Author + information
- Received March 20, 2012
- Revision received June 18, 2012
- Accepted July 3, 2012
- Published online November 13, 2012.
- Trine Holm Johannsen, MD, PhD⁎,†,
- Ruth Frikke-Schmidt, MD, DMSc⁎,†,
- Jesper Schou, MSc⁎,†,
- Børge G. Nordestgaard, MD, DMSc†,‡,§ and
- Anne Tybjærg-Hansen, MD, DMSc⁎,†,§,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Anne Tybjærg-Hansen, Department of Clinical Biochemistry KB3011, Section for Molecular Genetics, Rigshospitalet, Copenhagen University Hospital, Blegdamsvej 9, DK-2100 Copenhagen Ø, Denmark
Objectives This study tested whether genetic variation in the CETP gene is consistent with a protective effect of cholesteryl ester transfer protein (CETP) inhibition on risk of ischemic events and on total mortality, without the adverse effects reported for torcetrapib.
Background Torcetrapib, an inhibitor of CETP, increased risk of death and ischemic cardiovascular disease of those randomized to the drug, despite improving the lipid profile.
Methods The Copenhagen City Heart Study is a prospective cohort study of 10,261 individuals, aged 20 to 93 years, who were followed for up to 34 years (1976 to 2010). Of these, 2,087 developed ischemic heart disease, 1,064 developed ischemic cerebrovascular disease, and 3,807 died during follow-up. We selected 2 common genetic variants in CETP previously associated with reductions in CETP activity, thus mimicking the effect of pharmacological CETP inhibition.
Results In individuals carrying 4 versus 0 high-density lipoprotein cholesterol–increasing alleles, there was an increase in levels of high-density lipoprotein cholesterol of up to 14% (0.2 mmol/l), and concomitant decreases in triglycerides, low-density lipoprotein cholesterol, and non–high-density lipoprotein cholesterol of, respectively, 6% (0.1 mmol/l), 3% (0.1 mmol/l), and 4% (0.2 mmol/l) (p for trend 0.004 to <0.001). Corresponding hazard ratios were 0.76 (95% confidence interval [CI]: 0.68 to 0.85) for any ischemic vascular event, 0.74 (95% CI: 0.65 to 0.85) for ischemic heart disease, 0.65 (95% CI: 0.54 to 0.79) for myocardial infarction, 0.77 (95% CI: 0.65 to 0.93) for ischemic cerebrovascular disease, 0.71 (95% CI: 0.58 to 0.88) for ischemic stroke, and 0.88 (95% CI: 0.80 to 0.97) for total mortality. CETP genotypes did not associate with variation in markers of possible side effects previously reported for torcetrapib.
Conclusions Genetic CETP inhibition associates with reductions in risk of ischemic heart disease, myocardial infarction, ischemic cerebrovascular disease, and ischemic stroke, with a corresponding antiatherogenic lipid profile, and with increased longevity, without adverse effects.
Risk factors for atherosclerosis include elevated levels of low-density lipoprotein (LDL) and triglyceride-rich remnant lipoproteins (i.e., intermediate-density lipoproteins, very low-density lipoproteins, and chylomicron remnants), and decreased levels of high-density lipoproteins (HDL) (1–3). Cholesteryl ester transfer protein (CETP) facilitates exchange of cholesteryl esters for triglycerides between HDL and triglyceride-rich remnant lipoproteins (4), and although this exchange clearly may affect several atherogenic lipoproteins in plasma, any association with risk of ischemic heart disease (IHD) has traditionally been attributed to HDL cholesterol alone.
Because of the inverse relationship between HDL cholesterol and risk of IHD in observational studies, development of drugs to raise HDL cholesterol has been proposed, including inhibition of CETP. Recently, however, despite a 72% increase in HDL cholesterol, patients treated with the CETP-inhibitor torcetrapib in the ILLUMINATE (Investigation of Lipid Level Management to Understand Its Impact in Atherosclerotic Events) trial had a paradoxical increase in mortality and an increase in major cardiovascular events (5). The exact reason for the failure of torcetrapib is uncertain (6,7), but has led to a re-evaluation of CETP inhibition as a mechanism to raise HDL cholesterol and HDL cholesterol raising itself as an antiatherogenic strategy (8). Moreover, the recent premature terminations of the dal-OUTCOMES trial (9), due to lack of clinically meaningful efficacy on cardiovascular events of the CETP inhibitor dalcetrapib, and of the AIM-HIGH (Atherothrombosis Intervention in Metabolic Syndrome With Low HDL/High Triglycerides: Impact on Global Health Outcomes) study, due to lack of effect of niacin added to simvastatin in high-risk individuals with low HDL cholesterol and high triglycerides, further questions the value of CETP inhibition and HDL cholesterol–increasing therapy per se (10–12).
However, more recent studies designed to evaluate the side effects and overall safety profiles and the effects on lipid levels of 3 newer CETP inhibitors showed favorable lipid profiles and acceptable side effect profiles (13–15). Therefore, the most plausible explanations for the failure of torcetrapib are that pharmacological CETP inhibition either: 1) has off-target effects that might be specific for torcetrapib; or 2) causes a disproportionate increase in HDL cholesterol, which may be harmful.
We examined the hypothesis that genetic variation in CETP is consistent with a protective effect of CETP inhibition on risk of ischemic events and on total mortality, without the adverse effects previously reported for torcetrapib. To maximize power, we used genotype scores for lipids, lipoproteins, apolipoproteins, disease endpoints, and mortality for 2 very common single nucleotide polymorphisms (SNP) that were previously reported to associate with reductions in CETP activity (16–23), thus mimicking pharmacological CETP inhibition. Using individual participant data, we determined the association of the CETP polymorphisms −629 C>A and IVS1+279 G>A(Taq1B) as individual and combined genotypes, with: 1) changes in lipid, lipoprotein, and apolipoprotein levels; 2) risk of any ischemic event, IHD, myocardial infarction (MI), ischemic cerebrovascular disease (ICVD), ischemic stroke (IS), and mortality with up to 34 years of follow-up; and 3) changes in blood pressure, electrolytes, renal and liver function, and in the electrocardiogram affected by treatment with torcetrapib.
The CCHS (Copenhagen City Heart Study) is a population-based prospective study initiated from 1976 to 1978 (24–28). Participants aged 20 years or older were randomly invited from the Danish Central Population Register and re-examined from 1981 to 1983, 1991 to 1994, and 2001 to 2003. We included 10,261 participants in the present analysis, of which 2,087 developed IHD, and 1,064 developed ICVD; 3,807 died during follow-up. Follow-up was up to 34 years and was 100% complete. For a detailed description of study endpoints and follow-up, please see the Online Appendix.
All participants were white and of Danish descent. The study was approved by institutional review boards and by Danish ethical committees and conducted according to the Declaration of Helsinki. Written informed consent was obtained from all participants.
Genotyping and biochemical analyses
Genotyping was performed for 2 common SNPs in the CETP gene, rs1800775 C>A, in the promoter at position −629 relative to the transcriptional start site, and rs708272 G>A, the Taq1B polymorphism, in intervening sequence 1+279. For biochemical analyses and other covariates, please see the Online Appendix.
Stata software (version 10.1, Stata Corp., College Station, Texas) was used for all analyses. The Mann-Whitney U test or Kruskal-Wallis analysis of variance was used for continuous variables, and the Pearson chi-square test was used for categorical variables. For trend tests, the groups of subjects were classified according to CETP genotype or genotype combinations, coded as 0, 1, 2, 3, and 4. From the 2 CETP polymorphisms, we generated 4 genotype combinations based on the number of HDL-raising alleles in the individual genotype (Online Table 1): Noncarrier (no HDL-raising alleles = reference group: −629 CC+Taq1B GG, n = 2,621; white cell), 1 HDL-raising allele (−629 CA+Taq1B GG or −629 CC+Taq1B GA, n = 537 + 76 = 613; red cells), 2 alleles (−629 CA+Taq1B GA or −629 AA+Taq1B GG or −629 CC+Taq1B AA, n = 4510 + 45 + 12 = 4,567; yellow cells), 3 alleles (−629 CA+Taq1B AA or −629 AA+Taq1B GA, n = 37 + 471 = 508; blue cells), and 4 alleles (−629 AA+Taq1B AA, n = 1,952; green cell). Cuzick nonparametric test for trend was used to test for differences in lipid, lipoprotein, and apolipoprotein levels between CETP genotypes and genotype combinations.
Kaplan-Meier curves and tests for trend evaluated the cumulative incidence of any ischemic event as a function of age and CETP genotypes, individually and combined. Cox proportional hazard regression models, using age as time scale and delayed entry (left truncation), estimated hazard ratios (HRs) for any ischemic event, IHD, MI, ICVD, IS, and total mortality as a function of CETP genotypes and genotype combinations. Because age was the time scale, age was automatically adjusted for. Multifactorial adjustment was used for sex, body mass index, hypertension, diabetes mellitus, smoking, lipid-lowering therapy, alcohol consumption, physical inactivity, and for women, post-menopausal status and hormone replacement therapy.
As previously reported (16–20), CETP −629 C>A(rs1800775) and Taq1B G>A(rs708272) were in linkage disequilibrium (D′ = 0.97, R2 = 0.78) (Online Table 1), although in approximately 11% of individuals (6% of haplotypes), the minor alleles were not inherited on the same haplotype. The minor allele frequency was 0.49 for −629 C>A and 0.44 for Taq1B G>A. Selected clinical characteristics of the participants in the CCHS with and without events are shown in Online Table 2.
Plasma lipids, lipoproteins, and apolipoproteins
Relative to CC noncarriers of −629 C>A, the AA genotype was associated with 13% (0.2 mmol/l) higher HDL cholesterol levels, 6% (0.1 mmol/l) lower triglyceride levels, 3% (0.1 mmol/l) lower LDL cholesterol levels, and 3% (0.2 mmol/l) lower non-HDL cholesterol levels (p for trend 0.02 to <0.001) (Fig. 1). Relative to GG noncarriers of Taq1B G>A, the AA genotype was associated with similar variations in lipids and lipoproteins (p for trend ≤0.001).
However, both SNPs on the wild-type, Taq1B GG or −629 CC, background of the other SNP (corresponding to individuals framed in black in, respectively, row 1 [n = 2,709] and column 1 [n = 3,203] of Online Table 1), were also independently associated with increased levels of HDL cholesterol, with the largest increase of up to 17% in Taq1B AA homozygotes on the −629 CC wild-type background (p for trend = 0.02 and 0.03, respectively) (Online Fig. 1; genotypes indicated in left column). In addition, Taq1B genotype on the −629 CC wild-type background was associated with stepwise reductions in levels of lipoprotein(a) of up to 39% in AA homozygotes (p for trend = 0.02) (Online Fig. 1).
Collapsing all genotypes into 5 genotype groups with, respectively, 0 (=reference group), 1, 2, 3, or 4 HDL cholesterol-raising minor alleles (=0 to 4 A alleles) (Online Table 1) resulted in stepwise increases in HDL cholesterol of up to 14% (0.2 mmol/l) and concomitant stepwise decreases of up to 6% (0.1 mmol/l) in triglycerides, 3% (0.1 mmol/l) in LDL cholesterol, and 4% (0.2 mmol/dl) in non-HDL cholesterol in carriers of 4 A alleles compared with 0 A alleles (genotype −629 CC and Taq1B GG; p for trend 0.004 to <0.001) (Fig. 1).
Results for apolipoproteins A-I and B were similar to but attenuated compared with results for HDL cholesterol and non-HDL cholesterol, respectively, both overall (compare Fig. 1 with Online Fig. 2) and on the wild-type background of the other variant (compare Online Fig. 1 with Online Fig. 3). In contrast, CETP genotypes did not associate with apolipoprotein E levels (Online Figs. 2 and 3).
Risk of ischemic vascular disease
In total, 2,743 individuals developed ischemic vascular disease (IHD or ICVD) during 34 years of follow-up in the CCHS. There was a stepwise decrease in cumulative incidence of any ischemic event as a function of CETP genotype with the AA homozygotes having the lowest incidence for both = SNPs (Fig. 2, top and middle panel) and with a gradual reduction in risk as a function of the number of A alleles for the combined genotypes (Fig. 2, lower panel) (all p for trend <0.001).
Compared with the −629 wild-type (CC), the heterozygous (CA) and homozygous (AA) genotypes were associated with stepwise reductions in risk of ischemic events of 14% (hazard ratio [HR]: 0.86, 95% confidence interval [CI]: 0.78 to 0.94) and 21% (HR: 0.79, 95% CI: 0.72 to 0.88), respectively (Table 1). Compared with the Taq1B wild-type (GG), the heterozygous (GA) and homozygous (AA) genotypes were associated with similar stepwise reductions in risk of 15% (HR: 0.85, 95% CI: 0.78 to 0.93) and 23% (HR: 0.77, 95% CI: 0.69 to 0.86), respectively. For the combined genotypes, there was a gradual reduction in risk as a function of increasing number of A alleles of up to 24% (HR: 0.76, 95% CI: 0.68 to 0.85) for genotypes with 4 versus 0 A alleles (all p for trend <0.001) (Table 1).
Similar stepwise reductions in risk as a function of genotype were observed for IHD and ICVD separately and for the subgroups MI and IS (Fig. 3). For the combined genotypes, there were stepwise reductions in risks for all endpoints as a function of increasing number of A alleles of up to 26% (HR: 0.74, 95% CI: 0.65 to 0.85) for IHD, 35% (HR: 0.65, 95% CI: 0.54 to 0.79) for MI, 23% (HR: 0.77, 95% CI: 0.65 to 0.93) for ICVD, and 29% (HR: 0.71, 95% CI: 0.58 to 0.88) for IS, for genotypes with 4 versus 0 A alleles (p for trend 0.006 to <0.001) (Fig. 3).
In total, 3,807 participants died during follow-up. Compared with the −629 wild-type (CC), the heterozygous (CA) and homozygous (AA) genotypes were associated with a stepwise reduction in risk of overall mortality of 6% (HR: 0.94, 95% CI: 0.87 to 1.02) and 10% (HR: 0.90, 95% CI: 0.82 to 0.98), respectively (Fig. 4). The heterozygous (GA) and homozygous (AA) genotypes for Taq1B were associated with similar reductions in overall mortality of 3% (HR: 0.97, 95% CI: 0.90 to 1.04) and 11% (HR: 0.89, 95% CI: 0.82 to 0.98), respectively, compared with wild type (GG). For the combined genotypes, there was a stepwise reduction in overall mortality of up to 12% (HR: 0.88, 95% CI: 0.80 to 0.97) in the group with 4 versus 0 A alleles (all p for trend = 0.02) (Fig. 4).
CETP genotypes, individually or combined, were not consistently associated with variation in vital signs (systolic and diastolic blood pressure, heart rate), plasma levels of electrolytes (potassium, sodium, chloride), renal function (creatinine, estimated glomerular filtration rate), liver function (alanine aminotransferase, alkaline phosphatase, bilirubin, gamma-glutamyl transpeptidase), or other selected measures (duration of the QT interval corrected for heart rate, and high-sensitivity C-reactive protein) (Online Tables 3 and 4).
The principal findings of this study are that the combination of 2 common CETP polymorphisms previously shown to be associated with reduced CETP activity (16–23), not only predicted stepwise, per allele decreases in risk of IHD and MI, but also predicted corresponding decreases in risk of ICVD, IS, and mortality in the general population. In agreement with the protective effects on all ischemic endpoints and on mortality, the same polymorphisms were associated with an antiatherogenic lipid profile with stepwise, per allele increases in levels of HDL cholesterol and apolipoprotein A-I, and decreased levels of triglycerides, LDL cholesterol, non-HDL cholesterol, and apolipoprotein B. Finally, CETP genotypes did not associate with variation in systolic and diastolic blood pressure, plasma levels of electrolytes, renal function tests, liver function tests, or duration of the QT interval, all markers of possible side effects previously reported for torcetrapib. These data are novel and suggest that (genetic) inhibition of CETP reduces risk of all ischemic vascular events and increases longevity in a dose-dependent manner, without the side effects previously reported for torcetrapib.
This is the first study of CETP genotype scores with concomitant assessment of all ischemic endpoints including mortality, lipid, lipoprotein, and apolipoprotein levels, including lipoprotein(a) and apolipoprotein E, and possible adverse effects in a prospective study with a long follow-up in both sexes. Though consistent with the findings in previous studies on risk of coronary disease or MI (22,29,30), the present study is more definitive and extends findings to ICVD and IS, as well as to total mortality. Another strength is the analysis of 2 different = SNPs, which, although in partial linkage disequilibrium, allow calculation of meaningful genotype scores for lipoprotein and disease endpoints. Furthermore, because we included individual participant data as opposed to tabular meta-analyses (22,23), we could avoid some of the biases inherent to these types of studies. We included both sexes and up to 34 years of follow-up and had 2,087 incident IHD and 1,064 ICVD events, as opposed to a previous prospective study including approximately 18,000 women followed up to 15 years and including 198 incident MI events (29). For the individual genotypes, the observed per A allele increase in HDL cholesterol and the decrease in triglycerides and LDL cholesterol, as well as the observed reduction in risk of IHD and MI in the present study was larger per A allele than in a recent tabular meta-analysis (22), but very similar to the lipid and lipoprotein effects and hazard ratios observed for MI in the prospective Women's Genome Health Study (29). Finally, due to the extensive phenotyping of biochemical and other quantities in the present study, we were able to comprehensively evaluate markers of possible side effects of moderate CETP inhibition.
We cannot conclude that the reduction in risk per A allele can be explained by the associated increased HDL cholesterol levels per se, because we observed corresponding but lesser reductions in nonfasting triglycerides—a marker of cholesterol in remnant lipoproteins (24,26,31–33)—LDL cholesterol, and non-HDL cholesterol, all of which are atherogenic (34,35). In contrast, our analyses suggest that individuals with a lifelong increase in HDL cholesterol together with reductions in nonfasting triglycerides, LDL cholesterol, and non-HDL cholesterol due to low levels of CETP activity may be at reduced risk, not only for IHD and MI as suggested previously by some (22,29,30) but not all studies (36,37), but in addition for ICVD, and IS and they may also live longer. That the effect on risk of ischemic cardiovascular events in the present study is not likely explained by levels of HDL cholesterol per se is supported by recent genome-wide association studies demonstrating that genetic variants associated with isolated effects on levels of HDL cholesterol are not associated with cardiovascular risk (38,39).
Results from the present study suggest that in humans even moderate reductions in CETP activity may reduce risk of ischemic vascular events and mortality. Nonetheless, short-term pharmacological CETP inhibition with torcetrapib 60 mg daily increased both morbidity and mortality from cardiovascular events (5). A plausible explanation for this is that torcetrapib has off-target effects, mainly on blood pressure, which do not seem to be shared with other chemically dissimilar CETP inhibitors (13–15). To determine whether genetic reduction of CETP activity had similar adverse effects as reported for torcetrapib, we determined the effect of CETP genotypes, individually and combined, on blood pressure, heart rate, electrolytes, renal and liver function, QT interval, and high-sensitivity C-reactive protein. We observed no consistent differences as a function of genotypes on any parameters, in complete accordance with the findings discussed herein and with a recent study suggesting that the hypertensive effect of torcetrapib is likely an off-target effect (13–15,23).
Another possible explanation for the torcetrapib-induced adverse effects on cardiovascular endpoints is that torcetrapib caused a potentially harmful increase in HDL cholesterol. The change in lipid profile observed from baseline after 12 months of treatment with torcetrapib was a 72% increase in HDL cholesterol, a 25% reduction in LDL cholesterol, and a 9% reduction in fasting triglycerides (5). Corresponding changes in lipid profile for dalcetrapib were HDL cholesterol 31% increase, LDL cholesterol 2% decrease, triglycerides 3% decrease, and for anacetrapib and evacetrapib (as monotherapy), respectively, 138% and 132% increase in HDL cholesterol, 40% and 40% decrease in LDL cholesterol, and 7% and 20% decrease in triglycerides (13–15). Thus, although the improvement in antiatherogenic lipid profiles associated with pharmacological CETP inhibition were similar, the effect sizes on HDL cholesterol were, respectively, one-half–fold for dalcetrapib and 10-fold larger for both anacetrapib and evacetrapib versus torcetrapib. Nevertheless, only torcetrapib had adverse effects on cardiovascular endpoints, again suggesting that these were due to off-target effects.
Recently, Roche terminated the development of dalcetrapib after interim analysis of the phase III dal-OUTCOMES trial (9) showed a lack of clinically meaningful efficacy on cardiovascular events (40). The main differences between dalcetrapib and the newer CETP inhibitors is the more modest increase in HDL cholesterol, and—probably more important—the almost complete lack of effect on reducing LDL cholesterol and triglyceride-rich lipoproteins. Therefore, there is still hope that the newer CETP inhibitors may improve outcome.
Anacetrapib was reported to associate with a 39% reduction in lipoprotein(a) (13), an important additional lipoprotein effect that predicts a potential 12% reduction in cardiovascular risk (28). Although CETP genotypes overall did not associate with levels of lipoprotein(a), Taq1B genotype on the −629 CC wild-type background was associated with stepwise reductions in levels of lipoprotein(a) of up to 39% in AA homozygotes, suggesting that a reduction in levels of lipoprotein(a) might be a true effect of CETP inhibition.
We show that genetic CETP inhibition associates with reductions in risk of IHD, MI, ICVD, and IS, with a corresponding antiatherogenic lipid profile, and with increased longevity without adverse effects. These findings together with preliminary results for newer CETP inhibitors provide reassurance that pharmacological inhibition of CETP may reduce risk of ischemic vascular events and total mortality, when not accompanied by the off-target effects of torcetrapib.
The authors thank Mette Refstrup and Karin Møller Hansen for their persistent attention to the details of the large-scale genotyping. The authors are indebted to the staff and participants of the Copenhagen City Heart Study.
For supplemental methods and data, please see the online version of this article.
This work was supported by a Specific Targeted Research Project grant from the European Union, Sixth Framework Programme Priority (FP-2005-LIFESCIHEALTH-6) contract no. 037631; the Danish Medical Research Council; the Research Fund at Rigshospitalet, Copenhagen University Hospital; Chief Physician Johan Boserup and Lise Boserup's Fund; Ingeborg and Leo Dannin's Grant; Henry Hansen's and Wife's Grant; William Nielsen's Fund; and a grant from the Odd Fellow Order. Dr. Nordestgaard has received lecture and/or consultancy honoraria from AstraZeneca, Merck & Co., Inc., Pfizer Inc., Karo Bio, Omthera Pharmaceuticals, Inc., Abbott, Sanofi-Aventis, and Regeneron. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- cholesteryl ester transfer protein
- confidence interval
- high-density lipoprotein
- hazard ratio
- ischemic cerebrovascular disease
- ischemic heart disease
- ischemic stroke
- low-density lipoprotein
- myocardial infarction
- single nucleotide polymorphism
- Received March 20, 2012.
- Revision received June 18, 2012.
- Accepted July 3, 2012.
- American College of Cardiology Foundation
- Graham I.,
- Atar D.,
- Borch-Johnsen K.,
- et al.
- Grundy S.M.,
- Cleeman J.I.,
- Merz C.N.,
- et al.,
- Coordinating Committee of the National Cholesterol Education Program
- Chapman M.J.,
- Ginsberg H.N.,
- Amarenco P.,
- et al.,
- European Atherosclerosis Society Consensus Panel
- Tall A.R.,
- Breslow J.L.,
- Rubin E.M.
- Tall A.R.,
- Yvan-Charvet L.,
- Wang N.
- Stein E.A.,
- Roth E.M.,
- Rhyne J.M.,
- Burgess T.,
- Kallend D.,
- Robinson J.G.
- Dachet C.,
- Poirier O.,
- Cambien F.,
- Chapman J.,
- Rouis M.
- Klerkx A.H.,
- Tanck M.W.,
- Kastelein J.J.,
- et al.
- van Venrooij F.V.,
- Stolk R.P.,
- Banga J.D.,
- et al.,
- DALI Study Group
- Borggreve S.E.,
- Hillege H.L.,
- Wolffenbuttel B.H.,
- et al.,
- PREVEND Study Group
- Thompson J.F.,
- Lira M.E.,
- Durham L.K.,
- Clark R.W.,
- Bamberger M.J.,
- Milos P.M.
- Sofat R.,
- Hingorani A.D.,
- Smeeth L.,
- et al.
- Ridker P.M.,
- Paré G.,
- Parker A.N.,
- Zee R.Y.,
- Miletich J.P.,
- Chasman D.I.
- Boekholdt S.M.,
- Sacks F.M.,
- Jukema J.W.,
- et al.
- Zilversmit D.B.
- Langsted A.,
- Freiberg J.J.,
- Tybjærg-Hansen A.,
- Schnohr P.,
- Jensen G.B.,
- Nordestgaard B.G.
- Corsetti J.P.,
- Ryan D.,
- Rainwater D.L.,
- Moss A.J.,
- Zareba W.,
- Sparks C.E.
- Corella D.,
- Carrasco P.,
- Amiano P.,
- et al.