Low-Density Lipoprotein Size and Cardiovascular Disease: A Reappraisal

The major lipoprotein types, very low-density lipoprotein (VLDL), low-density lipoprotein (LDL), and high-density lipoprotein (HDL), are composed of many subgroups (1–3). Classifications are based on size (1), density, or apolipoprotein (apo) content (2), or a combination of these (3), and the subfractions that are isolated have distinct metabolic and other functional properties. Thus, it is entirely reasonable to think that subfractions of the major lipoproteins have diverse relationships to coronary heart disease (CHD). Because the classical lipid risk factors by no means perfectly predict CHD in patients, lipoprotein subfractionation has the potential to improve risk prediction. Among the various lipoprotein subfractionation systems, LDL size and density by far have attracted the most basic, clinical, and population research. Several companies offer LDL size measurements to physicians as a diagnostic test for CHD risk. Several lines of evidence implicate small, dense LDL in the etiology of CHD. Small, dense LDL is often increased in relative proportion or concentration in patients with CHD. Small, dense LDL has several characteristics that are linked to atherogenesis: long residence time in plasma, and enhanced oxidizability, arterial proteoglycan binding, and permeability through the endothelial barrier (4). Together, these findings have led to the hypothesis that small, dense LDL is a potent atherogenic lipoprotein that can be used to improve risk prediction, and evaluate response to lipid therapy (5–7). Furthermore, small, dense LDL is often part of a group of high-risk characteristics including high triglycerides, low HDL, diabetes, insulin resistance, obesity, and the metabolic syndrome (8–10). This has led logically to the concept that it contributes to the high rate of CHD in these groups. Yet, the very association between small, dense LDL and these other high-risk conditions challenges proponents of the hypothesis to show a direct, independent relationship between small, dense LDL and CHD. In this article, we evaluate evidence on LDL subclasses in risk assessment and therapy, reaching the conclusion that LDL subclass measurement does not add independent information to that conferred generically by the LDL concentration along with the other standard risk factors. LDL are spherical particles, 22–29 nm in diameter, composed of a core of esterified cholesterol and triglyceride, a surface lipid coat of unesterified cholesterol and phospholipid, an essential structural protein, apo B, and sometimes small apos, such as apo CIII and apo E that modulate LDL metabolism. Each LDL particle has one apo B molecule, which is recognized by LDL receptors that clear LDL from plasma. Thus, the LDL apo B concentration is the plasma concentration of LDL particles. The size of an LDL particle depends on how much lipid is in the core, and the lipid content naturally determines its density. Thus, smaller LDL is denser, larger LDL is lighter, and the two qualities are largely equivalent. Early studies using analytical ultracentrifugation revealed that distinct LDL subpopulations are present in each individual (11). Seven distinct LDL subpopulations were resolved by density-gradient ultracentrifugation and polyacrylamide-gradient gel electrophoresis (1, 12, 13). All people have LDL that is in a range of sizes that correspond to specific densities (8, 14). The size of the predominant LDL particles determines the classification: 22–25.5 nm is small, 25.6–26.5 nm intermediate, and 26.6–28.5 large (13, 14). Using mathematical modeling techniques that separated LDL size tracings on gradient-gel electrophoresis into Gaussian curves, Austin et al. (15) identified two subclass patterns: the classical category, Pattern A, is more than 25.5 nm, and Pattern B is 25.5 nm or less. Several metabolic pathways are involved in forming small dense LDL, as recently reviewed in detail by Berneis and Krauss (6). Small LDL may be formed by metabolic channeling of large VLDL (16), lipolysis of intermediate-density lipoprotein, and large LDL by hepatic lipase (17, 18), remodeling of LDL by cholesterol ester transfer protein (CETP) (19–21), secretion into plasma by the liver (22, 23), or a combination of these processes. Small, dense LDL has a low content of esterified and unesterified cholesterol, and phospholipid; the triglyceride content is either similar to (24) or greater than (25) large LDL. This reduced cholesterol content may be an effect of CETP (19, 20). CETP exchanges cholesterol ester and triglyceride among HDL, LDL, and VLDL. During this process, VLDL serves as an acceptor of cholesterol ester from HDL and LDL. LDL and HDL become enriched in triglyceride and depleted of cholesterol ester. Hydrolysis of the triglyceride in LDL by lipases further reduces the size of LDL (21, 26). LDL can be an acceptor as well as a donor of cholesterol ester, but in hypertriglyceridemia, CETP preferentially uses VLDL rather than LDL as an acceptor (20). This process may contribute to the relationship between hypertriglyceridemia and small LDL. Another proposed mechanism for the formation of small LDL is increased hepatic lipase that progressively lipolyzes intermediate-density lipoprotein to large and finally small LDL (17, 18). Increased hepatic lipase also lowers plasma HDL levels, perhaps explaining the relationship between small LDL particles and low HDL concentration. Analytical ultracentrifugation is the original gold standard to which subsequent methods have been calibrated and validated (27). Analytical ultracentrifugation measures the flotation velocity of LDL in a gravitational field; the faster the flotation velocity, the more lipid rich the LDL. This method is currently available only in a few research laboratories worldwide. Preparative ultracentrifugation separates discrete LDL subfractions that can be quantified and studied for chemical composition and function. The classic method separates seven LDL density fractions (1), and although it is time consuming, any laboratory that has an ultracentrifuge can do it. Preparative ultracentrifugation is the method that defines the LDL subclasses, and it is the de facto gold standard in the field. Nonequilibrium density gradient ultracentrifugation uses the same principles as analytical ultracentrifugation, and is simpler and less labor intensive. The vertical autoprofiling system is a well-validated example of this technique (28). Gradient gel electrophoresis is a simple, readily available method to determine LDL size (13). It uses a drop of whole plasma or serum, and multiple samples can be processed together (8, 29). It has been extensively validated using ultracentrifugation. Its limitation is that, in contrast to ultracentrifugation, it does not quantify the concentration of LDL particles of specific sizes, just the size of the predominant LDL species or the average size of LDL. Thus, changes in the predominant or average LDL size do not necessarily indicate changes in concentration of a particular species of LDL. For example, a selective reduction in large LDL concentration with no change in small LDL concentration would reduce the average size of LDL; this is often misinterpreted to mean an increase in the plasma concentration of small LDL particles. Nuclear magnetic resonance (NMR) measures the diameter and lipid concentration of LDL (30). Diameter is determined by a signal from the phospholipid surface coat of LDL, and concentration by the number of methyl groups on the cholesterol ester and triglyceride molecules within each of the four LDL subfractions resolved. Average LDL size is then calculated by the weighted average of the LDL subfractions. NMR computes the concentration of LDL subfractions using typical lipid contents of LDL subfractions in the published literature. NMR is the most rapid and convenient method for determining LDL size and subfraction concentration. However, it is limited by lack of published data on detailed procedures, calibration, and validation, which are expected when novel methods are established. The assumptions and calibration method that NMR uses to convert lipid signal intensity and size to LDL concentration have not been revealed, nor is it known whether these assumptions hold equally across diverse populations and during diet or drug therapy that affects the composition of LDL. For example, when LDL becomes cholesterol-ester poor and triglyceride rich, as in the generation of small LDL described above, the algorithms used may be incorrect as there is not likely to be a single triglyceride for cholesterol exchange in the final particle. Validation studies in large populations have not been published on LDL subfraction concentration by NMR and ultracentrifugation on the same samples, as have been long available for gradient gel electrophoresis. Mechanistic support for small LDL having a special atherogenicity depends on atherogenic actions being greater for small than for intermediate or large LDL. This case has not been proven, however. Both large and small LDL compared with intermediate size LDL have reduced affinity for the LDL receptor which clears LDL from plasma (31, 32). Decreased clearance of these forms of LDL by the liver and steroidogenic tissues is thought to lead to increased uptake by the arterial wall. In vivo, small LDL has a longer residence time in plasma than large LDL (22, 33). This may be caused by reduced exposure on small LDL of the region of apo B that binds to the LDL receptor, an interaction that is necessary to clear LDL from the circulation. The long residence time in plasma for small LDL could foster atherosclerosis if small LDL entered the arterial intima more readily than other LDL. Although this was found in experiments in rabbits (34), a study of transvascular transport of LDL in vivo in humans did not find a correlation with LDL size (35). This finding suggests that for every unit of time, large LDL is just as likely as small LDL to enter the arterial intima. Because large LDL has more cholesterol ester than small LDL, a large LDL particle would deposit more cholesterol into plaque than small LDL. Small LDL binds to arterial proteoglycan (36) in the arterial wall, but so does large cholesterol-rich LDL (37). Proteoglycan exists on the endothelial cell surface as well as inside the intima. Proteoglycan binding on endothelium may facilitate lipoprotein entry into the vascular intima, and binding to arterial intimal proteoglycan activates or accelerates plaque progression. Thus, it appears that both large and small LDL share undesirable characteristics. Oxidized LDL has atherogenic actions in the vascular wall including stimulation of foam cell formation and activation of inflammation. Circulating oxidized LDL is associated with increased risk of CHD in cross-sectional studies (38, 39). Small, dense LDL is depleted of vitamin E and is more rapidly oxidized in vitro (40), characteristics that could make small LDL more atherogenic. Furthermore, the susceptibility of small LDL to oxidation can be reversed by repletion with vitamin E. However, oxidizability of LDL was not independently associated with carotid intima-media thickness (IMT) in patients with combined hyperlipidemia (41). In a clinical trial in healthy men and women, vitamin E supplements reduced LDL oxidizability but did not reduce the progression of carotid IMT (42). Large trials of vitamin E and other antioxidant supplements definitively showed no reduction in CHD. Thus, a link between LDL oxidizability and human atherosclerosis and CHD remains to be established. Native, unoxidized LDL has direct atherogenic effects, for example to enhance activated monocytes to produce the inflammatory mediators TNF-α and IL-8 (43). Finally, large cholesterol-rich LDL is the predominant type of LDL in familial hypercholesterolemia (44), and it is firmly established that this LDL is responsible for their premature atherosclerosis. Thus, large and small LDL are atherogenic, and it is not possible to judge which if any is more harmful, overall. If LDL size is to be used to quantify risk of CHD, it first must be shown that it adds information on risk or is superior to the standard lipid risk factors used in national guidelines. Many studies found that small dense LDL is increased in concentration or relative proportion in CHD (5–7). However, small LDL size is substantially correlated with high triglycerides and low HDL, and mildly related to obesity and perhaps insulin resistance (4, 8–10). These metabolic connections complicate efforts to determine whether small LDL particles have an especially strong relationship to CHD, beyond their being simply LDL. In contrast, LDL cholesterol concentration has a very low correlation with other lipid risk factors and most nonlipid risk factors, and its establishment as the principal lipid risk factor for prevention and treatment has been straightforward (45). All established CHD risk factors have passed the test of multiple regression analysis that investigates the etiological relationship of potential risk factors to the disease. We have often encountered misunderstanding among colleagues about what multiple regression analysis can accomplish. It is a powerful technique that disentangles related variables, ideal when correlations are mild to moderate (i.e. r ≤ 0.7), but with a large population, it can be effective even when the correlation coefficients are higher. An instructive example is the case of LDL size and plasma triglycerides that are moderately inversely correlated (for example, r = −0.71) in a large prospective study of U.S. male physicians (46). In crude analysis, small LDL size and high triglycerides both were associated with myocardial infarction (MI). Multiple regression analysis tested the effect of LDL size across the concentration range of triglycerides, and the effect of triglycerides across the range of LDL sizes. If small LDL size is truly related to CHD, then it will be so whether triglyceride is high or low. Because the correlation between triglycerides and LDL size, −0.71, explains only 50% of the variance of each, there would be a wide range of LDL sizes among people with high triglyceride concentrations, and a wide range of triglyceride concentrations among people with small LDL to disentangle these two lipid variables. Figure 1 shows that the relative risk of MI is similar across tertiles of LDL size within each triglyceride tertile. In contrast, high triglyceride is strongly related to MI regardless of LDL size. For example, the relative risk is the same, 2.7, for high triglycerides with either small or large LDL size. Thus, LDL size does not give information beyond that given by triglycerides on risk of MI. This study demonstrates that triglycerides but not LDL size independently predicted first MI in U.S. male physicians. As we show subsequently, this is a typical for LDL size in between triglycerides, LDL size, and MI in a prospective study of U.S. physicians. This demonstrates that the association between small LDL size and MI only from the association between small LDL size and high triglyceride concentration. triglyceride concentration is independently related to MI regardless of LDL size. by The first studies that LDL size and CHD a cross-sectional Although this is not the ideal of by the clinical disease the findings from these original studies are with the more prospective studies described LDL size was in patients clinical CHD and in healthy In of the average LDL size was smaller in the than However, with of lipid and other risk factors in multiple regression analysis, the in LDL size in but one studies showed no in LDL size with or analysis and in two studies large LDL was associated with CHD et al. studied LDL size in men CHD with average lipid cholesterol mean triglycerides and no CHD risk factors to determine whether small LDL size could the than risk factors. to the larger LDL than the healthy and this finding in multiple regression Large LDL was associated with CHD in study but data on analysis are not available et al. found that the concentration of triglyceride in both and dense LDL was related to coronary atherosclerosis. Among large LDL is more the CHD is high Large LDL is more in with high of CHD, for example in compared with and in the compared with are limited by many factors that could studies of LDL size and disease convert to by studies of LDL size and disease convert to by studies are more than cross-sectional studies LDL size and other lipid risk factors are the of CHD, so the disease and its treatment do not the In of prospective small LDL predicted coronary or other CHD in However, for the of standard lipid risk factors or apo B, small LDL was not one or more other were independently For example, in the the relative risk for MI for small LDL from to = for lipid and nonlipid risk factors. In the plasma triglyceride, and the to HDL cholesterol were (46). study in found similar LDL size. In this a relative risk of = for small LDL in analysis was reduced to = for the to HDL cholesterol and triglyceride The independent were LDL cholesterol to HDL and studies of LDL size and disease not for in LDL size between and Small LDL size was not and did not enter the gradient gel electrophoresis. convert to by studies of LDL size and disease not for in LDL size between and Small LDL size was not and did not enter the gradient gel electrophoresis. convert to by The found that mean LDL size was nm smaller in than in This was reduced to nm and no longer = for the to HDL cholesterol which a In the the risk to small LDL particles may be independent of the in plasma lipoprotein lipid was by the of analysis that small LDL size was not a of CHD apo B was the only independent lipoprotein no independent association between LDL size and CHD was found among men and in the In men and women, LDL size did not predict CHD in either or the to HDL cholesterol was the independent In small LDL size was a of CHD only in the for plasma triglyceride and HDL cholesterol levels, large not small LDL size CHD along with low HDL This in the with in U.S. and patients a MI. not small LDL, was a strong of CHD and in both and risk = studies atherosclerosis with measurements of carotid or intima-media thickness (IMT) or with coronary cross-sectional studies found that small LDL was independently associated with carotid IMT in healthy men and in patients with familial combined hyperlipidemia and (41). Another study found the that large LDL was associated with carotid IMT in patients with hypercholesterolemia and or cholesterol there was no relationship with IMT studies that LDL size did not predict of coronary atherosclerosis although a high concentration of small LDL was but correlated = with progression In the that is from is that small LDL does not have a special relationship to CHD beyond its to LDL concentration. This could be more studied by or analysis of the individual a conclusion may well studies in with large of We think that it is likely that by triglycerides and other lipid risk factors is most studies that the risk associated with small LDL becomes or factors such as measurement or are not for this LDL size is a and measurement among the plasma and Another potential for LDL is to evaluate response to lipid and drug therapy for hyperlipidemia coronary in patients dense LDL treatment but not in LDL et al. found that in LDL larger and in lipase were correlated with in coronary treatment with and or and LDL increased and hepatic lipase in patients with small LDL and high hepatic lipase This group a in the hepatic lipase However, it is also that patients with the associated with low hepatic lipase LDL and during and this group showed the most of coronary The also found that this hepatic lipase conferred increased risk of CHD These an atherogenic for LDL. In two LDL size nor the plasma concentration of small LDL during or treatment was associated with progression of coronary In contrast, small LDL size during treatment was independently associated with of coronary these do not the of LDL size during lipid treatment in clinical This is in large clinical trials with clinical In this LDL size was not a of CHD or MI in patients with in the trial et al. found that dense LDL in patients with type This is an finding patients have a of dense LDL, and it likely explains of the reduction in CHD by therapy in patients with studies showed that reduce dense LDL in patients have a of this LDL and in patients with familial combined hyperlipidemia have a selective effect on any particular LDL et al. and other groups show that LDL type in a The mechanism is stimulation of across the of LDL and other apo B lipoproteins This may the finding that large LDL did not predict coronary in patients were with although it did in and the concentration of large and small LDL in patients with combined hyperlipidemia or of type the concentration of intermediate size LDL increased Because small LDL more than large LDL, the of LDL larger size particles lowers dense LDL concentration the of LDL to larger particles In one study increased the concentration of large LDL by NMR by about the same that it small LDL. of on LDL subfractions are with its to the plasma concentration of LDL particles about the by NMR of the individual LDL size subfractions. therapy with plasma concentration of large LDL with no change in small LDL; this is to an increase in clearance rate of large LDL from plasma compared with high mean LDL size the and subfractions in the The effect was a in LDL as both and reduce LDL size compared with it is to a clinical to or from such changes in LDL sizes during these In patients with small dense LDL or high triglycerides, the plasma concentration of VLDL remains high even when therapy has LDL cholesterol to the has this by using high triglyceride rather than the of small dense LDL as an of an VLDL concentration (45). that patients with high triglyceride have their be more to the of a low cholesterol, the used for VLDL LDL The of for any proposed risk factor is that it must add to risk assessment by or that it is but more LDL does not either of these studies that large and small LDL are atherogenic. In as much as any type of LDL is in the plasma LDL the standard clinical measurement of LDL be as The of response to lipid therapy is a reduction in the plasma concentration of atherogenic as by LDL and triglycerides, but by cholesterol or apo B cholesterol ester transfer coronary heart high-density intima-media low-density myocardial magnetic very low-density