Special Feature
Cholesterol Trials Update
Sources: Sacks FM, et al. Circulation 1998;97:1446-1452; West of Scotland Coronary Prevention Study. Circulation 1998;97:1440-1445; Pederson TR, et al. Circulation 1998;97:1453-1460.
Sacks and associates reanalyzed the care data to assess the effect of pravastatin on coronary events with respect to LDL cholesterol levels achieved during the study, as well as the relationship of HDL and triglicerides to clinical events. This reanalysis, using an expanded CAD end point tabulation, determined that the LDL level achieved during active treatment was a predictor of coronary events, whereas the extent reduction from baseline was not predictive, either expressed in absolute numbers or a percentage of baseline. Furthermore, Sacks et al observed that coronary event rate decreases were no different from placebo once an individual's LDL cholesterol was lowered below 125 mg/dL. It should be noted that approximately 80% of patients on active treatment started with an LDL level greater than 125; in the 20 percent whose baseline LDL was lower than 125 (as previously reported), no difference between pravastatin and placebo in clinical event rates was noted. Sacks et al stressed that these results correlate with a recent meta-analysis of lipid lowering, demonstrating that major coronary events among all trials are closely related to the average total cholesterol achieved (and in the CARE trial, LDL cholesterol). The overall relationship between cholesterol and events was curvilinear in that analysis. Furthermore, another meta-analysis demonstrated little to no relationship between serum total cholesterol and CHD death in the lowest 20-25% of cholesterol concentrations; in this CARE reanalysis, triglycerides had a weak relationship to coronary events and HDL had none. Sacks et al suggest that the range of optimal LDL goal of lipid-lowering therapy in secondary prevention is somewhere between 100 and 125 mg/dL.
A new analysis of the Western Scotland data demonstrated that pravastatin lowered coronary risk regardless of baseline lipid profile, and concluded that LDL cholesterol risk reduction in the range of 24% from baseline provides the maximum benefits of 40 mg of pravastatin. The investigators conclude from this analysis that lipid lowering per se may not explain all of the coronary event reductions, and they suggest that there may be other actions of pravastatin resulting in risk reduction. The latter is based on an overlap analysis of individuals who received placebo or active drug, and whose cholesterols were in similar range at study end. The pravastatin patients had a substantially lower event rate than the placebo group, who had a comparable LDL cholesterol level (although this level was not achieved with therapy but represented the individual's own baseline). Furthermore, these investigators demonstrated that the relative reduction of CAD events was comparable in all quintiles of baseline LDL cholesterol, suggesting that baseline LDL itself was a weak predictor of risk and indicating that the drug was effective in all LDL ranges of the population in the study. This analysis also used a composite of all cardiovascular end points, and censored individuals who had an event within the first six months of follow-up. Thus, this is a modifocation of the original primary end point trial results that were previously published. In another analysis, the Western Scotland investigators demonstrated that risk reduction from pravastatin was quite similar to what would have been predicted by the Framingham risk equation, which allows calculation of projected CAD mortality from a variety of risk factors, including cholesterol. Using this model, they conclude that event reduction could not be explained entirely by changes in plasma lipid levels. As in CARE, the actual or percent decrease in LDL from baseline were not independent predictors of CAD risk reduction. The investigators conclude that the lack of a clear and graded relationship between LDL risk reduction and events was unexpected, as further decreases in LDL below the mean decrease of approximately 24% were not associated with any significantly greater reduction of CAD risk.
Finally, Pederson and colleagues reassessed their data to determine whether baseline lipid levels predict coronary events and to determine whether specific lipoprotein alterations induced by simvastatin were related to clinical benefits. In the analysis, only baseline and year one lipid levels were analyzed. Placebo patients had a greater event rate than simvastatin patients; in this analysis, the end points extended beyond the primary trial end point to include a variety of CAD events. While decreases in LDL and HDL each contributed to reduction of risks, LDL was more important than HDL. Pederson et al conclude that the LDL levels at one year carry most of the prognostic information available from lipoprotein analysis. HDL and LDL cholesterol were also predictors of risk reduction with simvastatin, but Pederson et al conclude that triglyceride reduction is, at most, a minor contributor to the decrease in CAD events in this population. 4S, as with CARE and Western Scotland, also concludes that risk reduction produced by simvastatin is independent of baseline LDL levels, although 4S did not provide evidence for a lower limit or threshold beyond which a benefit of cholesterol reduction was no longer present. Pederson et al conclude that there is a continuous relationship between CAD mortality and baseline cholesterol. They do not share the opinion that there is no benefit of lowering LDL cholesterol beyond 125 mg/dL.
Grundy, in a cogent editorial comment, makes several points (Circulation 1998;97:1436-1439). He emphasizes that statin therapy has been shown to be safe and effective in high-risk patients and that the recent trials underscore the NCEP guidelines for lipid lowering, targeting LDL cholesterol as the primary factor. He appropriately points out that these three reports should be recognized as subgroup analyses; there are potential problems with dissecting datasets after the primary trial results have been reported. He discusses the difference between the findings from 4S, which is consistent with a curvilinear risk model for lipid lowering, as opposed to the CARE trial, which appears to demonstrate a threshold level below 125 mg/dL and does not model risk reduction in a curvilinear fashion. Grundy stresses that we simply do not know whether lowering LDL cholesterol below 125-130 mg/dL is beneficial in either high-risk primary prevention populations or in secondary prevention. He discusses the rationale of high-dose statin or combination therapy in patients whose LDL cholesterols have not achieved NCEP goals, and he basically concludes that the greater the absolute risk, the more aggressive the therapy should be, leaving clinician judgment as an important factor in determining how far one should push LDL cholesterol.
Comment by Jonathan Abrams, MD
The subgroup analyses from the three major statin trials are of considerable interest but in no way change the conclusion of the primary reports from each of these major studies. Arguments abound regarding how far one needs to lower LDL cholesterol, and it is unlikely that the reanalysis of the CARE data resolves this. Many would agree that a major new trial to assess the benefits of lipid lowering in the lower LDL cholesterol ranges (e.g., < 125 mg/dL) is indicated, and studies are underway to evaluate the effects of reductions in LDL beyond what has been already achieved. It would appear that for secondary prevention (i.e., individuals with established vascular disease), the issues are less cloudy. Lower is likely to be better, although the conundrum of an LDL level of 125 mg/dL or less remains unclear. I recommend pushing the LDL lower than 100 mg/dL in extremely high-risk patients who can be clinically defined in a variety of ways. At a minimum, this would include individuals with several CAD risk factors (especially diabetes), aggressive vascular disease, multiple procedures, relatively young age, etc. Whether one needs to be as aggressive in someone with a baseline LDL cholesterol of less than 130 mg/dL or a total cholesterol of 200 mg/dL or less remains problematic. Again, clinical judgment and assessment of overall global risk would be important. The question of target levels of individuals without overt vascular disease is also a difficult one when LDL levels are borderline high. Careful risk assessment and clinical judgment makes the most sense, particularly when one is confronted with a decision of increasing statins to high levels or using combination drug therapy for lipid lowering. A new Framingham risk model algorithm, published in Circulation (Wilson P. 1998;97:1837-1847), is strongly recommended to physicians who are caring for patients and use preventive cardiology in their practice. These guidelines should aid in decisions regarding pharmacological therapy of hyperlipidemia.
In conclusion, while the potential hazards of subgroup analysis are real, these reassessments of the three core lipid lowering trials (4S, CARE, Western Scotland) are useful and reinforce the powerful effect that the statin drugs have had on coronary artery disease, morbidity, and mortality. Issues as to whether statins have non-lipid lowering effects, how low is enough, and whether the risk event curves are curvilinear or if there is a treatment threshold may appear to be arcane to most practicing physicians. A solid database is in place for sound clinical decision making. The statin era clearly is at hand.
Dr. Abrams is Professor of Medicine, Division of Cardiology, University of New Mexico, Albuquerque.
You have reached your article limit for the month. Subscribe now to access this article plus other member-only content.
- Award-winning Medical Content
- Latest Advances & Development in Medicine
- Unbiased Content