Elevated High-Density Lipoprotein (HDL) Levels due to Hepatic Lipase Mutations Do Not Reduce Cardiovascular Disease Risk: Another Strike against the HDL Dogma

Hepatic lipase (HL) is a physiological regulator of lipid trafficking in plasma and one of the sparse connection points linking the high-density lipoprotein (HDL) system with the pathways of apolipoprotein (apo) B-containing lipoprotein disposal. From its central position on the hepatic sinusoid, HL influences lipolysis of remnant particles, thereby determining the rate of remnant clearance, the rate of low-density lipoprotein (LDL) formation, and the density and size of LDL (1). Reduced HL activity causes formation of buoyant, triglyceride (TG)-rich LDL and accumulation of partially lipolyzed β-very low density lipoproteins (2). These metabolic changes have a recognized influence on atherogenesis and may mediate the effect of HL on vascular disease. However, HL also modulates HDL cholesterol levels in plasma and the size distribution of HDL particles. Reduced HL activity results in higher HDL cholesterol levels and accumulation of larger TG-rich HDL. These changes also are expected to have a significant influence on vascular health. Because of its multiple influences on lipoproteins from different compartments and the potential for competing downstream effects of HL modulation, this enzyme is not currently a strong target for the development of specific inhibitors. However, some available drugs, such as the antidiabetic agents of the thiazolidinedione class or the antilipidemic agents of the niacin class, may affect plasma lipids partially through inhibition of this enzyme (3, 4). Like many other single genes with a significant influence on HDL metabolism and plasma HDL cholesterol levels, HL can be studied to determine whether sequence polymorphisms or mutations that influence HDL cholesterol presumably from birth confirm the notion of an inverse linearity between HDL levels, commonly distributed in the population as a consequence of a variety of genetic and environmental factors, and cardiovascular disease rates. The well-established notion that higher HDL cholesterol levels in populations predict reduced cardiovascular risk has led to a therapeutic philosophy of expecting cardiovascular benefits from HDL cholesterol elevations, irrespective of the mechanisms targeted. Such a notion has been, however, difficult to verify in situations where HDL cholesterol levels are predominantly under the control of a single gene, rather than of multiple genetic and environmental effectors. For example, a unique mutation in the gene for apoAI, the main structural protein of HDL, differing from the wild-type protein by a single amino acid change (apoAI173Arg→Cys, better known as apoAIMilano), confers low HDL cholesterol levels and paradoxical protection against cardiovascular disease (5). Indeed, efforts have been made to develop apoAIMilano as an injectable drug for the reduction of coronary atheroma burden (6). Similarly, common polymorphisms in the ATP-binding cassette type A1 (ABCA1), the gene whose protein is responsible for the bulk export of intracellular cholesterol into the HDL compartment, are accompanied by significant reductions in HDL cholesterol levels but have failed to show an association with increased cardiovascular risk (7). On the other side of the spectrum, elevations in HDL cholesterol due to mutations or polymorphisms in genes that regulate HDL remodeling [cholesteryl ester transfer protein (CETP)] or clearance [scavenger receptor type BI (SR-BI)] have not been clearly linked to vascular protection (8), whereas trials with an inhibitor of plasma CETP have suggested worsening of atherosclerosis despite significant increases (60% or better) in plasma HDL cholesterol achieved under treatment with this drug (9, 10). Although the CETP inhibitor tested had additional non-HDL-based effects (elevated blood pressure, increases in electrolyte and aldosterone levels) that may have caused worsening of vascular disease (11, 12), prudent investigators should take the available information as a strong suggestion that some HDL elevations are not therapeutic and are possibly harmful (13). A typical example of this concept in the experimental world is the mouse model engineered for the loss of the hepatic HDL receptor gene (SR-BI). Without the HDL receptor, the SR-BI null mouse shows a large accumulation of HDL cholesterol in plasma, an obvious consequence of interrupted plasma clearance (14). However, under conditions of hypercholesterolemia, these mice display a devastating atherogenic phenotype including coronary thrombosis and early death due to myocardial infarction, suggesting that the accumulating HDL may aggravate the growth of the atheroma (15). This issue of JCEM contains an interesting series of analyses from the investigators of the Copenhagen City Heart Study (CCHS), with results supporting the notion that individuals whose HDL cholesterol is elevated because of HL mutations do not enjoy the expected HDL-associated protective effects on ischemic disease rates (16). In this study, the authors use three different databases. The first one is from the CCHS, a prospective study with up to 28 yr of follow-up and a total of nearly 9,000 individuals characterized by phenotypic variables and HL genotypes. The subjects enrolled in this trial were investigated for association between HDL levels and incidence of ischemic vascular disease. In addition, two similar case-control studies were performed, one looking at 2110 patients with angina or myocardial infarction matched up to a 1 to 4 ratio with 4899 controls from the CCHS, and the other one with 769 patients with carotid artery stenosis of more than 50%, or history of stroke, matched in the same approach with 2836 controls from the CCHS. Finally, the investigators combined all cases with ischemic cardiovascular disease (ICD) from the prospective study and the two case-control studies (a total of 4626 cases and 7224 controls) in a Mendelian randomization study to look at the influence of HDL cholesterol levels directly determined by hepatic lipase mutations on the risk of ischemic vascular disease. The Mendelian randomization approach (17) deletes the bias of reverse causality or the possibility of undefined common factors that may link both predicting variable and predicted event (HDL cholesterol and ICD, in this case). The mutations studied were five previously reported amino acid substitutions of which only one (S267F) had an influence on HDL levels (18). In addition, the authors also tested a previously reported polymorphism in the promoter region of HL (−480c>t) with significant influence on plasma HDL cholesterol levels. The S267F mutation is known to inactivate completely the function of HL in homozygous carriers; however, only heterozygous subjects were identified in this study population, and the frequency of the rare allele was only 0.4%. Carriers of the rare allele showed significant increases in plasma HDL cholesterol levels compared with noncarriers (around 6 mg/dl in men, 13 mg/dl in women). On the other hand, the rare allele of the promoter polymorphism −480c>t was present in 21% of subjects, offering the possibility to investigate all three genotypes (CC, TT, CT), although the HDL effect was modest (+3–4% in heterozygous and +8–9% in homozygous subjects). The study starts off with the mandatory validation of the inverse relationship between plasma HDL cholesterol distribution and hazard ratio for ICD. The adjusted risk for subjects in the bottom quintile was 65% higher than that for the subjects in the top quintile, thus confirming the notion that HDL cholesterol levels influenced by multiple genetic and environmental variables either convey or associate with reduced cardiovascular disease burden. In the CCHS analysis, the hazard ratios for ICD (myocardial infarction, angina, stroke, transient ischemic attack, and amaurosis fugax) in the carriers of rare alleles of the six different HL genetic variants did not differ from those of the noncarriers. The same was true for both case-control studies. In the Mendelian randomization analysis, an increment in HDL cholesterol of 8 mg/dl, attributable to heterozygosity for the rare S267F HL allele, predicted a cardiovascular risk reduction of 13%, an assumption taken from the 44% lower event rate due to a 1 mmol/liter (38 mg/dl) increase in HDL cholesterol in the entire CCHS population. However, when all events and cases from the three trials were combined, the odds ratio for the S267F carriers was 19% higher than that of noncarriers, therefore completely deleting the benefits expected from higher HDL cholesterol levels. This was also seen for the promoter region deletion (−480c>t), which increased HDL cholesterol modestly and was expected to produce risk reductions ranging from 4% in heterozygous carriers to 9% in homozygous carriers. Contrary to expectation, the odds ratios were 4 and 8% higher than those of noncarriers, respectively. The importance of this study lies in the fact that the two functional HL mutations studied had an exclusive effect on HDL cholesterol levels, thereby eliminating the confusion that may arise from multiple effects on levels of LDL and remnant particles. The other four HL mutations investigated had no influence on any of the lipid parameters and no effect on cardiovascular risk. These observations provide a strong basis for the notion that HDL elevations due to HL mutations are not linked to cardiovascular benefits and may conversely increase cardiovascular risk. The value of this study is somewhat tempered by the low frequency of the HL allele with the largest influence on HDL and by the only modest effect on HDL of the well-distributed promoter polymorphism. However, the results agree with a previous analysis of the same cohort after a 7-yr follow-up, showing that ICD risk increased by up to 50% in subjects carrying homozygous HL promoter mutations (including the −480) that elevate HDL levels (19). What would be the reason why HDL cholesterol elevations due to multiple influences are protective against vascular disease, whereas increases due to single gene mutations can have neutral or even harmful effects? The answer likely relates to the degree of functionality of an HDL particle whose remodeling in the plasma compartment is greatly influenced by only one of the many factors that normally control reverse cholesterol transport. The TG-rich HDL induced by low HL activity may turn out to be a hypofunctional, nonfunctional, or dysfunctional particle. It is plausible to forecast situations where HDL cholesterol levels and HDL functionality do not go hand in hand. Indeed, the most likely occurrence of a nonfunctional or dysfunctional HDL may come in conditions of elevated plasma HDL cholesterol, as a consequence of impaired clearance of the HDL and its cargo from the plasma compartment. The results of the Mendelian randomization analysis pose a more fundamental question: is it possible that the well-established inverse relationship between HDL cholesterol and coronary heart disease rates in populations is actually a spurious one? In the lipid arena, the Mendelian randomization approach has served well the cause in support of the vascular benefits linked to LDL lowering by providing evidence that people carrying mutations in the gene for proprotein convertase subtilisin/kexin type 9 (PCSK9, a regulator of LDL receptor expression) leading to low lifetime LDL levels have significantly reduced risk of cardiovascular events when compared with age- and risk factor-matched controls with normal LDL levels (20). The fact that this was not the case in the study by Johannsen et al. (16) provides support to an idea that cannot yet be embraced but should not be dismissed, i.e. that elevated plasma HDL cholesterol levels and reduced risk of vascular disease in populations may be partially unrelated consequences of a healthy lifestyle. The authors are supported by National Institutes of Health Grants HL57986 and HL65709 (to S.F.) and HL65405 and HL53989 (to M.F.L.). Apolipoprotein cholesteryl ester transfer protein high-density lipoprotein hepatic lipase ischemic cardiovascular disease low-density lipoprotein scavenger receptor type BI triglyceride