|Home | About | Journals | Submit | Contact Us | Français|
Plasma high-density lipoprotein cholesterol (HDL-C) concentration is inversely related to the risk of cardiovascular disease (CVD). Inhibiting cholesteryl ester transfer protein (CETP) activity raises HDL-C and may be cardioprotective, but an initial clinical trial with a CETP inhibitor was prematurely stopped due to increased CVD in treated patients, raising concerns about this approach. Data relating circulating CETP concentrations to CVD incidence in the community are conflicting.
Plasma CETP activity was measured in 1978 Framingham Study participants (mean age 51 years, 54% women) who attended a routine examination in 1987–90 and were free of CVD. On follow-up (mean 15.1 years), 320 participants experienced a first CVD event (fatal or non-fatal coronary heart disease, cerebrovascular disease, peripheral vascular disease, or heart failure). In multivariable analyses adjusting for standard risk factors including HDL-C, plasma CETP activity was inversely related to the incidence of CVD events (hazards ratio [HR] for activity ≥median 0.72, 95% CI 0.57–0.90, p=0.004 [compared to <median]; HR per SD increment, 0.86, 95% CI 0.76–0.97, p=0.01). The inverse association of CETP activity with CVD incidence remained robust in time-dependent models updating standard risk factors every 4 years, and was maintained in analyses of incident ‘hard’ CVD events (myocardial infarction, stroke or heart failure).
In our prospective investigation of a community-based sample, lower plasma CETP activity was associated with greater CVD risk. These observations, if confirmed, challenge the concept that CETP inhibition may lower CVD risk.
In prospective epidemiological studies, lower blood levels of high-density lipoprotein cholesterol (HDL-C) are associated with a greater risk of coronary heart disease (CHD),1 a relationship that is graded, continuous and independent of the effect of low-density lipoprotein cholesterol (LDL-C). 1,2 Therefore, raising HDL-C has emerged as a key strategy for reducing the residual CHD risk in individuals optimally treated for elevated LDL-C.3
Attempts to raise HDL-C have recently centered on inhibiting cholesteryl ester transfer protein (CETP),4 a plasma glycoprotein that mediates the transfer of cholesteryl esters from HDL to apoprotein B (apoB)-containing lipoproteins in exchange for triglycerides. As a result of CETP action, HDL-C is decreased (impairing reverse cholesterol transport) and the cholesterol content of very low-density lipoprotein (VLDL) and LDL is increased, which could result in a pro-atherogenic profile especially when hepatic uptake of apo-B containing lipoproteins is impaired.5 Thus, it would seem intuitive that inhibiting CETP and maintaining higher levels of HDL-C would translate into a lower CHD risk.4,6,7 However, patients randomized to torcetrapib, the first CETP-inhibitor to be evaluated in a large clinical trial (ILLUMINATE), experienced a higher rate of cardiovascular (CVD) events.8 Suggested possible causes for the untoward outcome associated with torcetrapib treatment include ‘off-target’ blood pressure raising effects, induction of dysfunctional HDL by this agent, and/or a potential direct atherogenic effect of CETP inhibition.9–11 Studies also failed to demonstrate a reduction in progression of atherosclerosis in the carotid and coronary arteries with torcetrapib treatment despite impressive increases in HDL-C levels.12–14
Some genetic investigations have also raised questions about the desirability of inhibiting CETP. Although a recent meta-analysis noted a weak inverse association of common CETP gene variants (associated with lower CETP levels) with CVD risk,15 an updated meta-analysis separated community-based investigations from studies of high risk samples, and noted a direct association of CETP variants associated with lower CETP levels (and higher HDL-C) with CVD risk in the population-based samples.16 Some other studies focusing on partial genetic deficiency states have noted a paradoxical increased CVD risk in the setting of lower CETP activity.17–21
Also, other investigations relating circulating CETP levels to CVD incidence in the community have yielded conflicting results. Thus, two recent reports indicated that lower blood CETP concentrations were associated with a greater risk of CVD in patients with prior coronary artery disease,22 and in men with low triglyceride levels.23 In contrast, a case-control investigation from the EPIC-NORFOLK Study reported a positive association between plasma CETP levels and CVD that was confined to individuals with high triglyceride levels.24 These investigations were limited by the modest number of CVD events in some reports,22,23 and by the case-control design in others.23,24 Accordingly, we related plasma CETP activity to the incidence of CVD prospectively in the community-based Framingham Heart Study.
The design of the Framingham Offspring Study has been published elsewhere.25 Participants in this cohort are seen at the Heart Study clinic approximately every four years. At each examination cycle, participants undergo laboratory testing for CVD risk factors, anthropometry, and a standardized medical history and physical examination targeted at detection of CVD. A random subsample of participants (~50% of attendees) at the fourth examination cycle (1987–1990) underwent measurement of plasma CETP activity; a subsample was chosen for cost-containment combined with optimal use of nonrenewable plasma resources. Thus, of 4019 attendees at this examination, we excluded 2041 individuals for the following reasons hierarchically: prevalent CVD (n=343), missing values for covariates (n=15), loss to follow-up (n=16), plasma CETP activity not assayed (n=1623), and outlier CETP values (n=44; defined as values outside 1.5 times the interquartile range from the first or third quartile).26 After exclusions, 1978 participants remained eligible for the present investigation.
Participants who were excluded because they did not have plasma CETP activity measured did not differ from those with available CETP measurements in terms of their CVD risk factor profile (with the exception of a marginally lower prevalence of diabetes; see Appendix Table 1). Study participants provided written informed consent, and the study protocol was approved by the Institutional Review Board at the Boston University Medical Center.
CETP activity was determined using previously frozen (−80°C) EDTA plasma obtained from participants after an overnight fast. This measurement was performed using a CETP Activity Assay Kit (Roar Biomedical, Inc, New York, NY), as detailed previously.27 The method utilizes as a donor a synthetic phospholipid emulsion in which fluorescent cholesteryl ester [CE]) is solubilized and, as an accepter, native VLDL particles.28 In this assay the donor particle CE has been shown to be the preferred substrate for CETP in whole plasma and, indeed, does not compete with endogenous plasma HDL-CE as a substrate for CETP (unlike some radioisotopic assay methods that are affected by endogenous HDL-C concentrations).29 The reaction is linear (from 0.2 to 0.8 µl of normal human plasma) over 3 hours of incubation at 37°C,28,30 with CETP-mediated transfer of CE determined by an increase in fluorescence in acceptor VLDL particles (see Appendix 2 for further assay details). All measurements were made in duplicate and the intra-assay and interassay coefficients of variation were both <3%, as reported previously.27 Circulating lipids (total cholesterol, HDL-C and triglycerides) were measured using standardized methods.
All Framingham Heart Study participants are under continuous surveillance for the incidence of CVD events. A team consisting of 3 physicians reviews all relevant medical information, hospitalization records and physician office visits to adjudicate CVD events using standardized criteria.31 A separate group consisting of neurologists reviews and adjudicates all suspected cerebrovascular events. A diagnosis of CVD includes fatal or non-fatal coronary heart disease (CHD; recognized and unrecognized myocardial infarction [MI], stable or unstable angina, and CHD death), cerebrovascular disease (stroke, transient ischemic attack), peripheral vascular disease (intermittent claudication), and heart failure. In our primary analyses, we focused on incidence of a first CVD event as defined above, consistent with numerous prior Framingham reports.31 In secondary analyses, we related plasma CETP activity to the incidence of a first ‘hard’ CVD event, a composite endpoint that included fatal and non-fatal MI, unstable angina, stroke, or CHF; this combination of events was chosen to correspond roughly to the primary outcome in the ILLUMINATE trial.8
Given the near normal distribution of plasma CETP activity, we used untransformed values. We evaluated the correlation of plasma CETP activity with HDL-C using Pearson correlation coefficients. Because CETP activity distributions were quite similar in men and women, and we did not observe effect modification by sex upon formal testing, all analyses were performed for pooled sexes to maximize statistical power. We modeled plasma CETP activity as a binary variable (dichotomized at the median) in primary analyses. A median cut point is reasonable from a physiological perspective because pharmacological inhibitors reduce plasma CETP activity by about 35–65% in a dose-dependent manner,32–34 and partial genetic CETP deficiency (associated with increased CVD risk17–21) typically is associated with ~40–70% of normal activity.35–37 We also performed additional analyses modeling plasma CETP activity as a continuous variable.
We compared the cumulative incidence of CVD for groups with plasma CETP activity at or above versus below the median value (131 nmol/L/hr). After confirming that the assumption of proportionality of hazards was met, we used Cox proportional hazards regression to relate plasma CETP activity to incidence of a first CVD event on follow-up. Four sets of models were constructed hierarchically: a. adjusting for age and sex; b. additionally adjusting for baseline levels of standard CVD risk factors other than HDL-C, i.e., systolic blood pressure, hypertension treatment, total cholesterol, smoking and diabetes; c. adjusting for baseline standard risk factors noted above and for HDL-C; d. adjusting for standard risk factors including HDL-C in time-dependent analyses with updating of risk factors every four years (including adjustments for the use of lipid-lowering treatment on follow-up). To gain insights into potential nonlinearity of associations between plasma CETP activity and incidence of CVD, we examined generalized additive Cox models using penalized splines. These analyses also facilitate assessment of the dose-response relation between plasma CETP activity and CVD incidence. We also performed sensitivity analyses: evaluating tertiles of CETP activity (instead of the median cutpoint), and; analyzing the entire sample with plasma CETP activity assay without the exclusion of outliers.
Recent reports have underscored the context-specific associations of altered CETP activity, depending on sex, the level of HDL-C, triglycerides or absolute CVD risk.16,17,19,23 Therefore, we tested for effect modification by age, sex, high triglyceride ( ≥250 mg/dl), low HDL-C (<40 [men] or <50 mg/dl [women]), high HDL-C (≥60 mg/dl), high LDL-C (≥130 mg/dl), apolipoprotein B levels (dichotomized at median), and a high Framingham risk score (10-year risk of CHD ≥20%) by incorporating corresponding interaction terms in models with the binary plasma CETP activity variable. In secondary analyses, we related plasma CETP activity to incidence of a first ‘hard’ CVD event on follow-up.
A p-value <0.05 was used to indicate statistical significance. The authors had full access to the data and take responsibility for its integrity. All authors have read and agree to the manuscript as written.
The baseline characteristics of our sample overall and according to CETP activity are shown in Table 1. Participants with plasma CETP activity below the median were slightly older, more likely to be men, and had slightly lower mean LDL-C but higher mean TG levels compared to those with activity at or above the median. Although the two groups had similar HDL-C, overall CETP activity was inversely related to HDL-C in the sample (r= −0.07, p=0.001); this modest correlation is consistent with that observed in relatively normolipemic individuals.38
On follow-up (mean 15.1 years, maximum 19.6 years), 320 participants (124 women) experienced a first CVD event, including 222 individuals with a first hard CVD event. Table 2 and Figure 1 suggest that individuals with plasma CETP activity below the median experienced higher CVD incidence relative to those with activity at or above the median.
The results of analyses relating plasma CETP activity to CVD incidence are shown in Table 3. Plasma CETP activity at or above the median was associated with a 25–28% lower risk of CVD consistently across the various models. In models incorporating both HDL-C and CETP activity, HDL-C was associated inversely with CVD incidence (hazards rate [HR] per mg/dl increment of 0.98, 95% CI 0.97–0.99; HR per SD: 0.71, 95% CI 0.62–0.82, p<0.0001).
When modeled as a continuous variable, a SD increment in plasma CETP activity was associated with a 12–14% lower risk of CVD. Regression splines confirmed the linearity of the inverse association of CETP activity and CVD incidence (Figure 2). In sensitivity analyses analyzing tertiles instead of a median cutpoint, the inverse relation of CETP activity and incident CVD was maintained (HR per tertile increment 0.87, 95% CI 0.76–1.00, lowest tertile as referent; HR for second tertile 0.86, 95% CI 0.66–1.12, HR for highest tertile 0.77, 95% CI 0.58–1.00). In other sensitivity analyses evaluating the entire sample with plasma CETP activity without excluding outlier values, the results of our primary analyses remained robust.
Figure 3 displays the association of plasma CETP activity with incidence of ‘hard’ CVD events (complementing analyses of ‘all CVD’ shown in Table 3). Plasma CETP activity at or above the median was associated with a 27–30% lower risk of CVD, consistent with the primary analyses. We did not observe any effect modification by age, sex, high triglyceride, low or high HDL-C, high LDL-C or apolipoprotein B levels, or a high Framingham risk score (p values for all interactions exceeded 0.05).
The discovery of CETP and the elucidation of its diverse roles in lipid physiology have fuelled a contentious debate on its pro- and anti-atherosclerotic roles.4,39–42 Genetic epidemiological studies also have not resolved the issue.15,18,43,44 Genetic investigations focusing on partial CETP deficiency states have yielded conflicting results,20,21,45,46 sometimes from the same study cohort.20,45 Experimental approaches are also challenged by the lack of CETP in some species, diverse cholesterol transport mechanisms, and a different time frame for development of atherosclerosis in susceptible animals compared to humans.43 The controversy surrounding CETP inhibition as a strategy for raising HDL-C and lowering CVD risk has further intensified,9,11,47–50 in light of recent disappointing clinical trial data associated with one of several newly-synthesized inhibitors of CETP.8,12–14,50 Given this background, we evaluated prospectively the relations of plasma CETP activity to CVD incidence in a community setting.
Our principal finding was an inverse relation of plasma CETP activity and CVD incidence. While our findings are observational, the possibility that the observed relationship is a causal one is strengthened by its biological plausibility (see below), the demonstration of a linear dose-response (regression splines), the presence of a temporal relationship (plasma CETP activity was assessed before CVD events), and the consistency of the association in multiple analyses (incorporating covariates sequentially, including time-dependent models; modeling plasma CETP activity as a binary and as a continuous variable; and evaluation of all CVD and hard CVD events). The strength of the association (a 25–28% lower risk with plasma CETP activity values above the median [which corresponds to a hazard of 1.35–1.40 for plasma CETP activity values below the median, relative to those above]) is of a magnitude similar to that reported for partial CETP deficiency in some reports.17,18,20 We did not find evidence of effect modification by age, sex, HDL-C, LDL-C or triglyceride levels and the global CVD risk (as estimated by the Framingham risk score), in contrast to other reports that noted inverse associations of CETP levels (or of related genotypes) and CVD risk in select subgroups.16,17,19,23
Data on the relations of CETP mass and CVD are conflicting,22–24 as noted earlier, in part related to modest numbers of CVD events and varying study designs. We used a plasma CETP activity assay that was already available in our study cohort at an earlier examination27 (as opposed to CETP mass assays used by several investigators22–24), which facilitated a long-term prospective time-to-event analysis (20 years of follow-up) in this large community-based sample. It is noteworthy that circulating CETP mass and activity have been reported to be highly correlated (r of 0.85 or more) in several reports.51–53
We observed a very modest inverse relation of plasma CETP activity with HDL-C in our sample. This may be in part because HDL-C concentrations are determined by several enzymatic processes in addition to CETP (such as lecithin:cholesterol acyltransferase, hepatic lipase and phospholipid transfer protein)54,55 or it may arise because we measured CETP activity using a fabricated substrate that does not interact with the natural substrate in this reaction (i.e., plasma HDL-CE). Of note, the relations of CETP activity and HDL-C have been inconsistent in the literature with some investigations reporting no correlation54,56 whereas others have noted modest inverse correlations of a magnitude consistent with our findings.38
The failure of the CETP inhibitor, torcetrapib, has drawn attention to the complex and potential beneficial effects of CETP on reverse cholesterol transport at several levels. Irrespective of the toxicity issues of torcetrapib, in clinical trials this agent did produce a substantial increase in HDL-C (as predominantly cholesteryl esters) but did not appear to reduce atherosclerosis in either carotid50 or coronary vessels.14 Studies by Schwartz et al.57 have shown with multi-compartmental analysis in humans that very little cholesteryl ester is transported directly from HDL to the liver but, instead, is predominantly delivered to the liver from apoB-containing lipoproteins. This is in contrast to the fate of HDL-unesterified cholesterol which is readily taken up by the liver and secreted in bile or utilized as the primary precursor for bile acid synthesis.58 In summary, these kinds of results would support raising levels of HDL-C, but not blocking a natural egress pathway of HDL-cholesteryl esters to apoB-containing lipoprotein for delivery to the liver. Given the development of agents in the same class that seem to lack the off-target effects of torcetrapib, it is important to conduct additional studies to confirm our findings.
The moderate-sized community-based sample of middle-aged men and women, prospective design, and the assessment of CVD outcomes blinded to plasma CETP activity levels strengthen our study. Several limitations must be acknowledged. First, assessment of plasma CETP activity in vitro is challenging.59 We used a standardized assay and any measurement error is likely random, biasing us towards the null hypothesis of no association of CETP activity with CVD incidence. The ex vivo assessment of CETP activity may not reflect in vivo activity. Further, the activity assay used estimates rates of mass transfer of cholesteryl esters from donor particles (phospholipids) to acceptor particles (VLDL), and therefore reflects only one aspect of CETP activity. CETP is involved in lipid transfer reactions involving several different lipid particles.43,59,60 Second, we had single-occasion measurements of CETP activity, which would result in an underestimation of the strength of the association (regression dilution bias).61 Third, given the observational study design, our results cannot be extrapolated directly to the potential effects of pharmacological inhibition of CETP activity using available agents. Last, our sample was middle-aged, with an intermediate pre-test probability of CVD, and exclusively white, which would limit the generalizability of our results to age groups or ethnicities not studied.
In our prospective investigation of a moderate-sized community-based sample, lower plasma CETP activity was associated with greater risk of CVD. These observations, if confirmed, call into question the strategy of pharmacological inhibition of CETP to lower CVD risk.
Inhibiting cholesteryl ester transfer protein (CETP) activity raises HDL-C levels and may be cardioprotective. However, an initial clinical trial with a CETP inhibitor was prematurely stopped due to increased CVD events in treated patients. Data relating circulating CETP mass to CVD incidence in the community are also conflicting. So, we related routinely assayed plasma CETP activity to the incidence of CVD events on follow-up (average 15 years) in the offspring cohort of the Framingham Study. In multivariable analyses adjusting for standard CVD risk factors, plasma CETP activity was inversely related to the incidence of CVD events, a finding that remained robust in time-dependent models updating CVD risk factors every 4 years on follow-up, and was maintained in analyses of incident ‘hard’ CVD events. These observational data based on prospective follow-up of a large community-based sample require confirmation. If confirmed, our findings would call into question the use of CETP inhibition as a strategy for lowering CVD risk.
This work was supported by the National Heart, Lung and Blood Institute's Framingham Heart Study (Contract No. N01-HC-25195), HL-54776, 2 K24 HL04334 (Dr. Vasan), and contract 53-K06-5-10 from the US Department of Agriculture Research Service.