Triglyceride does not accumulate in the vessel wall. Its atherogenicity must be based, therefore, on the association, within lipoproteins, between triglyceride and cholesterol, the lipid that does accumulate in atherosclerotic plaque. Thus, proponents of an association between hypertriglyceridemia and atherosclerotic cardiovascular disease must postulate that disordered metabolism of the triglyceride-rich lipoproteins-very low density lipoprotein (VLDL) or chylomicron or both-is proatherogenic. This concept is not extreme: Even the well-accepted link between cholesterol and atherosclerotic cardiovascular disease relies on the accumulation of low-density lipoprotein (LDL) cholesterol, not simply plasma cholesterol, in lesions. The concept that several lipoprotein species can contribute cholesterol to lesion formation is not, therefore, unreasonable. Indeed, ample studies have shown that chylomicron remnants (the lipoprotein formed from chylomicrons after removal of much, but not all, of the chylomicron triglyceride) and VLDL (or its remnants) enter the subendothelial space of the vessel wall [5, 6]. These lipoproteins can also be internalized by monocyte macrophages, resulting in formation of foam cells. Recent studies in transgenic mice dramatically demonstrated the atherogenic potential of VLDL and chylomicron remnants, even when the mice were consuming normal chow diets [7].

If the delivery of cholesterol to the vessel wall by triglyceride-rich lipoproteins is, in fact, involved in the pathophysiology of atherosclerotic cardiovascular disease, epidemiologic data should show links between hypertriglyceridemia and this disease. When examining the role of triglyceride by univariate analyses, most prospective cohort studies have shown a positive relation between triglyceride levels and risk [2]. In multivariate analyses, however, triglyceride has often been shown to no longer be significantly associated with risk. Important exceptions may be seen in women with hypertriglyceridemia and patients with diabetes. I focus here only on three potential confounders of this statistical approach: the inverse relation between plasma levels of triglyceride and high-density lipoprotein (HDL) cholesterol, the potential atherogenicity of postprandial hypertriglyceridemia, and the biochemical and physical heterogeneity of triglyceride-rich lipoproteins.

High-density lipoprotein cholesterol is widely accepted as being protective against atherosclerotic cardiovascular disease [8]. The evidence has primarily been derived from prospective cohort studies and from secondary analyses of intervention trials in which lowering LDL cholesterol levels was the primary goal. The leading theory invokes a role for HDL in the "reverse cholesterol transport pathway," whereby cellular cholesterol can associate with HDL; in turn, the latter delivers cellular cholesterol to the liver for excretion. In most populations, triglyceride and HDL cholesterol levels are inversely related. Thus, analyses of triglyceride as a risk factor that use HDL cholesterol as a covariate will be confounded by the lack of independence of the two variables; both of these variables affect risk for atherosclerotic cardiovascular disease, but in opposite directions. Although recent studies in gene-knockout and transgenic mice support direct antiatherogenic roles for HDL [9, 10], the physiologic basis for the inverse relation between plasma levels of triglyceride and HDL cholesterol suggests that low HDL cholesterol levels may simply reflect proatherogenic disorders of triglyceride-rich lipoproteins. Increased levels of VLDL and chylomicrons stimulate the exchange of their triglyceride for cholesterol in HDL. Although this exchange, which is mediated in plasma by cholesteryl ester transfer protein [11], may be an alternate pathway for reverse cholesterol transport (with delivery of the cholesterol to the liver by VLDL and chylomicron remnants), movement of HDL cholesterol into triglyceride-rich lipoproteins could also result in the delivery of some cholesterol back to peripheral cells. Over time, such "short-circuiting" of the reverse cholesterol transport system could result in accumulation of cholesterol in tissues, including the arterial wall.

Essentially all of the epidemiologic data linking lipids and lipoproteins to risk for atherosclerotic cardiovascular disease are based on fasting levels [2]. Plasma levels of triglycerides vary widely throughout the day, however, because we absorb many grams of dietary fat after each meal. In the past decade, evidence for the atherogenicity of postprandial triglyceride-rich lipoproteins has increased. Several recent studies have indicated that postprandial levels of triglyceride and chylomicron remnants are predictive of the presence of atherosclerotic cardiovascular disease even when other risk factors, including HDL cholesterol, are considered [12, 13]. The role of triglyceride in atherogenesis might have been more readily demonstrated if epidemiologic studies had used nonfasting blood levels.

Another important reason for the difficulty in linking triglyceride levels to risk for atherosclerotic cardiovascular disease is the physical and biochemical heterogeneity of VLDL and chylomicrons. In particular, the size, number, and cholesterol content of these lipoproteins can vary greatly among individual persons. Size is important because very large particles cannot easily pass into the subendothelial space, the site of initiation of the atherogenic process. The number of triglyceride-rich particles in the bloodstream is also crucial, in part because the presence of more particles increases the probability that they will enter and be retained in the vessel wall, and in part because for any triglyceride level, more particles mean smaller, more atherogenic particles. Finally, the cholesterol content of the triglyceride-rich lipoproteins can vary significantly, and persons with more cholesterol-enriched particles will be at greater risk. Because of differences in the size, number, and cholesterol content of triglyceride-rich particles, patients with similar levels of plasma triglyceride may have very different risks for atherosclerotic cardiovascular disease.

So what is the clinician to do with the not uncommon, middle-aged patient who has hypertriglyceridemia? The National Cholesterol Education Program (NCEP) has not developed precise guidelines on which to base therapy [14]. Moreover, neither of the reports of two intervention trials gives a clearcut answer. The Helsinki Heart Study subgroup analysis [15] indicated that patients who had pure hypertriglyceridemia and were receiving gemfibrozil had a risk reduction similar to that of patients with either pure hypercholesterolemia or combined hyperlipidemia; however, the size of the subgroup was small. In the Stockholm Ischemic Heart Disease Study [16], the reductions in the rate of coronary events seen in patients treated with clofibrate and niacin correlated with reductions in triglyceride levels, but neither HDL nor LDL cholesterol levels were measured. In the face of these suggestive but incomplete data, one approach would be to use the NCEP guidelines, particularly focusing on the presence of other risk factors commonly associated with hypertriglyceridemia (hypertension, diabetes, low HDL cholesterol levels) to guide the intensity of therapy focused on LDL cholesterol.

Thus, if the patient is a man older than 45 years of age or a postmenopausal woman, it is likely that he or she will have two or more of the risk factors listed in the NCEP guidelines [14] and will require an LDL cholesterol level less than 130 mg/dL. If the patient has clinical atherosclerotic cardiovascular disease, an LDL cholesterol level less than 100 mg/dL would be the goal. When a patient with hypertriglyceridemia but not clinical atherosclerotic cardiovascular disease has an LDL cholesterol less than 130 mg/dL, however, the NCEP guidelines do not provide clear direction for therapy, regardless of the number of additional risk factors. Here, the LDL:HDL ratio might be helpful; both the Helsinki Study [17] and the Procam Study [18] indicated that patients with ratios exceeding 5.0 and triglyceride levels exceeding 200 mg/dL had markedly increased risk. It must be understood, however, that no data support the value of further lowering LDL cholesterol levels that are already less than 130 mg/dL in patients without clinical atherosclerotic cardiovascular disease. Measurement of apoprotein B levels (when standardization programs are widely available) might be of particular benefit in these patients [19].

Because of the multiple links between elevated triglyceride levels and risk for atherosclerotic cardiovascular disease, it seems prudent to screen for the presence of hypertriglyceridemia when determining a patient's risk for atherosclerotic cardiovascular disease. A therapeutic program for hypertriglyceridemia should obviously include diet, exercise, and weight loss programs: These are even more important in patients with hypertriglyceridemia than in persons with isolated hypercholesterolemia. When drugs are required, niacin is the logical first choice because of its beneficial effects on LDL cholesterol, HDL cholesterol, and triglyceride. Despite a higher incidence of adverse effects compared with other hypolipidemic agents, niacin therapy should be attempted unless it is specifically contraindicated. The hydroxymethylglutaryl CoA reductase inhibitors would be the next choice, given their efficacy in lowering LDL levels and their modest effects on triglyceride and HDL cholesterol. High doses of the most potent hydroxymethylglutaryl CoA reductase inhibitors seem to decrease triglyceride levels significantly. In patients who also have very high LDL cholesterol levels, these drugs could be the first choice for therapy. Gemfibrozil, the only fibrate available in the United States, has variable effects on LDL cholesterol and probably should be used first only if fasting triglyceride levels are greater than 400 to 500 mg/dL. The bile acid-binding resins are not recommended for monotherapy because they tend to increase triglyceride concentrations. Finally, when LDL cholesterol goals are met but triglyceride levels remain high and HDL cholesterol levels are low, gemfibrozil in combination with resins (or, in the patient at very high risk, in combination with a hydroxymethylglutaryl CoA reductase inhibitor) can be very effective. A triglyceride level goal of less than 200 mg/dL seems reasonable. An ongoing trial of triglyceride lowering [20] is examining this important management dilemma and may provide answers to the ultimate question: Does treating hypertriglyceridemia reduce the risk for atherosclerotic cardiovascular disease?