Jorge Plutzky, MD
The environment at the American Heart Association 2001 Scientific Sessions was somewhat subdued following the attacks on the World Trade Center in New York 2 months prior. This somber mood was offset by some of the clinically exciting data presented during the scientific sessions -- information that will have a far-reaching influence on current thinking and practice as well as on future research regarding preventive cardiology and mechanisms of atherosclerosis.
The Heart Protection Study (HPS)
Perhaps the most striking clinical trial results announced at this year's AHA meeting came from the HPS, run by the British Medical Research Council (MRC) and the British Heart Foundation (BHF), and known as the MRC/BHF Heart Protection Study.[1] HPS will be seen as a landmark study, both for its immediate impact and for its future implications.The anticipation and excitement regarding HPS stemmed in part from the questions it addressed, the treatment protocols employed, and from the sheer size of the cohort. In terms of current impact, the primary objectives of HPS address important and persistent issues regarding the use of statins and/or antioxidant vitamins in certain populations -- eg, those with average LDL cholesterol (LDL-C) levels and also those with diabetes. Its future impact will derive from the sheer size of this clinical trial - over 20,000 patients - and the opportunity to garner significant insights through subgroup analyses that will no doubt continue to emerge from this remarkable database.
HPS was designed to determine whether statins offer benefit to individuals with average-to-low LDL-C levels. The impact of therapy with an HMG-CoA reductase inhibitor (statin) was assessed in a large cohort of relatively low-risk patients with various prior disease histories for whom the benefit and safety of LDL-lowering remained unclear. Enrollment eligibility included patients with a history of coronary artery disease (CAD) and low-to-average total or LDL-C levels, persons at risk for CAD due to a history of other vascular disease (peripheral vascular disease or stroke); those who had a history of diabetes, and individuals who had been inadequately studied in the past (eg, patients > 70 years of age, females). Between July 1994 and April 1997, 20,536 individuals were assigned either to simvastatin (40 mg/day), vs placebo tablets, or to a cocktail of antioxidant vitamins (600 mg vitamin E, 250 mg vitamin C, and 20 mg beta-carotene) vs placebo capsules, for a mean duration of at least 5 years.
Prior to HPS, the suggestion was that a "J-shaped" curve exists for cholesterol levels and cardiovascular events/mortality - meaning that there exists an LDL-C level below which individuals actually have a worse prognosis, or at least do not have any further incremental benefit from LDL-lowering. The main counter-argument to the J-shape theory was that people with low total/LDL-C might have had some other concurrent illness, such as cancer, which accounted for their concomitant LDL-C level and adverse events and that therefore there is no "minimum" LDL-C level below which cholesterol-related events actually increase. A further hypothesis from post-hoc analyses of some early statin trials suggested that perhaps LDL-lowering was not linear, ie, that lower was not necessarily better as long as the LDL levels responded to some degree. HPS should also be able to resolve this question.
Beyond this important core issue, other unresolved issues were also taken on by the HPS dataset.
Is there statin benefit among people with diabetes, as suggested by other trials? Among the elderly? Among women?
- Do statins decrease stroke prospectively in patients with average LDL-C (again, suggested in meta-analyses and other trials)?
- Was there statin benefit among patients with just peripheral vascular disease (PVD) and no known CAD? Prior stroke but no CAD? (End points included stroke.)
In fact, the results of HPS are quite simple: The answer is "Yes" to all of the above questions, at least as far as the preliminary data released at this year's AHA meeting are concerned.
Results of the HPS
Statin therapy.Simvastatin treatment resulted in a constant decrement in LDL-C levels, regardless of baseline LDL-C level (see Table, below).Table. Simvastatin: Average LDL Difference
(mmol/L +/- baseline LDL)
Baseline Feature
Statin
(10,269)
Placebo
(10,267)
Difference in LDL
LDL (mmol/L)
<3.0 (116 mg/dL)
1.8
2.7
-0.90 +/- 0.02
>/= 3.0 < 3.5 (> 116 and < 136 mg/dL)
2.2
3.2
-0.96 +/- 0.03
>/= 3.5 (135 mg/dL)
2.7
3.7
-1.00 +/- 0.03
All Patients
2.3
3.3
-0.96 +/- 0.02
Congruent with this, the risk of cardiovascular events decreased significantly in all subgroups, irrespective of baseline LDL-C. Thus, people with diabetes did better on statin therapy, as did women, those over 75 years of age (~1200 patients in both arms combined), patients with known peripheral vascular disease, and those with prior stroke.
The stroke findings were particularly impressive. No tougher test exists for any medical therapy than to ask whether it exerts a mortality benefit. Although prior studies have suggested -- based on post-hoc analysis, meta-analysis, and even in some cases as a predefined prospective end point -- that statins decreased stroke event rates, the epidemiologic database prior to HPS had not suggested that total cholesterol levels were a strong predictor of stroke risk. In HPS, however, statin therapy lowered the risk of subsequent stroke by almost 30%. Thus, as was shown in the Scandinavian Simvastatin Survival Study (4S)[2] some 7 years ago in very-high-risk patients (prior myocardial infarction and elevated cholesterol), HPS demonstrated a mortality benefit in relatively low-risk patients. The ability to detect this benefit in patients with average-to-low LDL-C levels likely derives from the large size of the study.
Antioxidant vitamin therapy. In addition to statin therapy, HPS also had a vitamin arm. Proponents of antioxidant vitamins have long held out hope that the lack of positive data in prospective trials derived from the use of incorrect combinations. Thus, although the Heart Outcomes Protection Evaluation (HOPE) trial[3] demonstrated such a clear benefit from therapy with the angiotensin converting-enzyme (ACE) inhibitor ramipril that the Safety and Monitoring Committee called for early termination of the study, HOPE demonstrated no benefit from vitamin E. According to the vitamin proponents, this negative result occurred because vitamin C was not given with the vitamin E regimen. Metabolism of vitamin E itself can lead to oxidative species, and thus the outcome of vitamin E therapy may be limited unless accompanied by concomitant vitamin C to "scavenge" any free radicals produced from vitamin E. According to this argument, if vitamin E had been used in conjunction with vitamin C, the outcome would have been different, or so the reasoning went.
In response to this argument, HPS again offers the same simple answer -- a resounding "No." The cocktail of vitamins E, C, and beta-carotene offered no benefit whatsoever in these patients. This finding may prove quite valuable in another important area with a major impact on preventive measures - patient compliance. (or "adherence"). Americans love vitamins. Many clinicians have experienced the all too common phenomenon of appropriately screening and treating a high-risk patient and prescribing a statin, only to discover that the patient has remained on the latest "fad" vitamin. The physician has gone along with this approach since, although the vitamin may not help, it probably doesn't hurt, and the patient likes it. The problem arises when the patient returns 6 months later for the routine follow-up, and is no longer taking the statin but still carefully taking the vitamin.
In this regard, HPS offers an important message: we should focus on proven therapies with known benefits, especially therapies such as statins, which have such a strong safety record. As seen in HPS and in previous trials (and with other statins), simvastatin was quite safe, with very few side effects.
Implications of HPS
What impact will the results of HPS have on the practice of preventive cardiology or on the use of statins? Clearly, HPS adds considerable strength to the already impressive foundation of statin literature. While we await publication of the final results, we can also anticipate years of reports that will come from this database. However, there of course will be more immediate effects on clinical practice. Assuming that the final data are similar to what was presented at the AHA meeting, the clinician may feel that clinical trial evidence is now available to support what was already a common practice among more aggressively treating physicians - the use of statins in post-MI patients with average-to-low LDL-C levels (< 125-130 mg/dL).Perhaps another critical impact of HPS will be the choice of first-line drug therapy among diabetes patients, who often have average-to-low LDL-C levels, low HDL-C levels, and modestly to significantly elevated triglycerides. Although many physicians have been prescribing statins in such scenarios, in fact the only available evidence prior to HPS was from the Veterans Affairs HDL Intervention Trial (VA-HIT)[4] using the fibric acid gemfibrozil (Lopid). In that study, patients with mean LDL-C 110 mg/dL, HDL-C 32 mg/dL, and triglycerides 160 mg/dL who received Lopid 600 mg twice a day had a significant decrease in cardiovascular end points. This was not due to LDL-lowering, since those levels rose some 4 points, HDL rose 6 mg/dL, and triglycerides fell circa 25%. Not surprisingly, among patients with such numbers, 25% were diabetic and as many as 50% may have been insulin resistant. Although we must await the final HPS results for such diabetes patients, especially in terms of their baseline lipids and subsequent lipid changes, statins may well become first-line therapy in this group as well.
This widely pervasive effect of statins on outcomes led the AHA-designated discussant, Dr. Salim Yusuf,[5] McMaster University, Toronto, Ontario, Canada, to suggest that there may no longer be a need to check a lipid profile prior to initiation of a statin. Although this may not have been the consequence that HSP's leader, Dr. Rory Collins, intended to convey, it warrants comment, given its citation by various clinicians in discussion after the session.
Even with the widespread benefits apparently seen in HPS, lipid profiles remain of value. They allow the clinician to gauge risk, important information that may also guide therapy and help motivate the patient in other areas that have independent benefits, eg, diet and exercise. A lipid profile allows the physician to decide the degree of LDL-lowering needed, which may help with selection of the statin, as well as a basis in terms of follow-up. It also provides the physician with information regarding other aspects of the patient's dyslipidemia, ones that can represent additional targets for reducing risk, eg, raising HDL or lowering triglycerides. HPS does not eliminate the ongoing interest in improving HDL while lowering LDL, for example, through the use of combination therapy. The interest in doing so persists, given the obvious fact that simvastatin-treated patients in HPS were certainly not cured and events continued to happen.
C-reactive Protein (CRP) and Diabetes
The success of statins has spawned many studies considering the possible mechanisms by which statins alter the atherosclerotic process or its complication of plaque rupture. The 2001 Scientific Sessions provided a wealth of data regarding mechanisms of atherosclerosis. Particularly exciting is the ongoing convergence between cardiovascular epidemiology and vascular biology, given recent data suggesting that elevated markers of inflammation, such as CRP, predict the risk of future myocardial infarction. Evidence also exists to suggest that the benefits of statin therapy may derive to some extent from its anti-inflammatory effects, and a considerable number of presentations at this year's AHA meeting focused on these non-LDL, pleiotropic effects. Earlier basic science studies had suggested that perhaps CRP is not simply a marker for, but also a mediator of, inflammatory processes in the vasculature, and this theory is now supported by reports that CRP may increase uptake of oxidized LDL.This question of the role of CRP has also been extended to diabetes, a disease that, like atherosclerosis, is a chronic process in which inflammatory mechanisms may contribute to its natural history. Congruent with this were recent reports that inflammatory markers such as CRP may predict the occurrence of new-onset diabetes.
As a result of this expanding line of investigation, therapies for diabetes have begun to consider whether they might have an effect on inflammatory markers such as CRP and interleukin-6 (IL-6). The hypothesis is that it should be possible to prevent the onset of type 2 diabetes with drugs that improve insulin sensitivity, given the role of insulin resistance in this disease. Similarly, insulin sensitizers should provide good tools for investigating whether conversion occurs because of changes in inflammatory pathways. Several ongoing studies with insulin-sensitizing agents -- eg, thiazolidinediones (TZDs; pioglitazone [Actos] and rosiglitazone [Avandia]) or metformin (Glucophage) -- are considering the issues of conversion to diabetes and changes in inflammatory markers.
Haffner and colleagues[6] presented an important abstract at the AHA Sessions suggesting that rosiglitazone did in fact lower CRP levels. This abstract included results from ~ 350 patients with type 2 diabetes who were randomized to 1 of 3 regimens: rosiglitazone 4 mg/day, rosiglitazone 8 mg/day, or placebo. Effects of rosiglitazone on CRP levels were assessed (along with levels of matrix metalloproteinase 9). After 26 weeks of treatment, the placebo group demonstrated a small, nonsignificant decrease from baseline CRP levels (-13.9% [95% CI, -26.7, 1.2]). Conversely, statistically significant decreases from baseline were seen in both of the rosiglitazone treatment groups (rosiglitazone 4 mg/day, -40.7% [95% CI,-50.2, -29.3]; rosiglitazone 8 mg/day, -35.6% [95% CI, -44.7, -25.1]). Differences compared to placebo were significant for both treatment groups. Similar results have been seen in response to pioglitazone, with several abstracts at the AHA reporting anti-inflammatory effects in vitro and in animal models with these agents.
TZDs and PPARs
TZDs continue to generate considerable basic science interest, given that these compounds are ligands for peroxisomal proliferator-activated receptors, or PPARs. PPARs are members of the steroid hormone nuclear receptor superfamily, a group that includes the estrogen receptor and thyroid hormone receptor. Like all nuclear receptors, PPARs, when activated by their specific ligand, become transcription factors that control gene expression by binding to unique response elements in the promoter regions of certain target genes. TZDs are ligands for PPAR-gamma, which is known to control target genes essential to processes like adipogenesis, lipid metabolism, and, of course, glucose control. Fibric acids such as fenofibrate (Tricor) and gemfibrozil (Lopid) are ligands for PPAR-alpha, a nuclear receptor critically involved in fatty acid metabolism.The increasing interest in the role of PPARs and nuclear receptors in vascular biology and atherosclerosis was evident at the AHA meeting, as reflected by the focus on this topic in 3 of the sessions. Chris Glass,[7] whose laboratory has been a leader in PPAR-gamma studies, presented work from his laboratory demonstrating that PPAR agonists limited atherosclerosis in mouse models. Evidence was also presented demonstrating effects of PPARs in endothelial cells and vascular smooth muscle cells. Considerable interest also exists in the ways in which other nuclear receptors -- for example, the liver-X-receptor (LXR) -- may influence HDL levels through PPAR mechanisms. These lines of investigation may lead to new therapeutic interventions designed to improve lipid profiles and decrease atherosclerosis. Such efforts are certainly still warranted and offer exciting new avenues for investigation, even in the face of the tremendous data from HPS.