Published ahead of print on March 13, 2003, doi:10.1164/rccm.200207-755OC
© 2003 American Thoracic Society
A Genome-Wide Scan of Pulmonary Function Measures in the National Heart, Lung, and Blood Institute Family Heart StudyDepartments of Medicine, Neurology, and Biostatistics, Boston University Schools of Medicine and Public Health, Boston, Massachusetts; Division of Epidemiology, University of Minnesota School of Public Health, Minneapolis, Minnesota; Department of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, North Carolina; Pulmonary Division, LDS Hospital; Central Molecular Laboratory, University of Utah, Salt Lake City, Utah; and Division of Biostatistics, Washington University, St. Louis, Missouri Correspondence and requests for reprints should be addressed to Jemma B. Wilk, D.Sc., Boston University School of Medicine, B-601, 715 Albany Street, Boston, MA 02118. E-mail: jwilk{at}bu.edu
Spirometric measures of pulmonary function exhibited high heritability in the National Heart, Lung, and Blood Institute Family Heart Study. A genome scan of FEV1, FVC, and the ratio of FEV1/FVC was performed to identify chromosomal regions influencing these measures. The pulmonary traits were adjusted through multiple linear regression techniques for the effects of age, age2, body mass index, height, smoking status, and pack-years of smoking. The distribution of FEV1/FVC was transformed to account for nonnormality, and standardized residuals were used as the quantitative trait for variance component linkage analysis in GENEHUNTER (Whitehead Institute, Cambridge, MA). The genome scan identified regions on chromosomes 4 and 18 with logarithm of the odds favoring linkage (LOD) scores above 2.5, and these two chromosomes were further evaluated by incorporating additional marker genotyping. The FEV1/FVC ratio was linked to chromosome 4 around 28 centimorgans (cM; D4S1511) with a LOD score of 3.5, and the transformed ratio was linked to the same region with a LOD of 2.0. FEV1 and FVC were suggestively linked to regions on chromosome 18 with multipoint LOD scores of 2.4 for FEV1 and 1.5 for FVC at 31 cM (D18S843) and a LOD of 2.9 for FVC at 79 cM (D18S858).
Key Words: linkage • spirometry • genome Pulmonary function plays a crucial role in health and quality of life, and measures of pulmonary function are correlated with longevity. In healthy nonsmoking adults, lung function peaks between the ages of 20 to 25 years, remains stable for 5 to 10 years, and then declines with increasing age (1). Pulmonary function measured during middle age predicts subsequent development of chronic obstructive pulmonary disease (COPD) as well as cardiovascular and all-cause mortality (2). COPD results from a decline in pulmonary function as a result of inflammation of the airways (bronchitis and bronchiolitis) and destruction of lung parenchyma (emphysema). In the United States in 1997, COPD had an estimated prevalence of 6.6% of the population and was the fourth leading cause of death (3, 4). FEV1, FVC, and the ratio of FEV1/FVC are measures clinically used to characterize lung function. Several studies have shown evidence for substantial heritability of pulmonary function levels as measured by spirometry (5–7). Segregation analyses have provided evidence for a major gene effect influencing FEV1 in families ascertained through a proband with COPD (8) and a major gene influencing FVC in a population-based family study (9). However, other population-based family studies have shown that FEV1 is influenced by multiple genes and environmental factors (10, 11). The heritability estimates for FEV1, FVC, and the ratio of FEV1/FVC calculated from the Family Heart Study (FHS) randomly ascertained cohort were 0.52, 0.54, and 0.45, respectively, indicating a high heritability of these measures observed in this population. Segregation analysis of the same FHS sample indicated that a dominant major gene model best described the transmission of FEV1, although Mendelian transmission models for FVC could not be statistically differentiated (6). Few genes that influence the normal variation in pulmonary function or functional decline with aging have been identified. Evaluating linkage to quantitative lung function phenotypes is a powerful approach for finding genes influencing pulmonary variability and may be less subject to genetic heterogeneity than linkage to pulmonary disease (12). A genome scan for FEV1, FVC, and FEV1/FVC has been performed in the Framingham Heart Study. The strongest evidence of linkage to FEV1 occurred on chromosome 6 and for FVC on chromosome 21 (7). These linkages are not in regions of known or putative COPD-related genes. A genome scan performed in a cohort of families ascertained for severe early onset COPD identified a region on chromosome 2q (222 centimorgans [cM]) with significant evidence for linkage to the FEV1/FVC ratio (13). Regions on chromosome 1 (120 cM) and chromosome 17 (67 cM) also demonstrated linkage to the FEV1/FVC in the COPD cohort. In this study, we report the results of a genome scan of FEV1, FVC, and the ratio of FEV1/FVC in the NHLBI FHS. The FHS population is ideal for exploring linkage because of the large sample of extended families and evidence for substantial heritability of these traits.
Study participants were from the NHLBI FHS. Details of the design of FHS have been described previously (14). The Mammalian Genotyping Service (MGS) genotyped 393 autosomal microsatellite markers in the 401 largest white families. This analysis was performed using 391 pedigrees with 2,178 individuals that had both MGS genotyping data and spirometric measures, and an additional 702 individuals with genotypic data contributed to the calculation of identity by descent estimates. The University of Utah Central Biochemistry laboratory genotyped an additional 242 markers in sibships with coronary heart disease and a sample of sibships selected for high individual risk scores for coronary heart disease. A total of 1,327 persons were genotyped, and 483 of those had not been previously genotyped by MGS. The additional markers and subjects genotyped in Utah were used to follow up the regions with logarithm of the odds favoring linkage (LOD) scores greater than 2.5 in the MGS data. Markers were analyzed according to MGS published order (15). Spirometry was performed during the participant's clinical examination using a computerized volume-based spirometer and was reported at body temperature and pressure, saturated. FEV1 and FVC were measured. Participants who reported asthma (n = 199) or lung cancer (n = 4) in their personal medical history were excluded from this analysis. FEV1, FVC, and the ratio of FEV1/FVC were adjusted by multiple linear regression, separately for males and females, and standardized residuals (mean = 0, SD = 1) were calculated. The effects of age, age2, body mass index (kg/m2), height, dummy variables indicating cigarette smoking status (never, former, current), and cigarette pack-years for former and current smokers were retained in the final regression models. The covariates accounted for a total of 65%, 60%, and 25% of the variability in FEV1, FVC, and the ratio, respectively. The distribution of each of the traits was evaluated because departures from normality can increase type I error rates and inflate LOD scores obtained by variance component linkage analysis (16). Neither FEV1 nor FVC deviated substantially from a normal distribution. The ratio of FEV1/FVC had a kurtosis of 3.91, and thus, a ranking procedure in SAS (SAS Institute Inc., Cary, NC) was used to derive a normalized deviate of the ratio trait. Linkage analysis was performed on both the untransformed ratio residuals and the normalized deviate. The genome scan was performed using variance component linkage analysis for quantitative traits in the program GENEHUNTER (Whitehead Institute, Cambridge, MA), version 2.1 (17–19). The variance component method separates the total variation in a trait into components due to genetic and environmental effects. Linkage was evaluated by first estimating the likelihood of a polygenic model that allows for multiple genes of small effect. Models incorporating genetic data at a putative quantitative trait locus were then compared with the polygenic model. The LOD score was calculated as the log base 10 of the ratio of the likelihoods of the locus-specific and polygenic models. Multipoint LOD scores were calculated, as incorporating information from adjacent markers helps to reduce the false-positive rate and increase the power to detect true linkages. An additional description of the sample and methods is included in the online supplement.
Table 1 reports the descriptive characteristics of the sample stratified on sex. The mean values of FEV1, FVC, the ratio of FEV1/FVC, percent predicted values for the three pulmonary measures, age, height, body mass index, and pack-years are reported, as well as the proportion of former and current smokers. Males were more likely to have smoked and had higher mean pack-years of smoking than females. The mean values for the pulmonary measures and covariates in this sample for the genome scan are very similar to the mean values for the entire FHS cohort.
Genome-Wide Scan Results To asses type I error for this analysis, we performed a simulation using the variance components linkage analysis software package SOLAR (20). Using the FHS pedigree structure, trait values and genotypes for an unlinked marker were simulated. For individuals with missing pulmonary measures, the corresponding simulated trait data were also made missing. The simulation was based on a heritability of 46%, an average heritability for the pulmonary measures in this sample. In 400 replicates of this simulation, the maximum LOD score observed was 2.42. Only four replicates achieved a LOD score in excess of 1.5. Results of this genome-wide scan are reported where multipoint LOD scores in excess of 1.5 were identified. Figures 1A–1D show the genome-wide multipoint LOD score results for linkage analysis to FEV1, FVC, and both the untransformed and normalized deviate of the ratio of FEV1/FVC, respectively. Positions where LOD scores exceeded 1.5 for any of the pulmonary measures are reported in Table 2 , and the multipoint LOD scores for all measures at those positions are reported. The highest LOD score in the genome-wide scan of pulmonary measures was a 3.49 for the ratio of FEV1/FVC at 29.6 cM (near D4S403) on chromosome 4. The normalized deviate of the ratio trait also demonstrated suggestive evidence of linkage in this region with a LOD of 2.63. The results from the normalized deviate of the ratio, however, shifted the strongest evidence of linkage to 23.3 cM, a position 6.3 cM from the maximum LOD score for the untransformed trait. This region between 23.3 and 29.6 cM contained the highest LOD scores for both the untransformed and normalized ratio phenotypes. The highest multipoint LOD score observed for FEV1 was 2.01 on chromosome 3 at 119.5 cM (between markers D3S4529 and D3S3045). For FVC, the highest multipoint LOD score was 2.74 on chromosome 18 at 77.8 cM (near D18S851).
Results for Analyses Incorporating Utah Markers Chromosomes 4 and 18 were reanalyzed using both the original MGS markers and the marker data genotyped by the Utah laboratory, with the addition of coronary heart disease sib pairs (483 subjects) genotyped only in Utah. The multipoint LOD scores in the regions suggestive of linkage are reported in Table 3 . Incorporating the Utah genotyping data on chromosome 4 added one marker (D4S1511) in the region of linkage at 27.9 cM, between the MGS markers D4S403 at 25.9 cM and GATA70E01 at 35 cM. On chromosome 4, the magnitude of the LOD scores for the untransformed ratio was similar when the additional data were incorporated (3.5 at 29.3 cM), but the LOD score for the normalized deviate of the ratio was reduced by the additional data. The location of the maximum LOD score for the normalized trait shifted 4.6 to 27.9 cM, nearer to the peak observed for the untransformed trait. For the combined data, the distance between the positions of the peak LOD scores for the two definitions of the trait was reduced to 1.4 cM. The LOD score for the normalized ratio in the combined marker set was 2.01 compared with the 2.63 in the original marker set. The 1-LOD score support interval defines the region surrounding the maximum LOD score, bounded by the positions where the LOD scores equal the maximum LOD-1 (21). For the untransformed ratio, this support interval spans from approximately 23–34 cM, and for the normalized ratio, the support interval spans from 18–35 cM. From the variance components results, locus-specific heritability estimates for the ratio and normalized ratio were 24% and 17%, respectively. The multipoint LOD score results for both FEV1/FVC ratio phenotypes on chromosome 4 are shown in Figure 2 .
On chromosome 18, two of the markers genotyped by Utah were the same markers as genotyped in MGS. At markers D18S843 at 28.1 cM and D18S853 at 80.4 cM, the Utah genotyping was added only for individuals who were not previously genotyped at that marker by MGS. The overlapping sample was used to confirm that allele calling between the two laboratories was consistent. The additional genotyping data revealed a LOD score of 2.37 for FEV1 at 31 cM and a LOD score of 1.54 for FVC at 33.9 cM, whereas the original scan gave maximum LOD scores of 1.28 and 1.30 for FEV1 and FVC, respectively, in this region. The 1-LOD score support interval around the maximum LOD for FEV1 spans from approximately 22–38 cM, and for FVC, it spans from 24–46 cM. For FVC, the maximum LOD score on chromosome 18 was 2.88 at 78.7 cM (between D18S851 and D18S858). The additional marker data in this region increased the multipoint LOD score for FEV1 by 0.5 and increased the LOD score for FVC by just 0.14. The 1-LOD score support interval about the FVC locus spans from approximately 65–85 cM. The estimate of locus-specific heritability for the quantitative trait locus influencing FEV1 at 31 cM is 19% and for the quantitative trait locus influencing FVC at 78.7 cM is 16%. The multipoint LOD score results for FEV1 and FVC on chromosome 18 are shown in Figure 3 .
The results of this genome scan in the NHLBI FHS provide evidence for three loci influencing variability in spirometric measures of pulmonary function that meet established criteria for suggestive evidence of linkage (22). On chromosome 4, a locus was identified that was linked to the ratio of FEV1/FVC. The ratio of FEV1/FVC is a measure of the severity of airflow obstruction that is used clinically to identify obstructive pulmonary disease (23). A gene in this region may influence the acquisition of airway obstruction or may influence lung parenchymal elasticity. On chromosome 18, two distinct regions were linked to both FEV1 and FVC, but not to the FEV1/FVC ratio. These loci may have an influence on overall lung size, with proportional effects on airway and parenchymal size. On chromosome 4, evidence of linkage is observed to the FEV1/FVC ratio, a trait whose distribution deviates from the underlying assumption of normality that is made for variance component analysis. To address this non-normality in the distribution, a normalized deviate of the trait was also analyzed, and results with this trait provide suggestive linkage in the same region. The estimate of the locus-specific heritability for the region differs for the two trait distributions, with the original trait showing higher locus-specific heritability (24% compared with 17%). The two phenotypes defined here for the FEV1/FVC ratio represent distinctly different distribution patterns. Although studies have shown that leptokurtic trait distributions inflate LOD scores in variance components analysis, forcing the data to be normally distributed may remove some of the apparent effect of the quantitative trait locus and decrease power to detect a quantitative trait locus (16, 24). By examining the variance components results of both distributions, we identified a region on chromosome 4 where both distributions provide consistent results supporting linkage. The magnitude of the LOD scores generated by the two distributions support different interpretations for the strength of linkage to this region, but finding that the same region contains the highest genome-wide LOD scores for both of the FEV1/FVC ratio traits with LOD scores in excess of 2.0 provides support for a gene in the region around 28 cM on chromosome 4. A genome scan of these same pulmonary traits was performed in participants of the Framingham Heart Study (7). The results of the genome scan identified regions of chromosomes 6 and 21 to be most strongly linked to FEV1 and FVC, respectively. We are unable to confirm these linkages in FHS even though covariate adjustment was similar to that used by Framingham. Multipoint LOD scores in the region of the Framingham linkages did not exceed 1.0. The discrepancy may be due to differences in subject ascertainment and spirometric data collection. FHS families were ascertained at a single time point as opposed to the Framingham families, which were ascertained as two separate cohorts. Different spirometric equipment was used at different points in the Framingham study, and the means of the spirometric measures in FHS more closely resemble those in the Framingham offspring cohort than the original cohort. The similarity between the FHS participants and the offspring cohort of the Framingham study may reflect the association between temporal trends in body habitus, smoking behavior, or employment type, factors that may influence pulmonary function. The consistency in the spirometric equipment used to measure lung function and the relatively short time period during which subjects were ascertained is a strength of this study. The spirometric equipment was consistent across centers, and pulmonary function tests were sent to reading centers for measurement and interpretation (14). Family members who participated in clinical examinations that included spirometry were examined at the same field center. The FHS may have effectively reduced the variability in pulmonary function due to temporal trends and strength ened the ability to elucidate the underlying genetic components. The FHS estimates of the heritability of the pulmonary traits were higher than those obtained by the Framingham study, and FHS provided evidence for a major gene influencing FEV1 through segregation analysis, which was absent in Framingham data (6, 11). The inclusion of families who are at increased risk of coronary heart disease may have enriched the FHS sample for studies of pulmonary function by providing a broader distribution in the phenotypes than would be available from a random sample. Individuals with coronary heart disease may have reduced pulmonary function, and comparison with their unaffected family members provides a powerful discordant pair analysis. Loci identified in this study provided evidence for linkage in both the random and high-risk samples. Stratifying the families on the basis of their ascertainment produced LOD scores of 0.5 or higher in both the random and high-risk groups at the loci on chromosomes 4 and 18 (data not shown). This suggests that the effects on lung function from genes in these regions are not produced by an increased frequency of cardiovascular disease in the high-risk sample. To explore whether the loci identified on chromosomes 4 and 18 were related to COPD, we excluded participants who met a spirometric definition of moderate COPD. Ninety participants were classified as COPD cases because their FEV1 levels were less than 70% of predicted and FEV1/FVC levels were less than 90% of predicted. Removing these individuals from the linkage analysis resulted in lower LOD scores for the ratio phenotypes on chromosome 4. The LOD scores were reduced from 3.5 to 1.6 for the untransformed ratio and from 2.0 to 1.0 for the normalized ratio. The decrease in LOD score suggests that the COPD cases were accounting for up to half of the LOD score observed in the full sample and that a gene in this region may be related to obstructive disease. In contrast, the LOD scores for both loci on chromosome 18 increased with the exclusion of COPD cases. LOD scores of 2.4 for FEV1 at 31 cM and 2.9 for FVC at 79 cM in the full sample were both increased to LOD scores of 4.0. In addition, the LOD score for FEV1 at 79 cM increased from 1.4 to 2.9. These results suggest that genes underlying these loci on chromosome 18 may be related to normal variability of lung function through an influence on growth and development. This genome scan identified a region of linkage that was also reported in the genome scan of early onset COPD patients where linkage to the FEV1/FVC ratio was observed on chromosome 1 with a maximum multipoint LOD of 1.9 at 120 cM (13). In this study, a LOD score of 1.8 was observed with the normalized ratio trait at 123.1 cM and a LOD score of 1.7 with the untransformed ratio trait at 126.2 cM. Although these LOD scores are modest, the replication across studies suggests that further study of this region may be warranted to pursue genes influencing pulmonary obstruction. In summary, FHS found suggestive evidence of linkage to pulmonary traits in three regions. The locus on chromosome 4 in the region of D4S403 and D4S1511 shows linkage to the ratio of FEV1/FVC. Although the linkage evidence was attenuated somewhat with the additional genotyping data, these results suggest that further research in this region is warranted. One candidate gene in the region of this linkage is superoxide dismutase 3. Superoxide dismutase 3 is the main extracellular antioxidant enzyme in the lung and may play a role in attenuating tissue damage due to oxygen radicals. A polymorphism in superoxide dismutase 3 has been identified and was found in approximately 2% of a random population (25). Evaluating an association between superoxide dismutase 3 polymorphisms and the FEV1/FVC ratio may be a worthwhile follow-up to these results. On chromosome 18, two regions were identified to be linked to both FEV1 and FVC, and the evidence for linkage was enhanced after excluding COPD cases from the sample. These loci around 31 cM (near D18S843) and around 79 cM (near D18S851 and D18S858) may harbor genes influencing lung growth and development. At the positions where FEV1 or FVC achieved suggestive linkage in the full sample, the other trait also exhibited evidence of linkage with LOD scores of approximately 1.4–1.5, which strengthens the case for putative genes influencing pulmonary function measures in these regions. Although no obvious candidate genes are present in these regions, these loci deserve further evaluation.
This article is presented on behalf of the investigators of the NHLBI FHS. Participating institutions and principal staff of the study are as follows: Forsyth County/University of North Carolina/Wake Forest University: Gerardo Heiss, Stephen Rich, Greg Evans, James Pankow, H.A. Tyroler, Jeannette T. Bensen, Catherine Paton, Delilah Posey, and Amy Haire; University of Minnesota Field Center: Donna K. Arnett, Aaron R. Folsom, James Peacock, and Greg Feitl; Boston University and Framingham Field Center: R. Curtis Ellison, Richard H. Myers, Larry Atwood, Yuqing Zhang, Luc Djoussé, Jemma B. Wilk, and Greta Lee Splansky; University of Utah Field Center: Steven C. Hunt, Roger R. Williams (deceased), Paul N. Hopkins, Hilary Coon, and Jan Skuppin; Coordinating Center, Washington University, St. Louis: Michael A. Province, D.C. Rao, Ingrid B. Borecki, Yuling Hong, Mary Feitosa, Jeanne Cashman, and Avril Adelman; Central Biochemistry Laboratory, University of Minnesota: John H. Eckfeldt, Catherine Leiendecker Foster, Michael Y. Tsai, and Greg Rynders; Central Molecular Laboratory, University of Utah: Mark F. Leppert, Jean-Marc Lalouel, Tena Varvil, and Lisa Baird; National Heart, Lung, and Blood Institute—Project Office: Phyliss Sholinsky, Millicent Higgins (retired), Jacob Keller (retired), Sarah Knox, and Lorraine Silsbee.
Supported partially by the National Heart, Lung, and Blood Institute cooperative agreement grants U01 HL56563, U01HL56564, U01 HL56565, U01 HL56566, U01HL56567, U01 HL56568, and U01 HL56569. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Received in original form July 26, 2002; accepted in final form March 7, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
