Introduction

Abdominal aortic aneurysm (AAA) primarily affects men, with population-screening studies demonstrating an increased prevalence of approximately fivefold compared with women.1 Currently, the reasons for the male propensity for AAA are not known, but maybe linked to sex hormones, as in experimental models androgens have been positively linked to aortic dilatation.2 The production, metabolism and response to sex hormones are controlled by an array of enzymes and receptors, including steroid 5α reductase, aromatase (estrogen synthetase), androgen and estrogen receptors.3, 4 Steroid 5α reductase enzymes are responsible for the conversion of testosterone to the more potent androgen dihydrotestosterone.3 Aromatase (estrogen synthetase) is a cytochrome P450 enzyme that converts androgen precursor steroids to estrogens.4 The responses to circulating sex hormones are determined by the androgen and estrogen receptors. Genetic factors appear to have an important role in the production, metabolism and response to male sex hormones, and therefore maybe relevant to AAA, which has been shown to have inherited risk factors.5, 6 We selected four genes important in the production, metabolism and response to sex hormones to examine the association of genetic polymorphisms with aortic dilatation. We hypothesized that single-nucleotide polymorphisms (SNPs) in the steroid 5α reductase, subfamily A, polypeptide 1 (SRD5A1), cytochrome P450, family 19, subfamily A, polypeptide 1 (CYP19A1), androgen receptor (AR) and estrogen receptor 2 (ESR2) genes were associated with aortic dilatation and examined this within subset of the Health In Men Study (HIMS). We attempted to replicate any positive associations in an independent cohort from New Zealand.

Subjects and methods

Study design and subjects

HIMS consists of a cohort of men who originally participated in a trial of screening for AAA and has been previously described in detail.7, 8 For the current study, genotyping was undertaken in all men with AAAs from whom DNA was available (n=640) and 1071 randomly selected age-matched men without an AAA were taken as controls. Any SNPs found to be significantly associated with aortic diameter were further assessed in an independent cohort of subjects from New Zealand.9 These subjects included 513 men with large AAAs (80% had undergone aortic repair) and 269 healthy elderly men from the same region of Otago. All subjects included in both cohorts had undergone abdominal ultrasound. Ultrasound reproducibility was assessed during subject recruitment and 95% confidence intervals were <3 mm.9 The definitions of clinical risk factors such as hypertension, dyslipidemia, diabetes, coronary heart disease and waist-to-hip ratio were as previously described.7

Genotyping

The Haploview software package (http://sourceforge.net) was used to define the linkage disequilibrium blocks and to choose tagging SNPs within the location of SRD5A1 (n=14), CYP19A1 (n=39), AR (n=5) and ESR2 (n=16) genes using HapMap Phase II data utilizing a pairwise approach (minor allele frequency >5% and r2>0.8).10 Regions analyzed included the entire gene, plus additional sequences 10 kb upstream and downstream of the gene. With this approach, 100% of the variation in the genes was captured. Genotyping on the HIMS subjects was carried out using the Illumina Golden Gate assay on an Illumina BeadLab System at University of Western Australia. Genotype calls were made using Bead Studio Genotyping Module software package Version 3.1 (Illumina, San Diego, CA, USA). Monoallelic SNPs and those with genotyping efficiency <15% were excluded. As a result, findings for SRD5A1g.10386C>G (SNP ID rs4702379), SRD5A1g.720102C>T (SNP ID rs6872996), CYP19A1g.1667A>C (SNP ID rs2445768), CYP19A1g.16698C>T (SNP ID rs8025374), AR g.98970A>G (SNP ID rs2361634) and ESR2g.63762130G>T (SNP ID rs1152583) were excluded. Genotyping efficiency for the other 68 SNPs was between 97 and 100%. Genotyping in the New Zealand cohort was carried out using polymerase chain reaction as previously described.9

Statistical analysis

Hardy–Weinberg equilibrium was tested on a contingency table of observed verses predicted phenotype frequencies using a modified Markov-chain random-walk algorithm. To test our hypotheses, we investigated the association of genotyped SNPs and aortic diameter using linear regression adjusting for other risk factors for AAA (age, smoking, coronary heart disease, dyslipidemia, hypertension, diabetes, waist-to-hip ratio). Each of the bi-allelic SNPs was coded into three genotype classes and analyzed under codominant models (0=major allele homozygote, 1=heterozygote, 2=minor allele homozygote). Any significant codominant models were explored further (dominant and recessive models) to determine the best-fitting model using Akaiki information criteria. Haplotype frequencies were estimated from unphased genotype data using an expectation maximization algorithm under the assumption of Hardy–Weinberg equilibrium. Haplotypes were associated with aortic diameter using a generalized linear recessive model based on the initial genotyping results. Computations were undertaken using SimHap v1.0.2 (http://www.genepi.org.au/simhap.html) and SSPS 14.0 (SPSS Inc., Chicago, IL, USA). Genotypes associated with aortic diameter within the HIMS at a Bonferroni corrected P-value of <0.0007 (based on the 74 SNPs assessed) were examined in a second independent cohort.

Results

Association of genotypes with aortic diameter in the HIMS subjects

Genotyping was carried out in 1711 HIMS subjects of whom 640 (37%) had an AAA (Table 1). Out of 74, 68 (92%) SNPs passed quality assessments, were in Hardy–Weinberg equilibrium in controls, and therefore were assessed for association with aortic diameter. One SNP in CYP19A1 (CYP19A1g.49412370C>T; SNP ID rs1961177) was independently associated with aortic diameter in HIMS subjects (Supplementary Table). Median aortic diameters were 24.2 (interquartile range 21.1–31.3), 24.5 (inter-quartile range 21.1–32.7) and 30.6 mm (21.9–40.0) for men with CC (n=1317), TC (n=329) and TT (n=32) genotypes. CYP19A1g.49412370C>T (SNP ID rs1961177) was independently associated with aortic diameter under a recessive model (coefficient 5.058, SE 1.394, P=0.0003), including Bonferroni correction for multiple testing.

Table 1 Comparison of subjects with and without AAA undergoing genotyping
Full size table

Haplotype analysis in the HIMS subjects

A total of 14 linkage dysequilibrium blocks were identified within the CYP19A1 gene using HapMap Phase II data (Figure 1). We assessed the association of haplotype combination from these different blocks with aortic diameter in HIMS subjects. Four haplotypes were demonstrated to be significantly associated with aortic diameter, after adjusting for other risk factors and multiple testing (Table 2). In particular the relative common haplotype CTT defined by the g.49386747C>G (SNP ID rs17523922), g.49394252C>T (SNP ID rs3751591) and g.49412370C>T (SNP ID rs1961177), which was present in 10% of men, P=0.0001 (Table 2).

Figure 1
figure 1

Schematic diagram of linkage disequilibrium blocks within CYP19A1 gene. Black triangles represent linkage disequilibrium (LD) blocks; dark-grey represents strong LD; grey to white represents moderate to no LD among the single nucleotide polymorphisms.

Full size image
Table 2 The association of CYP19A1 haplotypes with aortic diameter
Full size table

Genotyping in the New Zealand men

We further examined g.49412370C>T (SNP ID rs1961177) in an independent cohort of 782 men from Otago, 513 (66%) of whom had an AAA. The patients with AAA in the New Zealand sample had a mean diameter of just under 6 cm, compared with <4 cm for the Western Australian men (Table 1). CYP19A1g.49412370C>T (SNP ID rs1961177) was not associated with aortic diameter in the New Zealand subjects (coefficient 0.076, SE 0.284, P=0.788).

Discussion

To our knowledge, this is the first published report examining the association of polymorphisms in genes determining circulating sex hormones and aortic diameter. As male gender is an important risk factor for AAA, we hypothesized that polymorphisms in four genes important in the production and action of sex hormones would be associated with aortic dilatation. One SNP within intron 1 of CYP19A1 was strongly associated with aortic diameter within the HIMS cohort, which included patients with small AAAs. The importance of this genetic locus in aortic dilatation was further supported by further analysis, which identified a haplotype, including this SNP as highly associated with aortic diameter in HIMS subjects. This genetic variation within CYP19A1 was not, however, associated with aortic diameter in the New Zealand subjects in which patients had much larger AAAs. Given the large number of SNPs examined and the lack of replication, our findings do not support a consistent association between polymorphisms in these genes and aortic diameter in men. Our findings do not, however, rule out a role of the one SNP highlighted in this study in early stage AAA formation in which different mechanisms may be involved compared with later-stage aneurysm progression. Also, our study did not examine all genes involved in sex hormone production and functions such as that of encoding estrogen receptor 1. The assessment of the SNP in CYP19A1 in another population screened for AAA in which aortic diameter and risk factors have been carefully assessed would be worthwhile when such a group becomes available.