Acute respiratory distress syndrome (ARDS) involves an intense inflammatory response in the lungs, with accumulation of both pro- and antiinflammatory cytokines in bronchoalveolar lavage fluid (BALF). Our goal was to determine how the balance between pro- and antiinflammatory mediators in the lungs changes before and after the onset of ARDS. We identified 23 patients at risk for ARDS and 46 with established ARDS and performed serial bronchoalveolar lavage (BAL). We used immunoassays to measure tumor necrosis factor α (TNF- α ) and soluble TNF- α receptors I and II; interleukin 1 β (IL-1 β ), IL-1 β receptor antagonist, and soluble IL-1 receptor II; IL-6 and soluble IL-6 receptor; and IL-10. We used sensitive bioassays to measure net TNF- α , IL-1 β , and IL-6 activity. Although individual cytokines increased before and after onset of ARDS, greater increases occurred in cognate receptors and/or antagonists, so that molar ratios of agonists/antagonists declined dramatically at the onset of ARDS. The molar ratios remained low for 7 d or longer, limiting the activity of soluble IL-1 β and TNF- α in the lungs at the onset of ARDS. This significant antiinflammatory response early in ARDS may provide a key mechanism for limiting the net inflammatory response in the lungs.
Keywords: acute respiratory distress syndrome; cytokine receptors; cytokines; IL-1; TNF
The acute respiratory distress syndrome (ARDS) is characterized by an acute inflammatory response in the lung parenchyma that is associated with severe injury to the epithelial and endothelial barriers (1). Cytokines play a critical role as signaling molecules that initiate, amplify, and perpetuate inflammatory responses on a local and systemic basis. Two of the most important early response cytokines are tumor necrosis factor α (TNF-α) and interleukin 1β (IL-1β). Both TNF-α and IL-1β are present in the bronchoalveolar lavage fluid (BALF) of patients at risk for ARDS and with established ARDS (2-4). The highest concentrations of TNF-α and IL-1β occur in the BALF from patients with sustained ARDS. The ratios of BALF to serum cytokine concentrations are typically elevated, suggesting a pulmonary origin (5, 6). Despite key roles for both TNF-α and IL-1β in the development and progression of inflammation, clinical trials directed at neutralizing TNF-α and IL-1β in patients with sepsis have not shown a significant improvement in outcome or the onset of lung injury (6, 7). In addition, single cytokine measurements in serum and BAL do not consistently predict either the onset or the outcome of ARDS (8). The failure of these efforts underscores an incomplete understanding of the pathophysiology of ARDS.
Specific and nonspecific cytokine antagonists were described soon after the initial discovery of cytokines and provide mechanisms for limiting the biologic effects of proinflammatory cytokines. These discoveries increased the understanding of the complexity of cytokine signaling pathways and inflammatory cascades. Examples of nonspecific antagonists include α2-macroglobulin and antiinflammatory cytokines such as IL-10, whereas specific antagonists include IL-1 receptor antagonist (IL-1ra), soluble IL-1 receptor II (sIL-1RII), soluble TNF receptor I (sTNF-RI) and soluble TNF receptor II (sTNF-RII). In contrast, the soluble IL-6 receptor (sIL-6R) is a specific agonist (9). Significant concentrations of sTNF-RI and sTNF-RII have been found in BALF from patients with ARDS (4). In addition, low concentrations of IL-10 and IL-1ra in BALF from patients with ARDS were found to be associated with increased mortality, suggesting an important role for antiinflammatory mediators in counter balancing the proinflammatory response (10).
Because the net inflammatory balance is of greater physiological and clinical importance than individual cytokine concentrations, ratios between proinflammatory and antiinflammatory mediators have been used to measure net pro- and antiinflammatory balance (11-16). For example, in inflammatory bowel disease an imbalance between pro- and antiinflammatory mediators has been suggested as a mechanism underlying persistent disease, and an elevation in the ratio between IL-1β and IL-1ra is an effective marker of active disease (11). Although pro- and antiinflammatory mediators have been found in the BALF of patients with ARDS, the relative production of pro- and antiinflammatory mediators, their net biological effects in BALF, and the relationships between ratios of pro- and antiinflammatory mediators and clinical parameters have not been examined in patients with ARDS.
The major goal of this study was to determine the relationships between early response cytokines in the lungs and their naturally occurring modulators to understand how these relationships change net biological activity in the lungs over time. Our second goal was to try to determine the relationship between molar ratios of agonist and antagonists and parameters of lung injury and outcome of patients at risk for ARDS and with established ARDS. We studied TNF-α and IL-1β because of their prominent roles as immediate response cytokines. We also studied IL-6 because TNF-α and IL-1β induce its production and because it has both pro- and antiinflammatory effects (17).
Patients with sepsis or trauma who were at risk for ARDS and patients with established ARDS from any cause were identified by prospective screening of all patients admitted to the Medical and Surgical Intensive Care Units of Harborview Medical Center (Seattle, WA) between February 14, 1994 and July 13, 1997. The specific criteria for sepsis and trauma risks and for ARDS have been described in prior studies from the Seattle ARDS research program (18-20). All patients with ARDS met the American European Consensus Conference definition of ARDS (21). The BAL techniques and handling of specimens also have been reported (18-20). The protocol was approved by the University of Washington Investigational Review Board. Either the patient or a responsible relative gave written informed consent.
All patients at risk for ARDS and with ongoing ARDS had daily measurements of vital signs and physiologic parameters according to standardized protocols, including hypoxemia, expressed as the PaO2 /Fi O2 ratio, tidal volume, and static compliance.
TNF-α, sTNF-RI, sTNF-RII, IL-1β, IL-1ra, sIL-1RII, IL-10, IL-6, and sIL-6R measurements were performed by enzyme immunoassay using commercially available reagents (R&D Systems, Minneapolis, MN). The assay sensitivities were < 20–50 pg/ml for all measured proteins. The concentration of total protein was measured by the bicinchoninic acid method (Pierce, Rockford, IL).
TNF-α bioactivity in BALF was measured with L929 cells as described previously with minor modifications (22, 23). A standard curve was created from dilutions of recombinant human TNF-α (R&D Systems) and concentrations of TNF-α were determined from the linear range of the standard curve. The specificity of the assay for TNF-α was confirmed by neutralization of the measured bioactivity with a murine anti-human TNF-α antibody (R&D Systems).
To assess the net biological activity of IL-1β, we measured the upregulation of intercellular adhesion molecule 1 (ICAM-1) on the human epithelial lung adenocarcinoma A549 cell line (American Type Culture Collection, Rockville, MD) as described (24). The IL-1β activity was calculated using results from the linear portion of the standard curve. The specificity of the assay for IL-1β in BALF was determined with an anti-human TNF-α antibody or a recombinant human IL-1 receptor antagonist as described (24).
IL-6 bioactivity in BALF was assayed by measuring the proliferation of the IL-6–dependent murine hybridoma cell line B9 (courtesy of P. Lansdorp, BC Cancer Research Center, Vancouver, British Columbia, Canada), as described (25, 26). The IL-6 bioactivity in each well was calculated from the linear portion of the standard curve.
For normally distributed data, the Student two-tailed t test was used for comparisons between two groups. The Mann–Whitney test was used for nonnormally distributed data. Correlations between cytokines, bioassays, and physiological parameters were calculated with Spearman correlation coefficients. The mortality associations were determined by Wilcoxon analysis. Multivariate analysis of the influence of IL-6 and sIL-6R was conducted by linear regression analysis.
We prospectively identified 46 patients with ARDS and 23 patients at risk for ARDS between January 1994 and November 1997. Seven of the at risk patients developed ARDS and were included in the ARDS data. The BALF from six normal volunteers was included for comparison. The clinical characteristics of the patient groups are summarized in Table 1. The at risk and ARDS populations were closely matched for age, sex, and APACHE II score. The number of patients with sepsis as a primary risk was similar in both groups, but there was a significantly higher percentage of patients with trauma in the at risk population (p < 0.05). The mortality of patients with ARDS was higher than the mortality of patients at risk for ARDS (23.9 versus 12.5%). The variation in the number of patients on different study days reflects changes in clinical status, including successful extubation or death, as well as the enrollment of some patients on Day 3 of ARDS.
At Risk | ARDS | |||
---|---|---|---|---|
n | 23 | 46 | ||
Age, yr | 46.6 ± 14.8 | 43.8 ± 16.6 | ||
Sex, % male | 59.1 | 59.1 | ||
Mortality, % deaths | 12.5 | 23.9 | ||
Primary risk, % patients | ||||
Sepsis | 41.7 | 34.8 | ||
Trauma | 58.3 | 39.1 | ||
Other | 0 | 20.0 | ||
APACHE II score* | 21.6 ± 6.1 | 21.6 ± 7.2 |
TNF-α was detectable by immunoassay in a minority of the BALF samples from patients at risk for ARDS (10.5% of BALF on Day 1 of at risk period, 7.7% of BALF on Day 3 of at risk period) (see Figure E1 in the online data supplement). TNF-α was also detectable in a minority of BALF specimens from patients with established ARDS from Days 1 to 7 (detectable in 28.6% of BALF on Day 1 of ARDS, 12.5% of BALF on Day 3, 6.9% of BALF on Day 7). TNF-α was not present in measurable concentrations in the BALF from normal volunteers. The measurable TNF-α concentrations were highest in the BALF on Day 1 of established ARDS and declined progressively after the onset of ARDS to undetectable levels after Day 7 of ARDS.
In contrast to TNF-α, sTNF-RI and sTNF-RII were detectable in nearly all of the BALF from patients at risk for ARDS and patients with established ARDS for up to 21 d after the onset of ARDS (Table 2). TNF receptors were also present in all of the normal BALF specimens. The median sTNF-RI and sTNF-RII concentrations increased more than 10-fold at the onset of ARDS (p < 0.05) and then declined during the course of ARDS (see Figures E2 and E3 in the online data supplement).
Normal Subjects(6) | At Risk | ARDS | ||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Day 1 (19) | Day 3 (14) | Day 1 (35) | Day 3 (40) | Day 7 (29) | Day 14 (15) | Day 21 (10) | ||||||||||
TNF-α | ND | 9.8 [2](4.6, 15.0) | 185.2 [1] (185.2, 185.2) | 13.0 [10] (9.1, 20.7) | 13.8 [5] (11.3, 21.4) | 932.6 [2] (8.4, 1,856.7) | ND | ND | ||||||||
sTNF-RI | 61.9(44.2, 76.1) | 96.0(53.4, 333.8) | 236.0 (80.4, 467.0)† | 650.1 (286.5, 1,019.9)‡ | 522.1 (161.5, 859.4)‡ | 136.7 (85.6, 313.0)‡ | 67.7 (40.1, 145.9) | 102.5 (55.5, 130.5) | ||||||||
sTNF-RII | 45.5(30.9, 51.8) | 155.3(50.4, 370.8) | 243.2 (85.6, 476.5)† | 782.4 (380.9, 1,091.4)‡ | 476.5 (210.4, 989.4)‡ | 141.5 (74.7, 328.9)‡ | 74.5 (43.7, 138.8) | 76.1 (56.4, 115.2)† | ||||||||
IL-1β | 6.0 [2](5.8, 6.2) | 5.4 [12](2.7, 13.2) | 6.3 [5] (4.4, 105.4) | 9.8 [23] (4.4, 18.9) | 4.3 [22] (2.8, 9.2) | 12.4 [12] (4.1, 37.1) | 5.0 [11] (2.1, 9.1) | 7.3 [7] (4.6, 8.4) | ||||||||
IL-1ra | 293.9(204.3, 649.3) | 504.8(326.8, 1,899.0) | 675.0 (355.2, 2,104.6) | 1,829.2 (1,124.3, 6,868.3)‡ | 2,601.7 (1,081.8, 4,105.1)‡ | 1,244.1 (576.7, 2,826.5)‡ | 972.9 (612.4, 1,417.7) | 1,189.7 (726.1, 1,621.0)† | ||||||||
sIL-1RII | 22.5 [2](19.6, 25.3) | 217.3 [16](94.8, 1,179.6)† | 397.8 [12] (111.0, 397.8)† | 1,511.4 [33] (662.3, 4,213.1)‡ | 1,142.0 [38] (394.4, 1,634.7)‡ | 133.7 [27] (53.0, 403.3)‡ | 76.5 [11] (31.1, 157.8)† | 49.0 [10] (25.4, 81.1)‡ | ||||||||
IL-6 | 3.2(1.9, 3.9) | 172.7(29.7, 1,034.3)† | 178.3 (37.8, 841.4)‡ | 1,229.5 (263.3, 3,232.6)‡ | 712.2 (144.6, 1,787.6)‡ | 80.5 (11.6, 258.3)‡ | 23.0 (6.3, 76.6)‡ | 19.3 (14.8, 51.1)‡ | ||||||||
sIL-6R | 129.3(60.5, 170.6) | 233.8(88.6, 857.7) | 232.8 (96.6, 564.2) | 564.2 (295.7, 1,014.3)‡ | 523.1 (225.0, 776.7)† | 308.2 (165.5, 477.1)† | 182.2 (90.0, 303.6) | 204.1 (109.0, 415.1) | ||||||||
IL-10 | ND | 1.7 [4](1.1, 2.9)‡ | 1.7 [12] (0.9, 2.6)‡ | 4.8 [28] (1.5, 9.3)‡ | 2.3 [29] (1.4, 5.4)‡ | 1.7 [12] (1.1, 2.1)† | 1.3 [10] (1.3, 2.1) | 1.6 [7] (0.8, 2.6)† |
The soluble TNF-α receptors, sTNF-RI and sTNF-RII, neutralize TNF-α with 1:1 stoichiometry. The molar TNF-α/ sTNF-RI and TNF-α/sTNF-RII ratios in patients at risk for ARDS were significantly lower on Days 1 and 2 than in normal volunteers (Figures 1 and 2). These ratios were even lower at the onset of ARDS, and on Days 1 and 3 were less than 15% of the values measured in normal volunteers (p < 0.005). In patients with sustained ARDS, the TNF-α/sTNF-RI and TNF-α/sTNF-RII ratios increased with time, reaching near normal values on Days 14 and 21 of ARDS.
A bioassay for TNF-α activity was used to measure the net balance between TNF-α and the soluble TNF-α receptors in BALF (Figure 3). Net TNF-α bioactivity was measured in BALF from 6 normal volunteers, 10 patients on Day 1 of ARDS, and 9 patients on Day 7 of ARDS. Net TNF-α bioactivity was not detectable in five of six normal volunteers. TNF-α bioactivity was detectable in all of the BALF samples from patients on Day 1 of ARDS (19.1 ± 25.1 pg/ml) and on Day 7 of ARDS (53.5 ± 39.3 pg/ml). The median TNF bioactivity increased significantly between Days 1 and 7 of ARDS (p = 0.034), which paralleled the increase in the median TNF-α/ sTNF-RI and TNF-α/sTNF-RII ratios from Day 1 to Day 7 of ARDS. Despite this similar pattern, there was no significant correlation between the results of the TNF-α bioassay for individual patients and the individual measurements of immunoreactive TNF-α, sTNF-RI, and sTNF-RII, or the ratios between TNF-α and either soluble receptor.
Interleukin 1β immunoreactivity was detectable in the BALF from normal volunteers, patients at risk for ARDS, and patients with established ARDS for up to 21 d after the onset of ARDS (Table 2). The concentrations of IL-1β in BALF from patients at risk for ARDS (at-risk Days 1 and 3) were elevated above those measured in BALF from normal volunteers. The IL-1β concentration in patients with established ARDS was highest at the onset of ARDS on Day 1 of ARDS (p < 0.05) (see Figure E4 in the online data supplement). Thereafter, the IL-1β concentration declined with time but remained elevated for up to 21 d in patients with sustained ARDS. Like the soluble TNF receptors, IL-1ra and sIL-1RII were detectable in the BALF from normal volunteers, patients at risk for ARDS, and patients with established ARDS (Table 2). The median concentrations of IL-1ra and sIL-1RII increased more than 5- and 100-fold, respectively, at the onset of ARDS and remained elevated for the first 3 d of ARDS (p < 0.005). Thereafter, the median concentrations of IL-1ra and sIL-1RII declined (see Figures E5 and E6 in the online data supplement).
The molar ratios of IL-1β/IL-1ra and IL-1β/sIL-1RII paralleled the decline in the molar ratios of TNF-α and its soluble receptors in patients at risk for ARDS and early after the onset of ARDS (Figures 4 and 5). The IL-1β/IL-1ra and IL-1β/sIL-1RII ratios on Days 1 and 3 of the at risk period were lower than the ratios measured in normal volunteers. These ratios were even lower in patients on Days 1 and 3 of ARDS (p < 0.05). The ratios increased by Days 14 and 21 of sustained ARDS, returning to values that were similar to those in normal volunteers.
The net bioactivity of IL-1β was measured in BALF from six normal patients, and in all patients at risk for ARDS or with established ARDS. The IL-1β bioactivity increased at the onset of ARDS and declined with time in patients with sustained ARDS (Figure 6). There was a significant relationship between immunoreactive IL-1β and IL-1β bioactivity in BALF on Day 1 of ARDS (r = 0.403, p = 0.018). This relationship was strongest on Day 3 of ARDS (r = 0.849, p < 0.001). When the results from all patients at all times were pooled, there was a strong and significant relationship between immunoreactive IL-1β and IL-1 bioactivity (r = 0.534, p < 0.001) (see Figure E7 in the online data supplement). There was no significant relationship between IL-1β bioactivity and either IL-1ra or sIL-1RII. There was also no relationship between IL-1β bioactivity and the ratios of IL-1β and either IL-1ra or sIL-1RII.
Immunoreactive IL-6 and sIL-6R were detectable in the BALF of normal volunteers, patients at risk for ARDS, and patients with established ARDS for up to 21 d after the onset of ARDS (Table 2). In the Day 1 and 3 BALF from patients at risk for ARDS, the median IL-6 concentrations detected by enzyme-linked immunosorbent assay (ELISA) were increased by 50-fold or more as compared with BALF from normal volunteers (p < 0.05 on Day 1 of at risk period). On Days 1 and 3 of established ARDS, the median IL-6 concentrations were approximately 5-fold higher than in BALF from patients at risk of ARDS, and 100-fold higher than in normal subjects (p < 0.005). In BALF from patients with sustained ARDS, the IL-6 concentration declined with time, but remained above the normal range for up to 21 d in patients with persistent ARDS (see Figure E8 in the online data supplement). The sIL-6R concentration was also increased in the BALF from patients at risk of and with established ARDS. Like IL-6, the sIL-6R concentrations were elevated on Days 1 and 3 in the BALF from patients at risk for ARDS (p < 0.05). The BALF sIL-6R concentration was increased on Days 1 and 3 of ARDS (p < 0.005), and fell with time in patients with persistent ARDS (see Figure E9 in the online data supplement).
In contrast to the low ratios of the respective soluble receptors or antagonists for TNF-α and IL-1β, the molar ratio of IL-6 to sIL-6R increased more than 10-fold in the BALF from patients at risk for ARDS (p < 0.05 on Day 1 of at risk period) (Figure 7). In patients with established ARDS, the IL-6/sIL-6R ratio was approximately 100-fold higher than in normal subjects (p < 0.005), and 3-fold higher than in patients at risk. The IL-6/sIL-6R ratio declined with time but remained significantly elevated at all times as compared with normal volunteers.
The IL-6 bioactivity was detectable in BALF from patients at risk of and with established ARDS (Figure 8). Net IL-6 bioactivity was measured in BALF from 3 normal volunteers, and subgroups of 13 patients on Day 1 of the at risk period for ARDS and 16 patients on Day 1 of ARDS. There was a significant relationship between the IL-6 bioactivity and immunoreactive IL-6 in BALF from patients at risk on Day 1 (r = 0.929, p < 0.001) as well as in the BALF of patients on Day 1 of established ARDS (r = 0.938, p < 0.001). When the data from all patients at all times were pooled, there was a strong linear relationship between IL-6 concentration and IL-6 bioactivity (r = 0.995, p < 0.001) (see Figure E10 in the online data supplement). There also was a significant relationship between IL-6 bioactivity and IL-6 soluble receptor in BALF from patients at risk on Day 1 (r = 0.9, p < 0.001), as well as in BALF from patients with established ARDS on Day 1 (r = 0.644, p = 0.007).
IL-10 was detectable by ELISA in low concentrations (< 20 pg/ml) in the BALF of patients at risk for ARDS and patients with established ARDS, but not in the BALF of normal volunteers (Table 2). The BALF IL-10 was present in low concentrations on Days 1 and 3 of at risk patients. On Days 1 and 3 of ARDS, however, the concentration of IL-10 was significantly higher (p < 0.005). The median concentration of IL-10 in BALF on Day 1 of ARDS was approximately 2.5-fold higher than the median concentration in BALF on either Day 1 or 3 of patients at risk for ARDS (see Figure E11 in the online data supplement). In BALF from patients with sustained ARDS, the concentration of IL-10 declined to nearly undetectable levels by Day 21 of ARDS.
The molar ratio of TNF-α/IL-10 declined like the molar ratios of TNF-α and its soluble receptors before and after the onset of ARDS (see Figure E12 in the online data supplement); however, the decline in the molar ratio of TNF-α/IL-10 was less dramatic. The ratios of TNF-α/IL-10 on Days 1 and 3 of the at risk period were 3-fold lower than the ratios measured in normal volunteers, but this change was not significant. The ratio was at least 4-fold lower in BALF from patients on Days 1 and 3 of ARDS compared with BALF from normal volunteers but this also was not significant. By Days 14 and 21 of sustained ARDS the ratios increased gradually but remained 2- to 3-fold lower than the ratios in BALF from normal volunteers.
To determine whether measures of cytokine balance were associated with clinical or physiological outcomes, we investigated the relationships between the ratios of proinflammatory/ antiinflammatory mediators and mortality, lung dysfunction, and lung injury. We used measures of lung compliance and hypoxemia expressed as the PaO2 /Fi O2 ratio as measures of lung dysfunction and the BALF total protein concentration as a measure of lung injury. In univariate analyses, on Day 7 of ARDS the concentrations of immunoreactive IL-1β and IL-6 and the ratio of IL-6 to sIL-6R were significantly related to subsequent mortality (p < 0.05). This was not true for measurements in BALF from at risk patients or in BALF immediately after the onset of ARDS (Days 1 and 3). There were significant positive relationships between the PaO2 /Fi O2 ratio and the TNF/sTNF-RII ratio on Day 1 of the at risk period, and with IL-1β/IL-1ra on Day 7 of ARDS (p < 0.05). There were significant positive relationships between lung compliance and the individual of IL-1β/IL-1ra, IL-1β/sIL-1RII, IL-6/sIL-6R, TNF/sTNF-RI, and TNF/sTNF-RII ratios on Days 1 and 3 of ARDS (see Table E1 in the online data supplement). The total protein concentration in BALF from patients at risk for ARDS and with established ARDS was significantly related to each of the individual soluble receptors and receptor antagonists. These relationships were present on all days studied, before and after the onset of ARDS (Table E1).
The main goal of this study was to determine the immunological and biological balance between cytokine agonists and antagonists in the BALF of patients at risk for ARDS and with established ARDS. We measured pro- and antiinflammatory cytokines (IL-1β, TNF-α, IL-6, and IL-10) and specific agonists and antagonists (IL-1ra, sIL-1RII, sTNF-RI and -RII, and sIL-6R) in patients at risk for ARDS and in patients with ARDS. We found a major antiinflammatory response that peaked early after the onset of ARDS, and exceeded the proinflammatory response. The antiinflammatory mediators included IL-1ra, IL-1RII, sTNF-RI, sTNF-RII, sIL-6R, and IL-10. Each of these reached maximal concentrations in BALF on Days 1 and 3 of ARDS, and returned to near normal values by Day 14 of ARDS. This prominent antiinflammatory response in the lungs provides a mechanism for limiting the intensity of the inflammatory response in the soluble phase of lung fluids before and after the onset of ARDS.
To evaluate the balance of pro- and antiinflammatory mediators, we calculated the molar ratios of cytokines and their respective soluble receptors and receptor antagonists. The use of molar ratios to reflect net biological activity is based on the known biochemical interactions of TNF-α, IL-1β, and IL-6 with their respective antagonists or agonists, each of which interacts on a 1:1 stoichiometric basis (27-32). Findings from both infectious and noninfectious inflammatory diseases suggest that the molar balance between pro- and antiinflammatory mediators may be a better measure of the net biological effects of cytokines and may have greater clinical relevance. In studies of rheumatoid arthritis, for example, mononuclear cells from patients with active rheumatic disease release significantly higher ratios of IL-1β to IL-1ra than do those from normal control subjects (14, 15). In addition, in patients with inflammatory bowel disease, the IL-1β/IL-1ra ratio is significantly higher than in normal subjects and is closely related to disease severity (11, 33).
We were surprised to find that the molar ratios of TNF-α and IL-1β and their soluble receptors and antagonists declined by one to two orders of magnitude at the onset of ARDS whereas the molar ratio of IL-6 to sIL-6R increased. Although IL-6 has mixed pro- and antiinflammatory effects, its effects are predominantly antiinflammatory (17). Thus, unlike other inflammatory diseases, the balance between TNF-α, IL-1β, and IL-6 with their respective antagonists varies throughout the course of ARDS in a pattern that suggests a net antiinflammatory effect in the soluble phase sampled by BAL at the onset of the disease.
We also found that IL-10 peaked on Day 1 of ARDS and slowly declined to undetectable levels by Day 21. IL-10 is an antiinflammatory cytokine and because TNF-α expression and IL-10 expression are closely related (34-37), it has been hypothesized that the ratio of IL-10 to TNF-α is important in determining risk for ARDS and outcome (38). We found that the molar ratio of TNF-α to IL-10 decreased at the onset of ARDS and rose to near normal levels as ARDS progressed. Taken together, the observations about the IL-10 concentration and TNF-α/IL-10 ratio suggest significant antiinflammatory effects in the airspaces at the onset of ARDS. These results are consistent with those of Donnelly and coworkers, who found elevated concentrations of IL-10 and IL-1ra in the BALF of 28 patients with ARDS (10). However, our results differ from those of Armstrong and Millar, who found significantly lower concentrations of IL-10 and an increase in the TNF-α/IL-10 ratio in a small group of patients with ARDS (n = 10), as compared with patients at risk for ARDS (38). Overall, our measurements of IL-10, IL-6/sIL-6R, and the ratios of TNF-α and IL-1β with their respective antagonists consistently support a predominantly antiinflammatory profile in the soluble environment in the lungs of patients at the onset of ARDS.
Although the ratios of cytokines and their soluble receptors and antagonists suggest an antiinflammatory balance in the lungs, ARDS is characterized by an explosive acute inflammatory response associated with infection, trauma, and other clinical risks. To further examine the relationships between pro- and antiinflammatory cytokines and to assess the validity of the cytokine agonist:antagonist ratios, we used specific bioassays to measure the net biological activity attributable to TNF-α, IL-1β, and IL-6. The results of these bioassays differed for each cytokine. The TNF-α bioassay showed low values of TNF-α bioactivity on Day 1 of ARDS that were significantly higher on Day 7 of ARDS, when the ratio of TNF-α to its inhibitors increased. The IL-1β bioassay showed measurable IL-1β activity at the onset of ARDS, and correlated most strongly with immunologically measured IL-1β. The IL-6 bioassay was strongly related to the measured IL-6 and sIL-6R values as well as to the IL-6/sIL-6R ratio.
The parallel rise between TNF-α bioactivity from Day 1 to Day 7 of ARDS and the rising ratios of TNF-α to its soluble receptors suggests that the molar ratio may be a more accurate measure of net TNF-α activity. Measurements of TNF-α concentration by immunological assays have several potential limitations. Specifically, the TNF-α immunoassays could detect biologically inactive precursors or be unable to detect active, protein-bound TNF-α. Most importantly, the immunoassays do not measure either the presence or the biological effects of the soluble TNF-α receptors.
In contrast to the findings with TNF-α, there was a significant relationship between the IL-1β bioassay and the immunologic measurement of IL-1β. The lack of a significant correlation between IL-1ra or sIL-1RII and the net biological activity of IL-1β is not surprising. IL-1 receptor I (IL-1RI) is the primary signal transducing receptor for IL-1β and is the site of inhibition by IL-1ra (28). Although not measured in this study, soluble IL-1RI is released during inflammation and maintains its ability to bind IL-1ra, significantly reducing the availability of IL-1ra and dampening its effect on IL-1β activity (29). In addition, the cell-bound IL-1RI receptor is extremely sensitive to IL-1β and requires only low concentrations of IL-1β to initiate transmembrane signaling because of postreceptor amplification through multiple phosphorylation steps (39). Thus, our results support the conclusion that despite the presence of significant concentrations of specific IL-1β antagonists, the concentration of free IL-1β remains the best immunologic measure of net IL-1β bioactivity in lung fluids.
The IL-6 bioassay had a strong relationship with both the soluble ligand and its soluble receptor as well as the IL-6/sIL-6R ratio. Linear regression analysis, however, did not identify a significant independent contribution of sIL-6R to the net IL-6 bioactivity, as measured in the B9 hybridoma assay. This result might seem surprising, as previous work demonstrated the sensitivity of B9 cells to the agonist effects of sIL-6R at concentrations as low as 5 ng/ml (40). This effect diminishes, however, at concentrations of IL-6 above 10 pg/ml (41). In BALF of patients at risk for ARDS and with established ARDS, the concentration of IL-6 was always higher than 10 pg/ml. Thus, our results support the conclusion that despite significant concentrations of sIL-6R in BALF, the contribution of sIL-6R to the net bioactivity of IL-6 in BALF from patients with ARDS is small, most likely because of the high IL-6 concentrations.
A second goal of this study was to determine whether relationships existed between the presence or activity of the inflammatory mediator systems and either the clinical severity or outcome of ARDS. We found significant relationships between lung compliance and the IL-1β/sIL-1RII, IL-6/sIL-6R, TNF/sTNF-RI, and TNF/sTNF-RII ratios on Days 1 and 3 of ARDS. There were also significant relationships between the degree of hypoxemia and TNF/sTNF-RII on Day 1 of the at risk period and with IL-1β/IL-1ra on Day 7 of ARDS. This may be a function of endothelial/epithelial capillary leak, as suggested by the strong relationship between soluble receptors and the total protein concentration measured in the BAL of patients at risk of ARDS and with established ARDS. This could reflect either local production in the lungs or movement from the plasma into the lungs when significant injury occurs to the endothelial–epithelial barrier (42).
The IL-6/sIL-6R ratio was the only molar ratio that was significantly related to outcome, with a higher ratio associated with a higher risk of death. As in prior studies, we found a significant relationship between IL-1β, IL-6, and subsequent mortality on Day 7 of ARDS (2, 5, 6). This finding is consistent with the central role of IL-1β in ARDS and is additional evidence that sustained inflammation in the lungs is associated with a worse outcome.
Several limitations of this study should be noted. First, the BAL method samples only soluble constituents in the airspaces. Biologically active tissue and membrane-bound cytokines and other ligands are not measured by this technique. Examples include membrane-bound TNF-α (43), soluble IL-1RI and IL-1R accessory protein (44), and soluble autoantibodies to IL-6 as well as a soluble form of signal-transducing gp130 protein (9, 45-47). From a clinical standpoint, the study population does not include patients with ARDS who improved rapidly and were extubated, or patients who were the most severely ill and were excluded for safety reasons. However, the patient population does provide an accurate representation of patients with ARDS who remain intubated and mechanically ventilated.
In summary, we have found that at the onset of ARDS, high concentrations of antiinflammatory cytokines produce markedly lower molar ratios of proinflammatory to antiinflammatory mediators in the lungs. This provides a mechanism for dampening the otherwise intense inflammatory response in the airspaces. These results highlight the complexity of the inflammatory response in the lungs of patients with ARDS and show that individual cytokine measurements cannot be considered in isolation but must be understood in the context of the balance between agonistic and antagonistic responses in the lungs.
The authors thank Ellen Caldwell and Todd Freudenberger for advice about the statistical analysis of the data.
Supported in part by grants HL30542 and AI29103 from the National Institutes of Health and the Medical Research Service of the Department of Veterans Affairs.
1. | Ware LB, Matthay MAThe acute respiratory distress syndrome. N Engl J Med342200013341349 |
2. | Siler TM, Swierkosz JE, Hyers TM, Fowler AA, Webster ROImmunoreactive interleukin-1 in bronchoalveolar lavage fluid of high-risk patients and patients with the adult respiratory distress syndrome. Exp Lung Res151989881894 |
3. | Schutte H, Lohmeyer J, Rosseau S, Ziegler S, Siebert C, Kielisch H, Pralle H, Grimminger F, Morr H, Seeger WBronchoalveolar and systemic cytokine profiles in patients with ARDS, severe pneumonia and cardiogenic pulmonary oedema. Eur Respir J9199618581867 |
4. | Suter PM, Suter S, Girardin E, Roux-Lombard P, Grau GE, Dayer JMHigh bronchoalveolar levels of tumor necrosis factor and its inhibitors, interleukin-1, interferon, and elastase, in patients with adult respiratory distress syndrome after trauma, shock, or sepsis. Am Rev Respir Dis145199210161022 |
5. | Meduri GU, Kohler G, Headley S, Tolley E, Stentz F, Postlethwaite AInflammatory cytokines in the BAL of patients with ARDS. Persistent elevation over time predicts poor outcome. Chest108199513031314 |
6. | Goodman RB, Strieter RM, Martin DP, Steinberg KP, Milberg JA, Maunder RJ, Kunkel SL, Walz A, Hudson LD, Martin TRInflammatory cytokines in patients with persistence of the acute respiratory distress syndrome. Am J Respir Crit Care Med1541996602611 |
7. | Opal SM, Cross AS. Clinical trials for severe sepsis. Past failures, and future hopes. Infect Dis Clin North Am 1999;13:285–297, vii. |
8. | Pittet JF, Mackersie RC, Martin TR, Matthay MABiological markers of acute lung injury: prognostic and pathogenetic significance. Am J Respir Crit Care Med155199711871205 |
9. | Muller-Newen G, Kuster A, Hemmann U, Keul R, Horsten U, Martens A, Graeve L, Wijdenes J, Heinrich PCSoluble IL-6 receptor potentiates the antagonistic activity of soluble gp130 on IL-6 responses. J Immunol161199863476355 |
10. | Donnelly SC, Strieter RM, Reid PT, Kunkel SL, Burdick MD, Armstrong I, Mackenzie A, Haslett CThe association between mortality rates and decreased concentrations of interleukin-10 and interleukin-1 receptor antagonist in the lung fluids of patients with the adult respiratory distress syndrome. Ann Intern Med1251996191196 |
11. | Casini-Raggi V, Kam L, Chong YJ, Fiocchi C, Pizarro TT, Cominelli FMucosal imbalance of IL-1 and IL-1 receptor antagonist in inflammatory bowel disease. A novel mechanism of chronic intestinal inflammation. J Immunol154199524342440 |
12. | Nicoletti F, Patti F, DiMarco R, Zaccone P, Nicoletti A, Meroni P, Reggio ACirculating serum levels of IL-1ra in patients with relapsing remitting multiple sclerosis are normal during remission phases but significantly increased either during exacerbations or in response to IFN-beta treatment. Cytokine81996395400 |
13. | Miller LC, Lynch EA, Isa S, Logan JW, Dinarello CA, Steere ACBalance of synovial fluid IL-1 beta and IL-1 receptor antagonist and recovery from Lyme arthritis. Lancet3411993146148 |
14. | Shingu M, Fujikawa Y, Wada T, Nonaka S, Nobunaga MIncreased IL-1 receptor antagonist (IL-1ra) production and decreased IL-1 beta/IL-1ra ratio in mononuclear cells from rheumatoid arthritis patients. Br J Rheumatol3419952430 |
15. | Chomarat P, Vannier E, Dechanet J, Rissoan MC, Banchereau J, Dinarello CA, Miossec PBalance of IL-1 receptor antagonist/IL-1 beta in rheumatoid synovium and its regulation by IL-4 and IL-10. J Immunol154199514321439 |
16. | Akalin H, Akdis AC, Mistik R, Helvaci S, Kilicturgay KCerebrospinal fluid interleukin-1 beta/interleukin-1 receptor antagonist balance and tumor necrosis factor-alpha concentrations in tuberculous, viral and acute bacterial meningitis. Scand J Infect Dis261994667674 |
17. | Tilg H, Dinarello CA, Mier JWIL-6 and APPs: anti-inflammatory and immunosuppressive mediators. Immunol Today181997428432 |
18. | Steinberg KP, Milberg JA, Martin TR, Maunder RJ, Cockrill BA, Hudson LDEvolution of bronchoalveolar cell populations in the adult respiratory distress syndrome. Am J Respir Crit Care Med1501994113122 |
19. | Matute-Bello G, Liles WC, Radella F, Steinberg KP, Ruzinski JT, Jonas M, Chi EY, Hudson LD, Martin TRNeutrophil apoptosis in the acute respiratory distress syndrome. Am J Respir Crit Care Med156199719691977 |
20. | Greene KE, Wright JR, Steinberg KP, Ruzinski JT, Caldwell E, Wong WB, Hull W, Whitsett JA, Akino T, Kuroki Y, Nagae H, Hudson LD, Martin TRSerial changes in surfactant-associated proteins in lung and serum before and after onset of ARDS. Am J Respir Crit Care Med160199918431850 |
21. | Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L, Lamy M, LeGall JR, Morris A, Spragg RReport of the American– European Consensus Conference on acute respiratory distress syndrome: definitions, mechanisms, relevant outcomes, and clinical trial coordination. Consensus Committee. J Crit Care919947281 |
22. | Martin TR, Ruzinski JT, Wilson CB, Skerrett SJEffects of endotoxin in the lungs of neonatal rats: age-dependent impairment of the inflammatory response. J Infect Dis1711995134144 |
23. | Flick DA, Gifford GEComparison of in vitro cell cytotoxic assays for tumor necrosis factor. J Immunol Methods681984167175 |
24. | Pugin J, Ricou B, Steinberg KP, Suter PM, Martin TRProinflammatory activity in bronchoalveolar lavage fluids from patients with ARDS, a prominent role for interleukin-1. Am J Respir Crit Care Med153199618501856 |
25. | Santhanam U, Avila C, Romero R, Viguet H, Ida N, Sakurai S, Sehgal PBCytokines in normal and abnormal parturition: elevated amniotic fluid interleukin-6 levels in women with premature rupture of membranes associated with intrauterine infection. Cytokine31991155163 |
26. | Aarden L, Lansdorp P, De Groot EA growth factor for B cell hybridomas produced by human monocytes. Lymphokines101985175185 |
27. | Heaney ML, Golde DWSoluble receptors in human disease. J Leukoc Biol641998135146 |
28. | Arend WP, Joslin FG, Thompson RC, Hannum CHAn IL-1 inhibitor from human monocytes. Production and characterization of biologic properties. J Immunol143198918511858 |
29. | Burger D, Chicheportiche R, Giri JG, Dayer JMThe inhibitory activity of human interleukin-1 receptor antagonist is enhanced by type II interleukin-1 soluble receptor and hindered by type I interleukin-1 soluble receptor. J Clin Invest9619953841 |
30. | Colotta F, Re F, Muzio M, Bertini R, Polentarutti N, Sironi M, Giri JG, Dower SK, Sims JE, Mantovani AInterleukin-1 type II receptor: a decoy target for IL-1 that is regulated by IL-4. Science2611993472475 |
31. | Murakami M, Hibi M, Nakagawa N, Nakagawa T, Yasukawa K, Yamanishi K, Taga T, Kishimoto TIL-6-induced homodimerization of gp130 and associated activation of a tyrosine kinase. Science260199318081810 |
32. | Taga T, Hibi M, Hirata Y, Yamasaki K, Yasukawa K, Matsuda T, Hirano T, Kishimoto TInterleukin-6 triggers the association of its receptor with a possible signal transducer, gp130. Cell581989573581 |
33. | Andus T, Daig R, Vogl D, Aschenbrenner E, Lock G, Hollerbach S, Kollinger M, Scholmerich J, Gross VImbalance of the interleukin 1 system in colonic mucosa—association with intestinal inflammation and interleukin 1 receptor antagonist genotype 2. Gut411997650657 |
34. | Bogdan C, Vodovotz Y, Nathan CMacrophage deactivation by interleukin 10. J Exp Med174199115491555 |
35. | Ralph P, Nakoinz I, Sampson-Johannes A, Fong S, Lowe D, Min HY, Lin LIL-10, T lymphocyte inhibitor of human blood cell production of IL-1 and tumor necrosis factor. J Immunol1481992808814 |
36. | Dickensheets HL, Freeman SL, Smith MF, Donnelly RPInterleukin-10 upregulates tumor necrosis factor receptor type-II (p75) gene expression in endotoxin-stimulated human monocytes. Blood90199741624171 |
37. | Wanidworanun C, Strober WPredominant role of tumor necrosis factor-alpha in human monocyte IL-10 synthesis. J Immunol151199368536861 |
38. | Armstrong L, Millar ABRelative production of tumour necrosis factor alpha and interleukin 10 in adult respiratory distress syndrome. Thorax521997442446 |
39. | Dinarello CA. Interleukin-1. In: Thomson A, editor. The cytokine handbook. San Diego, California: Academic Press; 1998. p. 35–72. |
40. | Thibault V, Terlain B, Gauldie JCharacterization and biologic activities of recombinant rat soluble interleukin-6 receptor. J Interferon Cytokine Res161996973981 |
41. | Diamant M, Hansen MB, Rieneck K, Svenson M, Yasukawa K, Bendtzen KStimulation of the B9 hybridoma cell line by soluble interleukin-6 receptors. J Immunol Methods1731994229235 |
42. | Munford RS, Pugin JNormal responses to injury prevent systemic inflammation and can be immunosuppressive. Am J Respir Crit Care Med1632001316321 |
43. | Armstrong L, Thickett DR, Christie SJ, Kendall H, Millar ABIncreased expression of functionally active membrane-associated tumor necrosis factor in acute respiratory distress syndrome. Am J Respir Cell Mol Biol2220006874 |
44. | Jensen LE, Muzio M, Mantovani A, Whitehead ASIL-1 signaling cascade in liver cells and the involvement of a soluble form of the IL-1 receptor accessory protein. J Immunol164200052775286 |
45. | Paysant J, Blanque R, Vasse M, Soria C, Soria J, Gardner CRFactors influencing the effect of the soluble IL-6 receptor on IL-6 responses in HepG2 hepatocytes. Cytokine122000774779 |
46. | Hansen MB, Svenson M, Diamant M, Abell K, Bendtzen KInterleukin-6 autoantibodies: possible biological and clinical significance. Leukemia9199511131115 |
47. | Homann C, Hansen MB, Graudal N, Hasselqvist P, Svenson M, Bendtzen K, Thomsen AC, Garred PAnti-interleukin-6 autoantibodies in plasma are associated with an increased frequency of infections and increased mortality of patients with alcoholic cirrhosis. Scand J Immunol441996623629 |
This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournals.org