Low-Density Lipoprotein Particle Concentration and Size as Determined by Nuclear Magnetic Resonance Spectroscopy as Predictors of Cardiovascular Disease in Women
Background— Nuclear magnetic resonance (NMR) offers an alternative, spectroscopic means of quantifying LDL and of measuring LDL particle size.
Methods and Results— We conducted a prospective nested case-control study among healthy middle-aged women to assess LDL particle size (NMR) and concentration (NMR) as risk factors for future myocardial infarction, stroke, or death of coronary heart disease. Median baseline levels of LDL particle concentration (NMR) were higher (1597 vs 1404 nmol/L; P= 0.0001) and LDL particle size (NMR) was lower (21.5 vs 21.8 nm; P=0.046) among women who subsequently had cardiovascular events (n=130) than among those who did not (n= 130). Of these 2 factors, LDL particle concentration (NMR) was the stronger predictor (relative risk for the highest compared with the lowest quartile=4.17, 95% CI 1.96–8.87). This compared with a relative risk of 3.11 (95% CI 1.55–6.26) for the ratio of total cholesterol to HDL cholesterol and a relative risk of 5.91 (95% CI 2.65–13.15) for C-reactive protein. The areas under the receiver operating characteristic curves for LDL particle concentration (NMR), total cholesterol to HDL cholesterol ratio, and C-reactive protein were 0.64, 0.64, and 0.66, respectively. LDL particle concentration (NMR) correlated with several traditionally assessed lipid and nonlipid risk factors, and thus adjustment for these tended to attenuate the magnitude of association between LDL particle concentration (NMR) and risk.
Conclusions— In this cohort, LDL particle concentration measured by NMR spectroscopy was a predictor of future cardiovascular risk. However, the magnitude of predictive value of LDL particle concentration (NMR) was not substantively different from that of the total cholesterol to HDL cholesterol ratio and was less than that of C-reactive protein.
Received May 20, 2002; revision received July 18, 2002; accepted July 19, 2002.
Chemical lipid measurement is routinely performed for cardiovascular risk assessment. Traditional methods for lipid analysis measure the cholesterol composition of LDL and HDL, whereas apolipoprotein B-100 has been used as a measure of the number, or concentration, of LDL particles.
Nuclear magnetic resonance (NMR) spectroscopy provides an alternative means of measuring lipoprotein levels in plasma,1,2⇓ with quantification based not on cholesterol or apolipoprotein content but on the detected amplitudes of spectral signals emitted by lipoprotein subclasses of different size. Because the signal amplitudes are not affected by differences in chemical composition, they are believed to give a direct indication of subclass particle concentrations. By acquiring plasma NMR spectra in a prescribed, automated manner and decomposing the data computationally, quantitative measures of LDL particle concentration and size can be obtained.3 Because the measurement methodology and basis of LDL quantification differ from what has been used in the past, it is unclear whether NMR spectroscopy offers any advantages over conventional measures of LDL in assessing cardiovascular disease risk.
To date, however, studies of NMR spectroscopy and cardiovascular risk have been scant and limited by their cross-sectional design.4 We therefore sought to determine whether baseline measures of LDL particle concentration (NMR) and size (NMR) could predict the future onset of cardiovascular disease. We also sought to compare the magnitude of predictive value associated with NMR with that of traditional chemical lipid measures and with C-reactive protein (CRP), an inflammatory biomarker also of use for global vascular risk prediction.5
The Women’s Health Study (WHS) is an ongoing randomized, double-blind, placebo-controlled trial of aspirin and vitamin E being conducted among middle-aged women with no history of cardiovascular disease or cancer.6 Blood samples were collected from 28 263 women at baseline in tubes containing EDTA and stored in liquid nitrogen until the time of analysis. Questionnaires were sent to WHS participants to elicit information on cardiovascular risk factors and incident cardiovascular events. For this analysis, case subjects were study participants from whom a baseline blood sample was obtained who subsequently had a cardiovascular event, as defined by death due to coronary heart disease, nonfatal myocardial infarction, or stroke. The mean follow-up period was 3 years.
For all cases of myocardial infarction, stroke, or death due to coronary heart disease, hospital records were obtained and reviewed. Myocardial infarction was classified as confirmed if symptoms met the criteria of the World Health Organization and if the event was associated with abnormal levels of cardiac enzymes or diagnostic ECG changes. Reported stroke was confirmed if the patient had a new neurological event that persisted for more than 24 hours; CT scans or MR images were available for the majority of women with stroke. Death due to coronary heart disease was confirmed by review of autopsy reports, death certificates, medical records, and circumstances of death.
For each woman with a confirmed cardiovascular event during follow-up, 1 control subject matched for age (within 1 year) and smoking status (current, former, or never) was selected from among the remaining study participants from whom a baseline blood sample had been obtained and who remained free of reported cardiovascular events during follow-up. With the use of these criteria, 130 cases and 130 controls were selected. The cases comprised 65 women who had a nonfatal myocardial infarction, 53 women who had a stroke, and 12 women who died of coronary heart disease.
Baseline plasma samples were thawed, and assays for total cholesterol, HDL cholesterol (HDL-C), triglycerides, and direct LDL cholesterol levels were performed on a Hitachi 911 analyzer (Roche Diagnostics) with reagents from Roche Diagnostics and Genzyme. Plasma levels of CRP were measured with a latex-enhanced immunonephelometric assay on a BN II analyzer (Dade Behring).7 Apolipoprotein B-100 was also measured with this device by immunoassay.
Lipoprotein subclass profiles were measured by proton NMR spectroscopy as described previously.1–3⇓⇓ In brief, the NMR method uses the characteristic signals broadcast by lipoprotein subclasses of different size as the basis of their quantification. Each subclass signal emanates from the aggregate number of terminal methyl groups on the lipids contained within the particle. Cholesterol esters and triglycerides in the particle core each contribute 3 methyl groups, and phospholipids and unesterified cholesterol in the surface shell each contribute 2 methyl groups. To a close approximation, the diameter of the particle determines the number of methyl groups present (and hence, the amplitude of the methyl NMR signal), irrespective of differences in lipid composition arising from, for example, variations in the relative amounts of cholesterol ester and triglyceride in the particle core, varying degrees of unsaturation of the lipid fatty acyl chains, or varying phospholipid composition. For this reason, the methyl NMR signal emitted by each subclass serves as a direct measure of the concentration of that subclass.
NMR spectra of each plasma specimen were acquired in duplicate at 47°C on an automated 400-MHz lipoprotein analyzer at LipoScience, Inc (Raleigh, NC), and the lipid methyl signal envelope was decomposed computationally to give the amplitudes of the contributing signals of 16 lipoprotein subclasses, among which are 4 LDL subclasses (IDL, 25±2 nm; large LDL, 22±0.7 nm; intermediate LDL, 20.5±0.7 nm; and small LDL, 19±0.7 nm). To obtain the conversion factors needed to relate these LDL signal amplitudes to particle concentrations, purified subclass standards were obtained and subjected to chemical lipid and NMR analysis. The subclass standards were isolated from a diverse group of normolipidemic and dyslipidemic individuals by a combination of ultracentrifugation and agarose gel chromatography and characterized for size distribution by electron microscopy. Particle concentrations (nanomoles of particles per liter) were derived for each subclass standard by measurement of the total core lipid concentration (cholesterol ester plus triglyceride) and by division of the volume occupied by these lipids by the calculated core volume per particle.8 Reported LDL particle concentrations (NMR) are the sums of the concentrations of the LDL subclasses (including IDL). For all biochemical and NMR analyses, samples were handled in a fully blinded fashion such that all investigators had no knowledge of case or control status.
Weighted average LDL particle sizes were computed as the sum of the diameter of each LDL subclass (excluding IDL) multiplied by its relative mass percentage as estimated from the amplitude of its methyl NMR signal. NMR LDL sizes are closely related to those estimated by gradient gel electrophoresis but are uniformly smaller by ≈5 nm, because they are referenced differently to diameters assessed by electron microscopy. To assess the correlation between LDL size measurements by NMR and gradient gel electrophoresis, frozen samples from a study of 21 normolipidemic men9 were measured by NMR and by electrophoresis in nondenaturing composite gradient gels in the laboratory of Dr David Rainwater (Southwest Foundation for Biomedical Research, San Antonio, Tex), as described previously.10 After electrophoresis, lipoprotein cholesterol was stained with Sudan black B, and absorbance profiles were determined with an LKB-Ultroscan XL laser densitometer. LDL particle sizes were calibrated with a standard curve that included thyroglobulin (17.0-nm diameter), carboxylated latex microspheres (38 nm, Duke Scientific), and 2 bands of LDL in a lyophilized plasma standard. We obtained estimates of median LDL diameter from the absorbance profiles by determining the particle diameter at which half of the absorbance came from larger and half from smaller LDL particles.10 The diameter of the most abundant LDL subclass, usually referred to as LDL peak particle diameter, was also measured.
Means and proportions for risk factors for cardiovascular events at baseline were calculated for cases and controls. Student’s t test was used to evaluate differences in means, and the χ2 statistic was used to compare proportions. Analysis of trends was used to test for any association between increasing levels of each plasma marker and the risk of future cardiovascular events, after the sample was divided into quartiles according to the distribution of each plasma marker. The Spearman coefficient was used to assess the correlation between plasma and other risk factors among control subjects.
Adjusted risk estimates were obtained with the use of logistic regression models that, in addition to accounting for the variables used in matching (age and smoking status), adjusted for random assignment to aspirin or vitamin E in the WHS and for several other risk factors for cardiovascular events, including chemically measured total cholesterol:HDL-C (TC:HDL-C) ratio and triglycerides, body mass index, a history of hypertension, a history of diabetes, a parental history of myocardial infarction before the age of 60 years, and use of hormone replacement therapy. Finally, on a post hoc basis and to estimate the clinical relevance of these parameters, we computed the area under receiver operating characteristic curves for prediction models based on LDL particle concentration (NMR), chemically measured TC:HDL-C, and CRP, the 3 strongest predictors of risk in this cohort.
All probability values were 2-tailed, and values of less than 0.05 were considered to indicate statistical significance. All confidence intervals were calculated at the 95% level.
The baseline clinical characteristics of the women who subsequently had cardiovascular events and those who remained free of reported cardiovascular events are shown in Table 1. Baseline LDL particle concentrations (NMR) and those of the individual contributing subclasses, LDL particle size (NMR), and levels of chemically determined lipids are shown in Table 2. Baseline LDL particle concentrations (NMR) were higher among women who subsequently had cardiovascular events than among those who remained free of cardiovascular events (P<0.001). Differences in levels of the small and intermediate-size LDL subclasses accounted for the relationship observed for total LDL particle concentration (NMR), because levels of the large LDL subclass did not differ significantly between cases and controls. The average LDL particle size (NMR) was smaller among cases than among controls (P<0.046). As expected, chemically determined LDL cholesterol levels (P=0.01), TC:HDL-C ratio (P<0.001), triglycerides (P=0.006), and apolipoprotein B-100 levels (P= 0.002) were higher among cases than among controls, whereas HDL-C levels were lower among cases than among controls (P=0.004). Consistent with prior data from this cohort, CRP levels were higher among cases than among controls (P<0.001).11
Table 3 shows the crude relative risks of future cardiovascular events associated with increasing quartiles of each plasma risk factor, adjusted for random treatment assignment. Of the NMR-determined lipoprotein parameters, total LDL particle concentration (NMR) was the strongest predictor, with a relative risk for women in the highest quartile compared with those in the lowest quartile of 4.17 (95% CI 1.96 to 8.87; P=0.0002). Thus, the magnitude of the predictive value of LDL particle concentration (NMR) was similar to that of the measured TC:HDL-C ratio (relative risk for women in the highest quartile compared with that for those in the lowest quartile of 3.11, 95% CI 1.55 to 6.26, P=0.002). The risk estimate in this cohort for CRP was 5.91 (95% CI 2.65 to 13.15, P<0.001).
The adjusted relative risks of future cardiovascular events associated with each quartile of LDL particle concentration (NMR) and LDL particle size (NMR) are shown in Table 4. Although attenuated in magnitude, increasing quartiles of LDL particle concentration (NMR) remained a predictor of future cardiovascular risk in analyses individually adjusted for either the TC:HDL-C ratio (P for trend=0.02), triglycerides (P for trend<0.001), other traditional nonlipid cardiovascular risk factors (P for trend=0.004), or CRP (P= 0.003). In contrast, LDL particle size (NMR) was no longer a significant predictor after adjustment for any of these parameters.
As shown in Table 5, LDL size (NMR) was inversely correlated with triglyceride levels (r=−0.59; P<0.001) and positively correlated with HDL-C (r=0.57, P<0.001), a finding consistent with previous reports. LDL particle concentration (NMR) was positively correlated with LDL cholesterol (r=0.72, P<0.001), TC:HDL-C ratio (r=0.64, P<0.001), apolipoprotein B-100 (r=0.70, P<0.001), and CRP (r=0.24, P=0.006). When LDL particle concentration (NMR) and TC:HDL-C ratio (the best chemical lipid predictor) were included together in a model without other variables, LDL particle concentration (NMR) remained a significant predictor of future cardiovascular risk, whereas the TC:HDL-C ratio did not; specifically in this analysis, the relative risk of future cardiovascular events with a 1-quartile increase in LDL particle concentration (NMR) was 41% (95% CI 5% to 88%; P=0.02), whereas the relative risk with a 1-quartile increase in TC:HDL-C ratio was 22% (95% CI −8% to 61%; P=0.17). Similarly, when LDL particle concentration (NMR) and apolipoprotein B-100 were included together in the same model, the relative risk of future cardiovascular events with a 1-quartile increase in LDL particle concentration (NMR) was 57% (95% CI 15% to 113%; P=0.004), whereas the relative risk with a 1-quartile increase in apolipoprotein B-100 was 2.2% (95% CI −24% to 37%; P=0.9).
To estimate the clinical relevance of these effects, we computed the area under receiver operating characteristic curves for prediction models based on LDL particle concentration (NMR) alone, chemically measured TC:HDL-C alone, and CRP alone. In these analyses, the area under the receiver operating characteristic curve for LDL particle concentration (NMR) was 0.64, which was similar to that seen for TC:HDL-C (0.64). The area under the receiver operating characteristic curve for CRP was 0.66.
Finally, the results of an analysis that compared NMR-assessed LDL sizes with median LDL diameters assessed by polyacrylamide gradient gel electrophoresis (PAGGE) among 21 healthy middle-aged men are presented in the Figure. As expected, NMR-assessed LDL sizes were uniformly smaller, but there was a high correlation between LDL size as assessed by each modality (r=0.89; P<0.001). A similar correlation (r=0.86; P<0.001) was seen between NMR-determined LDL size and LDL sizes from PAGGE measured in terms of peak particle diameter rather than median diameter.
In this prospective nested case-control study of apparently healthy middle-aged women, we found that baseline LDL particle concentration measured by NMR was a predictor of future cardiovascular risk, whereas LDL particle size (by NMR) was a weaker predictor. Although somewhat attenuated in adjusted analyses, the magnitude of predictive value associated with LDL particle concentration (NMR) was similar to that associated with standard lipid measurements, although less than that of CRP. To the best of our knowledge, this is the first study to assess LDL particle concentration (NMR) as a risk marker for future cardiovascular events in individuals without clinical cardiovascular disease. In a recent study of patients with cardiovascular disease from the Pravastatin Limitation of Atherosclerosis in the Coronaries (PLAC-I) trial,12 LDL particle concentrations (NMR) and levels of small LDL but not large LDL were found to predict angiographically measured disease progression independently of standard lipid levels. Our results are consistent with these in showing that the relation of LDL particle concentration (NMR) to cardiovascular events is mediated by the contributions of the small and intermediate-size LDL subclass particles.
LDL particle sizes (NMR), which are derived from the distribution of NMR-measured LDL subclasses, correlate highly with LDL sizes measured by PAGGE, expressed either as median LDL size10 or the more common LDL peak particle diameter.13–17⇓⇓⇓⇓ Most previous prospective studies using PAGGE to measure LDL particle size distribution have found a predominance of small, dense LDL to be associated with cardiovascular risk,13–16⇓⇓⇓ although this effect may be attenuated after adjustment for lipid parameters, in particular triglycerides or TC:HDL-C ratio.14–16⇓⇓ In contrast, a recent report from the secondary prevention Cholesterol and Recurrent Events trial suggested that large LDL particle size was associated with future recurrent events in adjusted analyses among those assigned to placebo, although this association was not present among those randomized to pravastatin.17 The present data for LDL size as assessed by NMR support the primary prevention data13–16⇓⇓⇓ and suggest that small LDL size (NMR) may be associated with cardiovascular risk, but not independently of other lipid risk factors.
Recently, an extended analysis of the PAGGE data from the large cohort of men in the Quebec Cardiovascular Study showed that the concentration of cholesterol in small LDL particles with a diameter <255 Å was a predictor of incident cardiovascular events, whereas LDL peak particle diameter was a weaker predictor that was not significant in multivariate analyses.18 Our findings are consistent with these in showing that NMR-determined concentrations of the smaller LDL particles or total LDL particle concentration (which, in contrast to LDL cholesterol, is weighted most heavily by contributions from the smaller, relatively cholesterol-poor particles) gives superior risk prediction compared with average LDL particle size.
Plasma levels of apolipoprotein B-100 have previously been used to provide an estimate of the number, or concentration, of LDL particles, and thus the strong correlation (r=0.70, P<0.0001) observed between LDL particle concentration (NMR) and plasma apolipoprotein B-100 in the present study is not surprising. Several studies, including the present report, have found plasma levels of apolipoprotein B-100 to be predictive of future cardiovascular risk.11,19–21⇓⇓⇓ In the present cohort, LDL particle concentration (NMR) appeared to be a stronger predictor than plasma apolipoprotein B-100 measured by automated immunoassay. One possible explanation for this unexpected finding is that because apolipoprotein B-100 is present on VLDL as well as LDL, levels of plasma apolipoprotein B-100 provide only an approximation of LDL particle number. In addition, it is noted that measured ratios of apolipoprotein B-100 to LDL cholesterol, which theoretically should be greater for small versus large LDL particles,22 do not always show the expected consistency of association with LDL size.23,24⇓ Further research is required to explain any potential differences between LDL particle concentration (NMR) and apolipoprotein B-100.
Although prospective in design, the present study has potential limitations, and thus these results require confirmation in other cohorts. First, the present study was conducted among women, and thus the predictive value of LDL particle concentration (NMR) among men remains unknown. Second, the HDL-C values in this cohort were somewhat lower than would be expected in the general population, which may limit the generalizability of our results. Third, our plasma samples were nonfasting, which may have affected the predictive power of plasma triglyceride levels; nonetheless, triglyceride levels remained a significant predictor of future cardiovascular risk in the present study. Fourth, we did not measure LDL size by polyacrylamide gradient gel electrophoresis in the Women’s Health Study cohort, and thus the results of NMR-determined LDL size in the present study should not be generalized to other techniques. Finally, the use of frozen samples may be a potential limitation of this study. Nonetheless, data from other settings suggest a negligible difference in NMR particle size and concentration measures on fresh and frozen samples. Moreover, any theoretical random misclassification due to frozen samples would tend to bias our data toward the null.
In summary, these prospective data are consistent with a role for LDL particle concentration (NMR) as a marker of cardiovascular risk prediction. As yet, relatively few studies have investigated the role of pharmacological therapies on NMR-derived lipoprotein measurements.25–27⇓⇓ The current data thus also support the need for studies evaluating the impact of lifestyle modification and lipid-lowering therapy on LDL particle concentration determined by NMR.
This study was supported by grants from the National Heart, Lung, and Blood Institute (HL58755 and HL43851) and the Leducq Foundation, Paris. Dr Ridker is the recipient of an Established Investigator Award from the American Heart Association and a Doris Duke Clinical Scientist Award from the Doris Duke Charitable Foundation. The authors wish to thank Dr Jonathan Cohen, for providing the samples used for substudy of NMR and gradient gel electrophoresis measurements of LDL size, and Dr David Rainwater, whose laboratory performed the gradient gel electrophoretic LDL size analyses.
Dr Otvos is employed by and has an equity ownership position in LipoScience, Inc, which performed the NMR lipoprotein analysis in this study. Dr Ridker is named as a coinventor on a pending patent application filed by the Brigham and Women’s Hospital on the use of markers of inflammation in coronary disease.
- ↵Otvos JD, Jeyarajah EJ, Bennett DW. Quantification of plasma lipoproteins by proton nuclear magnetic resonance spectroscopy. Clin Chem. 1991; 37: 377–386.
- ↵Otvos JD, Jeyarajah EJ, Bennett DW, et al. Development of a proton nuclear magnetic resonance spectroscopic method for determining plasma lipoprotein concentrations and subspecies distributions from a single, rapid measurement. Clin Chem. 1992; 38: 1632–1638.
- ↵Freedman DS, Otvos JD, Jeyarajah EJ, et al. Relation of lipoprotein subclasses as measured by proton nuclear magnetic resonance spectroscopy to coronary artery disease. Arterioscler Thromb Vasc Biol. 1998; 18: 1046–1053.
- ↵Ridker PM. High-sensitivity C-reactive protein (hs-CRP): a potential adjunct for global risk assessment in the primary prevention of cardiovascular disease. Circulation. 2001; 103: 1813–1818.
- ↵Buring JE, Hennekens CH. The Women’s Health Study: summary of the study design. J Myocardial Ischemia. 1992; 4: 27–29.
- ↵Rifai N, Tracy RP, Ridker PM. Clinical efficacy of an automated high-sensitivity C-reactive protein assay. Clin Chem. 1999; 45: 2136–2141.
- ↵Redgrave TG, Carlson LA. Changes in plasma very low density and low density lipoprotein content, composition, and size after a fatty meal in normo- and hypertriglyceridemic man. J Lipid Res. 1979; 20: 217–229.
- ↵Grundy SM, Vega GL, Otvos JD, et al. Hepatic lipase activity influences high density lipoprotein subclass distribution in normotriglyceridemic men: genetic and pharmacological evidence. J Lipid Res. 1999; 40: 229–234.
- ↵Rainwater DL, Moore H Jr, Shelledy WR, et al. Characterization of a composite gradient gel for the electrophoretic separation of lipoproteins. J Lipid Res. 1997; 38: 1261–1266.
- ↵Lamarche B, Tchernof A, Moorjani S, et al. Small, dense low-density lipoprotein particles as a predictor of the risk of ischemic heart disease in men: prospective results from the Quebec Cardiovascular Study. Circulation. 1997; 95: 69–75.
- ↵Kamigaki AS, Siscovick DS, Schwartz SM, et al. Low density lipoprotein particle size and risk of early-onset myocardial infarction in women. Am J Epidemiol. 2001; 153: 939–945.
- ↵St-Pierre AC, Ruel IL, Cantin B, et al. Comparison of various electrophoretic characteristics of LDL particles and their relationship to the risk of ischemic heart disease. Circulation. 2001; 104: 2295–2299.
- ↵Lamarche B, Moorjani S, Lupien PJ, et al. Apolipoprotein A-I and B levels and the risk of ischemic heart disease during a five-year follow-up of men in the Quebec cardiovascular study. Circulation. 1996; 94: 273–278.
- ↵McNamara JR, Small DM, Li Z, et al. Differences in LDL subspecies involve alterations in lipid composition and conformational changes in apolipoprotein B. J Lipid Res. 1996; 37: 1924–1935.