The ENHANCE (Effect of Ezetimibe Plus Simvastatin Versus Simvastatin Alone on Atherosclerosis in the Carotid Artery) study (1), a trial that used ultrasound to measure carotid intima-media thickening (CIMT) as a surrogate measure of atherosclerosis and compared moderate lipid lowering with simvastatin 80 mg versus the same statin and dose plus ezetimibe, generated unprecedented media publicity, patient and physician concern, and congressional involvement over the use of ezetimibe (2–4). It even brought into question the role of low-density lipoprotein cholesterol (LDL-C) as a proven surrogate for atherosclerosis. This issue of the Journal contains 2 more CIMT trials (5,6), which should assist in clarifying the role of CIMT as a potential guide for assessing subclinical atherosclerosis as an intermediate step in determining the potential benefit, or lack thereof, of lipid altering therapies.

The 2 new trials, 1 not-positive with fenofibrate (FIELD [Fenofibrate Intervention and Event Lowering in Diabetes] substudy) and 1 positive with ezetimibe (SANDS [Stop Atherosclerosis in Native Diabetics Study] substudy), have a number of differences and a number of similarities that need reviewing before placing them in context with ENHANCE and other lipid-altering CIMT trials.

The trials were carried out in middle-aged diabetics without existing clinical evidence of cardiovascular disease (CVD) and who were naïve to lipid-lowering therapy. Both had baseline CIMT in the 0.8- to 1.0-mm range consistent with significant carotid atherosclerosis. Though fairly small compared with some recent CIMT trials (1,7–9), they were of longer duration at 3 and 5 years, and thus patient years of observation were similar to these larger trials.

Both trials were read centrally and blinded by trained and experienced readers and showed internal consistency between the mid-point and final results. The amount of LDL-C was significantly lower between the treatment groups in both trials: 27 mg/dl between placebo and fenofibrate in the FIELD trial and 31 to 32 mg/dl between the standard and the aggressive arms in the SANDS trial (10). Within the aggressive treatment arm of the SANDS trial, the LDL-C reduction in the high-dose statin group did not differ from those on low-dose statin plus ezetimibe.

The trials differed in that the fenofibrate trial was double-blind, placebo-controlled, and the SANDS substudy was open-label and compared more aggressive to less aggressive LDL-C lowering. In addition, the SANDS substudy was a post hoc and not a pre-planned analysis and needs to be assessed as such.

The SANDS substudy showed a highly statistically significant difference (p < 0.0001) in CIMT in the ezetimibe-treated group compared with CIMT in the standard treated group. The results also indicated that equivalent LDL-C reduction, whether achieved with statin alone or in combination with ezetimibe, had the same beneficial effect on carotid atherosclerosis as assessed by CIMT.

The FIELD substudy showed no difference in CIMT between placebo and fenofibrate even after 5 years, the longest placebo-controlled trial ever to use CIMT. The 2-year data were consistent with the findings at 5 years.

As would be anticipated with such relatively small trials, neither showed significant differences, or even trends in differences, in CVD events.

So where do these divergent trials leave clinicians and patients who face real-life and real-time decisions with regard to achieving lipid goals (11,12), especially for those at highest risk such as with existing CVD, diabetes, or multiple risk factors?

Given all the recent conflicting data from vascular imaging, can we even use CIMT as a guide for assessing the potential benefit of lipid intervention? We do know from large epidemiological studies such as ARIC (Atherosclerosis Risk in Communities) (13) that there is a strong relationship between CIMT and CVD events. This relationship extends to CIMT and well-established CVD risk factors such as age, sex, LDL-C, blood pressure, and diabetes. This does not necessarily indicate that changes in CIMT over relatively short periods associated with modulation of these risk factors will unequivocally translate into clinical benefit. However there is some evidence from recent larger CIMT trials in high-risk subjects that adverse CIMT change does correlate with increased CVD events (7,9). In the RADIANCE 1 (Rating Atherosclerotic Disease Change by Imaging With a New CETP Inhibitor) trial (7), which assessed CIMT in 900 subjects with familial hypercholesterolemia (FH), there were 2.4% CVD events on atorvastatin alone and 5.1% in those on atorvastatin plus the cholesteryl-ester-transfer protein (CETP) inhibitor torcetrapib (p = 0.05). Although the primary CIMT end point was not significant, the secondary efficacy measure, annualized change in mean CIMT, indicated significant worsening with torcetrapib (p = 0.005). In a second, similarly sized FH CIMT trial, the CAPTIVATE (Efficacy and Safety of the ACAT Inhibitor CS-505 [Pactimibe] for Reducing the Progression of Carotid Artery Disease) study (9), the group on statin alone had 3.4% CVD events compared with 6.3% (p < 0.02) in those on statin plus an acetyl coenzyme A acetyltransferase (ACAT) inhibitor, pactimibe. The primary and secondary CIMT end points were all consistent with worsening of atherosclerosis with pactimibe.

For CIMT trials with agents, such as statins, or niacin, where efficacy in reducing CVD events is well proven, there are no large, robust CIMT trials that have shown reduced CVD events, even where treatment with statin was compared with treatment with placebo. In many of these trials, CIMT changes have been favorable for statin versus placebo or more aggressive LDL-C lowering statin therapy versus less. For example, the largest placebo-controlled statin trial, the METEOR (Measuring Effects on Intima-Media Thickness: An Evaluation of Rosuvastatin) study (8), in which the CIMT results were highly significant, the CVD events were not statistically different and numerically against rosuvastatin at 6 (0.9%) to 0. In trials comparing less versus more LDL-C reduction with statins, such as the ASAP (Effect of Aggressive Versus Conventional Lipid Lowering on Atherosclerosis Progression in Familial Hypercholesterolemia) (14) and SANDS (10) studies, even with significant CIMT results, CVD events have not shown similar findings or trends. A large 1-year trial, the CASHMERE (Carotid Atorvastatin Study in Hyperlipidemic Post-Menopausal Women: a Randomised Evaluation of Atorvastatin versus Placebo) trial (15,16), in 399 post-menopausal women, compared placebo to 80-mg atorvastatin and found no statistical difference in CIMT results. In fact, atorvastatin resulted in numerically more progression than placebo. How could a 50% reduction in LDL-C not be better than placebo on CIMT? The likely reason for "failure" in the CASHMERE trial was the lack of carotid atherosclerosis at entry, or development of disease during the trial. This is clearly demonstrated by the baseline CIMT of 0.69 mm and lack of any change a year later in the placebo group. A CIMT of 0.70 mm was the thinnest CIMT seen in ARIC (13) and was associated with low-risk women with no CVD and few risk factors. Thus even a 50% reduction in LDL-C with a powerful and widely proven statin could not alter the course of a disease that did not exist at the start of the trial or develop even on placebo! The same design flaw is also by far the most scientific reason for the "failure" of the ENHANCE study, where a 40% LDL-C reduction in the "control" group failed to show any progression in CIMT, despite that simvastatin was specifically selected as the "control" based on an FH population showing progression in a prior trial, ASAP, with the same drug and degree of LDL-C reduction. As in the CASHMERE trial, the entry cIMT in the ENHANCE study was 0.69 to 0.70 mm. This unexpectedly low CIMT possibly reflected regression of the lipid content in lesions, over the decades these FH patients received aggressive lipid lowering therapy as suggested by magnetic resonance image studies by Zhao et al. (17). More likely, carotid atherosclerosis was arrested by prior early- and long-term effective treatment (18) of these patients, who averaged only age 46 years when they entered the ENHANCE study. From the CASHMERE and ENHANCE trials, it is clear that if CIMT is to be considered for future trials, a baseline CIMT that is consistent with detectable disease is an absolute must, and subjects with prior lipid therapy, especially effective statin therapy, should not be included, because even if there is residual thickening, lesions will be lipid depleted with little chance that CIMT will progress or regress (7,9,19).

In summary, CIMT may be useful for detecting drugs that cause potential harm no matter the baseline CIMT of the patient population studied but cannot detect benefit unless the population meets critical entry criteria including having proven lipid-rich atherosclerotic burden.

With these factors in mind, let us re-examine the FIELD and SANDS substudies. Both trials met the criteria for success, enrolling treatment naïve patients with significant atherosclerosis based on baseline CIMT.

Neither trial showed CIMT changes that suggest harm. There were either too few CVD events or no difference between treatment arms in either trial to suggest harm. The SANDS substudy clearly showed the benefit on CIMT of more aggressive LDL-C lowering versus standard LDL-C reduction. Importantly, the group treated with ezetimibe had no less benefit than those achieving the same LDL-C reduction with statin alone. This would make physiological and pharmacological sense as the mechanism by which ezetimibe lowers circulating levels of LDL-C, presumably what drives atherosclerosis in any vascular bed, is ultimately the same as that for statins, namely up-regulation of the LDL receptor. In addition, when ezetimibe is added to statin, the "pleiotropic" effects on C-reactive protein, high-density lipoprotein cholesterol (HDL-C), and triglycerides are virtually identical to those of higher dose statin versus lower dose statin (1).

The critics (2–4) of the ENHANCE study ignored actual statistical results showing no difference between treatment arms to call small numerical differences of 0.0058 mm on simvastatin alone versus 0.0111 mm on simvastatin plus ezetimibe as showing potential "harm" or "an expensive placebo." By similar logic, the SANDS substudy results in which the same LDL-C reduction with ezetimibe achieved regression of –0.025 mm when compared with statin alone of –0.012 mm in CIMT could be interpreted as ezetimibe showing potentially enhanced benefit over statin monotherapy! This latest result from a CIMT trial, lacking the basic flaws of the ENHANCE study, is still only a small post hoc substudy, but it restores the balance doctors and patients need to make a decision as to ezetimibe's potential benefit, which ultimately will only be decided by the conclusion of an ongoing morbidity and mortality trial (20). For those patients needing additional LDL-C reduction because they cannot achieve recommended goals on statin alone, cannot tolerate higher dose statins, or perhaps are totally averse to statins, they should be reassured that, based on the 2 CIMT trials involving ezetimibe, the evidence shows that, even with design flaws in 1 trial that was neutral and the latest trial in which it showed benefit, the continued use of the drug is consistent with the well-proven LDL-C hypothesis.

For fenofibrate, however, the news may not be as good. The FIELD substudy met all criteria for assessing potential benefit on carotid atherosclerosis by CIMT. Even though the placebo-control group showed CIMT progression, although perhaps less than would have been anticipated in this diabetic population, the fenofibrate-treated group showed no difference after 5 years (p = 0.987). This was despite robust and significant differences in LDL-C and triglycerides.

This data taken together with the failure to meet the primary cardiovascular end point in the large FIELD trial (21), in which the drug was compared with placebo, although nonfatal myocardial infarction and coronary revascularizations were significantly reduced, should give clinicians concern as to the overall CVD benefit when added to statins. It would appear even more critical to demonstrate CVD benefit when fenofibrate is added to effective statin therapy, which alone reduces LDL-C and triglycerides and increases HDL-C. The definitive answer will hopefully be provided in a few years by the ACCORD (Action to Control Cardiovascular Risk in Diabetes) trial (22) in which 10,251 patients are enrolled in either the combination arm or on statin alone. Fenofibrate, a 33-year-old drug, is still both widely used and, at least in the U.S., as expensive as nongeneric statins, ezetimibe, bile acid sequestrants, and prescription extended-release niacin.

In the interim, both of these latest results negate the "recommendations" of the critics (2,3) of the ENHANCE study who suggested that the "thoughtful" clinician elect a "cautious" strategy that, after using a statin, adds drugs that have shown clinical benefit when added to statins such as fibrates. Fenofibrate, like other fibrates, is certainly an effective triglyceride-reducing agent to be used primarily in patients with high triglycerides (>400 mg/dl) when diet, weight loss, alcohol restriction, exercise, and glucose control fail to achieve significant reductions. In many such patients, use of fibrates will result in an increase in LDL-C as triglyceride-rich lipoprotein clearance is enhanced and a statin may be needed to be added. However in subjects with moderate triglyceride elevations (<400 mg/dl), especially when LDL-C, or non–HDL-C, is elevated, high-dose statin alone significantly reduces not just LDL-C but also triglycerides and raises HDL-C. The evidence for CVD event reduction from then adding a fibrate just does not exist, and until such evidence is produced, its routine addition to "beautify" triglycerides or HDL-C should be avoided.

 References