The cholesteryl ester transfer protein Taq1B gene polymorphism predicts clinical benefit of statin therapy in patients with significant coronary artery disease

The cholesteryl ester transfer protein Taq1B gene polymorphism predicts clinical benefit of statin therapy in patients with significant coronary artery disease John F. Carlquist, PhD,a,b,c Joseph B. Muhlestein, MD,a,c Benjamin D. Horne, MStat, MPH,a Noal I. Hart, BS,a,b Tami L. Bair, BS,a Henri O. F. Molhuizen, PhD,d,e and Jeffrey L. Anderson, MDa,c Salt Lake City, Utah, and Vlaardingen and Amsterdam, The Netherlands

Background

Cholesteryl ester transfer protein (CETP) regulates plasma lipid distribution. A polymorphism in the CETP gene (Taq1B) is associated with CETP activity, HDL concentration, atherosclerosis progression, and response to statins, and may influence cardiovascular (CV) events. We studied CETP Taq1B genotype, plasma HDL, and clinical events among all patients and patients stratified by statin treatment.

Methods

Consenting patients (n ⫽ 2531) with significant coronary artery disease (ⱖ1 lesion of ⱖ70% stenosis) undergoing coronary arteriography were genotyped, grouped by statin prescription at hospital discharge, and prospectively followed-up for the outcomes of all-cause mortality and myocardial infarction.

Results CETP Taq1B genotype frequencies were: B1B1, 32.9%; B1B2, 50.3%; and B2B2 16.8%. Plasma HDL was reduced for B1B1 patients (33 ⫾ 12 mg/dL, vs 36 ⫾ 13 mg/dL and 36 ⫾ 13 mg/dL for B1B2 and B2B2, respectively, P for trend ⫽ .003). Overall, event rates did not differ between genotypes. Event rates were similar among untreated (24.8%) and statin-treated (24.2%) B1 homozygotes (P ⫽ NS); statins significantly reduced events for B1B2 subjects (28.0% vs 21.0%, P ⫽ .009) and for B2B2 subjects (26.4% vs 17.4%, P ⫽ .048). Therapeutic benefit for B2 carriers remained after adjustment for covariates, and regression interaction analysis showed that B2 carriers experienced reduced events (relative risk [RR] 0.62, 95% CI 0.45– 0.86), but statins did not benefit those with B1B1 (RR 1.09, 95% CI 0.70 – 1.7; P for interaction ⫽ .02). Findings were similar for the end point of death alone, although a modest benefit was seen in B1B1 patients (RR 0.67, P ⫽ .10), in addition to the strong benefit for B1B2 (RR 0.53, P ⫽ .001) and B2B2 (RR 0.28, P ⫽ .001). Conclusions

The CETP Taq1B polymorphism is associated with differential HDL levels but no significant differential in CV risk in the absence of treatment. Importantly, however, CV event reduction by statin therapy is substantially enhanced in the presence of a B2 allele. Our findings suggest, for the first time, the potential of CETP Taq1B genotyping to enable more effective, pharmacogenetically directed therapy. (Am Heart J 2003;146:1007⫺14.)

See related Editorial on page 929.

Plasma lipid abnormalities are risk factors for coronary atherosclerosis and myocardial infarction (MI) and

From the aCardiovascular Department, and bMolecular Pathology, LDS Hospital, Salt Lake City, Utah, cDivision of Cardiology, University of Utah School of Medicine, Salt Lake City, Utah, dUnilever Health Institute, Vlaardingen, The Netherlands, and the eDepartment of Vascular Medicine, Academic Medical Center, Amsterdam, The Netherlands. Supported in part by a grant from the Deseret Foundation, Salt Lake City, UT: DF 00-384. Submitted August 13, 2002; accepted December 13, 2002. Reprint requests: John Carlquist, PhD, Cardiovascular Department, LDS Hospital, 8th Avenue and C Street, Salt Lake City, UT 84143. E-mail: [email protected]

may have genetic as well as dietary origins. Cholesteryl ester transfer protein (CETP) modifies the lipid composition of the plasma by transferring triglycerides and cholesterol esters between lipoproteins,1,2 thereby decreasing plasma high-density lipoprotein cholesterol (HDL-C) concentrations3 and increasing the proportion of lipids present in the atherogenic low-density lipoprotein cholesterol (LDL-C) and very-LDL fractions. Increased risk for atherosclerosis with increased CETP activity has been shown in CETP transgenic mouse

© 2003, Mosby, Inc. All rights reserved. 0002-8703/2003/$30.00 ⫹ 0 doi:10.1016/S0002-8703(03)00501-5

American Heart Journal December 2003

1008 Carlquist et al

Outcomes assessment

models,4,5 and there is inconclusive evidence of a similar risk in humans.6,7 CETP activity is associated with a single nucleotide polymorphism that creates a Taq1 restriction site in intron 1 of the CETP gene (the Taq1B polymorphism). The B1 allele (presence of the restriction site) at this locus is associated with increased plasma mass of CETP.8 One clinical study found decreased HDL and increased disease progression measured angiographically among patients with the CETP Taq1 B1B1 genotype,7 but further studies are required for confirmation. Statins (hydroxymethyl glutaryl coenzyme A inhibitors) have been shown to not only substantially reduce hyperlipidemia but also to improve endothelial function, stabilize and regress atherosclerotic plaque, and reduce cardiovascular (CV) events.9 Clinically, statins are being used increasingly for primary and secondary prevention and are effective in reducing CV risk, including morality and the need for coronary intervention.10 –14 A recent meta-analysis of clinical trials reported a 20% to 30% reduction in death and CV events among patients receiving statins.15 Whereas statins and CETP both affect plasma lipid concentration and composition, it is possible that interactions between them may occur. Two studies observed a reduction of CETP activity after initiation of statin therapy,16,17 and another study reported that the B1B1 genotype predicted favorable angiographic response to pravastatin.7 These observations suggest a potentially complex interaction among the CETP Taq1B polymorphism plasma lipids, lesion evolution and progression, and response to statin therapy. We hypothesized that the Taq1B genotype might influence occurrence of clinical events among patients with existing coronary atherosclerosis and response to statin therapy. We undertook to compare plasma lipid levels and clinical outcomes by Taq1B genotype among statin-treated and untreated patients.

Peripheral blood was collected in EDTA (ethylenediaminetetraacetic acid), and DNA was extracted as previously described.19 The CETP allelic genotypes arising from the presence (B1) or absence (B2) of the Taq1 restriction enzyme site within intron 1 of the CETP gene were determined as follows.20 A 1412 base pair segment was amplified by polymerase chain reaction and was followed by Taq1 digestion. The primer sequences were as follows: forward, 5⬘ ACA TAT TAA GCA ATT ATC CAG AT 3⬘; reverse, 5⬘ CAC TTG TGC AAC CCA TAC TTG ACT 3⬘. Reactions (15 ␮L) contained 0.1 ␮M of each primer, 100 ␮M of each dNTP (deoxynucleotide triphosphate), 0.1 to 0.5 ␮g genomic DNA, 0.5 U Taq DNA polymerase, and 0.1 mg/mL BSA (bovine serum albumin). The amplification protocol consisted of an initial denaturation step (94°C for 2 min) followed by 35 cycles, each cycle consisting of denaturation (94°C, 30 s), annealing (55°C, 30 s), and extension (72°C 1 min); a final extension at 72°C for 2 minutes was added after the last cycle. Amplified products were digested in a mixture of 2 ␮l reaction buffer (New England Biolabs, Beverly, Mass); 0.2 ␮L 100⫻ BSA; 0.2 ␮L Taq ␣1 (20 U/␮L, New England Biolabs); the 7 ␮l amplified product was diluted to a final volume of 20 ␮L. Products were separated by electrophoresis (2% agarose gel) and visualized by staining with ethidium bromide. The presence of 2 bands (670 bp and 750 bp) identified a homozygous B1 genotype; a single band (1412 bp), a B2 homozygote. A heterozygote was identified by the presence of all 3 bands. Electrophoretic patterns typical of each genotype are shown in Figure 1.

Methods

Assessment of statin prescription

Subjects

The Health Evaluation through Logical Processing (HELP) system is a proprietary clinical data repository that has been used to acquire reliable information regarding medication prescription and use within the Intermountain Health Care system.21 Statin prescription at discharge was determined for each subject using this database. Statin use before hospitalization was not known. Long-term compliance with therapy and the rate of drop-in statin use were estimated by telephone interview of a subset of the entire cohort (n ⫽ 343) at ⱖ2years after discharge.22

The study was approved by the hospital’s institutional review board. Between 1994 and 1998, a consecutive series (n ⫽ 2531) of consenting patients underwent coronary angiography at the LDS Hospital Cardiac Catheterization Laboratory and were found to have significant coronary artery disease (CAD), defined as ⱖ70% stenosis in ⱖ1 coronary artery. CAD status was determined by a cardiologist who was blinded to the study’s genetic variables. Patient demographics and clinical data were entered into a secured database as previously described.18 Briefly, clinical information was recorded at the time of angiography on paper-based standardized forms and transferred to an electronic database within 3 days. Results of laboratory tests and discharge International Classification of Disease codes were transferred electronically from the hospital database to the research database.

After baseline hospitalization, patients were prospectively followed-up until death, subsequent nonfatal MI, or end-ofstudy censor. The occurrence of incident MI was obtained electronically from records of the 22 hospitals in the Intermountain Health Care system. MI was defined as a presentation with characteristic symptoms and diagnostic electrocardiographic changes and/or elevation of cardiac markers. Allcause death (n ⫽ 352) was determined through a national Social Security database containing records current through December 2000.

CETP genotyping

Statistical considerations The ␹2 (categorical variables) and Student t test or 1-way analysis of variance (continuous variables) were used to compare baseline variables by mortality/MI, genotype, or statin therapy status. For variables evaluated across all three CETP

American Heart Journal Volume 146, Number 6

genotypes simultaneously, a test for trend was evaluated using ␹2 or analysis of variance. Multivariable logistic regression (SPSS V.10, SPSS Inc, Chicago, Ill) was used to adjust the genotype-statin effect on mortality/MI for age, sex, body mass index (kg/m2), total cholesterol (mg/dL), LDL-C (mg/dL), HDL-C (mg/dL), hypertension, diabetes, smoking, family history of early CAD, C-reactive protein (mg/dL), plasma total homocysteine (␮mol/L), renal failure, congestive heart failure, previous stroke, previous MI, clinical presentation at hospital admission, number of diseased coronary vessels (ⱖ70% stenosis), concomitant ␤-blocker and/or angiotensinconverting enzyme inhibitor prescription, and type of interventional treatment. Stepwise selection and unconditional entry were used to build the regression models. Logistic regression modeling evaluated the multiplicative interaction of statins with CETP genotype. Two-tailed P values are reported with .05 designated as nominally significant.

Carlquist et al 1009

Figure 1

Results Patient characteristics and genotype frequencies The demographic and key clinical characteristics of the patient sample overall and by genotype are given in Table I. The distribution of CETP genotypes was B1B1, 32.9%; B1B2, 50.3%; and B2B2, 16.8%. Overall allele frequencies in the study population were 0.58 for B1 and 0.42 for B2; these conformed to HardyWeinberg equilibrium. Of the entire patient sample, 720 (28%) were prescribed statin therapy at hospital discharge, and 1811 (72%) were not (Table II). By genotype, 29.7% of B1B1 participants were prescribed a statin, 28.1% of B1B2, and 27.0% B2B2 (P for trend ⫽ NS). Patient characteristics differed somewhat by treatment (eg, by age, diagnosis of hyperlipidemia, and concomitant therapy) (Table II). However, characteristics were generally similar by genotype, both overall and by treatment subgroup. In statin-treated patients, 61% of prescriptions were for simvastatin; 20%, atorvastatin; 8%, lovastatin; 7%, pravastatin; 4%, fluvastatin; and 0.5%, cerivastatin. Long-term compliance with statin prescription was found to be 77% at 3 years among a subset of the population (n ⫽ 343), and drop-in statin use was determined to be 40% among those who were not prescribed at discharge, as previously reported.22

CETP Taq1B genotype and plasma lipids Fasting plasma lipid levels at study entry were measured for 1464 study patients. Across B1B1, B1B2, and B2B2 genotypes, a significant trend was noted for HDL-C (33.4 ⫾ 12 mg/dL, 35.5 ⫾ 13 mg/dL and 36.4 ⫾ 13 mg/dL, respectively, P for trend ⫽ .003). For these respective genotypes, a near-significant trend was noted for total plasma cholesterol (185 ⫾ 48 mg/ dL, 188 ⫾ 47 mg/dL, and 193 ⫾ 45 mg/dL, P for trend ⫽ .053). A similar but weaker trend was observed for plasma LDL-C (113 ⫾ 36 mg/dL, 116 ⫾ 38 mg/dL, and 118 ⫾ 35 mg/dL, P for trend ⫽ .17). These observa-

Agarose electrophoresis gel illustrating the 3 possible CETP Taq1B genotypes.

tions confirm previous reports of relatively lower plasma HDL-C for B1B1 individuals and suggest other potential associations between genotype and the major plasma lipid fractions.

CETP genotype and overall risk for combined end points (all-cause mortality and/or MI) as a function of statin prescription A total of 638 events (all-cause death and/or MI) were identified during the period of the study (Table III). No difference in event rate was observed by genotype for patients not treated with statins (Table III). As expected, the overall event rate among statin-treated patients was significantly lower than among patients not receiving statins (21.5% for treated vs 26.7% for untreated; P ⫽ .006). Untreated and treated patients were next stratified by CETP genotype, and the total number of events was determined for each group (Figure 2). No difference in end point frequency was observed between untreated and treated patients with

American Heart Journal December 2003

1010 Carlquist et al

Table I. Baseline characteristics of patients categorized by Taq1B genotype (vs B1B1) Genotype

Demographics Age (y, mean ⫾ SD) Sex (% male) Cardiovascular risk factors BMI (kg/m2) (mean ⫾ SD) CRP (mg/dL) (mean ⫾ SD) Total HCY (␮mol/L) (mean ⫾ SD) Lipid levels (n ⫽ 1464) (mean ⫾ SD) TC (mg/dL) LDL (mg/dL) HDL (mg/dL) Triglycerides (mg/dL) Hyperlipidemia Diabetes Family history Hypertension Renal failure Smoking Congestive heart failure Prior stroke Prior MI Clinical presentation Stable angina‡ Unstable angina Myocardial infarction Treatment Medical‡ PCI CABG Number of vessels ⱖ70% stenosis 1‡ 2 3 Concomitant ␤-blocker therapy Concomitant ACE inhibitor therapy

Overall (n ⴝ 2531)

B1B1 (n ⴝ 789)

B1B2 (n ⴝ 1213)

B2B2 (n ⴝ 413)

65 ⫾ 11 77%

65 ⫾ 11 76%

65 ⫾ 11 77%

65 ⫾ 12 78%

28 ⫾ 7 2.0 ⫾ 1.9 15 ⫾ 7

29 ⫾ 9 2.0 ⫾ 1.8 14 ⫾ 6

28 ⫾ 5 2.0 ⫾ 1.9 15 ⫾ 6

29 ⫾ 5 2.2 ⫾ 2.2 15 ⫾ 7

188 ⫾ 47 115 ⫾ 37 35.0 ⫾ 13 179 ⫾ 144 57% 20% 37% 59% 6% 25% 14% 1.9% 25%

185 ⫾ 48 113 ⫾ 36 33.4 ⫾ 12 179 ⫾ 132 55% 19% 36% 58% 6% 26% 15% 1.5% 25%

188 ⫾ 47 116 ⫾ 38 35.5 ⫾ 13† 180 ⫾ 157 58% 22% 39% 59% 6% 25% 14% 2.0% 24%

193 ⫾ 45* 118 ⫾ 35 36.4 ⫾ 13† 174 ⫾ 125 55% 18% 36% 59% 5% 25% 13% 2.6% 27%

43% 34% 23%

45% 32% 23%

42% 36% 23%

44% 31% 25%

46% 29% 25%

45% 31% 24%

46% 28% 26%

45% 31% 24%

39% 27% 34% 62% 27%

39% 28% 33% 63% 29%

39% 26% 35% 62% 26%

39% 28% 33% 63% 24%

*P ⬍ .05. †P ⬍ .01. ‡Referent.

the B1B1 genotype (24.8% untreated vs 24.2% treated; adjusted relative risk [RR] 1.09, 95% CI 0.70 –1.7). Conversely, for the B1B2 genotype, statin therapy reduced the event rate relative to untreated patients (28.0% untreated vs 21.0% treated; adjusted RR 0.66, 95% CI 0.46 – 0.94). Similarly, for patients with the B2B2 genotype, statin treatment reduced the event rate from 26.4% to 17.4% (adjusted RR 0.49, 95% CI 0.24 – 0.97). It is of interest that, in the untreated group, there were actually fewer events among patients with the B1B1 genotype than for patients positive for the B2 allele (24.8% vs 27.6%, respectively), but this did not achieve significance (adjusted P ⫽ .065). Covariables entered into the final model included age, diabetes, C-reactive protein (mg/dL), plasma total homocysteine (␮mol/L), congestive heart

failure, previous MI, number of diseased coronary vessels (ⱖ70% stenosis), and type of interventional treatment. The observation of a differential treatment effect by genotype was confirmed by a formal test for interaction using multivariable regression (Figure 3). Overall for B2 carriers, statin therapy was highly effective in reducing CV events (27.6% vs 20.1%; adjusted RR 0.62, 95% CI 0.45– 0.86) but, as noted above, this was not the case for B1B1 patients (adjusted RR 1.09, 95% CI 0.70 –1.7). This treatment-by-genotype interaction on outcome was statistically significant (adjusted P for interaction ⫽ .02).

CETP, statins, and all-cause mortality For the end point of all-cause mortality, results were similar to those reported for the combined end points

American Heart Journal Volume 146, Number 6

Carlquist et al 1011

Table II. Baseline characteristics of patients categorized by treatment status and by Taq1B genotype (treated vs untreated) Overall

Genotype B1B1

Untreated (n ⴝ 1811) Demographics Age (y, mean ⫾ SD) Sex (% male) Cardiovascular risk factors BMI (kg/m2) (mean ⫾ SD) CRP (mg/dL)(mean ⫾ SD) Total HCY (␮mol/L) (mean ⫾ SD) Lipid levels (n ⫽ 1464)(mean ⫾ SD) TC (mg/dL) LDL (mg/dL) HDL (mg/dL) Triglycerides (mg/dL) Hyperlipidemia Diabetes Family history Hypertension Renal failure Smoking Congestive heart failure Prior stroke Prior MI Clinical presentation Stable angina§ Unstable angina Myocardial infarction Treatment Medical§ PCI CABG Number of vessels ⱖ70% stenosis 1§ 2 3 Concomitant ␤-blocker therapy Concomitant ACE inhibitor therapy

B1B2

B2B2

Treated Untreated Treated Untreated Treated Untreated Treated (n ⴝ 720) (n ⴝ 585) (n ⴝ 248) (n ⴝ 915) (n ⴝ 357) (n ⴝ 311) (n ⴝ 115)

66 ⫾ 11 77%

63 ⫾ 12‡ 76%

66 ⫾ 11 76%

63 ⫾ 12† 75%

66 ⫾ 11 78%

64 ⫾ 11* 75%

66 ⫾ 12 77%

63 ⫾ 11* 80%

28 ⫾ 7 2.0 ⫾ 1.9 15 ⫾ 6

29 ⫾ 5 2.1 ⫾ 1.8 14 ⫾ 7†

29 ⫾ 11 2.0 ⫾ 1.8 15 ⫾ 6

28 ⫾ 6 2.1 ⫾ 1.8 13 ⫾ 7

28 ⫾ 5 1.9 ⫾ 1.9 15 ⫾ 6

29 ⫾ 6 2.0 ⫾ 1.8 14 ⫾ 6

28 ⫾ 7 2.0 ⫾ 1.9 15 ⫾ 6

29 ⫾ 6 2.1 ⫾ 1.8 14 ⫾ 7

187 ⫾ 46 117 ⫾ 38 35 ⫾ 13 170 ⫾ 121 50% 19% 36% 57% 6% 25% 14% 1.7% 24%

189 ⫾ 49 183 ⫾ 45 189 ⫾ 54 188 ⫾ 46 187 ⫾ 48 193 ⫾ 46 193 ⫾ 42 112 ⫾ 36* 115 ⫾ 37 109 ⫾ 35 117 ⫾ 39 113 ⫾ 35 119 ⫾ 35 117 ⫾ 37 36 ⫾ 12 33 ⫾ 12 35 ⫾ 12* 35 ⫾ 14 36 ⫾ 12 36 ⫾ 14 37 ⫾ 12 199 ⫾ 186‡ 173 ⫾ 130 193 ⫾ 137 169 ⫾ 116 207 ⫾ 226† 167 ⫾ 118 189 ⫾ 139 73%‡ 50% 69%‡ 52% 74%‡ 46% 78%‡ 22% 19% 17% 20% 25% 17% 20% 41%‡ 36% 37% 36% 45%† 36% 38% 62%* 58% 59% 56% 65%† 59% 61% 5% 6% 4% 5% 6% 5% 5% 24% 27% 22% 25% 25% 24% 27% 13% 16% 13% 13% 14% 14% 11% 2.4% 1.4% 1.7% 1.8% 2.3% 2.0% 4.3% 28%* 25% 27% 22% 28%* 25% 30%

46% 33% 21%

36%* 34% 29%‡

47% 33% 20%

39% 31% 30%†

45% 35% 20%

34% 37%* 29%‡

45% 31% 24%

38% 33% 29%

50% 34% 26%

35% 43%‡ 22%

49% 26% 25%

36% 42%‡ 22%

50% 23% 27%

36% 40%‡ 24%

50% 23% 27%

32% 52%‡ 16%

38% 27% 35% 50% 22%

42% 27% 31% 93%‡ 37%‡

38% 28% 34% 50% 25%

43% 25% 32% 93%‡ 38%‡

39% 25% 36% 50% 22%

40% 29% 31% 93%‡ 35%‡

38% 28% 34% 51% 19%

44% 24% 32% 94%‡ 40%‡

*P ⬍ .05. †P ⬍ .01. ‡P ⬍ .001. §Referent.

Table III. Event rates in each treatment/genotype group Genotype B1B1

Death or MI MI Death

B1B2

B2B2

No statin (n ⴝ 585)

Statin (n ⴝ 248)

No statin (n ⴝ 915)

Statin (n ⴝ 357)

No statin (n ⴝ 311)

Statin (n ⴝ 115)

24.8% (145) 14.2% (83) 13.8% (81)

24.2% (60) 18.1% (45) 9.7% (24)

28.0% (256) 14.5% (133) 16.9% (155)

21.0% (75) 14.3% (51) 9.8% (35)

26.4% (82) 15.1% (47) 16.4% (51)

17.4% (20) 13.0% (15) 5.2% (6)

Data presented as percentage of events (number of events).

American Heart Journal December 2003

1012 Carlquist et al

Figure 2

toward mortality reduction for B1B1 carriers (from 13.8% to 9.7% [30% reduction], P ⫽ .10) and a significant reduction in events for the B1B2 heterozygotes (from 16.9% to 9.8% [42% reduction], P ⫽ .001) and for B2B2 homozygotes (from 16.4% to 5.3% [68% reduction], P ⫽ .001). Thus, for the end point of allcause mortality, these results support statin prescription at hospital discharge across genotypes. Nevertheless, they suggest potential quantitative differences in mortality benefit (adjusted P for interaction ⫽ .15).

Discussion

Frequency of events (all-cause mortality and MI) for statin-treated and untreated patients further separated into the 3 possible CETP Taq1B genotypes.

Figure 3

This study examined relationships between the CETP Taq1B genotype, plasma lipids, statin therapy, and clinical events for 2531 patients with severe, angiographically defined CAD. Consistent with other reports, we found an association between the B1B1 genotype and reduced HDL. We did not find a difference in incident clinical events by genotype in patients not prescribed statins. However, we show for the first time that the CETP Taq1B genotype is a significant predictor of clinical outcome after treatment with statins: reduction in the primary end point of death/MI was substantially enhanced in the presence of a B2 allele (P for interaction ⫽ .02). Reductions in mortality alone across genotypes support statin use irrespective of the Taq1B polymorphism, but they also suggest a quantitative, gene-dosage–related difference (B1B1 ⫽ 30% reduction in death, B1B2 ⫽ 42%, B2B2 ⫽ 68%). We do not believe these differences are likely due to chance. The genotype frequencies we observed match published population frequencies,23,24 suggesting that the study sample was an adequate representation of a general white population and arguing against ascertainment bias. Differences by genotype were found despite the proportionally smaller number of B2B2 individuals, suggesting a robust effect.

Previous studies

Relative risk of death/MI for statin treatment in the CETP Taq1 B1B1 and combined B1B2/B2B2 genotype groups (fully adjusted P for interaction ⫽ .02).

and are presented in Table III. A total of 352 deaths from all causes were recorded during the duration of the study. Across all genotypes, statin therapy significantly reduced death (from 15.8% to 9.0%, P ⬍ .001). By individual genotype, treatment produced a trend

Pharmacogenetic influences on statin therapy have been previously reported. In one study,25 a polymorphism in the stromelysin-1 promoter (5A/6A polymorphism) was found to be associated with clinical response to statins. In that study, treatment reduced clinical events only among individuals carrying a 6A allele, and event reduction was independent of changes in lipid levels or angiographic changes. Three studies found statin-induced plaque regression to be dependent upon gene polymorphisms, one in the angiotensin-converting enzyme I/D (insertion/deletion) locus, and another in the hepatic lipase gene promoter.26,27 Finally, Kuivenhoven et al,7 studying the same CETP Taq1B polymorphism as in the present study, found a statin-mediated reduction in lesion progression

American Heart Journal Volume 146, Number 6

for the carriers of the B1B1 genotype but not for the remainder of patients. Although the present study findings appear to conflict with those of Kuivenhoven et al, study differences may explain these apparent discrepancies. A principal difference between the studies is study end points: angiographic lesion regression (Kuivenhoven) versus adverse clinical events (present study). Whereas serial angiography has been found to correlate somewhat with clinical events, the correlation is highly variable and incomplete.28 Moreover, it is recognized that statin therapy promotes plaque stabilization, a process quantitatively more important in reducing clinical events than lesion regression.29 Thus, the end points of these studies are fundamentally different, and the same results should not necessarily be expected. It is of interest that the analysis of clinical events by Kuivenhoven et al revealed no difference in event rates between the placebo and treated groups of the B1B1 genotype (2.1% and 2.9%, respectively), findings similar to the present study. Among untreated patients, a minor trend toward fewer events was noted among the B1 homozygotes, and that genotype also had the lowest baseline levels of HDL-C (33 ⫾ 12 mg/dL). Although this appears paradoxical, low HDL may not be a good predictor of events for patients with established disease. In a recent analysis of the same patient sample included in the present study, decreased plasma HDL was not a predictor of cardiovascular-related clinical events in patients with preexisting disease.30 Alternatively, it is possible that the B1 (or linked) allele modifies mortality risk associated with decreased HDL. The relationship between HDL-C, CV events, and patient genotype warrants further study.

Study limitations The study has the limitations of nonrandomized, observational studies. Differences in baseline factors might modify or explain the observed results; however, in multivariate analysis, the genotype-statin interaction retained significance. The main limitation of this study is the lack of information regarding medications used before entry into the study and long-term compliance as well as “drop-in” use in patients not initially given statins. A follow-up (average, 3 years) of a subset of the entire cohort indicated 77% compliance with discharge statin prescription. Among this subset, drop-in usage of statin was found to be 40%. Although drop-out and drop-in medication usage would modify the true effect, the observed results would be an underestimate rather than an overestimate of the true effects and would tend to bias the study towards a negative rather than positive outcome. The robust interactions we observed, despite multivariable adjustments, suggest that the overall conclusions regarding

Carlquist et al 1013

genotype associations with outcome of statin therapy are likely valid. The use of all-cause rather than CV death as the primary study end point has advantages and disadvantages. CV death is a more specific end point to CAD pathology, and might be the preferred end point in a prospective trial when cause of death can be adjudicated on predetermined criteria. In the present study, criteria for classifying cardiac death were not established a priori. Determination of cardiac death from medical records is subject to errors,31 and use of death cause from death certificates has been found to be even more unreliable.32 Additionally, mechanisms of death that are unrelated to the study hypothesis should be spread randomly among the study’s patient subsets, which would tend to reduce rather than amplify associations. For these reasons, all-cause mortality represented the most conservative and reliable outcome to use for this study. When individual end points were examined, significant benefit was observed for B2 carriers with respect to all-cause mortality but not for nonfatal MI. An attractive explanation is that the B2 allele may have been associated with a reduction in the severity of MI, exchanging fatal for nonfatal MIs. Alternatively, failure to detect a reduction in nonfatal MIs may have arisen as a result of chance or of sample heterogeneity. In either event, the reduction of fatal and total (fatal plus nonfatal) events is of greater interest and relevance.

Conclusions Consistent with other reports, we found the CETP Taq1B polymorphism to be a determinant of plasma HDL concentration. We did not find a significant difference in clinical outcome (death/MI) by genotype in patients not prescribed lipid-lowering (statin) therapy. However, we did observe a significant interaction between genotype and treatment on the combined end point of all-cause mortality and nonfatal MI, and we observed a trend for all-cause mortality alone: carriers of the B2 allele benefited more than B1B1 homozygotes. Further, the reduction in adverse events by genotype showed a quantitative relationship with the number of B2 alleles present, suggesting a gene-dosage effect. Despite the dramatic pharmacogenetic findings in this study, it would be premature to recommend modification of the current recommendations for lipid lowering in patients with or at risk for CAD. Moreover, a treatment benefit with respect to all-cause mortality was seen for all genotypes despite the variable quantitative benefit. Further studies must prospectively confirm our findings of differential responses to lipid lowering as a function of CETP genotype. Should these findings be confirmed, the potential for individual tailoring of therapy based on genotype may have impor-

1014 Carlquist et al

tant clinical and economic implications for medical care (pharmacogenomics).

References 1. Yen FT, Deckelbaum RJ, Mann CJ, et al. Inhibition of cholesteryl ester transfer protein activity by monoclonal antibody: effects on cholesteryl ester formation and neutral lipid mass transfer in human plasma. J Clin Invest 1989;83:2018 –24. 2. Tall A. Plasma lipid transfer proteins. Ann Rev Biochem 1995;64: 235–57. 3. Hannuksela ML, Liinamaa MJ, Kesaniemi YA, et al. Relation of polymorphisms in the cholesteryl ester transfer protein gene to transfer protein activity and plasma lipoprotein levels in alcohol drinkers. Atherosclerosis 1994;110:35– 44. 4. Agellon LB, Walsh A, Hayek T, et al. Reduced high density lipoprotein cholesterol in human cholesteryl ester transfer protein transgenic mice. J Biol Chem 1991;266:10796 – 801. 5. Marotti KR, Castle CK, Boyle TP, et al. Severe atherosclerosis in transgenic mice expressing simian cholesteryl ester transfer protein. Nature 1993;364:73–5. 6. Bruce C, Chouinard RA, Tall AR. Plasma lipid transfer proteins, high-density lipoproteins, and reverse cholesterol transport. Annu Rev Nutr 1998;18:297–330. 7. Kuivenhoven JA, Jukema JW, Zwinderman AH, et al. The role of a common variant of the cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. N Engl J Med 1998;338: 86 –93. 8. Kuivenhoven JA, de Knijff P, Boer JM, et al. Heterogeneity at the CETP gene locus: influence on plasma CETP concentrations and HDL cholesterol levels. Arterioscler Thromb Vasc Biol 1997;17: 560 – 8. 9. Ballantyne CM, Herd JA, Dunn JK, et al. Effects of lipid lowering therapy on progression of coronary and carotid artery disease. Curr Op Lipidol 1997;8:354 – 61. 10. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet 1994;344:1383–9. 11. Ridker PM, Rifai N, Pfeffer MA, et al. Long-term effects of pravastatin on plasma concentration of C-reactive protein. Circulation 1999;100:230 –5. 12. Prevention of cardiovascular events and death with pravastatin in patients with coronary heart disease and a broad range of initial cholesterol levels. N Engl J Med 1998;339:1349 –57. 13. West of Scotland Coronary Prevention Study: identification of high-risk groups and comparison with other cardiovascular intervention trials. Lancet 1996;348:1339 – 42. 14. Downs JR, Clearfield M, Weis S, et al. Primary prevention of acute coronary events with lovastatin in men and women with average cholesterol levels: results of AFCAPS/TexCAPS. JAMA 1998;279: 1615–22. 15. Ross SD, Allen IE, Connelly JE, et al. Clinical outcomes in statin treatment trials: a meta-analysis. Arch Int Med 1999;159:1793–802. 16. Contacos C, Barter PJ, Vrga L, et al. Cholesteryl ester transfer in hypercholesterolaemia: fasting and postprandial studies with and without pravastatin. Atherosclerosis 1998;141:87–98.

American Heart Journal December 2003

17. Guerin M, Lassel TS, Le Goff W, et al. Action of atorvastatin in combined hyperlipidemia: preferential reduction of cholesteryl ester transfer from HDL to VLDL1 particles. Arterioscler Thromb Vasc Biol 2000;20:189 –97. 18. Taylor GS, Muhlestein JB, Wagner GS, et al. Implementation of a computerized cardiovascular information system in a private hospital setting. Am Heart J 1998;136:792– 803. 19. Anderson JL, King GJ, Thomson MJ, et al. A mutation in the methylenetetrahydrofolate reductase gene is not associated with increased risk for coronary artery disease or myocardial infarction. J Am Coll Cardiol 1997;30:1206 –11. 20. Drayna D, Lawn R. Multiple RFLPs at the human cholesteryl ester transfer protein (CETP) locus. Nucl Acids Res 1987;15:4698. 21. Evans RS, Pestotnik SL, Classen DC, et al. A computer-assisted management program for antibiotics and other antiinfective agents. N Engl J Med 1998;338:232– 8. 22. Muhlestein JB, Horne BD, Bair TL, et al. Usefulness of in-hospital prescription of statin agents after angiographic diagnosis of coronary artery disease in improving continued compliance and reduced mortality. Am J Cardiol 2001;87:257– 61. 23. Bernard S, Moulin P, Lagrost L, et al. Association between plasma HDL-cholesterol concentration and Taq 1 B CETP gene polymorphism in non-insulin-dependent diabetes mellitus. J Lipid Res 1998;39:59 – 65. 24. Ordovas JM, Cupples LA, Corella D, et al. Association of cholesteryl ester transfer protein-TaqIB polymorphism with variations in lipoprotein subclasses and coronary heart disease risk: the Framingham study. Arterioscler Thromb Vasc Biol 2000;20:1323–9. 25. de Maat MP, Jukema JW, Ye S, et al. Effect of the stromelysin-1 promoter on efficacy of pravastatin in coronary atherosclerosis and restenosis. Am J Cardiol 1999;83:852– 6. 26. Marian AJ, Safavi F, Ferlic L, et al. Interactions between angiotensin-I converting enzyme insertion/deletion polymorphism and response of plasma lipids and coronary atherosclerosis to treatment with fluvastatin: the lipoprotein and coronary atherosclerosis study. J Am Coll Cardiol 2000;35:89 –95. 27. Zambon A, Deeb SS, Brown BG, et al. Common hepatic lipase gene promoter variant determines clinical response to intensive lipid-lowering treatment. Circulation 2001;103:792– 8. 28. Brown BG, Zhao XQ, Sacco DE, et al. Atherosclerosis regression, plaque disruption, and cardiovascular events: a rationale for lipid lowering in coronary artery disease. Annu Rev Med 1993;44: 365–76. 29. Maron DJ, Fazio S, Linton MF. Current perspectives on statins. Circulation 2000;101:207–13. 30. Horne BD, Muhlestein JB, Carlquist JF, et al. Statin therapy, lipid levels, C-reactive protein and the survival of patients with angiographically severe coronary artery disease. J Am Coll Cardiol 2000;36:1774 – 80. 31. Gottlieb SS. Dead is dead—artificial definitions are no substitute. Lancet 1997;349:662– 663. 32. Lloyd-Jones DM, Martin DO, Larson MG, et al. Accuracy of death certificates for coding coronary heart disease as the cause of death. Ann Intern Med 1998;129:1020 – 6.