Polymorphisms in Transforming Growth Factor-β-Related Genes ALK1 and ENG Are Associated With Sporadic Brain Arteriovenous Malformations
Background and Purpose— Mutations in endoglin (ENG) and activin-like kinase (ALK1) cause hereditary hemorrhagic telangiectasias, disorders characterized by pulmonary and brain arteriovenous malformations (BAVMs). We investigated whether polymorphisms in these genes are also associated with sporadic BAVM.
Methods— A total of 177 sporadic BAVM patients and 129 controls (all subjects white) were genotyped for 2 variants in ALK1 and 7 variants in ENG.
Results— The ALK1 IVS3-35A>G polymorphism was associated with BAVM: (AnyA [AA+AG] genotype: odds ratio, 2.47; 95% CI, 1.38 to 4.44; P=0.002). Two ENG polymorphisms, ENG −1742A>G and ENG 207G>A, showed a trend toward association with BAVM that did not reach statistical significance.
Conclusions— A common polymorphism in ALK1 is associated with sporadic BAVM, suggesting that genetic variation in genes mutated in familial BAVM syndromes may play a role in sporadic BAVMs.
The etiology of brain arteriovenous malformations (BAVMs) remains poorly understood. Most are sporadic but also occur in hereditary hemorrhagic telangiectasias (HHT),1 familial disorders caused by loss-of-function mutations in endoglin (ENG) and activin-like kinase (ALK1, ACVLR1). Sporadic and HHT-related BAVMs share many clinical and morphological characteristics,2 suggesting common pathological mechanisms. We investigated whether sporadic BAVM is associated with ALK1 or ENG polymorphisms.
Materials and Methods
We compared polymorphism genotypes between BAVM patients and healthy controls of self-reported white race/ethnicity. BAVM patients were recruited as described previously.3,4 Patients with a family history or diagnosis of HHT were excluded. Controls were healthy volunteers from the same clinical catchment area with no chronic disease or medication and without significant past medical history.
Polymorphisms in ALK1 and ENG selected a priori by database searches (dbSNP) and via sequencing in 32 HHT patients (Table 1) were genotyped by template-directed dye-terminator incorporation assay with fluorescence-polarization detection.3,5
Variants polymorphic in the study cohort were tested for association with BAVM by univariable logistic regression. The criterion for statistical significance was P=0.05/6=0.0083 after Bonferroni correction for multiple testing. Variant genotypes were coded dichotomously for the model that best fit the data. The primary collapse was major allele homozygotes versus other genotypes (recessive). A dominant model was the best fit for ALK1 IVS3-35A>G. We tested appropriateness of the model by adding back the heterozygous term. Multivariable logistic regression controlling for age and sex was also performed. Possible joint effects of ALK1 IVS3-35A>G and ENG 207G>A were examined by using a composite variable in univariable analysis and running a multivariable analysis with both SNPs entered as predictors in the model; both approaches yielded similar results. Haplotype prediction, haplotype association analysis, and linkage disequilibrium (LD) analyses were performed using Haploview.
All polymorphisms were in Hardy-Weinberg equilibrium in cases and in controls. BAVM clinical features were similar to previously reported series (Table 2⇓);3,4 preliminary analyses revealed no significant associations with genotype. The ALK1 IVS3-35A>G polymorphism was associated with BAVM (univariable logistic regression; odds ratio [OR], 2.47; 95% CI, 1.38 to 4.44; P=0.002; AnyA genotype versus GG; Figure). This association remained significant (P=0.002) in a multivariable model incorporating age and gender (data not shown). Two ENG polymorphisms showed a trend toward association with BAVM, but the association did not reach statistical significance: ENG −1742A>G (OR, 1.7; 95% CI, 0.91 to 3.17; P=0.093, for AA genotype versus any other) and ENG 207G>A (OR, 1.88; 95% CI, 1.04 to 3.42; P=0.002 for GG genotype versus any other). ENG −1742A>G and ENG 207G>A were in LD (D′=0.95; r2=0.76). Haplotype prediction and haplotype association analysis was performed for the ENG locus but did not provide stronger support for association (data not shown). Because ALK1 and ENG proteins belong to the transforming growth factor-β (TGF-β) signaling cascade, we investigated the possibility of interaction between variants in their genes’ loci. The magnitude of the association with BAVM was modestly increased by combining the ALK1 and ENG genotypes with strongest effect (combined genotype variable: ALK1 IVS3-3 5AnyA+ENG 207GG; OR, 2.68; 95% CI, 1.64 to 4.39; P<0.001; Figure).
We report for the first time an association between sporadic BAVM and a common polymorphism in a gene mutated in HHT, consistent with the hypothesis that familial and sporadic BAVMs share common pathological mechanisms (ie, altered function of TGF-β receptor complexes).
Familial inheritance of AVMs outside HHT is rare.6 We excluded cases with HHT diagnosis or family history from our analyses to focus on sporadic disease. Although undiagnosed or future BAVM cannot be ruled out in our healthy control population, at an estimated prevalence of 10 per 100 000,7 it is unlikely to significantly confound our results.
Presence of the ALK1 IVS3-35A allele is associated with a 2.5-fold increase in likelihood of BAVM in whites. This association is strengthened in patients who are also homozygous for the ENG 207G allele, which, on its own, exhibits a trend with BAVM. It remains to be shown whether the association is attributable to the polymorphisms studied here or to other variants in LD with them. For the following SNPs, a plausible argument can be made that they are functional: the silent exonic ENG 207G>A variant reduces the predicted binding score for a splicing protein;8 the ENG −1742A>G promoter polymorphism alters a predicted CCAAT-binding protein site, potentially modulating transcription.
The BAVM-associated ALK1 genotypes are carried by the majority of subjects in the population. Interaction between multiple genetic and environmental factors likely underlies sporadic BAVM. Although confounding attributable to population substratification in our cohort cannot be ruled out, the large magnitude of the combined effect suggests clinical significance. A larger cohort will be necessary to test whether the findings in whites extend to other race/ethnic subgroups.
This work was supported in part by R01s NS34949, and NS41877, K24 NS02091, and P01 NS44155 (W.L.Y.), and GM61390 (K.M.G.). The authors thank UCSF BAVM Study Project, including: C.F. Dowd, V.V. Halbach, T. Hashimoto, R.T. Higashida, S.C. Johnston, D. Lind, and N.J. Quinnine.
- Received December 13, 2004.
- Revision received April 20, 2005.
- Accepted May 30, 2005.
Marchuk DA, Srinivasan S, Squire TL, Zawistowski JS. Vascular morphogenesis: tales of two syndromes. Hum Mol Genet. 2003; 12: R97–R112.
Matsubara S, Mandzia JL, ter Brugge K, Willinsky RA, Faughnan ME, Manzia JL. Angiographic and clinical characteristics of patients with cerebral arteriovenous malformations associated with hereditary hemorrhagic telangiectasia. AJNR Am J Neuroradiol. 2000; 21: 1016–1020.
Pawlikowska L, Tran MN, Achrol AS, McCulloch CE, Ha C, Lind DL, Hashimoto T, Zaroff J, Lawton MT, Marchuk DA, Kwok PY, Young WL. Polymorphisms in genes involved in inflammatory and angiogenic pathways and the risk of hemorrhagic presentation of brain arteriovenous malformations. Stroke. 2004; 35: 2294–2300.
Halim AX, Singh V, Johnston SC, Higashida RT, Dowd CF, Halbach VV, Lawton MT, Gress DR, McCulloch CE, Young WL. Characteristics of brain arteriovenous malformations with coexisting aneurysms: a comparison of two referral centers. Stroke. 2002; 33: 675–679.
Cartegni L, Wang J, Zhu Z, Zhang MQ, Krainer AR. ESEfinder: a web resource to identify exonic splicing enhancers. Nucleic Acids Res. 2003; 31: 3568–3571.