2. There were minimal numbers of neutrophils within the airways of either the ZDF or control animals exposed to isotonic sodium chloride solution alone, whereas the injection of 200 µg of LPS into the airways resulted in a large influx of neutrophils. There was no difference in the numbers of neutrophils within the BAL fluid of control or ZDF animals exposed to either isotonic sodium chloride solution or LPS.

The effect of the ZDF phenotype on LPS-induced neutrophil CD11b expression is shown in Figure 3. In the absence of LPS stimulation, about 10% of the neutrophils isolated from the airways expressed CD11b. Four hours after exposure to intratracheal LPS, about 70% of the neutrophils isolated from BAL fluid expressed CD11b (P<.001 vs neutrophils isolated from animals exposed to isotonic sodium chloride solution vehicle alone). There was no difference in the expression of CD11b in response to either intratracheal isotonic sodium chloride solution or LPS on comparison of the ZDF and control groups. Measurements of neutrophil CD18 expression mirrored that of CD11b (data not shown).

The frequency of type 2 diabetes mellitus in the population in general, and in hospitalized patients in particular, makes it imperative that clinicians have a better understanding of the effects of diabetes on the local and systemic inflammatory response to injury, proinflammatory mediators, and infection. The purpose of this study was to examine the effect of diabetes mellitus on the pulmonary microvascular response to intratracheal endotoxin exposure. The results of this study suggest that the pulmonary microvasculature of diabetic animals is significantly less permeable to plasma proteins both at baseline and in response to intratracheal LPS than is that of normoglycemic controls. This difference in microvascular protein extravasation occurs despite similar degrees of neutrophil adhesion molecule expression and neutrophil sequestration within the lungs of the diabetic and nondiabetic groups.

The intratracheal instillation of LPS into rats is a commonly used model of acute inflammatory lung injury, and, as such, it has been used to investigate the pathophysiological characteristics of acute respiratory distress syndrome.^{2 - 5} This experimental model is associated with hypoxemia, impaired surfactant production, neutrophil sequestration within the lungs, and pulmonary edema with parenchymal consolidation.^{2 - 3 ,6 - 8} The increased microvascular and epithelial permeability associated with this model has been at least temporally related to the recruitment of neutrophils into the lung^{3 ,6 - 7} and the release of proinflammatory mediators including cytokines,^{11 - 13} superoxide radical,⁷ and nitric oxide.⁶ The precise role that each of these proinflammatory mediators plays in the pathophysiological mechanism of this injury model remains to be defined. Furthermore, the impact of diabetes on the generation and activity of these mediators, particularly within the lung, is unknown.

Although in most tissues diabetes appears to enhance the loss of protein from the vascular space,^{14 - 17} several investigators have reported that the opposite is true in the lungs.^{18 - 19} The mechanism by which this may occur is unknown. Diabetes mellitus is associated with fundamental changes in the composition and structure of the microvascular barrier and interstitium that may affect the movement of macromolecules from the vascular space into the interstitium. For example, van den Born et al¹⁸ found that the heparan sulfate content within the blood vessels of diabetic animals was inversely correlated with the extravasation of radiolabeled albumin. The importance of this observation is supported by the fundamental role that heparan sulfate plays in the structure of the basement membrane and the previous studies relating a reduction in glomerular basement membrane heparan sulfate content to albuminuria in human and experimental diabetes.^{20 - 21} Other investigators have proposed that the nonenyzmatic glycosylation of basement membrane and interstitial proteins may be an important factor in the altered microvascular permeability that characterizes diabetic microangiopathy.^{22 - 23} Thickening of the pulmonary capillary basement membrane has been described in the ZDF rat²⁴ as well as the lungs of patients with diabetes.²⁵ Of note, however, there is no evidence to link either of these latter observations to the reduction in pulmonary microvascular permeability that characterizes the diabetic animals in the present study.

Numerous investigators have demonstrated alterations in the humoral and cellular components of the inflammatory response in clinical and experimental diabetes. Neutrophils harvested from diabetic patients or animals display a variety of defects in their function, including chemotaxis,²⁶ respiratory burst,^{26 - 28} phagocytosis,²⁹ and deformability.³⁰ In the present study, both the diabetic and the nondiabetic groups had similar degrees of neutrophil adhesion molecule expression and sequestration within the lungs in response to intratracheal endotoxin. This suggests that, at least in this experimental model, the ability of neutrophils from diabetic animals to respond to intratracheal LPS by expressing adhesion glycoproteins and emigrate into the airways appears to be intact. Other in vitro measures of neutrophil activation (such as respiratory burst) were not assessed in the present study. It is interesting to observe that pulmonary microvascular permeability remained unchanged in the diabetic animals exposed to LPS despite the movement of a large number of neutrophils through the pulmonary microvasculature into the interstitium and airways. This observation is consistent with those of Delclaux et al² and Martin et al,³¹ who reported that neutrophil migration across the alveolar capillary wall, in and of itself, does not necessarily induce a significant alteration of microvascular barrier permeability.

Diabetes also has important effects on humoral mediators of the acute inflammatory response, particularly arachidonic acid metabolites. Rats with streptozocin-induced diabetes mellitus have been shown to have abnormalities in essential fatty acid metabolism with impaired synthesis of arachidonic acid.^{32 - 33} Other investigators have demonstrated impaired uptake of arachidonic acid,³⁴ reduced levels of cyclooxygenase-1 message,³⁵ and diminished eicosanoid production.^{33 - 34} These biochemical abnormalities are likely to assume physiological importance given the well-recognized role that various eicosanoids play in pulmonary microvascular function during both normal states³⁶ and proinflammatory states, including endotoxemia.³⁷

In summary, these data suggest that the pulmonary microvascular response to intratracheal endotoxin is severely blunted in diabetic rats when compared with that of normoglycemic controls. This occurs despite similar degrees of neutrophil CD11b/CD18 expression and pulmonary neutrophil sequestration within the lungs. These studies support the contention that chronic illnesses such as diabetes mellitus may significantly alter the responsiveness of an organ to acute inflammatory stimuli. This assumes particular importance given the frequent incidence of chronic illnesses among critically ill patients.

This work was supported in part by a Merit Review Grant, Department of Veterans Affairs, Washington, DC (Dr Turnage).

Presented at the 19th Annual Meeting of the Surgical Infection Society, Seattle, Wash, May 1, 1999.

Reprints: Richard H. Turnage, MD, Department of Surgery, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75235-9060 (e-mail: Turnage.Richard_H@Dallas.VA.gov).

Daniel L. Traber, MD, Galveston, Tex: I would like to thank the Society for the honor of being able to discuss this very excellent paper. It is relevant to what has preceded here at this meeting, especially Dr Chaudry's presidential address as well as the symposium we had this morning, because we point out the importance of a correct animal model to look at sepsis, and this model and this study have several features about them that make them especially clinically relevant. First of all, it is a 2-hit model, since you saw there was some injury, not necessarily morphological injury, but certainly influx of polymorphonuclear cells with saline alone, and then secondly, of course, the second hit being the sepsis insult, and then thirdly, we have what is commonly seen in the clinical situation, an underlying disease process. And I have several questions that I would like to propose.

First of all, because the model is so important, I would like to ask a question related to the saline-alone group. As I calculate these data from the presentation—and I would like again to thank the group for having given me the manuscript well ahead of the time of this meeting—I know that in our institution and in many others, they use about 40 to 50 mL of saline as they are lavaging patients who are on ventilators, and that is exactly the amount if you extrapolated the size of a rat to the size of a human being, his 0.25 mL of saline as a control would be what a normal therapist would be giving to a patient, and yet he is seeing some rather interesting changes in microvascular permeability to albumin. I would like to know if he had data from unstimulated animals for the albumin leakage to be able to quantitate.

And then the other aspect of this study is that this disease entity, the diabetic situation, as well as the injury to the lung, are not associated with no systemic abnormalities, certainly there are also systemic responses that occur, and so important in a model of sepsis that we are indeed looking at a resuscitated animal. Perhaps it may not be possible to resuscitate to an end point, but whether or not this group has used resuscitation techniques and whether or not there are cardiovascular changes that were seen between the 2 groups of animals, at least from a systemic standpoint, might have been similar.

The authors have in a very elegant manner described morphological changes and some of the microvascular changes that were seen in the preparation, but were there any physiological changes? Mainly when we are dealing with ARDS [acute respiratory distress syndrome] in sepsis, we like to look at PF ratios [PaO₂/FIO₂]. Whether or not there was indeed a fall in PaO₂ is a major question.

Now, getting to the diabetic model, the major lesion appears to be the fact that these cells can migrate across the microvasculature and don't appear to be impeded, but they aren't being activated for some reason. I know that cell signaling nowadays by many individuals is thought to be the result of the selectin family of molecules. These contain carbohydrate materials whether or not this diabetic animal has a defect in its selectins. Of course, the other aspect of this is the damage that normally is seen with changes in permeability due to the release of oxygen free radicals. Is there some lesion in the diabetic animal that precludes their ability to have an oxidative burst from these cells? That may also be a reason why they have a systemic infection.

Dr Wright: To address your first question regarding to the saline-induced injury, we did not for these experiments perform any experiments where we just measured Evans blue dye extravasation without any type of vehicle administration. Evans blue dye does cross the endothelial barrier in the lung at a linear rate up to 1 hour, and some previous studies have shown where they may have a lower extravasation of dye than we did. Some of those were done at a 15-minute time point, and that may explain why their numbers are higher. With regard to the neutrophils within the airways, I am not sure; we just haven't done those experiments.

And regarding the cardiovascular status of these animals, we have not measured cardiac output or blood pressure or any other hemodynamic variables. Intratracheal endotoxin is not associated with a severe hemodynamic compromise as IV administration, although surely it does affect the animals' hemodynamics. It does affect the oxygenation, diffusion capacity, and compliance in the lung in a rat model, and this has been demonstrated, especially at 12 to 24 hours, following intratracheal endotoxin exposure. Whether or not that is present at 4 hours, I am not sure, and we have not measured that as of yet.

With regard to problems with selectins in diabetic animals, I am not aware of any defects in these animals, although membrane and surface proteins are glycosylated and cross-linked after exposure to prolonged hyperglycemia, but I don't know if these proteins in particular are affected by this process.

And with regard to the oxidative burst in these neutrophils, we are currently in the process of looking at that, but I don't have any data to show you as of yet. As you are probably well aware, previous work has demonstrated defects in neutrophil respiratory bursts, deformability, and chemotaxis and diabetes, and some of these changes have been demonstrated in animal models of diabetes such as this one as well.

Nicolas V. Christou, MD, Montreal, Quebec: What causes the damage in the lung? The neutrophils get there; they have got their adhesion molecules up-regulated. I don't know what causes the damage. Maybe you can enlighten us. So I am going to ask you 2 questions.

We spoke about relevancy. Throwing endotoxin down an animal's lung is not how our patients get challenged, so let's start getting relevant and throw bacteria, which is a replicating antigen, which also will modulate the endotoxin up and down depending on what happens with that infection. So if you haven't done it, can you speculate as to what would happen if you were to use bacteria as your challenge?

And second of all, you mentioned oxygen radicals. Do you have any data or have you measured the capability of these cells to respond to any trigger, either membrane associated with fMLP [N-formyl-methionyl-leucyl-phenylalanine] or PMA [phorbol-myristate acetate], to see if indeed once they are there, they are stuck, they can't go anywhere? If they really get pushed, will they cause the damage? I really would like to know what is causing the damage.

Dr Wright: A couple of other ideas as to what is causing this injury are other mediators, such as TNF [tumor necrosis factor], interleukins, prostaglandins, and that may be one difference that we are observing in the diabetics vs the nondiabetics. The diabetic animals do have altered fatty acid metabolism, which leads to an altered composition of membrane phospholipids, which then causes a relative deficiency of arachidonic acid. We have measured prostaglandin and thromboxane levels in these animals, and they are lower, both at baseline and in response to injury, than the nondiabetic animals, and we are still in the process of examining that.

With regard to a more relevant model, such as bacteria injection, we do have some preliminary studies where we have injected live organisms down the trachea of these animals and have just done some survival studies to see how long they last, and preliminarily, it looks like the diabetic animals do not survive as well. After about 4 or 5 days, most of those animals have died, whereas relatively few of the nondiabetic animals are killed by that.

Ronald V. Maier, MD, Seattle, Wash: If I understand you correctly from your preliminary data where you used basically a hydrostatic force, mechanical force, and showed decreased leak, and now in this series where you show that you do get neutrophils in the alveolar space, and from pathology it looks like they are activated and inflamed and in the interstitium and you have interstitial fluid but less leak through both mechanisms, I wonder if the explanation is as you mentioned, glycosylation causes lots of cross-linking of proteins, and whether you are just having basically a gluelike mechanism of your basement membrane so that you physically cannot get fluid across into the alveolar space. And I wonder if you have any evidence or data to support that.

These animals are young enough that they probably wouldn't show structural changes on morphology as far as thickening of the basement membrane, but from a functional standpoint as far as permeability, they may just have glycosylated cross-linking that prevents permeability or the increase with thromboxane or give them endotoxin.

Dr Wright: Right. We haven't done any experiments as of yet looking specifically at the cross-linking, although those are in our plans. With aminoguanidine, you can prevent that advanced glycosylation cross-linking of those products, but then you run into problems with nitric oxide, so you don't really know what is causing your injury or preventing it.

In addition, it appears to be very organ-specific, the changes in permeability. Increased glycosylation products and alterations in heparan sulfate, for instance, in basement membranes have been measured in different organs, say, in the kidney, where permeability is increased, or in the lung, where we are showing that it's decreased and other people have also shown it to be decreased. So it is very interesting how it is organ-specific as well as whatever mechanism is involved.

E. Patchen Dellinger, MD, Seattle: Are there any data suggesting that the incidence of ARDS in diabetic patients is different after trauma or in association with SIRS [systemic inflammatory response syndrome], or are there any data regarding the relative incidence at all?

Dr Wright: That is a very good question, and I have looked long and hard for that. I have not been able to find any good studies showing those differences. Talking to some of the pulmonologists, anecdotal cases, they will say that may be the case, that diabetics gets less ARDS, but I don't know of any studies that have shown that.

Dr Dellinger: Those SIS [Surgical Infection Society] members with trauma registries might think of searching their database to look for that.

Edwin A. Deitch, MD, Newark, NJ: Permeability is a surrogate for dysfunction. Do you have any O₂ levels, PaO₂ levels, or any other functional assays to say that the lungs of these diabetic animals work better, oxygenate better, ventilate better?

Dr Wright: No, we haven't measured oxygenation or blood arterial oxygen levels.

Cathy L. White Owen, MD, Cleveland, Ohio: Since you are using an isolated perfusion model, is there any possibility that you would consider looking at a cell-free system to try and sort out whether this is a PMN [neutrophil]–related issue vs a lung-related issue, and have you considered doing studies where you used diabetic lungs with normal neutrophils and normal lungs with diabetic neutrophils?

Dr Wright: No, we have not done that.