HDL cholesterol and cardiovascular risk: HDL Forum response to BMJ analysis

A recent systematic review and meta-regression analysis of 108 randomised trials  published in the British Medical Journal, February 16, 2009 concluded that ‘simply increasing the amount of circulating high density lipoprotein (HDL) cholesterol does not reduce the risk of coronary heart disease events, coronary heart disease deaths, or total deaths.’  Here, HDL Forum Editors Kerry-Anne Rye, John Chapman and Philip Barter give their response.

 Briel M, Ferreira-Gonzalez I, You JJ et al. Association between change in high density lipoprotein cholesterol and cardiovascular disease morbidity and mortality: systematic review and meta-regression analysis. BMJ 2009;338:b92. An accompanying editorial was published in the same issue (a3065).

 Overview of methodology and results

The authors searched for relevant studies that compared any lipid-modifying agent or diet with placebo or usual care, or compared more intensive versus less intensive regimens of lipid-modifying treatment. The studies were also required to have a randomised controlled design including targeted reduction in cardiovascular risk with a minimum of 6 months follow-up and report mortality or myocardial infarction (MI) separately for treatment groups. Studies that failed to report either change from baseline or follow-up concentrations of HDL cholesterol or low-density lipoprotein (LDL) cholesterol were excluded from the analysis. Study selection was undertaken in teams of two reviewers who independently evaluated the eligibility of trials. Data extraction was performed using standardised forms to extract the required relevant information from each trial; in 5 trials that failed to report exact lipid concentrations, data were abstracted from graphs.

 Lipid concentrations were reported as means and standard deviations. The clinical endpoints assessed were: total death, coronary heart disease (CHD) death and CHD events (a composite outcome of nonfatal MI and CHD death). Inverse variance weighted meta-regression analysis was used to investigate the association between differences in the change in HDL cholesterol and LDL cholesterol between treatment and control groups and risk ratios for clinical endpoints. Meta-regression was stratified by drug class to take account of any non-lipid effects of the drugs.

 In total, 108 eligible trials were included in the analysis, involving 146,890 treated subjects and 152,420 control subjects. Interventions included were predominantly statins (62 trials in total, 8 comparing more intensive vs. less intensive treatment) (Fig 1).

 Fig. 1 Studies included in meta-regression analysis

 

The analysis reported no significant association between the treatment-induced change in HDL cholesterol and risk ratios for CHD events, CHD death or total mortality after adjustment for changes in LDL cholesterol. 

 For all trials included, the change in HDL cholesterol explained <1% of variability in any of these outcomes; and the change in the quotient of LDL cholesterol and HDL cholesterol did not explain any more of variability in outcomes.  For example, the risk ratio for CHD events was increased by 16% on average (-4.2% to 36.9%, p=0.12) per 10 mg/dL (0.26 mmol/L) increase in HDL cholesterol. The change in the quotient for HDL cholesterol and LDL cholesterol explained 32% of the variability for CHD death, which was almost the same as that explained for the change in LDL cholesterol alone (33%).

 Sensitivity analyses which focused on interventions known to increase HDL cholesterol and excluded those with harmful effects, showed a significant association of change in HDL cholesterol and log risk ratio for CHD events in univariable analysis with a 29%, (51.7% to 6.6%, p=0.01) risk reduction for each 10 mg/dL increase in HDL cholesterol. However, the significance of this association was no longer detected in analyses adjusting for changes in LDL cholesterol. The authors concluded that these data add to evidence suggesting that reduction in outcomes was due to the association of changes in HDL cholesterol with changes in LDL cholesterol.

 HDL Forum response: Limitations of the analysis

 There are a number of limitations to this analysis, as also acknowledged by the authors (Fig 2). Chiefly, these relate to the study design and analysis. In this analysis, the authors used aggregated data, rather than individual subject data. However, the relation with subject averages across studies may not be the same as for subjects within individual trials. Additionally, the method of analysis, meta-regression analysis describes an observational association, and therefore risks bias by confounding. Regression analysis also ignores the effect of measurement errors in the data.

 Fig. 2 Limitations of study methodology

 

 Failure to take account of baseline triglycerides

 The changes in HDL cholesterol were very small (upper limit <10%) in most of the studies included in the analysis. This is not surprising, given that most were trials using a statin, in which the HDL cholesterol raising typically is in the region of 4-10%.1 However, the authors failed to take account of the effect of baseline triglycerides, which has been shown to influence the extent of HDL cholesterol raising associated with statin treatment. Patients with higher triglycerides tend to have a reduced HDL cholesterol raising response with statin treatment.1,2 In fact, the lipid phenotype of low HDL cholesterol and elevated triglycerides, mixed dyslipidemia, is typically observed in individuals with metabolic syndrome or type 2 diabetes, and would be expected to be prevalent in these trials. This failure to take into account the baseline lipid profile of patients in these trials is an important flaw in the methodology of this analysis.

 Classification of lipid-modifying agents

 Added to this, the classification of lipid-modifying interventions used by the authors combined treatments and diets that have important pharmacological differences or differences in mechanism of action, For example, in the category ‘niacin trials’, the authors included treatment with niacin plus a statin, fibrate or resin. This implicitly assumes that the changes in HDL cholesterol with different treatments are equivalent. However, HDL particles are highly heterogeneous in structure, metabolism and function, and the available evidence suggests that increases in HDL cholesterol with different agents are not entirely equivalent in terms of the HDL particle changes which underlie them.

 Inconsistent with weight of evidence

 This study tells little of the benefits of raising HDL cholesterol, and therefore should be viewed with some scepticism, especially given overwhelming evidence in animal models that HDL cholesterol raising interventions reduce atherosclerosis3,4 and the weight of epidemiological evidence in humans which clearly supports an independent inverse association between HDL cholesterol and risk for CHD events (Fig 3).5,6

 Fig. 3. Relationship between HDL cholesterol and CHD events. Data from the Framingham Study

 

 Analyses have also showed that HDL cholesterol is independently associated with risk for cardiovascular events in statin-treated patients independent of LDL cholesterol levels (Fig 4).7 Furthermore, the increase in HDL cholesterol associated with simvastatin treatment in the 4S study was predictive of benefit independent of the reduction in LDL cholesterol levels.8  A similar result was observed in the Lipid Research Clinics Primary Prevention Trial in which cholestyramine was used as the active agent. In this study a reduction in CHD events correlated positively with changes in LDL cholesterol levels and negatively with changes in HDL cholesterol. For every 1% increase in the concentration of HDL cholesterol there was a 0.6% reduction in CHD events that was independent of the changes in LDL cholesterol levels.9

 Fig. 4. HDL cholesterol and risk for cardiovascular events is independent of LDL cholesterol in statin-treated patients

 

 

This weight of evidence supporting HDL cholesterol and cardiovascular disease risk is acknowledged by guideline groups. For example, the US National Cholesterol Education Program Adult Treatment Panel III10 recognises HDL cholesterol as an independent cardiovascular risk factor and highlights the need for screening for low HDL cholesterol. 

 Ultimately, what we need to resolve this issue is a large prospective controlled study testing the hypothesis with effective and safe HDL cholesterol raising agents.

 References

 1. McTaggart F, Jones P. Effect of statins on high-density lipoproteins: a potential contribution to cardiovascular benefit. Cardiovasc Drugs Ther 2008;22:321-38.

2. Streja L, Packard CJ, Shepherd J et al. Factors affecting low-density lipoprotein and high-density lipoprotein cholesterol response to pravastatin in the West of Scotland Coronary Prevention Study (WOSCOPS). Am J Cardiol 2002;90:731-6.

3. Morehouse LA, Sugarman ED, Bourassa et al. (2007) Inhibition of CETP activity by torcetrapib reduces susceptibility to diet-induced atherosclerosis in New Zealand White rabbits. J Lipid Res 2007;48:1263-72.

4. Okamoto H, Yonemori F, Wakitani K et al. (2000) A cholesteryl ester transfer protein inhibitor attenuates atherosclerosis in rabbits. Nature 2000;406,203-7.

5. Gordon T, Castelli WP, Hjortland MC et al. High density lipoprotein as a protective factor against coronary heart disease. The Framingham Study. Am J Med 1977;62:707-14.

6. Assmann G. Dyslipidaemia and global cardiovascular risk: clinical issues
Eur Heart J Suppl 2006; 8: F40-F46.

7. Barter P, Gotto AM, LaRosa JC et al. HDL cholesterol, very low levels of LDL cholesterol, and cardiovascular events.  N Engl J Med. 2007;357:1301-10.

8. Pedersen TR, Olsson AG, Faergeman O et al. Lipoprotein changes and reduction in the incidence of major coronary heart disease events in the Scandinavian Simvastatin Survival Study (4S). Circulation 1998;97:1453-60.

9. The Lipid Research Clinics Coronary Primary Prevention Trial results. I. Reduction in incidence of coronary heart disease. JAMA 1984;251:351-64.

10. Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on detection, evaluation, and treatment of high blood pressure in adults (Adult Treatment Panel III). JAMA 2001;285:2486-97.

 


Download Attachment