HDL in Morbidity and Mortality: A 40+ Year Perspective

Epidemiology of High Density Lipoprotein (HDL)

The first large population-based study to show an inverse relationship between circulating HDL-cholesterol (HDL-C) concentrations and coronary heart disease (CHD) was a cross-sectional analysis of men of Japanese ancestry living in Hawaii (1). Four prospective epidemiological studies extended these finding showing that a 1 mg/dL increase in HDL-C was associated with a 2% decrease in CHD in men and a 3% decrease in CHD in women (2). The inverse relationship was independent of age, smoking, low density lipoprotein cholesterol (LDL-C), blood pressure, and body mass index (BMI), but was attenuated by adjustment for non-HDL-C (i.e., VLDL+LDL-C) concentrations. These findings suggested that raising HDL-C could be therapeutically beneficial but were insufficient to completely disentangle the close relationship between HDL and triglyceride-rich lipoproteins. The subsequent failure of most clinical trials aimed at raising HDL-C concentrations, especially the trials conducted with cholesteryl ester transfer protein (CETP) inhibitors, as well as detailed studies of the genetic factors underlying HDL-C variation have called into question the causal relationship between HDL-C and CHD, and have refuted the idea that raising HDL-C would automatically have therapeutic benefit. Nonetheless, HDL-C remains an important CHD risk factor that appears in American Heart Association and European Atherosclerosis Society guidelines to estimate risk and guide treatment. HDL-C measurements may integrate the effects of multiple risk factors including obesity, insulin resistance and hypertension. Moreover, an abundance of evidence indicates that the ability of HDL to mediate cholesterol efflux from macrophage foam cells does associate with protection from atherosclerosis, holding promise for future therapies targeting this mechanism. This brief review will provide a personal perspective on HDL research that has benefited from the contributions of many different laboratories (Table 1).

Table 1.

Milestones in HDL science.

Epidemiological studies reveal an independent inverse relationship between HDL-C concentrations and CHD (1976)

Elucidation of monogenic disorders affecting HDL-C concentrations including CETP deficiency (1989)

CETP inhibitors developed as potential antiatherogenic therapies (1990)

APOA1 overexpression reduces atherosclerosis in animal models (1991)

ABCA1 deficiency elucidated as the cause of Tangier Disease (1999)

ABCA1 found to be an LXR target (2000)

LXR activators reduce atherosclerosis (2002)

The first CETP inhibitor in clinical trials (torcetrapib) increases CHD and overall mortality (2007)

Dozens of novel SNPs found to associate with HDL concentrations in human GWAS (2010)

Cholesterol efflux capacity of HDL correlates inversely with CHD independent of HDL-C (2011)

Mendelian Randomization studies suggest that SNPs that associate with HDL-C concentrations do not associate with CHD (2012)

The CETP inhibitor anacetrapib produces a significant but moderate reduction in CHD. The sponsors decline to seek marketing approval (2017)

Reconstituted HDL preparations optimized for human phase 3 studies (2018)

Lessons from Genetics: CETP and ATP Binding Cassette Transporter (ABCA1)

The elucidation of monogenic disorders affecting circulating HDL-C concentrations provided first insights into genetic factors regulating HDL-C. CETP is a prime example. My lab purified CETP, and in collaboration with Ross Milne and Yves Marcel developed CETP mAbs. Injection of these antibodies into rabbits led to an increase in HDL-C concentrations, reflecting a delay in HDL-CE catabolism, providing the first in vivo evidence for a role of CETP in HDL metabolism. Using these antibodies and the recently elaborated CETP gene sequence, we collaborated with Mabuchi and Inazu to elucidate the first human genetic deficiency state of CETP in Japanese families, a single nucleotide change in the fourteenth intron of the CETP gene that resulted in an absence of plasma CETP (3). This mutation resulted in remarkably high HDL-C and apolipoprotein A-1 (apoA-1) concentrations and moderately reduced LDL-C and apolipoprotein B (apoB) concentrations. The elucidation of this lipoprotein phenotype, typically associated with low CHD risk, set off a stampede to develop CETP inhibitors in industry. A warning shot was our discovery, ironically using samples from the Hawaii-Japanese study, that genetic CETP deficiency caused by the intron 14 splicing defect or a reduced function CETP missense variant (D442G), might actually be associated with a modest excess of coronary heart disease (CHD), especially following adjustment for HDL-C concentrations (i.e., the expected protection from elevated HDL was not apparent). With Agellon and Breslow we used a human CETP minigene to prepare transgenic mice, confirming that CETP gene expression was induced by high cholesterol diets (later shown to be mediated through transcriptional induction by liver x receptor/retinoid x receptor (LXR/RXR) and resulted in decreased HDL-C and increased apoB and LDL-C concentrations. However, the introduction of the CETP transgene into a mouse with hypertriglyceridemia and accelerated atherosclerosis (caused by overexpression of human apolipoproteinC3) markedly lowered HDL-C, but did not affect atherosclerosis (4). Other human genetic studies and mouse models suggested a pro-atherogenic effect of CETP expression (5) and several companies persisted in the development of CETP inhibitors.

The history of the clinical development of CETP inhibitors has been reviewed elsewhere: the negative outcomes of most studies may be attributed to off-target toxicity, premature cessation of studies, and poor study design (6). Most recently, the REVEAL trial with the CETP inhibitor anacetrapib, the largest trial and the first to go to completion, did show a highly significant but modest reduction in CHD (7). The magnitude of the CHD benefit appeared to correlate with the reduction in non-HDL cholesterol levels. CETP inhibitors may still find a place in therapy based on their ability to lower LDL-C. However, in none of the CETP inhibitor trials was there any suggestion that the highly leveraged benefit predicted from increases in HDL-C of 40 mg/dL or more had affected the outcome. Along with the failed trials of niacin therapy, the so-called “HDL hypothesis” that increasing HDL-C levels therapeutically would consistently produce a major reduction in CHD risk has been thoroughly disproved.

Groups led by Assman, Brewer and Hayden made the important discovery that mutations in the ATP binding cassette transporter A1 (ABCA1) were the underlying cause of Tangier Disease, a disorder associated with very low HDL-C concentrations and accumulation of cholesterol loaded macrophage foam cells in tissues. Oram and we showed that ABCA1 is a phospholipid/cholesterol transporter that binds apoA-1, promoting cholesterol efflux from cultured cells including macrophage foam cells, while Parks demonstrated that ABCA1 initiates formation of HDL in the liver and small intestine (8). We showed that ABCA1 gene expression is induced by LXR/RXR transcription factors, providing a molecular mechanism to link the accumulation of cellular cholesterol to the upregulation of cellular cholesterol efflux (9). Thus the elucidation of Tangier Disease demonstrated the importance of cellular cholesterol efflux in the regulation of HDL formation and in macrophage cholesterol accumulation. However, Tangier Disease patients have not shown a consistent relationship to CHD, probably because these subjects also have reduced LDL-C concentrations.

Mendelian Randomization Studies

Human genome-wide association studies (GWAS) have uncovered a plethora of single nucleotide polymorphisms (SNPs) associated with HDL-C concentrations, leading to Mendelian Randomization (MR) studies to determine whether SNPs that associate with HDL-C also associate with CHD. Early studies using single variants and polygenic risk scores showed that LDL-C, but not HDL-C, is associated with CAD risk. Kathiresan et al. modeled the impact on CHD of 185 variants associated with LDL-C, HDL-C, or triglyceride (TG), and found that LDL-C and TG, but not HDL-C, are causally associated with CHD risk (10). One limitation of the MR approach is confounding due to pleiotropy. MR analysis of these data with error-weighting or exclusion of pleiotropic variants associated with body mass index or blood pressure found that HDL-C is associated with CHD (11). Multivariate MR of lipid traits in CHD using larger CHD datasets also detected an HDL-C association, but this association became nonsignificant after mathematical modeling of unmeasured pleiotropy; the TG relationship was also attenuated but remained significant (12). These somewhat equivocal results have been widely interpreted as indicating that HDL-C is not in the causal pathway of atherosclerosis.

While some of the SNPs associated with HDL-C were found in genes already known to regulate HDL, the majority were in or near novel genes, holding promise to reveal new mechanisms regulating HDL-C levels. A few studies have sought to elucidate the underlying mechanisms. One of these SNPs is in a gene called TTC39B that encodes a scaffolding protein; the lower expression variant associates with increased HDL-C concentrations. Our mouse studies showed that the decreased expression of TTC39B in enterocytes leads to stabilization of LXR and increased intestinal production of HDL (13). Interestingly, the reduced expression human TTC39B genetic variant but not other HDL-C SNPs, also associates with gallstone disease. Since SNPs in ABCG5/8 (LXR targets) strongly associate with gallstones, the TTC39B association may reflect LXR-mediated induction of ABCG5/8.

Cholesterol Efflux and CHD

The relevance of HDL-mediated macrophage cholesterol efflux to human atherosclerosis has been strongly supported by the discovery of Rader, Rothblat et al. that the ability of different patients’ HDL to promote cholesterol efflux from cultured macrophages [called the cholesterol efflux capacity (CEC)] is inversely related to the coronary atheroma burden (14). Subsequent studies have confirmed that the CEC of HDL is inversely related to incident coronary heart disease (CHD), even following adjustment for HDL-C concentrations, suggesting that CEC is measuring a key function of HDL relevant to CHD that is not well captured by HDL-C measurements. Studies from the Heinecke lab suggest that HDL particle number, as determined by ion mobility shift, may correlate better than HDL-C with CEC, coronary endothelial function, and CHD prediction (15). However, the relationship between CEC and CHD has not been consistent in all studies. With Shea, we recently assessed the relationship between macrophage cholesterol efflux and incident CHD using a nested case-control study design in a 10-year follow-up of the MESA cohort (16). In contrast to earlier studies that employed radioactive or fluorescent cholesterol analogs, we used a more specific measure of HDL-mediated changes in cholesterol mass in media of cultured macrophages, termed cholesterol mass efflux capacity (CMEC). These studies provide strong confirmation of the inverse relationship between cholesterol efflux capacity and CHD. They also show no relationship of CMEC to nonhemorrhagic stroke or peripheral artery disease. These findings suggest a particular impact of HDL-mediated cholesterol efflux on atherothrombosis in the coronary arteries.

Studies in Mice Show Anti-Atherogenic Effects of Increased HDL Production and Reverse Cholesterol Transport (RCT)

The evidence that HDL and RCT mediate anti-atherogenic effects has been strongly supported by animal studies. Fuster, Rubin, Breslow, Fisher and their colleagues showed that infusion of HDL or increased expression of APOA1 reduce atherosclerosis, both in progression and regression models (17, 18). Such studies have stimulated trials of infusion therapies of reconstituted HDL particles as a treatment for CHD in humans. The challenge has been to infuse sufficient amounts of reconstituted HDL of appropriate composition to stimulate cholesterol efflux without inducing hemolysis or hepato-toxicity (19).

Subsequent to the work on ABCA1 and Tangier Disease described above, Wang discovered that a second transporter ABCG1 (also an LXR target) promotes cholesterol efflux to HDL particles, especially larger HDL species, in contrast to ABCA1 that mainly promotes cholesterol efflux to lipid poor apoA-1 or small HDL particles (20). Rader et al. found that macrophage RCT (the transport of ³H-cholesterol from macrophages to the liver or feces) is dependent on macrophage ABCA1/G1 and that increased macrophage RCT in general correlates with reduced atherosclerosis. We have used Abca1^f/fAbcg1^f/f mice to probe the impact of defective efflux of cholesterol from different cell types. Marit Westerterp showed that knocking out these transporters in myeloid cells resulted in an increase in atherosclerosis, driven in part by increased inflammatory processes and, for endothelial cells, decreased endothelial nitric oxide synthase activity (21, 22). Studies by Yvan-Charvet and Murphy revealed an important role of ABCA1, ABCG1, APOE and LXRs in the regulation of cholesterol accumulation in hematopoietic stem and progenitor cells, leading to increased interleukin-3/granulocyte-macrophage colony-stimulating factor driven proliferation of hematopoietic stem and progenitor cells (HSPCs), increased myelopoiesis, and accumulation of monocytes and macrophages in lesions (23). Along with studies from the Swirski and Randolph labs, these findings have linked dyslipidemia and defective cholesterol efflux to HSPC proliferation, aberrant myelopoiesis, and atherogenesis. Our recent studies have shown that myeloid ABCA1/G1 deficiency leads to NLRP3 inflammasome activation and accelerated atherosclerosis with prominent features of neutrophil extracellular trap (NET) formation (24). These studies establish that NETosis may be secondary to myeloid cell inflammasome activation.

LXRs Coordinate Cholesterol Efflux and RCT and Suppress Atherosclerosis

Various molecules involved in macrophage cholesterol efflux, transport in blood and excretion into bile were found to be LXR/RXR targets, suggesting that LXRs might coordinate the overall process of reverse cholesterol transport (23). The Tontonoz lab showed that LXR activators are potently anti-atherogenic, showing that activation of RCT reduces atherosclerosis. We found that LXR activators suppress atherosclerosis, even in the absence of hematopoietic ABCA1/ABCG1 expression, pointing to the importance of cholesterol efflux-independent anti-inflammatory effects of LXRs in suppression of atherosclerosis. Our recent studies have shown on a genome-wide level that the mechanism of inflammatory repression involves the direct binding of LXRs to the promoter/enhancer regions of toll-like receptor 4 (TLR-4)-driven inflammatory genes (25). Most likely the determination of whether LXRs activate or repress gene transcription depends on the specific co-regulators and chromatin structures at different gene enhancers. Thus, it may be possible to design drugs that mediate anti-inflammatory effects while sparing adverse effects of LXR activators on LDL-C levels and fatty liver.

Role of HDL in the Immune System and Response to Infections

HDL and LXRs exert important immunosuppressive effects (23). HDL-mediated cholesterol efflux reduces TLR-4 levels and signaling in response to lipopolysaccharide in macrophages, resulting in decreased release of inflammatory cytokines (26). A secondary effect of HDL-mediated cholesterol efflux is to suppress Type 1 interferon responses by decreasing signaling via the macrophage interferon receptor. Thus, HDL acts to suppress innate immune responses. Conversely, the acute phase response suppresses macrophage cholesterol efflux and RCT at multiple nodes and this creates a positive feedback loop to further enhance the acute phase response (23). In the acute setting, this is beneficial in terms of antibacterial and antiviral immunity, but when chronic becomes dysfunctional. Acute infections may thus suppress HDL levels, cholesterol efflux, and RCT, enhancing the innate immune response to bacteria and viruses. During recovery from infection the increase in HDL and cholesterol efflux may help to resolve inflammatory responses. These effects may represent beneficial adaptions that help to enhance the resolution of microbial-driven inflammatory processes in vivo. However, during chronic sterile inflammation as occurs in obesity and atherosclerosis, the repression of cholesterol efflux and RCT may worsen the underlying diseases.

Summary and Future Perspective

While HDL-C measurements provide useful clinical information, the negative or suboptimal results of clinical trials indicate that HDL-C per se should no longer be viewed as a target for therapy. MR studies have suggested that HDL-C associated SNPs, once adjusted for other risk factors and pleiotropy, do not strongly associate with CHD. However, HDL-associated SNPs have the potential to reveal new biological functions and disease mechanisms linked to HDL-C. The ability of HDL to mediate cholesterol efflux from cultured cells correlates well with coronary atherothrombotic disease, suggesting that measuring CEC or CMEC may better predict the outcome of HDL therapeutic interventions than HDL-C. Accordingly, infusion of reconstituted HDL particles markedly increases HDL CEC and is under evaluation for prevention of recurrent atherothrombotic disease in the AEGIS-II trial. LXR activators targeted to macrophages may also induce macrophage cholesterol efflux and suppress inflammation with therapeutic benefit.

In sum, HDL research in the last 5 decades has helped to clarify the complex biology of HDL and its association with CHD, but much remains to be learned. The failure of clinical trials targeting HDL-C has helped to clear the way for the pursuit of other atherosclerosis treatments such as those targeting triglyceride-rich lipoproteins, antinflammatory mechanisms, or more relevant antiatherogenic functions of HDL linked to cholesterol efflux and inflammation.

 

Human Genes: ABCA1, ATP binding cassette subfamily A member 1; APOA1, apolipoprotein A1; CETP, cholesteryl ester transfer protein; TTC39B, tetratricopeptide repeat domain 39B.

Author Contributions: All authors confirmed they have contributed to the intellectual content of this paper and have met the following 4 requirements: (a) significant contributions to the conception and design, acquisition of data, or analysis and interpretation of data; (b) drafting or revising the article for intellectual content; (c) final approval of the published article; and (d) agreement to be accountable for all aspects of the article thus ensuring that questions related to the accuracy or integrity of any part of the article are appropriately investigated and resolved.

Authors’ Disclosures or Potential Conflicts of Interest: Upon manuscript submission, all authors completed the author disclosure form. Disclosures and/or potential conflicts of interest:

Employment or Leadership: None declared.

Consultant or Advisory Role: A.R. Tall, consultant to CSL, Amgen, and Astra-Zeneca, and on the SAB of Staten Biotechnologies and Fortico Biotech.

Stock Ownership: None declared.

Honoraria: None declared.

Research Funding: Supported by NIH grant HL 107653.

Expert Testimony: None declared.

Patents: None declared.