Jump to
Abstract
Abstract— Coronary artery disease (CAD) is the leading cause of mortality in the developed world. Although several CAD risk factors, including measures of lipid metabolism, obesity, and blood pressure, have a genetic basis, many genes for CAD susceptibility have yet to be identified. Coronary atherosclerosis is the major cause of CAD, but many with coronary atherosclerosis lack symptoms. Thus, a major limitation of using symptomatic CAD endpoints (eg, sudden coronary death, myocardial infarction) as a study outcome is substantial disease misclassification. Coronary artery calcification (CAC) is part of the atherosclerotic process and is an independent predictor of CAD endpoints. In the present study, CAC was noninvasively quantified by electron beam computed tomography. We performed genome-wide multipoint mode-of-inheritance-free linkage analysis on affected sib pairs, defined as being ≥ the 70th sex- and age-specific percentile for CAC quantity, in a sample of 29 families enriched for hypertension. Almost 95% of participants were asymptomatic for CAD. Our LOD score (log10 odds in favor of linkage) results provide evidence that chromosomal regions 6p21.3 (maximum LOD score=2.22, P=0.00070) and 10q21.3 (maximum LOD score=3.24, P=0.000057) may harbor genes associated with subclinical coronary atherosclerosis.
- atherosclerosis
- coronary artery disease
- coronary artery calcification
- electron beam computed tomography
- genome-wide scan
Atherosclerosis, the major cause of coronary artery disease (CAD), is influenced by the complex interplay of numerous environmental and genetic factors.1 Individual gene effects are thought to be relatively small with many depending on environmental and genetic contexts. Several CAD risk factors have a genetic basis.2–4⇓⇓ Known risk factors for CAD, however, fail to identify a large proportion of individuals with symptomatic CAD endpoints such as sudden coronary death and myocardial infarction (MI).5 A major limitation of using CAD endpoints as a study outcome is substantial disease misclassification, because many individuals with coronary atherosclerosis are asymptomatic. One half of sudden coronary deaths and one half of first myocardial infarctions occur in persons without previous symptoms.6
See page 359
Coronary artery calcification (CAC), a marker of atherosclerosis, can be quantified noninvasively and accurately by electron beam computed tomography (EBCT). CAC is an active and regulated process similar to bone mineralization.7 A direct relationship exists between CAC and both histologic and in vivo intravascular ultrasound measures of atherosclerotic plaque.8
CAC quantity is an independent predictor of angiographically defined CAD.9,10⇓ Debiased estimates for sensitivity and specificity to detect ≥50% stenosis are 97% and 72%, respectively.10 The inter- and intra-observer reliability for CAC measured by EBCT exceed 99% for all arteries combined.11 CAC predicts future CAD endpoints in asymptomatic and symptomatic adults.12,13⇓ Although many known CAD risk factors such as male sex, older age, smoking, abnormal lipid levels, high blood pressure, and ponderosity are related to CAC quantity, much variation in CAC quantity remains unexplained after accounting for these measures.14–18⇓⇓⇓⇓ Noise or artifact in EBCT measures of CAC quantity accounts for only a small amount of variability unexplained by CAD risk factors.14
Wagenknecht et al19 observed that CAC quantity clusters in families enriched for type 2 diabetes mellitus, independent of other risk factors, with an estimated heritability of 0.50. Apolipoprotein E genotype was found to influence the relationship between CAC presence and risk factors.20 Ellsworth et al21 reported a significant association between the S128R polymorphism of the E-selectin gene and presence of CAC in asymptomatic women 50 years old or younger. Pfohl et al22 reported an association between the insertion/deletion polymorphism of the angiotensin I-converting enzyme (ACE) gene and presence and quantity of CAC. Evidence for an interaction effect between paraoxonase-1 genotype and paraoxonase-2, and methylenetetrahydrofolate reductase genotypes has been reported.23 Despite these studies, the genetic basis of CAC is largely unknown.
No studies to date have reported on genome scans of CAC. We present here results from the first linkage genome scan searching for genes associated with CAC.
Methods
Subjects
Siblings in this study were participants in the Genetic Epidemiology Network of Arteriopathy (GENOA) study that began in 1995.24 Eligibility criteria for sibships included: at least two siblings with essential hypertension clinically diagnosed before age 60 years. All hypertensive and normotensive siblings in these sibships were eligible to participate. All GENOA participants were genotyped. Two hundred ninety-eight individuals (150 women) from 105 GENOA sibships (consisting of 375 individuals) also participated in the Epidemiology of Coronary Artery Calcification (ECAC) study17 and had EBCT examinations for CAC between 1992 and 1999, either before or during their participation in the GENOA study. Eligibility for the EBCT examination included not being pregnant or lactating and not having a history of coronary or noncoronary heart surgery. Because hypertension is positively associated with increased CAC levels,25 this study group is likely enriched for high levels of CAC. All siblings were white and from the Rochester, Minn, area. The final linkage analysis sample consisted of those sibships in which at least two siblings met our affected status criteria (see statistical analysis section). This requirement resulted in 29 sibships (94 individuals). Parental phenotype and genotype information were unavailable. Mayo Clinic and University of Michigan Institutional Review Boards approved study protocols, and participants gave written, informed consent after study procedures were explained.
Genotyping
A total of 370 highly polymorphic autosomal microsatellite markers (CHLC/Weber screening set 8.0) were genotyped by using standard polymerase chain reaction-based methods. Marker order was provided by the Center for Genetics at Marshfield Medical Research Foundation (www.marshmed.org/genetics). Distances between markers were based on families from the Center d’Etude du Polymorphisme Humain.26 Allele frequencies were estimated from the total GENOA sample.
EBCT
CAC was measured with a C-100 or C-150 EBCT scanner (Imatron Inc). A scan run consisted of 40 contiguous, 3-mm-thick transverse 2-dimensional image tomograms obtained from the level of the right branch of the pulmonary artery to the apex of the heart. A CAC focus was defined as an area of at least four adjacent pixels with CT number above 130 Hounsfield units in an epicardial artery. A score for each focus of CAC was calculated by multiplying the focus area (in mm2) by a density measure defined by the peak CT number in the focus as described by Agatston et al.27 Total CAC score was calculated as the sum of scores for all foci in the epicardial arteries.27
Statistical Analysis
We performed qualitative mode-of-inheritance-free linkage analysis. Affection status was defined by using sex- and age-specific CAC score percentiles estimated with the method of Royston.28 This method was applied to the first 1219 asymptomatic, white participants in the ECAC study who were selected independently of risk factor status,10 and coefficients from the model were then applied to data for the current study group. The age-specific 50th, 70th, and 90th percentiles are shown in Figure 1 for 623 women and in Figure 2 for 596 men in the ECAC study. (Also, see Tables I and II, http://atvb.ahajournals.org.) Individuals in the present study with CAC and an estimated sex- and age-specific percentile ≥70, which was chosen a priori to the linkage analysis (see Discussion), were considered affected. All others were considered unaffected.
Figure 1. The 50th, 70th, and 90th percentiles for coronary artery calcification score based on 623 asymptomatic, white women from the Epidemiology of Coronary Artery Calcification Study. The 50th percentile score is zero until age 63 years, the 70th percentile score is zero until age 55 years, and the 90th percentile score is zero until age 44 years.
Figure 2. The 50th, 70th and 90th percentiles for coronary artery calcification score based on 596 asymptomatic, white men from the Epidemiology of Coronary Artery Calcification Study. The 50th percentile score is zero until age 48 years, the 70th percentile score is zero until age 40 years, and the 90th percentile score is zero until age 28 years.
Multipoint linkage analysis was conducted with the maximum LOD score (MLS) method for affected sib pairs where LOD is log10 odds in favor of linkage.29,30⇓ Results are reported in terms of MLS values and their associated P values. Parameter optimizations for these analyses were restricted by using Holmans possible triangle model.31 Weighted multipoint MLS values were calculated by using GENEHUNTER version 2.0β.32
Because the maximum MLS value on chromosome 10 was achieved on the boundary of the parameter space for Z0, Z1, and Z2 (the estimated probability that two affected siblings share 0, 1, or 2 alleles identical by descent, respectively), there was concern about the accuracy of the MLS estimate at this location. We simulated 100 000 random replicate data sets on chromosome 10 using the method of gene dropping.33 The MLS value at position 91.8 cm in each random replicate data set was computed and then compared with the test statistic, the observed MLS value at 91.8 cm. An empirical P value was calculated and the corresponding 95% confidence interval (CI) constructed. Simulation studies were also performed to estimate the probability of observing our two largest MLS values in a genome-wide linkage scan in the absence of linkage. These genome-wide significance estimates were based on 10 000 random replicate data sets covering all 22 chromosomes.
Results
Twenty sibships had two affected siblings, 8 sibships had 3 affected siblings, and 1 sibship had 6 affected siblings. There were 70 affected siblings and 24 unaffected siblings in these 29 sibships, and only 5 participants had a history of MI. Selected sample descriptives for those classified as affected and unaffected are in Table 1. Genotype information was included on an additional 16 siblings in these sibships who were not examined with EBCT but were genotyped as part of the GENOA study. For these 16 individuals, their phenotypes were recorded as unknown.
Sample Descriptives by Affection Status for 94 Participants with CAC Data
The number of markers ranged from 30 on chromosome 1 to 6 on chromosomes 21 and 22. The average distance between markers ranged from 8.3 cm on chromosome 17 to 10.4 cm on chromosome 11. The maximum distance between adjacent markers ranged from 14 cm on chromosomes 11 and 14 to 26 cm on chromosome 9.
MLS results are illustrated in Figure 3. The largest multipoint MLS values, the corresponding parameter estimates for Z0, Z1, Z2 and location are presented for each chromosome in Table 2. A MLS value of 2.22 (P=0.00070) was observed on chromosome 6 at map position 76.4 cm (Figure 4) between markers D6S1053 and D6S1031. The overall largest MLS value was 3.24 (P=0.000057) at 91.8 cm between markers D10S1432 and D10S2327 on chromosome 10 (Figure 5). This MLS value was achieved on the boundary of the parameter space for Z0, Z1, and Z2 (Zˆ0= 0.00, Zˆ1= 0.50, and Zˆ2= 0.50). An empirical P value of 0.00012 (95% CI, 0.00010, 0.00014) was obtained at 91.8 cm based on a simulation study.
Figure 3. Multipoint MLS curves for chromosomes 1 to 22. The strongest evidence for linkage was observed on chromosomes 6 and 10.
Largest Multipoint MLS Value and Associated P Value, by Chromosome
Figure 4. Multipoint MLS curve for chromosome 6. Marker names and positions are given at the top of the graph. The peak of the MLS curve localized between D6S1053 and D6S1031 at map position 76.4 cm.
Figure 5. Multipoint MLS curve for chromosome 10. Marker names and positions are given at the top of the graph. The peak of the MLS curve localized between D10S1432 and D10S2327 at map position 91.8 cm.
Removing the 1 sibship with 6 affected siblings and repeating the genome wide linkage scan did not change the inferences. Excluding this sibship resulted in a slightly increased MLS value (P=0.00054) on chromosome 6 and a slightly decreased MLS value (P=0.00032) on chromosome 10 with the maximum likelihood estimates remaining on the boundary.
Four other MLS values greater than 1.0 were observed by using all 29 sibships. These were observed on chromosomes 2 (MLS=1.16 at 114.4 cm), 14 (MLS=1.16 at 95.4 cm), 15 (MLS=1.10 at 8.0 cm) and 17 (MLS=1.13 at 29.4 cm).
We calculated empirical genome-wide significance estimates of our largest MLS values. We estimated the probability (95% CI) of observing MLS values at least as large as 2.22 and 3.24 anywhere along the 22 chromosomes by chance alone to be 0.074 (0.067, 0.085) and 0.0090 (0.0064, 0.012), respectively.
Discussion
Our strategy for linkage analyses was dictated by the distribution of the CAC measurements. Approximately one third of the 298 participants did not have measurable CAC. Thus, standard linkage methods for quantitative traits, which are based on linear regression or variance components techniques, were inappropriate because such methods are highly sensitive to departures from normality in the distribution of the quantitative trait. Risch and Zhang34 advocate the use of highly discordant sib pairs in mapping quantitative trait loci. Ascertaining discordant sib pairs, with one sibling in the highest decile and one sibling in the lowest decile, is very difficult; we had very few such pairings. In fact, the large numbers of individuals with undetectable CAC made it very difficult to identify percentiles in the lower tail of the percentile distribution. Furthermore, the assumption that use of discordant sib pairs increases power to detect linkage may be suspect. Wagenknecht et al19 provided evidence that factors associated with presence and absence of CAC differ from those associated with quantity of CAC in those with measurable CAC. Should dissimilar mechanisms for CAC development and the extent of CAC quantity exist, including sib pairs discordant for their percentile scores could decrease the power to detect linkage.
We chose to focus on individuals above the 70th percentile of the CAC score distribution, thus selecting individuals in whom the CAC process had taken place for a prolonged period. The 70th percentile identifies those considered to be in the highest risk group for a future event. According to the Third Report of the National Cholesterol Education Program,35 anyone with a 10-year risk of CAD >20% is in the highest risk group. Raggi et al36 present estimated annual risks for fatal and non-fatal MI for different CAC percentiles. Those at or above the 70th percentile clearly meet National Cholesterol Education Program criteria for being at highest risk for an event. In our ECAC study, the 70th percentile cutoff was statistically significantly associated with occurrence of CAD events after an average of five years, after adjusting for Framingham risk (OR, 2.8; 95% CI, 1.2 to 6.4) (P.A. Peyser, unpublished data, 2001). Hoff et al37 recently reported that the 75th percentile is a very sensitive cutoff point for identifying subjects at greatest risk. The inferences from our study were the same with either the 60th or 80th percentile (data not shown).
A region on chromosome 10 is significantly linked and a region on chromosome 6 is suggestive of linkage to genes that influence CAC differences in our study group of individuals at high risk of hypertension and CAD. Although hypertension promotes atherosclerosis in the aorta and other arteries, the linkage findings apply only to detectable calcification in the coronary arteries. Candidate genes for CAC include genes involved in hypertension, the immune system, and bone mineralization processes. Candidate genes within the region of interest on chromosome 6 include collagen type XI α2 (OMIM #120290) and allograft inflammatory factor 1 (OMIM #601833). Candidate genes within the region of interest on chromosome 10 include collagen type XIII α1 (OMIM #120350) and bone morphogenetic protein receptor type 1A (OMIM #601299). Collagen plays a role in atherosclerosis by forming a fibrous cap around the lipid core and contributes to vascular stiffness. Allograft inflammatory factor 1 represents a cytokine-inducible, tissue-specific, and highly conserved transcript transiently expressed in response to vascular trauma.38 Bone morphogenetic protein receptors are involved in bone formation,39 and their associated proteins are expressed in calcified human atherosclerotic plaque.40 A candidate gene within the peak region on chromosome 17p12 is a gene linked to abdominal obesity-metabolic syndrome. This gene may be associated with leptin levels,41 which have been shown to influence the calcification of vascular cells.42 We found no evidence for linkage (MLS values <0.5) in the chromosomal regions for apolipoprotein E, E-selectin, angiotensin converting enzyme, paraoxonase-1, paraoxonase-2, or methylenetetrahydrofolate reductase, which are candidate genes for CAC.20–23⇓⇓⇓
Hypertension is independently associated with CAC after adjusting for other CAD risk factors.25 The magnitude of our linkage results, the modest number of sibships used in our analyses, and the high prevalence of hypertension may suggest that hypertension directly or indirectly increases the penetrance of genes influencing CAC. Interestingly, only modest evidence for linkage for hypertension status has been detected by using GENOA sibships in an unpublished genome-wide linkage scan (S.L. Kardia, unpublished data, 2001). Xu et al43 reported evidence of linkage for hypertension on chromosomes 3, 11, 15, 16, and 17 from a genome-wide scan in a large sample of Chinese hypertensive sib pairs. Our linkage peak for CAC on chromosome 17 (MLS=1.17) was observed in a very similar location (within 9 cm) to the peak results on chromosome 17 (MLS=2.16) from Xu et al.43 Krushkal et al44 observed evidence for linkage to genes that influence inter-individual systolic blood pressure variation on chromosomes 2, 5, 6, and 15. Although we observed suggestive linkage results on chromosomes 2, 6, and 15, the location of our peaks did not overlap described regions of Krushkal et al.44 Our peak region on chromosome 2 did, however, overlap the marker D2S1790 (MLS=1.01 at D2S1790), which was found to be significantly linked to diastolic blood pressure in two independent studies.4,45⇓
Studying a measure of subclinical atherosclerosis can identify genes acting through pathways of measurable atherosclerosis risk factors or through novel pathways that have not or cannot be directly measured in vivo. Our genome-wide linkage results provide regions of focus for future genetic studies of subclinical atherosclerosis. We identified two regions with stronger evidence of linkage and several regions with weaker evidence of linkage that require verification and additional testing. The CAC process has a complex etiology, likely influenced by the interaction of numerous environmental and genetic factors. Our findings, particularly the linkage result on chromosome 10, suggest a potentially strong genetic component for a high CAC percentile in individuals at increased risk for hypertension. Identification of genes that contribute to the CAC process should provide a better understanding of the origin of CAC and could ultimately establish a basis for improved prevention and treatment of asymptomatic coronary atherosclerosis.
Acknowledgments
This study was supported by United States National Institutes of Health grants R01 HL46292, U01 HL054481, U01 HL054464, U01 HL054457, and by General Clinical Research Center Grant M01 RR00585 also from the National Institutes of Health. L.A.L. was partially supported by a University of Michigan Regents Fellowship and a University of Michigan Rackham Dissertation Fellowship.
Footnotes
Received December 17, 2001; revision accepted January 14, 2002.
References
This Issue
Jump to
Article Tools
Share this Article
- Autosomal Genome-Wide Scan for Coronary Artery Calcification Loci in Sibships at High Risk for Hypertension
