Published ahead of print on April 7, 2004, doi:10.1164/rccm.200304-493OC
© 2004 American Thoracic Society
Whole Genome Scan for Obstructive Sleep Apnea and Obesity in African-American FamiliesWestern Australian Institute for Medical Research, Center for Medical Research, University of Western Australia, Perth, Australia; Department of Epidemiology and Biostatistics and Department of Pediatrics, Case Western Reserve University, and Rainbow Babies and Children's Hospital, Cleveland, Ohio; and Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, Massachusetts Correspondence and requests for reprints should be addressed to Lyle J. Palmer, Ph.D., Western Australian Institute for Medical Research, QE-II Medical Centre, B Block, Hospital Avenue, Nedlands 6009, Australia. E-mail: lyle{at}cyllene.uwa.edu.au
Obstructive sleep apnea (OSA) is a common, chronic disease associated with obesity. OSA and obesity are both prevalent in African Americans, who are also at increased risk for secondary complications. To identify susceptibility loci for OSA, we undertook a 9-centimorgans genome scan in 59 African-American pedigrees ascertained on the basis of either an affected individual with laboratory-confirmed disease or a proband who was a neighborhood control subject. Variance component linkage analysis was performed for the quantitative phenotypes apnea–hypopnea index (AHI) and body mass index. A candidate region on chromosome 8q (logarithm of odds [LOD] = 1.29, p = 0.006) gave the only evidence for linkage to the AHI. Body mass index was linked to multiple regions, most significantly to markers on chromosome 4q (LOD = 2.63, p = 0.0006) and 8q (LOD = 2.56, p = 0.0007). Evidence of linkage to the AHI was only slightly reduced after adjustment for body mass index. After adjustment for the AHI, some of the primary linkages to body mass index were greatly reduced whereas others remained suggestive. Our results suggest that there are both shared and unshared genetic factors underlying susceptibility to OSA and obesity, and that the genetic determinants of obesity in this population may be modulated by apnea severity.
Key Words: body mass index • genetic epidemiology • genome scan • obstructive sleep apnea • quantitative Obstructive sleep apnea (OSA) is a common disorder characterized by recurrent episodes of apnea (no airflow) and hypopnea (partially obstructed airflow) that occur during sleep, and is associated with oxygen desaturation, sleep fragmentation, and with symptoms of disruptive snoring and daytime sleepiness. Diagnosis is based on standard clinical criteria, and is generally validated by an overnight sleep study with measurement of the apnea–hypopnea index (AHI), the number of apneas and hypopneas per hour of sleep (1). OSA is associated with substantial comorbidity, including obesity, hypertension, diabetes, and cardiovascular disease (2, 3), highlighting its broad public health importance. OSA affects at least 2–4% of middle-aged adults (4), 2% of middle-school children (5), and more than 10% of the elderly. The prevalence of OSA may be particularly high among certain ethnic groups, such as African Americans, who also may be at increased liability for chronic cardiovascular morbidities and cardiovascular mortality. OSA appears to present at a younger age in African Americans than European Americans (6), and may also be more severe (7, 8). Conditions closely related to OSA such as obesity, hypertension, and Type 2 diabetes also occur more frequently in African Americans than European Americans (9–11). Hypertensive end organ disease (kidney and cerebral vascular disease) and cardiovascular mortality are two- to threefold more prevalent among African Americans than European Americans (11, 12). The increased severity of hypertension and associated comorbidities among African Americans may be due in part to unrecognized OSA (13). OSA and its clinical concomitants may thus represent a significantly greater public health burden in African-American populations, indicating that studies of the epidemiology and genetics of OSA in this ethnic group are an important priority. Familial aggregation of OSA has been demonstrated previously (14–17). Clinically apparent OSA likely results from multiple interacting genetic and environmental factors (18, 19). Obesity is the most characteristic feature of OSA in both European-American and African-American adults, and is most commonly measured by an elevated body mass index (BMI) (18). However, the nature of the causal pathways involved in the relationship between BMI and AHI is uncertain (14), and it may be that genetic factors predisposing to OSA differ qualitatively or quantitatively by ethnic group. To identify susceptibility loci for OSA that may be unique to African Americans, we conducted a genome-wide scan of 59 pedigrees of African-American heritage comprising 277 subjects. In these families, we have previously demonstrated increased risk of disease and elevated levels of OSA-associated pathophysiological traits among first-degree relatives of OSA probands (15, 16). As the mode of inheritance of OSA is complex and AHI and BMI are highly correlated, we conducted a model-free linkage analysis of both AHI and BMI. Given the close association of obesity and OSA, we performed linkage analysis of the AHI with and without BMI adjustment and vice versa.
Families The data analyzed are a sample from the Cleveland Family Study, a cohort assembled and monitored longitudinally to study the genetic epidemiology of OSA. This cohort includes 153 African-American pedigrees (n = 1,210 subjects); methods of recruitment and phenotypic assessment have been previously described (6, 15, 17) (see the online supplement). Families were selected for a whole genome scan on the basis of expected linkage informativity, as described previously (20). A total of 277 subjects representing 59 pedigrees were studied in the genome screen linkage analysis (mean pedigree size, 4.7; range, 2 to 10). Participants gave written informed consent; all protocols were approved by the institutional review boards of local hospitals.
Questionnaire
In-home Sleep Studies: Polysomnography
Anthropometry
Genotyping and Data Management
Statistical Analysis Linkage analyses of the genome scan data were performed using model-free methods that made no assumptions about the underlying genetic model for our quantitative outcomes. Variance component linkage analyses were performed using Sequential Oligogenic Linkage Analysis Routines (SOLAR) software package version 1.7.4 (Southwest Foundation for Biomedical Research, San Antonio, TX) (22), as previously described (20). SOLAR made use of all available genotypic and phenotypic information present for all members of the pedigrees. SOLAR uses maximum likelihood methods to partition the observed phenotypic variance for a given trait into (1) additive genetic variance attributable to an unobserved quantitative trait locus linked to a genotyped marker, (2) a residual polygenic component of variance, and (3) nongenetic components of variance (20, 22).
Multipoint analyses were based on a regression approach that yields a weighted average of the IBD (identity by descent) probabilities at 1-cM intervals across each chromosome (23). The narrow-sense heritability Covariates considered for inclusion in the variance components analyses were as follows: age, sex, history of any surgery that may modify upper airway patency (uvulopalatopharyngoplasty, tonsillectomy, or nasal septal surgery), self-reported alcohol consumption, pack-years of smoking, and height, together with polynomial and interaction terms. In certain analyses, the natural log of the AHI (ln[AHI]) or BMI were modeled as covariates. Covariates that were significant at p < 0.05 were retained in the models. The null hypothesis of no linkage at a specific chromosomal location was tested by comparison of a polygenic model with a model with genetic variance components for both a quantitative trait locus and polygenic factors. The difference between the two log10 likelihood values for these two models corresponds to a logarithm of odds (LOD) score for linkage. To avoid overinterpreting the significance of our results, simulations were performed with SOLAR to assess the statistical significance of the linkage results by estimating empirical p values. None of the pointwise p values found reached genome-wide significance at the 5% level, but are reported here to minimize type II error.
Demographics of OSA Families The characteristics of the study population (n = 277) included in the genome scan linkage analyses are presented in Table 1 . The extreme values for mean AHI and BMI in the probands are evident, as is the predominance of males (Table 1). Consistent with previous reports that the risk of developing OSA increases with age (18, 25), parents had higher mean AHI values than did siblings, and siblings had higher mean AHI values than did children of probands. There were more females (56.7%) than males in the study population as a whole.
Phenotypic Modeling Variance component analyses using FISHER (26) and SOLAR, adjusted for all important covariates, suggested that the narrow-sense heritability  of the ln(AHI) levels was 32.3% (SE = 11.7%), that is, additive genetic effects  contributed just under one-third of the total trait variance. The heritability was significantly greater than zero (p = 0.002). The variance component modeling suggested that sex (p < 0.0001) and age (p < 0.0000001) were significantly associated with ln(AHI), which was higher in males and in older subjects. No other potential covariate was significantly associated with ln(AHI). The h2N of BMI was 53.7% (SE = 10.6%) and was significantly greater than zero (p = 0.0000001). Linear (p = 0.0000006) and quadratic (p < 0.0000001) terms for age were highly significant for BMI; levels were higher in older subjects. None of the other potential covariates was significantly associated with BMI. ln(AHI) and BMI were significantly correlated with each other (p < 0.000001). The h2N of the ln(AHI) levels after inclusion of BMI as an additional covariate was reduced to 23.7% (SE = 11.8%); the h2N of BMI after inclusion of ln(AHI) as an additional covariate was reduced to 40.4% (SE = 12.7%). These associations were independent of the potential covariates modeled, and the covariance model ensured appropriate allowance for familial correlations.
Genome-wide Multipoint Variance Component Linkage Analysis
For ln(AHI), one multipoint LOD score above 1.0 was found on chromosome 8q24 (LOD = 1.29, empirical p = 0.006, 137 cM from the end of the short arm of the chromosome (pter) flanked by markers D8S1179 and D8S1128) (Table 2). After adjustment for BMI this LOD was reduced modestly (to 1.09; see Table 2), suggesting that this potential linkage was to gene(s) modulating AHI largely independently of BMI. For BMI, nine multipoint LOD scores above 1.0 were found on seven chromosomes. The highest LOD scores were 2.63 (empirical p = 0.0006) on chromosome 4q23 (105 cM from pter, at marker D4S1647), 2.56 (empirical p = 0.0007) on chromosome 8q21 (104 cM, at marker GATA8B01), and 2.08 (empirical p = 0.002) on chromosome 10q26 (171 cM, at marker D10S212) (Table 2). After adjustment for ln(AHI), five of these linkages were greatly reduced (LOD < 0.85), suggesting that any susceptibility loci in these regions modulating BMI may also act through a pathway involving AHI (Table 2). However, three linkages on chromosomes 2, 5, and 12 were only slightly reduced, suggesting that these potential linkages were to gene(s) modulating BMI largely independently of AHI (Table 2). The linkage to BMI on chromosome 8q was reduced by 63% after inclusion of ln(AHI) as a covariate but remained suggestive (LOD = 1.61) (Table 2), suggesting that any susceptibility genes in these regions may have both direct and indirect (via a pathway involving AHI) effects on BMI.
We have conducted the first whole genome scan for OSA in African-American families, and have found suggestive evidence of linkage to several chromosomal regions. We chose to present these results separately from our initial genome scan for AHI and BMI in European-American families (20) because the study populations, the results, and the implications of the two genome scans are different. As noted in our introduction, there are important clinical and epidemiological differences between African-American and other populations regarding OSA and obesity. Although relatively little is known about OSA in non-European populations, emerging data from the United States suggest that both old and young African Americans have higher levels of AHI than European Americans (6, 8, 17). Data from non-U.S. studies also provide evidence of ethnic differences in OSA prevalence and severity, as well as differences in the intermediate traits associated with OSA (27). The importance of understanding the genetic determinants of OSA in African-American populations is underscored by this group's high prevalence of obesity and OSA-associated comorbidities (13, 28), and by its different patterns of epidemiological risk factors for OSA compared with other ethnic groups. Obesity is more prevalent and is more epidemic in African Americans than in European Americans, with the greatest ethnic differences observed for females (29). Markers of subclinical cardiac disease in African Americans seem to be linked specifically to untreated OSA (30, 31). Thus, unrecognized and endemic OSA may be an important underlying risk factor in the genesis of chronic health conditions in the African-American population. As discussed below, there are also important differences in the results of the genome scans in these two populations. We initially performed extensive preliminary analyses and a formal investigation of heterogeneity among the African-American and European-American samples with a view to possibly pooling these data sets. Differences in allele frequencies and heterogeneity in linkage signal indicated the inability to validly pool data from each ethnic group. We were further motivated by the fact that there have been few whole genome scans for any complex human diseases in African-American populations, and none at all for sleep apnea phenotypes.
The importance of familial determinants in the regulation of BMI in humans is well established (32). A previous study of Nigerian, Jamaican, and African-American families found that the heritability of adult BMI ranged from 48 to 58% (33), consistent with our results A broad chromosomal region on chromosome 8q demonstrated some evidence of linkage to both ln(AHI) and BMI (Table 2). Interestingly, the linkages in this region to both ln(AHI) and BMI remained above 1.0 after adjustment for each other (Table 2). The highest observed multipoint LOD score for ln(AHI) was 1.29 at 8q24. This result is not significant at a genome-wide level, but indicates a linkage of potential interest in the context of a complex, multifactorial disease (34). The 8q region also gave promising evidence of linkage to BMI (LOD = 2.56, 8q21). Given the genetic distance between the linkages to AHI and BMI on chromosome 8q (about 32 cM), these linkage peaks are likely to represent two or more distinct loci. Alternatively, the chromosome 8q linkages may represent a single pleiotropic locus affecting both AHI and BMI independently, with additional effects on BMI modulated through the effects on AHI (Figure 2). The lack of correspondence between the two linkage peaks may simply reflect the imprecision of the AHI measure relative to BMI. Many regions throughout the human genome have been previously linked to obesity-related phenotypes (35). Evidence of linkage of obesity-related phenotypes to the three most promising regions for BMI in the current study, chromosomes 4q, 8q, and 10q, has been previously reported: the chromosome 4q23 region to abdominal subcutaneous fat and trunk-to-extremity skinfold ratio (36), 8q21 to BMI and leptin levels (37, 38), and 10q26 to BMI levels and to measures of abdominal fat (39). There have been two reported whole genome scans for obesity in African-American populations. Zhu and coworkers (40) studied 202 African-American families (n = 618 subjects) selected on the basis of high blood pressure, and found suggestive evidence of linkage to BMI on chromosomes 5p (LOD = 1.9) and 3q (LOD = 1.8). There was virtually no overlap with the linkage results of the current study, suggesting the possibility that the genetic determinants of obesity in families selected on the basis of OSA may differ from those found in families selected for high blood pressure. However, Chen and coworkers (41) reported strong linkage (LOD = 3.4) of 4q23 (between markers D4S1647 and D4S2623) to BMI in a whole genome scan of West African families (n = 691 subjects) selected on the basis of Type 2 diabetes. This result is similar in location to our linkage to BMI, suggesting that the 4q locus may be of particular interest for obesity susceptibility in populations of African origin. It is important for genetic studies to recognize ethnic differences in OSA, and such differences may also shed light on underlying genetic mechanisms for disease. Our study found different linkages from a previous genome scan of OSA-associated phenotypes in a sample of European-American families also drawn from the Cleveland Family Study (20). In that study, candidate regions on chromosomes 1p, 2p, 12p, and 19p gave most evidence for linkage to AHI; BMI was also linked to multiple regions, most significantly to markers on chromosomes 2p, 7p, and 12p (20). These differing results may reflect interpopulation heterogeneity between European-American and African-American populations due to differing environmental exposures, differing genetic determinants of OSA and obesity, or both. Neel (42) has proposed a genetic explanation for the greater prevalence of obesity and diabetes in some ethnic groups. The "thrifty gene" hypothesis suggests that different ethnic groups may have developed diverse molecular mechanisms to adapt to periods of unpredictable food supply (43). Our results suggest marked ethnic genetic heterogeneity for OSA and obesity, and are consistent with this hypothesis. The suggestion of genetic heterogeneity for BMI loci among different populations is consistent with observed differences in obesity patterns and lipid metabolism between African Americans and European Americans (44, 45).
Several biologically plausible candidate genes are located within the most promising chromosomal regions in our analysis. The 8q22 region contains three genes for carbonic anhydrase (CA) isoenzymes: CA1, CA2, and CA3 (46). The roles of CA in modulating respiratory control, and the role of CA inhibitors as potential treatments for conditions with underlying respiratory instability, including sleep periodic breathing and sleep apnea, have been the subjects of numerous animal and human studies (47–49). The chromosome 8q22 region also contains a core-binding factor (runt domain, As noted previously, obesity is a major risk factor for OSA in both European Americans and African Americans (20, 57). The coaggregation of OSA, central obesity, hypertension, and Type 2 diabetes suggests that OSA may be part of a "metabolic" syndrome (58, 59), which may be largely influenced by genes that influence insulin resistance and body fat distribution (60). Candidate genes for obesity are therefore relevant for studies of the genetics of OSA both because of the prominence of obesity in the OSA phenotype, and because of the potential impact of these genes on the expression of other traits of potential relevance to OSA. Previous genetic studies of obesity have not evaluated OSA, which may occur in as many as 66% of obese individuals. However, as our study suggests, the relationship between OSA and obesity may well be bidirectional. Although obesity is a risk factor for OSA, it is also plausible that OSA may increase risk for obesity. For example, OSA causes sleep fragmentation and sleepiness, effects that may promote weight gain via reduced physical activity and hypercytokinemia (61). The results of the phenotypic modeling and our whole genome scan were consistent with the hypothesis that obesity and OSA have both shared and unshared genetic determinants, and were consistent with our previous report of major genetic factors underlying AHI independently of BMI (62). Future linkage studies of obesity should take this potential interrelationship into account. This study has several potential limitations. Our sample, although intensively phenotyped, was of moderate size. Our limited power to detect modest effects may at least partially explain our modest linkages to AHI. Although the maximum LOD scores we observed were modest, this is consistent with our expectations regarding common, complex conditions such as OSA and obesity (63). The generalizability of linkage results in our OSA families to OSA in other populations with different exposures is undetermined. We have included known environmental determinants of OSA, such as alcohol consumption, as potential covariates in our analyses, but we have not formally tested for genotype-by-environment interactions. Although the quantitative trait we analyzed (the AHI) is the most common metric used to identify and quantify OSA, it is possible that more sophisticated indices of overnight breathing and sleep disruption may be more informative. We chose the AHI as the primary phenotype of interest because it is the metric most commonly used in both the clinical sleep field to identify disease and quantify its severity, and also in clinical and epidemiological research studies of sleep apnea. Definition of ethnicity in our study was based on self-report; self-report has proven useful in identifying important differences among ethnic groups in previous clinical, epidemiologic, and genetic studies (6, 64, 65). Finally, we have not adjusted for the multiple phenotypes analyzed, because AHI and BMI are closely correlated. The present study suggests the presence of multiple genetic determinants of the pathophysiological traits associated with OSA. Two of the important phenotypes associated with OSA (AHI and BMI) appear to be distinct traits with both shared and unshared genetic determinants. There are likely multiple genetic determinants of OSA, and these linkage analyses of quantitative phenotypes have identified several regions of interest on which to focus fine mapping efforts. Evaluation of the regions of linkage within our study with additional linkage markers will be required, and replication of our findings in larger samples will be necessary. Investigation of candidate genes in the regions of linkage will also be necessary. If novel genetic determinants of OSA can be identified, important new insights into OSA pathophysiology, and ultimately treatment, could result.
The authors are indebted to the dedicated staff of the Cleveland Family Study, including the superb field team: Kathryn Clark, Gregory Graham, Barbara O'Malia, Sunny Morton, Heather Rosebrock, and study coordinator Joan Aylor. The authors are grateful for the genotyping performed by the NHLBI Mammalian Genotyping Service and are especially thankful for the participation of the members of the OSA families. The authors thank Mr. Kim Carter for assistance with the graphics.
Supported in part by National Institutes of Health grants HL43680, HL07567, T32-HL07567, GM28356, U01 HL 66795, M01 RR00080-39, and U01 HL 66795. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: L.J.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.G.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; E.K.L. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.R.P. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; R.C.E. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; P.V.T. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; S.R. participated in a multicenter clinical trial of levalbuterol that reimbursed cost per child enrolled and the total fee collected for the cost of study was $15,000. Received in original form April 8, 2003; accepted in final form April 2, 2004
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
. 
. The heritability of AHI levels has not previously been estimated in an African-American population, and our results show strong genetic effects underlying the AHI 
. The heritability estimates for BMI and AHI in our African-American sample were consistent with a previous study of European Americans undertaken using the Cleveland Family Study population 
(
subunit) gene (CBFA2T1), which has been previously associated with BMI, percentage of body fat, and waist and hip circumference in a study of Pima Indians (
