© 2003 American Thoracic Society
Pleural Fluid Transforming Growth Factor–ß1 Correlates with Pleural Fibrosis in Experimental EmpyemaSection of Pulmonary and Critical Care, Department of Medicine, Pathology and Laboratory Medicine Services, Long Beach Veterans Affairs Medical Center, Long Beach; and University of California, Irvine, Irvine, California Correspondence and requests for reprints should be addressed to Scott Sasse, M.D., Pulmonary and Critical Care Medicine, Long Beach Veterans Affairs Medical Center, 5901 East 7th Street, Long Beach, CA 90822. E-mail: scott.sasse{at}med.va.gov
Transforming growth factor–ß1 (TGF-ß1) is a growth factor that is implicated in fibrosis of many organs. The purpose of this study was to determine the sequential levels of TGF-ß1 in the pleural fluid of rabbits that had undergone empyema induction, as fibrosis of the pleural space develops. Thirty-seven rabbits underwent empyema induction. Rabbits were sacrificed on Days 1, 2, 3, 4, 5, 6, and 8. Pleural fluid and viscera pleura specimens were collected at autopsy. TGF-ß1 levels were measured in pleural fluid using a commercially available ELISA kit, and pathologic specimens were scored for evidence of fibrosis (pleural thickness and number of fibroblasts). The median levels of pleural fluid TGF-ß1 increased from 8,100 pg/ml (Days 1 and 2) to 39,600 pg/ml (Day 8). Pleural fluid TGF-ß1 levels closely correlated with microscopic pleural thickness (r = 0.7, p < 0.001) and number of fibroblasts present in the visceral pleura (r = 0.68, p < 0.001). The first increase in pleural fluid levels of TGF-ß1 (Day 3) occurred before the increase in pleural thickness (Day 4) and before the increase in number of fibroblasts (Day 4). In conclusion, pleural fluid levels of TGF-ß1 rise in experimental empyema as pleural fibrosis develops. The rise in empyemic pleural fluid TGF-ß1 levels correlates with markers of pleural space fibrosis.
Key Words: empyema • transforming growth factor–ß1 • cytokines • pleural fluid Thoracic empyema is defined as an infection of the pleural space associated with the formation of thick, purulent, pleural fluid. As an empyema progresses, fibrosis of the pleural space develops. This fibrosis of the pleural space represents the most debilitating aspect of thoracic empyema. As fibrosis and loculations of the pleural space increase, more invasive therapeutic drainage options become necessary. If fibrosis of the pleural space continues to progress, a fibrothorax may ultimately develop (1). The exact mechanisms that cause fibrosis of the pleural space in empyema are unknown. Elucidating the key factors that lead to pleural fibrosis may lead to the development of treatment strategies, which could be used to arrest the progression of pleural fibrosis. The growth factor, transforming growth factor–ß (TGF-ß), which has both fibrogenic and immunosuppressive functions, may play a major role in the development of fibrosis of the pleural space with empyema. TGF-ß1 is a 25-kD dimeric protein, which is a member of a superfamily of growth factors (cytokines) of five identified isoforms (TGF-ß1–TGF-ß5). TGF-ß isoforms ß1–ß3 are found in mammalian species and are reported to have similar biological properties (2). Isoforms ß4 and ß5 are present in only nonmammalian species (3). TGF-ß1 is initially secreted in a latent form that requires activation before it manifests biological activity (3). TGF-ß is highly conserved between species, as there is a greater than 99% identity between TGF-ß1 sequences of most mammalian species (4). The gene for TGF-ß1 is located on chromosome 19, and TGF-ß1 is the isoform most implicated in fibrosis (2). TGF-ß has been implicated in fibrotic diseases involving many organs, including the kidney (5), liver (6), lung parenchyma (7–9), skin (10), and joints (11). TGF-ß is a multifunctional cytokine that has been demonstrated to be a potent chemoattractant factor for fibroblasts, a stimulator of extracellular matrix formation, and an immunosuppressant factor for lymphocytes (3). Normal human mesothelial cells synthesize and secrete TGF-ß (12). Normal human mesothelial cells have receptors for TGF-ß, suggesting autocrine functioning of TGF-ß in the pleura (13). Human pleural mesothelial cells (when stimulated by TGF-ß in vitro) have also been shown to secrete factors, such as plasminogen activator inhibitor, which inhibit fibrinolysis (14). TGF-ß1 has been measured and found to be elevated in pleural fluid of humans with fibrosis due to tuberculous pleural effusions (15) and in humans with parapneumonic effusions and sepsis (16). Immunohistochemical localization of TGF-ß has also been detected in the fibrotic lesions of asbestosis and pleural fibrosis (17). Recently, intrapleural injections of TGF-ß2 caused a pleurodesis in rabbits and sheep with normal pleura (18, 19). With the higher intrapleural doses of TGF-ß2, pleural effusions developed (18). We chose to examine the issue of pleural fibrosis using our previously developed model of empyema in the rabbit (20). Using this model, rabbits can be killed at sequential time points as an empyema progresses. Factors in the pleural fluid can then be measured and compared with the pathological changes that develop macroscopically and microscopically. We hypothesized that TGF-ß1 levels would increase over time in experimental empyema as fibrosis of the pleural space develops. In addition, we hypothesized that the thickness of the visceral pleura microscopically and the number of fibroblasts present in the visceral pleura would correlate with the level of TGF-ß1 in empyemic pleural fluid.
This study was approved by the animal studies committee of the Long Beach, VA Medical Center before the start of the study and conforms to National Institutes of Health guidelines for animal studies. An empyema was induced in 37 2.8- to 3.0-kg rabbits using methods previously described (20). In brief, 2 ml of 108 Pasteurella multocida bacteria (in 0.5% brain heart infusion agar) were injected into the right pleural space. The position of the catheter in the pleural space, before bacterial injection, was verified using a pressure transducer. Twenty-four hours after empyema induction, a diagnostic thoracentesis was performed, in which 1.5 ml of empyemic pleural fluid were removed. This fluid was analyzed for pH and glucose, as well as cultured on a blood agar plate to verify the presence of an empyema. After the diagnostic thoracentesis, rabbits received procaine penicillin G, 200,000 U intramuscularly daily to prevent early death of the rabbits from overwhelming sepsis. Rabbits were killed on Days 1, 2, 3, 4, 5, 6, and 8 after empyema induction. At autopsy, all remaining pleural fluid was aspirated from the right pleural space. Serum samples were collected from the animals on Days 1, 3, 5, and 8. The thorax was dissected en block and placed in a 10% formalin solution for 7 days. In a subset of five rabbits, a small-volume diagnostic thoracentesis (a total volume of less than 2 ml) was performed, on an every other day basis, to obtain additional empyemic pleural fluid for measurement of TGF-ß1. These additional diagnostic thoracenteses increased the total number of pleural fluid specimens for TGF-ß1 measurement to 45. An empyema was considered present if colonies of P. multocida grew on culture from pleural fluid or the pleural fluid pH was less than 7.15, if the pleural fluid glucose level was less than 40 mg/dl, and if the empyema gross score was 2 or greater. Pleural fluid glucose was measured using a glucometer (Accucheck II, model 792; Boehringer Mannheim, Indianapolis, IN). The glucose level was arbitrarily assigned a value of 20 mg/dl if a reading of low was present on the glucometer. The pleural fluid for measurement of pH was collected in an air-tight blood gas syringe and measured using an arterial blood gas analyzer (Model 845; Bayer Diagnostics, Tarrytown, NY). Pleural fluid obtained at autopsy was centrifuged at 2,000 rpm x 10 minutes. After removal of the pellet, the supernatant was stored at -10°C before measurement of TGF-ß1 (maximum storage time was 4 days). Levels of activated TGF-ß1 were measured from the supernatant, using an ELISA kit for measurement of human TGF-ß1 (R&D Systems, Minneapolis, MN). Triplicate readings from the spectrophotometer were averaged for each pleural fluid specimen. After placing the thorax in formalin for 7 days, the pleural space was evaluated, and a 1-cm thick tissue specimen of the middle lobe of the right lung was taken from 24 animals. The tissue specimens were fixed in paraffin and stained with hematoxylin and eosin. Four lung specimens (from four separate animals) were evaluated at each time point (with the exception of Days 2 and 3, in which only two pathologic specimens were obtained). In a blinded fashion (with the thorax identified by number only), a gross empyema score was determined as follows: 0 = normal pleural space, 1 = adhesions between the visceral and parietal pleura only, 2 = minimal pleural peel with only traces of exudative material present, 3 = moderate pleural peel with presence of exudative material, and 4 = pleural peel with extensive exudative material present.
Measures of Pleural Fibrosis Microscopically, the degree of fibrosis was assessed in two ways. First, the thickness (microns) between the mesothelial cell layer and the basal layer of the visceral pleura (corresponding to the internal elastic lamina) was measured using a Nikon Eclipse E400 photomicroscope with a calibrated micrometer eyepiece. Five randomly selected visceral pleura sites were chosen from each pathologic specimen (Figure 1) . The mean thickness of the five sites was then reported as the microscopic pleural thickness score.
Second, photomicrographs at x400 were taken of five sites of the visceral pleural (adjacent to the site where pleural thickness measurements had been obtained) for each pathologic specimen to determine a fibroblast score. Each photomicrograph encompassed an area of 0.07 mm2. The number of fibroblasts was then counted on each photomicrograph. Fibroblasts were counted from photomicrographs of hematoxylin and eosin–stained specimens after comparing the staining characteristics of fibroblasts stained with Masson's trichrome stain from a selected subset of pathologic specimens. Fibroblasts were chosen based on these cell characteristics: vesicular, round to ovoid nucleus, 10–12 microns in diameter with small karyocenters and stellate cytoplasmic processes, sometimes multiple, up to 12 microns in length. Care was taken to exclude mesothelial cells and endothelial cells from the scoring of fibroblasts (Figure 2) . In addition, a subset of specimens was also stained using immunohistochemistry for smooth muscle actin by a commercial laboratory (Impath, Los Angeles, CA) to verify further that fibroblasts were being stained and counted (Figure 3) . The presence of smooth muscle actin in cells with morphology as identified previously here has previously been reported to be specific for differentiating fibroblasts and myofibroblasts (21). Immunohistochemistry staining of visceral pleura specimens was compared with positive control subjects. The fibroblast score was then reported as the mean of five fibroblast counts for each pathologic specimen.
TGF-ß1 Immunohistochemistry Paraffin sections of the lung with attached visceral pleura of two animals autopsied on Day 6 after empyema induction was prepared. Three micron sections were deparaffinized, unmasked, and subjected to immunostaining as described by Flanders and colleagues (22). The primary antibody used for immunohistochemical staining was a goat polyclonal antibody to TGF-ß1 (R&D Systems). As control subjects, paraffin sections of the left lung from Day 1 after empyema induction were used, as well as specimens using the secondary antibody alone.
Statistical Analysis
Of the initial 37 rabbits that underwent empyema induction, 4 were excluded from analysis for the following reasons. One rabbit died of overwhelming sepsis and a bowel perforation less than 24 hours after empyema induction. Another two rabbits were excluded because of a lack of growth on pleural fluid culture from the initial diagnostic thoracentesis. One other rabbit was excluded from the analysis because of the lack of pleural fluid from the initial diagnostic thoracentesis, as well as an inadequate staining of the pathological specimens from this animal. Of the remaining 33 rabbits, pleural fluid pH was measured from the initial diagnostic thoracentesis in 30 animals. In 3 of the 33 animals, a pleural fluid pH was unable to be measured from the 24-hour diagnostic thoracentesis. In these three specimens, the pleural fluid clotted in one specimen, whereas the pleural fluid was too viscous to be injected into the blood gas analyzer in the remaining two specimens. The mean pleural fluid pH ± SD was 6.99 ± 0.16 in 30 animals. The mean pleural fluid glucose from 32 rabbits on initial diagnostic thoracentesis was 24.7 mg/dl. The glucose was not measured in one animal in error. The mean ± SD gross score of all 33 animals was 3.46 ± 1.0. In 6 of the 33 rabbits, no pleural fluid was present at autopsy for measurement of TGF-ß1. Pleural fluid was obtained at autopsy for measurement of TGF-ß1 from 27 rabbits. An additional 18 pleural fluid specimens were collected from the alternating day thoracenteses, which when combined with the specimens from the 27 rabbits at autopsy, yielded a total of 45 pleural fluid specimens. Activated pleural fluid TGF-ß1 levels were measured in 45 specimens. The median pleural fluid activated TGF-ß1 levels for Days 1, 2, 3, 4, 5, 6, and 8 are shown in Figure 4 . The median TGF-ß1 level rose from 8,100 pg/ml (on Days 1 and 2) to 39,600 pg/ml on Day 8. The median TGF-ß1 levels from Days 5, 6, and 8, were significantly greater than the median levels from Day 2 (p < 0.05).
The 4- to 5-day period represents the time during which early fibrosis first occurs in this model, as evidenced by increased pleural thickness and fibroblast infiltration of the visceral pleura. When the TGF-ß1 levels from Days 5, 6, and 8 were combined and compared with the TGF-ß1 levels from combined Days 1–4, the median value was significantly greater (28,572 vs. 9,201 pg/ml, respectively) (p < 0.001). As also shown in the graph, serum levels of TGF-ß1 remained low throughout the 8-day time period, and increased variability of the pleural fluid TGF-ß1 levels occurred at Day 8.
Measures of Degree of Fibrosis Present
The median microscopic pleural thickness score was measured from pathologic specimens of the visceral pleura in 30 animals. (In three animals, pathologic specimens were not obtained because of a laboratory error.) The median microscopic pleural thickness score versus day-after-empyema induction is shown in Figure 6 . As shown in Figure 6, the microscopic thickness of the space between the mesothelial cell layer and the elastic lamina layer of the visceral pleural is unchanged over the first 3 days. On Day 4, the thickness first starts to expand. Day 4 represents the time point at which the first signs of early fibrosis of the visceral pleura begin to occur microscopically (discussed later here). The median value from Day 8 was significantly greater than the individual values from Days 1–5, and the median value from Day 6 was significantly greater than the individual values from Days 1–4 (p < 0.05). The median value from Day 5 was significantly greater than the values from Days 1–3 (p < 0.05).
The median fibroblast score versus day-after-empyema induction is shown in Figure 7 . As shown in Figure 7, there is a marked influx of fibroblasts into the expanded visceral pleura. The number of fibroblasts remains at a low baseline level on Days 1–3, and first increases on Days 4 and 5. The median fibroblast score from Day 8 is significantly greater than the median values from Days 1–5, whereas the median fibroblast score from Day 6 is significantly greater than the scores from Days 1–3 (p < 0.05). The median values from Days 4 or 5 are significantly greater than the median values from Day 1 (p < 0.05).
Correlation with TGF-ß1 Levels in Pleural Fluid The graph demonstrating the correlation between the microscopic pleural thickness scores (microns) and the pleural fluid level of TGF- ß 1 (pg/ml) is shown in Figure 8 . A significant correlation was found between the level of pleural fluid TGF-ß1 and the microscopic pleural thickness (r = 0.7, p < 0.001).
The numbers on the graph indicate the day on which the rabbit was sacrificed, for collection of pleural fluid and pathological specimens. When day-after-empyema induction is added to TGF-ß1 as a second independent variable, multiple regression analysis demonstrates a slight improvement in the correlation (r = 0.8, p < 0.05 for both TGF-ß1 and day). A significant correlation was also found between the pleural fluid level of TGF-ß1 and the fibroblast score, as shown in Figure 9 (r = 0.68 and p < 0.001). Using multiple regression analysis, where day-after-empyema induction and TGF-ß1 are independent variables, led to a slight improvement in the correlation (r = 0.77, p < 0.05 for both TGF-ß1 and day). The levels of pleural fluid TGF-ß1 also correlated with the macroscopic pleural peel score (r = 0.621, p < 0.001).
TGF-ß1 Immunohistochemistry The results of initial TGF-ß immunohistochemistry staining of pathological specimens obtained on Day 6 after empyema induction revealed that anti–TGF-ß1 was localized to the viable inflammatory cells in the extrapleural, exudative, purulent material (adjacent to the visceral pleural mesothelial layer). In addition, staining was localized to macrophages within the pulmonary alveolar spaces. It should be noted, however, that background staining of the specimens was elevated, making it difficult to conclude with certainty that these are the only cell types that stain for TGF-ß1. The staining was compared with negative control specimens. Relative to the staining of the inflammatory cells in the exudative material, very little staining was present along the mesothelial cells of the visceral pleural.
In this study, we have demonstrated that pleural fluid levels of TGF-ß1 rise with time, as fibrosis of the pleura develops, in this rabbit model of empyema. The increasing levels of pleural fluid TGF-ß1 correlate closely with the increasing microscopic thickness of the visceral pleura. The pleural fluid TGF-ß1 levels also correlated with the number of fibroblasts present in the visceral pleura tissue. It has previously been shown that TGF-ß levels are elevated after fibrosis develops in the liver, lung parenchyma, skin, and kidney (2). TGF-ß is important for tissue remodeling and injury repair, as TGF-ß stimulates extracellular matrix production and fibroblast migration and inhibits further immune responses by lymphocytes. It has been hypothesized that sustained elevations of TGF-ß are required to cause fibrosis (2). In our experimental model of empyema, sustained elevations of TGF-ß1 in pleural fluid are present with the subsequent development of pleural fibrosis. We measured the levels of pleural fluid TGF-ß1 in the rabbit using an immunoassay kit for activated human TGF-ß1. We believe that the human assay for TGF-ß1 is accurate for the measurement of rabbit TGF-ß1 because of the very high degree of homology of amino acid sequences in TGF-ß1 from different mammalian species (4). To our knowledge, TGF-ß1 has not been measured previously in empyemic pleural fluid from animals or humans. In this study, levels of pleural fluid TGF-ß1 rose to very high levels compared with the pleural fluid TGF-ß1 levels measured in prior human pleural fluid studies. The median value in the pleural fluid at Day 8 in this study was greater than 38,000 pg/ml. In previous human studies, peak values were as high as 17,000 pg/ml in effusions after coronary artery bypass surgery, 9,500 pg/ml in parapneumonic effusions (16), and 43 pM in tuberculous pleural effusions (15). We speculate that the markedly elevated levels found in this study are due to a strong connective tissue response (as part of the repair process in the pleural space) that occurs as an empyema forms, with the subsequent development of fibrosis. The autocrine properties of TGF-ß1 (13) may also play a role in the development of the markedly elevated levels of pleural fluid TGF-ß1. Intrapleural injections of TGF-ß2 have now been shown to cause a pleurodesis in rabbits and sheep with normal pleura (18, 19). Pleural effusions also developed in the animals when intrapleural TGF-ß2 was administered in the highest doses. Our findings are consistent with these results, as pleural effusions develop with elevated pleural fluid TGF-ß levels using this model of empyema. It is possible that an additional cytokine (or cytokines), such as vascular endothelial growth factor (23, 24), is responsible for the pleural fluid accumulation in empyema. We also hypothesize that other cytokines play a role in the early stages of empyema formation (Days 1–3 in our model), as the TGF-ß1 levels remained low during these days, yet empyema formation progressed rapidly. These early additional cytokines may be needed as precursors for the subsequent rapid increase in TGF-ß1. The temporal relationship between the time point at which TGF-ß1 first began to rise and the time point at which our markers of fibrosis first began to increase also suggests an important role for TGF-ß1 in fibrosis of the pleural space. Pleural fluid levels of TGF-ß1 began to rise on Day 3, whereas our markers of fibrosis (microscopic pleural thickness and number of fibroblasts) began to increase on Day 4. In addition, in our previous studies, we reported a marked decrease in pleural fluid leukocyte count from Day 1 (40,000 leukocytes per mm3) to Day 4 (8,000 leukocytes per mm3) (20). We speculate that the immunosuppressive effects of TGF-ß1 on leukocytes may have influenced the rapid decline in pleural fluid leukocyte count. Although other models of fibrosis have been developed in the kidney, liver, and skin, this model is unique in that the pleural space (immediately adjacent to the site of fibrosis) accumulates pleural fluid, which is available for sampling. Currently, it is uncertain exactly which cells produce the majority of the TGF-ß1 that accumulates in the pleural space, although many cell types, including mesothelial cells (12), macrophages (25), lymphocytes, fibroblasts, endothelial cells (26) and bronchial epithelial cells (27), have been reported to secrete TGF-ß when appropriately stimulated. From our low levels of TGF-ß1 in the serum and our preliminary immunohistochemistry studies, it appears that TGF-ß1 either does not pass into the bloodstream from the pleural space or, more likely, TGF-ß1 is rapidly cleared from the lung parenchyma and bloodstream, given its reported half-life of less than 3 minutes (28). Future studies are needed to determine whether pleural fibrosis can be inhibited in empyema. If so, drainage of the pleural space could be accomplished with less invasive techniques. Possible methods for inhibition of TGF-ß1 include TGF-ß receptor blockers, antisense oligonucleotides, antisera to TGF-ß1, or administration of the proteoglycan agents decorin or betaglycan (2). In conclusion, we have shown that pleural fluid levels of TGF-ß1 rise significantly over time in experimental empyema, as pleural fibrosis develops. This rise in empyemic pleural fluid TGF-ß1 levels correlates with markers of the observed pleural fibrosis.
The authors thank these individuals for their technical assistance with this project: Richard Y. Kwok, Vu Nguyen, Ann Jones, Walter Thill, Phillip Neal, and Farhad Mazdiznian, M.D.
Supported by the Southern California Institute for Research and Education. Conflict of Interest Statement: S.A.S. has no declared conflict of interest; M.R.J. has no declared conflict of interest; G.D.K. has no declared conflict of interest. Received in original form February 19, 2002; accepted in final form July 2, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
