Impact of Inherited Genetic Variants Associated With Lipid Profile, Hypertension, and Coronary Artery Disease on the Risk of Intracranial and Abdominal Aortic AneurysmsClinical Perspective
Background—Epidemiological studies show that an unfavorable lipid profile and coronary artery disease (CAD) are risk traits for abdominal aortic aneurysms (AAAs) but not for intracranial aneurysms (IAs), and that hypertension is a main risk trait for IAs but not for AAAs. To evaluate these observations, we investigated single-nucleotide polymorphisms associated with serum lipid levels, hypertension, and CAD and tested their contribution to AAA and IA risk.
Methods and Results—We defined sets of single-nucleotide polymorphisms previously reported to be associated with serum lipid levels, CAD, and blood pressure. From previously collected genome-wide data, we extracted genotypes for these single-nucleotide polymorphism sets in 709 IA cases and 2692 controls and 807 AAA cases and 1905 controls (all of Dutch origin). We computed genetic scores for each individual by summing the observed number of risk alleles weighted by their previously published effect size. Using logistic regression, we tested the genetic scores for association with IAs and AAAs and found significant associations for genetic scores of total cholesterol (P=3.6×10-6), low-density lipoprotein-cholesterol (P=5.7×10-7), and CAD (P=0.0014) with AAAs and for the blood pressure score with IAs (P=0.0030).
Conclusions—We demonstrate that genetic risk profiles of lipid factors and CAD are associated with AAAs but not with IAs, and the genetic risk profile of blood pressure is associated with IAs but not with AAAs. These findings are consistent with epidemiological observations.
Intracranial aneurysms (IAs) and abdominal aortic aneurysms (AAAs) share a pathophysiological background (including upregulation of proteolytic pathways, inflammation, and loss of arterial wall matrix)1 and epidemiological risk factors, such as age and smoking.2,3 For both diseases, family history is a main risk factor, indicating that inherited genetic variation influences disease risk.2,3 The disease risk for first-degree relatives of patients is up to 8 times increased in AAAs4,5 and up to 7 times in IAs.6 Co-occurrence of the 2 diseases has been described, mainly within affected families, suggesting a shared genetic basis for these diseases.7
Recent genome-wide association studies (GWAS) have highlighted a role for a risk allele at locus 9p21.3 near CDKN2A and CDKN2B in both IAs and AAAs.8–10 In addition, risk alleles near the genes STARD13-KL, RBBP8, SOX17, CNNM2, and EDNRA are associated with IAs,8,9,11 whereas risk alleles near DAB2IP and LRP1 are associated with AAAs.12,13
Clinical Perspective on p 270
Epidemiological studies showed that high total cholesterol (TC), low-density lipoprotein-cholesterol (LDL-C), triglycerides, low high-density lipoprotein-cholesterol (HDL-C) levels, and coronary artery disease (CAD) are risk factors for AAAs.14–19 In contrast, hypercholesterolemia is not associated with increased risk of IAs,2,20 and no significant comorbidity of CAD and IAs has been observed, according to one study.20 Similarly, hypertension is a major risk factor for IA2 but is only weakly associated with AAAs.18
In this study, we test the effect of genetic variants associated with lipids, CAD, and blood pressure (BP) on IAs and AAA disease risk. This genetic design has some advantages compared with epidemiological studies, which may be more susceptible to confounding and reverse causation.21 We hypothesize that the genetic risk profiles of lipid factors and CAD are associated with AAAs but not with IAs, and the genetic risk profile of BP is associated with IAs but not with AAAs. In the future, this may help to identify additional genes or pathways in aneurysm pathogenesis.
We used data from Dutch subjects genotyped in previous GWAS.8–10,12,22–24 All studies were approved by the relevant medical ethical committees, and all participants provided written informed consent. All study populations were previously described in detail.8–10,12,22–24 Here, we give a brief description of each study population.
IA patients (n=786) were admitted to the Utrecht University Medical Centre (the Netherlands) between 1997 and 2007. The population consisted of 247 men and 539 women and included both patients with ruptured (727) and patients with only unruptured (59) IAs. Ruptured IAs were defined by symptoms suggestive of subarachnoid hemorrhage combined with subarachnoid blood on a computed tomography scan and a proven aneurysm at angiography (conventional angiogram, computed tomography angiogram, or MR-angiogram). Unruptured IAs were identified by computed tomography angiography, MR angiography, or conventional angiography in the absence of clinical or radiological signs of subarachnoid hemorrhage.8–10 As controls, we included 3110 Dutch subjects who were recruited as part of the Nijmegen Biomedical Study (n=1832) and the Nijmegen Bladder Cancer Study (n=1278).22,23 All case and control subjects were genotyped on Illumina CNV370 Duo BeadChips.
AAA patients (n=859) were recruited from 8 medical centers in the Netherlands, mainly when individuals visited their vascular surgeon in the outpatient clinic or, in some cases, during hospital admission for elective or emergency AAA surgery. An AAA was defined as an infrarenal aorta diameter of ≥30 mm. The cohort consisted of 772 men and 87 women. The mean AAA diameter was 58.4 mm. Of these patients, 530 had received surgery, of whom 43 after rupture. Genotyping was performed on Illumina HumanHap610 chips.12 Control subjects (n=2089) were ascertained via the Rotterdam Study, a population-based cohort of subjects ≥45 years of age recruited from a district in Rotterdam (the Netherlands). These subjects were genotyped on Illumina HumanHap550 chips.24
For the IA and AAA GWAS, PLINK version 1.0725 was used for quality control (QC) of both subjects and single-nucleotide polymorphisms (SNPs). After removal of SNPs with A/T or C/G alleles and SNPs that were not called in any individual, we performed sample QC and SNP QC as described below.
We performed sample QC after merging cases and controls using a subset of common, high-quality SNPs (as defined by SNPs without deviation from Hardy–Weinberg equilibrium [P>0.001], with high minor allele frequency [>20%] and low missingness [<1%]), and we performed pruning based on linkage disequilibrium (LD; r2>0.5). Subjects were removed on the basis of the following 3 criteria: genotype missingness (subjects with a call rate <95% were removed), heterozygosity (subjects were excluded if the inbreeding coefficient deviated >3 SD from the mean), and cryptic relatedness (by calculating identity-by-descent for each pair of individuals). In each pair with an identity-by-descent proportion of ≥5%, a subject was excluded if it exhibited distant relatedness with multiple individuals. For case–control pairs, we removed the control subject. In the remaining pairs, the subject with the lowest call rate was excluded.
We performed principal component (PC) analysis using EIGENSTRAT26 on the study subjects from the HapMap population of Northern and Western European ancestry (CEU).27 We excluded SNPs from 3 regions with known long-distance LD: the major histocompatibility region (chr6, 25.8–36 Mbp), the chromosome 8 inversion (chr8, 6–16 Mbp), and a chromosome 17 region (chr17, 40–45 Mbp). We created multidimensional scaling plots with the first 4 PC, using R version 2.11.28 On the basis of visual inspection of these plots, we excluded subjects that seemed to be outliers with respect to the CEU or the study population. After outlier removal, we recomputed PCs to include as covariates for logistic regression.
After sample QC, we excluded SNPs with >5% missing genotypes, a minor allele frequency <1%, genotype missingness higher than the minor allele frequency, and Hardy–Weinberg equilibrium deviation (P<1×10-6). We performed these QC steps in each study cohort separately and again after merging cases and controls. We also removed SNPs with a differential degree of missingness between cases and controls (P<1×10-5; χ2 test).
For imputation of ungenotyped SNPs, we used BEAGLE software version 3.0.4,29 with the HapMap phase II CEU population27 as the reference panel. Genotype probability scores were converted to allele dosages ranging from 0 to 2. After imputation, we removed SNPs with imputation quality scores (ie, ratio of the observed variance of the allele dosage to the expected variance) <0.1 or >1.1.
We performed logistic regression analyses of each SNP versus disease state in both the IA and the AAA study using PLINK. From the 10 PCs calculated in each study cohort, we included the PCs associated with case-control status (P<0.05; logistic regression) as covariates in this analysis.
We investigated the genetic overlap between IAs and AAAs and aneurysm risk traits (plasma lipid levels, BP, and CAD) by creating genetic scores for each trait in our study subjects on the basis of established risk SNPs for these traits. For plasma lipid levels, we listed 52 SNPs known to be associated with TC, 37 SNPs with LDL-C, 47 SNPs with HDL-C, and 31 with triglycerides.30 A total number of 26 SNPs are associated with systolic BP (SBP), 26 with diastolic BP (DBP),31,32 22 with mean arterial pressure (MAP), and 10 with pulse pressure.33 Another 31 risk SNPs are associated with CAD.34–36 As a negative control, we also used 180 SNPs associated with height.37 For each trait, we listed the associated SNPs with their corresponding risk alleles and published effect sizes. When the association was reported in multiple studies, we took the effect size from the largest study (ie, CARDIoGRAM consortium34 for previously identified CAD SNPs and International Consortium for Blood Pressure Genome-Wide Association Studies31 for previously identified BP SNPs). We only included SNPs that were independent (in low LD; r2<0.1). We used SNP Annotation and Proxy Search (SNAP)38 to calculate LD between SNPs (on the basis of genotype data from HapMap release 22) and to search for proxy SNPs in case risk SNPs were absent from our data.
For CAD, 30 published risk SNPs were independent. From these, 2 SNPs were not genotyped or imputed in our data because they were absent from HapMap. For 1 SNP (rs17465637 near MIA334), we used a perfect proxy (rs17011681; r2=1 in 1000 Genomes pilot 1). The other SNP (rs3798220 near LPA34) was excluded because no proxies were available. This resulted in a total number of 29 SNPs used for genetic score calculation.
To improve power for the BP scores, we constructed a composite BP score incorporating all SBP, DBP, MAP, and pulse pressure SNPs into a single score by including only independent SNPs with concordant effect directions across these 4 traits. This resulted in a genetic score of 35 SNPs after excluding 3 pulse pressure–associated SNPs with a discordant effect on DBP.33 For this genetic score calculation, we ascribed each SNP an equal weight because the effect sizes differed across these 4 BP traits.
For all SNPs associated with lipid factors, CAD, and BP, we listed the P values, odds ratio (OR), and direction of effect in our IA and AAA GWAS. In accordance with our hypothesis, we expect that the direction of effect of these SNPs is the same between each trait and IA/AAA (eg, BP increasing alleles increase AAA risk), with the exception of HDL-C increasing SNPs, which are expected to decrease AAA risk.
Using these SNPs, we calculated genetic scores for each trait in each individual of the IA and AAA cohort as follows:
Genetic score = β1x1 + β2x2 + … + βnxn,
where xi is the estimated allele dosage (between 0 and 2) in a given individual, and βi is the reported effect size (lipids, BP, and height) or the natural log of the reported OR (CAD) for the ith SNP.
We tested the resulting genetic scores for association with IAs and AAAs using logistic regression. Sex and PCs significantly associated with disease in the study cohorts were added as covariates in the analysis. We also adjusted for the 7 known IA risk SNPs (rs6841581 at locus 4q31, rs1333040 at 9p21, rs10958409 and rs9298506 at 8q11–12 [r2=0.058], rs12413409 at 10q24, rs9315204 at 13q13, and rs11661542 at 18q11)9,11 and for the 3 AAA risk SNPs (rs1466535 at 12q13, rs10757278 at 9p21, and rs7025486 at 9q33).12,13
To evaluate the effect of each genetic score, we divided the IA and AAA cohorts in quartiles on the basis of genetic score derived from each subject and calculated ORs and corresponding 95% confidence intervals (CIs) for IA/AAA disease risk in the highest quartile compared with that in the lowest quartile, using logistic regression. We adjusted for significantly associated PCs, known IA and AAA risk SNPs, and sex.
IA and AAA Cohorts
The IA cohort consisted of 709 cases and 2692 controls, and the AAA cohort contained 807 cases and 1905 controls. After imputation, both cohorts comprised 23 93 271 autosomal SNPs. Table 1 shows the baseline characteristics of the IA and AAA cohorts after QC.
For all SNPs associated with lipid factors, CAD, and BP, our IA and AAA GWAS results are listed in Tables I to VI in the online-only Data Supplement. Using these lists of SNPs, we constructed genetic scores for TC, LDL-C, HDL-C, triglycerides, CAD, SBP, DBP, MAP, pulse pressure, and composite BP and height, and we tested these for association to IAs and AAAs. As an example, the distribution of LDL-C-scores in the AAA cohorts is shown in the Figure. The distribution of all genetic scores in the IA and AAA cohort is shown in the Figure in the online-only Data Supplement. We estimated the effect size of these genetic scores by comparing the disease risk in the highest genetic score quartile with that in the lowest quartile (Table 2).
For IAs, we found associations for the genetic score of SBP (P=0.031), DBP (P=0.020), MAP (P=0.0065), and the composite BP score (P=0.0030). For the other genetic scores (TC, LDL-C, HDL-C, triglycerides, and CAD), none reached nominal significance. For the significantly associated genetic scores, the OR for IA risk in the highest versus the lowest genetic score quartile was 1.09 for the SBP score (95% CI, 1.00–1.19), 1.10 for the DBP score (95% CI, 1.01–1.20), 1.10 for the MAP score (95% CI, 1.01–1.19), and 1.11 for the composite BP score (95% CI, 1.02–1.21).
In AAAs, we found significant associations for genetic scores of TC (P=3.6×10-6), LDL-C (P=5.7×10-7), and CAD (P=1.4×10-3) and a weaker association with the HDL-C-score (P=0.020). We did not observe an association with the triglycerides and the BP scores. The OR for AAA risk in the highest versus the lowest genetic score quartile was 1.24 for the TC score (95% CI, 1.13–1.35), 1.21 for the LDL-C score (95% CI, 1.10–1.32), 0.94 for the HDL-C score (95% CI, 0.86–1.03), and 1.18 for the CAD score (95% CI, 1.06–1.32).
To test the impact of winner’s curse on the published effect estimates of the SNP associations, we repeated the analyses by recalculating the genetic risk scores without weighting each risk allele by the published effect size, computing effectively the total number of risk alleles in each individual. The results were essentially unchanged (online-only Data Supplement Table 7).
We explored the individual contributions of SNPs making up the scores that were significantly associated with IAs or AAAs. We observed that 1 SNP (rs6511720) associated with TC and LDL-C is also strongly associated with AAAs (P=3.9×10-5). This SNP is located at the LDLR gene, which is known to be associated with CAD34 and familial hypercholesterolemia.30 For the other SNPs in the genetic scores, the individual associations with IAs or AAAs were much weaker (online-only Data Supplement Tables I to VI).
We did not find evidence for a relation between the height-score with either IAs (P=0.67) or AAAs (P=0.20).
In this study, we used a genetic approach to study the role of epidemiological risk traits in IAs and AAAs. We found that genetic scores based on validated SNPs associated with CAD and lipid factors influence disease risk of AAAs but not of IAs. In contrast, genetic scores based on BP-associated SNPs are associated with IAs but not with AAAs.
The absence of association between genetic scores based on lipid risk alleles and IAs is consistent with previous epidemiological studies reporting that hypercholesterolemia is not associated with higher risk of IAs but possibly a risk-reducing factor.2 In contrast, we did find significant associations of TC and LDL-C scores to AAAs. Previous epidemiological studies considering lipid factors in AAAs report conflicting results and suffer from several limitations. For example, most studies did not adjust for lipid-lowering medication as a potential confounder,3,14,17 with the exception of some large studies that found an association for TC14 and HDL-C14,17 but not for triglycerides 14,17 and disagreed about an association for LDL-C.15,17 Overall, our results are largely consistent with large epidemiological studies, which adjusted for lipid-lowering medication.
To our knowledge, a shared genetic background of lipid factors and AAAs has not been described before. A recent GWAS reported an association between a common variant in LRP1 (rs1466535) and AAAs.13 Another variant mapping to the same gene (rs11613352) was found to be associated with HDL-C and triglycerides.30 However, close examination of the LRP1 locus reveals that these 2 SNPs are not in LD (r2=0.04). Furthermore, the AAA risk SNP at LRP1 was not associated with lipid levels in a previous meta-analysis.13 In this study, we found that one SNP (rs6511720) driving the association between the TC and the LDL-C score and AAAs maps to LDLR. This gene is of particular interest because mutations in LDLR are a well-known cause of familial hypercholesterolemia.39 Taken together, our findings suggest that a part of the (yet unknown) genetic risk variants for AAA may act through changes in lipid factors.
The association of the genetic CAD score with AAAs (and lack of association with IAs) may suggest a shared pathogenesis between CAD and AAAs. Epidemiologically, the risk for CAD is increased in the years after discovery of AAAs and after rupture of IAs.40,41 However, a history of myocardial infarction is a risk trait for AAAs,18 and CAD is frequently present at the time of diagnosis of AAAs.18,19 No significant comorbidity between CAD and IAs has been reported, although to our knowledge there has been only one study that investigated this relation.20 This study suggests a stronger association between CAD and AAAs than between CAD and IAs, which is consistent with our study findings.
Although the 9p21 locus is shared among CAD, IAs, and AAAs,10 we did not find evidence for additional genetic sharing between CAD and IAs. Considering the pleiotropic effects of 9p21 across a wide range of human conditions (including ones not associated with vascular diseases), sharing of this locus may not necessarily imply an overlap in disease etiology.
We observed a strong association of the composite BP score with IA, which was also supported by associations of the individual SBP, DBP, and MAP scores. The BP scores were not associated with AAAs, but the effect directions of these scores were similar for BP increase and AAA disease risk. The difference in effect of the BP scores on IA and AAA disease risk is in line with epidemiological findings: hypertension is one of the most prominent risk factors of IAs (OR, 2.6; 95% CI, 2.0–3.1),2 whereas for AAAs, a more modest effect on disease risk has been reported (OR, 1.3; 95% CI, 1.14–1.55).18 Of note, it has been suggested that hypertension is a risk factor for AAA rupture but not for development of AAAs.18 This finding can explain our study results, because our AAA study population is largely comprised of patients with unruptured aneurysms.
Our study findings are consistent with previous genetic studies on IAs and BP. One recent study discovered that a suggestive IA risk SNP at locus 5q23.2 in PRDM6 is also associated with high SBP.42 In addition, previous GWAS independently discovered that locus 10q24 is associated with IAs,9 and that the same SNP at this locus is also associated with SBP and DBP.31,32 From the genes in this region, CYP17A1 is a functional candidate gene for BP regulation, as it is involved in biosynthesis of mineralocorticoids and glucocorticoids.32 In contrast to IAs, no sharing of risk alleles has been reported for BP and AAAs.
In this study, we investigated the effect of SNPs associated with lipid factors, CAD, and BP on disease risk of IAs and AAAs. The genetic approach of this study has advantages compared with traditional epidemiological studies, which may suffer from potential confounding. In addition, reverse causation, where the disease influences the risk factor under study, may lead to false interpretations of causality. These limitations are less likely to play a role in genetic studies because alleles are randomly transmitted from both parents to offspring, are assumed to remain stable over a lifetime, and are not changed disease or by trait-modifying factors like medication.21
One limitation of our study is the modest sample size of our IA and AAA case–control collection, which puts a limit to what we can detect reliably. To illustrate this, we calculated the power for detecting a nominally significant association (P<0.05) in the IA analysis for a risk allele with a frequency of 10% using the genetic power calculator.43 The resulting power is 18% at a relative risk of 1.1 and 49% at a relative risk of 1.2 per allele, assuming additive effects. Power of the AAA analysis is virtually identical (because the sample sizes are similar). Under a model that multiple risk alleles have a cumulative effect on aneurysm risk, the power of detecting an association between the composite score and the outcome is much increased. The effective power increase will ultimately be dependent on the number of truly associated variants relative to the number of null variants (that may introduce noise) and the validity of the assumption of additivity and independence of the different variants. Needless to say, lack of an association with one of our genetic risk scores tested does not necessarily imply that none of the SNPs is truly associated. Finally, we note that the sample sizes of the IA and AAA cohorts are comparable, so the tests performed should represent a fair comparison between IAs and AAAs. That is to say, if we observe a significant effect in the IA case–control analysis, we should expect to have sufficient power for a similar effect to be detected in the AAA case–control analysis. The significant lipid–AAA association (P=3.6×10−6 for TC and P=5.7×10−7 for LDL-C) and the insignificant lipid–IA association (P=0.2 and P=0.6, respectively) seems, therefore, surprising and unexpected under the null. The association between the BP score and IAs (P=0.003) and the insignificant association for AAAs (P=0.2) are perhaps less striking, but the overall consistency between the (known) epidemiological relations and the observed genetic associations (namely, that hypertension is related to IAs and lipids and CAD are related to AAAs) is compelling.
We did not observe a significant association of the genetic score of human height, a highly polygenic trait, with IAs and AAAs. This reduces the likelihood that our results are because of population stratification, which is a possible confounder in genetic analyses.26 We also confirmed that the observed effects were not because of inflated effect sizes (winner’s curse) in the initial association studies.
In conclusion, this study demonstrates that the genetic scores for TC, LDL-C, HDL-C, and CAD are associated with the risk of AAAs but not with the risk of IAs, and genetic BP scores increase risk of IAs but not of AAAs. With this genetic approach, we confirm relationships observed in epidemiological studies. This study illustrates how genetic studies can help to elucidate the role of risk traits in aneurysm pathology. This study does not allow us to confirm whether the observed effects are actually mediated through these risk traits. Future research can include Mendelian randomization studies44 to determine whether these risk traits play a causal role in the disease mechanism of aneurysms.
We thank the individuals who participated in the study and whose contribution made this work possible. We also acknowledge Murat Gunel (Department of Genetics, Yale University School of Medicine, New Haven, CT) and Solveig Gretarsdottir (deCODE Genetics, Iceland) for sharing genotype data of the cases with intracranial aneurysms and abdominal aortic aneurysms.
Sources of Funding
Dr van ‘t Hof is supported by a grant of the Dutch Heart Foundation (National Health Service; project No. 2008B004). Dr Ruigrok is supported by a VENI grant from the Netherlands Organization for Scientific Research (NWO; project No. 91610016). Dr de Bakker is the recipient of a VIDI Award from the NWO. The generation and management of GWAS genotype data for the Rotterdam Study are supported by the NWO (No. 175.010.2005.011, 911-03-012). The Rotterdam Study is funded by the Research Institute for Diseases in the Elderly (014-93-015; RIDE2), the Netherlands Genomics Initiative/NWO project No. 050-060-810.
Guest Editor for this article was Nilesh J. Samani, MD.
The online-only Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.113.000022/-/DC1.
- Received June 28, 2012.
- Accepted April 12, 2013.
- © 2013 American Heart Association, Inc.
- Norman PE,
- Powell JT
- Feigin VL,
- Rinkel GJ,
- Lawes CM,
- Algra A,
- Bennett DA,
- van Gijn J,
- et al
- Ogata T,
- MacKean GL,
- Cole CW,
- Arthur C,
- Andreou P,
- Tromp G,
- et al
- Yasuno K,
- Bakircioğlu M,
- Low SK,
- Bilgüvar K,
- Gaál E,
- Ruigrok YM,
- et al
- Bown MJ,
- Jones GT,
- Harrison SC,
- Wright BJ,
- Bumpstead S,
- Baas AF,
- et al
- Forsdahl SH,
- Singh K,
- Solberg S,
- Jacobsen BK
- Reed D,
- Reed C,
- Stemmermann G,
- Hayashi T
- Cornuz J,
- Sidoti Pinto C,
- Tevaearai H,
- Egger M
- Nguyen TV,
- Chandrashekar K,
- Qin Z,
- Parent AD,
- Zhang J
- 28.↵R Core Team. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria; 2012. ISBN 3-900051-07-0. http://www.R-project.org/.
- Johnson AD,
- Handsaker RE,
- Pulit SL,
- Nizzari MM,
- O’Donnell CJ,
- de Bakker PI
- Freiberg MS,
- Arnold AM,
- Newman AB,
- Edwards MS,
- Kraemer KL,
- Kuller LH
- Nieuwkamp DJ,
- Algra A,
- Blomqvist P,
- Adami J,
- Buskens E,
- Koffijberg H,
- et al
- Purcell S,
- Cherny SS,
- Sham PC
- Davey Smith G,
- Ebrahim S
Intracranial aneurysms (IAs) and abdominal aortic aneurysms (AAAs) share some risk factors, such as age, smoking, and genes. In contrast, other risk factors seem to play a different role in the 2 diseases. An unfavorable lipid profile and coronary artery disease increase risk of AAAs but not of IAs, whereas hypertension predisposes patients to IAs but not to AAAs. The evidence for these associations from epidemiological studies is rather limited, and some of these studies even report conflicting results. Genetic studies offer an attractive alternative to dissect the relation between risk factors and disease because they are less likely to suffer from limitations like confounding and reverse causation than do epidemiological studies. In this study, we investigated genetic profiles on the basis of variants known to be associated with serum lipid levels, coronary artery disease, and blood pressure in 2 large cohorts of IA and AAA cases and controls. Consistent with the epidemiological observations, we show that genetic profiles of lipid factors and coronary artery disease are associated with AAAs but not with IAs, and the genetic profile of blood pressure is associated with IAs but not with AAAs. These findings suggest that these respective risk factors likely play a causal role in the onset of IAs and AAAs. Our study also illustrates the value of genetic studies for shedding light on the pathogenesis of cardiovascular diseases and the mechanisms they might share.