Abstract
Triglycerides, cholesterol, and their metabolism are linked due to shared packaging and transport within circulating lipoprotein particles. While a case for a causal role of cholesterol-carrying low-density lipoproteins (LDLs) in atherosclerosis is well made, the body of scientific evidence for a causal role of triglyceride-rich lipoproteins (TRLs) is rapidly growing, with multiple lines of evidence (old and new) providing robust support.
This review will discuss current perspectives and accumulated evidence that an overabundance of remnant lipoproteins stemming from intravascular remodeling of nascent TRLs—chylomicrons and very low-density lipoproteins (VLDL)—results in a proatherogenic milieu that augments cardiovascular risk. Basic mechanisms of TRL metabolism and clearance will be summarized, assay methods reviewed, and pivotal clinical studies highlighted.
Remnant lipoproteins are rendered highly atherogenic by their high cholesterol content, altered apolipoprotein composition, and physicochemical properties. The aggregate findings from multiple lines of evidence suggest that TRL remnants play a central role in residual cardiovascular risk.
Introduction
Triglycerides, cholesterol, and their metabolism are inextricably linked due to shared packaging and transport within circulating lipoprotein particles. While a causal relationship for cholesterol-carrying low-density lipoproteins (LDLs) in atherosclerosis is indisputable, the body of scientific evidence for a causal role of triglyceride-rich lipoproteins (TRLs) is rapidly growing, with multiple lines of evidence now adding robust support. Overabundance of these TRLs—chylomicrons and very low-density lipoproteins (VLDL)—results from increased production or impaired clearance and is clinically evident as hypertriglyceridemia conveniently detected upon routine lipid testing. The concept that hypertriglyceridemia signals clinically relevant abnormalities in lipid metabolism and a proatherogenic milieu has not been widely embraced by the general medical community. However, building on observations made decades ago, novel kinetic investigations, advances in laboratory methods for lipoprotein quantification, epidemiologic observations, genomic science, and clinical trial data have refocused attention on a potential central role of TRLs in residual cardiovascular risk, i.e., that risk mediated through pathways not addressed by our current LDL management algorithms. To understand these important developments, this narrative review will discuss the relevant biologic and clinical observations pertaining to TRLs and TRL ‘remnants’ and examine their role as therapeutic targets for cardiovascular disease prevention.
Contemporary Estimates of Hypertriglyceridemia Prevalence
The presence of hypertriglyceridemia varies greatly by country and region with limited contemporary comparative data. Nonetheless, a general prevalence estimate of 10% among adult populations has been suggested (1). In the US, over the period 2007–2014, this rate is about 1 in 4 having triglyceride (TG) concentrations >150 mg/dL (1.7 mmol/L) with important ethnic and racial differences, Mexican Americans having the highest prevalence, followed by non-Hispanic Whites and non-Hispanic Blacks (2, 3). Longitudinal data have shown a decline in prevalence that can be linked to increasing use of lipid lowering therapy, in particular, statins that have modest TG-lowering effects (4). Nonetheless, among high risk groups such as persons with diabetes or prior cardiovascular disease, these generally favorable trend data belie an important problem. In US adults with diabetes and LDL-C < 70 mg/dL (<1.8 mmol/L) on statin therapy, about 30% had residual hypertriglyceridemia (TG > 150 mg/dL; >1.7 mmol/L) (5). In a large population-based study (Ontario, Canada) of patients with established cardiovascular disease and controlled LDL-C, 1 in 4 had increased TGs (TG 135–499 mg/dL; 1.52–5.63 mmol/L) (6). While these data reveal opportunities for risk mitigation, the focus on hypertriglyceridemia diverts attention from the underlying physiologic abnormality, a shift in distribution of cholesterol trafficking lipoproteins, that is likely the more credible therapeutic target.
Basic Principles of Triglyceride-Rich Remnant Metabolism
In humans, cholesterol accumulation and only modest triglyceride accumulation characterizes atherosclerotic lesions. Thus, most conventional explanations for atherosclerosis susceptibility do not clearly implicate hypertriglyceridemia. Hypertriglyceridemia fuels TRL production, accelerates remodeling of cholesterol-carrying lipoproteins (7–9), and potentiates inflammatory pathways (10–12) but is unlikely to directly instigate atherosclerosis. TGs are primarily transported by TRLs comprising chylomicrons secreted by intestinal enterocytes, VLDL particles synthesized in the liver and more recently discovered to have intestinal origins (13), and their metabolic derivatives or ‘remnants’. Intravascular lipolysis of newly secreted TRLs gives rise, in the case of chylomicrons, to chylomicron remnants and in the case of VLDL to a spectrum of highly modified particles including remnants within the VLDL density range, intermediate-density lipoproteins (IDLs), and LDLs. These smaller highly modified TRL remnant particles are increasingly believed to be causative agents in atherosclerosis.
The metabolic fate of TRLs (degree of remodeling) and plasma abundance depends on the size of the TG pool, as well as hormonal, nutritional, and genetic traits in a given individual. Their atherogenic burden depends on structural properties that determine plasma residence time, size-related rates of arterial influx, residual cholesterol content, subendothelial retention time, and capacity to elicit foam cell formation and promote plaque formation. Basic knowledge of the origin of these lipoproteins and TRL metabolism pathways (topics discussed below) is essential to understanding their role in atherosclerosis. For a more detailed description, the reader is referred to several recent comprehensive reviews (14–16).
Chylomicrons (diameter >75 nm) are the largest and lowest-density lipoproteins synthesized and secreted from the small intestine after ingestion of dietary fat. Following release from the intestinal mucosa into the lymphatic system, newly secreted (nascent) chylomicrons enter the systemic circulation via the thoracic duct. Chylomicron TGs undergo rapid hydrolysis in a process much faster than VLDL TGs (17–19). Resulting remnants retain almost all the cholesterol contained in their precursors and acquire several molecules of apoE via exchange with high-density lipoprotein (HDL) in plasma or during transit within hepatic tissue (20, 21). Chylomicron remnants are ultimately cleared by binding the LDL receptor or LDL receptor protein-1 (LRP-1) via apoE interactions and via the heparan sulfate proteoglycan pathway in the liver (22–24). Within hepatocytes, cholesterol esters are digested by lysosomal enzymes, which release unesterified cholesterol that becomes available for repackaging in VLDL.
VLDLs are synthesized primarily in the liver and like chylomicrons fulfill a lipid transport function. Two major physically distinct species of VLDL exist: larger VLDL1 (50–80 nm diameter, circa 70% TG mass) and smaller VLDL2 (30–50 nm diameter, circa 30% TG mass). As a function of TG availability, hepatocytes vary the amount of lipid loaded onto lipoprotein particles in the endoplasmic reticulum and can assemble and secrete newly synthesized particles that span the entire size range (VLDL1-VLDL2-IDL-LDL). At normal TG concentrations (<100 mg/dL; <1.2 mmol/L), VLDL1 and VLDL2 circulate in approximately equal proportions. A number of factors influence VLDL1 secretion including hepatic TG stores, insulin resistance, chylomicron remnant clearance pathways, and regulatory pathways for de novo lipogenesis (25, 26). Synthesis of VLDL2 is more influenced by endogenous cholesterol synthesis, is less dependent on increased TGs and is less dominant in hypertriglyeridemia (27–30). Once released, progressive delipidation (removal of TGs) of both VLDLs and intravascular remodeling yields remnant particles in the IDL and LDL density intervals. Compared to chylomicron remnants, which are largely cleared by the liver, 25–75% of VLDL remnants do not undergo hepatic uptake but are rather converted to LDL (14).
Fisher first proposed (31), and Packard and colleagues confirmed (32), that ‘metabolic channeling’ within the VLDL-LDL delipidation cascade occurs such that VLDL1 and VLDL2 have differing rates of conversion to IDL and LDL. LDL derived from VLDL1 was also shown to have a slower clearance rate and longer duration to decay. It has been postulated that these slowly metabolized LDL particles of VLDL1 lineage are a major source of small dense LDL particles (7, 33), thereby providing a critical link between hypertriglyceridemia and other markers of atherogenic dyslipidemia. These observations also suggest that lipoprotein pedigree, whether newly secreted or the result of TRL catabolism, determines physicochemical properties and metabolic fate.
Remnants as Biologic Drivers of Atherosclerosis
Newly secreted chylomicrons, VLDLs, IDLs, and LDLs although structurally distinct from remnants may occupy the same density and size range as TRL remnants that have undergone extensive intravascular remodeling (Figure 1). Thus, size alone cannot be the determinant of remnant atherogenicity. The actions of 2 enzymes, lipoprotein lipase (LPL) and hepatic lipase (HL), and a specific transfer protein, cholesterol ester transfer protein (CETP), induce major structural changes that distinguish remnants from nonremnant lipoproteins and may enhance atherogenicity. Through rapid LPL-mediated (and to a lesser degree HL-mediated) removal of TGs and CETP-mediated exchange of TGs for cholesterol from LDL and HDL, remnant particles contain disproportionately more cholesterol than nascent chylomicrons or VLDL (34). Compared with their parent TRLs, with progressive catabolism, remnants also lose their complement of C apolipoproteins (including apoCIII) and become enriched in apoE (14). These changes appear due to reductions in volume and surface area that preferentially drive the movement of apoCs to HDL and the direct addition of apoE (35–37). However, hypertriglyceridemia alters this redistribution (38). Normal transfer of apoCIII to HDL is disrupted, and remnant apoCIII stores are retained. Hypertriglyceridemia also drives increased hepatic production of VLDL associated apoCIII (39). The net result is that apoCIII accumulates in VLDL, where it plays an inhibitory role on LPL and hepatic VLDL uptake ultimately leading to longer plasma residence time of these particles.
While chylomicrons and large VLDLs are prohibited from transcytosis by virtue of their size, smaller chylomicron and VLDL remnants can and do penetrate the artery wall (40–42). Using fluorescently labeled VLDL, Rutledge and colleagues showed that lipolysis of VLDL enhanced by more than 2-fold its ability to cross the endothelium and accumulate in the subendothelial space (43, 44). Upon entry, these particles become trapped by subendothelial proteoglycans and are susceptible to engulfment by vascular macrophages. As in the liver, apoE is a major determinant of macrophage remnant uptake by receptor-mediated endocytosis via binding to the LDL and LRP-1 receptors (16, 45, 46). In a process akin to LDL, ensuing lipid loading and foam cell formation is thought to elicit a similar cascade of events leading to plaque formation (34). However, the relative contribution of TRLs to this process requires further definition.
Even though remnant particles remain more triglyceride- than cholesterol-rich, their large size means 5–20 times more cholesterol may deposit per particle than LDL (47). Recent data from over 15 000 healthy individuals also indicate that the postprandial state may influence the cholesterol concentrations of TRL particles more so than LDL particles (48). Whether TRL pedigree (remnant vs nonremnant origin) impacts cholesterol delivery has not been well studied. However, TRL modification and longer residence in circulation can result in greater exposure to the vessel wall (44, 49). Unlike native LDL, which may exit the subendothelial space almost as rapidly as it enters, remnants efflux very slowly compared to their rate of entry with increased opportunity for internalization by macrophages and foam cell formation (44, 50). Also distinct from LDL, which is constitutively deficient in apoE and requires oxidation for uptake, remnants do not require oxidation to facilitate accumulation in macrophages due to their apoE enrichment . Another consequence of remnant retention is local TG lipolysis. Like adipocyte and myocytes, atherosclerotic plaque macrophages have the capacity to produces substantial quantities of LPL (51–53). LPL-mediated liberation of free fatty acids from accumulated remnants can induce production of proinflammatory intermediaries (cytokines, interleukins, and adhesion molecules) that accelerate the recruitment of leukocytes to areas of inflammation. Botham and Wheeler-Jones (54) provide an excellent mechanistic review of TRLs in vascular homeostasis. These concepts and experimental findings are consistent with the seminal Zilversmit hypothesis (55, 56), which proposed that TRLs, via remnant infiltration, act additively with LDL to direct atheroma propagation.
Remnant TRL Quantification: To Calculate or to Measure, and How?
As cholesterol trafficking by remnant TRLs is the favored mechanism in atherogenesis, optimal methods to measure remnant TRL cholesterol (TRL-C) are required for clinical studies and, ultimately, clinical practice. Accurate measurement and characterization of remnant cholesterol, however, has proven to be challenging for 2 major reasons: (1) due to their rapid and ongoing catabolism, TRL remnant size, quantity, and composition are highly dynamic; and (2) despite their reduced size and triglyceride content, TRL remnants are difficult to differentiate from their precursors (nonremnant lipoproteins). While no universally accepted definition of TRL remnants currently exists, one that is generally used as a goal for detection is postlipolytic partially TG-depleted particles derived from chylomicrons and VLDL that are relatively enriched in cholesteryl esters and apoE. It is also acknowledged that nomenclature in laboratory methods is highly variable. In this review, we use ‘TRL-C’ to reflect cholesterol content of all TRLs and specify ‘remnant TRL-C’ when referencing methods to isolate remnant cholesterol.
The simplest way to estimate TRL-C is by rearrangement of the Friedewald equation, which was initially developed to estimate LDL-C (i.e., LDL-C = total cholesterol − HDL cholesterol − triglyceride/5) (57). Using this equation, TRL-C (or VLDL-C as initially described) is estimated as 1/5 of triglyceride concentration in mg/dL (or 1/2.2 in mmol/L). However, by using a fixed ratio, the predicted value is equivalent to TG/5 and Friedewald estimated TRL-C does not add clinical information beyond TG concentrations alone. Consensus recommendations from the European Atherosclerosis Society and European Federation of Clinical Chemistry and Laboratory Medicine now recommend that a direct LDL-C measurement be used when the Friedewald formula is used for TRL-C estimation (58). Some have used this equation to estimate remnant TRL-C under the assumption that because of rapid catabolism, all circulating cholesterol in the VLDL component of the formula can be considered remnant cholesterol. This has not been well defined.
Apart from calculation, several direct methods are available by which TRL remnants are identified, isolated, and quantified on the basis of their density, charge, size, specific lipid components, apo composition, or apo specificity (59). Major methods that will be discussed include preparative ultrafiltration, immunoseparation, nuclear magnetic resonance (NMR), and direct homogenous assays, each of which have been shown in clinical studies to identify individuals with increased cardiovascular risk.
Preparative ultracentrifugation is considered the reference standard for lipoprotein separation and relies on relative flotation based on particle density. As remnant TRLs are smaller and denser than their precursors, those derived from VLDL are mainly found in the IDL range (d = 1.006–1.019 g/mL) with some present in the VLDL range (d = 0.930–1.006 g/mL). Chylomicron remnants are found in the chylomicron, VLDL or IDL range. Figure 1 illustrates this overlap and limitation, specifically, ultracentrifugation cannot differentiate TRL remnants from other nonremnant lipoprotein particles of similar density. In addition, most traditional ultracentrifugation methods are time intensive, taking several hours to days for completion. The vertical auto profile (VAP) technique is a modified approach that uses a vertical rotator for lipoprotein separation leveraging the shorter horizontal axis of the centrifuge tube rather than the longer vertical axis used in preparative ultracentrifugation (60). It suffers similar limitations in that remnants span multiple density zones and the method requires a special analyzer that is not available in most clinical laboratories.
Depiction of remnant particles by major apolipoproteins, size, nomenclature, and density. Abbreviations: CM, chylomicron; VLDL, very low-density lipoprotein; IDL, intermediate-density lipoprotein; LDL, low-density lipoprotein; HDL, high-density lipoprotein; LB-LDL, large buoyant LDL; SD-LDL, small dense LDL; Lp(a), lipoprotein(a). Adapted from Chait et al. (14).
Immunoseparation is a commercially and widely available method that detects ‘remnant-like particle cholesterol’ (RLP-C), which is the unbound fraction not adsorbed by anti-apoA1 or anti-apoB100 antibodies (61, 62). The anti-apoB100 antibody used does not bind apoE-rich VLDL or TRLs containing apoB48 (intestinally derived chylomicrons and VLDLs) and when combined with an apoA1 antibody, the immunoaffinity gel mixture adsorbs nascent apoB100 lipoproteins (LDL and VLDL) and apoA1-containing lipoproteins (HDL). The unbound fraction is then apoE enriched TRL remnants derived from chylomicrons and VLDL. The method requires pretreatment, specialized equipment, and is time-consuming and, although widely used, the epitope that -allows selective precipitation of apoB in nascent VLDL but not in remnants is not fully characterized and it is unclear whether remnants are under or overestimated by this approach (63, 64).
NMR relies on spectroscopic signals that differentiate lipoproteins based on their bulk lipid mass and requires no physical separation prior to analysis (65). Computational deconvolution of the proton NMR signal of lipid methyl groups is used for lipoprotein classification and subclass delineation. Testing can be quick (<1 minute) and does not require chemical pretreatment. Ongoing refinement of deconvolution algorithms has provided better resolution of TRL subclasses and offers unique information about particle size, particle composition, and particle number. These attributes permit ‘tagging’ of intermediate sized particles that may correspond to remnant lipoproteins. However, like ultracentrifugation, NMR separation of TRL particles does not selectively identify remnant lipoproteins that can span the size and lipid composition range, and this approach generally leads to higher estimates than other methods (14).
Direct homogenous assays offer the significant advantages of convenience and suitability for hospital- and clinic-based testing because these assays do not require pretreatment or skilled operation and can be performed on widely available automated analyzers. The method relies on specific interactions of enzymes and surfactants with lipoproteins (63). First-generation assays detected only a subfraction of TRL-C (<15%) when compared to calculated values (66) with newer assays directly measuring a substantially larger fraction (67). Scatterplots of earlier and newer generation assays from Denka performed in sequential analyses from the Copenhagen General Population Study cohorts illustrate this point (Fig. 2). The second- but not first-generation assay has been shown to associate with cardiovascular events (68, 69).
It is clear that no optimal way of accurately quantifying remnant TRLs currently exists. The lack of a universally accepted definition of remnant TRLs is perhaps the greatest barrier as consensus criteria on size, density, lipid, and apolipoprotein content are lacking. Yet, observations from numerous avenues of investigation demonstrate a link between increased remnant TRLs, assessed by a range of methods, and cardiovascular disease. Thus, there is a pressing need for both a consensus definition of TRL remnants, and accurate and reproducible methods for quantitation.
Clinical Studies of TRLs and Future Atherosclerotic Risk
Evidence from Observational Epidemiology
Numerous epidemiologic studies have demonstrated a positive association of remnant lipoprotein concentrations with incident cardiovascular events (Table 1). The largest of these derive from 3 Danish prospective cohorts (the Copenhagen General Population Study, the Copenhagen City Heart Study, and the Copenhagen Ischemic Heart Disease Study) that have provided a plethora of insights into the role of hypertriglyceridemia and remnant cholesterol, nonfasting concentrations in particular, as they relate to cardiovascular risk. Many other analyses have been conducted under fasting conditions that may underestimate risk associations, as fasting allows time for catabolism of dietary TGs and does not capture postprandial surges in remnants. Nonetheless, all but one of 17 major published reports from 13 independent cohorts demonstrated statistically significant associations of increased remnants with future vascular events. Null findings from the Atherosclerosis Risk in Communities Study (ARIC) (69) may relate to several factors, specifically fasting conditions, use of an earlier generation direct homogenous assay, and, importantly, adjustment for HDL in multivariable models. Whether or not to adjust for TG and HDL-C concentrations in risk models is controversial, since adjustment may overcontrol for these correlated risk factors and may be unnecessary as the weight of scientific evidence now suggests that TG and HDL-C concentrations are likely markers of aberrant remnant metabolism rather than independent causal agents.
Major prospective studies of remnant cholesterol and incident cardiovascular disease.
| Author | Year published | Sample size | Study setting | Fasting status | Measurement | Outcome | Adjustedarisk association HR or OR (95% CI) |
|---|---|---|---|---|---|---|---|
| Phillips NR et al. (70) | 1993 | 335 | Montreal Heart Study | Fasting | Ultracentrifugation: IDL-C+VLDL-C | Composite: MI, coronary revascularization, cardiac death | 1.02b (1.01–1.03) per 1 mg/dL increase |
| Kugiyama K et al. (71) | 1999 | 147 | Patients with Angiographic CAD | Fasting | Immunoseparation remnant TRL-C | Composite: MI, recurrent, or refractory angina requiring coronary revascularization, cardiac death | 6.38b (2.3–17.6) highest vs lowest tertile |
| McNamara JR et al. (72) | 2001 | 1567 | Framingham Heart Study, Women | Fasting | Immunoseparation remnant TRL-C | Composite: angina, MI, TS, TIA | 2.27b (1.37–3.77) highest quartile vs all others |
| Fukushima H et al. (73) | 2004 | 120 | Patients with Angiographic CAD and Diabetes | Fasting | Immunoseparation remnant TRL-C | Composite: MI, refractory angina requiring coronary revascularization, cardiac death | 2.2b (1.2–6.4) highest quartile vs all others |
| Imke C et al. (74) | 2005 | 1156 | Japanese-American Men | Fasting | Immunoseparation remnant TRL-C | Composite: MI, revascularization, sudden death | Normal TG: 1.009 (0.939–1.085) High TG: 1.011 (1.001–1.021) per mg/dl |
| Mora S et al. (75) | 2009 | 27 673 | Women’s Health Study | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, IS, coronary revascularization, CVD death | Small VLDL: 1.56 (1.27–1.91) Medium VLDL: 1.46 (1.17–1.82) Large VLDL: 1.77 (1.34–2.33) Total VLDL: 1.71 (1.38–2.12) highest vs lowest quintile |
| Varbo A et al. (76) | 2013 | 73 513 | Copenhagen General Population Study, Copenhagen City Heart Study, and Copenhagen Ischemic Heart Disease Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | Ischemic heart disease | 2.3 (1.7–3.1) highest vs lowest quintile |
| Joshi PH et al. (77) | 2016 | 4114 | Jackson Heart Study, Black Participants | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.18 (1.00–1.39) per SD increase |
| Joshi PH et al. (77) | 2016 | 818 | Framingham Offspring Cohort Study | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.46 (1.05–2.04) per SD increase |
| Lawler P et al. (78) | 2017 | 11 984 | Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, stroke, hospitalization for unstable angina, arterial revascularization, and CVD death | Small VLDL: 1.16 (0.99–1.36) Medium VLDL: 1.30 (1.10–1.53) Large VLDL: 1.17 (1.00–1.37) Total VLDL: 1.21 (1.04–1.41) per SD increase |
| Varbo A et al. (67) | 2018 | 106 216 | Copenhagen General Population Study, Stratified by Weight | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | MI | Normal Weight: 2.0 (1.3–3.2) Overweight: 1.9 (1.4–2.6) Obese: 2.3 (1.4–3.5) highest vs lowest quartile |
| Aday A et al. (79) | 2018 | 27 888 | Women’s Health Study | Nonfasting | NMR very small, small, medium, large, very large, and total VLDL particle concentration | PAD | Very Small VLDL: 1.01 (0.61–1.67) Small VLDL: 0.99 (0.62–1.59) Medium VLDL: 1.98 (1.15–3.41) Large VLDL: 1.58 (0.94–2.68) Very Large VLDL: 1.68 (1.05–2.55) Total VLDL: 1.39 (0.81–2.38) highest vs lowest tertile |
| Holmes M et al. (80) | 2018 | 4662 (Nested Case-Control) | Chinese Kadoorie Biobank (N = 512 891) | Nonfasting | NMR metabolomics remnant cholesterol (VLDL-C plus IDL-C) | MI, IS individually | 1.27 (1.15–1.39) for MI 1.20 (1.09–1.32) for IS per SD increase |
| Saeed A et al. (69) | 2018 | 9334 | Atherosclerosis Risk in Communities Study | Fasting | Direct homogenous assay for TRL-C, first generation | CHD, IS, CVD individually | b1.06 (0.88–1.27) for CHD b1.07 (0.78–1.45) for IS b1.05 (0.89–1.23) for CVD highest vs lowest quartile |
| Vallejo-Vaz et al. (81) | 2018 | 9993 | Treating to New Targets | Fasting | Calculated as non-HDL-C minus LDL-C | Composite: MI, total stroke, resuscitated cardiac arrest, and CVD death | Atorvastatin 10 mg: 1.48 (1.15–1.92) Atorvastatin 80 mg: 1.01 (0.76–1.34) highest vs lowest quintile |
| Varbo A et al. (82) | 2019 | 102 964 | Copenhagen General Population Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | IS | 1.99 (1.49–2.67) highest vs lowest quartile |
| Duran EK et al. (68) | 2020 | 976 (Case-Cohort) | Women’s Health Study (N = 27 552) | Nonfasting | Direct homogenous assay for TRL-C, second generation | Composite: MI, IS, PAD, and CVD death | 1.97 (95% CI: 1.26–3.08) highest vs lowest quartile |
| Author | Year published | Sample size | Study setting | Fasting status | Measurement | Outcome | Adjustedarisk association HR or OR (95% CI) |
|---|---|---|---|---|---|---|---|
| Phillips NR et al. (70) | 1993 | 335 | Montreal Heart Study | Fasting | Ultracentrifugation: IDL-C+VLDL-C | Composite: MI, coronary revascularization, cardiac death | 1.02b (1.01–1.03) per 1 mg/dL increase |
| Kugiyama K et al. (71) | 1999 | 147 | Patients with Angiographic CAD | Fasting | Immunoseparation remnant TRL-C | Composite: MI, recurrent, or refractory angina requiring coronary revascularization, cardiac death | 6.38b (2.3–17.6) highest vs lowest tertile |
| McNamara JR et al. (72) | 2001 | 1567 | Framingham Heart Study, Women | Fasting | Immunoseparation remnant TRL-C | Composite: angina, MI, TS, TIA | 2.27b (1.37–3.77) highest quartile vs all others |
| Fukushima H et al. (73) | 2004 | 120 | Patients with Angiographic CAD and Diabetes | Fasting | Immunoseparation remnant TRL-C | Composite: MI, refractory angina requiring coronary revascularization, cardiac death | 2.2b (1.2–6.4) highest quartile vs all others |
| Imke C et al. (74) | 2005 | 1156 | Japanese-American Men | Fasting | Immunoseparation remnant TRL-C | Composite: MI, revascularization, sudden death | Normal TG: 1.009 (0.939–1.085) High TG: 1.011 (1.001–1.021) per mg/dl |
| Mora S et al. (75) | 2009 | 27 673 | Women’s Health Study | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, IS, coronary revascularization, CVD death | Small VLDL: 1.56 (1.27–1.91) Medium VLDL: 1.46 (1.17–1.82) Large VLDL: 1.77 (1.34–2.33) Total VLDL: 1.71 (1.38–2.12) highest vs lowest quintile |
| Varbo A et al. (76) | 2013 | 73 513 | Copenhagen General Population Study, Copenhagen City Heart Study, and Copenhagen Ischemic Heart Disease Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | Ischemic heart disease | 2.3 (1.7–3.1) highest vs lowest quintile |
| Joshi PH et al. (77) | 2016 | 4114 | Jackson Heart Study, Black Participants | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.18 (1.00–1.39) per SD increase |
| Joshi PH et al. (77) | 2016 | 818 | Framingham Offspring Cohort Study | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.46 (1.05–2.04) per SD increase |
| Lawler P et al. (78) | 2017 | 11 984 | Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, stroke, hospitalization for unstable angina, arterial revascularization, and CVD death | Small VLDL: 1.16 (0.99–1.36) Medium VLDL: 1.30 (1.10–1.53) Large VLDL: 1.17 (1.00–1.37) Total VLDL: 1.21 (1.04–1.41) per SD increase |
| Varbo A et al. (67) | 2018 | 106 216 | Copenhagen General Population Study, Stratified by Weight | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | MI | Normal Weight: 2.0 (1.3–3.2) Overweight: 1.9 (1.4–2.6) Obese: 2.3 (1.4–3.5) highest vs lowest quartile |
| Aday A et al. (79) | 2018 | 27 888 | Women’s Health Study | Nonfasting | NMR very small, small, medium, large, very large, and total VLDL particle concentration | PAD | Very Small VLDL: 1.01 (0.61–1.67) Small VLDL: 0.99 (0.62–1.59) Medium VLDL: 1.98 (1.15–3.41) Large VLDL: 1.58 (0.94–2.68) Very Large VLDL: 1.68 (1.05–2.55) Total VLDL: 1.39 (0.81–2.38) highest vs lowest tertile |
| Holmes M et al. (80) | 2018 | 4662 (Nested Case-Control) | Chinese Kadoorie Biobank (N = 512 891) | Nonfasting | NMR metabolomics remnant cholesterol (VLDL-C plus IDL-C) | MI, IS individually | 1.27 (1.15–1.39) for MI 1.20 (1.09–1.32) for IS per SD increase |
| Saeed A et al. (69) | 2018 | 9334 | Atherosclerosis Risk in Communities Study | Fasting | Direct homogenous assay for TRL-C, first generation | CHD, IS, CVD individually | b1.06 (0.88–1.27) for CHD b1.07 (0.78–1.45) for IS b1.05 (0.89–1.23) for CVD highest vs lowest quartile |
| Vallejo-Vaz et al. (81) | 2018 | 9993 | Treating to New Targets | Fasting | Calculated as non-HDL-C minus LDL-C | Composite: MI, total stroke, resuscitated cardiac arrest, and CVD death | Atorvastatin 10 mg: 1.48 (1.15–1.92) Atorvastatin 80 mg: 1.01 (0.76–1.34) highest vs lowest quintile |
| Varbo A et al. (82) | 2019 | 102 964 | Copenhagen General Population Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | IS | 1.99 (1.49–2.67) highest vs lowest quartile |
| Duran EK et al. (68) | 2020 | 976 (Case-Cohort) | Women’s Health Study (N = 27 552) | Nonfasting | Direct homogenous assay for TRL-C, second generation | Composite: MI, IS, PAD, and CVD death | 1.97 (95% CI: 1.26–3.08) highest vs lowest quartile |
Adjusted for clinical risk factors.
Some studies present only models additionally adjusted for TG and/or HDL-C.
Abbreviations: HR, hazards ratio; OR, odds ratio; CAD, coronary artery disease; MI, myocardial infarction; TIA, transient ischemic attack; TS, total stroke; IS, ischemic stroke; ICH, intracranial hemorrhage; PAD, peripheral artery disease; CHD, coronary heart disease; CVD, cardiovascular disease; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; VLDL, very low-density-lipoprotein cholesterol; IDL-C, intermediate-density lipoprotein cholesterol; TRL, triglyceride-rich lipoproteins; VAP, vertical auto profile; NMR, nuclear magnetic resonance.
Major prospective studies of remnant cholesterol and incident cardiovascular disease.
| Author | Year published | Sample size | Study setting | Fasting status | Measurement | Outcome | Adjustedarisk association HR or OR (95% CI) |
|---|---|---|---|---|---|---|---|
| Phillips NR et al. (70) | 1993 | 335 | Montreal Heart Study | Fasting | Ultracentrifugation: IDL-C+VLDL-C | Composite: MI, coronary revascularization, cardiac death | 1.02b (1.01–1.03) per 1 mg/dL increase |
| Kugiyama K et al. (71) | 1999 | 147 | Patients with Angiographic CAD | Fasting | Immunoseparation remnant TRL-C | Composite: MI, recurrent, or refractory angina requiring coronary revascularization, cardiac death | 6.38b (2.3–17.6) highest vs lowest tertile |
| McNamara JR et al. (72) | 2001 | 1567 | Framingham Heart Study, Women | Fasting | Immunoseparation remnant TRL-C | Composite: angina, MI, TS, TIA | 2.27b (1.37–3.77) highest quartile vs all others |
| Fukushima H et al. (73) | 2004 | 120 | Patients with Angiographic CAD and Diabetes | Fasting | Immunoseparation remnant TRL-C | Composite: MI, refractory angina requiring coronary revascularization, cardiac death | 2.2b (1.2–6.4) highest quartile vs all others |
| Imke C et al. (74) | 2005 | 1156 | Japanese-American Men | Fasting | Immunoseparation remnant TRL-C | Composite: MI, revascularization, sudden death | Normal TG: 1.009 (0.939–1.085) High TG: 1.011 (1.001–1.021) per mg/dl |
| Mora S et al. (75) | 2009 | 27 673 | Women’s Health Study | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, IS, coronary revascularization, CVD death | Small VLDL: 1.56 (1.27–1.91) Medium VLDL: 1.46 (1.17–1.82) Large VLDL: 1.77 (1.34–2.33) Total VLDL: 1.71 (1.38–2.12) highest vs lowest quintile |
| Varbo A et al. (76) | 2013 | 73 513 | Copenhagen General Population Study, Copenhagen City Heart Study, and Copenhagen Ischemic Heart Disease Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | Ischemic heart disease | 2.3 (1.7–3.1) highest vs lowest quintile |
| Joshi PH et al. (77) | 2016 | 4114 | Jackson Heart Study, Black Participants | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.18 (1.00–1.39) per SD increase |
| Joshi PH et al. (77) | 2016 | 818 | Framingham Offspring Cohort Study | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.46 (1.05–2.04) per SD increase |
| Lawler P et al. (78) | 2017 | 11 984 | Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, stroke, hospitalization for unstable angina, arterial revascularization, and CVD death | Small VLDL: 1.16 (0.99–1.36) Medium VLDL: 1.30 (1.10–1.53) Large VLDL: 1.17 (1.00–1.37) Total VLDL: 1.21 (1.04–1.41) per SD increase |
| Varbo A et al. (67) | 2018 | 106 216 | Copenhagen General Population Study, Stratified by Weight | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | MI | Normal Weight: 2.0 (1.3–3.2) Overweight: 1.9 (1.4–2.6) Obese: 2.3 (1.4–3.5) highest vs lowest quartile |
| Aday A et al. (79) | 2018 | 27 888 | Women’s Health Study | Nonfasting | NMR very small, small, medium, large, very large, and total VLDL particle concentration | PAD | Very Small VLDL: 1.01 (0.61–1.67) Small VLDL: 0.99 (0.62–1.59) Medium VLDL: 1.98 (1.15–3.41) Large VLDL: 1.58 (0.94–2.68) Very Large VLDL: 1.68 (1.05–2.55) Total VLDL: 1.39 (0.81–2.38) highest vs lowest tertile |
| Holmes M et al. (80) | 2018 | 4662 (Nested Case-Control) | Chinese Kadoorie Biobank (N = 512 891) | Nonfasting | NMR metabolomics remnant cholesterol (VLDL-C plus IDL-C) | MI, IS individually | 1.27 (1.15–1.39) for MI 1.20 (1.09–1.32) for IS per SD increase |
| Saeed A et al. (69) | 2018 | 9334 | Atherosclerosis Risk in Communities Study | Fasting | Direct homogenous assay for TRL-C, first generation | CHD, IS, CVD individually | b1.06 (0.88–1.27) for CHD b1.07 (0.78–1.45) for IS b1.05 (0.89–1.23) for CVD highest vs lowest quartile |
| Vallejo-Vaz et al. (81) | 2018 | 9993 | Treating to New Targets | Fasting | Calculated as non-HDL-C minus LDL-C | Composite: MI, total stroke, resuscitated cardiac arrest, and CVD death | Atorvastatin 10 mg: 1.48 (1.15–1.92) Atorvastatin 80 mg: 1.01 (0.76–1.34) highest vs lowest quintile |
| Varbo A et al. (82) | 2019 | 102 964 | Copenhagen General Population Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | IS | 1.99 (1.49–2.67) highest vs lowest quartile |
| Duran EK et al. (68) | 2020 | 976 (Case-Cohort) | Women’s Health Study (N = 27 552) | Nonfasting | Direct homogenous assay for TRL-C, second generation | Composite: MI, IS, PAD, and CVD death | 1.97 (95% CI: 1.26–3.08) highest vs lowest quartile |
| Author | Year published | Sample size | Study setting | Fasting status | Measurement | Outcome | Adjustedarisk association HR or OR (95% CI) |
|---|---|---|---|---|---|---|---|
| Phillips NR et al. (70) | 1993 | 335 | Montreal Heart Study | Fasting | Ultracentrifugation: IDL-C+VLDL-C | Composite: MI, coronary revascularization, cardiac death | 1.02b (1.01–1.03) per 1 mg/dL increase |
| Kugiyama K et al. (71) | 1999 | 147 | Patients with Angiographic CAD | Fasting | Immunoseparation remnant TRL-C | Composite: MI, recurrent, or refractory angina requiring coronary revascularization, cardiac death | 6.38b (2.3–17.6) highest vs lowest tertile |
| McNamara JR et al. (72) | 2001 | 1567 | Framingham Heart Study, Women | Fasting | Immunoseparation remnant TRL-C | Composite: angina, MI, TS, TIA | 2.27b (1.37–3.77) highest quartile vs all others |
| Fukushima H et al. (73) | 2004 | 120 | Patients with Angiographic CAD and Diabetes | Fasting | Immunoseparation remnant TRL-C | Composite: MI, refractory angina requiring coronary revascularization, cardiac death | 2.2b (1.2–6.4) highest quartile vs all others |
| Imke C et al. (74) | 2005 | 1156 | Japanese-American Men | Fasting | Immunoseparation remnant TRL-C | Composite: MI, revascularization, sudden death | Normal TG: 1.009 (0.939–1.085) High TG: 1.011 (1.001–1.021) per mg/dl |
| Mora S et al. (75) | 2009 | 27 673 | Women’s Health Study | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, IS, coronary revascularization, CVD death | Small VLDL: 1.56 (1.27–1.91) Medium VLDL: 1.46 (1.17–1.82) Large VLDL: 1.77 (1.34–2.33) Total VLDL: 1.71 (1.38–2.12) highest vs lowest quintile |
| Varbo A et al. (76) | 2013 | 73 513 | Copenhagen General Population Study, Copenhagen City Heart Study, and Copenhagen Ischemic Heart Disease Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | Ischemic heart disease | 2.3 (1.7–3.1) highest vs lowest quintile |
| Joshi PH et al. (77) | 2016 | 4114 | Jackson Heart Study, Black Participants | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.18 (1.00–1.39) per SD increase |
| Joshi PH et al. (77) | 2016 | 818 | Framingham Offspring Cohort Study | Fasting | VAP: VLDL3 plus IDL-C | Composite: MI, CHD death, revascularization | 1.46 (1.05–2.04) per SD increase |
| Lawler P et al. (78) | 2017 | 11 984 | Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) | Nonfasting | NMR small, medium, large, and total VLDL particle concentration | Composite: MI, stroke, hospitalization for unstable angina, arterial revascularization, and CVD death | Small VLDL: 1.16 (0.99–1.36) Medium VLDL: 1.30 (1.10–1.53) Large VLDL: 1.17 (1.00–1.37) Total VLDL: 1.21 (1.04–1.41) per SD increase |
| Varbo A et al. (67) | 2018 | 106 216 | Copenhagen General Population Study, Stratified by Weight | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | MI | Normal Weight: 2.0 (1.3–3.2) Overweight: 1.9 (1.4–2.6) Obese: 2.3 (1.4–3.5) highest vs lowest quartile |
| Aday A et al. (79) | 2018 | 27 888 | Women’s Health Study | Nonfasting | NMR very small, small, medium, large, very large, and total VLDL particle concentration | PAD | Very Small VLDL: 1.01 (0.61–1.67) Small VLDL: 0.99 (0.62–1.59) Medium VLDL: 1.98 (1.15–3.41) Large VLDL: 1.58 (0.94–2.68) Very Large VLDL: 1.68 (1.05–2.55) Total VLDL: 1.39 (0.81–2.38) highest vs lowest tertile |
| Holmes M et al. (80) | 2018 | 4662 (Nested Case-Control) | Chinese Kadoorie Biobank (N = 512 891) | Nonfasting | NMR metabolomics remnant cholesterol (VLDL-C plus IDL-C) | MI, IS individually | 1.27 (1.15–1.39) for MI 1.20 (1.09–1.32) for IS per SD increase |
| Saeed A et al. (69) | 2018 | 9334 | Atherosclerosis Risk in Communities Study | Fasting | Direct homogenous assay for TRL-C, first generation | CHD, IS, CVD individually | b1.06 (0.88–1.27) for CHD b1.07 (0.78–1.45) for IS b1.05 (0.89–1.23) for CVD highest vs lowest quartile |
| Vallejo-Vaz et al. (81) | 2018 | 9993 | Treating to New Targets | Fasting | Calculated as non-HDL-C minus LDL-C | Composite: MI, total stroke, resuscitated cardiac arrest, and CVD death | Atorvastatin 10 mg: 1.48 (1.15–1.92) Atorvastatin 80 mg: 1.01 (0.76–1.34) highest vs lowest quintile |
| Varbo A et al. (82) | 2019 | 102 964 | Copenhagen General Population Study | Nonfasting | Calculated as TC minus HDL-C minus LDL-C | IS | 1.99 (1.49–2.67) highest vs lowest quartile |
| Duran EK et al. (68) | 2020 | 976 (Case-Cohort) | Women’s Health Study (N = 27 552) | Nonfasting | Direct homogenous assay for TRL-C, second generation | Composite: MI, IS, PAD, and CVD death | 1.97 (95% CI: 1.26–3.08) highest vs lowest quartile |
Adjusted for clinical risk factors.
Some studies present only models additionally adjusted for TG and/or HDL-C.
Abbreviations: HR, hazards ratio; OR, odds ratio; CAD, coronary artery disease; MI, myocardial infarction; TIA, transient ischemic attack; TS, total stroke; IS, ischemic stroke; ICH, intracranial hemorrhage; PAD, peripheral artery disease; CHD, coronary heart disease; CVD, cardiovascular disease; TC, total cholesterol; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; VLDL, very low-density-lipoprotein cholesterol; IDL-C, intermediate-density lipoprotein cholesterol; TRL, triglyceride-rich lipoproteins; VAP, vertical auto profile; NMR, nuclear magnetic resonance.
Evidence from Genetic Studies
Inferring causation from observational data is often problematic because it is not always known which of 2 variables is the cause, which is the effect, and whether both are common effects of a third unobserved variable. Genetic studies, in particular the Mendelian randomization (MR) approach, have been applied to address this uncertainty. Several MR analyses have now been published and support a causal role of TRLs in atherosclerosis. In the MR method, a single (or multiple) genetic variant(s) associated with a particular biomarker is used as a proxy for the biomarker. Outcomes are compared between the group having the affected allele(s) and groups with a reference allele(s). This approach may be considered to approximate a randomized clinical trial (RCT) in that genes are allocated randomly at birth, presumably independent of a range of potential confounding factors such as socioeconomic status and lifestyle choices. In our view, this assumption is flawed as awareness of reporter biomarker levels, especially lipid levels that are commonly measured, is analogous to unblinding in a clinical trial and may influence subsequent ‘postrandomization’ interventions, such as lipid lowering therapy or behavioral modification. These factors are not commonly modeled and may significantly alter genetic risk associations.
Another issue is genetic pleiotropy, whereby a single genetic variant may influence several traits. Almost all genetic variations in TG metabolism identified in human genetic studies also influence at least one other lipid trait, usually HDL-C (83). Three notable MR studies from Varbo et al. (76), Do et al. (84), and Holmes et al. (85) attempted to overcome this limitation by isolating variants that have limited pleiotropy, i.e., genetic effects on a single lipoprotein without effects on other lipoproteins. Varbo (76) et al. grouped genetic variants according to specificity for remnant TRL-C, specificity for both remnant TRL-C and HDL-C combined, specificity for HDL-C alone, or specificity for LDL-C alone. This study demonstrated that genetically increased remnant TRL-C was independently linked to coronary heart disease (CHD) with a roughly 3-fold excess risk per 39 mg/dL (1 mmol/L) increase in nonfasting TRL-C, while low HDL cholesterol was not. Do et al. (84) used multivariable MR to separate TG-associated effects on CHD risk from other lipid determinants. This analysis evaluated 185 common single nucleotide polymorphisms (SNPs) representing independent genetic loci that associate with at least 1 lipid trait. Associations with CHD were adjusted for SNP associations with other lipids. Their findings reaffirmed an isolated LDL-C genetic effect and demonstrated an isolated TG genetic effect, which was similar in magnitude to LDL-C after adjustment for genetic pleiotropy. An isolated genetic effect of HDL-C was not detected. Holmes et al. (85) used weighted risk scores of 135 SNPS associated with TG, HDL-C, and LDL-C. To account for genetic pleiotropy, 3 approaches were used: unrestricted risk scores (including all SNPs associated with the target lipid trait, regardless of association with other lipid traits), restricted risk scores (limited to a smaller number of SNPs associated only with the target lipid target), and statistical adjustment for plasma concentrations of nontarget lipids and statin use. The investigators found that genetically increased TG concentrations using both restricted and unrestricted risk scores were associated with CHD, but this risk was attenuated and the effect was null after adjustment for HDL-C. Given that 2 of 3 approaches provided evidence of causality, the authors concluded that TGs are likely causally related to coronary artery disease.
In another pivotal study, Ference and colleagues (86) utilized the large UK Biobank database (>650 000 participants; 91 129 with CHD). Using MR, they compared the impact of 2 genetic risk scores, 1 based on all known variants in the LPL gene associated with TG concentrations and 1 based on all known variants in the LDL-R gene associated with LDL-C concentrations. The exposure alleles were classified as those variants associated with lower TG or lower LDL, respectively, such that a higher score was expected to correlate with a lower risk of CHD. The scores were then standardized to each other by unit effects on a 10 mg/dL reduction in apoB-containing lipoproteins. For each 10-mg/dL lower plasma apoB concentration associated with variants in the LPL score, there was a corresponding 69.9-mg/dL lower TG concentration, with no change in LDL-C and a lower risk of CHD [odds ratio (OR): 0.771, 95% confidence interval (CI): 0.741–0.802]. Remarkably, for the same 10-mg/dL lower plasma apoB concentration associated with variants in the LDL-R score, there is a corresponding 14.1-mg/dL lower LDL-C concentration with no change in triglycerides, and a similar lower risk of CHD (OR: 0.773, 95% CI: 0.747–0.801). These findings indicate that the clinical benefit of lowering TGs is similar to the clinical benefit of lowering LDL-C per unit change in apoB and, further, that all apoB-containing lipoproteins, including TRLs and their remnants as well as LDL particles, have approximately the same per-particle effect on coronary risk.
Evidence from Fibrate and Omega-3 Fatty Acid Clinical Trials
Fibrates lower triglyceride concentrations by increasing VLDL clearance, principally in the VLDL1 density range, and by reducing postprandial lipemia (15, 87–89). Effects are likely mediated via their action on the peroxisome proliferator-activated receptor-α (PPAR-α) agonist response element within the apoCIII gene promoter region resulting in repressed transcription and, as a consequence, enhanced LPL activity (15). However, currently available fibrates are weak PPAR-α agonists with clinical utility limited by dose-dependent side effects and a lackluster performance in cardiovascular outcomes trials.
Five large RCTs have evaluated the effects of fibrates on cardiovascular risk (Table 2). Although early trials suggested benefit of fibrate monotherapy, the more recent FIELD (96) and ACCORD (98) studies showed no benefit of fenofibrate on cardiovascular outcomes in the setting of background statin therapy (unplanned drop-in in FIELD and by design in ACCORD). Importantly, none of these trials enrolled participants on the basis of hypertriglyceridemia and in each trial, post hoc subgroup analyses have suggested marked clinical benefit in this patient population (Table 2). For example, in meta-analyses evaluating subgroup effects (92, 99), consistently greater benefit was found in patients with high TG concentrations or mixed dyslipidemia (increased TG and low HDL-C). In these subgroups, fibrates appear to reduce cardiovascular risk by 28% (95% CI, 15–39%; P < 0.001) or 30% (95% CI, 19–40%, P < 0.0001), respectively, but only by 6% (95% CI, −2–13%, P = 0.13) in subjects without these lipid abnormalities. The ongoing PROMINENT study (100) is a placebo-controlled cardiovascular outcomes trial of pemafibrate, a potent and selective PPAR-α (SPPARM-α) agonist, in patients with diabetes and mixed dyslipidemia [TG: 200–499 mg/dL (2.26–5.64 mmol/L) and HDL-C: ≤40 mg/dL (1.03 mmol/L)]. The trial has fully recruited (n = 10 514) and is expected to report by early 2023. Importantly, Phase II trial data in European populations on background statin therapy, have shown that pemafibrate reduces TGs, calculated TRL-C, and measured TRL-C each by roughly 40% at 12 weeks using doses currently being studied in PROMINENT (unpublished data: Duran et al., presented at the European Atherosclerosis Society Congress 2020).
Effects of fibrates on cardiovascular events in large randomized controlled trials.
| Trial | Drug | Patient characteristics | CV outcomea | Trial duration (years) | RR reduction entire cohort | Atherogenic dyslipidemia subgroup | RR reduction subgroupb |
|---|---|---|---|---|---|---|---|
| HHS (90, 91) | Gemfibrozil | Non-HDL-C > 5.2 mmol/L No CHD Men | Nonfatal MI and CHD death | 5.0 | −34% (P < 0.02) | TG >204 mg/dL LDL-C/HDL-C ratio >5.0 | −71% (P = 0.005) |
| VA-HIT (92–94) | Gemfibrozil | HDL-C < 1.0 mmol/l CHD Men | Nonfatal MI and CHD death | 5.1 | −22% (P = 0.006) | TG >180 mg/dL <40 mg/dL | −30% (P < 0.05) |
| BIP (95) | Bezafibrate | Previous MI or angina Men and women | Fatal/nonfatal MI and sudden death | 6.2 | −7% (P = 0.26) | TG ≥200 mg/dL | −40% (P = 0.02) |
| FIELD (96, 97) | Fenofibrate | Type 2 diabetes Some patients receiving statins Men and women | MI, stroke, CVD death, coronary, or carotid revascularizationa | 5.0 | −11 (P = 0.035) | TG ≥204 mg/dL HDL-C < 40 mg/dL (men) or <50 mg/dL (women) | −27% (P = 0.005) |
| ACCORD (92, 98) | Fenofibrate | Type 2 diabetes CVD or >2 CVD risk factors Patients receiving simvastatin Men and women | Nonfatal MI, nonfatal stroke, and CVD death | 4.7 | −8% (P = 0.32) | TG ≥204 mg/dL HDL-C ≤ 34 mg/dL | −29% (P < 0.05) |
| Trial | Drug | Patient characteristics | CV outcomea | Trial duration (years) | RR reduction entire cohort | Atherogenic dyslipidemia subgroup | RR reduction subgroupb |
|---|---|---|---|---|---|---|---|
| HHS (90, 91) | Gemfibrozil | Non-HDL-C > 5.2 mmol/L No CHD Men | Nonfatal MI and CHD death | 5.0 | −34% (P < 0.02) | TG >204 mg/dL LDL-C/HDL-C ratio >5.0 | −71% (P = 0.005) |
| VA-HIT (92–94) | Gemfibrozil | HDL-C < 1.0 mmol/l CHD Men | Nonfatal MI and CHD death | 5.1 | −22% (P = 0.006) | TG >180 mg/dL <40 mg/dL | −30% (P < 0.05) |
| BIP (95) | Bezafibrate | Previous MI or angina Men and women | Fatal/nonfatal MI and sudden death | 6.2 | −7% (P = 0.26) | TG ≥200 mg/dL | −40% (P = 0.02) |
| FIELD (96, 97) | Fenofibrate | Type 2 diabetes Some patients receiving statins Men and women | MI, stroke, CVD death, coronary, or carotid revascularizationa | 5.0 | −11 (P = 0.035) | TG ≥204 mg/dL HDL-C < 40 mg/dL (men) or <50 mg/dL (women) | −27% (P = 0.005) |
| ACCORD (92, 98) | Fenofibrate | Type 2 diabetes CVD or >2 CVD risk factors Patients receiving simvastatin Men and women | Nonfatal MI, nonfatal stroke, and CVD death | 4.7 | −8% (P = 0.32) | TG ≥204 mg/dL HDL-C ≤ 34 mg/dL | −29% (P < 0.05) |
HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TG, triglycerides; T2D, Type 2 diabetes.
To convert TG from mg/dL to mmol/L, multiply by 0.0113; to convert HDL-C from mg/dL to mmol/L, multiply by 0.0259.
The CV outcome presented is the prespecified primary endpoint in all trials except FIELD. In this trial, the primary endpoint (CHD) was not reported in subgroups and the data are shown for the secondary endpoint of total CVD.
Risk reductions and P values for subgroups when not presented in publications by trial investigators were taken from the meta-analysis of Bruckert et al. (92).
Effects of fibrates on cardiovascular events in large randomized controlled trials.
| Trial | Drug | Patient characteristics | CV outcomea | Trial duration (years) | RR reduction entire cohort | Atherogenic dyslipidemia subgroup | RR reduction subgroupb |
|---|---|---|---|---|---|---|---|
| HHS (90, 91) | Gemfibrozil | Non-HDL-C > 5.2 mmol/L No CHD Men | Nonfatal MI and CHD death | 5.0 | −34% (P < 0.02) | TG >204 mg/dL LDL-C/HDL-C ratio >5.0 | −71% (P = 0.005) |
| VA-HIT (92–94) | Gemfibrozil | HDL-C < 1.0 mmol/l CHD Men | Nonfatal MI and CHD death | 5.1 | −22% (P = 0.006) | TG >180 mg/dL <40 mg/dL | −30% (P < 0.05) |
| BIP (95) | Bezafibrate | Previous MI or angina Men and women | Fatal/nonfatal MI and sudden death | 6.2 | −7% (P = 0.26) | TG ≥200 mg/dL | −40% (P = 0.02) |
| FIELD (96, 97) | Fenofibrate | Type 2 diabetes Some patients receiving statins Men and women | MI, stroke, CVD death, coronary, or carotid revascularizationa | 5.0 | −11 (P = 0.035) | TG ≥204 mg/dL HDL-C < 40 mg/dL (men) or <50 mg/dL (women) | −27% (P = 0.005) |
| ACCORD (92, 98) | Fenofibrate | Type 2 diabetes CVD or >2 CVD risk factors Patients receiving simvastatin Men and women | Nonfatal MI, nonfatal stroke, and CVD death | 4.7 | −8% (P = 0.32) | TG ≥204 mg/dL HDL-C ≤ 34 mg/dL | −29% (P < 0.05) |
| Trial | Drug | Patient characteristics | CV outcomea | Trial duration (years) | RR reduction entire cohort | Atherogenic dyslipidemia subgroup | RR reduction subgroupb |
|---|---|---|---|---|---|---|---|
| HHS (90, 91) | Gemfibrozil | Non-HDL-C > 5.2 mmol/L No CHD Men | Nonfatal MI and CHD death | 5.0 | −34% (P < 0.02) | TG >204 mg/dL LDL-C/HDL-C ratio >5.0 | −71% (P = 0.005) |
| VA-HIT (92–94) | Gemfibrozil | HDL-C < 1.0 mmol/l CHD Men | Nonfatal MI and CHD death | 5.1 | −22% (P = 0.006) | TG >180 mg/dL <40 mg/dL | −30% (P < 0.05) |
| BIP (95) | Bezafibrate | Previous MI or angina Men and women | Fatal/nonfatal MI and sudden death | 6.2 | −7% (P = 0.26) | TG ≥200 mg/dL | −40% (P = 0.02) |
| FIELD (96, 97) | Fenofibrate | Type 2 diabetes Some patients receiving statins Men and women | MI, stroke, CVD death, coronary, or carotid revascularizationa | 5.0 | −11 (P = 0.035) | TG ≥204 mg/dL HDL-C < 40 mg/dL (men) or <50 mg/dL (women) | −27% (P = 0.005) |
| ACCORD (92, 98) | Fenofibrate | Type 2 diabetes CVD or >2 CVD risk factors Patients receiving simvastatin Men and women | Nonfatal MI, nonfatal stroke, and CVD death | 4.7 | −8% (P = 0.32) | TG ≥204 mg/dL HDL-C ≤ 34 mg/dL | −29% (P < 0.05) |
HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; TG, triglycerides; T2D, Type 2 diabetes.
To convert TG from mg/dL to mmol/L, multiply by 0.0113; to convert HDL-C from mg/dL to mmol/L, multiply by 0.0259.
The CV outcome presented is the prespecified primary endpoint in all trials except FIELD. In this trial, the primary endpoint (CHD) was not reported in subgroups and the data are shown for the secondary endpoint of total CVD.
Risk reductions and P values for subgroups when not presented in publications by trial investigators were taken from the meta-analysis of Bruckert et al. (92).
Omega-3 fatty acids [eicosapentaenoic (EPA) and docosahexanoic (DHA) acids and their derivatives] given in high doses (>2 g/d) lower VLDL proportionate to dose and pretreatment TG concentration. In moderate hypertriglyceridemia, these agents reduce TGs by ∼20%. The landmark REDUCE-IT trial (101) showed a 25% risk reduction in major adverse cardiovascular events among patients with moderate hypertriglyceridemia assigned to icosapent ethyl at 4 g/d. Some suggest that this benefit is overestimated due to an increase in atherogenic lipoproteins and inflammatory markers in the comparison group that received a mineral oil placebo. Nonetheless, the large risk reductions cannot be easily or solely explained by placebo adjusted changes in these biomarkers and the clinical safety of employing mineral oil placebos has been recently reviewed (102). The notable finding that achieved TG concentrations in REDUCE-IT did not match achieved risk reduction was surprising but not unique for TG-lowering trials. Risk reductions in neither the HHS (103) nor VA-HIT (104) were correlated with achieved TG concentrations despite significant TG reduction in both of these fibrate trials. In HHS, the change in fasting TG with gemfibrozil therapy was not associated with the trial endpoint in either the total cohort or the subgroup with increased baseline concentrations [TG ≥ 177 mg/dL (2 mmol/L)] (103). Neither was on treatment fasting TG a significant predictor of the primary endpoint in VA-HIT [RR per 50 mg/dL 1.05 (0.98–0.97); P = 0.16] (104). As discussed in this review, risk reduction may better correlate with TRL-C concentrations; however, these data have not been available from these trials.
Conclusions and Future Directions
As discussed in this review, the scientific landscape with regard to the involvement of triglyceride-rich remnants in development of cardiovascular disease is rapidly changing. The lipoprotein system is comprised of a continuum of particles of differing size and composition that begin as chylomicrons and VLDL with catabolism to LDL. Hypertriglyceridemia is associated with aberrant TRL modification yielding remnants that are increasingly recognized as proatherogenic. The data now lead many to believe that TG-rich remnant particles by virtue of their excess cholesterol content, apolipoprotein composition, and structural properties are causal factors and along with their metabolic pathways are now potential targets for therapeutic modulation. Several novel agents (Table 3) are being tested in this regard and hold substantial promise to address residual cardiovascular risk that remains an important clinical problem. Continued scientific work must better inform our knowledge gaps, reliable biomarkers must be developed for use in clinical practice, and clinical trials must demonstrate treatment benefits to ultimately determine whether this line of investigation translates into new and effective paradigms for lipid management.
Emerging therapies for treatment of hypertriglyceridemia.
| Primary mechanism of TRL lowering | Therapeutic class | Examplesa |
|---|---|---|
| Suppression of TRL production | Selective PPAR modulators | Pemafibrate (K-877) Elafibranor (GFT505) |
| MTTP inhibitors | JuxtapidTM (lomitapide) Slx-4090 | |
| DGAT1 inhibitor | Pradigastat | |
| LPL activation | LPL activator | Ibrolipim |
| ApoC3 antisense oligonucleotide | Volanesorsen (ISIS 304801) | |
| ANGPTL3 inhibitor | Evinacumab (REGN1500) IONIS-ANGPTL3-LRx | |
| Complex effects | Prescription omega-3 fatty acid preparations | VascepaTM (EPA) EpanovaTM (EPA plus DHA) LovazaTM (EPA plus DHA) OmtrgTM (EPA plus DHA) Icosabutate (EPA) |
| Primary mechanism of TRL lowering | Therapeutic class | Examplesa |
|---|---|---|
| Suppression of TRL production | Selective PPAR modulators | Pemafibrate (K-877) Elafibranor (GFT505) |
| MTTP inhibitors | JuxtapidTM (lomitapide) Slx-4090 | |
| DGAT1 inhibitor | Pradigastat | |
| LPL activation | LPL activator | Ibrolipim |
| ApoC3 antisense oligonucleotide | Volanesorsen (ISIS 304801) | |
| ANGPTL3 inhibitor | Evinacumab (REGN1500) IONIS-ANGPTL3-LRx | |
| Complex effects | Prescription omega-3 fatty acid preparations | VascepaTM (EPA) EpanovaTM (EPA plus DHA) LovazaTM (EPA plus DHA) OmtrgTM (EPA plus DHA) Icosabutate (EPA) |
PPAR, peroxisome proliferator activator receptor; MTTP, microsomal triglyceride transfer protein; DGAT1, diacylglycerol acyltransferase 1; LPL, lipoprotein lipase; ApoC3, apolipoprotein C3; ANGPTL3, angiopoietin-like protein-3; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid.
Some already approved for clinical use in familial hypercholesterolemia or severe hypertriglyceridemia.
Emerging therapies for treatment of hypertriglyceridemia.
| Primary mechanism of TRL lowering | Therapeutic class | Examplesa |
|---|---|---|
| Suppression of TRL production | Selective PPAR modulators | Pemafibrate (K-877) Elafibranor (GFT505) |
| MTTP inhibitors | JuxtapidTM (lomitapide) Slx-4090 | |
| DGAT1 inhibitor | Pradigastat | |
| LPL activation | LPL activator | Ibrolipim |
| ApoC3 antisense oligonucleotide | Volanesorsen (ISIS 304801) | |
| ANGPTL3 inhibitor | Evinacumab (REGN1500) IONIS-ANGPTL3-LRx | |
| Complex effects | Prescription omega-3 fatty acid preparations | VascepaTM (EPA) EpanovaTM (EPA plus DHA) LovazaTM (EPA plus DHA) OmtrgTM (EPA plus DHA) Icosabutate (EPA) |
| Primary mechanism of TRL lowering | Therapeutic class | Examplesa |
|---|---|---|
| Suppression of TRL production | Selective PPAR modulators | Pemafibrate (K-877) Elafibranor (GFT505) |
| MTTP inhibitors | JuxtapidTM (lomitapide) Slx-4090 | |
| DGAT1 inhibitor | Pradigastat | |
| LPL activation | LPL activator | Ibrolipim |
| ApoC3 antisense oligonucleotide | Volanesorsen (ISIS 304801) | |
| ANGPTL3 inhibitor | Evinacumab (REGN1500) IONIS-ANGPTL3-LRx | |
| Complex effects | Prescription omega-3 fatty acid preparations | VascepaTM (EPA) EpanovaTM (EPA plus DHA) LovazaTM (EPA plus DHA) OmtrgTM (EPA plus DHA) Icosabutate (EPA) |
PPAR, peroxisome proliferator activator receptor; MTTP, microsomal triglyceride transfer protein; DGAT1, diacylglycerol acyltransferase 1; LPL, lipoprotein lipase; ApoC3, apolipoprotein C3; ANGPTL3, angiopoietin-like protein-3; EPA, eicosapentaenoic acid; DHA, docosahexaenoic acid.
Some already approved for clinical use in familial hypercholesterolemia or severe hypertriglyceridemia.
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: None declared.
Stock Ownership: None declared.
Honoraria: A.D. Pradhan, Medintelligence, Metabolic Endocrine Education Foundation, North American Center for Continuing Medical Education, Physician's Education Resource LLC.
Research Funding: E.K. Duran, the National Institutes of Health under Award Number 5T32HL007575-33; A.D. Pradhan, investigator-initiated research support from Kowa Research Institute, Kowa Pharmaceuticals Europe Co. Ltd, Denka Seiken, co-Principal Investigator of the PROMINENT trial (NCT03071692).
Expert Testimony: None declared.
Patents:None declared.
