Differential Effects of Chromosome 9p21 Variation on Subphenotypes of Intracranial Aneurysm
Background and Purpose— Recently, a genome-wide association study identified associations between single nucleotide polymorphisms on chromosome 9p21 and risk of harboring intracranial aneurysm (IA). Aneurysm characteristics or subphenotypes of IAs, such as history of subarachnoid hemorrhage, presence of multiple IAs and location of IAs, are clinically important. We investigated whether the association between 9p21 variation and risk of IA varied among these subphenotypes.
Methods— We conducted a case-control study of 981 cases and 699 controls in Japanese. Four single nucleotide polymorphisms tagging the 9p21 risk locus were genotyped. The OR and 95% CI were estimated using logistic regression analyses.
Results— Among the 4 single nucleotide polymorphisms, rs1333040 showed the strongest evidence of association with IA (P=1.5×10−6; per allele OR, 1.43; 95% CI, 1.24–1.66). None of the patient characteristics (gender, age, smoking, and hypertension) was a significant confounder or effect modifier of the association. Subgroup analyses of IA subphenotypes showed that among the most common sites of IAs, the association was strongest for IAs of the posterior communicating artery (OR, 1.69; 95% CI, 1.26–2.26) and not significant for IAs in the anterior communicating artery (OR, 1.22; 95% CI, 0.96–1.57). When dichotomizing IA sites, the association was stronger for IAs of the posterior circulation–posterior communicating artery group (OR, 1.73; 95% CI, 1.32–2.26) vs the anterior circulation group (OR, 1.28; 95% CI, 1.07–1.53). Heterogeneity in these ORs was significant (P=0.032). The associations did not vary when stratifying by history of subarachnoid hemorrhage (OR, 1.42; 95% CI, 1.18–1.71 for ruptured IA; OR, 1.27; 95% CI, 1.00–1.62 for unruptured IA) or by multiplicity of IA (OR, 1.57; 95% CI, 1.21–2.03 for multiple IAs; OR, 1.36; 95% CI, 1.15–1.61 for single IA).
Conclusions— Our results suggest that genetic influence on formation may vary between IA subphenotypes.
The rupture of an intracranial aneurysm (IA) is the most common cause of subarachnoid hemorrhage (SAH) associated with high morbidity and mortality. IA is implicated to be a multifactorial disease in which multiple genes and environmental factors influence disease risk.1 Recently, a multistage genome-wide association study of Finnish, Dutch, and Japanese cohorts identified associations between single nucleotide polymorphisms (SNPs) on chromosomes 2q33, 8q11, and 9p21 and the susceptibility to IA.2 Among the three loci, the 9p21 locus seems most intriguing because SNPs identified by the genome-wide association study of IA are in linkage disequilibrium with SNPs associated with several other arterial diseases (coronary artery disease, abdominal aortic aneurysm, and peripheral arterial disease),3 and the 9p21 risk locus overlaps a newly annotated noncoding RNA, called ANRIL (or CDKN2BAS).4
Patient characteristics such as gender, age, smoking habit, and history of hypertension are well-established risk factors affecting formation, growth, and rupture of IA.1 Furthermore, aneurysm characteristics such as the history of SAH, presence of multiple IAs, and location of IA (hereinafter, referred to as “subphenotypes” of IA) are clinically important because they are associated with the outcome of patients with IA.5,6 Genetic dissection of IA subphenotypes is essential but uncharted territory.7
We conducted a case-control study in the Japanese population to evaluate the possibility of confounding or effect modification of the 9p21 association through the environmental risk factors and to investigate whether the association between the 9p21 variation and risk of IA varied according to subphenotype.
Materials and Methods
All study participants were Japanese and recruited at Tokyo Women’s Medical University, Chiba University, and their affiliated hospitals. The Ethics Committee of Tokyo Woman’s Medical University, Chiba University, and Tokai University approved the study protocols and all participants gave written informed consent.
Nine hundred eighty-one IA patients included 322 familial IA patients (80 probands from nuclear families that had been used in our linkage study8 and 242 patients who had a family history of IA) and 659 sporadic IA patients. The presence of IA was confirmed by digital subtraction angiography, 3-dimensional computed tomography angiography, magnetic resonance angiography, or surgical findings. Phenotyping of IAs, such as aneurysm location and number of aneurysms, was performed by certified neurosurgeons. The patients harboring saccular IA were included, but those harboring fusiform or dissecting IA were excluded. All 699 controls were screened for not harboring IA by means of neuroradiological imaging such as MRI. IA patients and controls were classified as hypertensive if they had a medical history of hypertension or were currently receiving antihypertensive therapy. The subjects were divided into 2 groups according to their self-reported smoking habit: never-smokers and ever-smokers. The dataset was updated from our previous studies.2,8 Baseline characteristics of the study participants are summarized in Table 1.
All study participants were genotyped for 4 9p21 SNPs (rs1333040, rs2891168, rs2383207, and rs10757278). These SNPs were selected as tag SNPs capturing SNPs previously reported to show strong association to IA, coronary artery disease, abdominal aortic aneurysm, and peripheral arterial disease2,3,9–13 at r2>0.9 in phase II HapMap JPT (Japaneses in Tokyo, Japan) samples. Linkage disequilibrium structure and selection criterion of the SNPs are shown in Supplemental Figure I available online at http://stroke.ahajournals.org. Genotyping was performed by TaqMan SNP Genotyping Assays on the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems) or by direct sequencing with BigDye Terminators v3.1 Cycle Sequencing Kit on the Applied Biosystems 3730 DNA analyzer (Applied Biosystems).
The statistical significance of departure from Hardy-Weinberg equilibrium in control samples was examined by means of the exact test using PLINK software.14 We assessed the significance of association using the Cochran-Armitage trend test. To determine the most likely genetic model, additive and nonadditive effects were modeled in the context of logistic regression analysis by defining the SNP genotype predictor value x1 and x2 as and respectively, where “A” was the putative risk allele. If there is no evidence of nonadditivity (P>0.05), then the multiplicative model (additive in logit scale) is selected; otherwise, the dominance or recessive model is selected.
The stepwise logistic regression procedure was applied to assess the relative importance of the 4 SNPs. The significance level of 0.01 was necessary for entering a SNP into the model, and the significance level of 0.01 was necessary for a SNP to stay in the model at any iteration step. When applying the stepwise logistic regression procedure, missing genotypes were imputed via Beagle software15 with default settings. The accuracy of the imputation was assessed by means of the R2 values (ie, the estimates of the squared correlation between the imputed allele dosage and the true allele dosage).15 The resulting phased genotype data were used for haplotype analysis.
ORs of IA for the selected SNP and 2 risk factors (smoking habit and history of hypertension) were estimated by logistic regression analysis using a univariate analysis for each predictor, and then a multivariate model including all the predictors with adjustment for gender and age (<50, 50–59, 60–69, ≥70 years) to address possible confounding. We investigated whether the magnitude of the association between the 9p21 SNP and IA was modified by the risk factors. The joint association of the 9p21 genotype and each risk factor with IA was evaluated. Six categories were formed by combining the 9p21 genotype (AA, Aa, aa) and either smoking habit (ever, never) or history of hypertension (yes, no). Furthermore, interactions between 9p21 and risk factor were investigated by adding a product term into the multivariate logistic regression model to assess whether deviations from multiplicative joint effects were statistically significant (or to examine whether the joint effects were significantly greater or smaller than expected).
To evaluate whether the association between the 9p21 SNP and risk of harboring IA varied according to subphenotypes, we examined the association between the 9p21 SNP and IA according to subgroups stratified by history of SAH (ruptured or unruptured), multiplicity of IA (multiple or single), and location of the IA. When analyzing the history of SAH and location of IA, we excluded patients with multiple IAs because some patients with multiple IAs simultaneously had different types of aneurysms according to the SAH status or aneurysm location (eg, patients with ruptured and unruptured IAs). Thus, 713 patients harboring a single IA were analyzed.
For the subgroup analysis regarding location of IA, the definition of 2 sets of case-subgroups was as follows: (1) patients harboring IA at 3 common sites (anterior communicating artery [AcomA], middle cerebral artery [MCA], and posterior communicating artery [PcomA]), which we defined as the site where the number of patients was >100; and (2) patients grouped according to the International Study of Unruptured Intracranial Aneurysm5: (i) AC/MC/IC, which signify aneurysms in the AcomA, the anterior cerebral artery (ACA), the MCA, or the internal carotid artery (ICA) (not the cavernous portion) and (ii) Post-Pcomm, meaning the vertebrobasilar, the posterior cerebral arterial system, or the PcomA. Distribution of the IA subphenotypes is seen in Table 1.
To estimate subphenotype-specific ORs, the frequency of the 9p21 variant in each case-subgroup (eg, patients with ruptured or unruptured IA) was compared with the frequency in controls using a polytomous logistic regression. To assess whether the estimated subphenotype-specific ORs were significantly different, the frequencies of the 9p21 variant were compared between case-subgroups by dichotomous logistic regression (the case-subgroup heterogeneity test).16 For these subgroup analyses, the effects of gender and age were adjusted. When assessing the subgroup-specific effects stratified by SAH status, the case subgroup heterogeneity test with adjustment for IA site (6 categories: AcomA, ACA, MCA, ICA, PcomA, the posterior cerebral arterial system) was also implemented to keep out possible confounding effects attributable to complex relationships among IA subphenotypes. Similarly, when assessing the subgroup-specific effects stratified by IA sites, the effects of SAH status were adjusted (2 categories: ruptured, unruptured). All the logistic regression analyses were performed by SAS (SAS Institute).
The results of genotyping and association analysis for 4 SNPs on 9p21 are shown in Table 2. For the SNPs, the missing genotype rates were small (at most, 5.2% for rs2891168) and the genotypes for all SNPs were in the Hardy-Weinberg equilibrium in controls (P>0.05). All the SNPs showed highly significant associations with IA by means of the trend test (P=1.5×10−6 for rs1333040; P=7.4×10−5 for rs2891168; P=8.3×10−5 for rs2383207; P=4.1×10−5 for rs10757278). The effect of nonadditivity was not significant for any SNP (P>0.05); therefore, the multiplicative model was selected as an appropriate genetic model and applied in subsequent logistic regression analyses.
For the stepwise logistic regression analysis, the missing genotype imputation was implemented. The estimated R2 values were >0.98 for all the SNPs, indicating that accuracy of the imputation was considerably high. The stepwise logistic regression analysis identified rs1333040 as an independent predictor of IA among the 4 SNPs. The observed associations of the other 3 SNPs could be a result of linkage disequilibrium with rs1333040. Haplotype analysis reinforces this inference. All but 1 of 4 common haplotypes bearing T allele of rs1333040 (ie, H3, H5, and H6) was significantly associated with IA, whereas haplotype H2 bearing the C allele of rs1333040 and the risk alleles of the other 3 SNPs did not show significant association (Table 3). Thus, only rs1333040 was considered in further analyses.
The ORs obtained from both univariate and multivariate logistic regression analyses are shown in Table 4. The univariate OR for rs1333040 was 1.43 (95% CI, 1.24–1.66), which was similar to the OR of 1.44 (95% CI, 1.22–1.72) after the adjustment in the multivariate model that included gender, age, smoking, and hypertension. This result indicates that the association of this SNP is independent of the risk factors. However, the ORs for smoking habit and history of hypertension in the univariate model (1.33 [95% CI, 1.08–1.64] for smoking and 1.67 [95% CI, 1.36–2.06] for hypertension) were largely different from those in the multivariate model (2.32 [95% CI, 1.75–3.08] for smoking and 2.25 [95% CI, 1.78–2.86] for hypertension). This may be attributable to the correlations among the factors; smoking correlated negatively with age and female gender, and hypertension correlated positively with age (trend test P<0.0001).
The result of the joint association of rs1333040 and each risk factor is displayed in Supplemental Figure II available online at http://stroke.ahajournals.org, which shows that the effect of environmental exposure (smoking or hypertension) did not differ among subjects stratified by genotypes on a multiplicative (log-odds) scale and vice versa. Furthermore, the tests for deviations from multiplicative joint effects were not significant; the ORs for the genotype–smoking and genotype–hypertension interactions were 1.08 (95% CI, 0.76–1.53; P=0.67) and 1.22 (95% CI, 0.86–1.72; P=0.27), respectively.
The associations between rs1333040 and IA subdivided by aneurysmal subphenotypes were examined in the context of the polytomous logistic regression analyses. The subphenotype-specific ORs are shown in the Figure. The rs1333040 association was stronger for multiple IAs (OR, 1.57; 95% CI, 1.21–2.03) than for single IA (OR, 1.36; 95% CI, 1.15–1.61). However, the case-subgroup heterogeneity test did not show significant difference in OR between multiple and single IAs (P=0.29).
The subgroup analysis stratified by SAH status showed that rs1333040 was associated with ruptured IA (OR, 1.42; 95% CI, 1.18–1.71) and unruptured IA (OR, 1.27; 95% CI, 1.00–1.62). The difference in OR between ruptured and unruptured IAs was not significant in terms of the case-subgroup heterogeneity test (P=0.42). This was confirmed by the case-subgroup heterogeneity test with adjustment for IA site (P=0.69).
When stratifying patients with a single IA by 3 common sites, the association was strongest for PcomA (OR, 1.69; 95% CI, 1.26–2.26), secondly for MCA (OR, 1.36; 95% CI, 1.07–1.72), and not significant for AcomA (OR, 1.22; 95% CI, 0.96–1.57). Then, the difference in OR between PcomA and AcomA was suggestive by the case-subgroup heterogeneity test (P=0.067). However, the case-subgroup heterogeneity tests were not significant (P=0.50 for AcomA vs MCA and P=0.23 for MCA vs PcomA). When dichotomizing IA sites according to the grouping in the International Study of Unruptured Intracranial Aneurysm, the association was stronger for Post-Pcomm (OR, 1.73; 95% CI, 1.32–2.26) than for AC/MC/IC (OR, 1.28; 95% CI, 1.07–1.53). We found the difference between these ORs was statistically significant in the case-subgroup heterogeneity test (P=0.032). To consider a possible relationship between location of IA and SAH status, the difference among the site-specific ORs was also assessed with adjustment for SAH status. The resulting P values for the case-subgroup heterogeneity tests were P=0.055 for AcomA vs PcomA and P=0.031 for Post-Pcomm vs AC/MC/IC. These results indicate that the 9p21 SNP is more strongly associated with IA in the posterior circulation or PcomA than that in the anterior circulation.
We selected 4 9p21 SNPs to capture SNPs showing strong association with several arterial diseases in previous studies and then confirmed that the SNPs were associated with IA in Japanese. Among the 4 SNPs, rs1333040 showed the strongest evidence of association. The stepwise logistic regression analysis and the haplotype analysis reveal that the sequence variant tagged by the T allele of rs1333040 can explain the observed association with IA. Recently, the association between the 9p21 locus and IA was confirmed in a white population that was independent from populations analyzed in the original genome-wide association study.17
We found that the association was independent of established risk factors (gender, age, smoking, and hypertension). The presence of missing data on smoking and hypertension was a limitation of our study. However, similar findings were observed in case-control studies of IA and coronary artery disease; the effect of the 9p21 locus on risk of IA and coronary artery disease was not altered by adjustment for the risk factors.12,17,18 This may support the validity of our result.
The most important findings of this study are that rs1333040 was more strongly associated with IA of the PcomA and the posterior circulation than with IA in the anterior circulation. In terms of common sites, we found that the T allele of rs1333040 was associated with a 1.69-fold increased risk of IA in PcomA relative to the low-risk allele C, whereas the increased risk of IA in AcomA was only 1.22-fold.
To our knowledge, this is the first study showing that a common genetic variant with widely replicated evidence is associated with site distribution of IA. Epidemiological studies reinforce the hypothesis that genetic components influence site distribution of IA. There were considerably high concordance rates of IAs at identical or mirror-image sites in twins (70%)19 and in sibling pairs (69%),20 whereas the concordance rate in randomly selected pairs of patients was only 21%.19 These findings suggest that the 9p21 variant is one of the genetic factors predisposing to IAs in specific locations. Recently, Lindner et al21 demonstrated that the effects of established risk factors (gender, age, and alcohol intake) were different according to the site of aneurysms. This finding suggests that there is a possibility that the effects of genetic risk factors may also differ according to the site of aneurysms. The natural history of IAs in the anterior circulation and the posterior circulation and the PcomA are different. The posterior group of aneurysms is associated with a higher risk of rupture and poorer surgical and endovascular outcomes compared to the anterior circulation aneurysms.5,6 In the future, functional analyses focusing on differential gene expressions of aneurysm dome tissue subdivided by arterial sites may identify susceptibility genes and polymorphisms responsible for the formation of each subphenotype of IA and may help further understand the pathogenesis of this complex disease.
The effect of 9p21 SNP was not different when stratifying IA patients by SAH status. This finding suggests that the 9p21 risk locus may be involved in the formation of IA rather than in the precipitation of the aneurysm rupture.
It is known that subphenotypes of IAs show common characteristics, eg, the risk of aneurysm rupture is associated with aneurysm size and location.5,6 It is important to rule out possible confounding effects within the groups of IA subphenotypes. The analytical methods used in this study worked well for subgroup analyses regarding history of SAH and location of IA. The analysis for multiplicity of IA is more complex because of the presence of several subphenotypes in 1 patient eg, aneurysm at the AcomA and the MCA. Further sophisticated and large-scale analyses are needed to identify susceptibility genes of multiplicity of IA.
Some limitations of our study should be noted. All the current analyses are based on statistical analyses; therefore, functional analysis of rs1333040 or other SNP that is in linkage disequilibrium with rs1333040 is necessary. Although the diagnostic performance of the noninvasive imaging methods for the evaluation of IAs has been significantly improved, the use of magnetic resonance angiography and computed tomography angiography may be subject to misclassification of IA subphenotypes as compared to digital subtraction angiography and surgical findings. Subgroup analyses are subject to inflated false-positive rates attributable to multiple testing.22,23 The main conclusion of this study is based on 6 case-subgroup heterogeneity tests to assess whether the estimated subphenotype-specific ORs were significantly different. For simplicity, when we assume the tests are unrelated, 6 tests may produce a 26% chance of observing at least 1 significant result at the significance level of 0.05. Therefore, independent replication studies and meta-analyses integrating results from original and subsequent replication studies are indispensable to establish the credibility of the findings.22,23 We dichotomized aneurysm sites according to the grouping in the International Study of Unruptured Intracranial Aneurysm,5 but there may be controversy regarding the grouping of the PcomA aneurysms because the PcomA derives from the ICA.24 Therefore, both of the findings when IA patients are stratified according to the International Study of Unruptured Intracranial Aneurysm grouping and according to the common site will be corroborated.
One of the difficulties in genetic dissection of IA may be attributable to possible phenotypic heterogeneity.1,7 In most association studies, patients with different aneurysmal phenotypes were grouped together.7 If different predisposition underlies different subphenotypes of IA, then reducing phenotypic heterogeneity by restricting to specific subphenotype would be effective. We assessed whether the magnitude of the 9p21 association varied by aneurysmal subphenotypes. Our results of the 9p21 variation being associated with location of IA suggest that a genetic dissection of subphenotypes of IA is important and feasible. One of the most important aspects of IAs is whether they will rupture. The analytical methods implemented in this study may be useful to identify a genetic factor that allows prediction of aneurysmal rupture.
The authors thank all the study participants and supporting medical staff for making this study possible. The authors are grateful to Drs. Hideaki Onda, Taku Yoneyama, Hiroyuki Akagawa, Makiko Funamizu, and Midori Yamamoto for their contribution to sample and data collection. The authors thank Hiromi Kamura for her technical support.
Sources of Funding
This work was supported by KAKENHI (Grant-in-Aid for Scientific Research) on Priority Areas “Applied Genomics” from the Ministry of Education, Culture, Sports, Science, and Technology of Japan.
- Received April 16, 2010.
- Accepted May 13, 2010.
Bilguvar K, Yasuno K, Niemela M, Ruigrok YM, von Und Zu Fraunberg M, van Duijn CM, van den Berg LH, Mane S, Mason CE, Choi M, Gaal E, Bayri Y, Kolb L, Arlier Z, Ravuri S, Ronkainen A, Tajima A, Laakso A, Hata A, Kasuya H, Koivisto T, Rinne J, Ohman J, Breteler MM, Wijmenga C, State MW, Rinkel GJ, Hernesniemi J, Jaaskelainen JE, Palotie A, Inoue I, Lifton RP, Gunel M. Susceptibility loci for intracranial aneurysm in European and Japanese populations. Nat Genet. 2008; 40: 1472–1477.
Helgadottir A, Thorleifsson G, Magnusson KP, Gretarsdottir S, Steinthorsdottir V, Manolescu A, Jones GT, Rinkel GJ, Blankensteijn JD, Ronkainen A, Jaaskelainen JE, Kyo Y, Lenk GM, Sakalihasan N, Kostulas K, Gottsater A, Flex A, Stefansson H, Hansen T, Andersen G, Weinsheimer S, Borch-Johnsen K, Jorgensen T, Shah SH, Quyyumi AA, Granger CB, Reilly MP, Austin H, Levey AI, Vaccarino V, Palsdottir E, Walters GB, Jonsdottir T, Snorradottir S, Magnusdottir D, Gudmundsson G, Ferrell RE, Sveinbjornsdottir S, Hernesniemi J, Niemela M, Limet R, Andersen K, Sigurdsson G, Benediktsson R, Verhoeven EL, Teijink JA, Grobbee DE, Rader DJ, Collier DA, Pedersen O, Pola R, Hillert J, Lindblad B, Valdimarsson EM, Magnadottir HB, Wijmenga C, Tromp G, Baas AF, Ruigrok YM, van Rij AM, Kuivaniemi H, Powell JT, Matthiasson SE, Gulcher JR, Thorgeirsson G, Kong A, Thorsteinsdottir U, Stefansson K. The same sequence variant on 9p21 associates with myocardial infarction, abdominal aortic aneurysm and intracranial aneurysm. Nat Genet. 2008; 40: 217–224.
Pasmant E, Laurendeau I, Heron D, Vidaud M, Vidaud D, Bieche I. Characterization of a germ-line deletion, including the entire INK4/ARF locus, in a melanoma-neural system tumor family: identification of ANRIL, an antisense noncoding RNA whose expression coclusters with ARF. Cancer Res. 2007; 67: 3963–3969.
Wiebers DO, Whisnant JP, Huston J III, Meissner I, Brown RD Jr, Piepgras DG, Forbes GS, Thielen K, Nichols D, O'Fallon WM, Peacock J, Jaeger L, Kassell NF, Kongable-Beckman GL, Torner JC. International Study of Unruptured Intracranial Aneurysms Investigators. Unruptured intracranial aneurysms: natural history, clinical outcome, and risks of surgical and endovascular treatment. Lancet. 2003; 362: 103–110.
Wermer MJ, van der Schaaf IC, Algra A, Rinkel GJ. Risk of rupture of unruptured intracranial aneurysms in relation to patient and aneurysm characteristics: an updated meta-analysis. Stroke. 2007; 38: 1404–1410.
Ruigrok YM, Rinkel GJ. Genetics of intracranial aneurysms. Stroke. 2008; 39: 1049–1055.
McPherson R, Pertsemlidis A, Kavaslar N, Stewart A, Roberts R, Cox DR, Hinds DA, Pennacchio LA, Tybjaerg-Hansen A, Folsom AR, Boerwinkle E, Hobbs HH, Cohen JC. A common allele on chromosome 9 associated with coronary heart disease. Science. 2007; 316: 1488–1491.
Samani NJ, Erdmann J, Hall AS, Hengstenberg C, Mangino M, Mayer B, Dixon RJ, Meitinger T, Braund P, Wichmann HE, Barrett JH, Konig IR, Stevens SE, Szymczak S, Tregouet DA, Iles MM, Pahlke F, Pollard H, Lieb W, Cambien F, Fischer M, Ouwehand W, Blankenberg S, Balmforth AJ, Baessler A, Ball SG, Strom TM, Braenne I, Gieger C, Deloukas P, Tobin MD, Ziegler A, Thompson JR, Schunkert H. WTCCC and the Cardiogenics Consortium. Genomewide association analysis of coronary artery disease. N Engl J Med. 2007; 357: 443–453.
Broadbent HM, Peden JF, Lorkowski S, Goel A, Ongen H, Green F, Clarke R, Collins R, Franzosi MG, Tognoni G, Seedorf U, Rust S, Eriksson P, Hamsten A, Farrall M, Watkins H. PROCARDIS consortium. Susceptibility to coronary artery disease and diabetes is encoded by distinct, tightly linked SNP in the ANRIL locus on chromosome 9p. Hum Mol Genet. 2008; 17: 806–814.
Jarinova O, Stewart AF, Roberts R, Wells G, Lau P, Naing T, Buerki C, McLean BW, Cook RC, Parker JS, McPherson R. Functional analysis of the chromosome 9p21.3 coronary artery disease risk locus. Arterioscler Thromb Vasc Biol. 2009; 29: 1671–1677.
Begg CB, Zhang ZF. Statistical analysis of molecular epidemiology studies employing case-series. Cancer Epidemiol Biomarkers Prev. 1994; 3: 173–175.
Deka R, Koller DL, Lai D, Indugula SR, Sun G, Woo D, Sauerbeck L, Moomaw CJ, Hornung R, Connolly ES, Anderson C, Rouleau G, Meissner I, Bailey-Wilson JE, Huston Iii J, Brown RD, Kleindorfer DO, Flaherty ML, Langefeld CD, Foroud T, Broderick JP. The FIA Study Investigators. The relationship between smoking and replicated sequence variants on chromosomes 8 and 9 with familial intracranial aneurysm. Stroke. 2010; 41: 1132–1137.
Schunkert H, Gotz A, Braund P, McGinnis R, Tregouet DA, Mangino M, Linsel-Nitschke P, Cambien F, Hengstenberg C, Stark K, Blankenberg S, Tiret L, Ducimetiere P, Keniry A, Ghori MJ, Schreiber S, El Mokhtari NE, Hall AS, Dixon RJ, Goodall AH, Liptau H, Pollard H, Schwarz DF, Hothorn LA, Wichmann HE, Konig IR, Fischer M, Meisinger C, Ouwehand W, Deloukas P, Thompson JR, Erdmann J, Ziegler A, Samani NJ. Cardiogenics Consortium. Repeated replication and a prospective meta-analysis of the association between chromosome 9p21.3 and coronary artery disease. Circulation. 2008; 117: 1675–1684.
Lindner SH, Bor AS, Rinkel GJ. Differences in risk factors according to the site of intracranial aneurysms. J Neurol Neurosurg Psychiatry. 2010; 81: 116–118.