Published ahead of print on May 11, 2007, doi:10.1165/rcmb.2007-0057OC
© 2007 American Thoracic Society DOI: 10.1165/rcmb.2007-0057OC Cyclooxygenase-2 Deficiency Exacerbates Bleomycin-Induced Lung Dysfunction but Not FibrosisDivision of Intramural Research, National Institute of Environmental Health Sciences, National Institutes of Health; and CIIT Centers for Health Research, Research Triangle Park, North Carolina Correspondence and requests for reprints should be addressed to Darryl C. Zeldin, M.D., NIH/NIEHS, 111 T.W. Alexander Drive, Building 101, Room D236, Research Triangle Park, NC 27709. E-mail: zeldin{at}niehs.nih.gov
Cyclooxygenase (COX)-derived eicosanoids have been implicated in the pathogenesis of pulmonary fibrosis. Uncertainty regarding the influence of COX-2 on experimental pulmonary fibrosis prompted us to clarify the fibrotic and functional effects of intratracheal bleomycin administration in mice genetically deficient in COX-2. Further, the effects of airway-specific COX-1 overexpression on fibrotic and functional outcomes in wild-type and COX-2 knockout mice were assessed. Equivalent increases in airway cell influx, lung collagen content, and histopathologic evidence of fibrosis were observed in wild-type and COX-2 knockout mice 21 d after bleomycin treatment, suggesting that COX-2 deficiency did not alter the extent or severity of fibrosis in this model. However, bleomycin-induced alterations in respiratory mechanics were more severe in COX-2 knockout mice than in wild-type mice, as illustrated by a greater decrease in static compliance compared with genotype-matched, saline-treated control mice (26 ± 3% versus 11 ± 4% decreases for COX-2 knockout and wild-type mice, respectively; P < 0.05). The influence of COX-1 overexpression in airway Clara cells was also examined. Whereas the fibrotic effects of bleomycin were not altered in wild-type or COX-2 knockout mice overexpressing COX-1, the exaggerated lung function decrement in bleomycin-treated COX-2 knockout mice was prevented by COX-1 overexpression and coincided with decreased airway cysteinyl leukotriene levels. Collectively, these data suggest an important regulatory role for COX-2 in the maintenance of lung function in the setting of lung fibrosis, but not in the progression of the fibrotic process per se.
Key Words: cyclooxygenase • fibrosis • respiratory mechanics • prostaglandin • transgenic
Pulmonary fibrosis is a debilitating condition for which patient prognosis is extremely poor. The pathogenesis of pulmonary fibrosis has been ascribed to numerous mechanisms (1), yet the translation of the experimental study of such mechanisms to effective clinical therapies remains elusive. Considerable evidence suggests an imbalance between pro- and antifibrotic mediators and pathways as an underlying cause of fibrosis. Communication between alveolar epithelial cells and resident fibroblasts is important to the maintenance of normal lung architecture and gas exchange; for reasons not completely understood, certain conditions of epithelial cell injury result in defects in fibroblast proliferation and extracellular matrix deposition, leading to increased collagen production and fibrosis. The cyclooxygenase (COX) enzymes, COX-1 and COX-2, are expressed in the lung and have been shown to be important regulators of lung function and disease. With regard to pulmonary fibrosis, considerable attention has been afforded to COX-2 as a potentially critical enzyme whose prostaglandin (PG) products may be involved in limiting the fibrotic process (2). Indeed, fibroblasts isolated from fibrotic human lung express less COX-2 and produce less of the antifibrotic PGE2 than nonfibrotic fibroblasts (3, 4), and mice genetically deficient in COX-2 have been shown to be susceptible to pulmonary fibrosis induced by vanadium pentoxide (5). However, data regarding the influence of COX-2 deficiency on the fibrotic process in the more commonly used bleomcyin-induced pulmonary fibrosis model are equivocal, with separate reports describing exacerbation and no change of outcomes in COX-2–deficient mice relative to wild-type mice. In one report, COX-2 knockout mice displayed enhanced fibrosis relative to wild-type mice (6), while similar fibrotic outcomes were observed in COX-2 knockout and wild-type mice in another (7). Importantly, though, the study by Lovgren and coworkers (6) documented enhanced airway dysfunction that coincided with increased lung pathology in COX-2 knockout mice treated with bleomycin, thus introducing a critical physiological component to this experimental murine model that, as a result, more closely resembles human pulmonary fibrosis. To clarify the importance of COX-2 in the etiology of pulmonary fibrosis and associated lung mechanical alterations, the present study examined the fibrotic and functional responses of COX-2 knockout and wild-type mice to intratracheal bleomycin administration. Moreover, the influence of airway-specific overexpression of COX-1 on bleomycin-induced fibrotic and functional outcomes was assessed, as we have demonstrated that this genetic approach to manipulating the airway eicosanoid profile improves airway function in COX-2 knockout mice in a model of allergic airway disease (8). Results presented herein indicate that bleomycin-induced pulmonary fibrosis is not enhanced in the setting of COX-2 deficiency, but that respiratory mechanics in the fibrotic lung are adversely affected by the absence of this enzyme. In addition, COX-1 overexpression is shown to rescue the functional but not the fibrotic phenotype of bleomycin-treated COX-2 knockout mice, underscoring the complex interplay among eicosanoids in the regulation of lung function and disease in this model.
Animals and Treatments All studies were conducted in accordance with principles and procedures outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and were approved by the Animal Care and Use Committee of the National Institute of Environmental Health Sciences. Female mice from the F2 generation of a breeding strategy described in detail elsewhere (8) were used for study. Briefly, COX-2 null males on a hybrid C57BL/6-SvEv background (Taconic Farms, Germantown, NY) were bred to COX-1 transgenic (COX-1 Tr) females on a pure C57BL/6 background (human COX-1 expressed in airway Clara cells; from an in-house breeding colony at NIEHS). Approximately 50% of the F1 offspring were heterozygous at the COX-2 locus and also positive for the COX-1 transgene. Female COX-2 heterozygous/COX-1 Tr mice were then crossed to male COX-2 null/COX-1 WT (nontransgenic) mice to generate F2 mice for study. Approximately 25% of the F2 mice were COX-2–/– COX-1 Tr (COX-2 null/COX-1 Tr) and  25% were COX-2–/– COX-1 WT (COX-2 null/COX-1 WT). A similar scheme was used to generate COX-2+/+ COX-1 Tr and COX-2+/+ COX-1 WT mice. Female littermates were used in all studies because the COX-1 Tr mice are on a pure C57BL/6 background, whereas the COX-2 null mice are on a hybrid C57BL/6-SvEv background that has been intercrossed for > 20 generations. The F2 mice used for study were therefore murine COX-2 null or COX-2 sufficient (COX-2 WT), and similarly either human COX-1 transgene–positive (COX-1 Tr) or –negative (COX-1 WT). Thus, mice were of the following four genotypes: COX-2 WT/COX-1 WT, COX-2 null/COX-1 WT, COX-2 WT/COX-1 Tr, and COX-2 null/COX-1 Tr. For the purpose of clarity, these groups are referred to throughout this report as COX-2 WT, COX-2 null, COX-2 WT/COX-1 Tr, and COX-2 null/ COX-1 Tr, respectively. Female littermates between the ages of 3 and 6 mo were anesthetized with isoflurane/oxygen and were administered bleomycin sulfate (Sigma-Aldrich, St. Louis, MO) dissolved in sterile, endotoxin-free saline, or an equivalent volume of sterile saline (75 µl) as a vehicle control, via an oropharyngeal aspiration method described previously (9). Briefly, anesthetized mice were suspended by their upper incisors with a rubber band on a 60° incline board. The tongue was gently extended and the bleomycin or saline solution was pipetted into the mouth. Brief occlusion of the nose forced the animal to inhale through the mouth, thereby aspirating the solution into the respiratory tract in one or two breaths. Animals were subsequently removed from the board and observed closely until fully recovered from anesthesia. Bleomycin was administered at a dose of 1.0 mg/kg body weight. Mice were observed and weighed at regular intervals after aspiration of bleomycin or saline until assessment of lung function and tissue collection 21 d after dosing.
Assessment of Respiratory Mechanics
Tissue Collection and Sample Analysis
Lung Histopathology Fibrosis specifically associated with small- to medium-sized airways was determined by quantifying subepithelial collagen thickness around bronchioles with an internal diameter of 100–200 µm. Bronchioles that presented a closed circular or oval profile were selected from trichrome-stained sections of lung that were coded so that genotype and treatment information were unknown to the reader. Digitized photomicrographs of five randomly selected bronchioles from individual animals were obtained with a x10 objective using a Zeiss Axioskop 2 microscope (Carl Zeiss, Thornwood, NY) and the SPOT Advanced digital image analysis program (Diagnostic Instruments, Inc., Sterling Heights, MI). Airway wall measurements were taken by first centering an eight-point spoke wheel into the lumen of each bronchiole within the digitized photograph using the Freehand MX sofware program (Macromedia, San Francisco, CA). The images were then saved as TIFF files and the thickness of the collagen layer at the intersection of the airway wall with each of the eight spokes determined by a digital ruler using Adobe Photoshop 8.0 software (Adobe Systems, Inc., San Jose, CA). The average for each airway was calculated from these eight measurements, although individual intersections were excluded from analysis if they fell in a region of the airway wall that was closely adjacent to a vessel with abundant perivascular collagen. Average values for individual animals, expressed as airway wall collagen thickness in microns, were calculated from the values of the five airways analyzed.
Lung Collagen Assay
BAL Fluid Eicosanoid Analysis
Statistical Analysis
COX-2 Deficiency Does Not Alter Bleomycin-Induced Lung Fibrosis Administration of bleomycin to COX-2 WT and COX-2 null mice resulted in a decrease in body weight of 5% by 7 d after dosing, with no difference between the genotypes. Body weights of bleomycin-treated mice of both genotypes returned to pre-dosing levels by 14 d after dosing and increased to 5% above pre-dosing levels at 21 d after dosing. Saline-treated mice did not display any loss of body weight after dosing during the 21 d of study (data not shown). Relative to saline-treated controls, bleomycin-treated COX-2 WT and COX-2 null mice had increased total BAL fluid cells 21 d after dosing that were comprised primarily of macrophages/monocytes and lymphocytes (Figure 1A). Bleomycin induced a characteristic inflammatory and fibrotic response that included increased extracellular matrix deposition and areas of cell influx around airways and blood vessels that were readily apparent upon examination of histologic sections (Figure 1B). However, histopathologic quantification of the extent and degree of fibrosis did not reveal differences between bleomycin-treated COX-2 WT and COX-2 null mice (Figure 1C), nor did biochemical assessment of lung collagen content reveal differences between the genotypes (Figure 1D). These data suggest that COX-2–deficient mice were not any more or less susceptible to bleomycin-induced fibrosis than were COX-2 WT mice.
Airway-Specific Overexpression of COX-1 Does Not Alter Bleomycin-Induced Lung Fibrosis We have previously demonstrated that COX-1 overexpression in airway Clara cells does not influence inflammatory outcomes in a murine model of allergic airway inflammation (8). However, this does not preclude the possibility that alteration of airway eicosanoid levels by this approach will impact the development of bleomycin-induced pulmonary fibrosis, as these two models of lung disease are characterized by different pathogenic mechanisms and outcomes. Therefore, we studied mice harboring or lacking the human COX-1 transgene and similarly possessing or lacking the murine COX-2 gene. Similar to the COX-2 WT and COX-2 null mice described above, bleomycin-treated COX-2 WT/COX-1 Tr and COX-2 null/COX-1 Tr mice displayed a minor degree of weight loss ( 5%) 7 d after dosing and returned to pre-dosing body weights by 21 d. Saline-treated mice did not lose any weight during the 21 d of study (data not shown). Bleomycin induced an influx of inflammatory cells in the airways of COX-2 WT/COX-1 Tr and COX-2 null/COX-1 Tr mice that was similar in magnitude and composition (Figure 2A). Histopathologic changes resulting from bleomycin administration were also evident in lung sections from mice of both genotypes (Figure 2B) and did not differ quantitatively (Figure 2C). Lung collagen content was also increased after bleomcyin, but did not differ between the genotypes (Figure 2D).
Importantly, airway-specific overexpression of COX-1 did not alter bleomycin-induced inflammatory or fibrotic outcomes in either COX-2 WT or COX-2 null mice, as is illustrated by comparing the data presented in Figure 1 with that in Figure 2. For example, total lung collagen content in bleomycin-treated COX-2 WT, COX-2 null, COX-2 WT/COX-1 Tr, and COX-2 null/COX-1 Tr mice was 525 ± 12, 493 ± 30, 516 ± 22, and 536 ± 30 µg/lobe, respectively. Thus, alteration of airway eicosanoid levels via COX-1 overexpression is not likely a suitable approach to target the causative and/or progressive processes in the pathogenesis of pulmonary fibrosis.
Bleomycin-Induced Decrements in Lung Function Are Exaggerated in COX-2 Null Mice and Reversed by Airway-Specific COX-1 Overexpression
Airway overexpression of COX-1 was assessed for its effects on bleomycin-induced lung dysfunction. In COX-2 WT mice, overexpression of COX-1 did not appreciably alter the changes in Cst, E, or H resulting from bleomycin treatment. However, COX-1 overexpression prevented the lung function decline resulting from bleomycin treatment in COX-2 null mice. Specifically, the exaggerated decrease in Cst and increases in E and H observed in bleomycin-treated COX-2 null mice were absent in the presence of COX-1 overexpression (Figures 3A–3C; P < 0.05 for COX-2 null/COX-1 Tr versus COX-2 null for all three parameters).
Airway Fibrosis in Bleomycin-Treated Mice
Airway Eicosanoid Levels Airway PG and cysLT levels were also measured to determine if they may have contributed to the observed differences in lung function outcomes. Airway PG levels did not differ significantly among saline-treated groups, although mice that harbored the COX-1 transgene had approximately double the amount of PGE2 than mice that did not harbor the COX-1 transgene (Figure 5A), consistent with our previous observations in naïve animals (8). Compared with saline-treated mice, bleomycin-treated mice of all genotypes had increased airway 6-keto-PGF1  and PGE2 levels (Figure 5B). Whereas the increase in 6-keto-PGF1 was roughly equivalent among genotypes, mice that harbored the COX-1 transgene demonstrated a 3-fold increase in PGE2 levels, while those that did not harbor the COX-1 transgene did not (Figure 5B). Bleomycin-treated COX-1 Tr mice also demonstrated increased airway levels of PGF2 and PGD2 compared with littermates treated with saline (Figure 5B).
No differences in airway PG levels were observed between bleomycin-treated COX-2 WT and COX-2 null mice (Figure 5B). The lack of a difference in airway PGE2 levels between genotypes differs from the results of a recent study in which COX-2 null mice were unexpectedly found to have increased PGE2 levels compared with COX-2 WT mice after bleomycin administration (7). This discrepancy between studies may be the result of differences in the timing and methodology of airway PG assessment.
Airway cysLT levels in saline-treated mice were low (< 10 ng/ml BAL fluid) and did not differ among genotypes (Figure 5C). These levels were considerably increased in bleomycin-treated mice of all genotypes, albeit to varying degrees (Figure 5D). Whereas post-bleomycin PG levels did not differ, COX-2 null mice had increased airway cysLT levels (Figure 5D) compared with COX-2 WT mice, and this may have contributed to their exaggerated lung function decline. Airway overexpression of COX-1 resulted in an increase of airway PGF2
The main objective of this study was to clarify the effects of COX-2 deficiency on pulmonary fibrosis and associated lung functional changes. To do this, we employed an established model of fibrosis induced by administration of bleomycin and examined the biochemical and histologic indices of fibrosis and changes in lung mechanics that occurred in COX-2 null and wild-type littermate mice. Our data suggest that COX-2 deficiency did not exacerbate fibrosis in this model, but that it did result in a worsening of lung function decline as determined by invasive analysis of Cst, E, and H. Further, we demonstrated that fibrosis-associated decrements in lung function, but not fibrosis itself, could be prevented by airway COX-1 overexpression, indicating a complex interaction of eicosanoids in regulating various aspects of the lung injury in this model.
The role of COX-2 in the pathogenesis of pulmonary fibrosis has been examined in mouse models of vanadium pentoxide- and bleomycin-induced lung injury, with seemingly disparate results (3, 5–7). COX-2 null mice were shown to develop fibrosis after exposure to vanadium pentoxide, whereas COX-1 null and wild-type mice did not (5). Moreover, while an initial histologic assessment (four to six mice per genotype) suggested exacerbation of pulmonary fibrosis induced by 1 mg/kg bleomycin in COX-2 null mice (3), a more thorough analysis by the same research group later demonstrated no increase in the fibrotic outcomes in these mice relative to wild-type mice in response to this dose of bleomycin (7). A more recent study reported enhanced bleomycin-induced fibrosis in COX-2 null mice (6), although comparison of these findings with those of the previous studies is difficult due to the fact that the dose of bleomycin used in the latter study (0.05 units/mouse) was not adjusted for animal body weight and, according to our calculations, appears to have been greater in magnitude ( A decline in respiratory function is associated with poor prognosis in humans with pulmonary fibrosis and may be a more useful predictor of morbidity and mortality than traditional pathological endpoints assessed by biopsy (18, 19). Inclusion of lung function evaluation in murine models of pulmonary fibrosis is uncommon, yet this important analytical endpoint helps to more closely correlate these models to the human condition. In this regard, our observation of an exaggerated functional defect in the lungs of bleomycin-treated COX-2 null mice in the absence of biochemical or histologic evidence of enhanced fibrosis supports a critical role for this enzyme in regulating lung function in the setting of fibrosis. Decreased Cst and increased E and H were observed in bleomycin-treated mice, indicative of the characteristic increase in lung stiffness that leads to inefficient gas exchange and respiratory insufficiency in the fibrotic lung. These observations are consistent with other reports of decreased respiratory system compliance in bleomycin-treated experimental animals (6, 20, 21). The reason for exaggerated lung function decline in COX-2 null mice compared with COX-2 WT mice in the absence of a measurable difference in fibrosis may be related to events not quantifiable by traditional means of determining fibrotic endpoints such as differences in collagen maturity, cell-specific injury, effects on smooth muscle, and regional differences in collagen deposition. Differences in fibrosis-associated changes in other extracellular matrix components such as elastin may also have contributed to this phenomenon. With regard to regional differences in collagen deposition, our analysis of fibrosis associated with small- and medium-sized airways did not reveal an increased level of airway collagen in COX-2 null mice, suggesting that this possibility was not likely contributing to the more severe lung function decline in these mice. Regardless, the fact that functional alterations were more severe in COX-2 null mice than in COX-2 WT mice indicates that absence of this enzyme renders the lung prone to enhanced functional deficits that could not be predicted based on assessment of fibrosis alone, and underscores the necessity of including lung function endpoints in the overall appraisal of injury in lung fibrosis models. As we have documented an improvement of airway responsiveness via transgenic airway overexpression of COX-1 in COX-2 null mice in a model of allergic airway inflammation (8), we sought to determine if similar beneficial effects on lung function might be afforded by this approach in our fibrosis model. Whereas COX-1 overexpression was found to not influence the extent or severity of lung or airway-specific fibrosis in COX-2 WT or COX-2 null mice, functional alterations induced by bleomycin were attenuated. Specifically, decreased Cst and increased E and H were not as severe in COX-2 null mice harboring the COX-1 transgene. Beneficial effects of the COX-1 transgene were not as readily apparent in COX-2 WT mice, although the bleomycin-induced increases in E and H were alleviated somewhat. These findings suggest that airway overexpression of COX-1 was capable of rescuing the fibrosis-associated impairment of lung function in COX-2 null mice, an outcome that may have been related to effects on airway eicosanoid levels. Airway PG levels did not differ but airway cysLT levels were elevated in bleomycin-treated COX-2 null mice compared with COX-2 WT mice, coincident with their more severe impairment of lung function. Moreover, airway cysLT levels were lower and lung function decrements were less severe in bleomcyin-treated COX-2 null/COX-1 Tr mice compared with COX-2 null mice, further suggestive of a causal relationship between cysLTs and lung function decline. While their role in fibrosis-associated lung function decline has not been studied, the cysLTs have been shown to be essential to the pathogenesis of experimental lung fibrosis as mice either deficient in their production or treated with pharmacologic inhibitors of 5-lipoxygenase or cysLT1 receptor antagonists are protected from bleomycin-induced fibrosis (22–24). Further, the potential for cysLTs to impair dynamic pulmonary compliance in various species is well established (25–27) and suggests that reduction of airway cysLT levels or antagonism of cysLT receptors may improve lung function in a variety of disease states in which these pro-inflammatory eicosanoids are implicated. The most obvious of these is asthma, for which effective inhibitors of cysLT synthesis and antagonists of cysLT receptors have been developed and are currently in use. It remains to be determined whether pharmacologic inhibition of cysLT synthesis and/or receptor signaling might benefit fibrosis-associated lung function decline in the setting of clinical or experimental fibrosis. In our model of lung fibrosis, it appears as though the improvements in dynamic (measured as E and H) and static (measured as Cst) compliance afforded by airway overexpression of COX-1 in COX-2 null mice may have been mediated by the reduction of airway cysLT levels that occurred independent of changes in airway PG levels. These are interesting observations, given that the COX-1 transgene was expressed in airway Clara cells and therefore might not have been expected to influence Cst, a measurement normally associated with the mechanical properties of the distal airways and lung parenchyma. On this note, it is possible that expression in a more distal location such as alveolar type II cells (with expression driven by the surfactant protein A or C promoter) may have altered airway eicosanoid levels to a greater extent and/or in a region of the lung more relevant to the fibrotic and functional outcomes, and had an even more pronounced effect on fibrosis and associated functional alterations as a result. These possibilities remain to be tested and represent interesting avenues of future investigation. In the present study, shunting of arachidonic acid metabolism toward the 5-lipoxygenase pathway likely occurred in the lungs of bleomycin-treated COX-2 null mice, leading to the increased airway cysLT levels in these animals. It is interesting to note that this increase in cysLT levels occurred in the absence of a corresponding change in airway PG levels relative to those observed in COX-2 WT mice. These observations imply the likely involvement of transcellular biosynthetic processes in the resulting airway eicosanoid levels (28), similar to what we proposed as underlying the improved lung function resulting from COX-1 overexpression in COX-2 null mice in a model of allergic airway inflammation (8). Moreover, airway overexpression of COX-1 in COX-2 null mice resulted in decreased cysLT levels and a corresponding increase in PG levels after bleomycin treatment. These findings, together with the lung function outcomes, highlight the intricate interaction of eicosanoid-mediated effects in the regulation of lung injury and corresponding lung dysfunction. Confounding the situation is the recognition that an array of specific eicosanoid receptors expressed on a variety of cell types contributed to the outcomes and that eicosanoid levels and receptor expression patterns are likely dynamically regulated throughout the course of injury. That said, the airway eicosanoid profiles, determined at the same time point that lung function and fibrosis assessments were conducted, suggest that cysLTs promote lung dysfunction in the setting of fibrosis in COX-2 null mice.
Finally, it is prudent to discuss the results of the present study in the context of recently published data concerning the influence of COX-2 deficiency in the bleomycin-induced lung fibrosis model. Lovgren and colleagues (6) observed that absence of the prostacyclin receptor rendered mice as susceptible to bleomycin-induced lung fibrosis and dysfunction as did the absence of COX-2. Based on this data, these investigators postulated that prostacyclin is protective against fibrosis. However, prostacyclin levels in lung or BAL fluid of naïve or bleomycin-treated mice were not reported, thus precluding firm establishment of a role for prostacyclin in limiting the development of lung fibrosis and associated functional alterations. In the present study, airway prostacyclin levels (measured as the stable metabolite 6-keto-PGF1 Collectively, our data indicate that COX-2 deficiency does not influence the development of bleomycin-induced pulmonary fibrosis, but that functional impairments associated with fibrosis are exaggerated in the absence of this enzyme. Further, we show that the adverse functional effects associated with pulmonary fibrosis in COX-2 null mice are decreased by airway COX-1 overexpression and the ensuing alteration of airway eicosanoid levels. Based on our observations, we propose that targeted alteration of pulmonary eicosanoid levels may afford benefit in the treatment of fibrosis-associated lung function decline.
The authors are grateful to Sandy Ward for help with cell differential counting, to Dr. Shyamal Peddada for help with statistical analyses, and to Drs. Michael Fessler and Dianne Walters for helpful comments during preparation of this manuscript. This research was conducted in part at the National Institute of Environmental Health Sciences Inhalation Facility under contract to Alion Science and Technology.
This research was supported by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences. J.W.C. is the recipient of a Senior Research Training Fellowship from the American Lung Association of North Carolina. Originally Published in Press as DOI: 10.1165/rcmb.2007-0057OC on May 11, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form February 23, 2007 Accepted in final form April 13, 2007
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
CLINICAL RELEVANCE
