We conducted a study to determine whether high lipoprotein(a) levels predicted the risk of myocardial infarction (MI) and ischemic heart disease. Unlike other studies, we measured lipoprotein(a) levels shortly after sampling and corrected for regression dilution bias.
Despite the widespread use of traditional cardiovascular risk assessment to guide treatment interventions, myocardial infarction (MI) and ischemic heart disease (IHD) remain leading causes of morbidity and mortality in affluent societies. Increased lipoprotein(a) levels may explain part of the residual risk; thus, these values may be useful in identifying additional high-risk individuals.1
Lipoprotein(a) is made up of a low-density lipoprotein (LDL) cholesterol particle attached to apolipoprotein(a), which is a plasminogen-like glycoprotein.1 Several studies have shown that lipoprotein(a) may be a factor in the progression of thrombosis, atherosclerosis, or both, leading to IHD and MI.1-7 Despite these findings, lipoprotein(a) has not been used as a marker for increased risk of IHD and MI in clinical practice; this is partly because of contradictory findings in these previous studies. Furthermore, a lack of correction for regression dilution bias and measurements in long-term frozen samples have been limitations in earlier studies.8 There are also no absolute risk estimates available for the general population, which would assist in the use of lipoprotein(a) levels for risk assessment. In the Copenhagen City Heart Study, we evaluated whether the risk of MI could be predicted by excessively high levels of lipoprotein(a), correcting for regression dilution bias, and measuring lipoprotein(a) levels soon after sampling.
Subjects and methods
We included a total of 9330 individuals from the Danish general population who were examined between 1991 and 1994 and had no history of IHD. The end points of the study were the occurrence of IHD (including MI), death, or the beginning of 2004, whichever came first. Subjects who experienced an MI or symptoms of angina pectoris were considered to have IHD. At least 2 of the following criteria were required for a diagnosis of MI: electrocardiographic changes suggestive of MI, increased cardiac enzymes, and chest pain typical of MI. Follow-up was 100% complete.
At the start of the study, we used colorimetric and turbidimetric assays to assess apolipoprotein B, total cholesterol, LDL cholesterol, high-density lipoprotein cholesterol, and triglyceride levels. If the triglyceride level was <4 mmol/L, we used the Friedewald formula to determine LDL cholesterol, and if the level was >4 mmol/L, a direct measurement was used. An established in-house turbidimetric test was used to assess lipoprotein(a) levels. A sensitive immunoturbidimetric commercial assay, was used to measure lipoprotein(a) again in 4609 subjects from 2001 to 2003, which allowed for the correction of regression dilution bias.8 Hazard ratios (HRs) were adjusted for regression dilution bias. Absolute 10-year risks were estimated for men and women separately using the regression coefficients from a Poisson regression model.
Figure 1. Relative risk of myocardial infarction and ischemic heart disease by levels of lipoprotein(a) in the
general population. Hazard ratios were adjusted for age or multifactorially for age, total cholesterol, low-
density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, apolipoprotein B, body
mass index, hypertension, diabetes mellitus, smoking, lipid-lowering therapy, and, for women, also
menopause and hormone replacement therapy. Hazard ratios were also adjusted for regression dilution
bias. P values are test for trend of hazard ratios.
A total of 1142 subjects with IHD were included in the analysis, 498 of whom experienced an MI. The median follow-up period was 10 years. As shown in Figure 1, the higher the lipoprotein(a) level, the greater the risk of IHD and MI in both sexes (trend tests ranged from <.001 to = .02). Multifactorially adjusted HRs for MI for increased lipoprotein(a) levels in women were 1.1 (95% confidence interval [CI], 0.6-1.9) for 5 to 29 mg/dL (22nd-66th percentile); 1.7 (95% CI, 1.0-3.1) for 30 to 84 mg/dL (67th-89th percentile); 2.6 (95% CI, 1.2-5.9) for 85 to 119 mg/dL (90th-95th percentile); and 3.6 (95% CI, 1.7-7.7) for >120 mg/dL (>95th percentile) versus levels <5 mg/dL (<22nd percentile). For men, the HRs were as follows: 1.5 (95% CI, 0.9-2.3); 1.6 (95% CI, 1.0-2.6); 2.6 (95% CI, 1.2-5.5); and 3.7 (95% CI, 1.7-8.0). The results for IHD were similar, but attenuated (Figure 1). No significant interactions were observed between lipoprotein(a) and any covariates, including, age and sex, on risk of MI and IHD.
Figure 2. Absolute 10-year risk of myocardial infarction by lipoprotein(a) levels, sex,
smoking, hypertension, and age.
The absolute 10-year risk of MI and IHD increased with increasing lipoprotein(a) levels in both sexes and also with increasing age and a history of hypertension or current smoking (Figures 2 and 3). In women, the highest absolute 10-year risks of MI of 10% and 20% were found in hypertensive smokers aged 60 years and older and with lipoprotein(a) levels <5 mg/dL (<22nd percentile) and ≥120 mg/dL (>95th percentile), respectively. Equivalent values for men were 19% and 35%. Although the absolute increase in risk from low to extreme levels of lipoprotein(a) was slightly attenuated compared with that observed for MI, the absolute 10-year risks of IHD were higher than those for MI.
Figure 3. Absolute 10-year risk of ischemic heart disease by lipoprotein(a) levels, sex,
smoking, hypertension, and age.
Lipoprotein(a) levels above the 95th percentile were shown to increase the risk of MI 3- to 4-fold in subjects enrolled in the current study, with absolute 10-year risks of 35% and 20% in high-risk men and women, respectively. Lipoprotein(a) is a liver-derived circulating particle composed of an LDL cholesterol particle bound to a plasminogen-like apolipoprotein(a) particle.1 It has been associated with the development of atherogenesis and shown to interfere with fibrinolysis; thus, it may be a factor in the development of IHD and MI.
A positive correlation between the risk of IHD and high lipoprotein(a) levels has been shown in most prospective studies.1-7,9,10 The risk estimates in earlier studies2-4,6,7,9,10 were not as high as those shown in our study, probably because we corrected for regression dilution bias, focused on MI rather than on IHD, concentrated on extremely high lipoprotein(a) levels, and measured lipoprotein(a) levels soon after sampling. A risk ratio of 1.7 was shown for IHD in a meta-analysis of 18 studies that compared subjects in the upper and lower tertiles of the lipoprotein(a) distribution.3 The risk estimates in our study would not be significantly different if we considered the risk of IHD as a function of upper versus lower lipoprotein(a) tertile.
Whereas our study showed a 2.4-fold increased risk of IHD in men and women with increased lipoprotein(a) levels (Figure 1), 1 prior study showed a 1.9-fold increased risk of IHD among women with lipoprotein(a) levels above the 95th percentile.7 In that study, correction was not made for regression dilution bias, and long-term frozen samples were used, which increased the possibility of differential degradation of lipoprotein(a) isoforms, resulting in lower risk estimates.11 A 2.3-fold increased risk of cardiovascular death was shown among men with lipoprotein(a) levels above the 90th percentile in another study, and correction was made for regression dilution bias, although long-term frozen samples were used.12 These results are similar to the findings in our study.
An interaction between lipoprotein(a) and LDL cholesterol levels has been shown to have an effect on the risk of IHD.4,5,7 These results were not borne out in our study, however, as testing for interaction was not significant, and categorizing the subjects by the median LDL cholesterol level showed similar results in both strata. Although some studies have suggested a threshold effect,5,7 our findings showed a stepwise increase in risk of MI and IHD with increasing levels of lipoprotein(a) in both sexes, with no evidence of a threshold effect.
One of the limitations of our study was that only a white population was evaluated, and distributions of lipoprotein(a) are known to differ among ethnic groups. As a result, our findings may not be applicable to other ethnic groups. Misclassification of a few cases or controls cannot be ruled out. The slightly attenuated risk estimates observed for IHD compared with those for MI could be the result of misclassification simply because MI is a harder end point that is less subjectively assessed than a diagnosis of angina pectoris and IHD. It is unlikely that selection bias affected our results because participants were randomly selected from the general population and follow-up was 100% complete.
Lipoprotein(a) is notoriously difficult to measure because of a large number of different-sized isoforms of the apolipoprotein(a) moiety.13 Whether our measurements were affected by apolipoprotein(a) isoform size is not known. If measurements are affected by apolipoprotein(a) isoform size, however, concentrations of large isoforms are overestimated, whereas concentrations of small isoforms are underestimated.13 In light of the inverse association between apolipoprotein(a) isoform size and plasma lipoprotein(a) concentration, this would tend to result in underestimation of high concentrations and overestimation of low concentrations, decreasing risk estimates. The true risk of MI and IHD resulting from extremely high lipoprotein(a) levels, therefore, may be even greater than those seen in our study. The lipoprotein(a) assay that we used is not standardized internationally for accuracy, but this would not change the risk estimates based on percentiles, especially because the same technician measured all of the samples in our study using only 1 autoanalyzer, 1 calibration, and 1 batch of antibodies.
The findings of our study show that the risk of MI and IHD is increased in European men and women with high levels of lipoprotein(a). We also report absolute 10-year risk estimates of MI and IHD, which may be used to advise patients, provided that the percentile cut points for the lipoprotein(a) distribution for a given lipoprotein(a) assay are known.
The use of niacin may decrease lipoprotein(a) levels,14 although a reduction in the incidence of IHD resulting from niacin treatment has not been shown in randomized studies. Furthermore, niacin may not be well tolerated by some individuals, and it affects other lipoprotein levels as well. Lipoprotein(a) levels, however, may be used to identify high-risk patients who would benefit from other aggressive preventive measures, such as the use of statins to lower elevated cholesterol levels. A significant decrease in cholesterol levels in patients with hypercholesterolemia has been shown to reduce the ability of lipoprotein(a) to cause IHD.15
Our study assessed whether extremely high lipoprotein(a) levels could predict MI in the general population. We corrected for regression dilution bias and used lipoprotein(a) measurements soon after sampling. Results showed a stepwise increase in the risk of MI with increasing lipoprotein(a) levels, and levels above the 95th percentile predicted a 3- to 4-fold increase in the risk of MI in the general population and conferred an absolute 10-year risk of 20% and 35% in high-risk women and men, respectively.
DISCLOSURE The study was supported by the Danish Heart Foundation and the Ib Mogens Kristiansen (IMK) fund. There are no other financial relationships that might represent a conflict of interest with the content of this article.