Apolipoprotein B and apolipoprotein A-I: risk indicators of coronary heart disease and targets for lipid-modifying therapy


Professor Göran Walldius, AstraZeneca, Pepparedsleden 1, 431 83 Mölndal, Sweden (fax: +46 317763802; e-mail: goran.walldius@astrazeneca.com).


Although LDL cholesterol (LDL-C) is associated with an increased risk of coronary heart disease, other lipoproteins and their constituents, apolipoproteins, may play an important role in atherosclerosis. Elevated levels of apolipoprotein (apo) B, a constituent of atherogenic lipoproteins, and reduced levels of apo A-I, a component of anti-atherogenic HDL, are associated with increased cardiac events. Apo B, apo A-I and the apo B/apo A-I ratio have been reported as better predictors of cardiovascular events than LDL-C and they even retain their predictive power in patients receiving lipid-modifying therapy. Measurement of these apolipoproteins could improve cardiovascular risk prediction.


For over three decades it has been recognized that a high level of total blood cholesterol, particularly in the form of LDL cholesterol (LDL-C), is a major risk factor for developing coronary heart disease (CHD) [1–4]. However, as more recent research has expanded our understanding of lipoprotein function and metabolism, it has become apparent that LDL-C is not the only lipoprotein species involved in atherogenesis. A considerable proportion of patients with atherosclerotic disease have levels of LDL-C and total cholesterol (TC) within the recommended range [5, 6], and some patients who achieve significant LDL-C reduction with lipid-lowering therapy still develop CHD [7].

Other lipid parameters are also associated with elevated cardiovascular risk, and it has been suggested that LDL-C and TC may not be the best discriminants for the presence of coronary artery disease (CAD) [5]. Elevated levels of intermediate-density lipoprotein (IDL) and very low-density lipoprotein (VLDL) are also associated with increased cardiovascular risk, as are low levels of HDL and high levels of plasma triglyceride (TG) [8–11].

Apolipoproteins are important components of lipoprotein particles, and there is accumulating evidence that measurement of various forms of apolipoproteins may improve the prediction of the risk of cardiovascular disease [5, 10–12]. The involvement of apolipoproteins in regulating the synthesis and metabolism of lipoprotein particles is gradually being defined.

This paper will review the role of apolipoproteins in atherogenesis. The clinical relevance of apolipoprotein levels to CHD risk and the effects of lipid-modifying drugs on apolipoprotein levels will be examined. In particular, apolipoproteins B and A-I (apo B and apo A-I) will be discussed as likely candidates for clinical use in the near future.

Apolipoproteins, lipoprotein metabolism and atherogenesis

Lipoprotein particles are made up of an insoluble lipid core surrounded by a coat of phospholipid, free cholesterol and apolipoproteins. Each class of lipoprotein particles is associated with distinctive apolipoproteins (Table 1) that, in addition to stabilizing lipoprotein structure, play an essential role in regulating metabolism. Some of the apolipoproteins act as ligands to tissue receptors, whilst others activate or inhibit enzymes involved in metabolic steps in the circulation or tissues. Figure 1 summarizes the main pathways of lipoprotein metabolism and the apolipoproteins involved. The various apolipoproteins are active at different points in these pathways and understanding their roles in the development of CHD is a focus of current research interest.

Table 1.  Major apolipoprotein components of human plasma lipoproteins
LipoproteinMajor apolipoprotein components
  1. VLDL, very low-density lipoprotein; IDL, intermediate-density lipoprotein.

ChylomicronB-48, C, E
VLDLB-100, C, E
IDLB-100, E
 Subfraction 2A-I, A-II, C
 Subfraction 3A-II, A-I, C
Figure 1.

Summary of the general pathways of lipoprotein metabolism (adapted from Ref. [13]). Apo, apolipoprotein; CETP, cholesteryl ester transfer protein; HL, hepatic lipase; IDL, intermediate-density lipoprotein; LCAT, lecithin cholesterol acyl transferase; LL, lipoprotein lipase; VLDL, very low-density lipoprotein.

Apolipoprotein B

Apolipoprotein B exists in two forms, apo B-48 and apo B-100. Apo B-48 is synthesized in the intestine, where it is complexed with dietary TG and free cholesterol absorbed from the gut lumen to form chylomicron particles. These are metabolized in the circulation and in the liver (Fig. 1). Apo B-100 is synthesized in the liver and is present in LDL, IDL and VLDL particles [14]. Only one apo B molecule is present in each of these lipoprotein particles [15] and therefore the total apo B value indicates the total number of potentially atherogenic lipoproteins (Fig. 2) [5, 11]. Apo B is essential for the binding of LDL particles to the LDL receptor, allowing cells to internalize LDL and thus absorb cholesterol (Fig. 1). An excess of apo B-containing particles is a main trigger in the atherogenic process. Small dense LDL particles are considered more atherogenic than large buoyant LDL molecules, as they are easily internalized into the subintimal space where they adhere to matrix proteoglycans [16], are oxidized, and increase the risk of atherothrombotic events [17].

Figure 2.

Atherogenic and anti-atherogenic lipoproteins. Apo B, apolipoprotein B; Apo A-I, apolipoprotein A-I; VLDL, very low-density lipoprotein; IDL, intermediate-density lipoprotein; TG, triglyceride; C, cholesterol.

The concentration of plasma apo B particles is highly correlated with the level of non-HDL cholesterol (non-HDL-C), defined as TC minus HDL-C [18]. As HDL is known to be protective against cardiovascular risk, non-HDL-C reflects the fraction of blood cholesterol that is not contained in atheroprotective lipoproteins. Therefore, non-HDL-C has been recognized by the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) guidelines as a target for lipid-lowering therapy [19]. Non-HDL-C has been found to predict nonfatal myocardial infarction (MI) and angina pectoris [20]. However, apo B has been found to be a better predictor of risk than non-HDL-C [21], as discussed in more detail below.

Individuals with seemingly low or normal LDL-C levels can still be at increased risk of cardiovascular events. In these patients, the risk of cardiovascular events appears to be more closely related to an increased number of small, dense LDL particles in addition to hypertriglyceridaemia and a low level of protective HDL-C [17, 22], a combination known as the atherogenic lipid triad. Thus, LDL-C levels are not always a good or adequate indicator of cardiovascular risk, and it has been noted that the level of apo B and/or apo B/apo A-I in the plasma may be a better predictor [5, 11, 21, 23, 24]. Target levels for apo B have now been included in a table on treatment goals in an update of the NCEP ATP III guidelines [25]. From several perspectives, including pathophysiology, diagnosis, assessment of therapy and methodological soundness, there are powerful arguments to support the routine use of apo B in the clinical setting [26]. As discussed below, using apo B and apo A-I, expressed as the apo B/apo A-I ratio, seems to be a very effective way of characterizing cardiovascular risk in any patient irrespective of their lipoprotein abnormality [11, 21]. Patients with diabetes or the metabolic syndrome can have normal LDL-C levels but possess aspects of the atherogenic lipid profile [27], and these individuals often have a high ratio of apo B/apo A-I, which is a strong indicator of cardiovascular risk [11, 24].

Apolipoprotein A

Apolipoprotein A has two major forms, apo A-I and apo A-II. Apo A-I is the major apolipoprotein associated with HDL-C (Table 1). Levels of apo A-I are strongly correlated with those of HDL-C, and expression of apo A-I may be largely responsible for determining the plasma level of HDL [28]. Apo A-I also acts as a cofactor for lecithin cholesterol acyl transferase (LCAT) [29], which is important in removing excess cholesterol from tissues and incorporating it into HDL for reverse transport to the liver (Fig. 1) [14]. Furthermore, apo A-I is the ligand for the ATP-binding cassette (ABC) protein, ABCA1, and hence is involved in the docking procedure by which excess cholesterol in peripheral cells is externalized to HDL [30–33] for further reverse cholesterol transport either directly or indirectly via LDL back to the liver [34, 35].

Experiments in transgenic mice have shown that Apo A-II inhibits hepatic and lipoprotein lipase (LL) activity [36]. This effect tends to increase plasma TG and reduce plasma HDL. Thus, although apo A-I is consistently protective against cardiovascular risk, the influence of apo A-II is still unclear [37].

HDL exists as particles of different sizes, with HDL-2 being the largest and containing the most lipid in its core. HDL-3 particles are smaller and pre-β-HDL is the smallest, and these may be the most active particles in taking up peripheral cholesterol [38]. Apolipoprotein composition can be used to separate HDL into subpopulations: HDL containing apo A-I and apo A-II (HDL A-I : A-II), and HDL containing apo A-I but not apo A-II (HDL A-I) [37]. HDL A-I is more effective than HDL A-I : A-II in promoting cholesterol efflux [37], which is consistent with the atheroprotective effect of apo A-I on LCAT [29, 35]. The Prospective Epidemiological Study of Myocardial Infarction (PRIME) study examined the association between the incidence of CHD and several HDL-related parameters, including HDL-C itself, apo A-I, HDL A-I, and HDL A-I : A-II [39]. All four parameters were related to CHD risk, however, apo A-I was the strongest predictor [39]. In addition, the use of apo A-I for predicting CAD has been confirmed by other studies [40–42].

Other apolipoproteins

Apolipoprotein C (apo C) is associated with chylomicrons, VLDL-C and HDL-C (Table 1) [14]. Several major subtypes are known and all modulate the metabolism of TG-rich lipoproteins, principally IDL and VLDL, in a variety of complex ways [43, 44].

Apolipoprotein C-I has an inhibitory action on the uptake of VLDL via hepatic receptors [44], it is the major plasma inhibitor of cholesteryl ester transfer protein (CETP) and it interferes directly with fatty acid uptake [45]. Apo C-II is a major activator of LL [46] and therefore regulates TG levels by stimulating TG hydrolysis, a lipolytic process that liberates fatty acids from the lipoprotein particle and makes them available for uptake in the periphery, mainly by adipose tissue. Apo C-III inhibits the lipolysis of TG-rich lipoproteins and interferes with their clearance from the circulation [44, 45]. High levels of apo C-III have been associated with increased risk of atherogenesis and cardiovascular events [47]. A high apo C-III level is also found in subjects with hypertriglyceridaemia as part of the metabolic syndrome [48, 49]. In the Monitored Atherosclerosis Regression Study (MARS) and the Cholesterol Lowering Atherosclerosis Study (CLAS), a high apo C-III level was strongly related to atherosclerotic lesion progression [50, 51]. Apolipoprotein C-IV is contained in TG-rich lipoproteins, such as chylomicrons, and also in HDL. It is involved in the regulation of lipid absorption.

Apolipoprotein E (apo E) is a constituent of VLDL, IDL and chylomicrons (Fig. 1). There is increasing evidence that it protects against atherogenesis through a variety of mechanisms [52, 53]. The apo E gene is polymorphic and there are three common apo E proteins: apo E2, apo E3 and apo E4 [54]. The genetic variation in apo E has a strong effect on its anti-atherogenic properties [52, 54]. The E4 allele is disproportionately common in patients with CAD, indicating a pro-atherogenic effect [55, 56]. Apo E genotyping may be clinically helpful in defining the CHD risk of patients and predicting their likely responses to therapeutics [56], however, further studies are required to confirm the clinical use of apo E. Apolipoprotein D is primarily associated with HDL, but its role in lipid metabolism has yet to be defined [57].

Apolipoprotein a (apo[a]), which is bound to LDL to form lipoprotein(a) [Lp(a)], shares 80% of its amino acid sequence with plasminogen and competitively inhibits the surface binding and activation of plasminogen [58]. It is strongly genetically determined, with large variability in different populations [59]. Elevated Lp(a) has been associated with arterial wall thickening [58, 60], and high blood concentrations of Lp(a) and apo(a) have been suggested as risk factors for atherosclerosis [58, 60, 61]. High levels of Lp(a) and early signs of atherosclerosis, such as increased intima-media thickness (IMT), have been found in children and young adults with a parental history of premature MI [62]. In subjects with high Lp(a) and a moderately elevated or high LDL-C, CHD risk is clearly increased and treatment is necessary to lower LDL-C. Of all lipid-lowering drugs, only nicotinic acid has been shown to significantly reduce the levels of Lp(a) [63]. The role of Lp(a) in the pathogenesis of atherosclerotic cardiovascular disease and its use as a predictor of CHD events has recently been reviewed [64]. International standardization of Lp(a) is ongoing [65].

Alaupovic et al. [66, 67] have developed methods by which several subspecies of combinations of apolipoproteins B, A, C, and E are measured and their relations to cardiovascular disease are studied.

Apolipoproteins B and A-I as markers of atherosclerosis

Clinical studies have established that elevated cholesterol [1–4, 11] and TG levels [11, 68–71], and low levels of HDL-C [11, 71, 72] are associated with increased CHD risk. The relationships between apolipoproteins and atherosclerotic disease have been less well examined, but studies are beginning to investigate this area, particularly for apo B and apo A-I. Although the first studies were presented about 20 years ago [73–75], more recent clinical trials using softer end-points like coronary angiography, angina pectoris and nonfatal MI have been published and reviewed [76], with most pointing to the importance of apo B and apo A-I as risk indicators.

Measurement of apolipoprotein levels has methodological advantages over measurement of LDL-C [26, 77, 78]. In the routine clinical setting and also in major treatment trials like those of the statins, LDL-C is most often calculated from multiple parameters, usually using the Friedewald formula [79]. This approach is inherently subject to compound errors for the LDL-C values [80] and is especially problematic in the low range of LDL-C. According to Scharnagl et al. [81], LDL-C calculations are frequently inaccurate at levels <3 mmol L−1, and can result in clinical misjudgement [82]. Furthermore, measurements are not standardized and LDL-C cannot be calculated in subjects with TG levels above 4.5 mmol L−1. In contrast, apo B and apo A-I can be measured directly, accurately and precisely [11, 77]. Moreover, measurement of apo B and apo A-I does not require fasting blood samples [11], is internationally standardized [83, 84], inexpensive to perform, may be conducted on frozen samples and can be automated, as was carried out in the Apolipoprotein-related MOrtality RISk (AMORIS) study [11, 85, 86].

Endothelial function and imaging studies

Apolipoprotein B has been found to be an independent predictor of endothelial vasodilatory function [87], increased carotid IMT and arterial stiffness in healthy subjects [88–90]. Furthermore, apo B was also identified as an independent predictor of carotid IMT in patients with familial combined hyperlipidaemia [91]. Cuomo et al. [62] also reported that the IMT of the common carotid artery was pathologically increased, and levels of apo B and Lp(a) elevated, in young relatives of MI patients. An angiographic study demonstrated that women with significant CAD, defined as stenosis of more than 60% in at least one coronary artery, had higher levels of apo B, LDL-C, TG and TC and lower levels of HDL-C than women without CAD (Table 2) [92]. The apo B level was found to be the best predictor of the extent of CAD after correction for age, as assessed by the number of stenotic arteries, and was the only statistically significant predictor of the presence of CAD in patients without dyslipidaemia (Table 2) [92]. Similarly, Snehalatha et al. [93] found that apo B and apo A-I were more common and strongly related to the presence of angiographically proven CHD than LDL-C in Asian Indians with diabetes. In addition, in male survivors of MI under 45 years of age, Hamsten et al. [94] found that 18% of the variation in atherosclerosis score, based on angiography and adjusted for age, smoking habits and body weight, was related to the level of apo B. Addition of other lipoproteins variables, hypertension, and glucose intolerance did not add to the prediction [94]. In the CLAS study, men with previous coronary bypass surgery and progressive atherosclerosis were randomized to cholesterol-lowering therapy or placebo, and the thickness of the carotid artery wall was measured using ultrasound after 2 and 4 years of treatment [95]. Regression analysis showed that reduced apo B, increased apo C-III and increased HDL-C levels were significantly associated with reductions in artery wall thickness [95].

Table 2.  Fasting plasma lipids and lipoproteins in women with (n = 160) and without (n = 129) coronary artery disease (CAD) as assessed by angiography (adapted from Ref. [92] with kind permission from the publisher)
 Without CADWith CADP-valuea
  1. Data are expressed as mean ± SD. aP-value by logistic regression using age as a covariate. bLogarithmically transformed. cTC ≤6.5 mmol L−1 and TG ≤2.3 mmol L−1. Apo, apolipoprotein; TC, total cholesterol; HDL-C, HDL cholesterol; TG, triglyceride; LDL-C, LDL cholesterol; NS, not statistically significant.

All subjects
 TC (mmol L−1)6.38 ± 1.22 (n = 129)7.01 ± 1.19 (n = 160)0.0011
 TGb (mmol L−1)1.71 ± 0.93 (n = 113)1.98 ± 0.84 (n = 145)0.0065
 HDL-C (mmol L−1)1.37 ± 0.38 (n = 111)1.28 ± 0.28 (n = 146)0.019
 LDL-C (mmol L−1)4.13 ± 1.13 (n = 111)4.74 ± 1.09 (n = 146)0.0008
 Apo A-I (g L−1)1.53 ± 0.34 (n = 96)1.49 ± 0.24 (n = 131)NS
 Apo B (g L−1)1.25 ± 0.34 (n = 96)1.48 ± 0.32 (n = 131)<0.0001
Normolipidaemicc subjects
 TC (mmol L−1)5.63 ± 0.69 (n = 75)5.87 ± 0.54 (n = 52)NS
 TGb (mmol L−1)1.37 ± 0.42 (n = 66)1.43 ± 0.47 (n = 50)NS
 HDL-C (mmol L−1)1.35 ± 0.33 (n = 64)1.35 ± 0.31 (n = 50)NS
 LDL-C (mmol L−1)3.58 ± 0.78 (n = 64)3.89 ± 0.59 (n = 50)0.054
 Apo A-I (g L−1)1.48 ± 0.33 (n = 58)1.49 ± 0.28 (n = 49)NS
 Apo B (g L−1)1.08 ± 0.28 (n = 58)1.21 ± 0.20 (n = 49)0.019

CHD events related to apolipoprotein B and apolipoprotein A-I levels

In the Bogalusa Heart Study, children of parents with a history of MI had high apo B and low apo A-I levels, and thus a high apo B/apo A-I ratio, whereas their LDL-C and HDL-C levels were not outside normal limits [96]. Subsequent clinical studies have investigated the predictive value of apolipoprotein levels in cardiovascular risk. In a case–control analysis of data from the Cholesterol And Recurrent Events (CARE) trial, baseline levels of lipid and apolipoprotein parameters were compared between a group of patients who had subsequent MI or coronary death (cases) and a group who did not have such events (controls) in 5 years of follow-up [47]. The baseline plasma levels of apo B in VLDL, apo C-III in VLDL and LDL, and apo E in HDL were demonstrated to be significant independent predictors of subsequent coronary events [47]. In patients with unstable angina in the European Concerted Action on Thrombosis and Disabilities (ECAT) Angina Pectoris study, baseline levels of HDL-C and apo A-I were found to be the strongest predictors of MI, independent of other coronary risk factors and lipid measurements [97]. Apo B was also associated with coronary events, but in this study the association was not independent of HDL-C [97]. In addition, Lundstam et al. [98] reported that low levels of apo A-I and HDL-C were significantly related to an increased risk of total or cardiac mortality in patients who had previously developed angiogram-positive angina pectoris.

Apolipoprotein B was also found to be an independent predictor of ischaemic heart disease (IHD) in the prospective Quebec Cardiovascular Study [12]. Subjects in this study were stratified according to baseline apo B level, and men in the highest tertile for apo B were found to be at greater risk of IHD than men in the lowest tertile (P < 0.0005) [12]. The relationship between baseline apo B level and IHD risk was found to be independent of TG and HDL-C, and apo B level was a better predictor of IHD than the TC/HDL-C ratio.

The large prospective AMORIS study, which published its first report of the relationship between lipid levels and cardiovascular events in 1992 [99], was designed specifically to assess the power of apo B, apo A-I and the apo B/apo A-I ratio for predicting fatal acute MI or sudden death, and whether apo B and apo A-I added predictive power over and above that of TG, TC and LDL-C levels [11]. Raised apo B levels, an increased apo B/apo A-I ratio and low levels of apo A-I were highly predictive of risk of fatal MI in univariate analyses. Compared with subjects in the lowest quartile, those in the highest quartile for apo B had almost a threefold increase in risk; for the apo B/apo A-I ratio the increase in risk was almost fourfold in men and threefold in women. Risk for subjects in the highest quartile of apo A-I was less than half that of subjects in the lowest quartile. The results were similar for males and females, and in patients above and below 70 years of age. Furthermore, in multivariate analyses, high apo B levels, high apo B/apo A-I ratio and low levels of apo A-I were stronger predictors of risk than LDL-C, TC and TG levels [11]. When adjusting for TC, TG and age there was a dose–response relationship showing that the highest risk for both males and females is found in those patients with the highest apo B and lowest apo A-I levels [11]. Follow-up has now exceeded 99 months with results based on 1267 acute MI deaths in males and 586 in females. Previously unpublished AMORIS data indicates that the risk of acute MI associated with elevated apo B and low apo A-I levels, expressed here as acute MI deaths/1000 observation years (Fig. 3a,b; males and females, respectively) or as the risk ratio adjusted for TC, TG, and age (Fig. 4a,b; males and females, respectively), is even stronger than that reported in earlier papers. Based on all these findings, the ratio of apo B/apo A-I, i.e. the balance between potentially atherogenic cholesterol-rich apo B-containing particles and the anti-atherogenic apo A-I-rich particles, is proposed as the best integrated measure of cardiac risk related to lipoprotein profiles. Furthermore, the apo B/apo A-I ratio has also been shown to be superior to the LDL-C/HDL-C and TC/HDL-C ratios in predicting cardiac risk [21, 100]. The apo B/apo A-I ratio is preferred as the apolipoprotein determinations are independent measures compared with the conventionally used lipid ratios that are based on calculated numbers [11]. The risk of fatal MI in subjects with LDL-C <3 mmol L−1 (115 mg dL−1) was related to elevated TG values, and especially to a high apo B/apo A-I ratio, but not to LDL-C values (Fig. 5). Again, this strengthens the use of the apo B/apo A-I balance as an integrated risk ratio irrespective of the levels of TC and LDL-C.

Figure 3.

Risk of fatal acute myocardial infarction (MI) in the AMORIS study in relation to increasing apo B and decreasing apo A-I expressed as acute MI deaths/1000 observation years in (a) males >30 years (n = 1267) and (b) females >30 years (n = 586).

Figure 4.

Risk of fatal acute myocardial infarction (MI) in the AMORIS study in relation to increasing apo B and decreasing apo A-I expressed as risk ratios adjusted for TG, TC, and log transformed for age in (a) males >30 years (n = 1267) and (b) females >30 years (n = 586). The degree of significance in relation to the cell corresponding to the highest apo A-I and lowest apo B = 1.0. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Figure 5.

Risk of fatal acute myocardial infarction (MI) in the AMORIS study in males and females above 40 years of age (n = 198), with LDL-C values below 3.0 mmol L−1. The risk ratio in relation to tertile values for triglycerides, apo B/apo A-I, and LDL-C is shown and adjusted for fasting and log transformed for age by tertiles. NS > 0.05, ****P < 0.0001 (tested versus reference cell).

Moss et al. [101] analysed the predictive role of multiple thrombotic/fibrinolytic risk variables as well as several lipoprotein-related variables in a 4-year prospective study of 1045 post-MI patients, and reported that apo B was a stronger predictor of cardiac risk than LDL-C. In multivariate analyses, apo B as well as apo A-I and D-dimer remained significant predictors of cardiac risk.

Talmud et al. [10] have recently confirmed the results from the AMORIS study. These investigators followed 2508 middle-aged men, free of coronary disease at baseline, for 6 years. The study found that apo B was a better predictor of cardiac risk than LDL-C, and that both apo B with TG and the apo B/apo A-I ratio had the strongest associations with CHD. The effectiveness of the Jungner formula [11, 100] to estimate LDL-C and HDL-C for CHD risk prediction was also confirmed [10].

Prediction of risk in patients on lipid-lowering therapy using apolipoprotein B and apolipoprotein A-I

When lipoprotein levels are modified by drug therapy, they may lose their predictive power. However, the predictive value of apo B and apo A-I is maintained in patients on lipid-modifying therapy. In the Air Force/Texas Coronary Atherosclerosis Preventions Study (AFCAPS/TexCAPS), a primary prevention trial using lovastatin in patients with mild-to-moderate elevations in TC and LDL-C and below-average HDL-C, baseline levels of apo B, apo A-I and the apo B/apo A-I ratio were significantly correlated with the risk of an acute major coronary event; in contrast, levels of LDL-C or TC were not (Table 3) [102]. After 1 year of lipid-lowering treatment, all the lipid parameters had lost their predictive power, whilst apo B, apo A-I and the apo B/apo A-I ratio were still significantly associated with coronary risk (Table 3) [102]. Furthermore, the apo B/apo A-I ratio had the strongest relation with coronary risk both in placebo and statin-treated groups [102].

Table 3.  Association between lipid parameters and risk of acute major coronary event in the AFCAPS/TexCAPS study (adapted from Ref. [102] with kind permission from the publisher)
Lipid parameterAssociationa between lipid parameters and risk of acute major coronary event (P-value)
Value at baselineValue at 1 year
  1. aP-values from logistic regression adjusted for age, sex, marital status, hypertension, smoking and family history of CHD. Treatment group was included as a covariate. bInteraction between treatment group and apo A-I was significant (P = 0.043). Apo, apolipoprotein; TC, total cholesterol; HDL-C, HDL cholesterol; TG, triglyceride; LDL-C, LDL cholesterol; NS, not statistically significant.

Apo A-I0.008b0.013
Apo B0.002<0.001
Apo B/apo A-I<0.001<0.001

In the Scandinavian Simvastatin Survival Study (4S), both apo B and LDL-C predicted CHD risk for patients in the placebo group with similar statistical strengths [103]. In another secondary prevention study, on-treatment levels of TC, LDL-C and TG did not predict cardiovascular events for 848 patients with angiographically proven CAD responding to statin treatment with at least a 30% reduction in TC [104]. In contrast, on-treatment levels of apo A-I were predictive of future CAD risk in both men and women, and on-treatment apo B levels were predictive of risk when the study population was analysed as a whole [104].

Hence, in patients receiving lipid-modifying therapy, it appears that determination of subsequent cardiovascular risk is likely to be more accurate if based on assessment of apo B and apo A-I levels rather than LDL-C. As apo B summarizes atherogenic lipoproteins, especially the number of small dense lipoprotein particles, and apo A-I represents the atheroprotective capacity, the two apolipoproteins seem to integrate the whole lipoprotein balance and the changes occurring during lipid-lowering treatments. Again this supports the adoption of the apo B/apo A-I ratio, and confirms its use for predicting CHD risk and forecasting treatment success [11]. Sniderman et al. [21] reached similar conclusions in a review of the evidence for apolipoproteins versus lipids as indices of cardiac risk and as targets for statin treatment. Based on observational and statin treatment studies, nine comparisons in all, the authors concluded that apo B was stronger than LDL-C, and that the apo B/apo A-I ratio was the strongest predictor of cardiac risk. Therefore, these apolipoproteins should be used as targets for lipid-lowering therapy.

Normal values for apolipoprotein B and apolipoprotein A-I and target levels for treatment

Before using apo B, apo A-I, and the apo B/apo A-I ratio for predicting cardiac risk or as targets of therapy, it is important to determine what are considered ‘normal’ values. Data from several different populations have been published [86, 105–108]. The Canadian Cardiovascular Society has developed guidelines for how to evaluate and treat patients based on their apolipoprotein B value [109], and it has also been suggested that measurement of apo B would simplify follow-up of patients with dyslipidaemia [110]. The use of apo B and apo A-I as useful risk markers was highlighted in the recently published joint Swedish and Norwegian guidelines [111, 112], although the main focus still remains on LDL-C and HDL-C levels [111].

The methods for measuring apo B and apo A-I are now internationally standardized by the World Health Organization–International Federation of Clinical Chemistry (WHO-IFCC) [83, 84]. Values of apo B >1.2 g L−1 [107], and for apo A-I <1.2 g L−1 [108], have been proposed as cut values. Recently, Grundy has recommended ‘target’ levels for apo B and related these values to the absolute cardiac risk determined according to the NCEP ATP III guidelines [25]. Results from the AMORIS study [11] would tend to confirm the suitability of the values proposed by Grundy for apo B levels. Thus, in those at greatest risk, a target apo B level below 0.9 g L−1 is recommended regardless of gender. However, as males in general have lower apo A-I values than females [86], it may be more relevant to have different cut values for males (<1.15 g L−1) and females (<1.25 g L−1). Regarding cut values for the apo B/apo A-I ratio, <0.9 and <0.8 may be used to define a risk level for males and females, respectively. It should be noted that new guidelines have to be developed and agreed before a more widespread use of apo B and apo A-I can be implemented in general clinical practice [11, 113, 114]. However, the apo B/apo A-I ratio retains its predictive power for patients receiving treatment and therefore may represent the best indicator of the efficacy of lipid-lowering therapy.

Apolipoproteins and current risk assessment

Neither the updated Joint European guidelines [115] nor NCEP ATP III [19] use apo B or apo A-I as targets of therapy. Although apo A-I has been shown to be a better predictor of CHD than HDL-C [39, 40], the NCEP ATP III guidelines indicate that its independent predictive power beyond that of HDL-C is uncertain [19]. There is, however, more evidence to support the superiority of apo B over LDL-C levels for risk prediction. Indeed, apo B was reported as a better predictor of CHD events compared with LDL-C in over 180 000 patients assessed in the Northwick Park Heart Study [10], AMORIS [11] and the Quebec Cardiovascular Health Study [12]. Furthermore, unlike LDL-C, apo B levels remain predictive of CHD events in patients receiving treatment [102, 104]. The NCEP ATP III guidelines have suggested that the evidence to support the adoption of apo B goals for treatment is insufficient [19]. However, at the time the NCEP ATP III guidelines were developed, much of the new data supporting the use of apo B had not been published. Instead, the NCEP ATP III guidelines recommend non-HDL-C levels as a surrogate for apo B-containing particles when TG levels are elevated [19]. However, it has been suggested that non-HDL-C is a poor surrogate for apo B [21]. Although there is a strong positive correlation between non-HDL-C and apo B, there is a great dispersion around the line of identity, especially in the middle to low part of the range for non-HDL-C [116]. The Chair of the NCEP ATP III guideline committee indicated, in an editorial commenting on the use of non-HDL-C targets [25], that apo B may well be a better marker of risk than both non-HDL-C and LDL-C and he provided target values for all three variables. Finally, it has also been shown that apo B/apo A-I was superior to non-HDL-C/HDL-C for predicting CHD risk based on the AMORIS results [21].

The Systematic COronary Risk Evaluation (SCORE and SCORECARD) system now implemented by the Joint European guidelines uses conventional risk factors, such as LDL-C and HDL-C [115, 117]. However, it may be possible to incorporate new risk factors, such as apolipoproteins, in the interactive, electronic version of SCORE, known as SCORECARD. Interestingly, TC/HDL-C is used in these guidelines, which indicates that lipid ratios are considered of value. As apo B/apo A-I is a better predictor of risk than TC/HDL-C [21] it would be logical, and should be possible, to replace TC/HDL-C with apo B/apo A-I as an indicator of abnormal lipoprotein balance.

The NCEP ATP III guidelines indicate that standardized methodologies for measuring apo B and apo A-I are not yet widely available or appropriately standardized for clinical use [19], and therefore this could limit the use of apolipoproteins as a target or in risk-assessment algorithms. However, it should be noted that apo B and apo A-I are internationally standardized according to WHO-IFCC [83, 84], which is not the case for current methods of either calculating LDL-C according to the Friedewald formula [79] or for methods used for the direct determination of LDL-C levels [26, 118, 119].

Improving the apolipoprotein B and apolipoprotein A-I profile with lipid-modifying drugs

Previous work on the effects of lipid-modifying drugs has concentrated on LDL-C, TG and HDL-C. In light of recent studies showing that apolipoproteins are closely involved in atherogenesis, there is a need to explore the effects of lipid-modifying therapy on the apolipoprotein profile. Measures, such as the apo B/apo A-I ratio, may be a particularly useful way of comparing treatments, as this parameter can integrate the opposing effects of lipid-modifying drugs on pro- and anti-atherogenic lipoproteins.


The primary action of statins in reducing blood cholesterol is the inhibition of HMG CoA reductase, which reduces hepatocyte cholesterol concentration and stimulates hepatic uptake of LDL-C from the circulation [120]. However, evidence is growing that statins can also exert pleiotropic effects [121], reducing C-reactive protein independently of LDL-C reductions [122].

Statins can alter the synthesis and metabolism of lipoproteins and, in cultured hepatoblastoma cells, atorvastatin reduced expression of microsomal TG transfer protein and accelerated intracellular apo B degradation, resulting in a reduction in apo B secretion [123]. Studies in small series of patients have shown that lovastatin can reduce the production of apo B-LDL [124] and apo B-VLDL [125]. Furthermore, in patients with type II diabetes, statin treatment decreased the postprandial production of large apo B-containing chylomicron particles, which are thought to be particularly atherogenic [126]. Apo A-I metabolism may also be influenced by statin therapy, as cultured hepatoblastoma cells display increased expression of apo A-I in the presence of statins [127].

Decreases in apo B and increases in apo A-I during statin therapy have been reported in large clinical trials. The Atorvastatin Comparative Cholesterol Efficacy and Safety Study (ACCESS) in patients with hypercholesterolaemia reported reductions in plasma apo B levels following statin therapy [18]. In the ACCESS study, the most efficacious statin used for lowering LDL-C levels, atorvastatin, also produced the largest reductions in apo B compared with fluvastatin, lovastatin, pravastatin and simvastatin [18] (Table 4). Although atorvastatin reduced LDL-C levels from 178 to 102 mg dL−1 [i.e. to below current target levels of 3.0 mmol L−1 (115 mg dL−1)], apo B fell from 1.70 to 1.14 g L−1. These apo B levels would still be considered high according to AMORIS [11] and compared with targets recommended for high-risk patients [25]. These results may therefore indicate that patients could be undertreated if LDL-C is used as the marker of risk [21]. However, it is known that there may be greater errors in LDL-C values when levels are in the normal to low range [80–82, 116], and this would support the use of apo B, which has a coefficient of variation (CV) below 3–5%. Insull et al. [128] also reported a greater reduction in apo B levels with atorvastatin compared with simvastatin in patients with mixed dyslipidaemia, which is consistent with atorvastatin's greater effect on LDL-C. In the Treat To Target (3T) study, both atorvastatin and simvastatin were associated with increases in apo A-I and HDL-C [129]. These effects, but not changes in LDL-C or TG, were significantly correlated with changes in C-reactive protein, suggesting that changes in apo A-I and/or HDL may be associated with the anti-inflammatory properties of statins.

Table 4.  Comparison of various statins for modifying LDL-C, non-HDL-C and apo B levels (adapted from Ref. [18])
Baseline (mmol L−1)Reduction at 54 weeks (%)Baseline (mmol L−1)Reduction at 54 weeks (%)Baseline (mmol L−1)Reduction at 54 weeks (%)
  1. Mean baseline levels and percentage reductions after 54 weeks of treatment are shown. Apo, apolipoprotein; non-HDL-C, non-HDL cholesterol; LDL-C, LDL cholesterol.


Although atorvastatin has been shown to be superior to lovastatin, pravastatin and simvastatin in terms of apo B reduction, it has been reported to have a lesser effect on apo A-I than simvastatin in patients with hypercholesterolaemia [130]. The effect on apo A-I was smaller with the high dose of atorvastatin than with the lower dose, and a similar pattern was seen with HDL-C [130], indicating that different statins may have differential effects on the various apolipoproteins. In patients with low HDL-C levels, fenofibrate given 200 mg day−1 increased HDL-C and apo A-I levels significantly more than atorvastatin 10 mg day−1 [131].

The newer agent, rosuvastatin, has demonstrated superior efficacy over existing statins in reducing LDL-C [132, 133], and has also been reported to reduce apo B levels significantly more than atorvastatin, simvastatin and pravastatin [132–135]. This provides further support for an association between greater apo B reduction and greater LDL-C reduction. Rosuvastatin demonstrated superior increases in HDL-C compared with atorvastatin [132], and increased apo A-I levels in patients with primary hypercholesterolaemia [136] and heterozygous familial hypercholesterolaemia [133, 137]. Recently published studies have also shown that rosuvastatin produced significantly greater reductions in the apo B/apo A-I ratio compared with atorvastatin [133, 136, 138], simvastatin and pravastatin (Fig. 6) [133, 138, 139]. Given that changes in HDL-C, and therefore apo A-I, do not always appear to be related to dose, or occur in parallel to those produced by rosuvastatin [140, 141], increasing doses of less efficacious statins may not result in a similar reduction in the apo B/apo A-I ratio. Indeed, both HDL-C and apo A-I tend to fall with increasing doses of atorvastatin [140, 141]. Furthermore, increasing statin doses could also raise the potential for adverse events. Therefore, these results indicate that rosuvastatin, at starting and higher doses, has great potential for correcting the abnormal balance between atherogenic and anti-atherogenic lipoproteins that are closely related to cardiovascular risk, including fatal MI.

Figure 6.

Percentage change in apo B/apo A-I ratio after 12 weeks treatment with rosuvastatin, atorvastatin, pravastatin and simvastatin (adapted from Rader et al. [138]).


Bezafibrate has been demonstrated to reduce apo B levels by 9% in patients who had survived a previous MI [142]. In this study, on-trial apo B and HDL-3 cholesterol values were independent predictors of different estimates of coronary atherosclerosis, whereas small dense LDL and/or VLDL lipids were unrelated to disease progression. Gemfibrozil and bezafibrate have been reported to have no effect on apo B levels in a study of 29 patients with type IIb hyperlipoproteinaemia [143]. This finding is consistent with the clinical properties of fibrates, which predominantly affect TG and have limited efficacy in lowering LDL-C. Fibrates also increase HDL-C, and might therefore be expected to influence apo A-I levels; indeed, fibrates have been shown to increase apo A-I levels by induction of the apo A-I gene in transgenic rabbits [144] and in cultured human hepatocytes [145]. However, in clinical studies, fibrates seem to vary in their effects on apo A-I; in one study bezafibrate increased apo A-I by 8% whilst gemfibrozil had no effect [143]. Micronized fenofibrate has shown a 17% lowering of apo B and a 10% increase in apo A-I in patients with low HDL-C at baseline [131]. In patients with type 2 diabetes and dyslipidaemia, fenofibrate also improved endothelial function, and multiple stepwise regression analysis indicated that the on-trial concentration of apo B was the only significant predictor of improvement in flow-mediated vasodilation (r = −0.56; P < 0.0007) [146]. Fibrates may also increase apo A-II and decrease apo C-III via their agonistic action on peroxisome proliferator-activated receptors, and further information on their effects is provided in an important overview of the use of fibrates for reducing CHD in diabetic patients [147].

Niacin (nicotinic acid)

Studies in cultured hepatoblastoma cells have shown that niacin accelerates hepatic intracellular post-translational degradation of apo B by selectively reducing TG synthesis [148], and that niacin also selectively decreases hepatic removal of HDL A-I [149]. These two effects would be expected to lead to decreased plasma levels of apo B-containing lipoproteins, and increased plasma levels of HDL A-I. Such effects have been observed in clinical studies of extended-release niacin (niacin-ER), with reported apo B reductions of 12, 20, 15 and 19%, after 16 [63], 25 [150], 48 [151] and 96 weeks of treatment [152], respectively. Niacin-ER increases apo A-I levels in a dose-dependent manner [63, 153], and selectively increases HDL A-I particles [154]. Compared with gemfibrozil, niacin-ER is more effective in increasing apo A-I [153, 155] and decreasing Lp(a) [155], but there is no statistically significant difference between these agents for their effect on apo B [155].


Cholestyramine has been shown to reduce the levels of apo B and to increase the levels of apo A-I either as monotherapy or in combination with statins [156]. In this study, the changes in apo B and apo A-I levels correlated well with the changes in LDL-C and HDL-C, respectively.

Combination therapy with statins

Statins and other lipid-lowering drugs influence lipoprotein metabolism by different mechanisms, suggesting that combination therapy may have the potential to result in greater improvements in the apolipoprotein profile compared with monotherapy. The effects of combination treatment with lovastatin or niacin plus colestipol were compared with conventional therapy (placebo in patients without elevated LDL-C at baseline, or colestipol alone in patients whose baseline LDL-C exceeded the 90th percentile for their age) in a double-blind study of men with high baseline levels of apo B and documented CAD [157]. Addition of niacin or lovastatin to usual therapy resulted in significant increases in apo A-I and further reductions in apo B levels (Table 5). Colestipol plus either lovastatin or niacin reduced the frequency of progression of coronary lesions, increased the frequency of regression, and reduced the incidence of cardiovascular events compared with the control group. Moreover, changes in plasma apo B levels correlated well with the degree of CAD regression.

Table 5.  Effects of combination therapy on apolipoprotein levels
Study treatmentPercentage change from baseline
Apo BApo A-IApo A-II
  1. aThe control group received either placebo (if they did not have elevated LDL-C at baseline) or colestipol alone (if their LDL-C exceeded the 90th percentile for their age). NS, not statistically significant. *P < 0.05; **P < 0.01; ***P < 0.001.

Brown et al. [134]
 Control groupa−5***NSNS
 Colestipol plus lovastatin−35***+7**NS
 Colestipol plus niacin−28***+14**−6*
Guyton et al. [151]
 Niacin-ER monotherapy−15**Not reportedNot reported
 Niacin-ER plus statin−26**Not reportedNot reported

In an open label extension study of niacin-ER, some patients also received concomitant therapy with statins, resulting in an additional 11% reduction in apo B levels [151]. A combination of niacin-ER and lovastatin has shown equal apo B-lowering effects compared with atorvastatin 10 mg, but a greater ability to increase apo A-I (14% vs. 2%, respectively) [158]. Pravastatin in combination with cholestyramine has also been shown to potentiate reductions in apo B and increases in apo A-I, in parallel with changes in LDL-C and HDL-C, respectively [156].

However, combination therapy may be associated with an increased incidence of adverse events, including liver and muscle toxicity [159–161]. Although combination therapy may be a valuable option in those cases where greater efficacy is required than can be achieved with one drug alone, it should be used with due consideration for the increased risk of side effects. Statin monotherapy is generally well tolerated except for the severe complications induced by cerivastatin used at high doses or in combination with gemfibrozil, which resulted in its withdrawal [162]. Indeed, recent clinical trials suggest that the benefits of enhanced efficacy across the lipid profile outweigh the risk of toxicity when consideration is given to the choice of agents to be combined and doses are kept as low as possible [163]. As statins have also been shown to be effective in improving the apolipoprotein profile, they will most likely be the basic treatment upon which other lipid-lowering or other cardioprotective agents are added.


Apolipoprotein metabolism is closely associated with the development of atherosclerosis, and the roles played by the various apolipoprotein types are beginning to be delineated. Elevated levels of apo B and/or low levels of apo A-I have consistently been associated with an elevated risk of cardiovascular events in clinical studies, and baseline apo B level has been demonstrated to be a better predictor of cardiovascular risk than LDL-C. Apo B and apo A-I also retain their predictive power in patients receiving lipid-modifying therapy. As apo B and apo A-I appear to have opposing effects on atherogenic risk, the ratio between the two, indicating the balance between potentially atherogenic versus atheroprotective cholesterol-rich particles, may be a more useful measure of risk than either parameter alone. Apolipoprotein levels are also straightforward to measure, do not require fasting blood samples, and the analyses are standardized and are easily automated [11, 77, 86]. Apo B and apo A-I may thus have methodological advantages over LDL-C and HDL-C as measures of cardiovascular risk, in addition to their greater predictive power. It has been suggested that measurement of apo B and apo A-I could significantly improve the assessment of cardiovascular risk, especially in patients without elevated LDL-C, and these markers should be included in revisions to the international guidelines for lipid-modifying treatment [11]. Thus, instead of having to measure TC, TG, LDL-C, HDL-C, non-HDL-C and lipid ratios as recommended by the NCEP ATP III guidelines [19], it may suffice to measure apo B/apo A-I, and possibly also TG, to effectively evaluate cardiac risk and to monitor the effects of lipid-lowering therapy.

The apolipoprotein profile is clearly a valuable indicator of cardiovascular risk, and provides information that may be used to guide the treatment of individual patients in clinical practice. Treatment regimens should be assessed according to their effects on apolipoproteins, as well as on the established lipoprotein measures. New guidelines should be developed including apo B and apo A-I as important predictors of cardiac risk and as markers of lipid-lowering therapy.

Conflict of interest statement

Göran Walldius is an adjunct Professor at Karolinska Institute, Stockholm, Sweden. He is employed by AstraZeneca, Mölndal, Sweden. He has no other relations with any other pharmaceutical industries.

Ingmar Jungner, via ‘Gunnar och Ingmar Jungners Stiftelse för Laboratoriemedicin’, has received research grants from Bure Hälsa & Sjukvård Ltd, Stockholm, AstraZeneca, Mölndal, Uppsala University, Uppsala, Sweden, and from McGill University, Montreal, Canada.

There are no other conflicts of interests for Göran Walldius or Ingmar Jungner.