INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Idiopathic interstitial pneumonia (IIP) is a group of lung diseases of unknown etiology characterized by the accumulation of inflammatory cells and fibrosis in pulmonary parenchyma and interstitium. It has been pointed out that IIP has several subgroups such as usual and nonspecific interstitial pneumonia (UIP and NSIP) (1). As a consequence of inflammation, IIP is accompanied by an increased deposition of collagen, which leads to a progressive loss of alveolar-capillary units. In this process, several kinds of lung cells produce cytokines, which can modulate pulmonary fibrosis (4, 5). Although transforming growth factor- (TGF-), platelet-derived growth factor (PDGF), tumor necrosis factor- (TNF-) and basic fibroblast growth factor are well known to have fibrogenic functions (4, 6), inhibitory mediators including IL-6 are also important in limiting fibroproliferative responses. Interleukin-6 (IL-6) is produced in the lung by interstitial fibroblasts (9), alveolar macrophages (10), large-vessel endothelial cells (11), and bronchial epithelial cells (12). It is elevated in chronic inflammatory processes of the lung such as allogeneic transplantation (13) and bleomycin-induced fibrosis (14) in animal models, and a variety of human interstitial lung diseases including pneumoconiosis, sarcoidosis, pigeon breeders' disease, and idiopathic pulmonary fibrosis (IPF) (15).

The inhibitory effect of IL-6 was reported in a mouse model of hypersensitivity pneumonitis (19), in which a reduction of fibrosis was achieved by administration of IL-6. However, IL-6 shows a fibrogenic effect on a certain subset of fibroblasts obtained from patients with IPF and those from murine lung (20, 21). A direct relationship between IL-6 and lymphocytosis in the lung was revealed by experiments using local overexpression of IL-6 and IL-6 receptor (IL-6R) genes in rat (22, 23). In the bronchial tree of rat transferred with IL-6 or IL-6R genes, prominent changes were a massive infiltration of lymphocytes in the pulmonary parenchyma with a little fibrosis on histology and a concurrently increased number of lymphocytes in BAL cells. The majority of the infiltrating lymphocytes were helper and suppressor subsets of T lymphocytes (22). In contrast, local overexpression of TGF- and PDGF-B genes induces significant proliferation of fibroblasts and deposition of collagen fibrils (23). These animal studies provide evidence that IL-6 can be closely associated with lymphocytosis and minimal fibrosis as observed in certain kinds of human lung diseases.

Among subgroups of IIP, NSIP is distinguished from UIP by a temporal uniformity of interstitial inflammation and/or fibrosis on histology (1, 24), and the different findings on radiology (25, 26). NSIP has a relative lymphocytosis with a predominance of a suppressor subset of T lymphocytes in BAL cells when compared with that of UIP (3). One of the distinctive clinical features of NSIP is a better prognosis than that of UIP (27, 28). These findings imply that NSIP has a different pathogenesis from UIP. In this study, we measured IL-6 concentrations and cellular components in BAL fluid from patients with NSIP and UIP to evaluate the relation of IL-6 with the lymphocytosis in IIP as evidenced by the IL-6 gene-transferred animal studies. In addition, the sources of IL-6 synthesis were analyzed in the lung biopsy specimens of patients with NSIP and UIP.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Populations

Study subjects included 7 patients with NSIP, 16 patients with UIP, and 45 normal control subjects. We had 51 patients with IIP from 1984 to 1995 in our hospital (29). Among them, 26 UIP and 8 NSIP were confirmed by open lung or video-guided thoracoscopic lung biopsies. We successfully retrieved BAL fluid in 16 patients with UIP and 7 patients with NSIP, who were enrolled in this study. The institutional review board for human studies approved the protocol and informed written consent was obtained from all subjects before the BAL procedure. After conducting a clinical evaluation including pulmonary function tests, we performed BAL and open lung or video-guided thoracoscopic biopsies in the area mostly involved showing a ground glass appearance on chest PA and computed tomogram in the patients with IIP. We avoided the area of severe fibrosis such as macroscopic honeycombing. BAL was performed 3 d before lung biopsy in all patients with IIP without any immunosuppressive therapies given. In normal subjects, BAL was performed in the right middle lobe. For proof of histology, three local pathologists and one outside pathologist reviewed the slides, being informed only of the age and sex of the patients and the presence of unilateral or bilateral pulmonary disease by the chest PA and the computed tomogram. Only when both the local and outside pathologists agreed on the diagnosis of NSIP and UIP were the patients enrolled in the study.

The pathological criteria were as follows (3): UIP is characterized by patchy scarring of the lung parenchyma with intervening normal or nearly normal alveoli. The most fibrotic zones show honeycombing with complete destruction of the architecture and in these regions, inflammation, mucostasis, and metaplastic epithelial changes may all be prominent. The subpleural and paraseptal distribution, the patchy character, and the temporal heterogeneity are the most helpful features in establishing the diagnosis of UIP. Uniform involvement of lung parenchyma, marked chronic inflammation in the interstitium, and prominent pericentral scarring are all features that negate a diagnosis of UIP. NSIP represents the heterogeneous group of interstitial pneumonia that histologically does not fit other categories. However, it was easy to differentiate NSIP from either lymphocytic interstitial pneumonia (LIP) or desquamative interstitial pneumonia (DIP), by identifying an absence of desquamation of cells into alveolar space, as in DIP, or an absence of a variable spectrum of lymphoid cells, as in LIP. The key histopathological feature of NSIP in comparison with UIP is an absence of temporal heterogeneity of the fibrotic lesions (24). The seven patients with NSIP were subcategorized into three groups according to the classification system of Katzenstein and Fiorelli (24). Group I was characterized by a cellular interstitial pneumonia with little fibrosis. Group II was characterized by a cellular interstitial pneumonia with significant fibrosis. Group III was characterized by architectural derangement due to dense interstitial fibrosis by collagen deposition and a lesser degree of inflammation. None of the patients with NSIP and UIP had any evidence of the underlying collagen vascular diseases either clinically or by laboratory analysis.

Bronchoalveolar Lavage Procedure and Bioassay of Interleukin-6

The study subjects were pretreated with atropine (0.5 mg) and diazepam (10 mg) immediately before the BAL procedure. Instillation of 50 ml of normal saline into bronchial trees four times and gentle suction with a negative pressure below 50 mm Hg were done with a fiberoptic bronchoscope (Olympus B2-10; Olympus Optical Co., Tokyo, Japan) (30). The total volume of recovered fluids was measured, then supernatant was separated from cell pellets by centrifugation at 500 × g for 5 min. The supernatants were frozen and stored at 80° C until IL-6 and protein assay. A differential count of 500 cells was done on slides prepared by cytocentrifuge and Diff-Quik stain (Scientific Products, Gibbstowne, NJ). No complication was experienced during the BAL procedures.

IL-6 in BAL fluid was measured as previously described (31) with a minor modification. A murine B9 hybridoma cell line was grown in Dulbecco's modified Eagle medium (DMEM) containing 2-mercaptoethanol (5 × 10⁵ M), garamycin (10 µg/ml), and 5% fetal calf serum (FCS) (B9 medium) and recombinant human IL-6 (Genzyme, MA). Cells were harvested and washed twice in B9 medium. Then 5,000 cells in 100 µl of B9 medium were cultured with 100 µl of standard IL-6 solutions or BAL fluids in a flat-bottomed 96-well plate. In each plate, a titration curve of standard IL-6 was included. BAL fluids were also prepared in six two-fold dilutions with 100 µl of B9 medium. The plates were incubated in a humidified, 5% CO₂ incubator for 72 h at 37° C. Proliferation of B9 cells was measured by 3-dimethylthiazole diphenyltetrazolium bromide (MTT) colorimetric assay (32). Optical density was read with an ELISA reader (Molecular Devices, CA). A dual-wavelength setting of 570 nm and 630 nm was used to subtract background noise. The experiment was performed in triplicate. The activity of IL-6 resulting in half the maximal MTT incorporation of the standard IL-6 preparation was defined as one unit. The detection limit was approximately 0.05 U/ml. A value below this limit was regarded as 0 U/ml. The intraassay coefficient of variance was 13.2% and interassay coefficient of variance was 10.5%. Protein concentration of BAL samples was measured for standardization using a micro BCA protein assay kit (Pierce, Rockford, IL).

Immunohistochemical Stain

Immediately after lung biopsy specimens were obtained from the patients with NSIP (n = 7), UIP (n = 7), and benign mediastinal tumors (n = 3), tissues were embedded in OCT medium (Tissue Tek, Miles, IN). Cryostat sections of 5 µm thickness were dried on poly-L-lysine-treated slides for 24 h, then fixed with cold acetone for 15 min and immersed in 15% sucrose for 30 min. The sections were hydrated with Tris-buffered saline (50 mM, pH 7.6, TBS) for 10 min. Endogenous biotin was blocked with 0.1% avidin and 0.01% biotin in TBS (Biotin Blocking System, DAKO X590, CA) for 10 min. The slides were treated with 3% rabbit serum for 15 min to reduce nonspecific stain, then incubated overnight at 4° C with a mouse monoclonal antibody (Mab) for human IL-6 (10 µg/ml; Genzyme), epithelium (5 dilution, DAKO J1713), and macrophges (25 dilution; Serotec, MCA 519, Oxford, UK). A mouse immunoglobulin G (IgG₁) myeloma protein (MOPC-21; Sigma, St. Louis, MO) was used as a negative control. After the sections were rinsed five times in TBS for 5 min, biotin-conjugated rabbit anti-mouse immunoglobulin (200 dilution, DAKO E413) and stravidine-alkaline phosphatase (200 dilution, DAKO D0396) were applied sequentially for 30 min at room temperature. After the sections were treated with 5 mM levamisol (Sigma) in TBS for 5 min, color was developed with Fast-red/naphthol phosphate (DAKO FAST RED, K597) in the presence of 5 mM levamisol. The sections were counterstained with methyl green or hematoxylin. Lipopolysaccharide (10 µg/ml)-stimulated peripheral blood monocytes were used as a positive control cells for IL-6 stain.

Statistics

An SPSS/PC⁺ program (Chicago, IL) was used. The differences between groups were compared by the nonparametric Kruskal-Wallis test for continuous data, and if found significant, the Mann-Whitney U test was applied to compare any two groups. The relationships of IL-6 concentrations with the percentages of cellular components of BAL cells were analyzed with Spearman's rank correlation. The 95 percentile of IL-6 level in normal control subjects was calculated from grouped data. The difference was considered significant when the p value was less than 0.05. Data were expressed as mean ± SEM.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Clinical Profiles of Study Subjects and Cellular Proportions and IL-6 Concentrations in BAL Fluid

Clinical profiles of the study subjects are summarized in Table 1. When age and smoking history were matched between groups, females were predominant in the NSIP group when compared with the UIP group and normal control subjects. Both patients with NSIP and UIP had a significant reduction in FVC, total lung capacity (TLC), and diffusing capacity of carbon monoxide (DL_CO) when compared with normal control subjects (p < 0.01, respectively). The cellular profile of BAL fluid is summarized in Table 2. Recovery rate of BAL fluid was lower in patients with UIP than that in normal control subjects (p < 0.05), but was not different between patients with UIP and NSIP. Cell number retrieved in BAL fluid was significantly increased in both patients with NSIP and UIP compared with that of normal control subjects (p < 0.01). The numbers of lymphocytes, neutrophils, and eosinophils were also significantly higher in patients with NSIP and UIP than in normal control subjects (p < 0.01, respectively). Patients with NSIP had a significantly greater number of lymphocytes than patients with UIP (p < 0.01). There was, however, no difference in number of neutrophils, eosinophils, and macrophages between patients with NSIP and UIP.

IL-6 activity was detected in the BAL fluid of all patients with NSIP, and 12 of 16 patients with UIP and 17 of 45 control subjects. Mean IL-6 level in patients with NSIP was higher than in patients with UIP (252.5 ± 65.0 U/ml versus 26.8 ± 11.8 U/ml, p < 0.01), which was also significantly higher than that of normal control subjects (2.1 ± 0.9 U/ml, p < 0.01) (Figure 1). The specificity of bioassay using B9 cell proliferation for IL-6 was checked by a blocking study using an anti-human IL-6 monoclonal antibody (B-E8; Bender MedSystems, Vienna, Austria). Four BAL fluid with 40.0 U, 36.5 U, 7.3 U, and 2.2 U of IL-6 activity were mixed with various concentrations of the antibody (10 pg to 100 ng/ml), then incubated at 37° C for 1 h. The proliferation of B9 cells by the untreated BAL samples was compared with that by the antibody-pretreated respective ones. The IL-6 activity of the BAL fluids was inhibited by treatment with the antibody in a dose-dependent manner (Figure 2). The pretreatment with 100 ng/ml of the antibody reduced the IL-6 activity of the BAL samples by 91.3 ± 0.7%.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Concentrations of IL-6 in BAL fluid obtained from patients with nonspecific interstitial pneumonia/fibrosis (NSIP, n = 7), usual interstitial pneumonia (UIP, n = 16), and normal control subjects (n = 45). The crossbars represent the mean values. The significant differences are illustrated. Statistical analysis was done by Mann-Whitney U test.

View larger version (23K): [in this window] [in a new window]   Figure 2.   A blocking study to check the specificity of a bioassay with B9 cell proliferation for IL-6 measurement in BAL fluid. Pretreatment with a neutralizing anti-human IL-6 monoclonal antibody inhibits the IL-6 activity of four BAL fluids (40.0 U, 36.5 U, 7.3 U, and 2.2 U/ml, respectively) in a dose-dependent manner.

All of the patients with NSIP had IL-6 levels greater than 10.1 U/ml, which was the 95 percentile of normal control subjects. Of 16 patients with UIP 7 had IL-6 levels greater than 10.1 U/ml (high-UIP group) and the other nine patients had levels that were lower (low-UIP group) (Figure 1). The high-UIP group had a significantly greater number of lymphocytes when compared with those of the low-UIP group (p < 0.05), whereas neutrophils, eosinophils and macrophages were not different in number between the high- and low-UIP group (Table 2). The IL-6 level of BAL samples was standardized with the respective protein concentration. The mean protein level of BAL fluid was 481.8 ± 46.2 µg/ml in patients with NSIP, 343.8 ± 53.3 µg/ml in patients with UIP, and 138.5 ± 10.8 µg/ml in normal control subjects, respectively. The mean ratio of IL-6/protein of BAL fluid was 489.0 ± 114.6 U/mg in patients with NSIP, 79.4 ± 32.8 U/mg in patients with UIP, and 15.1 ± 5.5 U/mg, respectively. The ratio of IL-6/protein of BAL fluid exhibited the same pattern of difference when compared among patients with NSIP, UIP, and normal control subjects as the levels of IL-6 did (Table 2). The mean level of IL-6 in BAL fluid was not different between nonsmokers and smokers in the patients with UIP (25.6 ± 16.6 U/ml versus 28.4 ± 17.8 U/ml), and between those in the normal control subjects (2.0 ± 1.3 U/ml versus 2.2 ± 0.8 U/ml). In patients with NSIP, IL-6 levels significantly correlated with the cell number retrieved in BAL fluid (r = 0.89; p < 0.01) and with the number of lymphocytes (r = 0.93, p < 0.01) (Figure 3), but not with those of the other inflammatory cells in BAL fluid. When patients with NSIP and UIP were included together, IL-6 levels also correlated with the cell number retrieved in BAL fluid (n = 23, r = 0.44; p < 0.05) and with the number of lymphocytes (n = 23, r = 0.81; p < 0.001). However, IL-6 levels did not correlate with the number of BAL lymphocytes when observed in patients with UIP only (n = 16, r = 0.08, p = 0.758). In addition, there was no correlation between the ages and the levels of IL-6 in patients with NSIP (r = 0.16, p = 0.46), or between patients with UIP (r = 0.30, p = 0.25) and between normal control subjects (r = 0.14, p = 0.36).

View larger version (9K): [in this window] [in a new window]   Figure 3.   The relationship between IL-6 concentrations and numbers of cells retrieved (left panel ) and those of lymphocytes (right panel ) in BAL fluid from patients with NSIP. Statistical analysis was done by Spearman's rank test.

Localization of IL-6 by Immunohistochemical Stains

Examination of hematoxylin-eosin-stained sections of biopsy specimens from the seven patients with NSIP showed a variable degree of interstitial inflammation with or without fibrosis, which was uniform within each case. Areas of honeycombing were not found in all cases of NSIP. One case was categorized as group I, four cases as group II, and two cases as group III according to the classification by Katzenstein and Fiorelli (24). Areas of honeycombing were present in all of seven patients with UIP. Lung tissues obtained from three patients with mediastinal benign tumors showed normal findings by light microscopic examination. IL-6 was detected in BAL fluid of four of the seven patients with UIP. The epithelial cells of the small airway were strongly stained for IL-6 in six of seven patients with NSIP (Figure 4B) and in five of seven patients with UIP, whereas they were weakly stained in one of three normal lung tissues (Figure 4C). Negative control experiments using mouse myeloma IgG showed no stain in all study specimens (Figure 4A). Alveolar mononuclear cells were focally stained for IL-6 in five of seven patients with NSIP (Figure 4D) and three of seven patients with UIP, but not in normal lung tissues (Figure 4E). IL-6 was not detected on the endothelium of blood vessels in any subject.

View larger version (158K): [in this window] [in a new window]   View larger version (165K): [in this window] [in a new window]   View larger version (129K): [in this window] [in a new window]   View larger version (137K): [in this window] [in a new window]   View larger version (131K): [in this window] [in a new window]   Figure 4.   Micrographs of lung sections from a subject with nonspecific interstitial pneumonia/fibrosis. Monoclonal antibodies for human IL-6, biotin-conjugated rabbit anti-mouse immunoglobulin, and stravidin-alkaline phosphatase were applied in sequence. Then Fast-red/naphthol phosphate and methyl green were used for color development and counterstain, respectively. Stain for IL-6 protein (red ) is shown on bronchial epithelium of NSIP (B). Negative control experiment using mouse myeloma IgG antibody (A) and stain for IL-6 protein on normal lung (C) showed no stain, respectively. Stain for IL-6 protein is shown on alveolar macrophages of NSIP (D), but not on those of normal lung (E). Bar = 200 µm for A, B, and C and 50 µm for D and E.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We confirmed that the patients with NSIP have a relative hypercellularity with lymphocytosis in their BAL fluid. In our study, the cellularity of lung lesions was estimated by a numerical analysis of inflammatory cells in BAL fluid. BAL has been proved to be a useful method to permit sampling of cells from alveolar areas. The cellular profiles of BAL fluid accurately reflect the cellularity of the peripheral parts of the lung in diffuse interstitial lung diseases (33, 34). The proportions of various inflammatory and immune cells in BAL fluid are correlated with those cell populations that are isolated from the respective lung biopsies of idiopathic pulmonary fibrosis and sarcoidosis (33). Our observation of BAL lymphocytosis in the patients with NSIP is consistent with the result of Nagai and coworkers (3), and can be explained by the pathological finding that lymphocytes and plasma cells are the main cells infiltrating in alveolar septa and around bronchioles in most cases of NSIP (24).

We observed that IL-6 levels and those standardized with the respective protein concentrations increased in the BAL fluid of both patients with NSIP and some patients with of UIP when compared with those of normal control subjects. These observations are in agreement with the study of Lesur and colleagues (15), in which standardized concentrations of IL-6 with epithelial lining fluid volumes were elevated in BAL fluid from patients with idiopathic pulmonary fibrosis. Our additional result was that the levels of IL-6 were well correlated with the number of lymphocytes in BAL fluid when analyzed in patients with NSIP alone as well as when done in patients with NSIP and UIP as a group. There are two implications of these results: first, IL-6 is a functional mediator to accumulate lymphocytes into the lung lesions; second, the infiltrating lymphocytes synthesize IL-6 in situ. A direct effect of IL-6 on lymphocyte infiltration in the human lung can be conjectured from the rat experiments using a local overexpression of IL-6 and IL-6R genes in the lung. Intratracheal administration of rat IL-6 cDNA induces profound bronchiolar lymphocytic hyperplasia and massive lymphocytic infiltration in lung parenchyma with a concurrent lymphocytosis in BAL fluid (22). Furthermore, alveolar macrophages and epithelial cells, which are expressed with IL-6 and IL-6R genes, also induce the histological changes of lymphocytic interstitial pneumonia without a proliferation of fibroblast in rat (23), as observed in some patients with NSIP. These results can support our result of a close relation between the levels of IL-6 and the increased number of lymphocytes in BAL fluid of patients with NSIP.

The exact role of IL-6 in IIP is still unknown. Several studies have presented evidence that IL-6 shares antiinflammatory properties in lung. In a mouse model of hypersensitivity pneumonitis by the instillation of thermophilic actinomycete, direct intratracheal infusion of antibody for IL-6 brought about a significant increase in fibrosis whereas fibrosis was reduced by treatment with IL-6 (19). In a human study (16), IL-6 concentrations were elevated in BAL fluid of patients with pigeon breeders' lung, and when IL-6 concentrations were standardized with urea or albumin in BAL fluid, asymptomatic pigeon breeders had a significantly higher concentration of IL-6 than symptomatic pigeon breeders. These results imply that the effect of IL-6 is to inhibit the progress of the inflammatory reaction in hypersensitivity pneumonitis. One possible mechanism of IL-6 to down-regulate the inflammatory responses in lung diseases is to prevent the macrophages from producing proinflammatory cytokines such as interleukin-1 (IL-1) and tumor necrosis factor (TNF) (36).

It is still unknown whether lung lymphocytes of IIP produce IL-6 or not. Wallace and colleagues (37) observed that a T helper cell type 2 (Th2-like) pattern of immune response predominated in the pulmonary interstitium of patients with IPF. They demonstrated that IL-4 and IL-5 mRNA and proteins expressed predominantly over interferon-gamma (IFN-) in the lung tissues of patients with IIP. They did not observe IL-6 expression, but it is possible that IL-6 can be expressed in the pulmonary interstitium of patients with IIP because Th2-like lymphocytes produce IL-6 in humans. Our study, however, showed no expression of IL-6 protein in the sites other than in epithelial cells and macrophages. This finding can exclude the possibility of lymphocytes as a source of IL-6 in patients with IIP, which is in agreement with the study of Emura and colleagues (38) that BAL T cells of patients with IPF did not produce IL-6 in vitro. In addition, a predominant subset of CD8-positive cells over CD4-positive cells in BAL fluid of patients with NSIP (3) suggests that the lung lymphocytes of NSIP have cytotoxic and suppresser functions. The proliferating and differentiating effect of IL-6 on cytotoxic T lymphocytes (39) might be an explanation of CD8⁺ lymphocytosis in BAL of patients with NSIP. However, further studies are needed to determine whether the epithelium and the macrophages of patients with IIP produce a greater amount of IL-6 than those of normal control subjects, and whether the lymphocytes of patients with IIP respond to IL-6 much more than those of normal control subjects do.

Taken together, levels of IL-6 were elevated in the BAL fluid of all patients with NSIP who had hypercellularity and lymphocytosis in BAL fluid, and the major sources of IL-6 were epithelial cells and macrophages. These findings strongly suggest that the lymphocytosis in the lesion of patients with NSIP is associated with the increased amount of IL-6 produced by the epithelial cells and the macrophages.