This study examines the hypothesis that the cigarette smoke-induced inflammatory process is amplified in severe emphysema and explores the association of this response with latent adenoviral infection. Lung tissue from patients with similar smoking histories and either no (n = 7), mild (n = 7), or severe emphysema (n = 7) was obtained by lung resection. Numbers of polymorphonuclear cells (PMN), macrophages, B cells, CD4, CD8 lymphocytes, and eosinophils present in tissue and airspaces and of epithelial cells expressing adenoviral E1A protein were determined using quantitative techniques. Severe emphysema was associated with an absolute increase in the total number of inflammatory cells in the lung tissue and airspaces. The computed tomography (CT) determined extent of lung destruction was related to the number of cells/m2 surface area by R2 values that ranged from 0.858 (CD8 cells) to 0.483 (B cells) in the tissue and 0.630 (CD4 cells) to 0.198 (B cells) in the airspaces. These changes were associated with a 5- to 40-fold increase in the number of alveolar epithelial cells expressing adenoviral E1A protein in mild and severe disease, respectively. We conclude that cigarette smoke-induced lung inflammation is amplified in severe emphysema and that latent expression of the adenoviral E1A protein expressed by alveolar epithelial cells influenced this amplification process.
Keywords: emphysema; lung inflammation; latent adenoviral infection; cigarette smoking
The cigarette smoking habit is the major risk factor for emphysema, but only a fraction of the smoking population develops this complication (1-3). Studies based on autopsy (4, 5), lung resection (3, 6), biopsy (7-9), and bronchoalveolar lavage (10, 11) have established that all smokers develop peripheral lung inflammation. This process underlies the pathogenesis of emphysema (12), and we postulate that amplification of this cigarette smoke-induced inflammatory response in some individuals accounts for the fact that only a minority of heavy smokers develop emphysema.
Recent studies suggest that latent adenoviral infection (i.e., the persistence of adenoviral DNA and protein without replication of a complete virus) is one of the factors that can amplify lung inflammation. The adenovirus targets lung epithelium (13), its DNA persists in higher quantities in lungs of patients with chronic obstructive pulmonary disease (COPD) (14), and the adenoviral transactivating protein (E1A) has been demonstrated in the epithelium of the conducting airways, bronchial glands, and alveolar epithelium (15). Others have shown that E1A protein expression causes resting cells to enter the cell cycle (16), that it modulates the cell susceptibility to lysis by tumor necrosis factor-α (TNF-α) (17), and that it increases transcription of host genes by associating with the DNA binding domains of host transcription factors (18). Transfection of E1A into A549 cells results in excess expression of ICAM-I and interleukin-8 (IL-8) following challenge (19, 20) and guinea pigs with latent adenoviral infection develop an exaggerated inflammatory response (21) and excess emphysema when challenged with cigarette smoke (22).
The expression of viral proteins by lung epithelial cells also amplifies the lung inflammatory response through adaptive immune mechanisms. Recent studies show that transgenic mice that express an influenza viral protein in their alveolar epithelial cells develop an epithelial-driven inflammatory response when they are subsequently challenged by adoptive transfer of CD8+ T cells sensitized in vitro to the same viral protein (23, 24). This response is MHC restricted but is independent of the mechanisms used by CD8 cytotoxic T cells to lyse virus-infected cells. As it represents a specific response to epithelial expression of a viral protein expressed in the absence of a lytic viral infection, it may be relevant to the role proposed for latent viral infection in the lung.
The present study was designed to test the hypothesis that the lung inflammatory response is amplified during emphysematous lung destruction using quantitative histological methods to compare lung tissue from heavy smokers with advanced emphysema with lung tissue from patients who had smoked similar amounts but were emphysema free. The same lung tissue was also examined for the expression of adenoviral E1A protein to determine if its expression was related to the amplification of the inflammatory response and the progression of emphysema.
Tissue from the patients with severe emphysema (n = 7) was obtained by lung volume reduction surgery (LVRS) at the University of Pittsburgh. Lung tissue from the patients with normal lung function and no emphysema (n = 7) and mild emphysema (n = 7) was obtained from lung lobes resected for tumor at St. Paul's Hospital in Vancouver, British Columbia, Canada. All three groups were heavy smokers and had similar age and sex distributions. The procedures used to obtain the tissue were approved by the institutional review boards of both the University of British Columbia and the University of Pittsburgh. All patients signed informed consent allowing their preoperative lung function measurements, CT scans, and resected tissue to be used for the purposes of this study.
The methods used to obtain the measurements of forced expiratory volume (FEV1), subdivisions of lung volume (total lung capacity [TLC], functional residual capacity [FRC], and residual volume [RV]), and diffusing capacity for carbon monoxide (Dl CO) reported here met the standard set by the American Thoracic Society and have been described in our previous publications (3, 25).
The severity of the emphysema was determined using a recently described technique where the computed tomography (CT)-determined lung volume is used as the reference volume for the quantitative histology performed on the resected lung tissue. This method has been detailed elsewhere and provides measurements of lung weight and volume in situ as well as estimates of the surface/volume ratio and total surface area of both lungs (26).
The histology was examined by observers who were blinded to all information except that present on the slide. The volume fraction Vv of the airspace and the parenchymal tissue surrounding the airspace excluding conducting airways and larger vessels was determined by point counting at 100× magnification and the Vv of these spaces taken up by each cell type was determined by point counting at 400× magnification. This process established the volume fraction (Vv) of either airspace or tissue taken up by the cell type of interest (29). The total volume of the specifically stained cells was determined by multiplying this histologically determined Vv by the CT-determined reference volume of total lung tissue and airspace (29). The number of cells of each type was then determined by dividing the volume of all cells of a particular type by the volume of an individual cell (29-31). The number of cells present in the tissue and airspace compartments was expressed either as the total number present in the entire lung or as cells/m2 surface area.
The data concerning patient demographics, pulmonary function and CT scans (Tables 1 and 2) were analyzed using a one-way analysis of variance. The data for each cell type (Table 3) were analyzed using a single function analysis of variance with a post hoc t test to determine differences between groups (Microsoft Excel). The number of cells present/m2 surface area was related to the percent of the lung occupied by emphysema determined by CT (Figure 1 and Table 4) using a standard curve fitting technique (Cricket Graph, Cricket Software, Malvern, PA).
|Age, yr||61 ± 2||66 ± 3||68 ± 2|
|Sex ratio, M/F||6/1||6/1||6/1|
|Pack-years||52 ± 8||28 ± 6||67 ± 13*|
|FEV1, %P||87 ± 5||84 ± 5||30 ± 2†|
|RV, %P||146 ± 13||154 ± 25||211 ± 21‡|
|FRC, %P||134 ± 9||157 ± 9||186 ± 11§|
|TLC, %P||114 ± 6||122 ± 6||126 ± 6|
|Dl CO, %P||94 ± 5||78 ± 13||32 ± 6‡|
|Lung volume, ml||5135 ± 210||6210 ± 388||6857 ± 629*|
|Lung weight, g||1059 ± 49||1085 ± 33||844 ± 55†|
|Emphysema, %||1.7 ± 0.4||13 ± 2*||43 ± 4‡|
|Surface area, m2||131 ± 7||119 ± 3||65 ± 3‡|
|Surface area/volume||21.3 ± 1||15 ± 2*||5 ± 1‡|
|PMN, ×1012||24 ± 8.3||20 ± 5.1||29 ± 5.9||50 ± 30||140 ± 29*||300 ± 50*|
|Macro, ×1012||4.5 ± 1.8||270 ± 80||4.4 ± 2.5||330 ± 66||71 ± 19*||4000 ± 400*|
|Eos, ×108||25 ± 8||3.4 ± 1.2||21 ± 7.9||1.0 ± 0.3||220 ± 50*||7.8 ± 1.4†|
|CD4, ×1012||45 ± 10||58 ± 30||44 ± 13||76 ± 18||330 ± 58*||750 ± 89*|
|CD8, ×1012||31 ± 6||40 ± 17||34 ± 12||46 ± 14||250 ± 51*||1400 ± 590†|
|CD20, ×1012||0 ± 0||0 ± 0||0 ± 0||0 ± 0||47 ± 29||2.6 ± 2.4|
Table 1 summarizes the data concerning age, sex, smoking history, and lung function for the three groups of patients. The control subjects had smoked amounts similar to those with severe disease and those with mild disease had smoked less than either of the other two groups. The lung function tests show that those who had severe disease had a markedly reduced forced expiratory volume in 1 s (FEV1) (p < 0.005), a reduction in the diffusing capacity (Dl CO) (p < 0.005), and an increased RV (p < 0.05) compared with the control subjects and those with mild disease. They also had an increased FRC (p < 0.005) compared with control subjects, but there was no difference in TLC between these three groups.
The data obtained using computed tomography and lung morphometry (Table 2) show that the proportion of the lung taken up by emphysema increased from 1.7 ± 0.4% of the total lung volume in control subjects to 13 ± 2% in mild and 43 ± 4% in severe disease. The combined volume of lung tissue and air was increased (p < 0.05) and lung weight decreased (p < 0.05) when severe emphysema was present. Table 2 also shows that the surface-to-volume ratio of the entire lung decreased in mild (p < 0.05) and further decreased in severe disease (p < 0.0005) compared with control subjects. The total lung surface area, on the other hand, was similar in the control subjects and those with mild disease but was severely reduced in the patients with advanced emphysema (p < 0.0005).
Table 3 shows the total number of inflammatory cells present in the lung in each of the three groups in which severe emphysematous destruction is associated with an absolute increase in polymorphonuclear cells (PMN), macrophages, eosinophils, and CD4 and CD8 cells. Although the CD20 or B cells showed a similar trend, the result was not significant.
Figure 1A shows a photomicrograph of macrophages/monocytes in alveolar tissue and airspaces from one of the patients, and Figure 1B shows the relationship between the number of macrophages present in tissue and airspaces and the percent of the lung occupied by emphysema for all of the patients in the study. This shows a sharp increase in the macrophage number/m2 in the airspaces beginning when approximately 30% of the lung volume is occupied by emphysematous lesions. The number of macrophages in the tissue also increased, but this is not apparent in Figure 1B because the numbers in the tissue are much lower than in the airspaces.
Table 4 shows the data concerning the R2 values for the equation relating the number of each type of cell/m2 surface area and the percentage of the lung taken up by emphysematous destruction. These values ranged from 0.858 (CD8) to 0.483 (CD20) cells in peripheral lung tissue and from 0.630 (CD4) to 0.198 (CD20) in the airspace. These values suggest a strong correlation between the number of inflammatory cells present/m2 surface area and emphysematous destruction of the lung for all cells except the eosinophils and CD20 cells.
Figure 2A shows an example of epithelial cells on the alveolar surface of the lung of a patient with severe emphysema containing E1A protein and Figure 2B shows that there was an approximate 5-fold increase in these E1A-containing epithelial cells in the patients with mild emphysema compared with the control subjects and a 41-fold increase in these cells in the lungs from the patients with severe emphysema. This increase in E1A-containing cells was also associated with an increased expression of ICAM-1 by the lung surface epithelium (Figure 2B).
The data reported here show that all of the inflammatory cell types present except the B cells were increased in the lungs of patients with severe emphysema even though these patients had not smoked more than the control subjects. The total number of inflammatory cells present in the lung was used as a measure of the intensity of the inflammatory reaction rather than the cell number per gram lung or square meter of surface area because both lung weight and surface area are reduced by emphysema. The quantitative techniques used in this study are dependent on a random lung sample and might have been artificially increased in tissue obtained by LVRS where the surgeon preferentially samples diseased tissue. To guard against this error, the total number of cells was recalculated using only the volume of emphysematous lung tissue as the reference volume. This correction reduced the reference volume to 43 ± 4% of its original value but still resulted in an increase in the same cell types with similar estimates of significance (data not shown). There was also a relationship between the total amount of emphysematous destruction present on the lung CT scan and the number of each cell type present in the tissue and air spaces/m2 lung surface area. The R2 values of this relationship were above 0.8 for CD4, CD8, and MACS, and 0.749 for PMN in the tissue with slightly lower values for the airspace. Therefore, we conclude that there is an increase in the intensity of the inflammatory response in the lung tissue being destroyed by emphysema, and all of the cells that have been implicated in the pathogenesis of emphysema are present in increased numbers (12, 32-34).
There was a 5-fold increase in the number of alveolar lung epithelial cells expressing adenoviral E1A protein in mild emphysema and a 41-fold increase in these cells in severe emphysema that paralleled the increase in inflammatory cells in the tissue and airspaces of the lung. This observation is consistent with the hypothesis that a persistent intracellular pathogen such as the adenovirus may be capable of amplifying cigarette smoke-induced inflammation in humans. The adenovirus replicates in type II cells during acute infections and viral DNA and protein persist in the nuclei of lung epithelial cells long after viral replication has stopped (14, 15, 35). Viral DNA also persists in tonsils (36) and peripheral blood lymphocytes (37) following acute lytic infection. The alveolar type II cell targeted by the adenovirus has a cuboidal structure that covers 7% of the alveolar surface area of the normal lung (31). The wide dispersion of these cells and low sensitivity of immunohistochemistry probably account for the small numbers of E1A-positive cells detected in our studies. This result is consistent with studies in an animal model of latent adenoviral infection showing that E1A expression by low numbers of epithelial cells was associated with an amplified cigarette smoke-induced lung inflammation (21, 22, 35).
Walker and coworkers (38, 39) have confirmed the earlier observation of Damiano and coworkers (40) that the PMN enter the alveolar space through the junctions between type I and type II alveolar cells. This places the type II cell at an ideal site to amplify inflammatory cell migration. In vitro studies have shown that a type II-like A549 cell line transfected with E1A produces excess IL-8 and ICAM-1 following challenge (19, 20), and the present study shows there is also excess alveolar surface epithelial cell expression of ICAM-1 in severe emphysema. These data are consistent with the hypothesis that a latent infection of a small number of type II alveolar epithelial cells can amplify the passage of inflammatory cells along their migratory pathway and increase their number in the tissue and airspace. The adenovirus avoids destruction by the host immune response by expressing its E3/19K protein, which binds to host MHC class I protein and lowers the expression of viral antigens on the cell surface preventing recognition by cytotoxic T cells. As the strength of binding between viral E3/19K and host HLA proteins is HLA dependent (41), the genetic makeup of the host could influence the observed difference in E1A expression by host epithelial cells and its effect on the inflammatory process.
The concept that the expression of a viral protein by alveolar epithelial cells in the absence of a replicating viral infection can amplify lung inflammation has received recent support from studies in transgenic mice (23, 24). These authors developed a transgenic model in which a gene coding for an influenza viral protein was put under the control of a promoter for the lung surfactant protein gene to create mice that constitutively expressed this viral protein in their alveolar type II epithelial cells. They then showed that adoptive transfer of CD8 lymphocyte clones sensitized in vitro to the same viral protein produced a remarkable degree of lung inflammation in these animals. Interestingly, they described a novel response that was not associated with epithelial destruction by granzymes or FAS expression, but it was caused by antigen-specific MHC Class 1-restricted CD8 cells interacting with the epithelial cells. This interaction involved TNF signaling because recipient mice lacking the TNF-αR1 receptor did not respond. Sensitized CD8+ lymphocytes stimulated the epithelial cells expressing the viral protein in vitro to produce monocyte chemoattractant protein 1 (MCP-1) and the macrophage inflammatory protein (M1P2), which could account for a very large influx of alveolar macrophages during this response.
These transgenic experiments differ sharply from those reported here in that every type II epithelial cell in the mouse constitutively expresses the protein, whereas only a small fraction of the human alveolar cells express the viral protein. However, they illustrate that an adaptive immune response to a viral protein expressed by the epithelial cells can influence the inflammatory process in the absence of viral replication. We suggest that the increase in expression of adenoviral E1A protein by the lung epithelial cells in the human lungs with severe emphysema reported here may have a similar effect through both an adaptive immune mechanism illustrated by the transgenic experiments (23, 24) and the direct effect of the E1A on ICAM-1 (19) and IL-8 (20) expression. A more complete understanding of the effect of residual intracellular pathogens such as the adenovirus on the host response could shed important light on cigarette smoke-induced inflammation and may explain why only a fraction of heavy smokers develop emphysema.
|1.||Ryder RC, Dunnill MS, Andersen JAA quantitative study of bronchial mucus gland volume, emphysema and smoking in a necropsy population. J Pathol10419715971|
|2.||Thurlbeck WM, Ryder RC, Steinberg NA comparative study of the severity of emphysema in necropsy populations in three different countries. Am Rev Respir Dis1091974239248|
|3.||Hogg JC, Wright JL, Wiggs BR, Coxson HO, Opazo Saez AO, Pare PD. Lung structure and function in cigarette smokers. Thorax 1994;49: 473–478.|
|4.||McLean KAPathogenesis of pulmonary emphysema. Am J Med2519586274|
|5.||Niewoehner DE, Kleinerman J, Rice DBPathologic changes in the peripheral airways of young cigarette smokers. N Engl J Med2911974755758|
|6.||Wright JL, Lawson LM, Pare PD, Wiggs BR, Kennedy S, Hogg JCMorphology of peripheral airways in current smokers and ex-smokers. Am Rev Respir Dis1271983474477|
|7.||O'Shaughnessy TC, Ansari TW, Barnes NC, Jeffery PKInflammation in bronchial biopsies of subjects with chronic bronchitis: inverse relationship of CD8+ T lymphocytes with FEV1. Am J Respir Crit Care Med1551997382387|
|8.||Saetta M, Di Stefano A, Turato G, Facchini FM, Corbino L, Mapp CE, Maestrelli P, Ciaccia A, Fabbri LMCD8+ T-lymphocytes in peripheral airways of smokers with chronic obstructive pulmonary disease. Am J Respir Crit Care Med1571998822826|
|9.||Ollerenshaw SL, Woolcock AJCharacteristics of the inflammation in biopsies from large airways in subjects with asthma and chronic airflow limitation. Am Rev Respir Dis1451992922927|
|10.||Hunninghake GW, Crystal RG. Cigarette smoking and lung destruction: accumulation of neutrophils in the lungs of cigarette smokers. Am Rev Respir Dis 1993:128:833–838.|
|11.||Gadek JE, Fells JA, Crystal RGCigarette smoking induces a functional antiprotease deficiency in the lower respiratory tract of humans. Science206197913151316|
|12.||Janoff ABiochemical links between cigarette smoking and pulmonary emphysema. J Appl Physiol551983285293|
|13.||Hogg JC, Irving WL, Porter H, Evans M, Dunnill MS, Fleming KIn situ hybridization studies of adenoviral infections of the lung and their relationship to follicular bronchiectasis. Am Rev Respir Dis139198915311535|
|14.||Matsuse T, Hayashi S, Kuwano K, Keunecke H, Jefferies WA, Hogg JCLatent adenoviral infection in the pathogenesis of chronic airways obstruction. Am Rev Respir Dis1461992177184|
|15.||Elliott WM, Hayashi S, Hogg JCImmunodetection of adenoviral E1A proteins in human lung tissue. Am J Respir Cell Mol Biol121995642648|
|16.||Whyte P, Buchkovich KJ, Horowitz JM, Friend SH, Raybuck M, Weinberg RA, Harlow EAssociation between an oncogene and an anti-oncogene: the adenovirus E1A proteins bind the retinoblastoma gene product. Nature3341988124129|
|17.||Duerksen-Hughes P, Wold WSM, Gooding LRAdenovirus E1A renders infected cells sensitive to cytolysis by tumor necrosis factor. J Immunol143198941934200|
|18.||Liu F, Green MRPromoter targeting by adenovirus E1a through interaction with different cellular DNA-binding domains.Nature3681994520525|
|19.||Keicho N, Elliott WM, Hogg JC, Hayashi SAdenovirus E1A gene dysregulates ICAM-1 expression in transformed pulmonary epithelial cells. Am J Respir Cell Mol Biol1619972330|
|20.||Keicho N, Elliott WM, Hogg JC, Hayashi SAdenovirus E1A upregulates interleukin-8 expression induced by endotoxin in pulmonary epithelial cells. Am J Physiol2721997L1046L1052|
|21.||Vitalis TZ, Kern I, Croome A, Behzad H, Hayashi S, Hogg JCThe effect of latent adenovirus 5 infection on cigarette smoke-induced lung inflammation. Eur Respir J111998664669|
|22.||Meshi B, Vitalis TZ, Ionescu D, Elliott WM, Nang X-D, Hayashi S, Hogg JC. Emphysematous lung destruction by cigarette smoke: the effect of latent adenoviral infection on the lung inflammatory response. Am J Respir Cell Mol Biol (In press).|
|23.||Zhao MQ, Stoler MH, Liu AN, Wei B, Soguero C, Hahn YS, Enelow RIAlveolar epithelial cell chemokine expression triggered by antigen-specific cytolytic CD8+ T cell recognition. J Clin Invest1062000R49R58|
|24.||Liu AN, Mohammed AZ, Rice WR, Fiedeldey DT, Liebermann JS, Whitsett JA, Braciale TJ, Enelow RIPerforin-independent CD8+ T-cell-mediated cytotoxicity of alveolar epithelial cells is preferentially mediated by tumor necrosis factor-α: relative sensitivity to Fas ligand. Am J Respir Cell Mol Biol201999849858|
|25.||Sciurba FC, Rogers RM, Keenan RJ, Slivka WA, Gorcsan JImprovement in pulmonary function and elastic recoil after lung-reduction surgery for diffuse emphysema. N Engl J Med334199610951099|
|26.||Coxson HO, Rogers RM, Whittall KP, D'Yachkova Y, Paré PD, Sciurba FC, Hogg JCA quantification of the lung surface area in emphysema using computed tomography. Am J Respir Crit Care Med1591999851856|
|27.||Cordell JL, Falini B, Erber WN, Ghosh AK, Abdulaziz Z, MacDonald S, Pulford KAF, Stein H, Mason DYImmunoenzymatic labeling of monoclonal antibodies using immune complexes of alkaline phosphatase and monoclonal anti-alkaline phosphatase (APAAP complexes). J Histochem Cytochem321984219229|
|28.||Hansel FK. 1953. Clinical allergy. St. Louis, MO: CV Mosby.|
|29.||Coxson HO, Hogg JC, Mayo JR, Behzad H, Whittall KP, Schwartz DAQuantification of idiopathic pulmonary fibrosis using computed tomography and histology. Am J Respir Crit Care Med155199716491656|
|30.||Schmid-Schonbein GW, Shih YY, Chien SMorphometry of human leukocytes. Blood561980866875|
|31.||Crapo JD, Barry BE, Gehr P, Bachofen M, Weibel ERCell number and cell characteristics of the normal lung. Am Rev Respir Dis1261982332337|
|32.||Shapiro SDElastolytic metalloproteinases produced by human mononuclear phagocytes: potential roles in destructive lung disease. Am J Respir Crit Care Med1501994S160S164|
|33.||Finkelstein R, Fraser RS, Ghezzo H, Cosio MGAlveolar inflammation and its relation to emphysema in smokers. Am J Respir Crit Care Med152199516661672|
|34.||Saetta M, Di Stefano A, Maestrelli P, Turato G, Ruggieri MP, Roggeri A, Calcagni P, Mapp CE, Ciaccia A, Fabbri LMAirway eosinophilia in chronic bronchitis during exacerbations. Am J Respir Crit Care Med150199416461652|
|35.||Vitalis TZ, Keicho N, Itabashi S, Hayashi S, Hogg JCA model of latent adenovirus 5 infection in the guinea pig (Cavia porcellus). Am J Respir Cell Mol Biol141996225231|
|36.||Green M, Wold WSM, Mackey JK, Rigden PAnalysis of human tonsil and cancer DNAs and RNAs for DNA sequences of group C (serotypes 1,2,5,6) human adenoviruses. Proc Natl Acad Sci USA76197966066610|
|37.||Horvath J, Palkonyay L, Weber JGroup C adenovirus DNA sequences in human lymphoid cells. J Virol591986189192|
|38.||Behzad A, Chu F, Walker DCFibroblasts are in a position to provide directional information to migrating neutrophils during pneumonia in rabbit lungs. Microvas Res511996303316|
|39.||Walker DC, Behzad AR, Chu FNeutrophil migration through pre-existing holes in the basal lamina of alveolar capillaries and epithelium during streptoccocal pneumonia. Microvas Res501995397416|
|40.||Damiano VV, Cohen A, Tsang A, Batra G, Petersen RA morphologic study of the influx of neutrophils into dog lung alveoli after lavage with sterile saline. Am J Pathol1001980349364|
|41.||Severinsson L, Martens I, Peterson PADifferential association between two human MHC class 1 antigens and an adenoviral glycoprotein. J Immunol137198610031009|
Dr. Retamales is a Visiting Scientist from the Department of Pathology, Faculty of Medicine, University of Chile, Santiago, Chile. Dr. Coxson is supported by a Canadian Lung Association-Boehringer-Ingelheim Fellowship. Supported by grants from the Medical Research Council of Canada 7246, NIH HLBI, and the George Love Research Fund.
This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournal.org