The Role of Tumor Necrosis Factor-α in the Pathogenesis of Aspiration Pneumonitis in Rats 

All procedures performed on rats in this study were approved by the State University of New York at Buffalo's Institutional Animal Care and Use Committee and complied with all state, Federal, and National Institutes of Health regulations.

Four different injury vehicles were prepared for this study:(1) normal saline, pH = 5.3;(2) normal saline adjusted to pH = 1.25 with reagent HCl (acid);(3) 40 mg/ml gastric particles in normal saline, pH = 5.3 (particulate); and (4) 40 mg/ml gastric particles in normal saline adjusted to pH = 1.25 with reagent HCl (acidic particulate). [11] All four of these injury vehicles were instilled into the lungs of the test animals at a dose of 1.2 ml/kg. In addition, a fifth injury group was included, normal saline adjusted to pH = 1.25 with reagent HCl and delivered at a dose of 2.4 ml/kg (high-volume acid). The gastric particles were prepared by collecting the stomach contents from necropsied rats, washing three times in normal saline, pH = 5.3 and filtering through sterile, gauze sponges (to remove large "chunks"). [11] After autoclaving for 25 min (20 psi, 121 [degree sign]C), the filtrate was centrifuged at 3000 rpm for 2 min at room temperature (2,000g, GS-6R centrifuge with GH-3.8 rotor, Beckman Instruments, Palo Alto, CA), the supernatant discarded, and the pellet weighed (particle concentrations reported are calculated based on this wet weight). A stock solution of gastric particles was prepared by suspending the pellet in normal saline, pH = 5.3 to 100 mg/ml, and was used to prepare the particulate-containing injury vehicles.

Anesthesia was induced in male, 250–300-g, Long-Evans rats (Harlan Sprague-Dawley, Indianapolis, IN) in a glass chamber with 3–4% halothane in 50% O^2/airand maintained with 1.5% halothane, delivered by nose cone with spontaneous breathing. Gas concentrations were monitored continuously using a RASCAL II Ramon light-scattering spectrophotometer (Ohmeda, Salt Lake City, UT). Animals were placed on a 60-degree upright dissection board and suspended by the front teeth, exposing the ventral side of the neck. [9] A 2-cm midline incision was made to expose the trachea into which an 18-gauge, 1.25-inch Teflon catheter (Becton Dickinson Infusion Therapy Systems, Sandy, UT) was inserted to just above the carina and secured with 1–0 silk. [7] A 1-ml syringe was used to instill the appropriate injury vehicle into the lungs through the catheter. One minute later, the catheter was removed and the trachea and neck incisions were closed with 6–0 and 4–0 silk, respectively. The animal was removed from the inclined board and allowed to awaken in a cage maintained at 37 [degree sign]C with a servocontrolled heat lamp while spontaneously breathing room air. Food and water were provided ad libitum until harvest.

In experiments to determine TNF-[Greek small letter alpha] activity in the lungs following aspiration injury, bronchoalveolar lavage was performed at various times following instillation of the aspirant (each time point represents one animal because this was a terminal procedure). The animal was anesthetized by intraperitoneal injection of 260 mg of sodium pentobarbital, and the neck incision was reopened to expose the trachea. A 16-gauge, 1.25-inch Teflon catheter was inserted into the trachea, distal to the inoculation incision, and secured in position with a suture. A longitudinal incision was made through the diaphragm and sternum and the rib cage was spread open, carefully, to avoid puncturing the lungs. A three-way stopcock with two 20-ml syringes attached, one empty and the other containing 14 ml sterile, phosphate-buffered saline (PBS)(137 mM NaCl, 8 mM Na²HPO⁴, 1.5 mM KH²PO⁴, 2.7 mM KCl, 0.9 mM CaCl², 0.5 mM MgCl², pH = 7.4), was attached to the catheter and 2 x 7-ml PBS instilled into the lungs and recovered. The collected lavage fluid was pooled, centrifuged at 2100 rpm for 5 min at 4 [degree sign]C (700g, GS-6R centrifuge with GH-3.8 rotor, Beckman Instruments), and the supernatant aliquoted and stored at -20 [degree sign]C.

WEHI 164, subclone 13, cells, a TNF-[Greek small letter alpha]-sensitive cell line derived from a mouse fibrosarcoma (a generous gift from Dr. Steven L. Kunkel, Department of Pathology, University of Michigan, Ann Arbor, Michigan), were cultured to approximately 90% confluency in culture medium containing RPMI-1640, 2 mM L-glutamine, 10% fetal calf serum (GIBCO-BRL, Grand Island, NY), and 3 [micro sign]g/ml gentamicin (Sigma Chemical, St. Louis, MO) in T75 flasks at 37 [degree sign]C in 5% CO². Cells were prepared for the assay by detaching with 0.25% trypsin and 0.02% EDTA (Sigma Chemical, St. Louis, MO), adding 10 ml culture medium per flask, centrifuging at 800g for 5 min at room temperature (GS-6R centrifuge with GH-3.8 rotor, Beckman Instruments), and resuspending in culture medium supplemented with 1 [micro sign]g/ml actinomycin D (Calbiochem, La Jolla, CA) to a concentration of 500,000 cells/ml. One hundred microliters of cell suspension was added to each well of a flat-bottom 96-well plate (Sarstedt, Newton, NC) containing 100 [micro sign]l of two-fold serial dilutions of unknown samples, in triplicate, or known concentrations of human rTNF-[Greek small letter alpha] standards (R & D Systems, Minneapolis, MN) in diluting medium: RPMI-1640, 2 mM L-glutamine, 1% fetal calf serum, and 15 mM HEPES (Sigma Chemical). Following 20 h of incubation at 37 [degree sign]C in 5% CO (²), 20 [micro sign]l of filter sterilized 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT; Sigma Chemical) in PBS (5 mg/ml) was added to each well. After an additional 4-h incubation, 150 [micro sign]l was removed from each well and replaced with 100 [micro sign]l of 0.04-N HCl in isopropanol. The plates were incubated for 20 h in the dark at room temperature, and the absorbance at 570 nm (OD⁵⁷⁰) was measured using a 3550-UV microplate reader (Bio-Rad, Hercules, CA). A standard curve (reverse sigmoid in shape) of the OD⁵⁷⁰versus log TNF-[Greek small letter alpha] concentration was plotted and the TNF-[Greek small letter alpha] concentration of each sample was determined from the dilution closest to the inflection point of the standard curve. This assay has a detection limit of approximately 1 pg/ml. [19]

The assay is based on the specific cytotoxicity of the WEHI 164, subclone 13, cells to TNF-[Greek small letter alpha]. MTT is used as a cell viability indicator, being taken up and cleaved to a dark-blue formazan product by active mitochondria, which is quantitated spectrophotometrically. Increasing TNF-[Greek small letter alpha] concentration results in increased cell death and, therefore, a reduced absorbence at 570 nm. [20]

All reagents and equipment were maintained RNase-free during experiments involving RNA extraction. Whole lungs were dissected at 0, 1, 2, and 4 h after instillation of the injury vehicle, following flushing of the pulmonary circulation with 10 ml saline, via right-ventricle injection. The harvested lungs were snap-frozen in liquid nitrogen. Total RNA was extracted from whole-lung homogenates using a guanidinium isothiocyanate/chloroform-based technique (TRIZOL Reagent, GIBCO-BRL), followed by isopropanol precipitation. Purified rat-lung total RNA, 12 [micro sign]g, was fractionated on a 1% agarose formaldehyde gel and transferred to a nylon membrane (Zetabind, CUNO, Meriden, CT) by capillary action. A full-length cDNA probe for the rat TNF-[Greek small letter alpha] gene (708 base pairs) was PCR-cloned from an IgG immune complex-injured rat lung using the following primers: 5′-ATGAGCACGGAAAGCATGATCCGA (sense); 5′-tcacagagcaatgactccaaagta (antisense), with the following cycle conditions: 94 [degree sign]C x 30 s, 52 [degree sign]C x 1 min, 72 [degree sign]C x 1 min 30 s, for 35 cycles. The amplified product was sequenced and confirmed to be rat TNF-[Greek small letter alpha]. [21] This cDNA probe was radiolabeled with [(³²) P]dCTP (Pharmacia, Piscataway, NJ) by random priming. Radioactivity of the probe was determined by an LS6500 scintillation counter (Beckman Instruments) and 1.5 x 10⁷cpm was applied to the blot with hybridization occurring at 65 [degree sign]C for 16 h. Autoradiography of the blots was performed at -70 [degree sign]C on X-OMAT-AR film (Eastman Kodak, Rochester, NY). The resultant autoradiograph was digitized and analyzed with a microprocessor-controlled GS-700 imaging densitometer and Molecular Analyst version 1.5 software (Bio-Rad). The TNF-[Greek small letter alpha] mRNA and corresponding 28S ribosomal RNA bands were integrated from the acquired image and the TNF-[Greek small letter alpha] mRNA bands normalized to a constant 28S ribosomal RNA level. [22]

The affinity-purified IgG fraction of goat anti-rat TNF-[Greek small letter alpha] antisera was administered intratracheally to assess whether "locally expressed" TNF-[Greek small letter alpha] was required for full expression of gastric aspiration injury. [12,23] The production, purification, and characterization of this polyclonal antibody has been fully described. [24] Anti-TNF-[Greek small letter alpha], or control goat IgG (Lampire Biological Laboratories, Pipersville, PA), was administered intratracheally (1.5 or 3.0 mg/kg) with the injury vehicle, in the case of normal saline and particulate injuries, or 1 min following injury vehicle instillation at 0.6 ml/kg for acid, high-volume acid, and acidic particulate injuries via the same intratracheal catheter. Treatment with anti-TNF-[Greek small letter alpha] was not administered with the acid-containing injury vehicles because of the potential for inactivation of the antibody. Another group of animals was included in these experiments in which no injury or treatment solution was instilled into the lungs to provide basal levels of the indices measured.

Alveolar-capillary wall integrity was quantified by measuring the translocation of radiolabeled albumin from the blood into the lung. Blood gases were measured to assess gas-exchange efficiency. Immediately prior to injury, 100,000 cpm of¹²⁵I-BSA (Dupont NEN Research Products, Boston, MA) in 1 ml 2% BSA was injected into the penile vein. At 5 h after injury, the animals were exposed to 80% oxygen, breathing spontaneously, for 10 min then anesthetized with halothane (as described previously except with 80% oxygen in air as the carrier gas). A midline peritoneal and thoracic incision was made and approximately 500 [micro sign]l of blood collected from the abdominal aorta with a heparinized syringe and 22-gauge needle. One milliliter of venous blood was collected from the inferior vena cava. The vena cava was cut to exsanguinate the animal, and the pulmonary vasculature was cleared of residual blood by injecting 10 ml of sterile PBS into the beating heart's right ventricle. The lung was excised and its¹²⁵I activity determined using a 1282 CompuGamma CS gamma counter (Wallac, Gaithersburg, MD). The activity in the 1-ml blood sample was also measured and the lung albumin permeability index (PI) calculated as a simple ratio of the activity in the lung (minus background) to that in 1 ml blood (minus background). [7]

The arterial blood sample was injected into an ABL4 blood gas analyzer (Radiometer America, Westlake, OH) and the gas-exchange efficiency calculated as Pa^O(2)/FI^O(2), where Pa^O(2) is the arterial oxygen partial pressure in millimeters of mercury and FI^O(2) is the inspired oxygen concentration (80%). [25]

Myeloperoxidase (MPO) activity in lung homogenates was measured as a surrogate marker of lung neutrophil infiltration. After counting on the gamma counter, harvested lungs were homogenized on ice using a Polytron TP-2000 tissue homogenizer (Brinkman Instruments, Westbury, NY) in 3 ml buffer containing 50 mM KH²PO⁴, 13.7 mM hexadecyltrimethylammonium bromide, and 5 mM EDTA, pH = 6.0. The homogenate was then sonicated on ice for 4 x 10-s pulses using a Sonifier Cell Disruptor 350 with microtip probe (Branson Ultrasonics, Danbury, CT) and centrifuged at 3200 rpm for 30 min at 4 [degree sign]C (2300g, GS-6R centrifuge with GH-3.8 rotor, Beckman Instruments). The resultant supernatant was assayed for MPO activity by combining 50 [micro sign]l of sample with 1.5 ml assay buffer containing 50 mM KH²PO⁴, 176 [micro sign]M H²O², and 525 [micro sign]M o-dianisidine dihydrochloride, pH = 6.0, in a cuvette and continually recording the absorbance at 460 nm for 2 min using a DU-650 spectrophotometer (Beckman Instruments). The spectrophotometer was blanked with 50 mM phosphate buffer before reading the sample. MPO activity was expressed as the absorbance change (at 460 nm) per minute (ABS/min) over the linear portion of the curve. [26]

Intergroup comparisons (i.e., TNF-[Greek small letter alpha] bioactivity from particulate-injured vs. acidic particulate-injured at 1 h after injury, or PI from acid-injured IgG-treated vs. anti-TNF-[Greek small letter alpha] antisera-treated) were made using the Student two-sample t test after F tests to determine if equal or unequal variance assumptions were appropriate. To guard against the possibility of the data not being normally distributed, Mann-Whitney tests were also conducted. All comparisons were performed using two-tailed tests. To test for interaction between acid and particulate injury, with respect to the injury (PI, Pa^O(2)/FI^O(2)) or inflammation (MPO) parameters, two-factor analyses of variance were performed among the groups receiving similar treatments (i.e., IgG or anti-TNF-[Greek small letter alpha]). The effect of normal saline, acid, or high-volume acid aspiration on TNF-[Greek small letter alpha] bioactivity in bronchoalveolar lavage fluid was analyzed at each harvest time point for significant differences from no injury (0 h) by only the nonparametric Mann-Whitney test because the data were decidedly non-normal. Values are expressed as mean +/- SD, with n = total number of animals in each group. Differences were accepted as significant for P < 0.05.

All study groups demonstrated a TNF-[Greek small letter alpha] response that was detectable in the lavage fluid by the WEHI assay, albeit with varying kinetics and amplitude. Figure 1displays the time course of experiments with normal saline, acid, and high-volume acid as the injury vehicles. All three injury regimes produced TNF-[Greek small letter alpha] peaks that resolved to baseline by 4 h after injury. The response from normal saline peaked at 1 h after injury, 95.8 +/- 133.8 pg/ml (P < 0.01 compared with no-injury levels), whereas the peak response from acid and high-volume acid did not occur until 2 h after injury, 48.3 +/- 62.4 and 82.6 +/- 124.8 pg/ml, respectively (P < 0.025 compared with no-injury levels). Groups receiving gastric particles produced a much more robust and sustained TNF-[Greek small letter alpha] response (Figure 2). Particulate and acidic-particulate groups produced greater amounts of lavage fluid TNF-[Greek small letter alpha] at all time points after injury compared with those injuries that did not contain gastric particles (P < 0.0001). At 2 h after injury, lavage fluid TNF-[Greek small letter alpha] levels in the particulate group were 390-fold greater than those in the normal saline group (19.3 +/- 7.9 ng/ml vs. 0.05 +/- 0.08 ng/ml), 400-fold greater than the acid group, and 235-fold greater than the high-volume acid group. The elevated TNF-[Greek small letter alpha] response was sustained through 4 h after injury in the particulate group (28.0 +/- 7.8 ng/ml) and the acidic-particulate group (32.7 +/- 10.8 ng/ml), a time at which the normal saline, acid, and high-volume acid lavage fluid TNF-[Greek small letter alpha] levels had returned to baseline (4.3 +/- 3.4, 8.2 +/- 8.9, and 6.7 +/- 5.1 pg/ml, respectively). Interestingly, adding an acidic component to the gastric particulate-treated lungs initially delayed the onset of the TNF-[Greek small letter alpha] response (at 0.5 h, particulate = 300.8 +/- 185.0 pg/ml vs. acidic particulate = 87.6 +/- 115.5 pg/ml, P < 0.05; and at 1 h, particulate = 5.4 +/- 3.0 ng/ml vs. acidic particulate = 2.1 +/- 0.9 ng/ml, P < 0.05). It did not affect the peak plateau from 2 to 4 after injury, but it did hasten the TNF-[Greek small letter alpha] response's decline (at 8 h, particulate = 9.6 +/- 1.7 ng/ml vs. acidic particulate = 2.0 +/- 0.3 ng/ml, P < 0.0002).

To determine the transcriptional state of TNF-[Greek small letter alpha] production in injured lung tissue, experiments were performed with each of the injury groups and analyzed for TNF-[Greek small letter alpha] mRNA using Northern blots. Only the lungs from animals injured with aspirant vehicles that contained gastric particles (particulate and acidic particulate) demonstrated robust upregulation of TNF-[Greek small letter alpha] mRNA (Figure 3). This upregulation peaked between 1 and 2 h after injury and began to diminish by 4 h after injury. Expression of TNF-[Greek small letter alpha] mRNA could not be consistently detected in homogenized lung samples that had been injured with normal saline, acid, or high-volume acid using this technique (data not shown).

To assess the role that TNF-[Greek small letter alpha] plays in the pathogenesis of gastric aspiration lung injury, experiments with anti-rat TNF-[Greek small letter alpha] antibody treatment were conducted on animals with acid, high-volume acid, particulate, or acidic-particulate solutions instilled into the lungs. Normal saline injury was not included as an experimental group because the lung injury that results, as assessed by lung protein permeability and blood-gas analysis, is not sufficiently different from untreated animals to detect effects of anti-TNF-[Greek small letter alpha] antisera treatment (PI = 0.40 +/- 0.13 vs. 0.28 +/- 0.09, and Pa^O(2)/FI^O(2)= 503.9 +/- 22.1 mmHg vs. 550.6 +/- 56.1 mmHg). Instillation of nonspecific goat IgG in the same manner as the anti-TNF-[Greek small letter alpha] antibody treatment was used as the control in all injury groups.

Treatment with anti-TNF-[Greek small letter alpha] antibody produced a significant reduction in albumin permeability of the alveolar-capillary wall produced by acid and high-volume acid, but not particulate or acidic-particulate lung injury (Figure 4). The PI was reduced 21.8% in the acid (0.86 +/- 0.28 vs. 1.10 +/- 0.36, P < 0.05) and 27.5% in the high-volume acid injury group (1.32 +/- 0.22 vs. 1.82 +/- 0.29, P < 0.0005). No reduction in this index of lung injury was demonstrable by anti-TNF-[Greek small letter alpha] antibody treatment of particulate pulmonary injury (0.95 +/- 0.29 vs. 1.00 +/- 0.23). Also, doubling of the anti-TNF-[Greek small letter alpha] dose, from 1.5 mg/kg to 3.0 mg/kg, still did not produce a significant reduction in particulate lung injury, as assessed by PI. The IgG-treated acidic particulate injury group had a PI of 5.07 +/- 1.19, which represented a 4.6-fold, synergistic increase (P < 0.0001) in PI over the similarly treated acid-injured animals, and a 5.0-fold increase in PI over the particulate-injured group. Anti-TNF-[Greek small letter alpha] antibody treatment had no significant effect on this synergistic increase (P < 0.0001) in lung injury from the combined gastric components.

Impairment of lung function (i.e., blood oxygenation) by the different injury models was also assessed. Acid injury alone resulted in a 57% decrease in pulmonary efficiency, as measured by Pa^O(2)/FI^O(2)(Figure 5), compared with uninjured animals (238.7 +/- 99.9 mmHg vs. 550.6 +/- 56.1 mmHg, P < 0.0001). Increasing the volume of the acid instillate yielded a 73% reduction (149.8 +/- 49.9 mmHg, P < 0.0001). However, particulate injury only caused an 11% reduction in Pa^O(2)/FI^O(2)(490.9 +/- 78.1 mmHg, P < 0.05), even though pulmonary albumin permeability was increased to the same magnitude as the acid injury. The acidic-particulate group demonstrated a synergistic increase (P < 0.02) in lung injury, as measured by this parameter, yielding an 86% decrease (75.5 +/- 15.6 mmHg, P < 0.0001). The effect of anti-TNF-[Greek small letter alpha] treatment on the pulmonary efficiency for each of the injury regimes mirrored the pattern for PI. Both the acid and the high-volume acid groups responded with increased gas exchange (by 77% in the acid group, to 423.2 +/- 104.4 mmHg, P < 0.0001, and by 90% in the high-volume acid group, to 284.1 +/- 126.9 mmHg, P < 0.01), whereas none of the groups containing gastric particles experienced a significant change when compared to their associated nonspecific IgG control groups. As with the lung protein permeability parameter, anti-TNF-[Greek small letter alpha] treatment did not effect the synergistic interaction (P < 0.0001) resulting from the combination of the acid and particulate injury modalities in regards to pulmonary gas-exchange efficiency.

Previous work from our laboratory has demonstrated that alveolar neutrophil infiltration is involved in acid, gastric-particulate, and acidic-particulate aspiration injury. [7,9] MPO activity in lung homogenates from injured animals treated with anti-TNF-[Greek small letter alpha] antibody or nonspecific IgG was determined to ascertain TNF-[Greek small letter alpha's] involvement in neutrophil infiltration (Figure 6). The acid and high-volume acid groups were the only groups in which anti-TNF-[Greek small letter alpha] antibody reduced neutrophil infiltration (0.29 +/- 0.14 vs. 0.58 +/- 0.43 ABS/min for acid and 0.29 +/- 0.07 vs. 0.44 +/- 0.19 ABS/min for high-volume acid, P < 0.05). Anti-TNF-[Greek small letter alpha] antibody treatment did not change the MPO activity in the particulate or acidic-particulate groups. Comparing the IgG-treated groups, the relationship of the MPO activity of the acidic-particulate group (0.96 +/- 0.35 ABS/min) to that of the acid alone (0.58 +/- 0.43 ABS/min) or particulate alone (0.71 +/- 0.23 ABS/min) groups demonstrated no interactive effects (P < 0.30) between the acid and particulate injuries with respect to neutrophil infiltration. This lack of an interactive effect was also true when comparing the same injury groups that had been treated with anti-TNF-[Greek small letter alpha] antibody (P < 0.19).

(Table 1) offers a semiquantitative, "visual" presentation of how different injury-induced TNF-[Greek small letter alpha] responses (as determined in lavage fluid) compare with the resulting injury and inflammation parameters measured. Clearly, a direct correlation between the level of TNF-[Greek small letter alpha] and the lung injury resulting from acid, particulate, or acidic particulate instillation cannot be drawn.

Aspiration of gastric components leads to an acute pulmonary inflammatory injury that is characterized by proteinaceous edema in the lung, a decrease in pulmonary gas-exchange efficiency, and neutrophil infiltration into the interstitial and alveolar space from the intravascular compartment, as well as various systemic inflammatory responses. [6–9,11,16–18,25,27–37] Patients that experience this condition are at risk for developing ARDS, with its associated high mortality rate. [3,4,38] There is considerable evidence that the proinflammatory cytokine, TNF-[Greek small letter alpha], plays an important role in the pathophysiology associated with acute pulmonary inflammation resulting from a variety of lung insults, including gastric aspiration. [10,12–16,18,23,24,39,40] This study has examined the role of TNF-[Greek small letter alpha] in the pathogenesis of aspiration pneumonitis by characterizing the TNF-[Greek small letter alpha] response to different gastric components and determining if it is required for full expression of the resulting lung injury. This has been done with the goal of developing therapeutic strategies for this potentially lethal complication of anesthesia and surgery.

Previously, we have demonstrated that the rat aciaspiration pneumonitis model involves a biphasic injury pattern, with an "early phase" (0–1 h after aspiration) that is associated with the direct chemical effects of the acid, and an additional "late phase" (4–6 h after aspiration) inflammatory injury. Neutrophils have been clearly established, by morphometric analysis and depletion studies, as necessary for the additional inflammatory lung injury of the "late phase." [7,9] These results have been confirmed by other laboratories. [27,31,33] We have also reported that aspiration of washed and filtered gastric particulate matter causes an inflammatory lung injury similar to the "second phase" of the acid aspiration. Maximal leakage of protein occurs at 4–6 h after deposition of the food particles into the trachea and corresponds to infiltration of large numbers of neutrophils into the lung. There is not a significant "first phase" injury as with acid aspiration. [11] The data presented here support those previous results (Figure 4 and Figure 6).

In this study a small increase in TNF-[Greek small letter alpha] levels in the bronchoalveolar lavage fluid of rats receiving an instillate of normal saline into the lung was detected using a cytotoxicity bioassay. The temporal relationship of TNF-[Greek small letter alpha] release differs based on the pH of the instillate (with a low pH, pH = 1.25, TNF-[Greek small letter alpha] levels peak later than normal saline, pH = 5.3;Figure 1). Others have implicated TNF-[Greek small letter alpha] involvement in acid-aspiration injury but do not find detectable levels of TNF-[Greek small letter alpha] in the bronchoalveolar lavage fluid from acid-injured lungs. [17,36,37] The discrepancy between these reports and ours may be caused by their use of the L929 mouse fibroblast bioassay, which is not as sensitive for TNF-[Greek small letter alpha] detection as the WEHI bioassay. [19] However, detectable increases in serum TNF-[Greek small letter alpha] levels have been reported 1–2 h after injury. [17,18,37] In the current study, upregulation of TNF-[Greek small letter alpha] mRNA was not consistently detected in the normal saline-instilled animals at either pH. This result is consistent with the report from Ohara et al., [34] in which upregulation of TNF-[Greek small letter alpha] mRNA could not be detected in alveolar macrophages (most likely source of TNF-[Greek small letter alpha] in acute lung injury [12]) isolated from acid-injured mouse lungs.

When gastric particulate material is instilled into the lungs, a large (4-log increase over uninjured controls at 4 h after instillation) and sustained (>8 h) TNF-[Greek small letter alpha] response is detected in the bronchoalveolar lavage fluid (Figure 2). TNF-[Greek small letter alpha] mRNA is also upregulated as a result of this injury (Figure 3). Other studies have demonstrated a similar TNF-[Greek small letter alpha] response using other types of particulate insult. [13,15,39] The acidic-particulate combination injury results in a TNF-[Greek small letter alpha] response, as assessed in the bronchoalveolar lavage fluid, as well as with mRNA upregulation, that closely mirrors that of particulate alone in terms of intensity. However, the addition of the acidic component to the particulate injury inhibits the initial TNF-[Greek small letter alpha] response and accelerates its decline (Figure 2).

Goldman et al. [16] demonstrated in an earlier study that the pulmonary protein permeability, neutrophil infiltration into the lung, and lung wet/dry weight of rats injured by acid aspiration can be reduced by intraperitoneal administration of anti-TNF-[Greek small letter alpha] antisera. Similar results have been obtained in this study (Figure 4 and Figure 6). In addition, intratracheal administration of anti-TNF-[Greek small letter alpha] antibody improves blood oxygenation in the animals injured with acid alone (Figure 5).

This study clearly supports the hypothesis that TNF-[Greek small letter alpha] plays an important role in inflammatory processes involved in acid-aspiration pneumonitis. The paradigm that recognition cytokines (i.e., TNF-[Greek small letter alpha], interleukin [IL]-1[Greek small letter beta]) "prime" the inflammatory response and lead to the elaboration of other inflammatory regulators (i.e., chemokines, eicosenoids) is consistent with the results of this injury model. [10,41] Inhibition of TNF-[Greek small letter alpha] limits the recruitment of neutrophils and decreases lung injury associated with acid aspiration. Because TNF-[Greek small letter alpha] is not by itself directly chemotactic, the obvious conclusion is that the cytokine either leads to the generation of direct regulators of neutrophil infiltration and activation by paracrine or autocrine mechanisms or enhances the action of these chemotactic agents. [10] Additionally, the observation that TNF-[Greek small letter alpha] levels in the bronchoalveolar lavage fluid peaks later in both the low pH models (normal saline and particulate) suggests that a low pH aspirate may initially directly inhibit TNF-[Greek small letter alpha] release. Thus, we postulate that indirect effects of the insult lead to the later transient increases in TNF-[Greek small letter alpha] levels. A possible mechanism may involve the stimulation of alveolar macrophages and release of preformed TNF-[Greek small letter alpha] by cellular products or debris caused by low-pH damage to the cells lining the airways or alveoli. This stimulation appears to be transient and does not result in appreciable upregulation of TNF-[Greek small letter alpha] mRNA. This is consistent with the limited lung injury demonstrated by the low-pH model if there are no other additional insults.

Treatment of animals with anti-TNF-[Greek small letter alpha] antibody that had been injured by instillation of gastric particles into the lungs demonstrated no significant changes in lung protein permeability, gas-exchange efficiency, or neutrophil lung infiltration. These results are different than those obtained in a lung-injury study using a model of inhalation of quartz particles model. [13] In that study, rats pretreated with anti-TNF-[Greek small letter alpha] antibody have diminished infiltration of neutrophils into the lungs compared with rats pretreated with nonimmune IgG. The different types and size of particle, as well as the timing of the measurements (24 h vs. 5 h after injury in our experiments), may be the reasons for the differing results.

Considering the robust nature of the TNF-[Greek small letter alpha] response in both gastric-particle models, it is surprising that inhibition of the bioactivity of the TNF-[Greek small letter alpha] did not decrease the lung injury in either group of animals. There are several possible explanations for these findings. First, the experimental conditions may not have been able to prevent the biologic effects of a sustained release of TNF-[Greek small letter alpha] with the antibody strategy that we employed. To demonstrate the requirement for TNF-[Greek small letter alpha] in the pathogenesis of the gastric-particulate injury may require larger dosages or continuous administration of the antibody.

Second, proinflammatory cytokine cascades have a significant amount of redundancy. Because of the overlapping actions of the many inflammatory cytokines, interpretation of data derived from inhibition studies may not be straightforward. Clinically, this can lead to many of the failures associated with therapies based on experimental data associated with cytokine cascades in disease (e.g., strategies to improve outcomes in sepsis). [42] This may be particularly important when the stimulus is multifactorial (e.g., low pH and particulate). The most obvious and direct example of this is the overlapping responses of IL-1 [Greek small letter beta] and TNF-[Greek small letter alpha]. These initiation cytokines have very similar actions in inducing subsequent cytokines that lead to inflammatory effector cell infiltration. [43] Additionally, inhibiting the bioactivity of one proinflammatory cytokine may lead to the overexpression of another. Thus, concurrent use of several blocking agents (e.g., anti-IL-1 [Greek small letter beta] and TNF-[Greek small letter alpha]) may be required to demonstrate overlapping actions in the pathogenesis of the lung injuries.

Finally, TNF-[Greek small letter alpha] also has a number of potentially beneficial effects involving the injury associated with the inflammatory response. This may also be why therapeutic strategies to decrease the bioactivity of TNF-[Greek small letter alpha] clinically in systemic inflammatory injury syndromes has failed. For example, sustained TNF-[Greek small letter alpha] is an important stimulant of IL-10, a cytokine that can prevent some of the untoward effects of acute inflammation and shift the inflammatory response to a less acute, chronic injury or reparative mode. [44] Additionally, TNF-[Greek small letter alpha] has been demonstrated to increase levels of antioxidants in the lung that are predicted to protect the lung from injury associated with oxidant-mediated damage. [45] Interestingly, in several models of lung infection, increasing TNF-[Greek small letter alpha] levels has been shown to be protective locally. [14] Thus, decreasing TNF-[Greek small letter alpha] activity may directly decrease lung inflammation but also directly or indirectly decrease the innate antiinflammatory defenses from being upregulated, leading to no detectable differences in lung injury.

An interesting difference between the acidic injury and the particulate injury emerged in the blood-gas data. Gastric particulate pulmonary injury produced a minimal (11%) reduction in oxygen-exchange efficiency, as measured by Pa^O(2)/FI^O(2), whereas the effect of acid aspiration was quite substantial, a 57% reduction (Figure 5). This was despite the fact that the acid- and particulate-injury regimens used in this study produced the same increase in pulmonary protein permeability (Figure 4).

Clearly, the mechanism of lung-function impairment (with regard to blood oxygenation) is different between the two gastric components and cannot be explained merely in terms of alveolar edema impeding gas diffusion. If this were the case, the pulmonary protein permeability measurements would have correlated with the gas-exchange indexes for the two different aspirate modalities. Alterations in the normal hypoxic pulmonary vasoconstriction response may offer a putative mechanism. As a result of the injury insult, acid, or particulate, the integrity of the alveolar-capillary barrier is compromised, producing proteinaceous pulmonary edema at the site of insult exposure. The resulting decrease in oxygen exchange would normally lead to a decrease in blood flow to the pulmonary vasculature of the injured area. This response prevents venous admixture and is controlled by endothelial vasomediators. [46–48] In gastric particulate aspiration-induced inflammation this mechanism appears to be intact, whereas in the acid-injured lung, it does not. Stimulation of vasodilating mediators (i.e., NO) by the acid insult, or inhibition of vasoconstricting mediators (i.e., thromboxanes), may result in venous admixture. Likely sources of these mediators would be the vascular endothelium, alveolar macrophages, and neutrophils marginated in the pulmonary vasculature. [41]

Previously, we have demonstrated that aspiration of gastric particles suspended in an acid solution leads to a resultant injury to the alveolar-capillary boundary, as measured by protein leakage into the lung, that is synergistic compared with the two injury modalities alone. [11] The results of this study corroborate these previous findings (Figure 4) and also demonstrate a synergistic decrease in gas-exchange efficiency when the two gastric-aspirate components are combined (Figure 5). These results suggest that the combined injury insult produces such an overwhelming pulmonary edema that either blood oxygenation cannot be improved by hypoxic pulmonary vasoconstriction or the normal response mechanism is inhibited by the acid exposure.

Clearly, aspiration of gastric components produces a sequela of pulmonary inflammation that is dependent on the constituents of the aspirant. Aspiration of stomach acid, gastric particles, or their combination produces inflammatory injuries that are uniquely manifested and have different underlying mechanisms driving their pathogenic and reparative courses. This is demonstrated in this study by the lack of correlation between lung protein permeability and gas- exchange efficiency changes induced by particulate aspiration injury, whereas these lung-injury parameters mirror each other in response to acid alone or acidic particulate aspiration. The synergistic increase in lung injury produced by the combination of the acidic and particulate components of gastric contents may, in part, be responsible for the potential for development of ARDS following an aspiration event. Interestingly, although elaboration of the proinflammatory cytokine, TNF-[Greek small letter alpha], has been demonstrated to occur with each of the tested aspiration modalities, these changes in TNF-[Greek small letter alpha] expression or presentation are not reflected in the resultant lung-injury parameters tested. Therefore, TNF-[Greek small letter alpha] does not appear to be the singular mitigating factor producing the synergistic increase in lung injury. However, the evidence from this study does not exclude the possibility that the interaction of TNF-[Greek small letter alpha] with other inflammatory cascade components may be responsible for that process.

Finally, there are other potential risk factors that can confound the course of events. Introduction of throat-colonizing bacteria into the lung either by the aspiration event itself or intubation presents an entirely different stimulus for the immune/inflammatory system to deal with. Hyperoxia exposure may pose a risk, as well. We previously demonstrated that exposure to as little as 50% oxygen for 5 h or 98% oxygen for 2 h produced an exacerbation of the lung injury in this rat model of acid aspiration. Exposing the injured lung to high concentrations of oxygen, as is often the case in the critical-care setting, can add an additional insult to the preexisting ones and aggravate the situation with hyperoxic modification of the body's defenses. [25]

Elucidating the cytokine cascades involved in "directing" each inflammatory response will clearly help us understand why some gastric-aspiration victims successfully recover on their own and why others develop ARDS. This knowledge can help us develop therapies to prevent this potentially lethal downward spiral.