Published ahead of print on July 17, 2008, doi:10.1164/rccm.200804-550OC
© 2008 American Thoracic Society doi: 10.1164/rccm.200804-550OC
Telomere Shortening in Familial and Sporadic Pulmonary Fibrosis1 Eugene McDermott Center for Human Growth and Development, University of Texas Southwestern Medical Center, Dallas, Texas; 2 Division of Pulmonary and Critical Care Medicine, University of Washington Medical Center, Seattle, Washington; and 3 Division of Pulmonary and Critical Care Medicine, University of Texas Southwestern Medical Center, Dallas, Texas Correspondence and requests for reprints should be addressed to Christine Kim Garcia, M.D., Ph.D., University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd., Dallas, TX 75390-8591. E-mail: christine.garcia{at}utsouthwestern.edu
Rationale: Heterozygous mutations in the coding regions of the telomerase genes, TERT and TERC, have been found in familial and sporadic cases of idiopathic interstitial pneumonia. All affected patients with mutations have short telomeres. Objectives: To test whether telomere shortening is a frequent mechanism underlying pulmonary fibrosis, we have characterized telomere lengths in subjects with familial or sporadic disease who do not have coding mutations in TERT or TERC. Methods: Using a modified Southern blot assay, the telomerase restriction fragment length method, and a quantitative polymerase chain reaction assay we have measured telomere lengths of genomic DNA isolated from circulating leukocytes from normal control subjects and subjects with pulmonary fibrosis. Measurements and Main Results: All affected patients with telomerase mutations, including case subjects heterozygous for newly reported mutations in TERT, have short telomere lengths. A significantly higher proportion of probands with familial pulmonary fibrosis (24%) and sporadic case subjects (23%) in which no coding mutation in TERT or TERC was found had telomere lengths less than the 10th percentile when compared with control subjects (P = 2.6 x 10–8). Pulmonary fibrosis affectation status was significantly associated with telomerase restriction fragment lengths, even after controlling for age, sex, and ethnicity (P = 6.1 x 10–11). Overall, 25% of sporadic cases and 37% of familial cases of pulmonary fibrosis had telomere lengths less than the 10th percentile. Conclusions: A significant fraction of individuals with pulmonary fibrosis have short telomere lengths that cannot be explained by coding mutations in telomerase. Telomere shortening of circulating leukocytes may be a marker for an increased predisposition toward the development of this age-associated disease.
Key Words: idiopathic pulmonary fibrosis • pulmonary fibrosis • telomere length • aging • interstitial lung disease
The idiopathic interstitial pneumonias are characterized by damage to the lung parenchyma by a combination of fibrosis and inflammation. The prototype of these diseases is idiopathic pulmonary fibrosis (IPF), the prevalence and annual incidence of which increase dramatically with age (1). A clue to the genetic underpinnings of the familial subtype of this disorder emerged from the discovery that a subset (  15%) of patients with IPF is heterozygous for mutations in the genes encoding the protein component (TERT) and the RNA component (TERC) of telomerase, a ribonucleoprotein enzyme that catalyzes the addition of hexameric nucleotide repeats to the ends of linear chromosomes (2, 3). One mutation in TERT was found in a subject with the sporadic form of the disease who had no family history of IPF (2). The mutations in TERT segregate not only with individuals that met the strict clinical diagnosis of IPF (4), but also with several with pulmonary fibrosis favoring the upper lobes of the lung and others with unclassified pulmonary disease (2). The missense, frameshift, and splice site mutations in TERT found in the familial and sporadic cases of pulmonary fibrosis span the entire coding regions of the gene but cluster in conserved domains. One frameshift mutation, V747fs, which is predicted to be missing half the reverse transcriptase domain, has little enzymatic activity in a recombinant in vitro assay; cotranslation of various ratios of plasmids encoding this mutation and the wild-type TERT protein suggests a mechanism of haploinsufficiency (2). The missense mutations in TERT that have been identified in patients with IPF have between 30 and 100% wild-type telomerase activity in a rabbit reticulocyte in vitro assay (2). The telomere lengths of circulating leukocytes of individuals with these mutations and pulmonary fibrosis are reproducibly shorter than those of age-matched control subjects (2, 3). Telomere shortening is a common feature of dyskeratosis congenita and bone marrow failure syndromes, two diseases previously associated with mutations in DKC1, TERT, TERC, or TINF2 (5) (reviewed in Garcia and coworkers [6]). Irrespective of the gene in which a mutation is found, patients with dyskeratosis congenita have short telomeres in their circulating leukocytes (7). Mutations in TERT are present in up to 4% of individuals with acquired aplastic anemia (8), and yet short telomere lengths are found in 34% of patients with this disease (9). In the patients with aplastic anemia, the severity of disease is directly related to the degree of telomere shortening; moreover, a lack of response to immunosuppressive agents is related to shorter telomere length (10, 11). In this study, we have sequenced the genes encoding telomerase, TERT and TERC, in patients with the familial and sporadic forms of idiopathic interstitial pneumonias. We have compared telomere lengths in circulating leukocytes of both familial and sporadic case subjects with those of normal control subjects. All patients with pulmonary fibrosis in whom mutations were identified had telomere lengths that were less than the 10th percentile when compared with age-matched control subjects. In addition, we found that 20–25% of subjects with familial or sporadic pulmonary fibrosis who did not have any detectable mutations in telomerase also had telomere lengths less than the 10th percentile. These findings suggest that telomere shortening is a more generalized feature of IPF.
Human Subjects This study was approved by the University of Texas Southwestern Medical Center Institutional Review Board (Dallas, TX). Written informed consent was obtained from all subjects. Each participant completed a medical questionnaire. Medical records were obtained when available. Each sporadic case and at least one member of each kindred with familial pulmonary fibrosis carried a diagnosis of idiopathic interstitial pneumonia or unclassifiable interstitial pneumonia in concordance with established criteria (4). Kindreds with familial pulmonary fibrosis were defined as those in which there was at least one other relative of the proband who was affected with an interstitial lung disease; all sporadic case subjects had no affected first- or second-degree family members. Sporadic cases of pulmonary fibrosis due to known causes, associated with collagen vascular disease, sarcoidosis, and other diffuse parenchymal lung disease were excluded from study. All case subjects and affected family members were diagnosed at an age of 21 years or more. A subset of the probands (n = 46) and sporadic case subjects (n = 42) included in this study were also included in a previous study (2). Patients with idiopathic, familial, or anorexigen exposure associated pulmonary arterial hypertension and who fulfilled the following criteria were included: (1) right heart catheterization measurement of mean pulmonary artery pressure greater than 25 mm Hg, pulmonary capillary wedge pressure less than 15 mm Hg, and pulmonary vascular resistance greater than 3 Wood units; (2) a ratio of total lung capacity to forced vital capacity exceeding 70% plus a normal chest X-ray or a ratio of total lung capacity to forced vital capacity exceeding 50% plus a normal computed tomography scan of the chest; and (3) exclusion of pulmonary arterial hypertension associated with collagen vascular disease, congenital shunts, HIV infection, portal hypertension, and all other forms of pulmonary hypertension including pulmonary venous hypertension or that associated with lung disease, chronic thromboembolism, or disorders of the pulmonary vasculature. Genomic DNA samples for the population of normal control subjects (n = 201; age, 19–89 yr) were obtained from a sample of unrelated, multiethnic individuals from Dallas, Texas from H. Hobbs. The ethnicity of each subject was self-assigned. Genomic DNA was isolated from circulating leukocytes with an Autopure LS (Qiagen, Valencia, CA). Buccal swabs were obtained with sterile BD Falcon SWUBE applicators (BD Biosciences, San Jose, CA); DNA was isolated with Gentra (Qiagen) reagents.
Sequencing and Mutation Analysis
Telomerase Repeat Amplification Protocol Assays
Terminal Restriction Fragment Length Analysis of Telomere Length
Quantitative Polymerase Chain Reaction Determination of Telomere Length
Statistical Methods
Six New Mutations in TERT in Subjects with Adult-Onset Pulmonary Fibrosis We sequenced the coding and flanking intronic sequences of TERT and TERC in 25 probands and 34 sporadic subjects with pulmonary fibrosis and identified six new mutations in TERT—five in probands with familial pulmonary fibrosis (Figure 1A) and one in a patient with sporadic pulmonary fibrosis (the lung phenotype is defined in METHODS). Additional clinical findings of the subjects and their relevant family members are provided in Table E1 (see the online supplement). Missense mutations were found in the probands of family F55 (K1050E), F80 (P702L), F106 (L1019F), F107 (V694M), and F119 (P704S). All three members of family F80 with IPF were heterozygous for the same missense mutation as the proband (P702L). One of the patients with sporadic disease had a missense mutation in the C-terminal region of TERT (S957R). All of these missense mutations involve highly conserved residues in regions of the protein (Figures 1B and 1C) that have postulated roles in enzymatic activity, processivity, and cellular location of the protein (15–17). None of the mutations were detected in a multiethnic panel of 528 individuals sequenced as control subjects (8). All of the mutations had not been identified previously except V694M, which was found in a 34-year-old subject with moderate aplastic anemia (8). The 60-year-old proband of family F107 with this mutation has a normal complete blood count, which does not suggest any bone marrow dysfunction (see Table E1). Electropherograms of all the sequence mutations are shown in Figure E2.
To determine the effect of these amino acid substitutions in TERT on enzymatic activity, each mutation was reproduced by site-directed mutagenesis and tested in vitro after coexpressing the recombinant telomerase protein and RNA. For these studies we used the telomere repeat amplification protocol (TRAP) assay. As seen in Figure 1D, the missense mutations had between 23 and 88% of the telomerase activity of the wild-type protein.
Comparison of Telomere Lengths to Those of a Normal Population
Next, we examined the telomere lengths of individuals from the families in Figure 1 and described by Tsakiri and coworkers (2), in which a TERT or TERC mutation was not identified (noncarriers). Six of 76 noncarriers fell below the 10th percentile predicted line. In contrast, 52 of the 61 TERT/TERC mutation carriers fell below this threshold. All the family members who had been diagnosed with pulmonary fibrosis were mutation carriers who were more than 48 years of age and their telomere lengths were uniformly less than the 10th percentile when compared with the control subjects (Figure 2B).
Telomere Shortening in Pulmonary Fibrosis Subjects without Mutations in TERT or TERC
We also analyzed the telomere lengths in our sporadic case subjects with pulmonary fibrosis who did not have a mutation in the coding region of telomerase. Figure 3B shows that some subjects, such as CKG059 and CKG973, have telomere lengths similar to those of age- and ethnicity-matched control subjects. Others, such as CKG030 and CKG545, have telomere lengths that are shorter than those of matched control subjects and are similar in length to the sporadic case who is heterozygous for the TERT S957R mutation (4.5, 4.4, and 4.1 kb, respectively). We determined the distribution of telomere lengths in familial (n = 59) and sporadic (n = 73) case subjects with pulmonary fibrosis who did not have a detectable mutation in the telomerase genes. The majority of case subjects were male and white (Table 1), reflecting the demographics of this disease (1, 24). In general, affected individuals with telomerase mutations were similar to the familial and sporadic case subjects without telomerase mutations regarding smoking status, age at diagnosis, pathologic subtype of idiopathic interstitial pneumonia, and other comorbidities (Table 2). A comparison of the telomere lengths, as determined by the TRFL assay, of the probands and sporadic cases with normal subjects is provided in Figures 4B and 4C, respectively. As expected, all probands with mutations in TERT or TERC had mean TRFLs below the 10th percentile prediction line. In addition, 14 of the 59 probands (24%) with familial pulmonary fibrosis who did not have a mutation in telomerase had mean TRFLs below the 10th percentile prediction line; this is more than was observed in control subjects (P = 8.0 x 10–6). Some of these subjects had TRFLs that are just as short as those with mutations in TERT or TERC. If a more stringent cutoff was used, such as the 1st or 5th percentile, there were still significantly more subjects without telomerase mutations below these thresholds than control subjects (data not shown). Similarly, there were more sporadic case subjects without coding mutations in TERT or TERC who had telomere lengths that were below the 10th percentile prediction line than would be expected by chance (17 of 73 sporadic case subjects without telomerase mutations vs. 8 of 201 control subjects; P = 2.6 x 10–6) (Figure 4C). No consistent distinguishing phenotype was seen for familial and sporadic case subjects with short telomere lengths versus those whose telomere lengths were greater than the 10th percentile.
Telomere length of genomic DNA was also determined using an independent quantitative PCR method (see METHODS). The average telomere length was determined by assessing the ratio of telomere copy number repeats to a single copy gene, β2-globin (T/S ratio) in experimental samples relative to a reference sample. We observed a good correlation between this method and the TRFL method for determining telomere length (P < 2.2 x 10–16; and see Figure E3). For control subjects, the relative T/S ratio decreased with age by 0.0096 unit/year, generally consistent with other reports of rates of attrition (14, 25, 26). The 10th and 90th percentile predicted bands were determined from the linear regression model (Figure 4D). The distributions for the probands of kindreds with familial pulmonary fibrosis and for the sporadic pulmonary fibrosis case subjects are shown in Figures 4E and 4F, respectively. As expected, all the probands of the familial cases with mutations in TERT or TERC fell below the 10th percentile prediction line (Figure 4E). When this band was used as an arbitrary cutoff, we again observed a significant excess of subjects with familial or sporadic pulmonary fibrosis with telomere lengths below this boundary (Table 3).
Consistent with telomere shortening with age, the percentage of short telomeres (determined from the TRFL Southern blots) also increased with age. All probands with telomerase mutations had logit(p) scores greater than the 90th percentile (Figure 5B), and again we found a significantly greater than expected number of probands and sporadic case subjects without telomerase mutations with logit(p) scores above the 90th percentile predicted band (P = 9.5 x 10–9) (Figures 5B and 5C; and see Table 3). All the probands of the familial cases with mean telomere lengths less than the10th percentile had logit(p) values greater than the 90th percentile.
Multiple regression analysis showed a correlation between telomere length and sex after controlling for age and ethnicity (P = 4.9 x 10–4). There was a small difference in mean TRFLs between men (5.25 kb) and women (5.45 kb) with pulmonary fibrosis without telomerase mutations (P = 0.004) (Table 4). Smoking has been associated with short telomere lengths (27, 28). In contrast with sex, we did not find a correlation between telomere length and smoking by multivariate analysis in the cases after controlling for sex and ethnicity. Because ethnicity was previously reported to be associated with telomere length (29), we analyzed white male and female case and control subjects separately and found similar results in these smaller subgroups when using all three methods for assessing telomere length (see Table E2).
To determine the specificity of the short telomere phenotype we measured telomere length by the quantitative PCR method for a cohort of patients with idiopathic, familial, or anorexigen-associated pulmonary arterial hypertension (see METHODS for a description of this patient population) (Figure 6). There was not a significant excess of subjects in this patient population below the 10th percentile predicted line when compared with control subjects, either for the entire group or sex- and ethnicity-matched case and control subjects.
In this article, we show short telomere lengths of circulating leukocytes in 25% or more of all subjects with familial or sporadic idiopathic interstitial pneumonia. The findings of the inherited mutations in telomerase in families and individuals with IPF initially suggested that telomerase dysfunction is important for the molecular pathogenesis of this disease. Here we find that of 71 unrelated probands, 12 (17%) with heterozygous mutations in TERT or TERC and an additional 14 (20%) probands without telomerase mutations have mean telomere lengths of circulating leukocytes that are shorter than the 10th percentile by the TRFL assay. Therefore, 37% of probands with the familial form of the disease have evidence of short telomeres. Similarly, of 75 unrelated sporadic cases, 2 (3%) have mutations in TERT and 17 have TRFLs less than the 10th percentile; so 25% of sporadic cases have short telomeres. These findings suggest that short telomere lengths are commonly associated with both the familial and sporadic forms of pulmonary fibrosis and can be only partially explained at the molecular level by coding mutations in TERT and TERC. The sequencing of both genes has been limited to the coding regions and their surrounding intronic splice sites. We cannot rule out noncoding mutations or small deletions in either gene. In one expanded family, the short telomere phenotype segregates with pulmonary fibrosis as an autosomal dominant trait but does not cosegregate with haplotypes about TERT and TERC, suggesting that other genetic loci may contribute to this trait (data not shown). All families collected with familial pulmonary fibrosis have at least one affected member carrying a diagnosis of an idiopathic interstitial pneumonia or unclassifiable interstitial pneumonia; other affected family members have interstitial lung disease. Although IPF is the most common diagnosis among the affecteds, it is not the only diagnosis. In fact, only 65% of case subjects with coding mutations in TERT or TERC meet the clinical definition of IPF. Other mutation carriers within the same family have been diagnosed with nonspecific interstitial pneumonitis, granulomatous lung disease, and coal worker pneumoconiosis. Similarly, the occurrence of pathologic findings of diverse subtypes of non–usual interstitial pneumonitis in the same family has been reported for other cohorts of familial pulmonary fibrosis kindreds (30, 31). Granulomatous lung disease, as seen in the proband of family F106, has been associated with telomerase mutations earlier; we previously described one of the affected individuals in family F71 with chronic hypersensitivity whose open lung biopsy showed usual interstitial pneumonitis with features of noncaseating granulomas (2). The mutations in telomerase appear to increase the susceptibility to interstitial lung disease in general and are not associated with one particular clinicopathologic subtype. This raises the possibility that although the diagnosis of a specific interstitial lung disease may differ for individual family members with telomerase mutations due to different environmental or occupational exposures, the identified genetic predisposition leads to a tissue repair response of fibrosis in reaction to injury. We also found that IPF is the most common, but not the only, pulmonary diagnosis seen in those familial and sporadic case subjects without telomerase mutations and whose telomere lengths are less than the 10th percentile. A diagnosis of IPF was found in 50–85% of these groups; other diagnoses such as nonspecific interstitial pneumonitis and cryptogenic organizing pneumonia were made by open lung biopsy. Patients in this study were collected on the basis of a diagnosis of idiopathic interstitial pneumonia, but we again found that telomere shortening was not associated with one particular clinicopathologic subtype. A collected cohort of patients with idiopathic, familial, or anorexigen-associated pulmonary arterial hypertension did not demonstrate an excess of individuals with short telomere lengths, suggesting some specificity of the association between pulmonary fibrosis and short telomeres. Additional cohorts of patients with different pulmonary phenotypes will be needed to delineate the range of pulmonary diagnoses associated with telomere shortening. Although most individuals who carry a heterozygous mutation in TERT or TERC have short telomeres, 15% do not fall below the 10th percentile by the TRFL assay. Analysis of these individuals by the quantitative PCR method also demonstrates that they fall within the normal range (data not shown). In contrast, all of the individuals with diagnoses of pulmonary fibrosis and who carry a heterozygous mutation in TERT or TERC are more than 48 years of age and have telomere lengths below the 10th percentile predicted line. This strongly suggests that the pulmonary fibrosis phenotype is related to older age and short telomere lengths in this molecularly defined group of patients. Leukocyte telomere length may be influenced by other genetic, intrinsic, tissue-specific, or environmental effects. It is currently unknown whether the other mutation carriers have subclinical manifestations of the disease or whether they demonstrate incomplete penetrance. It is also not known whether the nature and degree of telomere shortening seen in circulating leukocytes is representative of the lung cells within these subjects. For eight control subjects and four subjects heterozygous for the TERT P702L mutation, we see a correlation between telomere lengths of DNA isolated from circulating leukocytes and oral buccal epithelial cells (see Figure E4), suggesting that the germline mutations have a global effect on telomere shortening. Because most of the case subjects in this study have been collected or referred from lung transplantation centers, many are severely affected, having demonstrated progressive worsening of disease despite treatment or withdrawal from presumed culprit exposures. An important question concerns whether mutations in telomerase or telomere length offer any prognostic information regarding the natural history of the disease. Although we have found that more subjects with the telomerase mutation had died or undergone transplantation over the course of this study than those without mutations in TERT or TERC (50 vs. 32%), these results are not statistically significant. Similarly, there is a trend in that more familial and sporadic case subjects without telomerase mutations below the 10th percentile had died or undergone lung transplantation than those with telomere lengths greater than the 10th percentile (43–53% vs. 25–31%), but these trends are not significant in both groups. We did not find a significant correlation between telomere length and diffusion capacity measurements for these individuals. Prospective studies and/or analysis of larger cohorts will be needed to determine whether telomerase mutations or telomere shortening are associated with rate of progression or severity of the pulmonary fibrosis phenotype. Smoking is a known risk factor for IPF (32) and for the development of interstitial lung disease in at-risk individuals in kindreds with the familial form of the disease (30, 31). Those with mutations in TERT or TERC had a lower cumulative amount of cigarette smoking than those without mutations, 12.5 versus 19.2 pack-years. It may be that smoking of any level (even small amounts) may increase the risk of developing pulmonary fibrosis in those with an inherited predisposition. Similar results have been found in studies of carriers of a major lung cancer susceptibility locus identified by linkage analysis (33). Cigarette smoking and other environmental effects are likely important modifiers for the development of organ-specific disease in subjects with a global risk of telomerase dysfunction. One of the mutations reported in this study, V694M, was identified in a 60-year-old smoker with no evidence of bone marrow dysfunction and whose 80-year-old mother had a history of pulmonary fibrosis. This same mutation has been reported in a 34-year-old man with moderate aplastic anemia that did not respond to immunosuppression and has short telomere lengths as measured by flow–fluorescence in situ hybridization (8). The development of pulmonary fibrosis versus bone marrow dysfunction in telomerase mutation carriers may be related to secondary "hits," such as environmental toxins (cigarette smoking), fibrosis-prone intrinsic host mesenchymal responses to injury (34, 35), or other susceptibilities. Understanding the influences on the development of lung disease in telomerase mutation carriers will be important to delineate. In epidemiologic studies, IPF is diagnosed more commonly in males than females (1). Of the 20 case subjects with pulmonary fibrosis and telomerase mutations for which we had available DNA, 17 are male. The male-to-female ratio of 5.7:1 in this group suggests a lower penetrance of the pulmonary fibrosis phenotype in females with these dominantly inherited telomerase mutations. Women with TERT mutations and pulmonary fibrosis tend to be, on average, 11.9 years older than their male counterparts (66.0 vs. 54.1 yr for women and men, respectively) and many do not fit the narrow clinical diagnosis of IPF, having apical lung-predominant pulmonary fibrosis. The male-to-female ratio is 1.4:1 and 1.3:1 for the familial and sporadic case subjects without telomerase mutations, respectively. We did observe shorter mean age-adjusted telomere lengths for the men in all groups (Table 4) and found a significant correlation between telomere length and sex after controlling for age and ethnicity by multiple regression. These findings suggest that some of the observed excess of male cases may be explained by their shorter telomere lengths. Both assays used in this study are straightforward and can measure telomere lengths of genomic DNA samples. We included only case subjects for whom we have good-quality DNA isolated from circulating leukocytes. The TRFL assay provides an indirect, rather than a direct, measure of telomere length because the undigested lengths of DNA contain telomeres and subtelomeric segments. To minimize the size of subtelomeric sequences, we digested the genomic DNA with six different 4-bp restriction enzymes. We observed a comparable rate of telomere length attrition with age (17 bp/yr) in normal individuals of this age range, as have other investigators who have used this same method (18–22). The quantitative PCR measurement of telomere length offers an independent method for estimating telomere length from genomic DNA and we found a good correlation between these two methods. It had been previously shown that a quantitative PCR method for estimating telomere length is fast, inexpensive, and requires significantly less DNA than measurement by Southern blotting (13). From the TRFL assay, but not the quantitative PCR assay, we can estimate the percentage of short telomeres in each sample which is the most biologically relevant measure (36). Telomere length is known to be a heritable trait with parental effects (37). Twin studies suggest that telomere size, as measured by the TRFL assay, displays a heritability of 36–78% (22, 38). The quantitative trait of TRFLs has been mapped to various loci in normal subjects (38) or in small families with heart disease (19), but the causative genetic variants within these genomic intervals have not been pinpointed. Understanding the genetic underpinnings of telomere shortening in pulmonary fibrosis may lead to a more complete understanding of how this process contributes to the risk of developing irreversible lung scarring. In addition, such studies may more clearly define the pulmonary as well as other organ phenotypes associated with telomere shortening in aging humans.
The authors thank the affected individuals and their families for their participation in this study; Holly Brookman, Erica Solis, and especially Melissa Nolasco for excellent technical assistance; Dr. Helen Hobbs for the control DNA samples; David Leonard for assistance with the statistical analysis; Dr. Yolanda Mageto, Dr. Craig Glazer, Dr. Borna Mehrad, Dr. Hal Collard, Dr. Edward Garrity, Dr. Keith Meyer, Dr. Varsha Taskar, Dr. Christopher Blewett, Dr. Carlos Girod, Dr. John Fitzgerald, Martha Kingman, and Barbi Estes for patient referrals; and Helen Hobbs, Jonathan Cohen, Jerry Shay, and Woody Wright for helpful discussions.
Supported by National Institutes of Health grant K23-RR020632. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Originally Published in Press as DOI: 10.1164/rccm.200804-550OC on July 17, 2008 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 April 11, 2008; accepted in final form July 17, 2008
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
