There has been much interest in the significance of the lipoproteins in atherosclerosis. Large numbers of studies have been carried out, different populations have been examined, various diets have been tried, and endless pages of statistics have been published. Several laboratory assays that have general, but sometimes not unanimous, acceptance as predictors of atherosclerotic risk have emerged from these data. Some of these risk factors include cigarette smoking, fibrinogen (which can be elevated in part due to cigarette smoking but is still a risk factor in nonsmokers), diabetes mellitus, hypertension, and various serum lipids. Discussion follows of laboratory tests currently used to assess the level of coronary heart disease (CHD) risk induced by various risk factors.

Serum (total) cholesterol. Total serum cholesterol comprises all of the cholesterol found in various lipoproteins. Cholesterol is the major component of LDLs and a minority component of VLDLs and HDLs. Since LDL has consistently been associated with risk of atherosclerosis, and since LDL is difficult to measure, serum total cholesterol has been used for many years as a substitute. There is general agreement that a strong correlation exists between considerably elevated serum cholesterol levels and an increased tendency for atherosclerosis. Disadvantages include the following:

1. There is considerable overlap between cholesterol values found in populations and individuals at normal risk for atherosclerosis and those at higher risk. This leads to controversy over what values should be established as “normal” for the serum cholesterol reference range. A related problem is a significant difference between the reference range values for cholesterol based on “ideal” populations (i.e., derived from populations with a low incidence of atherosclerosis) compared with reference ranges from populations with a higher incidence of atherosclerosis (e.g., unselected clinically asymptomatic persons in the United States). This has led to objections that data from many persons with significant but subclinical disease are being used to help derive the reference values for populations with higher risk of CHD.

Whereas the upper limit of statistically derived U.S. values is about 275-300 mg/100 ml (7.2-7.6 mmol/L), some investigators favor 225 mg/100 ml (5.85 mmol/L) as the acceptable upper limit since that is the value representing average risk for CHD in the Framingham study. However, the average risk for CHD in the U.S. population of the Framingham study is higher than the average risk for a low-risk population. The National Institutes of Health (NIH) Consensus Conference on cholesterol in heart disease in 1984 proposed age-related limits based on degree of CHD risk. The NIH Conference guidelines were widely adopted. In some studies, serum cholesterol (as well as triglyceride) reference values are sex related as well as age related.

To make matters more confusing, many investigators believe that 200 mg/100 ml (5.15 mmol/L) should be considered the acceptable upper limit because that is the approximate upper limit for low-risk populations. The Expert Panel of the National Cholesterol Education Program (NCEP, 1987) chose 200 mg/100 ml without regard to age or sex (Table 22-6). Although the NCEP advocated use of total cholesterol as the basic screening test for CHD risk, they recommended that therapy should be based on LDL cholesterol values.

The NCEP Guidelines seem to be replacing the NIH Consensus Guidelines. One possible drawback is lack of consideration of HDL cholesterol effects (discussed later), which may be important since HDL is an independent risk factor.

2. Serum cholesterol can have a within-day variation that averages about 8% (range, 4%-17%) above or below mean values during any 24-hour period (±8% variation represents about ±20 mg/100 ml if the mean value is 250 mg/100 ml).
3. Day-to-day cholesterol values in the same individual can fluctuate by 10%-15% (literature range, 3%-44%).
4. Serum cholesterol values may decrease as much as 10% (literature range 7%-15%) when a patient changes from the erect to the recumbent position, as would occur if blood were drawn from an outpatient and again from the same person as a hospital inpatient. Two studies have shown less than 5% average difference between serum cholesterol obtained from venipuncture and from fingerstick capillary specimens in the same patient.
5. Various lipid fractions are considerably altered during major illnesses. For example, total cholesterol values often begin to decrease 24-48 hours after an acute myocardial infarction (MI). Values most often reach their nadir in 7-10 days with results as much as 30%-50% below previous levels. The effect may persist to some extent as long as 6-8 weeks. In one study not all patients experienced postinfarct cholesterol decrease. Although theoretically one could obtain valid cholesterol results within 24 hours after onset of acute MI, the true time of onset is often not known. Surgery has been shown to induce similar changes in total cholesterol to those following acute MI. HDL cholesterol also temporarily fell in some studies but not in others. Triglyceride levels were relatively unchanged in some studies and increased in others. In bacterial sepsis and in viral infections, total cholesterol levels tend to fall and triglyceride levels tend to increase. Besides effects of illness there are also effects of posture and diet change, stress, medications, and other factors that make hospital conditions different from outpatient basal condition. For example, thiazide diuretic therapy is reported to increase total cholesterol levels about 5%-8% (range, 0%-12%) and decrease HDL cholesterol to a similar degree. However, several studies reported return to baseline levels by 1 year. Certain medications can interfere with cholesterol assay. For example, high serum levels of ascorbic acid (vitamin C) can reduce cholesterol levels considerably using certain assay methods.
6. Certain diseases are well-known causes of hypercholesterolemia; these include biliary cirrhosis, hypothyroidism, and the nephrotic syndrome. A high-cholesterol diet is another important factor that must be considered.
7. Total cholesterol becomes somewhat increased during pregnancy. In our hospital, data from 100 consecutive patients admitted for delivery showed 16% with values less than 200 mg/100 ml (5.2 mmol/L); the lowest value was 169 mg/100 ml (4.4 mmol/L). Thirty-five percent were between 200-250 mg/1000 ml (5.2-6.5 mmol/L); 36% were between 250-300 (6.5-7.8 mmol/L); 10% were between 300-350 (7.8-9.1 mmol/L); and 3% were between 350-400 (9.1-10.4 mmol/L), with the highest being 371 (9.6 mmol/L). On retesting several patients 3-4 months after delivery, all had values considerably less than previously, although the degree of decrease varied considerably.
All of the major lipoprotein fractions, including chylomicrons, contain some cholesterol. Therefore, an increase in any of these fractions rather than in LDL alone potentially can elevate serum total cholesterol values. Of course, for lipoproteins with low cholesterol content the degree of elevation must be relatively great before the total cholesterol value becomes elevated above reference range.

In summary, according to the NCEP, 200 mg/100 ml (5.15 mmol/L) is the upper acceptable limit for any age. Lipid values obtained during hospitalization may be misleading, and borderline or mildly elevated values obtained on a reasonably healthy outpatient may have to be repeated over a period of time to obtain a more accurate baseline. Changes between one specimen and the next up to 20-30 mg/100 ml (0.52-0.78 mmol/L), or even more—may be due to physiologic variation rather than alterations from disease or therapy.

Although one would expect cholesterol in food to raise postprandial serum cholesterol values, actually serum cholesterol levels are very little affected by food intake from any single meal. Cholesterol specimens are traditionally collected fasting in the early morning because serial cholesterol specimens should all be drawn at the same time of day after the patient has been in the same body position (upright or recumbent) and because triglyceride (which is greatly affected by food intake) or HDL cholesterol assay are frequently performed on the same specimen.

Cholesterol assay on plasma using EDTA anticoagulant is reported to be 3.0-4.7 mg/100 ml (0.078-0.12 mmol/L) lower than assay on serum (depending on the concentration of EDTA).

Low-density lipoprotein cholesterol. The LDL (beta electrophoretic) fraction has been shown in various studies to have a better correlation with risk of atherosclerosis than total serum cholesterol alone, although the degree of improvement is not marked. As noted previously, the NCEP bases its therapy recommendations on LDL values. The major disadvantage of this approach is difficulty in isolating and measuring LDL. The most reliable method is ultracentrifugation. Since ultracentrifugation is available only in a relatively few laboratories and is expensive, it has been standard procedure to estimate LDLs as LDL cholesterol by means of the Friedewald formula. This formula estimates LDL cholesterol from results of triglyceride, total cholesterol, and HDL cholesterol.

One report suggests that modifying the formula by dividing triglyceride by 6 rather than 5 produces a more accurate estimate of LDL levels. A disadvantage of the Friedewald formula is dependence on results of three different tests. Inaccuracy in one or more of the test results can significantly affect the formula calculations. In addition, the formula cannot be used if the triglyceride level is greater than 400 mg/100 ml (4.52 mmol/L).

High-density lipoprotein cholesterol. Several large-scale studies have suggested that HDL levels (measured as HDL cholesterol) have a strong inverse correlation with risk of atherosclerotic CHD (the higher the HDL level, the less the risk). HDL seems to be a risk factor that is independent of LDL or total cholesterol. Some believe that HDL cholesterol assay has as good or better correlation with risk of CHD than total or LDL cholesterol. In general, the Framingham study suggested that every 20 mg/100 ml (5.2 mmol/L) reduction of HDL cholesterol corresponds to approximately a doubling of CHD risk. Disadvantages include certain technical problems that affect HDL assay, although methodology is becoming more simple and reliable. These problems include different methods that produce different results and need for two procedure steps (separation or extraction of HDL from other lipoproteins and then measurement of the cholesterol component), all of which produce rather poor correlation of results among laboratories. Ascorbic acid (vitamin C) may interfere (5%-15% decrease) with some test methods but not others. Reliability of risk prediction is heavily dependent on accurate HDL assay, since a relatively small change in assay values produces a relatively large change in predicted risk. HDL values are age and sex dependent. HDL values tend to decrease temporarily after acute MI, as do total serum cholesterol values. Hypothyroidism elevates HDL values and hyperthyroidism decreases them; therefore, in thyroid disease HDL values are not reliable in estimating risk of CHD. The possible effects of other illnesses are not as well known. Certain antihypertensive medications (thiazides, beta-blockers without intrinsic sympathomimetic activity, sympathicolytic agents) decrease HDL by a small but significant degree.

Since serum total cholesterol and HDL are independent risk factors, some patients may have values for one that suggest abnormality but values for the other that remain within reference limits. As independent risk factors, a favorable value for one does not entirely cancel the unfavorable effect of the other.

Serum cholesterol/high-density lipoprotein cholesterol ratio. Some investigators use the serum cholesterol/HDL cholesterol ratio as a convenient way to visualize the joint contribution of risk from these important risk factors. The ratio for normal risk is 5, for double risk is 10, and for triple risk is 20. Some believe that the ratio is the best single currently available predictor of CHD risk. Others believe that the ratio does not adequately demonstrate the independent contributions of the two factors and may be misleading in cases in which one or both factors may be abnormal, but the ratio does not suggest the actual degree of abnormality.

It should be mentioned that some uncertainty exists whether mortality data involving total cholesterol and HDL cholesterol is still valid in persons over age 60, and if so, to what degree.

Apolipoproteins. Apolipoprotein A (apo A) is uniquely associated with HDL, and measurement of apolipoprotein A1 (apo A1) has been proposed as a better index of atherogenic risk than assay of HDL cholesterol. Apolipoprotein B (apo B) comprises most of the protein component of LDL, which is composed of a core of cholesterol esters covered by a thin layer of phospholipids and free cholesterol around which is wrapped a chainlike molecule of the principal subgroup of apo B known as apo B100. Apo B100 is also the major B apolipoprotein component of VLDL. The apo B subgroup known as apo B48 (produced by the intestine) is a major structural protein in chylomicrons. Some research suggests that apo B may have a role of its own in cholesterol synthesis and that apo B measurement may provide a better indication of atherosclerotic risk than LDL cholesterol measurement. The apo A1/apo B ratio has been reported by some to be the best single predictor of CHD. However, there is some controversy over the role of apoprotein assay in current management of CHD. In my experience the total cholesterol/HDL ratio and the Apo A1/Apo B ratio, done simultaneously, gave approximately the same CHD risk assessment in the great majority of patients. The apoproteins have been quantitated mostly by immunoassay. Apo E4 gene has been proposed as a risk factor for Alzheimer’s disease. Apoprotein assay is still not widely available or widely used, and quality control surveys have shown problems in accuracy between laboratories with one international survey finding within-lab coefficients of variation (CVs) of 5%-10% and between-lab CVs of 15%-30%.

Triglyceride (TG). Triglyceride (TG) is found primarily in chylomicrons and in VLDLs. In fasting plasma, chylomicrons are usually absent, so TG provides a reasonably good estimate of VLDL. The usefulness of VLDL or TG as an indicator of risk for CHD has been very controversial. The majority opinion in the early 1980s was that TG levels do not of themselves have a strong predictive value for CHD. The majority opinion in the early 1990s cautiously suggests that such an independent role is possible but is not yet unequivocally proven. Several large studies reported a strong correlation between increased TG and increased CHD values. However, when the effect of other risk factors was considered, there was thought to be less evidence of an independent TG role. There is a roughly inverse relationship between TG and HDL levels, so that elevated TG levels tend to be associated with low HDL levels (which are known to be associated with increased risk for CHD). Currently, the major use of TG assay still is to calculate LDL using the Friedewald formula, to help screen for hyperlipidemia, and to help establish lipoprotein phenotypes.

Other factors that influence TG levels are frequently present. Nonfasting specimens are a frequent source of elevated TG levels. Postprandial TG levels increase about 2 hours (range, 2-10 hours) after food intake with average maximal effect at 4-6 hours. Therefore, a 12- to 16-hour fast is recommended before obtaining a specimen. Within-day variation for triglyceride averages about ± 40% (range, 26%-64%), with between-day average variation about ± 25%-50% (range, 18%-100%). Obesity, severe acute stress (trauma, sepsis, burns, acute MI) pregnancy, estrogen therapy, alcohol intake, glucocorticoid therapy, high-fat diet, and a considerable number of diseases (e.g., diabetes, acute pancreatitis, nephrotic syndrome, gout, and uremia) increase TG levels. Levels more than 1,000 mg/100 ml (11.29 mmol/L) interfere with many laboratory tests, and predispose for acute pancreatitis. There are also certain laboratory technical problems that may falsely decrease or increase TG values. High alkaline phosphatase levels increase TG levels to some degree in all TG methods. All TG methods actually measure glycerol rather than triglyceride, so that glycerol that is not part of TG (from a variety of etiologies) can falsely increase the result unless a “blank” is prepared and subtracted. Increased bilirubin, uric acid, or vitamin C levels interfere with some TG methods.

Plasma TG fasting values of 250 mg/100 (2.82 mmol/L) were considered to be the upper limit of normal in adults by an NIH Consensus Conference on hypertriglyceridemia in 1993. Fasting values more than 500 mg/100 ml (5.65 mmol/L) were considered definitely abnormal. Most laboratories perform TG assays on serum rather than plasma and apply the NIH cutoff values to the results, although serum values are about 2%-4% less than results obtained from plasma.

Lipoprotein (a) [Lp(a)] Lipoprotein (a) [Lp(a)] is a lipoprotein particle produced in the liver and composed of two components: one closely resembling LDL in structure which, like LDL, is partially wrapped by a chainlike apo B100 molecule, and an apolipoprotein (a) glycoprotein molecule covalently linked to apo B100 by a single disulfide bond. Apo (a) has a structure rather similar to plasminogen, which is the precursor molecule of the anticoagulant enzyme plasmin. The apo (a) gene is located on the long arm of chromosome 6 next to the gene for plasminogen. However, there are at least 6 alleles (isoforms) of apo (a), so that small variations in the structure and size of apo (a)—and therefore of Lp (a)—may occur. The apo (a) isoforms are inherited in a codominant fashion and Lp(a) is inherited as a autosomal dominant. In Europeans, Lp(a) distribution is considerably skewed toward the lower side of value distribution; while in African Americans there is a gaussian bell-shaped value distribution that is relative to Europeans results in a greater number of elevated values. Familial hypercholesterolemia, chronic renal failure requiring dialysis, the nephrotic syndrome, and postmenopausal decreased estrogen levels (in females) are associated with higher Lp(a) levels. Chronic alcoholism may decrease Lp(a) levels.

There now are a number of studies indicating that Lp(a) elevation is a very significant independent risk factor for atherosclerosis, especially for CHD and probably for stroke and abdominal aneurisms. About 10% of the general population have elevated levels of Lp(a). Lp(a) values over 30 mg/100 ml increase CHD risk two to threefold. When high levels of LDL and Lp(a) coexist, this raises the relative CHD risk up to fivefold. However, a few studies deny that Lp(a) is an important independent risk factor.

Lp(a) can be quantitated by a variety of immunoassay methods. Concentration has been reported as total Lp(a) mass; this includes both the lipid (HDL) and protein (apo[A]) components of Lp(a). The majority of the population has values less than 20 mg/dl (0.2 g/L). Elevation above 30 mg/dL (0.3 g/L) is associated with a twofold or more increase in CHD risk. Concentration has also been reported as apo(a) protein mass. Elevation above 0.5-0.7 g/L increases risk for CHD. However, these cutoff points were established in predominately European populations and may not be exactly applicable to other racial populations. There are problems with assay standardization (since currently there is no international standard material) and significant variations between laboratories and various assays. There is also a potential problem because apo(a) and plasminogen have considerable structural similarities, and therefore antibodies against either molecule may have some degree of cross-reaction. Postprandial specimens are reported to be 11%-13% lower than fasting specimens.

Summary. The most widely used current procedure to estimate risk of coronary heart disease is to obtain serum or plasma total cholesterol levels (as a substitute for LDL assay) and HDL cholesterol levels. If desired, the total cholesterol/HDL ratio can be calculated, and LDL cholesterol levels can be derived from the same data plus TG assay by means of the Friedewald formula. These studies are best performed when the patient is in a basal state. It is important to note that many investigators caution that such studies may be misleading when performed on hospitalized patients, due to the effects of disease and the hospital environment. Accuracy of total cholesterol, HDL, TG, and apolipoprotein measurements is increased if two or preferably three specimens are obtained, each specimen at least 1 week apart (some prefer 1 month apart), each obtained fasting at the same time of the day to establish an average value to compensate for physiologic and laboratory-induced fluctuations in lipoprotein measurements.