INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Pleural diseases are characterized by the presence of specific subsets of leukocytes within the pleural space (1). Leukocyte extravasation implies the interaction of endothelial cells with leukocytes through cell-adhesion molecules, together with the existence of a chemotactic or haptotactic gradient through which leukocytes then move (2).

Interleukin-8 (IL-8) is recognized as a potent and quite specific chemoattractant for neutrophils; although its chemotactic activity has also been demonstrated for basophils and T lymphocytes, the effect of IL-8 on T cells seems to require the presence of additional leukocyte populations, according to a recent study (3).

IL-8 is a chemokine belonging to the family of the CxC cytokines that is produced by a variety of cell types in response to other inflammatory stimuli (4, 5). It also induces activation of neutrophils and the expression or activation of adhesion molecules in the neutrophil cell membrane (6).

Neutrophils, in addition to their microbicidal effect, release reactive oxygen and nitrogen species into the extracellular millieu, in addition to releasing their granular content, which contains several serine and neutral proteases that can produce tissue injury. Among the neutrophil proteases, elastase has been implicated in both chronic and acute inflammatory damage, and oxidant species, partly produced by the action of another granular enzyme, myeloperoxidase (MPO), potentiate its effects (7, 8). Therefore, it is likely that the products derived from neutrophil activation are related to the evolution of parapneumonic pleural effusions from the noncomplicated to the complicated state.

Taking into account that several cells in the pleural space can release IL-8, and considering the role of this chemokine in neutrophil migration and activation, we assessed IL-8, differential leukocyte count, and two indicators of neutrophil degranulation (neutrophil elastase [NE] and myeloperoxidase [MPO]) in blood and pleural fluid (PF) of patients with pleural effusions of various etiologies, in order to: (1) determine whether PF IL-8 is produced in the pleural space; and (2) establish the relation of this cytokine to the neutrophil number and activation state in infectious pleural effusions.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Groups

The groups studied consisted of 219 patients (59% male, 41% female; age range: 15 to 95 yr) with newly diagnosed pleural effusion who were consecutively admitted to the emergency area of an 800-bed university hospital. PF and blood were obtained from the patients during diagnostic thoracentesis after informed consent was obtained. The study was approved by the Comissió d'Ética i Assaigs Clínics dels Hospitals Vall d'Hebron.

Effusions were categorized as exudates or transudates according to the Light criteria (9).

Empyema (n = 32, 86% males, 14% females; mean age: 55 yr) was defined as an effusion that met one or more of the following criteria: purulent fluid on macroscopic examination, positive Gram stain and/ or growth of organisms in culture, and PF pH < 7.2 or glucose < 60 mg/dl in association with pneumonia. Parapneumonic effusions (n = 30, 67% males, 33% females; mean age: 56 yr) were those with a glucose concentration > 60 mg/dl and pH > 7.2, and no organisms seen on Gram stain or found on PF culture in a patient with pneumonia. Tuberculous effusions (n = 42, 43% males, 57% females; mean age: 32 yr) were defined as those with a positive Ziehl-Nielssen stain or Löwenstein-Jensen culture of PF or pleural biopsy specimens, or an adenosine deaminase (ADA) catalytic concentration in PF > 43 U/L plus lymphocytes as > 50% of the total cell count. Malignant effusions (n = 28, 53% males, 47% females; mean age: 71 yr) were exudates associated with the presence of malignant cells in PF or a pleural histology positive for neoplasia. The primary neoplasia was in the lung in 21% of cases, chest in 14%, ovary in 11%, and pleura in 8%, and was gastrointestinal or hepatic in 3% or of unknown origin in 36%. Except in cases of neoplasia of primary pleural origin (mesothelioma), the neoplastic pleural effusion was diagnosed in the late stage of the illness. Transudates (n = 36, 59% males 41% females; mean age: 71 yr) were defined as effusions with a protein concentration < 3 g/dl, and were associated with congestive heart failure. Patients with exudative effusions that did not meet the aforementioned criteria were considered to have nonspecific pleural effusion (n = 51, 59% males and 41% females; mean age: 66 yr).

Sample Processing and Leukocyte Count

Blood and PF were collected in tubes containing the tripotassium salt of ethylene diamine tetraacetic acid (K₃EDTA). Total and differential leukocyte counts were immediately performed on blood samples in a Coulter S-Plus IV cell counter (Coulter, Hialeah, FL) and in PF by cytocentrifugation (Cytospin 2; Shandon Instruments, Sewickley, PA) at 2,000 × g for 8 min, followed by May-Grunwald-Giemsa staining.

For IL-8, NE and MPO determination, blood and PF were spun at 1,800 × g and 4° C within 1 h of collection; the plasma and PF supernatants were kept in aliquots at 30° C until analysis.

IL-8 Immunoassay

Immunoreactive IL-8 in plasma and PF was quantitated with an enyme-linked immunosorbent assay (ELISA; Bender Medsystems, Vienna, Austria), using a murine monoclonal antihuman IL-8 antibody as the capture antibody and a horseradish peroxidase-conjugated polyclonal antibody as the labeling antibody. In our laboratory, the detection limit of the method was 8 ng/L. Concentrations of IL-8 below the detection limit were equaled to 1 ng/L for statistical purposes.

NE and MPO Determination

NE was assessed with a homogeneous immunoactivation method (IMAC; Merck, Darmstadt, Germany) that measures both free and ₁-antiproteinase-bound human NE. The method was adapted to a Hitachi 917 autoanalyzer, and sample blanks were processed to correct for endogenous oxidants and antioxidants.

A competitive double-antibody radioimmunoassay (MPO; Pharmacia, Uppsala, Sweden) with a ¹²⁵I-labeled antigen was used to quantify MPO.

Statistical Analysis

The Kolmogorov-Smirnov test was used to check the distribution fitting of the data. Significant differences between raw data and normal distribution were observed. Log₁₀-transformed PF data, but not blood data, fitted a normal distribution. Therefore, nonparametric methods were used when comparing blood values among groups (Kruskal- Wallis analysis of variance [ANOVA]) and investigating for correlations among blood and PF values (Spearman's rank correlation coefficient). Parametric methods were employed with the log₁₀-transformed PF data: ANOVA with Bonferronis's correction was used to make comparisons among groups, and the univariate Pearson's correlation was used to test the relationships between two variables. ANOVA were done with and without including patients' age and gender as covariates.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Leukocyte and Inflammatory Markers in Blood

Blood neutrophil and lymphocyte counts, as well as plasma IL-8, NE, and MPO concentrations found in each group, are summarized in Table 1. No significant differences among groups were detected in any of the components studied. IL-8 plasma concentrations were below the detection limit in 68.8% of patients with empyema, and 50.0% of those with parapneumonic, 64.3% with tuberculous, and 78.6% with neoplastic effusions; 58.3% with transudates; and 52.9 % with nonspecific effusions.

PF Differential Leukocyte Count, and IL-8, NE, and MPO Concentrations

The PF characteristics corresponding to each group of patients are shown in Table 2. The patients with empyema had higher numbers of neutrophils than those in any of the other groups (p < 0.001). Parapneumonic PF showed higher neutrophil counts than tuberculous (p = 0.004), or neoplastic PF or transudates (p < 0.001). The neutrophil counts in transudates were also significantly lower than in nonspecific effusions (p < 0.001). Statistically significant differences in lymphocyte count were observed only in the comparison of tuberculous PFs with transudates (p < 0.001) and empyema (p = 0.004).

Figure 1 shows the individual IL-8 PF concentrations. As was seen with the neutrophil count, the highest IL-8 values (median: P₂₅ to P₇₅) were found in empyema (9,544 ng/L; 1,372 to 105,688 ng/L; p < 0.001). Although the parapneumonic and tuberculous groups showed very similar concentrations (104 ng/L; 10 to 239 ng/L; and 138 ng/L; 28 to 287 ng/L, respectively), statistical analysis evidenced significant differences only between tuberculous PFs and transudates (13 ng/L; < 8 to 43 ng/L; p < 0.001) and between tuberculous and nonspecific effusions (22 ng/L; < 8 to 88 ng/L; p = 0.002).

View larger version (12K): [in this window] [in a new window]   Figure 1.   Pleural fluid IL-8 concentrations in the different diagnostic groups. Pneumo = parapneumonic; TB = tuberculous; Neop = neoplastic; Trans = transudates; Nonspec = nonspecific. Horizontal line = median. *p < 0.001 with respect to each of the other groups; p < 0.001 compared with transudates and p = 0.002 compared with nonspecific effusions.

The NE and MPO concentrations within the different diagnostic groups are shown in Figures 2 and 3. Concentrations of both these enzymes changed in parallel among the groups, and were much higher in empyema (NE: 23,040 µg/L, 2,295 to 125,025 µg/L; MPO: 31,440 µg/L, 6,000 to 182,425 µg/L) than in any of the other groups (p < 0.001). Surprisingly, parapneumonic and tuberculous PFs presented very similar levels of NE (218 µg/L, 58 to 974 µg/L; 280 µg/L, 84 to 686 µg/L, respectively) and MPO (807 µg/L, 282 to 1,940 µg/L; 1,462 µg/L, 606 to 4,134 µg/L). Concentrations of NE and MPO in neoplastic and nonspecific effusions were lower than in tuberculous PFs (neoplastic NE: 65 µg/L, 27 to 196 µg/L; MPO: 249 µg/L, 158 to 612 µg/L; nonspecific NE: 71 µg/L, 28 to 182 µg/L; MPO: 306 µg/L, 156 to 530 µg/L), although significant differences in MPO were evidenced only between tuberculous and neoplastic PF and in NE and MPO between tuberculous and nonspecific PF. Transudates showed the lowest levels: NE 20 µg/L, 12 to 39 µg/L; p < 0.001 compared with parapneumonic or tuberculous; PF: MPO 81 µg/L, 54 to 130 µg/L; p < 0.001 compared with all of the other groups. The inclusion of age and gender as covariates in the statistical analysis did not change the significance of the differences.

View larger version (15K): [in this window] [in a new window]   Figure 2.   NE concentrations in pleural effusions of different etiologies. Pneumo = parapneumonic; TB = tuberculous; Neop = neoplastic; Trans = transudates; Nonspec = nonspecific. Horizontal line = median. *p < 0.001 compared with all of the other groups; p < 0.001 compared with transudates; p < 0.001 with respect to transudates and nonspecific effusions.

View larger version (15K): [in this window] [in a new window]   Figure 3.   MPO in pleural fluid of patients with different types of pleural effusion. Pneumo = parapneumonic; TB = tuberculous; Neop = neoplastic; Trans = transudates; Nonspec = nonspecific. Horizontal line = median. *p < 0.001 with respect to each of the other groups; p < 0.001 compared with neoplastic effusions, transudates, and nonspecific effusions; significantly lower than all the other groups (p < 0.001).

Relationship Between Blood and PF Inflammatory Cells and Inflammatory Mediators

To determine whether PF levels of the inflammatory mediators studied were influenced by an increased state of permeability, we also analyzed the correlation between IL-8, NE, MPO in blood and PF and lymphocyte and neutrophil count. The highest correlation was found between plasma and PF IL-8 in transudates (r = 0.6745, p < 0.001). Other significant but weaker correlations were observed in parapneumonic patients for IL-8 (r = 0.4802, p = 0.007), MPO (r = 0.4023, p = 0.027), and lymphocyte count (r = 0.4502, p = 0.013); in tuberculous patients for NE (r = 0.3842, p = 0.012) and MPO (r = 0.3459, p = 0.025), and in patients with nonspecific effusions for IL-8 (r = 0.3326, p = 0.017) and lymphocyte count (r = 0.5436, p < 0.001).

Correlation Analysis of Inflammatory Mediators in PF

The significant correlations found between PF inflammatory cells, IL-8, and neutrophil markers within each group are summarized in Table 3. No significant correlation was found between IL-8 and neutrophil number in any group. In contrast, IL-8 was highly correlated with NE and MPO in empyema, and a significant correlation was also observed between IL-8 and both NE and MPO in the tuberculous and nonspecific groups, but not in the parapneumonic group. The neutrophil count showed a high degree of association with NE and MPO in empyema and parapneumonic PF, and a weaker association in the nonspecific effusions. As expected, the degree of correlation between NE and MPO was very high in all the groups except in transudates, in which all patients had low values of both enzymes.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The factors that guide migration of the different leukocyte types to the pleural space are not completely understood. In recent years, some authors have analyzed the chemotactic cytokine IL-8 in PFs (10), focusing on relating IL-8 to neutrophil influx, but not to markers of neutrophil degranulation. Our study, designed to assess this aspect of the inflammatory response in the pleural space, showed a close relationship between IL-8, NE, and MPO in empyema and tuberculous effusions.

Production of IL-8 in the Pleural Space

We first investigated whether IL-8 in blood contributed to the PF concentrations of this cytokine. Among the exudates studied, we observed significantly positive, although weak, correlations between blood and PF IL-8 concentrations only in parapneumonic and nonspecific effusions. This relationship was closer in the transudates. Moreover, IL-8 was undetectable in more than half of the plasma samples independently of the diagnosis, whereas it was detected in a high proportion of pleural exudate samples. These results taken together favor the concept of a local production of IL-8 in exudative effusions, rather than a passive diffusion of this chemokine from plasma to the pleural compartment. Miller and associates (12) observed lower IL-8 concentrations in transudates than in other types of pleural effusion, both when absolute levels of IL-8 and levels normalized for the protein content were compared, thus also suggesting a compartmentalization of this chemokine in the pleural space. This is not surprising, since several cell types in or near the pleural space have been shown to produce IL-8, including pleural mesothelial cells, alveolar macrophages (AM), lung fibroblasts, lymphocytes, and neutrophils (5, 6).

IL-8 Concentration in Infectious PFs

The finding of higher IL-8 concentrations in PF in empyema than in any of the other groups was consistent with the massive neutrophilic infiltration of PF in this condition. However, the fact that no differences were observed between parapneumonic and tuberculous PF is in clear disparity with the predominance of neutrophils in parapneumonic PF and lymphocytes in the latter group. In contrast, Antony and colleagues (10), reported significantly higher IL-8 in parapneumonic than in tuberculous PF, although only seven patients were included in this group in their study. In a recent work, all infectious PFs (empyema, parapneumonic and tuberculous) were considered collectively, and showed median IL-8 levels that were higher than in noninfectious pleural effusions (14). Our study also evidenced higher IL-8 levels in infectious than in noninfectious pleural effusions, independently of whether the infection was caused by an intracellular or an extracellular pathogen.

The presence of high IL-8 concentrations in PF of tuberculous patients could be explained by the fact that monocytes/ macrophages are among the predominant cells in tuberculous granuloma and in light of the showing by Friedland and coworkers (15) that human monocyte cell lines, following phagocytosis of Mycobacterium tuberculosis, release from eight to 15 times more IL-8 than do the same cells stimulated with lipopolysaccharide (LPS), which is regarded as a powerful inducer of IL-8. Furthermore, Zhang and coworkers (16) demonstrated that IL-8 protein secretion is increased in supernatants of macrophages and in broncoalveolar lavage fluid (BALF) from patients with pulmonary tuberculosis as compared with normal controls, and that M tuberculosis and its cell-wall components stimulated IL-8 synthesis and release in AM. Considering the chemoattractant effects of this cytokine on neutrophils and lymphocytes, and particularly activated T lymphocytes (17), these data support the concept that IL-8 may be an important contributor to granuloma formation by recruiting both acute and chronic inflammatory cells to the site of the infection.

Relation of IL-8 to the Number of Neutrophils in PFs

Although neutrophil chemotactic activity was the first biologic property of IL-8 to be identified, considerable evidence indicates that it is not the only neutrophil chemotaxin in the pleural space. Broaddus and colleagues (11) showed that neutrophil chemotaxis of empyema fluids decreased by 65 ± 5% after treatment with anti-IL-8 F(ab')antibody, whereas Antony and associates (10) found only 32.3 ± 6.4% suppression in similar neutralization experiments. In other studies (12), removal of IL-8 from PF resulted in changes in neutrophil chemotactic activity that were very dissimilar and depended on the cause of the effusion, suggesting that the contribution of IL-8 to neutrophil movement into the pleural space may vary in different disease states.

If IL-8 were the major factor contributing to neutrophil influx into the pleural space, a strong correlation between PF neutrophil count and IL-8 would be expected. The results in regard to this are not in complete agreement. Some authors describe a positive correlation when all of the fluids studied are analyzed together (11), but not when various fluid groups are analyzed separately (13); others have observed a correlation only in empyema (10). In accordance with Miller and coworkers' results (12), we could not demonstrate a correlation between PF IL-8 and neutrophil count, even in empyema.

The degree of association between IL-8 and neutrophil infiltration found in other biologic fluids varies. In BALF a positive correlation has been described in patients with pneumonia (18) and diffuse panbronchiolitis (19), but not in patients with sarcoidosis (20). Similarly, no relationship was found in synovial fluid from patients with rheumatoid arthritis (21), and the degree of association in cerebrospinal fluid (CSF) differed according to the causative agent of infection (22). Interestingly, Yokoyama and colleagues (26), describe a positive correlation between CSF IL-8 and neutrophil number in aseptic meningitis only within 12 h of disease onset, but not later, and suggest that IL-8 triggers a rapid but transient migration of neutrophils into the CSF. The kinetics of IL-8 production and elimination in other body compartments is not known, but these differences in time-dependent IL-8 release and neutrophil migration could account for the lack of correlation observed in some body fluids.

NE and MPO in the Different Diagnostic Groups

With regard to the neutrophil-derived enzymes NE and MPO, we observed a profile similar to that for IL-8. As expected, and in agreement with other studies, we found the highest PF concentrations of both enzymes in empyema. However, the finding of similar levels of NE and MPO in the parapneumonic and tuberculous groups is surprising, taking into account the two groups' differences in neutrophil count. Increases in NE in tuberculous PF have also been reported by other authors (27).

The reason for the lack of any difference between the parapneumonic and tuberculous groups in either NE or MPO levels is not clear. It is generally accepted that NE and MPO are specific neutrophil markers, because of the high content of these enzymatic proteins in the azurophil granules of these cells. Elastase has been found as a constitutive product of T cells (30), and the production of an NE-like activity in activated lymphocytes has recently been described (31). On the other hand, both enzymes have been found in monocytes (although in concentrations much lower than in neutrophils), although maturation of monocytes to macrophages apparently leads to loss of most of the intracellular content of elastase and MPO (32). The presence of activated T-lymphocytes in tuberculous PF has been documented; moreover it is well known that T lymphocytes and monocytes/macrophages are the main infiltrating cells in tuberculous granulomas. Consequently, both cell types, together with neutrophils, can contribute to the PF NE concentration in tuberculosis. A lymphocytic origin for this NE seems unlikely, however, because increases in NE were coincident with those of MPO in our tuberculous effusions. Whether NE and MPO are related to monocyte/macrophage maturation, and therefore to the extent of granuloma formation, or whether they represent a higher degree of neutrophil degranulation in tuberculous pleural effusions, is a matter of speculation.

Relation of IL-8 with NE and MPO

To our knowledge, there are no reported studies focusing on the relationship between IL-8 levels and neutrophil degranulation status in PFs. Our results demonstrate this relationship, at least in empyema, thus suggesting a greater role for IL-8 in neutrophil degranulation than in neutrophil recruitment in these fluids. In tuberculous pleural effusions both NE and MPO correlated positively with IL-8 but not with neutrophil count, thus suggesting that in this group these enzymes are released by cells other than neutrophils. However a higher degree of neutrophil activation or destruction in tuberculous fluids, either at the moment of PF analysis or previously, at the beginning of the infection, cannot be excluded.

The effect of IL-8 on neutrophil degranulation has been demonstrated in vitro, and there are also data to support its occurrence in vivo (3, 5, 6). In recent observations, Boutten and associates (18) reported a close correlation between IL-8 and NE in the BALF of patients with pneumonia, and interestingly, Chanez and colleagues (35) were able to demonstrate a significant correlation between BALF IL-8 and MPO, but not between IL-8 and neutrophil count, in patients with asthma and chronic bronchitisresults that coincide with ours in empyematous PF.

Although similar concentrations of IL-8, NE, and MPO have been found in tuberculous and parapneumonic effusions, it is conceivable that the pathophysiologic significance of these results could differ in chronic and in acute infections. In parapneumonic PFs, the inflammatory mediators seem to reflect a high degree of neutrophil degranulation that could be responsible for complication of the disease. In tuberculous effusions, on the other hand, these mediators could be related to the extent of granuloma formation; therefore, increases in these inflammatory mediators in tuberculous PF could indicate either an effective pleural response to limit the infection or a detrimental effect on the pleural space.

In summary, we have shown that levels of IL-8, NE, and MPO are higher in infectious than in noninfectious PFs, although it is likely that these inflammatory mediators have a different cellular origin in tuberculous than in parapneumonic effusions. Furthermore, our findings demonstrate a relationship between IL-8 and markers of neutrophil (or leukocyte) degranulation, but not between IL-8 and the number of neutrophils or lymphocytes present in empyematous or tuberculous pleural effusions. Additional prospective studies will help to determine whether these inflammatory markers reflect the evolution of the infection toward control of the pathogen or toward the development of pleural fibrous complications.