Ischemic heart disease (IHD) is the most common cause

Transcription

1 Genetic Variation in ABCA1 Predicts Ischemic Heart Disease in the General Population Ruth Frikke-Schmidt, Børge G. Nordestgaard, Gorm B. Jensen, Rolf Steffensen, Anne Tybjærg-Hansen Downloaded from by guest on May 9, 2018 Objective We tested the hypothesis that 6 nonsynonymous single nucleotide polymorphisms (SNPs) in ATP-Binding- Cassette transporter A1 (ABCA1) affect risk of ischemic heart disease (IHD) in the general population. Methods and Results We genotyped 9259 individuals from the Danish general population followed for 25 years. Two SNPs (V771M and V825I) were previously associated with increases in HDL-C, 1 (R1587K) with decreased HDL-C, whereas 3 (R219K, I883M and E1172D) did not affect HDL-C levels. Despite this, 5 out of 6 SNPs (V771M, V825I, I883M, E1172D, R1587K) predicted increased risk of IHD. Similar results were obtained in a verification sample with 932 IHD cases versus 7999 controls. A stepwise regression approach identified V771M, I883M, and E1172D as the most important predictors of IHD and additive effects on IHD risk were present for V771M/I883M and I883M/E1172D pairs. Conclusions We show that 3 of 6 nonsynonymous SNPs in ABCA1 predict risk of IHD in the general population. (Arterioscler Thromb Vasc Biol. 2008;28: ) Key Words: atherosclerosis cardiovascular diseases genetics lipids lipoproteins Ischemic heart disease (IHD) is the most common cause of death in developed countries, 1 and a low level of plasma high-density lipoprotein cholesterol (HDL-C) is a major risk factor for IHD in the general population. 2 An important atheroprotective property of the HDL particle is thought to be its key function in reverse cholesterol transport mobilizing cellular cholesterol from arterial wall macrophages to lipid poor plasma apolipoproteins, a transport that is mediated by the transmembrane ATP-Binding- Cassette transporter A1 (ABCA1). 3 Mutations in the ABCA1 gene cause rare mendelian HDL deficiency syndromes characterized in the homozygous form by almost complete absence of HDL particles in plasma, but only a modest if any increase in risk of IHD, and in the heterozygous form by half-normal HDL-C levels. 4 7 Recently, we and others have shown that both rare and common variants in the ABCA1 gene also contribute to HDL-C levels in the general population. 8,9 Whether common genetic variation in ABCA1 predicts risk of IHD in the general population is currently unknown. This study was performed to test the hypothesis that 6 nonsynonymous single nucleotide polymorphisms (SNPs) in ABCA1 affect risk of IHD in the general population. Methods Subjects General Population Sample The Copenhagen City Heart Study is a prospective study of the Danish general population initiated in 1976 to 1978 with follow-up examinations in 1981 to 1983 and 1991 to ,11 Individuals were selected based on the Central Population Register Code to reflect the adult Danish general population aged 20 to 80 years. In the present study we included 9259 individuals from the 1991 to 1994 examination, whom we genotyped for all nonsynonymous SNPs (R219K, V771M, V825I, I883M, E1172D, R1587K) identified by resequencing ABCA1 in 190 individuals of Danish ancestry. 8 Information on diagnosis of IHD (n 1170; World Health Organization; International Classification of Diseases, 8th edition: codes 410 to 414; 10th edition: codes I20-I25) was collected and verified until 31st December 2000 by reviewing all hospital admissions and diagnoses entered in the national Danish Patient Registry, all causes of death entered in the national Danish Causes of Death Registry, and medical records from hospitals and general practitioners. IHD was determined by experienced cardiologists according to the guidelines of the European Society of Cardiology on the basis of previous myocardial infarction or characteristic symptoms of angina pectoris based on location, character and duration of pain, and the relation of pain to exercise. 12 A diagnosis of myocardial infarction required the presence of at least 2 of the following criteria: characteristic chest pain, elevated cardiac enzymes, and electrocardiographic changes indicative of myocardial infarction. Original received February 27, 2007; final version accepted October 4, From the Department of Clinical Biochemistry (R.F.-S., A.T.-H.), Rigshospitalet, Copenhagen University Hospital, University of Copenhagen; the Department of Clinical Biochemistry (B.G.N.), Herlev University Hospital, University of Copenhagen; the Copenhagen City Heart Study (B.G.N., G.B.J., A.T.-H.), Bispebjerg University Hospital, University of Copenhagen; and the Department of Medicine B (R.S.), Hillerød Hospital, Hillerød, Denmark. Correspondence to Anne Tybjærg-Hansen, MD, DMSc, Chief Physician and Associate Professor, Department of Clinical Biochemistry KB 3011, Section for Molecular Genetics, Rigshospitalet, Copenhagen University Hospital, Blegdamsvej 9, DK-2100 Copenhagen Ø, Denmark. anne.tybjaerg.hansen@rh.regionh.dk 2007 American Heart Association, Inc. Arterioscler Thromb Vasc Biol. is available at DOI: /ATVBAHA

2 Frikke-Schmidt et al ABCA1 and Ischemic Heart Disease 181 Table 1. Characteristics of Individuals in the General Population (The Copenhagen City Heart Study) Women Men Downloaded from by guest on May 9, 2018 Participants Without IHD (n 4510) Patients With IHD A second population comprised 992 consecutive patients from the greater Copenhagen area referred for coronary angiography to Copenhagen University Hospital, during the period from 1991 through Of these, 948 (26% women) had documented IHD based on characteristic symptoms of angina pectoris, 12 plus at least 1 of the following: 70% stenosis of at least 1 coronary artery or 50% stenosis of the left main coronary artery on coronary angiography (n 767), a previous myocardial infarction (n 494), or a positive exercise electrocardiography test. Studies were approved by institutional review boards and Danish ethical committees: Nos /91, Copenhagen and Frederiksberg committee, and KA 93125, Copenhagen County committee, and conducted according to the Declaration of Helsinki. Informed consent was obtained from participants. Roughly 99% were white and of Danish descent. Study Designs Prospective Study A prospective study was conducted using IHD as end point. The cohort was participants in The Copenhagen City Heart Study who attended the 1991 to 1994 examination, and for whom combined genotypes for all 6 nonsynonymous SNPs (R219K, V771M, V825I, I883M, E1172D, R1587K) were available as well as all clinical and biochemical data (n 9028 out of n 9259; Table 1). All end points were recorded in the follow-up period 1976 to Follow-up time was 25 years ( person-years). Individuals diagnosed with IHD before entry into the study (n 63) were excluded, leaving 1107 incident IHD cases (n 455 women and n 652 men), and a total of 8965 individuals for all further analyses. Follow-up was 100% complete. Please see Data Supplements, Expanded Methods for description of risk factors. Case Control Study (Verification Sample) To retest whether SNP genotype was associated with risk of IHD in an independent sample, a case control study was conducted. The cases were consecutive patients referred for coronary angiography at Copenhagen University Hospital in 1991 through 1994 with documented IHD (see above), and for whom genotypes for all 6 nonsynonymous SNPs were available as well as all clinical and biochemical data (n 932 out of n 948). Patients with IHD (n 245 women and n 687 men) were compared with unmatched controls from the general population without IHD (n 4598 women and n 3401 men). Participants With IHD (n 455) Participants Without IHD (n 3348) Participants With IHD (n 652) Age, years * * Cholesterol, mmol/l * * LDL cholesterol, mmol/l * * Apolipoprotein B, mg/dl * * HDL cholesterol, mmol/l * * Apolipoprotein AI, mg/dl Triglycerides, mmol/l * * Body mass index, kg/m * * Smoking, % * Diabetes mellitus, % 3 9* 5 11* Hypertension, % 48 77* 56 76* Cholesterol lowering therapy, % 0.5 3* 0.5 4* Values are mean SEM or percentage. IHD indicates ischemic heart disease. Participants without IHD were compared with participants with IHD by Mann Whitney U test or Pearson 2 test. *P 0.001; P SNP Genotyping The ABI PRISM 7900HT Sequence Detection System (Applied Biosystem Inc) was used to genotype for all 6 nonsynonymous SNPs (R219K, V771M, V825I, I883M, E1172D, R1587K) identified by resequencing ABCA1 in 190 individuals, as previously described. 8 Biochemical Analyses Colorimetric and turbidimetric assays (Hitachi autoanalyzer) were used to measure plasma levels of total cholesterol, HDL-C, triglycerides, and apolipoproteins B and -AI (all Boehringer Mannheim GmbH). Statistical Analyses We used the statistical software package Stata (STATA Corp). Two-sided probability values 0.05 were considered significant. Pearson 2 test, t test, and Mann Whitney U test were used for 2-group comparisons. Kaplan Meier curves and log-rank tests evaluated the cumulative incidence of IHD as a function of SNP genotypes. With the use of left truncation (or delayed entry), Cox proportional hazards regression models with age as time scale estimated HRs for IHD. 13 Covariates were dichotomized (smoking, diabetes, hypertension), or tertilized (body mass index, HDL-C and low-density lipoprotein cholesterol [LDL-C]), and incorporated into the regression models. A stepwise backward Cox regression model was used to select the best set of SNPs for prediction of IHD. To account for Type I error, a verification strategy was applied in an independent case-control study. Logistic regression was used for the verification study. For expanded Statistical Analyses please see Data Supplements, Expanded Methods. Results Characteristics of Individuals in the Prospective Study The general population cohort (The Copenhagen City Heart Study) experienced 1107 incident IHD events during 25 years of follow-up and person-years. All risk factors were increased in individuals who later developed IHD compared with those without IHD (Table 1, all probability values 0.05). Allele Frequencies Minor allele frequencies in the general population for the 6 nonsynonymous SNPs previously identified by resequencing

3 182 Arterioscler Thromb Vasc Biol. January 2008 Downloaded from by guest on May 9, 2018 Figure 1. Pairwise linkage disequilibria between 6 nonsynonymous ABCA1 SNPs. r 2 values below and D values above the gray diagonal. indicates negative D. the entire ABCA1 gene in 190 individuals of Danish descent were: 0.26 (R219K), 0.03 (V771M), 0.03 (E1172D), 0.06 (V825I), 0.12 (I883M), and 0.24 (R1587K). 8 Similar allele frequencies have been reported for other White populations. 14 Please see Data Supplements, Expanded Results for description of Hardy-Weinberg equilibrium. Linkage Disequilibrium and Haplotypes Figure 1 illustrates the pairwise linkage disequilibrium (LD) pattern among the six nonsynomymous SNPs in ABCA1, with D above and r 2 below the diagonal. A strong positive D was present for R219K/V771M, M825I/I883M, and E1172D/ R1587K (D 0.9), whereas a strong negative D was present for R219K/V825I, V771M/V825I, and V771M/I883M (D 0.9). The r 2 for the V825/I883M pair was 0.44 whereas the remaining SNP pairs had r (Figure 1). Haplotypes are described in the Data Supplements, Expanded Results and Table III. Risk of IHD Prospective Study The cumulative incidence of IHD as a function of age was increased for V771M (GA AA versus GG, borderline P 0.06), V825I (GA AA versus GG; P 0.02), I883M (AG GG versus AA, P 0.01), E1172D (GC CC versus GG, P 0.03), and for R1587K (AA versus GG, borderline P 0.06), but not for R219K (Figure 2). The age adjusted hazard ratios (HRs) for IHD were: V771M (GA AA versus GG) 1.2 (95% confidence interval [CI] 1.0 to 1.5), V825I (GA AA versus GG) 1.2 (1.0 to 1.5), I883M (AG GG versus AA) 1.2 (1.0 to 1.4), E1172D (GC CC versus GG) 1.3 (1.0 to 1.6) and R1587K (AA versus GG) 1.2 (1.0 to 1.6), respectively (Table 2). Adjusting for HDL-C or for LDL-C, smoking, diabetes, hypertension, and body mass index did not Figure 2. Cumulative incidence of IHD by ABCA1 genotypes. Probability values are for overall log-rank tests. For V771M, V825I, I883M, and E1172D, heterozygotes and homozygotes for the variant allele were pooled (combined genotype in red, common genotype in green). For R219K and R1587K, all 3 genotypes were presented (homozygote in red, heterozygote in black, common genotype in green).

4 Frikke-Schmidt et al ABCA1 and Ischemic Heart Disease 183 Table 2. Risk of Ischemic Heart Disease as a Function of ABCA1 Genotype in the Prospective Study Downloaded from by guest on May 9, 2018 Number of Events Incidence Rate (95% CI) Hazard Ratio (95% CI) n, % Observed Expected per Person-Years Age Adjusted HDL-C Adjusted Multifactorially Adjusted R219K GG 4863 (54) (59 70) GA 3461 (39) (58 71) 1.0 ( ) 1.0 ( ) 1.0 ( ) AA 641 (7) (50 80) 1.0 ( ) 1.0 ( ) 1.0 ( ) V771M GG 8379 (93) (60 68) GA AA 586 (7) (60 93)* 1.2 ( ) 1.3 ( ) 1.2 ( ) M825I GG 7959 (89) (59 67) GA AA 1006 (11) (64 89) 1.2 ( ) 1.2 ( ) 1.2 ( ) I883M AA 6934 (77) (58 66) AG GG 2031 (23) (64 81) 1.2 ( ) 1.2 ( ) 1.2 ( ) E1172D GG 8469 (94) (59 67) GC CC 496 (6) (68 106) 1.3 ( ) 1.3 ( ) 1.3 ( ) R1587K GG 5216 (58) (58 67) GA 3213 (36) (58 71) 1.0 ( ) 1.0 ( ) 1.0 ( ) AA 536 (6) (65 102)* 1.2 ( ) 1.2 ( ) 1.3 ( ) Nonincident cases (n 63) were excluded, leaving 8965 individuals for the survival analyses and Cox regression. CI indicates confidence interval; HDL, high-density lipoprotein; LDL, low-density lipoprotein. In all 3 Cox regression models (age-adjusted, HDL cholesterol adjusted, and multifactorially adjusted) age is adjusted for by incorporating age in the baseline hazard function (left truncation). HDL cholesterol adjusted: HDL cholesterol in tertiles was adjusted for in the Cox regression model. Multifactorially adjusted: Smoking, diabetes mellitus, hypertension dichotomized, and body mass index and LDL cholesterol in tertiles were adjusted for in the Cox regression model. Log-rank test vs most common genotype: *P 0.06; P Rs numbers: R219K (rs ); V771M (rs ); V825I (rs ); I883M (rs ); E1172D (rs ); R1587K (rs ). alter these HRs substantially. There was evidence for a statistically significant interaction between gender and V771M (P 0.04) in the prediction of IHD, whereas the remaining 5 SNPs did not interact with gender (all probability values 0.52). Please see all gender specific cumulative incidence plots and HRs in the Data Supplements, Figure I and Table I. Case Control Study (Verification Sample) The significant effects in the prospective study were retested in an independent case-control study comprising 932 cases with IHD and 7999 controls without IHD. The age-adjusted odds ratios (ORs) were: V771M (GA AA versus GG) 1.2 (0.9 to 1.5), V825I (GA AA versus GG) 1.2 (0.9 to 1.4), I883M (AG GG versus AA) 1.2 (1.0 to 1.4), E1172D (GC CC versus GG) 1.1 (0.8 to 1.4), and R1587K (GA versus GG) 1.2 (1.0 to 1.4). Please see the Data Supplements, Table II for frequencies and gender specific results. Stepwise Regression Approach To determine which ABCA1 SNPs were independent predictors and not caused by LD among SNPs, a stepwise Cox regression approach was performed. The final prediction model included the 3 noncorrelated SNPs, V771M, I883M, and E1172D (HRs: V771M, 1.2 [1.0 to 1.6]; I883M, 1.2 [1.0 to 1.4]; E1172D, 1.2 [1.0 to 1.6]). Performing gender specific stepwise regression, V771M and I883M were the best predictors in women (HRs: V771M, 1.6 [1.2 to 2.3]; I883M, 1.3 [1.1 to 1.6]), whereas I883M and E1172D were best in men (HRs: I883M, 1.1 (1.0 to 1.4); E1172D, 1.3 (1.0 to 1.7). Using standard statistical tests, there was no evidence for interactions among pairs of the 3 SNPs identified from the stepwise regression approach (all probability values 0.21). However, because additive effects of these significant SNPs might have clinical relevance, the pairwise combinations of these 3 SNPs are shown and described in the Data Supplements, Expanded Results and Figure II. ABCA1 SNPs and HDL-C Levels In genders combined, V771M and V825I were associated with increases in HDL-C of 0.04 and 0.05 mmol/l, respectively (Figure 3, upper panel). R1587K was associated with a decrease in HDL-C of 0.03 mmol/l (P 0.005), whereas R219K, I883M, and E1172D were not associated with changes in HDL-C levels (Figure 3, upper panel). Gender specific effects are shown in middle and lower panels of Figure 3.

5 184 Arterioscler Thromb Vasc Biol. January 2008 Downloaded from by guest on May 9, 2018 Figure 3. Relative increase or decrease in plasma HDL-C as a function of ABCA1 SNPs. Heterozygotes and homozygotes for the variant allele were pooled for all 6 SNPs. *P 0.05, **P Discussion The principal finding of this study is that common genetic variation in ABCA1, a well-known HDL gene, predicts risk of IHD in the general population. Although most SNPs had significant effects on plasma HDL-C levels, the association between ABCA1 SNPs and risk of IHD was not related to their association with HDL-C. Mechanistically, the results are highly plausible. ABCA1 is a transmembrane protein of key importance for reverse cholesterol transport, 3 a major pathway for removing excess cholesterol from peripheral tissues back to the liver. When cholesterol accumulates in the arterial wall atherosclerosis develops and may eventually lead to IHD and possibly to early death. ABCA1 in macrophages within the arterial wall facilitates removal of cholesterol and is likely to have antiatherogenic effects. Therefore, genetic variation in ABCA1 could influence the speed by which atherosclerosis develops, and thus the risk of IHD, exactly as was observed in the present study. Several studies have reported associations between V825I/ I883M and increased plasma HDL-C levels 8,15,16 and associations between R1587K and a decreased HDL-C or apoai. 8,14,17 For in silico prediction of ABCA1 SNPs please see the Data Supplements Expanded Results. The rare allele of the R219K SNP has previously been reported to be associated with decreased 14,17 or increased risk of coronary heart disease. 18 The present largest prospective results as well as a recent case control study did not find any association with atherosclerosis susceptibility. 19 In support of I883M and E1172D as 2 of the 3 most important ABCA1 SNPs for IHD prediction, a previous study of 2028 White men found increased frequencies in cases compared with controls, and also detected increased risk of future IHD events associated with the 1172D allele. 18 The pairwise LD structure of the SNPs is important for the interpretation of the association results. Because the effects on IHD are modest and concern almost all SNPs, one might suspect that this is attributable to LD. Where D is useful in detecting recombination and haplotype structure of the gene, r 2 describes how closely correlated the genotype is between a pair of SNPs and is thus a useful statistic for determining whether an effect associated with one SNP could be detected by genotyping a second SNP. 20 In general, the present ABCA1 SNPs are weakly correlated and do not serve as good genotype markers for each other (all r 2 0.8). There seem, however, to be several smaller relatively nonrecombinant regions as illustrated by the high D for several of the SNP pairs, either positive (the 2 rare alleles segregate together) or negative (the rare allele at one SNP segregates together with the common allele at the second SNP). The fact that the stepwise regression approach identified the V771M, I883M, and E1172D SNPs, which have very weak association between rare alleles (low r 2 or negative D ), as important for the final IHD prediction model, supports that the observed findings for these 3 SNPs are not caused by LD. However, the single site result on IHD risk for V825I is most likely attributable to LD with I883M, and the findings for R1587K are most likely attributable to LD with E1172D, the latter also supported by the haplotype analysis presented in the Data Supplements, Table III and Expanded Results. In the present study, risk of IHD predicted by SNPs in ABCA1 was independent of plasma HDL-C levels: SNPs predicting increased risk were associated with either increases or decreases in HDL-C, or no effect on HDL-C. In most cases the risk associated with genetic variation in ABCA1 was not explained by the inverse relationship between HDL-C and risk of IHD, as observed for other important genes in reverse cholesterol transport, cholesteryl

6 Frikke-Schmidt et al ABCA1 and Ischemic Heart Disease 185 Downloaded from by guest on May 9, 2018 ester transfer protein, 21,22 and hepatic lipase. 23,24 This is in good agreement with experimental studies, because several observations favor antiatherosclerotic effects of ABCA1 beyond plasma HDL-C levels: (1) In mice, macrophage specific deficiency of ABCA1 had a minimal effect on reducing plasma HDL-C levels, but resulted in significantly increased atherosclerosis. 25,26 Thus, macrophage ABCA1 appears to contribute little to bulk lipidation of HDL and therefore to plasma HDL-C levels, 27 but seems to be important in the prevention of atherosclerotic lesion development locally in the arterial wall. 25 (2) Macrophage ABCA1 appears to possess antiinflammatory properties 28 mediating engulfment of apoptotic cells by macrophages. 29,30 (3) Platelet ABCA1 might be involved in the modulation and suppression of platelet activation. 31 (4) ABCA1 mediates efflux of cellular phospholipids and cholesterol to lipid-poor apolipoproteins, such as apoai and apoe, but in contrast to ABCG1 and ABCG4, it interacts poorly with HDL-2 and HDL-3 particles that constitute the bulk of plasma HDL. 32 (5) Tangier disease (homozygosity or compound heterozygosity for mutations in ABCA1) is associated with extremely low levels of HDL-C, but despite this with only moderate increases in risk of IHD. 4 This is only partly explained by the low levels of LDL-C in some of these patients, and might suggest that risk of IHD attributable to ABCA1 mutations may not, or may only partly, be reflected in HDL-C levels in these individuals as well. (6) We have recently shown that a common mutation (frequency 0.4%) in ABCA1, K776N, predicted a 2- to 3-fold increase in risk in the general population independent of HDL-C levels. 33 (7) Finally, in the present study adjustments for HDL-C levels did not substantially alter the risk estimates, supporting that ABCA1 has effects on risk of IHD independent of HDL-C levels. Taken together, these results suggest that mutations in ABCA1 may have proatherosclerotic effects independent of HDL-C levels. Conclusion We show that 3 of 6 nonsynonymous SNPs in ABCA1 predict risk of IHD in the general population. Acknowledgments We thank Mette Refstrup for her persistent attention to the details of the large-scale genotyping. We also thank the subjects who participated in the study. Sources of Funding This work was supported by The Danish Heart Foundation, The Danish Medical Research Council, Ingeborg, and Leo Dannin s Grant, the Research Fund at Rigshospitalet, Copenhagen University Hospital, and a Specific Targeted Research Project grant from the European Union, Sixth Framework Programme Priority [FP LIFESCIHEALTH-6], contract # None. Disclosures References 1. Murray CJL, Lopez AD. Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet. 1997;349: Stampfer MJ, Sacks FM, Salvini S, Willett WC, Hennekens CH. A prospective study of cholesterol, apolipoproteins, and the risk of myocardial infarction. N Engl J Med. 1991;325: Tall AR, Breslow JL, Rubin EM. Genetic disorders affecting plasma high-density lipoproteins. In: Scriver CR, Beaudet AL, Valle D, Sly WS, eds. The Metabolic and Molecular Bases of Inherited Disease. 8th ed. New York: McGraw-Hill; p Assman G, von Eckardstein A, Bryan Brewer H Jr. Familial analphalipoproteinemia: Tangier disease. In: Scriver CR, Beaudet AL, Valle D, Sly WS, eds. The Metabolic and Molecular Bases of Inherited Disease. 8th ed. New York: McGraw-Hill; p Rust S, Rosier M, Funke H, Real J, Amoura Z, Piette JC, Deleuze JF, Brewer HB, Duverger N, Denefle P, Assmann G. Tangier disease is caused by mutations in the gene encoding ATP-binding cassette transporter 1. Nat Genet. 1999;22: Bodzioch M, Orso E, Klucken J, Langmann T, Bottcher A, Diederich W, Drobnik W, Barlage S, Buchler C, Porsch-Ozcurumez M, Kaminski WE, Hahmann HW, Oette K, Rothe G, Aslanidis C, Lackner KJ, Schmitz G. The gene encoding ATP-binding cassette transporter 1 is mutated in Tangier disease. Nat Genet. 1999;22: Brooks-Wilson A, Marcil M, Clee SM, Zhang LH, Roomp K, van Dam M, Yu L, Brewer C, Collins JA, Molhuizen HO, Loubser O, Ouelette BF, Fichter K, Ashbourne-Excoffon KJ, Sensen CW, Scherer S, Mott S, Denis M, Martindale D, Frohlich J, Morgan K, Koop B, Pimstone S, Kastelein JJ, Hayden MR. Mutations in ABC1 in Tangier disease and familial high-density lipoprotein deficiency. Nat Genet. 1999;22: Frikke-Schmidt R, Nordestgaard BG, Jensen GB, Tybjærg-Hansen A. Genetic variation in ABC transporter A1 contributes to HDL-cholesterol in the general population. J Clin Invest. 2004;114: Cohen JC, Kiss RS, Pertsemlidis A, Marcel YL, McPherson R, Hobbs HH. Multiple rare alleles contribute to low plasma levels of HDL cholesterol. Science. 2004;305: Appleyard M, Hansen AT, Jensen G, Schnohr P, Nyboe J. The Copenhagen City Heart Study. Østerbroundersøgelsen. A book of tables with data from the first examination ( ) and a five year follow-up ( ). The Copenhagen City Heart Study Group. Scand J Soc Med Suppl. 1989;41: Schnohr P, Jensen G, Lange P, Scharling H, Appleyard M. The Copenhagen City Heart Study, Østerbroundersøgelsen, tables with data from the third examination Eur Heart J. 2001;3(Suppl H): Julian DG, Bertrand ME, Hjalmarson A, Fox K, Simoons ML, Ceremuzynski L, Masere A, Mernertz T, Meyer J, Pyorala K, Rehnqvist N, Tavazzi L, Toutouzas P, Treasure T. Management of stable angina pectoris - Recommendations of the Task Force of the European Society of Cardiology. Eur Heart J. 1997;18: Klein JP, Moeschberger ML. Refinements of the semiparametric proportional hazards model. Survival Analysis. Techniques for Censored and Truncated Data. 2nd ed. New York: Springer-Verlag, Inc.; p Tregouet DA, Ricard S, Nicaud V, Arnould I, Soubigou S, Rosier M, Duverger N, Poirier O, Mace S, Kee F, Morrison C, Denefle P, Tiret L, Evans A, Deleuze JF, Cambien F. In-depth haplotype analysis of ABCA1 gene polymorphisms in relation to plasma apoa1 levels and myocardial infarction. Arterioscler Thromb Vasc Biol. 2004;24: Wang J, Burnett JR, Near S, Young K, Zinman B, Hanley AJ, Connelly PW, Harris SB, Hegele RA. Common and rare ABCA1 variants affecting plasma HDL cholesterol. Arterioscler Thromb Vasc Biol. 2000;20: Harada T, Imai Y, Nojiri T, Morita H, Hayashi D, Maemura K, Fukino K, Kawanami D, Nishimura G, Tsushima K, Monzen K, Yamazaki T, Mitsuyama S, Shintani T, Watanabe N, Seto K, Sugiyama T, Nakamura F, Ohno M, Hirata Y, Yamazaki T, Nagai R. A common Ile 823 Met variant of ATP-binding cassette transporter A1 gene (ABCA1) alters high density lipoprotein cholesterol level in Japanese population. Atherosclerosis. 2003;169: Clee SM, Zwinderman AH, Engert JC, Zwarts KY, Molhuizen HO, Roomp K, Jukema JW, van Wijland M, van Dam M, Hudson TJ, Brooks- Wilson A, Genest JJ, Kastelein JJ, Hayden MR. Common genetic variation in ABCA1 is associated with altered lipoprotein levels and a modified risk for coronary artery disease. Circulation. 2001;103: Brousseau ME, Bodzioch M, Schaefer EJ, Goldkamp AL, Kielar D, Probst M, Ordovas JM, Aslanidis C, Lackner KJ, Bloomfield RH, Collins D, Robins SJ, Wilson PW, Schmitz G. Common variants in the gene encoding ATP-binding cassette transporter 1 in men with low HDL cholesterol levels and coronary heart disease. Atherosclerosis. 2001;154:

7 186 Arterioscler Thromb Vasc Biol. January 2008 Downloaded from by guest on May 9, Morgan TM, Krumholz HM, Lifton RP, Spertus JA. Nonvalidation of reported genetic risk factors for acute coronary syndrome in a large-scale replication study. JAMA. 2007;297: Jarvik GP, Hatsukami TS, Carlson C, Richter RJ, Jampsa R, Brophy VH, Margolin S, Rieder M, Nickerson D, Schellenberg GD, Heagerty PJ, Furlong CE. Paraoxonase activity, but not haplotype utilizing the linkage disequilibrium structure, predicts vascular disease. Arterioscler Thromb Vasc Biol. 2003;23: Borggreve SE, Hillege HL, Wolffenbuttel BHR, de Jong PE, Zuurman MW, van der Steege G, van Tol A, Dullaart RPF, on behalf of the PREVEND Study Group. An increased coronary risk is paradoxically associated with common cholesteryl ester transfer protein gene variations that relate to higher high-density lipoprotein cholesterol: A population-based study. J Clin Endocrinol Metab. 2006;91: Agerholm-Larsen B, Nordestgaard BG, Steffensen R, Jensen G, Tybjærg- Hansen A. Elevated HDL cholesterol is a risk factor for ischemic heart disease in white women when caused by a common mutation in the cholesteryl ester transfer protein gene. Circulation. 2000;101: Andersen RV, Wittrup HH, Tybjærg-Hansen A, Steffensen R, Schnohr P, Nordestgaard BG. Hepatic lipase mutations, elevated high-density lipoprotein cholesterol, and increased risk of ischemic heart disease: the Copenhagen City Heart Study. J Am Coll Cardiol. 2003;41: Hokanson JE, Cheng S, Snell-Bergeon JK, Fijal BA, Grow MA, Hung C, Erlich HA, Ehrlich J, Eckel RH, Rewers M. A common promoter polymorphism in the hepatic lipase gene (LIPC-480C T) is associated with an increase in coronary calcification in type 1 diabetes. Diabetes. 2002; 51: Van Eck M, Bos IS, Kaminski WE, Orso E, Rothe G, Twisk J, Bottcher A, Van Amersfoort ES, Christiansen-Weber TA, Fung-Leung WP, Van Berkel TJC, Schmitz G. Leukocyte ABCA1 controls susceptibility to atherosclerosis and macrophage recruitment into tissues. Proc Natl Acad Sci U S A. 2002;99: Aiello RJ, Brees D, Bourassa PA, Royer L, Lindsey S, Coskran T, Haghpassand M, Francone OL. Increased atherosclerosis in hyperlipidemic mice with inactivation of ABCA1 in macrophages. Arterioscler Thromb Vasc Biol. 2002;22: Haghpassand M, Bourassa PA, Francone OL, Aiello RJ. Monocyte/macrophage expression of ABCA1 has minimal contribution to plasma HDL levels. J Clin Invest. 2001;108: Schmitz G, Kaminski WE, Porsch-Ozcurumez M, Klucken J, Orso E, Bodzioch M, Buchler C, Drobnik W. ATP-Binding Cassette transporter A1 (ABCA1) in macrophages: A dual function in inflammation and lipid metabolism? Pathobiology. 1999;67: Hamon Y, Broccardo C, Chambenoit O, Luciani MF, Toti F, Chaslin S, Freyssinet J-M, Devaux PF, McNeish J, Marguet D, Chimini G. ABC1 promotes engulfment of apoptotic cells and transbilayer redistribution of phosphatidylserine. Nat Cell Biol. 2000;2: Luciani MF, Chimini G. The ATP binding cassette transporter ABC1, is required for engulfment of corpses generated by apoptotic cell death. EMBO J. 1996;15: Schmitz G, Buechler C. ABCA1: regulation, trafficking and association with heteromeric proteins. Ann Med. 2002;34: Wang N, Lan D, Chen W, Matsuura F, Tall AR. ATP-binding cassette transporters G1 and G4 mediate cellular cholesterol efflux to high-density lipoproteins. Proc Natl Acad Sci U S A. 2004;101: Frikke-Schmidt R, Nordestgaard BG, Schnohr P, Steffensen R, Tybjærg- Hansen A. Mutation in ABCA1 predicted risk of ischemic heart disease in the Copenhagen City Heart Study Population. J Am Coll Cardiol. 2005;46:

8 Downloaded from by guest on May 9, 2018 Genetic Variation in ABCA1 Predicts Ischemic Heart Disease in the General Population Ruth Frikke-Schmidt, Børge G. Nordestgaard, Gorm B. Jensen, Rolf Steffensen and Anne Tybjærg-Hansen Arterioscler Thromb Vasc Biol. 2008;28: ; originally published online October 19, 2007; doi: /ATVBAHA Arteriosclerosis, Thrombosis, and Vascular Biology is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX Copyright 2007 American Heart Association, Inc. All rights reserved. Print ISSN: Online ISSN: The online version of this article, along with updated information and services, is located on the World Wide Web at: Data Supplement (unedited) at: Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Arteriosclerosis, Thrombosis, and Vascular Biology can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: Subscriptions: Information about subscribing to Arteriosclerosis, Thrombosis, and Vascular Biology is online at:

9 Data Supplements Expanded Methods Statistical analyses Pairwise LD, haplotype frequencies, and associations were estimated using Haploview ( The Haploview program uses a two marker expectation maximization (EM) algorithm to estimate the maximum-likelihood values of the four gamete frequencies 1, from which the correlation coefficient r 2 and D is calculated 2. Risk factors in the prospective study The risk factors, diabetes mellitus, smoking and hypertension were dichotomized and defined as ever-diabetics (self-reported disease, use of insulin, use of oral hypoglycaemic drugs and/or nonfasting plasma glucose 11.1 mmol/l at any of the three examinations), ever-smokers (ex-smoker or current smoker at any of the three examinations), ever-hypertensives (systolic blood pressure 140 mmhg or diastolic blood pressure 90 mmhg and/or use of antihypertensive drugs at any of the three examinations). Body mass index was tertilized, and was obtained from the examination of the Copenhagen City Heart Study where DNA was collected. Expanded Results Hardy-Weinberg equilibrium The relative genotype frequencies of R219K, V771M, V825I, I883M, and R1587K were in Hardy- Weinberg equilibrium (P-values ranging from ). The relative genotype frequencies for E1172D deviated from Hardy-Weinberg equilibrium, due to a larger number of homozygotes 1

10 observed for the minor allele than expected (15 observed vs. 7 expected, P=0.004). For this reason, we sequenced 299 basepairs spanning this SNP in all 15 homozygotes and all 490 heterozygotes, and found 100% concordance with our previous results obtained by Taqman analyses. Pairwise combinations of ABCA1 SNPs identified from a stepwise Cox regression approach as independent predictors of IHD Pairwise combinations of the important SNPs identified from the stepwise Cox regression were V771M/I883M, V771M/E1172D, and I883M/E1172D, and additive effects were detected for the V771M/I883M and I883M/E1172D pairs as described in the following. In genders combined, the cumulative incidence of IHD as a function of age was increased for the combined genotypes V771M/I883M GGAG and GAAG versus the most common genotype GGAA (log-rank test: P=0.02 and P=0.008, respectively), and the corresponding multifactorially adjusted HRs were 1.2 (95% confidence interval (CI) ) and 1.9 ( ), respectively (Figure II, upper left panel). At the age of 80 years, about 30% and 40%, respectively, of V771M/I883M GGAG and GAAG carriers had IHD compared with about 20% for the common GGAA genotype. The cumulative incidence was also increased for the combined genotypes I883M/E1172D AGGG, AAGC, and AGCC versus the most common genotype AAGG (log-rank test: P=0.008, P=0.02, and P=0.04, respectively), and the corresponding multifactorially adjusted HRs were 1.2 ( ), 1.4 ( ) and 3.2 (CI ), respectively (Figure II, upper right panel). At the age of 80 years, about 26%, 30%, and 60%, respectively, of I883M/E1172D AGGG, AAGC, and AGCC carriers had IHD compared with about 20% for the common AAGG genotype. For significant gender specific effects, please see Figure II, lower left and right panels. 2

11 Case-control study (verification sample) In genders combined, the increased risks associated with the V771M/I883M GGAG genotype and with the I883M/E1172D AGGG genotype (Data Supplements Figure II, upper left and right panels) were verified in the case-control study (V771M/I883M GGAG vs. GGAA, odds ratio (OR) 1.2 ( ); I883M/E1172D AGGG vs. AAGG, OR 1.2 ( )). In women the increased risk associated with the V771M/I883M GGAG genotype (Data Supplements Figure II, lower left panel) was verified in the case-control study (V771M/I883M GGAG vs. GGAA, OR 1.4 (95% CI ), whereas the increased risks for men associated with the I883M/E1172D AAGC (Data Supplements Figure II, lower right panel) showed a similar trend in the case-control study (I883M/E1172D AAGC vs. AAGG, OR 1.3 ( )). Haplotype association testing Haplotype frequencies were similar to those reported for other Caucacian populations 3. Haplotype analyses identified two risk increasing haplotypes, one carrying the minor alleles of E1172D and R1587K (GGGAGG versus GGGACA, P=0.004/0.03 after 1000 permutations), and the other the minor alleles of M825I, I883M and R1587K (GGGAGG versus GGAGGA, P=0.04/0.33) (Table III). After gender stratification, these results were only present in men. Because the two rare alleles of E1172D and R1587K segregate together (D =0.95), the effect of the E1172D SNP obtained from the stepwise Cox regression, was detected in the haplotype analyses, also exclusively in men. A similar situation occurred with the second significant haplotype, GGAGGA, where the effect of the I883M SNP obtained from the stepwise Cox regression, was detected for this particular haplotype. In support of the results from the stepwise regression approach, that the 1587K allele did not independently predict risk, the haplotype harbouring the rare allele of the R1587K with the common alleles of the five remaining SNPs (GGGAGA), was not associated with any increased risk. The 3

12 effects of the V771M and I883M SNPs obtained from the stepwise Cox regression were not detected by the haplotype analyses, because rare alleles of these SNPs were located on several different haplotypes (Table III and Table 5 in reference 4). In contrast to the stepwise Cox regression approach, the haplotype association test is obviously only sensitive when there is high LD between SNPs in a given gene. The power of haplotype analysis decreases when SNPs are located on several different haplotypes as in the present study for V771M and I883M. Identification of all possible functional candidate SNPs using HapMap data Based on our own resequencing data in 190 white individuals of Danish ancestry, and on the Caucasian part of HapMap ( we can exclude that the observed effects of these six nonsynonymous SNPs are due to linkage disequilibrium with other nonsynonymous SNPs, with 5 promoter variants, or with variants in well-established splice donor or splice acceptor sites 4. In silico prediction of ABCA1 SNPs All six ABCA1 SNPs were predicted to be functionally benign using PolyPhen (Polymorphism Phenotyping; and PANTHER 5, 6, perhaps with the exception of V771M which had a substitution position-specific evolutionary (subpsec) score of 2.86, suggestive of a functional effect 5. For three of these SNPs, R219K, V771M and I883M, cholesterol efflux to apolipoprotein AI was determined in mutagenesis studies in vitro 5. In contrast to both the in vivo results in our previous study which suggest that V825I and not I883M increases high density lipoprotein cholesterol (HDL-C) 4, and the in silico predictions by both PolyPhen and PANTHER, I883M was associated with a reduction in cholesterol efflux. Cholesterol efflux was normal for R219K and V771M, although with a tendency towards an increase for V771M 5. Based 4

13 on these findings and the findings in our study, one might speculate that V771M causes a functional change in ABCA1 associated with increased cholesterol efflux and reflected in an increase in HDL- C in plasma. 5

14 References (1) Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005;21: (2) Pritchard JK, Przeworski M. Linkage disequilibrium in humans: models and data. Am J Hum Genet 2001;69:1-14. (3) Tregouet DA, Ricard S, Nicaud V, Arnould I, Soubigou S, Rosier M, Duverger N, Poirier O, Mace S, Kee F, Morrison C, Denefle P, Tiret L, Evans A, Deleuze JF, Cambien F. In-depth haplotype analysis of ABCA1 gene polymorphisms in relation to plasma apoa1 levels and myocardial infarction. Arterioscler Thromb Vasc Biol 2004;24: (4) Frikke-Schmidt R, Nordestgaard BG, Jensen GB, Tybjærg-Hansen A. Genetic variation in ABC transporter A1 contributes to HDL-cholesterol in the general population. J Clin Invest 2004;114: (5) Brunham LR, Singaraja R, Pape TD, Kejariwal A, Thomas PD, Hayden MR. Accurate prediction of the functional significance of single nucleotide polymorphisms and mutations in the ABCA1 gene. PLoS Genetics 2005;1: (6) Thomas PD, Campbell MJ, Kejariwal A, Mi H, Karlak B, Daverman R, Diemer K, Muruganujan A, Narechania A. PANTHER: A library of protein families and subfamilies indexed by function. Genome Res 2003;13:

15 Data Supplements Table I. Gender specific risk of ischemic heart disease as a function of ABCA1 genotype in the prospective study. Number of events Incidence Rate (95% Hazard Ratio (95% CI) Observed Expected CI) per 10,000 person-years Age adjusted HDL-C adjusted Multifactorially adjusted R219K N (%) Women GG 2,658 (54) (39-51) GA 1,967 (40) (41-55) 1.1 ( ) 1.1 ( ) 1.1 ( ) AA 340 (7) (35-71) 1.1 ( ) 1.1 ( ) 1.1 ( ) Men GG 2,205 (55) (80-98) GA 1,494 (37) (77-100) 1.0 ( ) 1.0 ( ) 1.0 ( ) AA 301 (8) (59-109) 0.9 ( ) 0.9 ( ) 0.9 ( ) V771M Women GG 4,665 (94) (41-50) GA+AA 300 (6) (45-87)* 1.6 ( ) 1.7 ( ) 1.5 ( ) Men GG 3,714 (93) (81-95) GA+AA 286 (7) (65-117) 1.0 ( ) 1.0 ( ) 1.0 ( ) M825I Women GG 4,399 (89) (41-50) GA+AA 566 (11) (44-73) 1.3 ( ) 1.4 ( ) 1.3 ( ) Men GG 3,560 (89) (79-94) GA+AA 440 (11) (80-126) 1.2 ( ) 1.2 ( ) 1.2 ( ) I883M Women AA 3,812 (77) (39-49) AG+GG 1,153 (23) (46-66) 1.3 ( ) 1.3 ( ) 1.3 ( ) 7

16 Men AA 3,122 (78) (78-94) AG+GG 878 (22) (82-113) 1.2 ( ) 1.2 ( ) 1.2 ( ) E1172D Women GG 4,693 (95) (42-51) GC+CC 272 (5) (36-78) 1.2 ( ) 1.2 ( ) 1.2 ( ) Men GG 3,776 (94) (79-93) GC+CC 224 (6) (94-166) 1.3 ( ) 1.3 ( ) 1.3 ( ) R1587K Women GG 2,857 (58) (41-52) GA 1,802 (36) (40-54) 1.0 ( ) 1.0 ( ) 1.0 ( ) AA 306 (6) (36-74) 1.1 ( ) 1.1 ( ) 1.1 ( ) Men GG 2,359 (59) (75-92) GA 1,411 (35) (79-102) 1.1 ( ) 1.1 ( ) 1.1 ( ) AA 230 (6) (93-164) 1.4 ( ) 1.3 ( ) 1.4 ( ) Non-incident cases (n=63) were excluded, leaving 8,965 individuals for the survival analyses and Cox regression. CI= confidence interval; HDL= high-density lipoprotein; LDL=low-density lipoprotein. In all three Cox regression models (age adjusted, HDL cholesterol adjusted and multifactorially adjusted) age is adjusted for by incorporating age in the baseline hazard function (left truncation). HDL cholesterol adjusted: HDL cholesterol in tertiles was adjusted for in the Cox regression model. Multifactorially adjusted: Smoking, diabetes mellitus, hypertension dichotomized, and body mass index and LDL cholesterol in tertiles were adjusted for in the Cox regression model. Log-rank test vs. most common genotype: * P<0.01; P<

17 Data Supplements Table II. Verification of significant results from the prospective study in an independent case-control study. Number of individuals Odds Ratio (95% CI) Age adjusted Ischemic Controls heart disease V771M N (%) All GG 864 (93) 7,491 (94) 1 GA+AA 68 (7) 508 (6) 1.2 ( ) Women GG 229 (93) 4,332 (94) 1 GA+AA 16 (7) 266 (6) 1.1 ( ) Men GG 635 (92) 3,159 (93) 1 GA+AA 52 (8) 242 (7) 1.1 ( ) V825I All GG 815 (87) 7,121 (89) 1 GA+AA 117 (13) 878 (11) 1.2 ( ) Women GG 217 (89) 4,085 (89) 1 GA+AA 28 (11) 513 (11) 1.0 ( ) Men GG 598 (87) 3,036 (89) 1 GA+AA 89 (13) 365 (11) 1.2 ( ) I883M All AA 702 (75) 6,211 (78) 1 AG+GG 230 (25) 1,788 (22) 1.2 ( ) Women AA 181 (74) 3,549 (77) 1 AG+GG 64 (26) 1,049 (23) 1.2 ( ) Men AA 521 (76) 2,662 (78) 1 AG+GG 166 (24) 739 (22) 1.1 ( ) E1172D All GG 881 (95) 7,579 (95) 1 GC+CC 51 (5) 420 (5) 1.1 ( ) Women GG 232 (95) 4,350 (95) 1 GC+CC 13 (5) 248 (5) 1.0 ( ) Men GG 649 (94) 3,229 (95) 1 GC+CC 38 (6) 172 (5) 1.1 ( ) R1587K All GG 509 (55) 4,672 (58) 1 GA 367 (39) 2,868 (36) 1.2 ( ) AA 56 (6) 459 (6) 1.2 ( ) 9

18 Women GG 142 (58) 2,651 (58) 1 GA 90 (37) 1,668 (36) 1.0 ( ) AA 13 (5) 279 (6) 0.9 ( ) Men GG 367 (53) 2,021 (59) 1 GA 277 (40) 1,200 (35) 1.3 ( )* AA 43 (6) 180 (5) 1.3 ( ) Nine hundred and thirty two patients with ischemic heart disease and severe atherosclerosis (see methods for diagnostic criteria) were compared with 7,999 healthy controls. Chi-square test vs. most common genotype: *P<0.01, P<

19 Data Supplements Table III. ABCA1 haplotype frequencies and association with risk of ischemic heart disease in the general population. Haplotype * Without IHD With IHD P-value P-value after permutation All N=15,716 N=2,214 GGGAGG AGGAGG GGGAGA AGGAGA GGAGGG AGGGGG AAGAGG AGGGGA GGGACA GGAGGA Women N=9,020 N=910 GGGAGG AGGAGG GGGAGA AGGAGA GGAGGG AGGGGG AAGAGG AGGGGA GGGACA GGAGGA Men N=6,696 N=1304 GGGAGG AGGAGG GGGAGA AGGAGA GGAGGG AGGGGG AAGAGG AGGGGA GGGACA GGAGGA * Haplotypes of the six nonsynonymous SNPs in the present study in the following order (left to right): R219K (G>A), V771M (G>A), V825I (G>A), I883M (A>G), E1172D (G>C), and R1587K (G>A). Only estimated haplotypes with frequencies above 1% were included, and the reference group was the most common haplotype (GGGAGG). P-values for haplotype frequencies in cases vs. controls by chisquare test. P-values obtained after 1000 permutations using the permutation testing option of Haploview. Number of haplotypes, i.e. twice the number of participants without IHD and with IHD in the study. IHD = ischemic heart disease. 11

20 Figure Legends Data Supplements Figure I. Cumulative incidence of IHD by ABCA1 genotypes separately for women and men. P-values are for overall log-rank tests. For V771M, V825I, I883M and E1172D, heterozygotes and homozygotes for the variant allele were pooled (combined genotype in red, common genotype in green). For R219K and R1587K, all three genotypes were presented (homozygote in red, heterozygote in black, common genotype in green). Data Supplements Figure II. Risk of IHD by pairs of ABCA1 SNPs. P-values for the two group comparison log-rank tests are adjacent to the corresponding Kaplan-Meier plots. Kaplan-Meier plots are shown for cells with significant HRs except for the I883M/E1172D AGCC genotype in men, because only 2 incident events occurred in this group. HR=hazard ratio, CI=confidence interval, n=number of individuals. 12