Low levels of HDL cholesterol are predictive of an increased risk of cardiovascular disease (CVD). Although this relationship is consistent and well established, in clinical practice a low HDL-cholesterol concentration rarely stands alone and is most-often found in sedentary patients who are overweight or obese (particularly abdominally obese with ectopic fat deposition), are insulin resistant with hypertriglyceridemia, or who show evidence of systemic inflammation. A reduction in the size of HDL particles is also a hallmark of this constellation of metabolic abnormalities, often referred to as the metabolic syndrome. Thus, although the relationship between low HDL-cholesterol level and CVD is strong, whether this association reflects a causal link is debatable. Mackey et al. have now shown that, beyond HDL-cholesterol levels, the number of circulating HDL particles in the bloodstream could be a strong marker of cardiovascular risk.1

Approximately 20 years ago, Otvos and colleagues reported that HDL particle size and the number of circulating HDL particles per unit of plasma could be measured using NMR spectroscopy.2 This technology is now widely used in large-scale studies. NMR spectroscopy has also improved our understanding of HDL structure and function, and its potential relevance as a therapeutic target. The associations between HDL-cholesterol level, HDL particle number, carotid atherosclerosis, and incident cardiovascular events have been investigated in the Multi-Ethnic Study of Atherosclerosis (MESA), a report from which has been published by Mackey et al. in the Journal of the American College of Cardiology.1 As expected, in univariate analyses, both HDL-cholesterol level and HDL particle number were strongly and inversely associated with carotid intima–media thickness (cIMT) as well as with the risk of cardiovascular events (227 events in 5,598 participants). However, after accounting for covariates including triglyceride levels, HDL particle size, and HDL particle number, the association between HDL-cholesterol level, cIMT, and cardiovascular events was no longer significant.1 These results suggest that the correlates of a low HDL-cholesterol concentration could explain, to a significant extent, the association between HDL-cholesterol level and CVD. When the MESA investigators examined the confounding effects of the same covariates in the relationship between HDL particle number and cIMT or risk of cardiovascular events (replacing HDL particle number with HDL-cholesterol level in the list of covariates), the associations remained highly significant thereby making HDL particle number an independent marker of cardiovascular risk, over and above HDL-cholesterol level.

In the European Prospective Investigation into Cancer and Nutrition (EPIC)-Norfolk study3 (822 events in 2,223 participants), we reported an association between HDL particle number measured by NMR spectroscopy very similar to that subsequently observed in MESA.1 Again, none of the above-mentioned covariates attenuated the relationship between HDL particle number and cardiovascular risk.3 Thus, the results of the MESA1 and EPIC-Norfolk3 studies provide robust and consistent evidence that some parameters of HDL other than HDL-cholesterol level are strong markers of cardiovascular risk. Furthermore, these studies show that the static measurement of HDL-cholesterol level might not be sufficient to capture fully the risk associated with low HDL-cholesterol concentration. In other words, an individual could have a low HDL-cholesterol level yet not be at increased cardiovascular risk if HDL function is not impaired. On the other hand, having a high HDL-cholesterol level might not be sufficient to prevent the development of atherosclerosis if the individual carries dysfunctional HDL particles. In parallel to the MESA1 and EPIC-Norfolk studies,3 the results of other prospective studies have shown that measurement of HDL functionality, such as the ability to remove cholesterol from lipid-laden macrophages4 and HDL anti-inflammatory properties,5 could also be predictors of CVD risk. Investigation of the relationship between pre-β-HDL levels, HDL metabolomics, and CVD will add to our knowledge on this subject (Figure 1). Whether a high HDL particle number is a reliable surrogate of HDL functionality, and explains its cardioprotective effects, is currently unknown and requires investigation.

Figure 1
figure 1

Association between lifestyle and genetic factors with traditional measures, emerging biomarkers of HDL, and biomarkers of HDL functionality.

Full size image

we need to target other features of HDL structure, composition, and function

Raising HDL-cholesterol levels by pharmacotherapy to reduce cardiovascular risk is an appealing strategy. However, large-scale clinical trials have failed to provide evidence that raising HDL-cholesterol level per se has clinical benefits. For example, in May 2012, Roche reported the failure of the cholesteryl ester transfer protein (CETP) inhibitor dalcetrapib, which raises HDL-cholesterol levels by 30–35%, to provide cardiovascular benefits in the dal-OUTCOMES study.6 Investigating the impact of dalcetrapib on parameters of HDL metabolism, Ballantyne and colleagues reported that dalcetrapib 600 mg per day (the dose used in the dal-OUTCOMES study) increased apolipoprotein A-1 level and HDL particle number by only 12.6% and 9.3%, respectively.7 The more-potent CETP inhibitor anacetrapib, which can raise HDL-cholesterol level by up to 140%, is currently being tested in the large, phase III Randomized Evaluation of the Effects of Anacetrapib Through Lipid-modification (REVEAL) trial (NCT01252953). However, results from epidemiological studies in which both HDL-cholesterol level and HDL particle number have been measured suggest that we need to target other features of HDL structure, composition, and function. As a proof of concept, investigators of the Veterans Affairs High-Density Lipoprotein Intervention Trial (VA-HIT)8 reported that individuals who achieved a high HDL particle number with the fibrate gemfibrozil were those at the lowest risk of cardiovascular events. No relationship was observed between HDL-cholesterol level, HDL particle size, or apolipoprotein A-1 level and CVD risk.8 Loss of intra-abdominal adipose tissue and ectopic fat through healthy eating and regular physical activity is likely to be the most 'physiological' way of increasing HDL-cholesterol levels. However, whether this increase will translate into improvements in HDL function needs to be established.

The lack of an independent association between HDL-cholesterol level and risk of CVD has been supported by data from a large-scale genetic association study.9 According to the concept of Mendelian randomization, if HDL-cholesterol concentration (or any other biomarker) is an independent risk factor for CVD then the genetic determinants of this phenotype would be associated to a similar extent with the onset of CVD. In a meta-analysis that included 20 studies and more than 100,000 participants, Voight and colleagues showed that a genetic risk score strongly associated with HDL-cholesterol level did not predict CVD risk.9 Although the results of this study do not support the use of HDL-cholesterol level as a marker of cardiovascular risk, studies aimed at identifying the genetic variants associated with HDL concentration, composition, and functionality and their association with CVD risk will provide clues about the relevance of these features of HDL. Interestingly, in one such study, several loci were found to be associated with HDL-cholesterol level and HDL particle size (on the CETP, LIPC, and PLTP genes, for example), but no single nucleotide polymorphisms associated with HDL particle number were identified.10 These results provide evidence that HDL-cholesterol level and HDL-related biomarkers do not provide the same information, and that the latter could be distinctly related to CVD events.10

The story of HDL research is not a fairy tale. Although the concept of HDL as a target to reduce CVD risk is currently under criticism, we need to keep in mind that some other features of HDL, beyond HDL cholesterol could also be relevant to the management of CVD. The puzzle of HDL is complex, certainly more so than for LDL. For those who have judged the book by its cover, this story has had a rather unhappy ending. However, as we unveil new chapters on the epidemiology of biomarkers of HDL concentration, composition, and functionality, the story is far from over. The first book on HDL was interesting, but left us perplexed and unsatisfied. The sequel could be even more fascinating, and lead to major therapeutic advances in our battle against CVD.