- Email: [email protected]
Cytokines and the pathogenesis of nosocomial pneumonia
Cytokines and the pathogenesis of nosocomial pneumonia Steven G. Muehlstedt, MD, Chad J. Richardson, MD, Michael A. West, MD, PhD, Mark Lyte, PhD, and Jorge L. Rodriguez, MD, Minneapolis, Minn
Background. Nosocomial pneumonia (NP) in injured patients is a significant clinical problem. We hypothesize that the pathogenesis of NP in injured patients involves an imbalanced cytokine response within the alveolar airspace that may inhibit effector cell function. Methods. Proinflammatory (IL-8) and anti-inflammatory (IL-10) levels were measured in bronchoalveolar lavage (BAL) fluid from multitrauma patients on admission, 24, 48, and 72 hours post-injury and following lipopolysaccharide (LPS) induction of alveolar cells. Patients were compared based on IL-8 levels and the development of NP. Results. A high level of IL-8 on admission was associated with the development of NP. In addition, levels of IL-8 were significantly greater in NP-positive patients at all time points. The IL-10 levels decreased from admission values in NP-negative patients but increased in NP-positive patients. Furthermore, a high level of IL-10 ( > 120 pg/mL) at 72 hours post-injury was associated with the development of NP. Alveolar cells from NP-positive patients produced significantly more IL-10 in response to LPS than cells from NP-negative patients. Conclusions. The pathogenesis of NP in injured patients involves an early and severe IL-8 process within the lung followed by an exaggerated IL-10 response that may inhibit effector cell function. (Surgery 2001;130:602-11.) From the Department of Surgery, Hennepin County Medical Center, affiliated with the University of Minnesota Medical School
MULTI-TRAUMA PATIENTS REQUIRING mechanical ventilation represent a unique subset of critically ill patients in which the incidence of nosocomial pneumonia (NP) approaches 45%. Despite the identification of at-risk patients, NP remains prevalent and a leading cause of morbidity.1 The failure of current prophylactic therapy likely is due to an incomplete understanding of the disease process. As such, research directed at defining the pathogenesis of NP within the trauma setting is paramount. With this in mind, we previously observed that traumatic injury induces a pro-inflammatory (IL-8) cytokine response within the alveolar airspace of injured patients.2,3 Furthermore, IL-8 production occurs locally within the lung and correlates with the influx of inflammatory cells, pulmonary dysfunction, and the development of NP.2 Presented at the 58th Annual Meeting of the Central Surgical Association, Tucson, Ariz, March 7-10, 2001. Reprint requests: Jorge L. Rodriguez, MD, Hennepin County Medical Center, Department of Surgery, 701 Park Ave S, Minneapolis, MN 55415. Copyright © 2001 by Mosby, Inc. 0039-6060/2001/$35.00 + 0 11/6/117105 doi:10.1067/msy.2001.117105
602 SURGERY
Pro- and anti-inflammatory cytokines have been measured in the serum of injured patients.4 It is assumed these mediators regulate one another to avoid organ injury and dysfunction. Previously we have demonstrated that bronchoalveolar lavage (BAL) fluid containing large amounts of IL-8 inhibits effector cell chemoattraction and superoxide production implying the presence of an inhibitory factor.5 A potent anti-inflammatory cytokine (IL-10) has been shown to inhibit a variety of effector cell functions including the production of proinflammatory cytokines (TNF-α, IL-1, IL-6, and IL-8) and the ability to present antigen.6,7 We hypothesize that the pathogenesis of NP in injured patients involves an imbalanced cytokine response within the lung that may inhibit effector cell function. MATERIAL AND METHODS The Internal Review Board of Hennepin County Medical Center approved this protocol for research involving human subjects. Multitrauma patients admitted within 2 hours of injury who required mechanical ventilation were included in the study provided there was no history of infection or aspiration. Demographic
Surgery Volume 130, Number 4
data that were collected included age, gender, pulmonary comorbidity (active smoker, asthma, chronic obstructive pulmonary disease) and nonpulmonary comorbidity. Also assessed were the mechanism of injury, injury severity score (ISS), the presence of head injury, severity of chest injury as assessed by the chest anatomical injury score (AIS), hypotension (systolic blood pressure < 90 mm Hg) or emergent intubation in the field. Data collected upon admission to the surgical intensive care unit (SICU) included systolic blood pressure, temperature, blood pH, Pa O2, PaCO2, serum bicarbonate concentration (HCO3), ratio of arterial oxygen tension to inspired oxygen concentration (PaO2/FiO2), mode of ventilation, tidal volume, positive end-expiratory pressure (PEEP), plateau pressure (Pplat), static compliance (tidal volume/Pplat), and lung injury score (LIS).8 In the SICU, information recorded daily included the PaO2/FiO2 ratio, static lung compliance, LIS, antibiotics use, and blood product transfusion. Acute lung injury was defined as PaO2/FiO2 less than 300 mm Hg regardless of PEEP. The simultaneous presence of 4 or more of the following criteria within 8 days of admission were used to define NP: (1) temperature > 101.5°F; (2) white blood cell count > 15,000/mm3; (3) inflammatory cells (> 25 WBC/low power field and < 10 epithelial cells/low power field) and bacteria with Gram stain; (4) positive quantitative sputum culture; (5) new or changing infiltrate on chest x-ray (CXR). BAL samples were obtained through the endotracheal tube using sterile technique, at admission, 24, 48, and 72 hours post-injury. A suction catheter (Medline Industries, Mundelein, Ill) was wedged, and 10 mL of sterile saline solution (0.9% NaCl, 37°C) was instilled, aspirated with suction, and discarded. Next, 30 mL of sterile saline solution was instilled and aspirated into a sterile specimen collector (Allegiance Healthcare, McGaw Park, Ill). Recovery rate was 82% ± 10%. Samples were chilled at 4°C immediately and processed within 10 minutes of collection. For each sample, a Cytospin (Shandon, Pittsburgh, Pa) slide was prepared, stained with Wright’s stain, and a differential obtained. Samples were strained through a 60-mesh steel screen (Sigma, St Louis, Mo) to remove mucus and centrifuged at 3000 rpm for 5 minutes at 4°C. Cell-free supernatant was removed and stored at –70°C for cytokine analysis (baseline values). The cell pellet was then re-suspended in phosphatebuffered saline solution without calcium (Gibco, Rockville, Md). White blood cells were counted by hemacytometer and viability was assessed by trypan blue exclusion (Sigma). All samples were more
Muehlstedt et al 603
than 95% viable. RPMI 1640 medium (Gibco) supplemented with 10% pooled human AB serum (Atlanta Biologicals, Norcross, Ga), 100 U/mL penicillin, and 100 g/mL streptomycin (Gibco) was added to re-suspended cells achieving a final concentration of 106 WBC/mL. One and a half mL of this solution (1.5 million cells) were cultured for 24 hours at 37°C in 5% CO2, in the presence or absence of 10 ng/mL Escherichia coli 0111:B4 lipopolysaccharide (LPS; Sigma). Following incubation the samples were stored at –70°C pending analysis. Supernatants were assayed for IL-8 and IL10 using commercial enzyme-linked immunosorbent assay kits according to manufacturer’s directions (OptEIA, BD PharMingen, San Diego, Calif). Detection limits for IL-8 and IL-10 were 3.1 pg/mL and 7.8 pg/mL, respectively. Values below these levels were considered zero for statistical analysis. An anti-inflammatory (AI) index was calculated daily for each patient. This was defined as the ratio of baseline IL-10 (ng/mL) to baseline IL-8 (ng/mL), multiplied by 1000. Outcome data included duration of mechanical ventilation, the development of acute respiratory distress syndrome (ARDS; defined as PaO2/FiO2 < 200 mm Hg regardless of PEEP, bilateral CXR infiltrates, and pulmonary capillary wedge pressure ≤ 18 mm Hg), intensive care unit length of stay, hospital length of stay, and mortality. Patients were categorized for analysis into those who developed NP (NP-positive) and those who did not (NP-negative), as well as by IL-8 levels on admission (> 10,000 and < 10,000 pg/mL). Continuous variables were examined by Student t test and expressed as mean + SD. Binary data were assessed by chi-square analysis and expressed as a percentage. Cytokine data, analyzed by the MannWhitney U test, were reported as mean ± standard error of the mean. A P value < .05 was considered significant. RESULTS A total of 49 patients met criteria for the study. The patient population primarily was young (42 + 21 years) males (81.6%) injured in blunt trauma (96%) with significant injury (ISS 31.5 ± 9.1). Of the 49 patients, 22 (44%) developed NP within 8 days of admission. Average time to diagnosis was 5 ± 1.3 days. Microbial isolates primarily were gramnegative pathogens (60%), including Pseudomonas, Haemophilus influenzae, Enterobacter, and Klebsiella in decreasing order of frequency. Overall the most common individual isolate was Staphylococcus aureus, accounting for 31% of pathogens. All NPpositive patients had an inflammatory Gram stain,
604 Muehlstedt et al
Surgery October 2001
Table I. Demographic data for pneumoniapositive and pneumonia-negative patients
Age (years) Gender (male) Mechanism of injury (blunt) ISS Head injury Chest AIS Pulmonary comorbidity Nonpulmonary comorbidity
Pneumoniapositive (n = 22)
Pneumonianegative (n = 27)
43 ± 22 73% 100%
40 ± 21 89% 93%
33 ± 9 68% 2.0 ± 1.5 9%*
31 ± 10 52% 1.5 ± 1.7 37%
36%
33%
ISS, Injury severity score; AIS, anatomical injury score. *P < .05.
a positive sputum culture, and a new or changing infiltrate on CXR. All patients satisfied at least 4 criteria and 82% satisfied 5 criteria. Demographically, NP-positive patients were similar to NP-negative patients, the only exception being more pulmonary comorbidity among the NP-negative group (Table I). Prior to hospital admission, NP-positive patients were more often emergently intubated (18% vs 4%, P < .05) and hypotensive (68% vs 15%, P < .05). There were no clinical differences between groups on admission to the SICU (Table II). Within the first 72 hours of injury, patients who ultimately developed pneumonia experienced significantly more acute lung injury than patients who did not (59% vs 30%, P < .05). In addition, whereas the static compliance of NP-negative patients improved, that of NP-positive patients worsened. After admission, the LIS was consistently greater in NP-positive patients than NP-negative patients, reaching statistical significance at 24, 48, and 72 hours post-injury (Table III). The NP-positive group received significantly fewer prophylactic antibiotics than NP-negative patients (55% vs 81%, P < .05), however the incidence of blood transfusions was similar between groups (55% and 56%, respectively). The IL-8 concentration in BAL supernatant (baseline) was compared between groups. Significantly greater levels of IL-8 were measured in the BAL of NP-positive patients at all time points. In addition, the BAL of NP-positive patients contained a significantly greater percentage of neutrophils on admission (87% vs 49%, P < .05) and at 24 hours post-injury (92% vs 81%, P < .05). Also of interest, IL-8 levels were highest on admission and decreased thereafter both within each group and individually (Fig 1).
Table II. Admission clinical data for pneumoniapositive and pneumonia-negative patients Pneumoniapositive (n = 22) Systolic blood 111 ± 18 pressure (mm Hg) Temperature (°F) 97.2 ± 1.8 pH 7.32 ± 0.09 PaO2 (mm Hg) 206.1 ± 132.0 PaCO2 (mm Hg) 37.7 ± 8.9 HCO3 (mEq/L) 19.6 ± 5.0 PaO2/FiO2 ratio 360 ± 189 AC ventilation 95% Tidal volume 11.4 ± 0.6 (mL/kg) PEEP (cm H2O) 4.6 ± 2.3 Pplat (cm H2O) 22.7 ± 6.7 Compliance 47.9 ± 14.3 (mL/cm H2O) LIS 1.4 ± 0.8
Pneumonianegative (n = 27) 127 ± 21 97.6 ± 1.2 7.32 ± 0.08 194.3 ± 88.7 40.4 ± 10.1 21.1 ± 7.7 382 ± 218 96% 11.3 ± 0.4 4.5 ± 2.5 21.9 ± 5.1 54.4 ± 15.9 1.2 ± 0.8
AC, Assist control; PEEP, positive end-expiratory pressure; Pplat, plateau pressure; Compliance, static compliance; LIS, lung injury score.
When admission IL-8 levels categorized patients, 24 patients had a high IL-8 level (> 10,000 pg/mL) and 25 patients had a low IL-8 level (< 10,000 pg/mL). A high IL-8 level was associated strongly with the development of NP (76% of high IL-8 patients vs 12.5% of low IL-8 patients, P < .001). The static compliances of high and low IL-8 patients were similar on admission (56 + 16 vs 58 + 19 mL/cm H2O), however by 24 hours post-injury the static compliance of high IL-8 patients was significantly less than low IL-8 patients (48 + 18 vs 57 + 15 mL/cm H2O, P < .05), and this relationship persisted at 48 hours (48 + 14 vs 51 + 18 mL/cm H2O) and 72 hours post-injury (47 + 13 vs 55 + 18 mL/cm H 2 O). Similarly, while the PaO2/FiO2 ratios of both the high and low IL-8 groups were similar on admission (361 + 170 vs 385 + 160), the ratio among high IL-8 patients was much lower at 24 hours (283 + 127 vs 380 + 180), 48 hours (298 + 165 vs 328 + 148), and 72 hours (261 + 132 vs 289 + 110) post-injury. Finally, a high IL-8 level (> 10,000 pg/mL) on admission was associated strongly with a high IL-10 level (> 120 pg/mL) 72 hours post-injury (P < .05), and a high IL-10 level on day 3 correlated with the development of NP (92% of high IL-10 patients vs 18% of low IL-10 patients, P < .05). Baseline IL-10 was compared between groups (Fig 2). Following admission, IL-10 levels were significantly greater in NP-positive patients at all time points. Furthermore, whereas IL-10 levels
Muehlstedt et al 605
Surgery Volume 130, Number 4
Table III. Pulmonary physiology data for pneumonia-positive and pneumonia-negative patients Admission Static compliance Pneumonia + Pneumonia – PaO2/FiO2 ratio Pneumonia + Pneumonia – Lung injury score Pneumonia + Pneumonia –
24 hours
48 hours
72 hours
48 ± 14 54 ± 16
47 ± 16 53 ± 17
48 ± 18 52 ± 17
45 ± 12* 58 ± 17
360 ± 189 382 ± 218
310 ± 138 363 ± 186
275 ± 125 351 ± 169
263 ± 107 295 ± 133
1.4 ± 0.8 1.2 ± 0.8
1.8 ± 0.7* 1.2 ± 0.8
2.0 ± 0.7* 1.3 ± 0.8
2.1 ± 0.6* 1.4 ± 1.1
Pneumonia +, pneumonia-positive; pneumonia –, pneumonia-negative. *P < .05.
Fig 1. Levels of the pro-inflammatory cytokine IL-8 measured in the bronchoalveolar fluid of patients who develop pneumonia and those who do not.
Fig 2. Levels of the anti-inflammatory cytokine IL-10 measured in the bronchoalveolar fluid of patients who develop pneumonia and those who do not.
decreased from admission values in the NP-negative group, levels steadily increased in NP-positive patients, reaching a maximum at 72 hours. The AI index was compared between groups, and similar to IL-10, AI index values diverged. While the AI index of NP-negative patients fell, the AI index of NP-positive patients rose (Fig 3). The IL-8 level produced by BAL cells following LPS stimulation was compared (Fig 4). Initially, white blood cells harvested from NP-positive patients produced significantly more IL-8 than cells from NPnegative patients. At 48 and 72 hours, however, levels were similar. Levels of IL-10 produced in response to LPS were compared in Fig 5. Whereas initial levels were similar between groups, the amount of IL-10 produced by NP-positive patients was significantly greater at 48 and 72 hours post-injury. The incidence of ARDS in the overall population was 6% and statistically was not different between groups. On average, NP-positive patients had more ventilator days and greater intensive care unit length of stay and hospital length of stay than NP-negative patients with similar injuries. Overall
mortality was 12% (6/49) and was not statistically different between groups (Table IV). Causes of death were severe head injury (1 NP-positive, 3 NPnegative) and multiple organ system failure (1 NPpositive, 1 NP-negative). DISCUSSION The presence of NP in mechanically ventilated trauma patients remains problematic despite improvements in the care of critically ill patients. The 44% incidence in the current study, while significant, is consistent with published reports.1,9 Furthermore, this study supports that patients who develop NP are more often emergently intubated, hypotensive, and victims of head trauma.1 This study identifies distinct cytokine trends that precede NP within the lungs of injured patients. Individuals who develop NP have significantly greater levels of IL-8 within their alveolar airspace, the highest of which are within 2 hours of injury. Moreover when admission IL-8 levels categorized patients, a strong association was observed between a high level of IL-8 and the subsequent develop-
606 Muehlstedt et al
Fig 3. The anti-inflammatory index (ratio of IL-10 to IL-8, multiplied by 1000) of patients who develop pneumonia and those who do not.
Fig 4. Levels of IL-8 produced following LPS stimulation of bronchoalveolar lavage white blood cells from pneumonia-positive and pneumonia-negative patients.
ment of pulmonary dysfunction and NP. The IL-8 is a potent neutrophil activator and chemoattractant and this is consistent with the greater percentage of neutrophils seen in admission samples from NPpositive patients.10,11 The pulmonary dysfunction that also is more prevalent among this group likely is initiated by this process. In fact, the presence of anti-IL-8 antibodies in BAL fluid has been associated with survival in human ARDS patients, and IL-8 inhibition has been shown protective of acute lung injury in animal models of trauma.12,13 The LIS, statistically greater in the NP-positive group following admission, may be a valuable clinical tool for identifying this exaggerated pro-inflammatory response to trauma.8 As mentioned, the greatest IL-8 levels were measured within 2 hours of injury. This implies that the initiating event is the trauma itself and not the
Surgery October 2001
Fig 5. Levels of IL-10 produced following LPS stimulation of bronchoalveolar lavage white blood cells from pneumonia-positive and pneumonia-negative patients.
result of mechanical ventilation or transfusions. Although induced by trauma, the actual source of IL-8 remains debatable. We previously have demonstrated the presence of IL-8 mRNA within the BAL of injured patients, implying it is locally produced and is not initiated in the systemic circulation.2,3 Furthermore, the ability of BAL cellular contents to produce substantial amounts of IL-8 in response to LPS indicates that these cells have the capacity to produce the IL-8 levels found after trauma. It also is interesting that as more polymorphonuclear leukocytes emigrate into the alveolar airspace, basal IL-8 levels fall. For these reasons, we believe the primary source of IL-8 is the activated tissue macrophage and not the polymorphonuclear leukocyte population or activated endothelium, which have been suggested.14 Increasing levels of IL-10 are found within the alveolar airspace of NP-positive patients such that at 72 hours post-injury, high levels of IL-10 correlate with the development of pneumonia. This is in contrast to NP-negative patients, whose levels of this anti-inflammatory cytokine decrease from admission values. Furthermore, by 48 hours postinjury, alveolar cells from NP-positive patients produce significantly more IL-10 following LPS stimulation than cells from NP-negative patients. The fact that alveolar cells from NP-positive patients initially produce more IL-8 in response to LPS and later more IL-10 may indicate a shift of these cells to an anti-inflammatory state. This is reflected by the increased anti-inflammatory ratio of these patients. In the lung, IL-10 is produced primarily by the alveolar macrophage.15 Originally described as cytokine synthesis inhibitory factor, IL-10 has been shown to inhibit a variety of effector cell func-
Muehlstedt et al 607
Surgery Volume 130, Number 4
Fig 6. Schematic representation of the lung’s response to injury. Trauma induces a pro-inflammatory process within the lung that is balanced by an anti-inflammatory response. If exaggerated, high levels of IL-8 recruit and activate neutrophils that cause acute lung injury. In contrast, high levels of IL-10 inhibit effector cell function, predisposing patients to pneumonia.
Table IV. Outcome data for pneumonia-positive and pneumonia-negative patients
ARDS Ventilated days Intensive care days Hospital days Mortality
Pneumoniapositive (n = 22)
Pneumonianegative (n = 27)
9% 15 ± 10 19 ± 12 26 ± 15 9%
4% 9 ± 11 12 ± 13 20 ± 16 15%
ARDS, Acute respiratory distress syndrome.
tions including IL-8 synthesis.6 In addition to the inhibition of cytokine production, however, IL-10 down-regulates the expression of human leukocyte antigen DR on effector cells, diminishing their capacity to recognize and present antigen.7 As such, IL-10 diminishes the ability of pulmonary macrophages to ingest and kill bacteria and has been correlated with infection in injured patients.16,17 In an animal model of pneumonia, IL-10 attenuated the pro-inflammatory cytokine response within the lung but hampered effective clearance of bacteria.18 Furthermore, since attempts to eradicate bacteria both systemically and locally in human patients have not reduced substantially the incidence of NP, then perhaps it is not primarily the microbial load that is the problem, but rather the capacity of the local organ to
effectively clear bacteria.19 We believe this posttraumatic dysfunction exists and is mediated in part by IL-10. It is reasonable to suspect that IL-10 levels are reflective of the initial pro-inflammatory process, whereby more IL-10 is produced in response to more IL-8. This seems reasonable, as high levels of IL-8 on admission are associated with high levels of IL-10 measured 3 days post-injury. This however does not explain why IL-10 levels continue to increase in NP-positive patients and decrease in NP-negative patients, both in the face of decreasing IL-8 levels. It appears that the IL-10 response of NP-positive patients is exaggerated. It is possible that when a critical IL-8 level is reached within the alveolar airspace, a pronounced IL-10 response results. Another possibility is that certain individuals have unique cellular or endothelial populations, or perhaps, are genetically predisposed to such cytokine production following trauma. Catecholamines recently have been found to induce de novo IL-10 synthesis by monocytes.20 This is interesting because NP-positive patients more often were hypotensive in the field, and although speculative, they may have experienced a greater release of endogenous catecholamines following injury. Why certain individuals display an exaggerated alveolar cytokine response to trauma remains unknown. Demographically there were no differ-
608 Muehlstedt et al
ences in age, gender, mechanism of injury, or injury severity among patients who developed pneumonia and those who did not. The groups also were similar in terms of the incidence and severity of chest injury. Only pulmonary comorbidity differed between groups, the majority of which was cigarette smoking. Pulmonary comorbidities such as smoking, asthma, and chronic obstructive pulmonary disease are well-defined risk factors for communityacquired pneumonia and pneumonia in critically ill medical patients but not among the trauma population. In fact, this study identified more pulmonary comorbidity among patients that did not develop pneumonia. This would seem to indicate that although pneumonia can develop in multiple populations, the pathogenesis among injured patients might be unique. In cigarette smokers, the ability of alveolar macrophages to produce cytokines is reduced although the ability to phagocytize and kill bacteria is maintained.21,22 More pulmonary comorbidity among NP-negative patients therefore is consistent with the hypothesis that an exaggerated cytokine response is responsible for the predisposition to infection in injured patients. Antibiotic interaction with cytokine production is described in the literature. Various antibiotics have been shown to both diminish and augment the proinflammatory cytokine response of effector cells to a stimulus.23,24 Most studies involve systemic cells, unfortunately, and not those of the alveolar airspace. Regardless, initial BAL samples were collected prior to systemic antibiotic prophylaxis in most cases, and it is the IL-8 level of these samples that correlates strongly with the development of NP. The impact of antibiotics on subsequent cytokine measurements is difficult to answer, however we found no correlation between antibiotic usage and cytokine levels at any time point (data not shown). As with any clinical study, this study has flaws. First, we have chosen a very select patient population and conclusions may not apply to less injured patients. Although the initial proinflammatory response is most likely present prior to the institution of mechanical ventilation, it is not possible for this study to determine the effect of mechanical ventilation on later cytokine production. Second, in regards to the diagnosis of NP, we made no attempt to compare our culture methods to protected specimen brush technique. Also, although sterile technique was implemented, it is possible that lavage may have induced a local cytokine response or introduced bacteria. Third, the use of an ex-vivo model and LPS in nanogram concentrations is not physiologic. Finally, the nature of
Surgery October 2001
cytokines themselves, with their short half-lives and complex interactions, undoubtedly complicate analysis. For these reasons it is imperative to standardize sampling intervals, technique, and analysis. Given these limitations, this study’s goal was not to identify isolated pneumonia predictors, but rather to identify the dynamic trends of cytokine production that precede development. Trauma induces an IL-8 cytokine response in the lungs of injured patients. This pro-inflammatory response appears exaggerated in certain individuals, and it is associated with pulmonary inflammation and dysfunction evident by the influx of neutrophils, decreased compliance, and early acute lung injury. Teleologically it seems reasonable that this pro-inflammatory state induces an anti-inflammatory response to attenuate the process. If this anti-inflammatory response is similarly exaggerated, IL-10 levels may be capable not only of inhibiting cytokine production, but also the ability of effector cells to recognize and process antigen as well, a state referred to as immunoparalysis.25 We believe this model is accurate (Fig 6). As such, future study should identify individuals predisposed to these responses and correlate cytokine levels with effector cell function to better define the disease process in these patients. REFERENCES 1. Rodriguez JL, Gibbons KJ, Bitzer LG, Dechert RE, Steinberg SM, Flint LM. Pneumonia: incidence, risk factors, and outcome in injured patients. J Trauma 1991;31:907-14. 2. Rodriguez JL, Miller CG, DeForge LE, Kelty L, Shanley CJ, Bartlett RH, Remick DG. Local production of interleukin-8 is associated with nosocomial pneumonia. J Trauma 1992;33:74-82. 3. Keel M, Ecknauer E, Stocker R, Ungethum U, Steckholzer U, Kenney J, Gallati H, et al. Different pattern of local and systemic release of proinflammatory and anti-inflammatory mediators in severely injured patients with chest trauma. J Trauma 1996;40:907-14. 4. Sherry RM, Cue JL, Goddard JK, Parramore JB, DiPiro JT. Interluekin-10 is associated with the development of sepsis in trauma patients. J Trauma 1996;40:613-7. 5. Rodriguez JL, Moore NP, Miller CG, Cox C, Garner WL, Smith DJ, Remick DG, et al. Traumatic injury: the local organ (lung) alters cell function. Am Coll Surg 1993; Surgical forum volume XLIV;124. 6. Fiorentino DF, Zlotnik A, Mosmann TR, Howard M, O’Garra A. IL-10 inhibits cytokine production by activated macrophages. J Immunol 1991;147:3815-22. 7. de Waal Malefyt R, Haanen J, Speits H, et al. Interleukin 10 (IL-10) and viral IL-10 strongly reduce antigen-specific human T cell proliferation by diminishing the antigen-presenting capacity of monocytes via downregulation of class II major histocompatibility complex expression. J Exp Med 1991;174:915-24. 8. Murray JF, Matthay MA, Luce JM, Flick MR. Pulmonary perspectives: an expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988;138:720-3.
Surgery Volume 130, Number 4
9. Fabian TC, Boucher BA, Croce MA, et al. Pneumonia and stress ulceration in severely injured trauma patients: a prospective evaluation of the effects of stress ulcer prophylaxis. Arch Surg 1993;128:185-92. 10. Baggiolini M, Walz A, Junkel SL. Neutrophil-activating peptide-1/interleukin 8, a novel cytokine that activates neutrophils. J Clin Invest 1989;84:1045-9. 11. Kunkel SL, Standiford T, Kasahara K, Strieter RM. Interleukin-8: the major neutrophilic chemotactic factor in the lung. Exp Lung Res 1991;17:17-23. 12. Kurdowska A, Miller EJ, Noble JM, et al. Anti-IL-8 autoantibodies in alveolar fluid from patients with the adult respiratory distress syndrome. J Immunol 1996;157:2699-706. 13. Sekido N, Mukaida N, Harada A, et al. Prevention of lung reperfusion injury in rabbits by a monoclonal antibody against interleukin-8. Nature 1993;365:654-7. 14. Schroder JM, Christophers E. Secretion of novel and homologous neutrophil-activating peptides by LPS-stimulated human endothelial cells. J Immunol 1989;142:244-51. 15. Armstrong L, Millar AB. Relative production of tumor necrosis factor a and interleukin 10 in adult respiratory distress syndrome. Thorax 1997;52:442-6. 16. Steinhauser ML, Hogaboam CM, Kunkel SL, Lukacs NW, Strieter RM, Standiford TJ. IL-10 is a major mediator of sepsis-induced impairment in lung antibacterial host defense. J Immunol 1999;162:392-9. 17. Sachse C, Prigge M, Cramer G, Pallua N, Henkel E. Association between reduced human leukocyte antigen (HLA)-DR expression on blood monocytes and increased plasma level of interleukin-10 in patients with severe burns. Clin Chem Lab Med 1999;37:193-8. 18. van de Poll T, Marchant A, Keogh CV, Goldman M, Lowry SF. Interleukin-10 impairs host defense in murine pneumococcal pneumonia. J Infect Dis 1996;174:994-1000. 19. Gastinne H, Wolff M, Delatour F, Faurisson F, Chevret S. A controlled trial in intensive care units of selective decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Med 1992;326:594-9. 20. Platzer C, Docke W-D, Volk H-D, Prosch S. Catecholamines trigger IL-10 release in acute systemic stress reaction by direct stimulation of its promoter/enhancer activity in monocytic cells. J Neuroimmol 2000;105:31-8. 21. McCrea KA, Ensor JE, Nall K, Bleecker ER, Hasday JD. Altered cytokine regulation in the lungs of cigarette smokers. Am J Respir Crit Care Med 1994;150:696-703. 22. Baldwin GC, Tashkin DP, Buckley DM, Park AN, Dubinett SM, Roth MD. Marijuana and cocaine impair alveolar macrophage function and cytokine production. Am J Respir Crit Care Med 1997;156:1606-13. 23. Oishi K, Sonoda F, Kobayashi S, Iwagaki A, Nagatake T, Matsushima K, Matsumoto K. Role of interleukin-8 (IL)-8 and an inhibitory effect of erythromycin on IL-8 release in the airways of patients with chronic airway diseases. Infect Immun 1994;62:4145-52. 24. Morikawa K, Watabe H, Araake M, Morikawa S. Modulatory effect of antibiotics on cytokine production by human monocytes in vitro. Antimicrob Agents Chemother 1996;40:1366-70. 25. Volk H-D, Reinke P, Docke W-D. Clinical Aspects: from systemic inflammation to ‘immunoparalysis’. In Jack RS (ed). Chem Immunol. Basel: Karger;2000;162-77.
DISCUSSION Dr Andrew R. Peitzman (Pittsburgh, Pa). It is an honor to discuss another in a series of important and well-designed studies investigating pulmonary response to injury and
Muehlstedt et al 609
infection from Dr Rodriguez and his group. In the interest of time I will not review what was an excellent presentation; however, I have a few comments and several questions. First, infection is an imbalance between host factors, local tissue injury, and virulence of the infecting organism. You have attempted to clarify the effects of local tissue injury for us in this study. Why did you choose to study only IL-8 and IL-10? Are we naive in doing so? We know TNF, IL-1, IL-6, and other cytokines are active as well. Do you have such data? Your earlier studies showed that local, meaning pulmonary, cytokine responses did not always correlate with systemic cytokine or messenger RNA levels. As many of your outcome criteria such as mortality, length of stay, and so on reflect systemic markers, do you have systemic levels of IL-8, IL-10, or their mRNA levels? Dr Muehlstedt. First was a question of why we chose to study IL-8 and IL-10. Our lab has studied a variety of other cytokines both systemically and within the alveolar airspace, including TNF, TNF soluble receptor, IL-1, IL-1 receptor antagonist, and IL-6. TNF was not included as levels decrease rapidly after injury and therefore are difficult to compare between patients admitted different times from injury. Both IL-1 and IL-6 were not included, because we previously demonstrated no correlation between these cytokines and pulmonary dysfunction or infection. Furthermore, although systemic levels of certain cytokines correlate with injury severity, we found no correlation between the magnitude of the cytokine response within the systemic circulation and that of the alveolar airspace, implying that cytokines within the alveolar airspace are not initiated in the systemic circulation. We acknowledge that IL-8 and IL-10 are not the only inflammatory cytokines involved in the lung’s cytokine response to trauma, however, we do believe that they are two of the primary mediators of this response and subsequent pulmonary dysfunction. We know this since, following trauma, anti-IL-8 has been shown protective of acute lung injury and IL-10 is involved in the resolution of pulmonary inflammation. Dr Andrew R. Peitzman. Second, you stated that pulmonary co-morbidity was greater in the pneumonia-free group. Yet the group with pneumonia had more acute lung injury, lower static compliance, higher lung injury scores, and longer time on the ventilator. Please reconcile this, since your observation seems counterintuitive. Dr Muehlstedt. The observation that patients who develop NP have significantly less pulmonary comorbidity is counterintuitive. If our hypothesis is correct, and the pathogenesis of nosocomial pneumonia involves an initially severe pro-inflammatory process, then it may be that a history of repetitive inflammatory insults to the lung, whether it be from cigarette smoke, asthma, or pulmonary disease, diminishes the capacity of pulmonary effector cells to mount an exaggerated pro-inflammatory
610 Muehlstedt et al
response. We are currently looking at this issue and hope to provide an answer soon. Dr Andrew R. Peitzman. Your AI index is an interesting use of new data. Is there any other data to corroborate its use? Is it statistically legitimate? If you simply look at these trends evaluating only IL-10 without your AI index, are your conclusions unchanged? Dr Muehlstedt. To my knowledge an AI index has not been described before. The fact that the AI index increases in patients who ultimately develop pneumonia is an important point of this paper as it may indicate a shift of alveolar cells to a primarily anti-inflammatory state. We have not studied this index prospectively but intend to do so in the future. Dr Andrew R. Peitzman. You state that you have not defined the source of IL-8 as tissue macrophage, PMN, or activated endothelium. Would immunohistochemistry do this? Dr Muehlstedt. Using both immunohistochemistry and PCR, we have previously demonstrated that alveolar macrophages are the primary producers of IL-8 following injury. We also think IL-10 is produced primarily by these cells and current literature supports this assumption. Dr Andrew R. Peitzman. In the discussion section of the manuscript, the authors suggest that more IL-10 is produced in response to greater IL-8 production initially. Is this truly cause-and-effect or simply overall up-regulation of the inflammatory cytokines? In your discussion you state, and I quote, “The fact that alveolar cells from pneumonia-positive patients initially produce more IL-8 in response to LPS and later more IL-10 may indicate a shift of these cells to an antiinflammatory state.” This is a critical observation and must be elaborated upon. The key to the risk of infection, multiple organ failure, and often survival in the trauma patient is this balance between the pro-inflammatory cytokines and the anti-inflammatory cytokines. This is both a temporal phenomenal and relative effect, as shown in this study. If the pro-inflammatory cytokines predominate for a prolonged period, the patient develops SIRS. If the anti-inflammatory cytokines predominate, CARS results, or, as the author stated in the paper, an immunoparalysis. This results in patients susceptible to infection, as shown in the study. Dr Raymond Pollak (Peoria, Ill). I thought this was a fascinating paper because it demonstrates that the host response to injury may well be dependent on the cytokine gene profile that exists in the genome. We recently have looked at this in terms of gender and ethnicity, and it is clear that when looking at cytokine gene polymorphisms, the way a person responds to injury, whether it is trauma, burn injury, or transplantation, may well be predicated by the subject’s cytokine gene profile. So I was curious as to whether you had looked at these, using PCR-based techniques, which
Surgery October 2001
are relatively easy to do? Some individuals may be predisposed to developing pro-inflammatory cytokine responses whereas others might be predisposed to developing the opposite response. Dr Muehlstedt. Why a certain subset of trauma patients develops an exaggerated IL-8 response to trauma is unknown. Many hypotheses can be made. We do plan on pursuing genetic studies in injured patients, as it may be that certain individuals are genetically predisposed to this response. Catecholamines also may be involved. Although we have not measured levels in these patients, catecholamines have been associated with a pro-inflammatory cytokine response and are capable of inducing IL-10 production. In this study, pneumonia patients more often were hypotensive and in need of vasopressors. It is possible that elevated levels of both endogenous and exogenous catecholamines are responsible for the increased levels of IL-10 found within the alveolar airspace of these patients. Dr Mark A. Malangoni (Cleveland, Ohio). Your premise is based on the fact that this abnormal cytokine profile is important in the development of pneumonia. I wonder if you examined your data looking at it at a different way, that is, evaluate patients who developed the abnormal cytokine profile—and I am assuming that there must be a few in your group who didn’t develop pneumonia. Did that change the results? For people who had the abnormal ratio, did all of them develop pneumonia? The second comment I had was that the occurrence of the pneumonias in this group was very early. What we see in these patients is a different bacterial flora as the cause for the pneumonia than in patients who develop pneumonia later on in their course. Did you try to correlate anything with the microbiology of this disease, and do you have any information about patients who develop late pneumonias? Dr Muehlstedt. Why we chose to categorize patients into two groups, pneumonia-positive and pneumonianegative, as opposed to isolating patients based on initial levels of IL-8 is a good question. When we separate patients into initially high and low IL-8 groups, we find that 85% of those with high levels develop pneumonia. The purpose of this study, however, was not to identify isolated cytokine predictors of nosocomial pneumonia as there are significant problems with doing this. The short half-lives of cytokines, variations in sampling intervals and technique, and the lack of standardization among ELISA kits makes current cytokine analysis unsuitable for human diagnosis. Therefore, although appealing, it would be unreasonable to conclude a certain level of IL8 is predictive of pneumonia. Our goal was to identify dynamic trends of cytokine production that precede pneumonia. By looking at cytokine trends over time, using strictly defined sampling intervals, the potential error of isolated measurements is minimized.
Surgery Volume 130, Number 4 Dr Thomas Vargish (Chicago, Ill). I have two questions, the first of which is a follow-up to the one you have just heard. Even though you have defined the time period for the change in the cytokine levels, how did that correlate with the timing of the onset of pneumonia in these patients? Did you see the pattern occurring in close correlation with those changes, or was it something that happened later on? The second one is the following: As I recall looking at your data, the group that got pneumonia seemed to be intubated earlier. I wonder whether or not this had to do with severity of injury and the possibility that they aspirated or their airway management was more complicated. What role did that play in the cytokine changes you saw and in the development of pneumonia? Dr Muehlstedt. Times from injury to intubation were not statistically different between groups. Furthermore, we excluded patients with pre-existing infections and those with documented aspiration, as those patients would undoubtedly complicate our analysis. Burn patients were not included given the high incidence of inhalation injury in this group. Dr James Tyburski (Detroit, Mich). I have one question, can you elaborate a little bit on any primary lung injuries? In some of the data that I am going to present later, we looked at IL-6 in these patients, and it was clear that the production was much higher locally if the lung itself had some primary pathology and was relatively low even with other extrapulmonary injuries. Can you elaborate more on primary pulmonary injuries? Dr Muehlstedt. We defined acute lung injury as a PaO2/FiO2 ratio less than 300 regardless of the level of PEEP. Patients who developed NP had significantly more acute lung injury early after injury. These also are patients with a more pronounced IL-8 response to injury and a greater influx of neutrophils into the alveolar airspace. The involvement of IL-8 in the pathogenesis of
Muehlstedt et al 611
acute lung injury is well documented in the literature, explaining this observation. The two groups were no different in terms of injury severity as measured by the ISS and chest AIS. This is important since chest injury and contusion have been shown to illicit a local cytokine response. Furthermore, both groups have a similar incidence of head injury and this is relevant because the incidence of pneumonia in these patients is quite high. Dr Josef E. Fischer (Cincinnati, Ohio). I have followed the work of this group with great interest. I must say that although this paper is intriguing, it is almost counterintuitive. Since you limited your examination to just 2 cytokines, are you placing more emphasis on the fact that IL-10 was developed later? If you had looked at some other cytokines—and IL-6 comes to mind—you might not have had to be as convoluted in your interpretation, which really weakens whatever it is you are trying to prove. Dr Joseph S. Solomkin (Cincinnati, Ohio). The question I have has to do with the criteria for diagnosing the pneumonia. When you did the BAL, did you do quantitative cultures? What was the correlation with that data? I think that BAL is a much more compelling case for infection than using the CDC criteria in patients who already had substantial underlying acute lung disease. Dr Muehlstedt. The pneumonia diagnosis was made clinically by our ICU team. Standard CDC criteria for the diagnosis of NP were used. Cultures were quantitative. A portion of each admission BAL sample was sent for culture and all cultures were negative. In no instance, therefore, was a patient infected prior to admission. Finally, we did not look at the incidence of late pneumonia in these patients. I would like to thank the Association for the opportunity to present these data and the discussants for their questions.
Background. Nosocomial pneumonia (NP) in injured patients is a significant clinical problem. We hypothesize that the pathogenesis of NP in injured patients involves an imbalanced cytokine response within the alveolar airspace that may inhibit effector cell function. Methods. Proinflammatory (IL-8) and anti-inflammatory (IL-10) levels were measured in bronchoalveolar lavage (BAL) fluid from multitrauma patients on admission, 24, 48, and 72 hours post-injury and following lipopolysaccharide (LPS) induction of alveolar cells. Patients were compared based on IL-8 levels and the development of NP. Results. A high level of IL-8 on admission was associated with the development of NP. In addition, levels of IL-8 were significantly greater in NP-positive patients at all time points. The IL-10 levels decreased from admission values in NP-negative patients but increased in NP-positive patients. Furthermore, a high level of IL-10 ( > 120 pg/mL) at 72 hours post-injury was associated with the development of NP. Alveolar cells from NP-positive patients produced significantly more IL-10 in response to LPS than cells from NP-negative patients. Conclusions. The pathogenesis of NP in injured patients involves an early and severe IL-8 process within the lung followed by an exaggerated IL-10 response that may inhibit effector cell function. (Surgery 2001;130:602-11.) From the Department of Surgery, Hennepin County Medical Center, affiliated with the University of Minnesota Medical School
MULTI-TRAUMA PATIENTS REQUIRING mechanical ventilation represent a unique subset of critically ill patients in which the incidence of nosocomial pneumonia (NP) approaches 45%. Despite the identification of at-risk patients, NP remains prevalent and a leading cause of morbidity.1 The failure of current prophylactic therapy likely is due to an incomplete understanding of the disease process. As such, research directed at defining the pathogenesis of NP within the trauma setting is paramount. With this in mind, we previously observed that traumatic injury induces a pro-inflammatory (IL-8) cytokine response within the alveolar airspace of injured patients.2,3 Furthermore, IL-8 production occurs locally within the lung and correlates with the influx of inflammatory cells, pulmonary dysfunction, and the development of NP.2 Presented at the 58th Annual Meeting of the Central Surgical Association, Tucson, Ariz, March 7-10, 2001. Reprint requests: Jorge L. Rodriguez, MD, Hennepin County Medical Center, Department of Surgery, 701 Park Ave S, Minneapolis, MN 55415. Copyright © 2001 by Mosby, Inc. 0039-6060/2001/$35.00 + 0 11/6/117105 doi:10.1067/msy.2001.117105
602 SURGERY
Pro- and anti-inflammatory cytokines have been measured in the serum of injured patients.4 It is assumed these mediators regulate one another to avoid organ injury and dysfunction. Previously we have demonstrated that bronchoalveolar lavage (BAL) fluid containing large amounts of IL-8 inhibits effector cell chemoattraction and superoxide production implying the presence of an inhibitory factor.5 A potent anti-inflammatory cytokine (IL-10) has been shown to inhibit a variety of effector cell functions including the production of proinflammatory cytokines (TNF-α, IL-1, IL-6, and IL-8) and the ability to present antigen.6,7 We hypothesize that the pathogenesis of NP in injured patients involves an imbalanced cytokine response within the lung that may inhibit effector cell function. MATERIAL AND METHODS The Internal Review Board of Hennepin County Medical Center approved this protocol for research involving human subjects. Multitrauma patients admitted within 2 hours of injury who required mechanical ventilation were included in the study provided there was no history of infection or aspiration. Demographic
Surgery Volume 130, Number 4
data that were collected included age, gender, pulmonary comorbidity (active smoker, asthma, chronic obstructive pulmonary disease) and nonpulmonary comorbidity. Also assessed were the mechanism of injury, injury severity score (ISS), the presence of head injury, severity of chest injury as assessed by the chest anatomical injury score (AIS), hypotension (systolic blood pressure < 90 mm Hg) or emergent intubation in the field. Data collected upon admission to the surgical intensive care unit (SICU) included systolic blood pressure, temperature, blood pH, Pa O2, PaCO2, serum bicarbonate concentration (HCO3), ratio of arterial oxygen tension to inspired oxygen concentration (PaO2/FiO2), mode of ventilation, tidal volume, positive end-expiratory pressure (PEEP), plateau pressure (Pplat), static compliance (tidal volume/Pplat), and lung injury score (LIS).8 In the SICU, information recorded daily included the PaO2/FiO2 ratio, static lung compliance, LIS, antibiotics use, and blood product transfusion. Acute lung injury was defined as PaO2/FiO2 less than 300 mm Hg regardless of PEEP. The simultaneous presence of 4 or more of the following criteria within 8 days of admission were used to define NP: (1) temperature > 101.5°F; (2) white blood cell count > 15,000/mm3; (3) inflammatory cells (> 25 WBC/low power field and < 10 epithelial cells/low power field) and bacteria with Gram stain; (4) positive quantitative sputum culture; (5) new or changing infiltrate on chest x-ray (CXR). BAL samples were obtained through the endotracheal tube using sterile technique, at admission, 24, 48, and 72 hours post-injury. A suction catheter (Medline Industries, Mundelein, Ill) was wedged, and 10 mL of sterile saline solution (0.9% NaCl, 37°C) was instilled, aspirated with suction, and discarded. Next, 30 mL of sterile saline solution was instilled and aspirated into a sterile specimen collector (Allegiance Healthcare, McGaw Park, Ill). Recovery rate was 82% ± 10%. Samples were chilled at 4°C immediately and processed within 10 minutes of collection. For each sample, a Cytospin (Shandon, Pittsburgh, Pa) slide was prepared, stained with Wright’s stain, and a differential obtained. Samples were strained through a 60-mesh steel screen (Sigma, St Louis, Mo) to remove mucus and centrifuged at 3000 rpm for 5 minutes at 4°C. Cell-free supernatant was removed and stored at –70°C for cytokine analysis (baseline values). The cell pellet was then re-suspended in phosphatebuffered saline solution without calcium (Gibco, Rockville, Md). White blood cells were counted by hemacytometer and viability was assessed by trypan blue exclusion (Sigma). All samples were more
Muehlstedt et al 603
than 95% viable. RPMI 1640 medium (Gibco) supplemented with 10% pooled human AB serum (Atlanta Biologicals, Norcross, Ga), 100 U/mL penicillin, and 100 g/mL streptomycin (Gibco) was added to re-suspended cells achieving a final concentration of 106 WBC/mL. One and a half mL of this solution (1.5 million cells) were cultured for 24 hours at 37°C in 5% CO2, in the presence or absence of 10 ng/mL Escherichia coli 0111:B4 lipopolysaccharide (LPS; Sigma). Following incubation the samples were stored at –70°C pending analysis. Supernatants were assayed for IL-8 and IL10 using commercial enzyme-linked immunosorbent assay kits according to manufacturer’s directions (OptEIA, BD PharMingen, San Diego, Calif). Detection limits for IL-8 and IL-10 were 3.1 pg/mL and 7.8 pg/mL, respectively. Values below these levels were considered zero for statistical analysis. An anti-inflammatory (AI) index was calculated daily for each patient. This was defined as the ratio of baseline IL-10 (ng/mL) to baseline IL-8 (ng/mL), multiplied by 1000. Outcome data included duration of mechanical ventilation, the development of acute respiratory distress syndrome (ARDS; defined as PaO2/FiO2 < 200 mm Hg regardless of PEEP, bilateral CXR infiltrates, and pulmonary capillary wedge pressure ≤ 18 mm Hg), intensive care unit length of stay, hospital length of stay, and mortality. Patients were categorized for analysis into those who developed NP (NP-positive) and those who did not (NP-negative), as well as by IL-8 levels on admission (> 10,000 and < 10,000 pg/mL). Continuous variables were examined by Student t test and expressed as mean + SD. Binary data were assessed by chi-square analysis and expressed as a percentage. Cytokine data, analyzed by the MannWhitney U test, were reported as mean ± standard error of the mean. A P value < .05 was considered significant. RESULTS A total of 49 patients met criteria for the study. The patient population primarily was young (42 + 21 years) males (81.6%) injured in blunt trauma (96%) with significant injury (ISS 31.5 ± 9.1). Of the 49 patients, 22 (44%) developed NP within 8 days of admission. Average time to diagnosis was 5 ± 1.3 days. Microbial isolates primarily were gramnegative pathogens (60%), including Pseudomonas, Haemophilus influenzae, Enterobacter, and Klebsiella in decreasing order of frequency. Overall the most common individual isolate was Staphylococcus aureus, accounting for 31% of pathogens. All NPpositive patients had an inflammatory Gram stain,
604 Muehlstedt et al
Surgery October 2001
Table I. Demographic data for pneumoniapositive and pneumonia-negative patients
Age (years) Gender (male) Mechanism of injury (blunt) ISS Head injury Chest AIS Pulmonary comorbidity Nonpulmonary comorbidity
Pneumoniapositive (n = 22)
Pneumonianegative (n = 27)
43 ± 22 73% 100%
40 ± 21 89% 93%
33 ± 9 68% 2.0 ± 1.5 9%*
31 ± 10 52% 1.5 ± 1.7 37%
36%
33%
ISS, Injury severity score; AIS, anatomical injury score. *P < .05.
a positive sputum culture, and a new or changing infiltrate on CXR. All patients satisfied at least 4 criteria and 82% satisfied 5 criteria. Demographically, NP-positive patients were similar to NP-negative patients, the only exception being more pulmonary comorbidity among the NP-negative group (Table I). Prior to hospital admission, NP-positive patients were more often emergently intubated (18% vs 4%, P < .05) and hypotensive (68% vs 15%, P < .05). There were no clinical differences between groups on admission to the SICU (Table II). Within the first 72 hours of injury, patients who ultimately developed pneumonia experienced significantly more acute lung injury than patients who did not (59% vs 30%, P < .05). In addition, whereas the static compliance of NP-negative patients improved, that of NP-positive patients worsened. After admission, the LIS was consistently greater in NP-positive patients than NP-negative patients, reaching statistical significance at 24, 48, and 72 hours post-injury (Table III). The NP-positive group received significantly fewer prophylactic antibiotics than NP-negative patients (55% vs 81%, P < .05), however the incidence of blood transfusions was similar between groups (55% and 56%, respectively). The IL-8 concentration in BAL supernatant (baseline) was compared between groups. Significantly greater levels of IL-8 were measured in the BAL of NP-positive patients at all time points. In addition, the BAL of NP-positive patients contained a significantly greater percentage of neutrophils on admission (87% vs 49%, P < .05) and at 24 hours post-injury (92% vs 81%, P < .05). Also of interest, IL-8 levels were highest on admission and decreased thereafter both within each group and individually (Fig 1).
Table II. Admission clinical data for pneumoniapositive and pneumonia-negative patients Pneumoniapositive (n = 22) Systolic blood 111 ± 18 pressure (mm Hg) Temperature (°F) 97.2 ± 1.8 pH 7.32 ± 0.09 PaO2 (mm Hg) 206.1 ± 132.0 PaCO2 (mm Hg) 37.7 ± 8.9 HCO3 (mEq/L) 19.6 ± 5.0 PaO2/FiO2 ratio 360 ± 189 AC ventilation 95% Tidal volume 11.4 ± 0.6 (mL/kg) PEEP (cm H2O) 4.6 ± 2.3 Pplat (cm H2O) 22.7 ± 6.7 Compliance 47.9 ± 14.3 (mL/cm H2O) LIS 1.4 ± 0.8
Pneumonianegative (n = 27) 127 ± 21 97.6 ± 1.2 7.32 ± 0.08 194.3 ± 88.7 40.4 ± 10.1 21.1 ± 7.7 382 ± 218 96% 11.3 ± 0.4 4.5 ± 2.5 21.9 ± 5.1 54.4 ± 15.9 1.2 ± 0.8
AC, Assist control; PEEP, positive end-expiratory pressure; Pplat, plateau pressure; Compliance, static compliance; LIS, lung injury score.
When admission IL-8 levels categorized patients, 24 patients had a high IL-8 level (> 10,000 pg/mL) and 25 patients had a low IL-8 level (< 10,000 pg/mL). A high IL-8 level was associated strongly with the development of NP (76% of high IL-8 patients vs 12.5% of low IL-8 patients, P < .001). The static compliances of high and low IL-8 patients were similar on admission (56 + 16 vs 58 + 19 mL/cm H2O), however by 24 hours post-injury the static compliance of high IL-8 patients was significantly less than low IL-8 patients (48 + 18 vs 57 + 15 mL/cm H2O, P < .05), and this relationship persisted at 48 hours (48 + 14 vs 51 + 18 mL/cm H2O) and 72 hours post-injury (47 + 13 vs 55 + 18 mL/cm H 2 O). Similarly, while the PaO2/FiO2 ratios of both the high and low IL-8 groups were similar on admission (361 + 170 vs 385 + 160), the ratio among high IL-8 patients was much lower at 24 hours (283 + 127 vs 380 + 180), 48 hours (298 + 165 vs 328 + 148), and 72 hours (261 + 132 vs 289 + 110) post-injury. Finally, a high IL-8 level (> 10,000 pg/mL) on admission was associated strongly with a high IL-10 level (> 120 pg/mL) 72 hours post-injury (P < .05), and a high IL-10 level on day 3 correlated with the development of NP (92% of high IL-10 patients vs 18% of low IL-10 patients, P < .05). Baseline IL-10 was compared between groups (Fig 2). Following admission, IL-10 levels were significantly greater in NP-positive patients at all time points. Furthermore, whereas IL-10 levels
Muehlstedt et al 605
Surgery Volume 130, Number 4
Table III. Pulmonary physiology data for pneumonia-positive and pneumonia-negative patients Admission Static compliance Pneumonia + Pneumonia – PaO2/FiO2 ratio Pneumonia + Pneumonia – Lung injury score Pneumonia + Pneumonia –
24 hours
48 hours
72 hours
48 ± 14 54 ± 16
47 ± 16 53 ± 17
48 ± 18 52 ± 17
45 ± 12* 58 ± 17
360 ± 189 382 ± 218
310 ± 138 363 ± 186
275 ± 125 351 ± 169
263 ± 107 295 ± 133
1.4 ± 0.8 1.2 ± 0.8
1.8 ± 0.7* 1.2 ± 0.8
2.0 ± 0.7* 1.3 ± 0.8
2.1 ± 0.6* 1.4 ± 1.1
Pneumonia +, pneumonia-positive; pneumonia –, pneumonia-negative. *P < .05.
Fig 1. Levels of the pro-inflammatory cytokine IL-8 measured in the bronchoalveolar fluid of patients who develop pneumonia and those who do not.
Fig 2. Levels of the anti-inflammatory cytokine IL-10 measured in the bronchoalveolar fluid of patients who develop pneumonia and those who do not.
decreased from admission values in the NP-negative group, levels steadily increased in NP-positive patients, reaching a maximum at 72 hours. The AI index was compared between groups, and similar to IL-10, AI index values diverged. While the AI index of NP-negative patients fell, the AI index of NP-positive patients rose (Fig 3). The IL-8 level produced by BAL cells following LPS stimulation was compared (Fig 4). Initially, white blood cells harvested from NP-positive patients produced significantly more IL-8 than cells from NPnegative patients. At 48 and 72 hours, however, levels were similar. Levels of IL-10 produced in response to LPS were compared in Fig 5. Whereas initial levels were similar between groups, the amount of IL-10 produced by NP-positive patients was significantly greater at 48 and 72 hours post-injury. The incidence of ARDS in the overall population was 6% and statistically was not different between groups. On average, NP-positive patients had more ventilator days and greater intensive care unit length of stay and hospital length of stay than NP-negative patients with similar injuries. Overall
mortality was 12% (6/49) and was not statistically different between groups (Table IV). Causes of death were severe head injury (1 NP-positive, 3 NPnegative) and multiple organ system failure (1 NPpositive, 1 NP-negative). DISCUSSION The presence of NP in mechanically ventilated trauma patients remains problematic despite improvements in the care of critically ill patients. The 44% incidence in the current study, while significant, is consistent with published reports.1,9 Furthermore, this study supports that patients who develop NP are more often emergently intubated, hypotensive, and victims of head trauma.1 This study identifies distinct cytokine trends that precede NP within the lungs of injured patients. Individuals who develop NP have significantly greater levels of IL-8 within their alveolar airspace, the highest of which are within 2 hours of injury. Moreover when admission IL-8 levels categorized patients, a strong association was observed between a high level of IL-8 and the subsequent develop-
606 Muehlstedt et al
Fig 3. The anti-inflammatory index (ratio of IL-10 to IL-8, multiplied by 1000) of patients who develop pneumonia and those who do not.
Fig 4. Levels of IL-8 produced following LPS stimulation of bronchoalveolar lavage white blood cells from pneumonia-positive and pneumonia-negative patients.
ment of pulmonary dysfunction and NP. The IL-8 is a potent neutrophil activator and chemoattractant and this is consistent with the greater percentage of neutrophils seen in admission samples from NPpositive patients.10,11 The pulmonary dysfunction that also is more prevalent among this group likely is initiated by this process. In fact, the presence of anti-IL-8 antibodies in BAL fluid has been associated with survival in human ARDS patients, and IL-8 inhibition has been shown protective of acute lung injury in animal models of trauma.12,13 The LIS, statistically greater in the NP-positive group following admission, may be a valuable clinical tool for identifying this exaggerated pro-inflammatory response to trauma.8 As mentioned, the greatest IL-8 levels were measured within 2 hours of injury. This implies that the initiating event is the trauma itself and not the
Surgery October 2001
Fig 5. Levels of IL-10 produced following LPS stimulation of bronchoalveolar lavage white blood cells from pneumonia-positive and pneumonia-negative patients.
result of mechanical ventilation or transfusions. Although induced by trauma, the actual source of IL-8 remains debatable. We previously have demonstrated the presence of IL-8 mRNA within the BAL of injured patients, implying it is locally produced and is not initiated in the systemic circulation.2,3 Furthermore, the ability of BAL cellular contents to produce substantial amounts of IL-8 in response to LPS indicates that these cells have the capacity to produce the IL-8 levels found after trauma. It also is interesting that as more polymorphonuclear leukocytes emigrate into the alveolar airspace, basal IL-8 levels fall. For these reasons, we believe the primary source of IL-8 is the activated tissue macrophage and not the polymorphonuclear leukocyte population or activated endothelium, which have been suggested.14 Increasing levels of IL-10 are found within the alveolar airspace of NP-positive patients such that at 72 hours post-injury, high levels of IL-10 correlate with the development of pneumonia. This is in contrast to NP-negative patients, whose levels of this anti-inflammatory cytokine decrease from admission values. Furthermore, by 48 hours postinjury, alveolar cells from NP-positive patients produce significantly more IL-10 following LPS stimulation than cells from NP-negative patients. The fact that alveolar cells from NP-positive patients initially produce more IL-8 in response to LPS and later more IL-10 may indicate a shift of these cells to an anti-inflammatory state. This is reflected by the increased anti-inflammatory ratio of these patients. In the lung, IL-10 is produced primarily by the alveolar macrophage.15 Originally described as cytokine synthesis inhibitory factor, IL-10 has been shown to inhibit a variety of effector cell func-
Muehlstedt et al 607
Surgery Volume 130, Number 4
Fig 6. Schematic representation of the lung’s response to injury. Trauma induces a pro-inflammatory process within the lung that is balanced by an anti-inflammatory response. If exaggerated, high levels of IL-8 recruit and activate neutrophils that cause acute lung injury. In contrast, high levels of IL-10 inhibit effector cell function, predisposing patients to pneumonia.
Table IV. Outcome data for pneumonia-positive and pneumonia-negative patients
ARDS Ventilated days Intensive care days Hospital days Mortality
Pneumoniapositive (n = 22)
Pneumonianegative (n = 27)
9% 15 ± 10 19 ± 12 26 ± 15 9%
4% 9 ± 11 12 ± 13 20 ± 16 15%
ARDS, Acute respiratory distress syndrome.
tions including IL-8 synthesis.6 In addition to the inhibition of cytokine production, however, IL-10 down-regulates the expression of human leukocyte antigen DR on effector cells, diminishing their capacity to recognize and present antigen.7 As such, IL-10 diminishes the ability of pulmonary macrophages to ingest and kill bacteria and has been correlated with infection in injured patients.16,17 In an animal model of pneumonia, IL-10 attenuated the pro-inflammatory cytokine response within the lung but hampered effective clearance of bacteria.18 Furthermore, since attempts to eradicate bacteria both systemically and locally in human patients have not reduced substantially the incidence of NP, then perhaps it is not primarily the microbial load that is the problem, but rather the capacity of the local organ to
effectively clear bacteria.19 We believe this posttraumatic dysfunction exists and is mediated in part by IL-10. It is reasonable to suspect that IL-10 levels are reflective of the initial pro-inflammatory process, whereby more IL-10 is produced in response to more IL-8. This seems reasonable, as high levels of IL-8 on admission are associated with high levels of IL-10 measured 3 days post-injury. This however does not explain why IL-10 levels continue to increase in NP-positive patients and decrease in NP-negative patients, both in the face of decreasing IL-8 levels. It appears that the IL-10 response of NP-positive patients is exaggerated. It is possible that when a critical IL-8 level is reached within the alveolar airspace, a pronounced IL-10 response results. Another possibility is that certain individuals have unique cellular or endothelial populations, or perhaps, are genetically predisposed to such cytokine production following trauma. Catecholamines recently have been found to induce de novo IL-10 synthesis by monocytes.20 This is interesting because NP-positive patients more often were hypotensive in the field, and although speculative, they may have experienced a greater release of endogenous catecholamines following injury. Why certain individuals display an exaggerated alveolar cytokine response to trauma remains unknown. Demographically there were no differ-
608 Muehlstedt et al
ences in age, gender, mechanism of injury, or injury severity among patients who developed pneumonia and those who did not. The groups also were similar in terms of the incidence and severity of chest injury. Only pulmonary comorbidity differed between groups, the majority of which was cigarette smoking. Pulmonary comorbidities such as smoking, asthma, and chronic obstructive pulmonary disease are well-defined risk factors for communityacquired pneumonia and pneumonia in critically ill medical patients but not among the trauma population. In fact, this study identified more pulmonary comorbidity among patients that did not develop pneumonia. This would seem to indicate that although pneumonia can develop in multiple populations, the pathogenesis among injured patients might be unique. In cigarette smokers, the ability of alveolar macrophages to produce cytokines is reduced although the ability to phagocytize and kill bacteria is maintained.21,22 More pulmonary comorbidity among NP-negative patients therefore is consistent with the hypothesis that an exaggerated cytokine response is responsible for the predisposition to infection in injured patients. Antibiotic interaction with cytokine production is described in the literature. Various antibiotics have been shown to both diminish and augment the proinflammatory cytokine response of effector cells to a stimulus.23,24 Most studies involve systemic cells, unfortunately, and not those of the alveolar airspace. Regardless, initial BAL samples were collected prior to systemic antibiotic prophylaxis in most cases, and it is the IL-8 level of these samples that correlates strongly with the development of NP. The impact of antibiotics on subsequent cytokine measurements is difficult to answer, however we found no correlation between antibiotic usage and cytokine levels at any time point (data not shown). As with any clinical study, this study has flaws. First, we have chosen a very select patient population and conclusions may not apply to less injured patients. Although the initial proinflammatory response is most likely present prior to the institution of mechanical ventilation, it is not possible for this study to determine the effect of mechanical ventilation on later cytokine production. Second, in regards to the diagnosis of NP, we made no attempt to compare our culture methods to protected specimen brush technique. Also, although sterile technique was implemented, it is possible that lavage may have induced a local cytokine response or introduced bacteria. Third, the use of an ex-vivo model and LPS in nanogram concentrations is not physiologic. Finally, the nature of
Surgery October 2001
cytokines themselves, with their short half-lives and complex interactions, undoubtedly complicate analysis. For these reasons it is imperative to standardize sampling intervals, technique, and analysis. Given these limitations, this study’s goal was not to identify isolated pneumonia predictors, but rather to identify the dynamic trends of cytokine production that precede development. Trauma induces an IL-8 cytokine response in the lungs of injured patients. This pro-inflammatory response appears exaggerated in certain individuals, and it is associated with pulmonary inflammation and dysfunction evident by the influx of neutrophils, decreased compliance, and early acute lung injury. Teleologically it seems reasonable that this pro-inflammatory state induces an anti-inflammatory response to attenuate the process. If this anti-inflammatory response is similarly exaggerated, IL-10 levels may be capable not only of inhibiting cytokine production, but also the ability of effector cells to recognize and process antigen as well, a state referred to as immunoparalysis.25 We believe this model is accurate (Fig 6). As such, future study should identify individuals predisposed to these responses and correlate cytokine levels with effector cell function to better define the disease process in these patients. REFERENCES 1. Rodriguez JL, Gibbons KJ, Bitzer LG, Dechert RE, Steinberg SM, Flint LM. Pneumonia: incidence, risk factors, and outcome in injured patients. J Trauma 1991;31:907-14. 2. Rodriguez JL, Miller CG, DeForge LE, Kelty L, Shanley CJ, Bartlett RH, Remick DG. Local production of interleukin-8 is associated with nosocomial pneumonia. J Trauma 1992;33:74-82. 3. Keel M, Ecknauer E, Stocker R, Ungethum U, Steckholzer U, Kenney J, Gallati H, et al. Different pattern of local and systemic release of proinflammatory and anti-inflammatory mediators in severely injured patients with chest trauma. J Trauma 1996;40:907-14. 4. Sherry RM, Cue JL, Goddard JK, Parramore JB, DiPiro JT. Interluekin-10 is associated with the development of sepsis in trauma patients. J Trauma 1996;40:613-7. 5. Rodriguez JL, Moore NP, Miller CG, Cox C, Garner WL, Smith DJ, Remick DG, et al. Traumatic injury: the local organ (lung) alters cell function. Am Coll Surg 1993; Surgical forum volume XLIV;124. 6. Fiorentino DF, Zlotnik A, Mosmann TR, Howard M, O’Garra A. IL-10 inhibits cytokine production by activated macrophages. J Immunol 1991;147:3815-22. 7. de Waal Malefyt R, Haanen J, Speits H, et al. Interleukin 10 (IL-10) and viral IL-10 strongly reduce antigen-specific human T cell proliferation by diminishing the antigen-presenting capacity of monocytes via downregulation of class II major histocompatibility complex expression. J Exp Med 1991;174:915-24. 8. Murray JF, Matthay MA, Luce JM, Flick MR. Pulmonary perspectives: an expanded definition of the adult respiratory distress syndrome. Am Rev Respir Dis 1988;138:720-3.
Surgery Volume 130, Number 4
9. Fabian TC, Boucher BA, Croce MA, et al. Pneumonia and stress ulceration in severely injured trauma patients: a prospective evaluation of the effects of stress ulcer prophylaxis. Arch Surg 1993;128:185-92. 10. Baggiolini M, Walz A, Junkel SL. Neutrophil-activating peptide-1/interleukin 8, a novel cytokine that activates neutrophils. J Clin Invest 1989;84:1045-9. 11. Kunkel SL, Standiford T, Kasahara K, Strieter RM. Interleukin-8: the major neutrophilic chemotactic factor in the lung. Exp Lung Res 1991;17:17-23. 12. Kurdowska A, Miller EJ, Noble JM, et al. Anti-IL-8 autoantibodies in alveolar fluid from patients with the adult respiratory distress syndrome. J Immunol 1996;157:2699-706. 13. Sekido N, Mukaida N, Harada A, et al. Prevention of lung reperfusion injury in rabbits by a monoclonal antibody against interleukin-8. Nature 1993;365:654-7. 14. Schroder JM, Christophers E. Secretion of novel and homologous neutrophil-activating peptides by LPS-stimulated human endothelial cells. J Immunol 1989;142:244-51. 15. Armstrong L, Millar AB. Relative production of tumor necrosis factor a and interleukin 10 in adult respiratory distress syndrome. Thorax 1997;52:442-6. 16. Steinhauser ML, Hogaboam CM, Kunkel SL, Lukacs NW, Strieter RM, Standiford TJ. IL-10 is a major mediator of sepsis-induced impairment in lung antibacterial host defense. J Immunol 1999;162:392-9. 17. Sachse C, Prigge M, Cramer G, Pallua N, Henkel E. Association between reduced human leukocyte antigen (HLA)-DR expression on blood monocytes and increased plasma level of interleukin-10 in patients with severe burns. Clin Chem Lab Med 1999;37:193-8. 18. van de Poll T, Marchant A, Keogh CV, Goldman M, Lowry SF. Interleukin-10 impairs host defense in murine pneumococcal pneumonia. J Infect Dis 1996;174:994-1000. 19. Gastinne H, Wolff M, Delatour F, Faurisson F, Chevret S. A controlled trial in intensive care units of selective decontamination of the digestive tract with nonabsorbable antibiotics. N Engl J Med 1992;326:594-9. 20. Platzer C, Docke W-D, Volk H-D, Prosch S. Catecholamines trigger IL-10 release in acute systemic stress reaction by direct stimulation of its promoter/enhancer activity in monocytic cells. J Neuroimmol 2000;105:31-8. 21. McCrea KA, Ensor JE, Nall K, Bleecker ER, Hasday JD. Altered cytokine regulation in the lungs of cigarette smokers. Am J Respir Crit Care Med 1994;150:696-703. 22. Baldwin GC, Tashkin DP, Buckley DM, Park AN, Dubinett SM, Roth MD. Marijuana and cocaine impair alveolar macrophage function and cytokine production. Am J Respir Crit Care Med 1997;156:1606-13. 23. Oishi K, Sonoda F, Kobayashi S, Iwagaki A, Nagatake T, Matsushima K, Matsumoto K. Role of interleukin-8 (IL)-8 and an inhibitory effect of erythromycin on IL-8 release in the airways of patients with chronic airway diseases. Infect Immun 1994;62:4145-52. 24. Morikawa K, Watabe H, Araake M, Morikawa S. Modulatory effect of antibiotics on cytokine production by human monocytes in vitro. Antimicrob Agents Chemother 1996;40:1366-70. 25. Volk H-D, Reinke P, Docke W-D. Clinical Aspects: from systemic inflammation to ‘immunoparalysis’. In Jack RS (ed). Chem Immunol. Basel: Karger;2000;162-77.
DISCUSSION Dr Andrew R. Peitzman (Pittsburgh, Pa). It is an honor to discuss another in a series of important and well-designed studies investigating pulmonary response to injury and
Muehlstedt et al 609
infection from Dr Rodriguez and his group. In the interest of time I will not review what was an excellent presentation; however, I have a few comments and several questions. First, infection is an imbalance between host factors, local tissue injury, and virulence of the infecting organism. You have attempted to clarify the effects of local tissue injury for us in this study. Why did you choose to study only IL-8 and IL-10? Are we naive in doing so? We know TNF, IL-1, IL-6, and other cytokines are active as well. Do you have such data? Your earlier studies showed that local, meaning pulmonary, cytokine responses did not always correlate with systemic cytokine or messenger RNA levels. As many of your outcome criteria such as mortality, length of stay, and so on reflect systemic markers, do you have systemic levels of IL-8, IL-10, or their mRNA levels? Dr Muehlstedt. First was a question of why we chose to study IL-8 and IL-10. Our lab has studied a variety of other cytokines both systemically and within the alveolar airspace, including TNF, TNF soluble receptor, IL-1, IL-1 receptor antagonist, and IL-6. TNF was not included as levels decrease rapidly after injury and therefore are difficult to compare between patients admitted different times from injury. Both IL-1 and IL-6 were not included, because we previously demonstrated no correlation between these cytokines and pulmonary dysfunction or infection. Furthermore, although systemic levels of certain cytokines correlate with injury severity, we found no correlation between the magnitude of the cytokine response within the systemic circulation and that of the alveolar airspace, implying that cytokines within the alveolar airspace are not initiated in the systemic circulation. We acknowledge that IL-8 and IL-10 are not the only inflammatory cytokines involved in the lung’s cytokine response to trauma, however, we do believe that they are two of the primary mediators of this response and subsequent pulmonary dysfunction. We know this since, following trauma, anti-IL-8 has been shown protective of acute lung injury and IL-10 is involved in the resolution of pulmonary inflammation. Dr Andrew R. Peitzman. Second, you stated that pulmonary co-morbidity was greater in the pneumonia-free group. Yet the group with pneumonia had more acute lung injury, lower static compliance, higher lung injury scores, and longer time on the ventilator. Please reconcile this, since your observation seems counterintuitive. Dr Muehlstedt. The observation that patients who develop NP have significantly less pulmonary comorbidity is counterintuitive. If our hypothesis is correct, and the pathogenesis of nosocomial pneumonia involves an initially severe pro-inflammatory process, then it may be that a history of repetitive inflammatory insults to the lung, whether it be from cigarette smoke, asthma, or pulmonary disease, diminishes the capacity of pulmonary effector cells to mount an exaggerated pro-inflammatory
610 Muehlstedt et al
response. We are currently looking at this issue and hope to provide an answer soon. Dr Andrew R. Peitzman. Your AI index is an interesting use of new data. Is there any other data to corroborate its use? Is it statistically legitimate? If you simply look at these trends evaluating only IL-10 without your AI index, are your conclusions unchanged? Dr Muehlstedt. To my knowledge an AI index has not been described before. The fact that the AI index increases in patients who ultimately develop pneumonia is an important point of this paper as it may indicate a shift of alveolar cells to a primarily anti-inflammatory state. We have not studied this index prospectively but intend to do so in the future. Dr Andrew R. Peitzman. You state that you have not defined the source of IL-8 as tissue macrophage, PMN, or activated endothelium. Would immunohistochemistry do this? Dr Muehlstedt. Using both immunohistochemistry and PCR, we have previously demonstrated that alveolar macrophages are the primary producers of IL-8 following injury. We also think IL-10 is produced primarily by these cells and current literature supports this assumption. Dr Andrew R. Peitzman. In the discussion section of the manuscript, the authors suggest that more IL-10 is produced in response to greater IL-8 production initially. Is this truly cause-and-effect or simply overall up-regulation of the inflammatory cytokines? In your discussion you state, and I quote, “The fact that alveolar cells from pneumonia-positive patients initially produce more IL-8 in response to LPS and later more IL-10 may indicate a shift of these cells to an antiinflammatory state.” This is a critical observation and must be elaborated upon. The key to the risk of infection, multiple organ failure, and often survival in the trauma patient is this balance between the pro-inflammatory cytokines and the anti-inflammatory cytokines. This is both a temporal phenomenal and relative effect, as shown in this study. If the pro-inflammatory cytokines predominate for a prolonged period, the patient develops SIRS. If the anti-inflammatory cytokines predominate, CARS results, or, as the author stated in the paper, an immunoparalysis. This results in patients susceptible to infection, as shown in the study. Dr Raymond Pollak (Peoria, Ill). I thought this was a fascinating paper because it demonstrates that the host response to injury may well be dependent on the cytokine gene profile that exists in the genome. We recently have looked at this in terms of gender and ethnicity, and it is clear that when looking at cytokine gene polymorphisms, the way a person responds to injury, whether it is trauma, burn injury, or transplantation, may well be predicated by the subject’s cytokine gene profile. So I was curious as to whether you had looked at these, using PCR-based techniques, which
Surgery October 2001
are relatively easy to do? Some individuals may be predisposed to developing pro-inflammatory cytokine responses whereas others might be predisposed to developing the opposite response. Dr Muehlstedt. Why a certain subset of trauma patients develops an exaggerated IL-8 response to trauma is unknown. Many hypotheses can be made. We do plan on pursuing genetic studies in injured patients, as it may be that certain individuals are genetically predisposed to this response. Catecholamines also may be involved. Although we have not measured levels in these patients, catecholamines have been associated with a pro-inflammatory cytokine response and are capable of inducing IL-10 production. In this study, pneumonia patients more often were hypotensive and in need of vasopressors. It is possible that elevated levels of both endogenous and exogenous catecholamines are responsible for the increased levels of IL-10 found within the alveolar airspace of these patients. Dr Mark A. Malangoni (Cleveland, Ohio). Your premise is based on the fact that this abnormal cytokine profile is important in the development of pneumonia. I wonder if you examined your data looking at it at a different way, that is, evaluate patients who developed the abnormal cytokine profile—and I am assuming that there must be a few in your group who didn’t develop pneumonia. Did that change the results? For people who had the abnormal ratio, did all of them develop pneumonia? The second comment I had was that the occurrence of the pneumonias in this group was very early. What we see in these patients is a different bacterial flora as the cause for the pneumonia than in patients who develop pneumonia later on in their course. Did you try to correlate anything with the microbiology of this disease, and do you have any information about patients who develop late pneumonias? Dr Muehlstedt. Why we chose to categorize patients into two groups, pneumonia-positive and pneumonianegative, as opposed to isolating patients based on initial levels of IL-8 is a good question. When we separate patients into initially high and low IL-8 groups, we find that 85% of those with high levels develop pneumonia. The purpose of this study, however, was not to identify isolated cytokine predictors of nosocomial pneumonia as there are significant problems with doing this. The short half-lives of cytokines, variations in sampling intervals and technique, and the lack of standardization among ELISA kits makes current cytokine analysis unsuitable for human diagnosis. Therefore, although appealing, it would be unreasonable to conclude a certain level of IL8 is predictive of pneumonia. Our goal was to identify dynamic trends of cytokine production that precede pneumonia. By looking at cytokine trends over time, using strictly defined sampling intervals, the potential error of isolated measurements is minimized.
Surgery Volume 130, Number 4 Dr Thomas Vargish (Chicago, Ill). I have two questions, the first of which is a follow-up to the one you have just heard. Even though you have defined the time period for the change in the cytokine levels, how did that correlate with the timing of the onset of pneumonia in these patients? Did you see the pattern occurring in close correlation with those changes, or was it something that happened later on? The second one is the following: As I recall looking at your data, the group that got pneumonia seemed to be intubated earlier. I wonder whether or not this had to do with severity of injury and the possibility that they aspirated or their airway management was more complicated. What role did that play in the cytokine changes you saw and in the development of pneumonia? Dr Muehlstedt. Times from injury to intubation were not statistically different between groups. Furthermore, we excluded patients with pre-existing infections and those with documented aspiration, as those patients would undoubtedly complicate our analysis. Burn patients were not included given the high incidence of inhalation injury in this group. Dr James Tyburski (Detroit, Mich). I have one question, can you elaborate a little bit on any primary lung injuries? In some of the data that I am going to present later, we looked at IL-6 in these patients, and it was clear that the production was much higher locally if the lung itself had some primary pathology and was relatively low even with other extrapulmonary injuries. Can you elaborate more on primary pulmonary injuries? Dr Muehlstedt. We defined acute lung injury as a PaO2/FiO2 ratio less than 300 regardless of the level of PEEP. Patients who developed NP had significantly more acute lung injury early after injury. These also are patients with a more pronounced IL-8 response to injury and a greater influx of neutrophils into the alveolar airspace. The involvement of IL-8 in the pathogenesis of
Muehlstedt et al 611
acute lung injury is well documented in the literature, explaining this observation. The two groups were no different in terms of injury severity as measured by the ISS and chest AIS. This is important since chest injury and contusion have been shown to illicit a local cytokine response. Furthermore, both groups have a similar incidence of head injury and this is relevant because the incidence of pneumonia in these patients is quite high. Dr Josef E. Fischer (Cincinnati, Ohio). I have followed the work of this group with great interest. I must say that although this paper is intriguing, it is almost counterintuitive. Since you limited your examination to just 2 cytokines, are you placing more emphasis on the fact that IL-10 was developed later? If you had looked at some other cytokines—and IL-6 comes to mind—you might not have had to be as convoluted in your interpretation, which really weakens whatever it is you are trying to prove. Dr Joseph S. Solomkin (Cincinnati, Ohio). The question I have has to do with the criteria for diagnosing the pneumonia. When you did the BAL, did you do quantitative cultures? What was the correlation with that data? I think that BAL is a much more compelling case for infection than using the CDC criteria in patients who already had substantial underlying acute lung disease. Dr Muehlstedt. The pneumonia diagnosis was made clinically by our ICU team. Standard CDC criteria for the diagnosis of NP were used. Cultures were quantitative. A portion of each admission BAL sample was sent for culture and all cultures were negative. In no instance, therefore, was a patient infected prior to admission. Finally, we did not look at the incidence of late pneumonia in these patients. I would like to thank the Association for the opportunity to present these data and the discussants for their questions.