Lipids and Lipoproteins and Risk of Different Vascular Events in the ...

DOI: 10.1161/CIRCULATIONAHA.111.073684

1

Lipids and Lipoproteins and Risk of Different Vascular Events in

the MRC/BHF Heart Protection Study

Running title: Parish et al.; Lipids, lipoproteins and vascular events

Sarah Parish, DPhil1; Alison Offer, PhD1; Robert Clarke, FRCP1; Jemma C. Hopewell, PhD1;

Michael R. Hill, DPhil1; James Otvos, PhD2; Jane Armitage, FRCP1; Rory Collins, FMedSci1

on behalf of the Heart Protection Study Collaborative Group

1Clinical Trial Service Unit and Epidemiological Studies Unit, University of Oxford, Oxford,

United Kingdom; 2LipoScience Inc., Raleigh, NC

Correspondence:

Sarah Parish, DPhil

Heart Protection Study

Clinical Trial Service Unit & Epidemiological Studies Unit

Richard Doll Building

Old Road Campus, Roosevelt Drive

Oxford OX3 7LF, UK

Tel: +44 186 574 3743

Fax: +44 186 574 3985

E-mail: [email protected]

Journal Subject Codes: [7] Chronic ischemic heart disease; [8] Epidemiology; [135] Risk Factors; [90] Lipid and lipoprotein metabolism

by guest on February 11, 2018http://circ.ahajournals.org/

Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from


Dow

nloaded from

http://circ.ahajournals.org/



























2

Abstract:

Background - Low density lipoprotein (LDL) and high density lipoprotein (HDL) cholesterol are

established risk factors for vascular disease, but lipoprotein particle concentrations may be

stronger determinants of risk.

Methods and Results - Associations between vascular events and baseline concentrations of

cholesterol fractions, apolipoproteins B and A1, and lipoprotein particles assessed by nuclear

magnetic resonance were considered in the Heart Protection Study randomized trial of

simvastatin versus placebo (more than 5000 vascular events during 5.3 years follow-up among

20,000 participants). Major occlusive coronary events were equally strongly associated with the

cholesterol- and particle-based total LDL measures; adjusted hazard ratios per one standard

deviation higher level were 1.25 (95%CI 1.16 1.34) for LDL-cholesterol, 1.22 (1.14-1.32) for

non-HDL-cholesterol, 1.23 (1.15-1.33) for apolipoprotein B and 1.25 (1.16 1.35) for LDL-

particle number. Given total LDL particle number, the distribution between small and large

particles did not add predictive value. Associations of these different LDL-related measures

were similar with arterial revascularization procedures, but much weaker or non-existent with

ischemic stroke and other cardiac event (mainly heart failure). After adjustment for LDL particle

number, the hazard ratios for major occlusive coronary event per one standard deviation higher

level were 0.91 (95%CI 0.86-0.96) for HDL-cholesterol and 0.89 (0.85-0.93) for HDL particle

number. Other cardiac events were inversely associated with total (hazard ratio 0.84 [95%CI

0.79-0.90]) and small HDL-particle number (0.82 [0.76-0.89]), but only very weakly associated

with HDL-cholesterol (0.94 [0.88-1.00]).

Conclusions - In a population at 2% average coronary event risk per year, cholesterol,

apolipoprotein, and particle measures of LDL were strongly correlated and had similar predictive

value for incident major occlusive vascular events. It is unclear whether the associations between

HDL particle numbers and other cardiac events represent a causal or reverse-causal effect.

Clinical Trial Registration Information - http://isrctn.org/; Identifier: ISRCTN48489393.

Key words: coronary heart disease, lipids, lipoproteins, risk prediction, lipoprotein particles


Dow

nloaded from



3

Introduction

Large observational studies indicate strong positive associations between low density lipoprotein

(LDL) particles, which carry cholesterol, and the risk of coronary heart disease (CHD).1,2

Randomized trials have demonstrated that lowering LDL cholesterol (LDL-C) with statins

reduces the risk of CHD death, non-fatal MI, ischemic stroke, and the need for revascularization

procedures, with less benefit apparent on deaths from other cardiac disease (including sudden

death, arrhythmia, and heart failure).3,4,5 Observational studies have, however, typically reported

weaker associations of LDL-C (or related measures) with ischemic stroke than with CHD,1,2

which is discordant with the findings of randomized trials where the benefits of LDL-lowering

were similar for CHD and stroke.3,6

High-density lipoprotein (HDL) cholesterol (HDL-C) is inversely associated with CHD

in observational studies (largely among individuals without prior CHD).1,2 However, HDL-C is

negatively correlated with apolipoprotein B (apoB) and with non-HDL cholesterol (non-HDL-

C);2 hence, some of its predictive value is not independent of these LDL-related measures. In the

large Emerging Risk Factors Collaboration (ERFC) meta-analysis of prospective observational

studies of individuals without prior vascular disease, the lower CHD risk associated with one

standard deviation (1 SD) higher HDL-C attenuated from 29% to 22% after adjustment for non-

HDL-C and was only about half the size of the 56% higher risk associated with 1 SD higher non-

HDL-C.2

The extent to which LDL and HDL particles influence risk through atherosclerosis,

inflammation and other postulated mechanisms may vary according to subtypes of lipoprotein

particle.7-9 Small dense LDL particles are believed to be particularly hazardous, but it is unclear

whether these associations are independent of other lipid measures.8,10-12 Studies comparing


Dow

nloaded from



4

different HDL subtypes (based on size, density or apolipoprotein content)13 have yielded

conflicting results about the relevance of different subtypes.14 Discordance in results may be

attributable to small study sizes and variable allowance for associations with the LDL system.

Pre-existing disease in participants may also be a confounding factor, since cardiac disease can

reduce the functionality of HDL particles.15

LDL and HDL are traditionally quantified by measuring the cholesterol they contain

(LDL-C and HDL-C), but inter-individual variations in the cholesterol content of LDL and HDL

particles could mean that alternative analytic measures are more informative.16-18 One approach

is to measure the predominant protein moiety on the particles: apoB for LDL and apolipoprotein

A1 (apoA1) for HDL. LDL particles and their precursors each carry one molecule of apoB on

their surface (with about 90% of apoB being on LDL particles); thus, plasma apoB

concentrations provide a good estimate of LDL particle concentrations. Plasma apoA1

concentrations, in contrast, are not proportional to HDL particle concentrations because HDL

particles contain variable numbers (2 to 5) of apoA1 molecules. Nuclear magnetic resonance

(NMR) spectroscopy offers another way to quantify lipoproteins, providing particle

concentrations of LDL (LDL-P) and HDL (HDL-P) and their subclasses.19 The present analyses

investigate the associations of cholesterol, apolipoprotein, and particle measures of LDL and

HDL, as well as subclass levels, with more than 5000 vascular outcomes in 20,000 high-risk

individuals during 5 years of follow-up in the MRC/BHF Heart Protection Study (HPS).

Methods

Recruitment and Eligibility Criteria

Details of the HPS have been reported previously.20,21 Between 1994-1997 in the United


Dow

nloaded from



5

Kingdom, 20,536 men and women aged 40-80 years were recruited and assigned randomly to

receive 40 mg simvastatin daily or matching placebo (and separately, in a 2 x 2 factorial design,

to receive antioxidant vitamins or placebo capsules). To be eligible, participants had non-fasting

blood total cholesterol concentrations of at least 3.5 mmol/L (135 mg/dL) and either a previous

diagnosis of CHD, cerebrovascular disease, other occlusive disease of non-coronary arteries,

diabetes mellitus (type I or II), or were men aged 65 years or older undergoing treatment for

hypertension. Individuals were excluded if their doctor considered statin therapy clearly

indicated or contraindicated. At initial screening, written consent to participate was obtained, and

a non-fasting blood sample was taken (“screening sample”), and participants then began a “run-

in” phase involving 4 weeks of placebo followed by 4-6 weeks of 40 mg simvastatin daily plus

vitamin supplementation. At the end of this run-in period, a non-fasting blood sample was taken

(“randomization sample”) and compliant and eligible individuals were randomly assigned

treatment for approximately 5 years.

Laboratory Measurements

Blood samples taken into heparinised vacutainers at screening and randomization visits were

chilled to about 4 C and couriered overnight to the coordinating central laboratory for separation,

assay and long-term storage in liquid nitrogen (<-80 C). Analyses for this report were based on

the screening samples for participants subsequently randomly allocated placebo-simvastatin at

the randomization visit, and on the randomization samples (after taking the simvastatin regimen)

for participants who were subsequently randomly allocated active-simvastatin. Plasma lipid

fractions (including HDL-C, apoB and apoA1), triglycerides, creatinine, N-BNP and C-reactive

protein (CRP) were assayed using previously reported methods.4,21,22 As methods for LDL-C

have progressed, LDL-C was re-assayed using the N-geneous method (Genzyme Diagnostics).23


Dow

nloaded from



6

Lipoprotein particle profiles were measured by NMR spectroscopy using the LipoProfile-3

algorithm at LipoScience, Inc (Raleigh, NC). LDL and HDL particle subclasses ( mol/L) were

quantified from the amplitudes of their spectroscopically-distinct lipid methyl group NMR

signals, and weighted-average LDL and HDL sizes were derived from the sum of the diameter of

each subclass multiplied by its relative mass percentage based on the amplitude of its methyl

NMR signal.19 Diameter range estimates for the subclasses were: small LDL (LDL-Psmall) 18-

21.2 nm, large LDL (LDL-Plarge) 21.2-23 nm, intermediate density lipoprotein (IDL-P) 23-27

nm, small HDL (HDL-Psmall) 7.3-8.2 nm, medium HDL (HDL-Pmedium) 8.2-8.8 nm, large HDL

(HDL-Plarge) 8.8-13 nm. LDL-P and HDL-P are the totals of the particle number concentrations

of the LDL and HDL subclasses respectively, and VLDL-P denotes lipoprotein particles of

diameter >27 nm. The glomerular filtration rate was estimated using the Modification of Diet in

Renal Disease formula.24 During the study, blood was collected annually from a random sample

of participants attending follow-up. Repeat NMR and chemistry measurements in the 4000

participants with available samples taken while compliant with their allocated statin treatment

allowed assessment of the reproducibility of each lipid measure.

Follow-up, determination of events

Participants were followed up for the incidence of events, including myocardial infarction (MI),

stroke, vascular procedures, death and hospital admissions for other cardiac events (ascertained

as previously reported in detail);20 this report includes a mean of 5.3 years follow-up until 11th

November 2001, when study treatment ended. As most deaths from heart failure and sudden

death were believed to have an underlying coronary cause, they had been included in the pre-

specified trial outcome “major coronary event” (non-fatal MI or coronary death). However, as

some such deaths may have had non-occlusive aetiologies (eg, arrhythmia), the following


Dow

nloaded from



7

outcome categories were used in the present analyses: “major occlusive coronary event” defined

as non-fatal MI or coronary death other than from heart failure or sudden death (corresponding to

the revised definition of major coronary event in the recent Cholesterol Treatment Trialists’

meta-analyses3), and “other cardiac event” defined as hospitalization or death due to heart

failure, sudden death or non-coronary cardiac death.

Compliance

Among those allocated placebo, non-compliance resulting from the uptake of statin therapy

increased during the course of the study, and was greater amongst those with higher baseline

cholesterol levels (3% versus 29% at 3 years in bottom versus top quintile of baseline non-HDL-

C). This bias would tend to attenuate the observed associations of risk with LDL-related

measures but not affect the associations with HDL-related measures. Since a determinant of

such statin use may have been a high LDL-C or non-HDL-C level, usage may be more directly

related to cholesterol levels than to the levels of other LDL-related measures (such as LDL-P or

apoB), and so lead to greater attenuation of associations with LDL-C and non-HDL-C than of

associations with LDL-P and apoB (see Supplemental Material). When considering comparisons

between LDL-related measures, particular emphasis has, therefore, been placed on the results in

the statin arm to reduce the potential for such confounding.

Statistical analysis

Hazard ratios (HRs) were calculated for the first post-randomization occurrence of each outcome

per 1 SD higher level of the baseline lipid-related factor using Cox proportional hazards

regression. Standard adjustment in risk analyses included: simvastatin and vitamin allocation,

baseline 5-year age group by sex, prior history of vascular disease or diabetes (as a 3-way

variable for CHD [MI, other CHD, no CHD] and a 5-way variable for cerebrovascular disease,


Dow

nloaded from



8

peripheral vascular disease and diabetes [each alone, multiple, none]), treated hypertension,

medication use (ACE inhibitors, anti-coagulants, beta-blockers, bronchodilators, calcium

antagonists, digoxin, diuretics, hormone replacement therapy, hypoglycaemics, inhaled steroids,

insulin), cigarette smoking status (current, ex-, never), systolic blood pressure (SBP) and

estimated glomerular filtration rate (eGRF) as continuous variables, and N-BNP as a categorical

variable with 5 groups (cut-points: 400, 1000, 2000 and 5000 pg/ml, based around the ESC

guidelines which grade N-BNP<400 pg/ml as “heart failure unlikely”, and N-BNP>2000 pg/ml

as “heart failure likely”25). The median value was substituted for the few missing values for any

covariate (<0.7%). The SD of a lipid-related factor was estimated by the root mean square error

in a regression of the factor on the standard adjustment terms. For LDL subclasses, the SD of the

total particle concentration was used so that all the hazard ratios were per the same unit of

particle number, facilitating comparison of whether the hazard per particle varied by subclass;

and likewise hazard ratios for HDL subclasses were calculated per SD of HDL-P. Likelihood

ratio tests, yielding 2 statistics and corresponding p-values, were used to assess evidence of

association following progressive inclusion of terms. Pearson correlation coefficients were used,

except in correlations involving N-BNP when Spearman were used. All analyses were based on

the 20,021 participants with complete information on concentrations of NMR particles,

cholesterol fractions and apolipoproteins.

Results

During a mean follow-up of 5.3 years, 1796 participants had a major occlusive coronary event,

2187 had a revascularization, 1043 had an “other cardiac event” and 995 had a presumed

ischemic stroke (Table 1). History of occlusive vascular disease, and indicators of heart


Dow

nloaded from



9

failure25,26 were major determinants of the incidence of vascular events (Supplemental Table 1),

highlighting the importance of adjusting for these factors in all analyses. The mean

concentrations of the directly measured chemistries and the NMR lipoprotein subclasses are

shown in Supplemental Table 2, together with self-correlations which indicate the

reproducibility of measurements in samples collected from the same individual a few years apart.

All of the LDL-related factors, with the exception of IDL-P, showed similar self-correlations of

about 0.65 on statin and 0.75 off statin. The self-correlations of the different HDL-related

factors showed greater variation, but were similar on and off statin (eg off statin self-correlations

0.83, 0.67, 0.73 and 0.60, respectively, for HDL-C, apoA1, HDL particles and HDL-Psmall).

Correlations between LDL- and HDL-related measures

The four total LDL-related measures were strongly correlated with each other in measurements

made in the same samples (Table 2): non-HDL-C and apoB were the most strongly correlated

(0.93), while LDL-P correlated most strongly with apoB (0.84) and least strongly with LDL-C

(0.79). The three total HDL-related measures were less strongly correlated with each other

(pairwise correlation coefficients between 0.61 and 0.81: Table 2). The LDL- and HDL-related

measures were considerably inter-correlated. LDL and HDL size were positively correlated with

each other (0.53), and both were negatively correlated with LDL-P and triglycerides and

positively correlated with HDL-P. HDL-C was more strongly negatively correlated with several

LDL-related measures than was HDL-P (eg, correlations respectively of, -0.34 and -0.10 with

LDL-P and of -0.46 and 0.06 with log triglycerides).

LDL-related associations

Figure 1 compares the hazard ratios and 2 statistics for the strengths of association of LDL-C,

non-HDL-C, apoB and LDL-P for several different outcomes, after adjustment for baseline


Dow

nloaded from



10

covariates. Major occlusive coronary event and revascularization showed clear associations with

all four LDL-related measures in both the statin and placebo arms, while other cardiac event and

ischemic stroke showed only weak associations with these measures. In the statin arm, the LDL

particle measures (apoB and LDL-P) and the cholesterol measures (LDL-C and non-HDL-C)

showed similar strengths of association with major occlusive coronary event: HR 1.23 (95% CI

1.15 1.33) and 1.25 (1.16 1.35) for apoB and LDL-P; and 1.25 (1.16 1.34) and 1.22

(1.14 1.32) for LDL-C and non-HDL-C, respectively (Figure 1). Findings were similar for

revascularization (Figure 1), and for the combined outcome of major occlusive coronary event

or revascularization (HRs between 1.19 and 1.20, Supplemental Figure 1). In the placebo arm,

the associations of all of the measures with the combined outcome were weaker than in the statin

arm, but (given that some bias might arise from compliance associations) there was no clear

evidence of any differences between the measures. The separate hazard ratios for major

occlusive coronary event and revascularization (Figure 1) were statistically compatible with the

results for the combined outcome (Supplemental Figure 1). Including VLDL-P with LDL-P

made little difference, but total cholesterol was a weaker predictor than LDL-C or non-HDL-C

(Supplemental Figure 1).

Table 3 shows the associations of LDL subclasses with major occlusive coronary event

in the statin arm. Considered singly, the association with LDL-Psmall was stronger than that with

LDL-Plarge but both were weaker than the association with LDL-P. The 23 for the strength of the

association with the three LDL subclasses jointly was the same as that for LDL-P alone (32.4

versus 32.3), indicating that the LDL subclasses did not contribute any additional predictive

value over LDL-P. A similar pattern was seen for revascularization. The strengths of the

associations with the main LDL-P subclasses individually corresponded to their respective


Dow

nloaded from



11

correlations with LDL-P (correlation coefficient 0.76 for LDL-Psmall, 0.16 for LDL-Plarge; Table

2).

HDL-related associations

Figure 2 contrasts the associations of different HDL-related measures, given LDL-P, with

various vascular outcomes in all participants. (Associations in the statin and placebo arms

separately were similar and are given in Supplemental Figure 2.) For major occlusive coronary

event, the HRs were around 0.90, with HDL-P showing a slightly stronger association than

HDL-C and apoA1 ( 21 22.5, 13.2 and 13.1, respectively): whereas, for revascularization, only

HDL-C showed much association ( 21 15.9). For both outcomes, however, the confidence

intervals for all three measures overlapped.

Table 4 shows the associations of the HDL subclasses with major occlusive coronary

event in all participants, both with and without adjustment for LDL-P. Without such adjustment,

HDL-C and HDL-P showed similar strengths of association with major occlusive coronary event

risk. However, these associations derived partly from negative correlations with LDL-related

measures (correlations of HDL-Plarge and HDL-C with LDL-P of -0.44 and -0.34, respectively;

Table 2). Hence, given LDL-P, the strengths of association with major occlusive coronary event

were reduced, becoming slightly weaker for HDL-C than for HDL-P. The 23 for the strength of

association with the three HDL subclasses jointly was minimally greater than that for HDL-P

alone (24.0 versus 22.5), indicating that, given LDL-P, the HDL subclasses did not contribute

additional predictive value over HDL-P for major occlusive coronary event.

Ischemic stroke showed no significant association with any of the HDL-related measures.

Other cardiac events were strongly associated with HDL-P and with mean HDL particle size

given HDL-P (HRs 0.84 [95%CI 0.79-0.90] and 1.13 [1.05-1.21], respectively; Figure 2).


Dow

nloaded from



12

Supplemental Table 3 shows that these associations derived from inverse associations of risk

with HDL-Psmall and HDL-Pmedium (HRs 0.82 [95%CI 0.76-0.89] and 0.81 [0.74-0.89],

respectively, in analyses of the subclasses jointly). This pattern of association was most apparent

in the absence of adjustment for N-BNP, suggesting that pre-existing cardiac insufficiency (as

indicated by higher N-BNP) was also associated with this HDL pattern. Lack of adjustment for

N-BNP also impacted on the relative strengths of the associations of the HDL-related measures

with major occlusive coronary event, causing the HDL-P association to appear stronger (Figure

3). The associations of HDL-related measures with any major coronary event (ie, including 449

heart failure or sudden deaths believed to have an underlying coronary cause) also weakly

reflected the pattern seen for other cardiac event.

Influence of prior disease

The adjustments for prior disease (included in all analyses) had little net effect on the LDL-

related associations. By contrast, HDL-C and HDL-P associations were considerably attenuated

by the adjustment for prior disease-related factors. For example, 2 values for the associations

with major occlusive coronary event were reduced by 50-60% after further adjustment beyond

age, sex, allocated treatment arm, smoking and blood pressure (Supplemental Table 4). The

attenuation was somewhat greater for HDL-P than for HDL-C, largely through the additional

association of HDL-P with N-BNP (Supplemental Table 4, Table 2).

Findings in this investigation did not differ by sex (Supplemental Figures 3-4), and

almost all participants (97%) were of white ethnicity.

Discussion

There was no clear evidence in this large study of a difference in the strengths of the associations


Dow

nloaded from



13

of any of the particle or cholesterol-based LDL-related measures with major occlusive coronary

events or revascularization (1580 such events in the statin arm; 2010 in the placebo arm). The

lack of additional predictive value from LDL subclasses or LDL mean size among people with

prior vascular disease in this study is consistent with the results from the most informative of the

previous studies, which were in populations without prior cardiovascular disease (EPIC-Norfolk

with 1003 coronary artery disease cases;18 Women’s Health Study [WHS] with 1015

cardiovascular cases17). Other studies have reported mixed results,10 but the variety of

approaches to measurement and statistical analysis make a consensus summary of them difficult.

Furthermore, improvements in assay methods and standardisation23,27 may have altered the

relative predictive values of different measures.

The strengths of the associations of the LDL-P subclasses with occlusive events in the

present study broadly reflected their correlations with LDL-P. LDL-Psmall was the subclass most

strongly correlated with LDL-P, but its association with major occlusive coronary event risk was

weaker than that of LDL-P. The lack of added predictive value from the subclasses cannot be

attributed to differences in reproducibility since the reproducibility of the LDL subclasses (in

samples collected a few years apart) was similar to that of the four total LDL-related measures.

These four measures were strongly correlated with each other (~0.8 or higher) in the present

study, whereas the correlations between cholesterol and particle measures in the EPIC-Norfolk

and WHS studies, were somewhat weaker (eg ~ 0.63 between LDL-P and LDL-C).

The ERFC meta-analysis of over 4000 CHD events in people without prior vascular

disease, also found that non-HDL-C and apoB (and LDL-C, which was available for about 2000

cases) had similar strengths of association with CHD.2 This finding, however, differs from those

of earlier large retrospective case-control studies in acute MI which had reported that


Dow

nloaded from



14

apolipoproteins were superior predictors of MI than cholesterol fractions,28,29 but this may be

explained by distortions of HDL-C levels in the hours following an acute MI when

apolipoproteins may be more robust measures. The hazard ratio for major occlusive coronary

event per unit non-HDL-C (1.22 per 0.34 mg/dl in the statin arm) in the present study was low

compared to that found in the ERFC meta-analysis (1.56 per 0.29 mg/dl after correction for

within-person variation). In the present study, within-person variation together with some

variation in statin use (even in the statin arm), resulted in correlations of only about 0.55 between

baseline and mid-study non-HDL-C values (data not shown), which would attenuate associations

with risk. The weaker hazard ratios observed in this high-risk population would also be

consistent with the weaker hazard ratios seen at older ages and higher SBP in previous large

meta-analyses.1,2

The weaker and indefinite associations with ischemic stroke in the present study are also

consistent with the relative strengths of the associations of non-HDL-C with coronary events and

with ischemic stroke in the ERFC (where, based on 1800 strokes, the log hazard ratios were only

a quarter as great for ischemic stroke as for CHD). NMR-measured lipoprotein subclass

measurements did not provide any insight into the discordance between the weak observational

findings and the effect of LDL-lowering therapy on ischemic stroke.

HDL-C and HDL size were more strongly correlated than apoA1 or HDL-P with the total

LDL-related measures. Thus, adjustment for LDL-P attenuated the risk association with HDL-C

more than it attenuated the risk associations with apoA1 and HDL-P. The weak levels of

association of all HDL measures with CHD, however, limited the power to discriminate between

these measures. The correlation of baseline HDL-C with repeat measurements obtained during

follow-up was strong (0.82), so allowance for within-person variation30 would only increase the


Dow

nloaded from



15

strength of association modestly (eg from 0.90 to 0.88). The modest hazard ratios for the HDL-

related measures found in this high-risk population (by comparison with ERFC) may indicate a

tendency to weaker associations in less healthy individuals, as suggested by the marked trends

with age and SBP in the ERFC analyses.2 Such attenuation may be a manifestation of HDL

becoming dysfunctional in individuals with existing cardiovascular disease.15,31-34 Adjustment for

prior disease, including for N-BNP as a marker of heart failure, considerably attenuated

associations with HDL-related measures, highlighting the potential for confounding by pre-

existing disease and for conflicting observations from studies with differing degrees of

adjustment.

A strong inverse association was seen between HDL-Psmall and both N-BNP and other

cardiac event risk (Table 2, Supplemental Table 3). As cardiac events in this study

predominantly occurred in participants who had pre-existing cardiac disease at entry, it is not

possible to determine whether the association represents a causal or reverse-causal effect (ie

whether higher HDL-Psmall protects against the risk of a cardiac event,7 or whether lower HDL-

Psmall is a consequence of the pre-existing disease). Small and large mean HDL size were

associated with different forms of prior disease: small mean HDL size was associated with high

triglycerides and LDL-P and with other features of metabolic syndrome (Table 2), as previously

noted in the EPIC-Norfolk study,35 while large mean HDL size was associated with high N-BNP

and other cardiac event risk.

The current large study was prospective in design and, in contrast to previous studies,

was able to look separately at associations of lipoprotein subclasses with coronary events, other

cardiac events, ischemic strokes and revascularizations. Another strength of this study was that

it included detailed adjustment for prior vascular disease and other established risk factors. It


Dow

nloaded from



16

extends previous results, showing similar strengths of risk association with different LDL and

HDL measures in a population at higher risk of occlusive coronary event (average 2% per

annum). Published studies have suggested that the magnitude of some associations of lipid

levels with risk may attenuate with increasing levels of risk factors (such as age and SBP) and,

thus, it is a limitation of this study that the observed associations may be representative only of

this type of high-risk population.

Conclusions

In a wide range of individuals at varying levels of cardiovascular risk, lipoprotein particles,

apolipoproteins, and cholesterol fractions of LDL are of similar predictive value for major

occlusive coronary events, and LDL subclasses provide little additional information.

Lipoprotein particle number measurements could be considered as an alternative to LDL and

HDL cholesterol measurements for risk prediction if they became available at a similar cost.

Measurement of a particle-based measure for LDL, such as apoB or LDL-P, also makes more

biological sense since it is particles that attach to the arterial wall.36 Associations of HDL-related

measures with risk are considerably influenced by prior disease and, in particular, interpretation

of the relationship between HDL particle measures, prior disease and risk of other cardiac events

requires further study.

Acknowledgements: The most important acknowledgement is to the participants in the study,

and to the steering committee and collaborators listed in a previous report.20 We also gratefully

acknowledge the staff of the Wolfson laboratories in the Clinical Trial Service Unit for their help

with processing, storage and retrieval of blood samples.


Dow

nloaded from



17

Funding Sources: The Heart Protection Study was funded by the UK Medical Research

Council, British Heart Foundation, Merck & Co (manufacturers of simvastatin), and Roche

Vitamins Ltd (manufacturers of vitamins). NMR lipoprotein measurements were funded by

LipoScience, Inc. Jemma Hopewell acknowledges support from the Oxford British Heart

Foundation Centre of Research Excellence.

Conflict of Interest Disclosures: The Clinical Trial Service Unit has a policy of not accepting

honoraria or other payments from the pharmaceutical industry, except for the reimbursement of

costs to participate in scientific meetings. James Otvos is an employee and shareholder of

LipoScience, Inc.

References:

1. Prospective Studies Collaboration. Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta-analysis of individual data from 61 prospective studies with 55 000 vascular deaths. Lancet. 2007;370:1829-1839. 2. Di Angelantonio E, Sarwar N, Perry P, Kaptoge S, Ray KK, Thompson A, Wood AM, Lewington S, Sattar N, Packard CJ, Collins R, Thompson SG, Danesh J. Major lipids, apolipoproteins, and risk of vascular disease. JAMA. 2009;302:1993-2000. 3. Baigent C, Blackwell L, Emberson J, Holland LE, Reith C, Bhala N, Peto R, Barnes EH, Keech A, Simes J, Collins R. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet. 2010;376:1670-1681. 4. Emberson JR, Ng LL, Armitage J, Bowman L, Parish S, Collins R. N-terminal Pro-B-type natriuretic peptide, vascular disease risk, and cholesterol reduction among 20,536 patients in the MRC/BHF heart protection study. J Am Coll Cardiol. 2007;49:311-319. 5. Cleland JGF, McMurray JJV, Kjekshus J, Cornel JH, Dunselman P, Fonseca C, Hjalmarson A, Korewicki J, Lindberg M, Ranjith N, van Veldhuisen DJ, Waagstein F, Wedel H, Wikstrand J, on behalf of the CORONA Study Group. Plasma Concentration of Amino-Terminal Pro-Brain Natriuretic Peptide in Chronic Heart Failure: Prediction of Cardiovascular Events and Interaction With the Effects of Rosuvastatin: A Report From CORONA (Controlled Rosuvastatin Multinational Trial in Heart Failure). J Am Coll Cardiol. 2009;54:1850-1859. 6. Amarenco P, Goldstein LB, Szarek M, Sillesen H, Rudolph AE, Callahan A 3rd, Hennerici M, Simunovic L, Zivin JA, Welch KM. Effects of intense low-density lipoprotein cholesterol reduction in patients with stroke or transient ischemic attack: the Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) trial. Stroke. 2007;38:3198-3204.


Dow

nloaded from



18

7. Kontush A, Chapman MJ. Antiatherogenic small, dense HDL--guardian angel of the arterial wall? Nat Clin Pract Cardiovasc Med. 2006;3:144-153. 8. Rizzo M, Berneis K. Low-density lipoprotein size and cardiovascular risk assessment. QJM. 2006; 99:1-14. 9. Francis GA. The complexity of HDL. Biochim Biophys Acta. 2010;1801:1286-1293. 10. Ip S, Lichtenstein AH, Chung M, Lau J, Balk EM. Systematic review: association of low-density lipoprotein subfractions with cardiovascular outcomes. Ann Intern Med. 2009;150:474-484. 11. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) Final Report . Circulation. 2002;106:3143. 12. Mora S, Szklo M, Otvos JD, Greenland P, Psaty BM, Goff DC Jr, O'Leary DH, Saad MF, Tsai MY, Sharrett AR. LDL particle subclasses, LDL particle size, and carotid atherosclerosis in the Multi-Ethnic Study of Atherosclerosis (MESA). Atherosclerosis. 2007;192:211-217. 13. Rosenson RS, Brewer HBJ, Chapman MJ, Fazio S, Hussain MM, Kontush A, Krauss RM, Otvos JD, Remaley AT, Schaefer EJ. HDL Measures, Particle Heterogeneity, Proposed Nomenclature, and Relation to Atherosclerotic Cardiovascular Events. Clin Chem. 2011;57:392-410. 14. Barter P, Kastelein J, Nunn A, Hobbs R. High density lipoproteins (HDLs) and atherosclerosis; the unanswered questions. Atherosclerosis. 2003;168:195-211. 15. Dodani S, Grice DG, Joshi S. Is HDL function as important as HDL quantity in the coronary artery disease risk assessment? J Clin Lipidol. 2009;3:70-77. 16. Otvos JD, Mora S, Shalaurova I, Greenland P, Mackey RH, Goff DC. Clinical implications of discordance between low-density lipoprotein cholesterol and particle number. J Clin Lipidol. 2011;5:105-113. 17. Mora S, Otvos JD, Rifai N, Rosenson RS, Buring JE, Ridker PM. Lipoprotein particle profiles by nuclear magnetic resonance compared with standard lipids and apolipoproteins in predicting incident cardiovascular disease in women. Circulation. 2009;119:931-939. 18. El Harchaoui K, van der Steeg WA, Stroes ES, Kuivenhoven JA, Otvos JD, Wareham NJ, Hutten BA, Kastelein JJ, Khaw KT, Boekholdt SM. Value of low-density lipoprotein particle number and size as predictors of coronary artery disease in apparently healthy men and women: the EPIC-Norfolk Prospective Population Study. J Am Coll Cardiol. 2007;49:547-553. 19. Jeyarajah EJ, Cromwell WC, Otvos JD. Lipoprotein particle analysis by nuclear magnetic


Dow

nloaded from



19

resonance spectroscopy. Clin Lab Med. 2006;26:847-870. 20. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:7-22. 21. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol-lowering therapy and of antioxidant vitamin supplementation in a wide range of patients at increased risk of coronary heart disease death: early safety and efficacy experience. Eur Heart J. 1999;20:725-741. 22. Heart Protection Study Collaborative Group. C-reactive protein concentration and the vascular benefits of statin therapy: an analysis of 20 536 patients in the Heart Protection Study. Lancet. 2011;377:469-476. 23. Nauck M, Warnick GR, Rifai N. Methods for measurement of LDL-cholesterol: a critical assessment of direct measurement by homogeneous assays versus calculation. Clin Chem. 2002;48:236-254. 24. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130:461-470. 25. Dickstein K, Cohen-Solal A, Filippatos G, McMurray JJ, Ponikowski P, Poole-Wilson PA, Stromberg A, van Veldhuisen DJ, Atar D, Hoes AW, Keren A, Mebazaa A, Nieminen M, Priori SG, Swedberg K, Vahanian A, Camm J, De Caterina R, Dean V, Dickstein K, Filippatos G, Funck-Brentano C, Hellemans I, Kristensen SD, McGregor K, Sechtem U, Silber S, Tendera M, Widimsky P, Zamorano JL. ESC Guidelines for the diagnosis and treatment of acute and chronic heart failure 2008: the Task Force for the Diagnosis and Treatment of Acute and Chronic Heart Failure 2008 of the European Society of Cardiology. Developed in collaboration with the Heart Failure Association of the ESC (HFA) and endorsed by the European Society of Intensive Care Medicine (ESICM). Eur Heart J. 2008;29:2388-2442. 26. Struthers A, Lang C. The potential to improve primary prevention in the future by using BNP/N-BNP as an indicator of silent 'pancardiac' target organ damage: BNP/N-BNP could become for the heart what microalbuminuria is for the kidney. Eur Heart J. 2007;28:1678-1682. 27. Contois JH, McConnell JP, Sethi AA, Csako G, Devaraj S, Hoefner DM, Warnick GR. Apolipoprotein B and Cardiovascular Disease Risk: Position Statement from the AACC Lipoproteins and Vascular Diseases Division Working Group on Best Practices. Clin Chem. 2009;55:407-419. 28. Parish S, Peto R, Palmer A, Clarke R, Lewington S, Offer A, Whitlock G, Clark S, Youngman L, Sleight P, Collins R. The joint effects of apolipoprotein B, apolipoprotein A1, LDL cholesterol, and HDL cholesterol on risk: 3510 cases of acute myocardial infarction and 9805 controls. Eur Heart J. 2009;30:2137-2146.


Dow

nloaded from



20

29. McQueen MJ, Hawken S, Wang X, Ounpuu S, Sniderman A, Probstfield J, Steyn K, Sanderson JE, Hasani M, Volkova E, Kazmi K, Yusuf S. Lipids, lipoproteins, and apolipoproteins as risk markers of myocardial infarction in 52 countries (the INTERHEART study): a case-control study. Lancet. 2008;372:224-233. 30. Clarke R, Shipley M, Lewington S, Youngman L, Collins R, Marmot M, Peto R. Underestimation of risk associations due to regression dilution in long-term follow-up of prospective studies. Am J Epidemiol. 1999;150:341-353. 31. Sviridov D, Mukhamedova N, Remaley AT, Chin-Dusting J, Nestel P. Antiatherogenic functionality of high density lipoprotein: how much versus how good. J Atheroscler Thromb. 2008;15:52-62. 32. Feng H, Li XA. Dysfunctional high-density lipoprotein. Curr Opin Endocrinol Diabetes Obes. 2009;16:156-162. 33. Nicholls SJ, Zheng L, Hazen SL. Formation of Dysfunctional High-Density Lipoprotein by Myeloperoxidase. Trends in Cardiovascular Medicine. 2005;15:212-219. 34. Fogelman AM. When good cholesterol goes bad. Nature Medicine. 2004;10:902-903. 35. El Harchaoui K, Arsenault BJ, Franssen R, Despres JP, Hovingh GK, Stroes ES, Otvos JD, Wareham NJ, Kastelein JJ, Khaw KT, Boekholdt SM. High-density lipoprotein particle size and concentration and coronary risk. Ann Intern Med. 2009;150:84-93. 36. Arsenault BJ, Boekholdt SM, Kastelein JJ. Lipid parameters for measuring risk of cardiovascular disease. Nat Rev Cardiol. 2011;8:197-206.


Dow

nloaded from



21

Table 1. Numbers of participants suffering various vascular events

Type of vascular event Statin arm (10033)

Placebo arm (9988)

Major occlusive coronary event *†

Non-fatal MI 388 598

Death from MI 148 197

Death from CHD (excl. heart failure and sudden death) 245 286

Subtotal 757 1039

Revascularization

Coronary revascularization 542 743

Non-coronary revascularization: aortic 67 62

Non-coronary revascularization: carotid 44 83

Non-coronary revascularization: peripheral 421 474

Subtotal 970 1217

Other cardiac event

Non-fatal heart failure 305 337

Death from heart failure (with underlying CHD cause)† 64 81

Death from heart failure (without underlying CHD cause) 6 10

Sudden death† 150 154

Other non-coronary cardiac death 20 16

Subtotal 505 538

Ischemic stroke (including stroke of unknown origin)

Non-fatal ischemic stroke 371 488

Fatal ischemic stroke 76 101

Subtotal 430 565 *Major occlusive coronary event corresponds to the revised definition of major coronary event used in the latest cycle of analyses from the Cholesterol Treatment Trialists collaboration1 †Major coronary event (as defined in the protocol of the Heart Protection Study clinical trial) is composed of these events


Dow

nloaded from


22

Table 2. Correlation coefficients* between measurements of different lipid-related factors within the same sample in 20 021 participants

LDL-C

Non-HDL-C ApoB LDL-P LDL-

Psmall LDL-Plarge

IDL-P LDL size HDL-C ApoA1 HDL-P HDL-

Psmall HDL-Pmedium

HDL-Plarge

HDL size

log (triglyceride)

LDL-related factors

LDL-C 0.87 0.87 0.79 0.39 0.42 0.36 0.06 -0.07 0.05 -0.02 0.17 -0.08 -0.22 -0.30 0.20 Non-HDL-C 0.87 0.93 0.80 0.52 0.19 0.44 -0.15 -0.18 0.05 0.03 0.24 -0.06 -0.28 -0.33 0.48 ApoB 0.87 0.93 0.84 0.54 0.24 0.38 -0.13 -0.18 0.03 -0.01 0.24 -0.10 -0.29 -0.34 0.39 LDL-P 0.79 0.80 0.84 0.76 0.16 0.34 -0.29 -0.34 -0.14 -0.10 0.30 -0.18 -0.44 -0.52 0.37

LDL-Psmall 0.39 0.52 0.54 0.76 -0.48 0.13 -0.76 -0.62 -0.36 -0.18 0.40 -0.26 -0.66 -0.65 0.58

LDL-Plarge 0.42 0.19 0.24 0.16 -0.48 -0.11 0.76 0.44 0.28 0.07 -0.22 0.09 0.39 0.27 -0.41 IDL-P 0.36 0.44 0.38 0.34 0.13 -0.11 -0.08 0.02 0.13 0.16 0.09 0.12 -0.07 -0.08 0.15 LDL size 0.06 -0.15 -0.13 -0.29 -0.76 0.76 -0.08 0.59 0.35 0.16 -0.38 0.26 0.56 0.53 -0.55

HDL-related factors HDL-C -0.07 -0.18 -0.18 -0.34 -0.62 0.44 0.02 0.59 0.81 0.61 -0.20 0.43 0.84 0.70 -0.46

ApoA1 0.05 0.05 0.03 -0.14 -0.36 0.28 0.13 0.35 0.81 0.77 0.07 0.42 0.67 0.48 -0.08 HDL-P -0.02 0.03 -0.01 -0.10 -0.18 0.07 0.16 0.16 0.61 0.77 0.28 0.58 0.45 0.28 0.06

HDL-Psmall 0.17 0.24 0.24 0.30 0.40 -0.22 0.09 -0.38 -0.20 0.07 0.28 -0.52 -0.33 -0.50 0.31

HDL-Pmedium -0.08 -0.06 -0.10 -0.18 -0.26 0.09 0.12 0.26 0.43 0.42 0.58 -0.52 0.29 0.34 -0.04

HDL-Plarge -0.22 -0.28 -0.29 -0.44 -0.66 0.39 -0.07 0.56 0.84 0.67 0.45 -0.33 0.29 0.89 -0.39 HDL size -0.30 -0.33 -0.34 -0.52 -0.65 0.27 -0.08 0.53 0.70 0.48 0.28 -0.50 0.34 0.89 -0.38

Other factors log (triglyceride) 0.20 0.48 0.39 0.37 0.58 -0.41 0.15 -0.55 -0.46 -0.08 0.06 0.31 -0.04 -0.39 -0.38 log (N-BNP)† -0.07 -0.11 -0.10 0.01 -0.11 -0.15 0.09 0.07 0.13 0.14 0.08 -0.03 0.14 -0.10 0.01 0.15 log (CRP)† 0.04 0.06 0.08 0.12 0.16 -0.06 -0.01 -0.12 -0.15 -0.13 -0.09 -0.01 0.00 -0.16 -0.11 0.08

BMI§ 0.04 0.09 0.08 0.13 0.22 -0.16 0.04 -0.20 -0.23 -0.14 -0.04 0.15 -0.07 -0.24 -0.24 0.21

eGFR†‡ -0.08 -0.12 -0.09 -0.08 -0.10 0.05 -0.01 0.06 0.10 0.05 0.09 -0.06 0.09 0.11 0.10 -0.14 * Adjusted for age, sex, simvastatin and vitamin allocation, prior disease, SBP, smoking , eGFR, log(N-BNP) and medication; † Available in screening samples only (n=9988); ‡ Not adjusted for eGFR; § Available for 19857 participants


Dow

nloaded from


23

Table 3. Association of major occlusive coronary event and revascularization with LDL-related factors* in the statin arm

Major occlusive coronary event Revascularization

HR (95%CI) per 1SD† 12 p value HR (95%CI) per 1SD† 12 p value

LDL-P 1.25 (1.16-1.35) 32.3 <0.0001 1.15 (1.07-1.23) 15.3 <0.0001 LDL size 0.92 (0.85-1.00) 4.1 0.04 0.92 (0.86-0.99) 5.2 0.02 LDL size given LDL-P 0.97 (0.89-1.05) 0.5 0.47 0.95 (0.88-1.02) 2.1 0.15 LDL subclasses‡ singly LDL-Psmall 1.20 (1.11-1.30) 20.0 <0.0001 1.14 (1.06-1.22) 12.8 0.0003 LDL-Plarge 1.08 (0.96-1.23) 1.6 0.21 1.02 (0.91-1.14) 0.1 0.71 IDL-P 1.53 (1.08-2.16) 5.5 0.02 1.13 (0.82-1.54) 0.5 0.47 LDL subclasses‡ jointly 32 32 LDL-Psmall 1.25 (1.14-1.36) 1.17 (1.08-1.26) LDL-Plarge 1.23 (1.08-1.41) 32.4 <0.0001 1.12 (0.99-1.26) 16.3 0.001 IDL-P 1.32 (0.94-1.86) 1.03 (0.75-1.42) 22 22 Additional predictive value of subclasses over LDL-P§

0.1 0.93 0.9 0.62

* Analyses are adjusted for age, sex, simvastatin and vitamin allocation, smoking, prior disease, SBP, eGFR, medication and N-BNP; † All SDs are as in figure 1; ‡ Hazard ratios for all LDL subclasses are per 0.34 μmol/l (1SD of LDL-P); § Calculated from the difference between the 3

2 for the subclasses jointly and the 12 for

LDL-P alone


Dow

nloaded from


24

Table 4. Association of major occlusive coronary event with HDL-related factors in all participants*

Not adjusted for LDL-P Adjusted for LDL-P

HR (95%CI) per 1SD† 12 p value � HR (95%CI) per 1SD† 12 p value

HDL-C 0.87 (0.83-0.92) 30.1 <0.0001 0.91 (0.86-0.96) 13.2 0.0003 ApoA1 0.89 (0.85-0.94) 19.4 <0.0001 0.91 (0.87-0.96) 13.1 0.0003 HDL-P 0.88 (0.83-0.92) 28.7 <0.0001 0.89 (0.85-0.93) 22.5 <0.0001 HDL size 0.91 (0.87-0.95) 16.0 <0.0001 0.97 (0.92-1.03) 0.8 0.36 HDL size given HDL-P 0.94 (0.89-0.99) 6.3 0.01 1.01 (0.96-1.07) 0.2 0.66 HDL subclasses‡ singly HDL-Psmall 1.03 (0.98-1.09) 1.3 0.26 0.98 (0.93-1.03) 0.6 0.44

HDL-Pmedium 0.88 (0.83-0.93) 22.7 <0.0001 0.90 (0.85-0.95) 13.6 0.0002

HDL-Plarge 0.79 (0.71-0.88) 20.3 <0.0001 0.89 (0.79-1.00) 3.8 0.05

HDL subclasses‡ jointly

HDL-Psmall 0.93 (0.87-0.99) 32 0.90 (0.85-0.96) 32

HDL-Pmedium 0.87 (0.82-0.93) 38.0 <0.0001 0.87 (0.81-0.93) 24.0 <0.0001 HDL-Plarge 0.82 (0.73-0.91) 0.92 (0.81-1.03) 22 22 Additional predictive value of subclasses over HDL-P§

9.3 0.01 1.6 0.45

* Analyses are adjusted for age, sex, simvastatin and vitamin allocation, smoking, prior disease, SBP, eGFR, medication and N-BNP; † All SDs are as in figure 2. ‡ Hazard ratios for all HDL subclasses are per 5.18 μmol/l (1SD of HDL-P); § Calculated from the difference between the 32 for the subclasses jointly and the 1

2 for HDL-P alone


Dow

nloaded from


25

Figure Legends:

Figure 1. Comparison of the predictive strengths of LDL-related measures for vascular

events. *SD: adjusted standard deviation calculated across both treatment arms:

LDL-C: 0.73 mmol/l (0.28 g/l), non-HDL-C: 0.89 mmol/l (0.34 g/l), apoB: 0.20 g/l, LDL-P:

0.34 μmol/l and LDL size: 0.59 nm.

Figure 2. Comparison of the predictive strengths of HDL-related measures, given LDL-P, for

vascular events (all participants).*SD: adjusted standard deviation calculated across both

treatment arms: HDL-C: 0.29 mmol/l (0.11 g/l), apoA1: 0.18 g/l, HDL-P: 5.18 μmol/l and

HDL size: 0.44 nm.

Figure 3. Comparison of the predictive strengths of HDL-related measures, given LDL-P, for

vascular events with and without adjustment for N-BNP (all participants). Major coronary

event is the combination of non-fatal MI or CHD death, death from heart failure (with

underlying cause CHD) and sudden death.


Dow

nloaded from



Dow

nloaded from


Armitage and Rory CollinsSarah Parish, Alison Offer, Robert Clarke, Jemma C. Hopewell, Michael R. Hill, James Otvos, Jane

Protection StudyLipids and Lipoproteins and Risk of Different Vascular Events in the MRC/BHF Heart

Print ISSN: 0009-7322. Online ISSN: 1524-4539 Copyright © 2012 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Circulation published online April 26, 2012;Circulation.

http://circ.ahajournals.org/content/early/2012/04/20/CIRCULATIONAHA.111.073684World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circ.ahajournals.org/content/suppl/2013/10/14/CIRCULATIONAHA.111.073684.DC2 http://circ.ahajournals.org/content/suppl/2012/04/20/CIRCULATIONAHA.111.073684.DC1

Data Supplement (unedited) at:

http://circ.ahajournals.org//subscriptions/

is online at: Circulation Information about subscribing to Subscriptions:

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

document. Permissions and Rights Question and Answer available in the

Permissions in the middle column of the Web page under Services. Further information about this process isOnce the online version of the published article for which permission is being requested is located, click Request

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Circulation Requests for permissions to reproduce figures, tables, or portions of articles originally published inPermissions:


Dow

nloaded from

http://circ.ahajournals.org/content/early/2012/04/20/CIRCULATIONAHA.111.073684

http://circ.ahajournals.org/content/suppl/2012/04/20/CIRCULATIONAHA.111.073684.DC1

http://circ.ahajournals.org/content/suppl/2013/10/14/CIRCULATIONAHA.111.073684.DC2

http://www.ahajournals.org/site/rights/

http://www.lww.com/reprints

http://circ.ahajournals.org//subscriptions/


SUPPLEMENTAL MATERIAL Table of Contents Supplemental Methods and Results Supplemental Tables Supplemental Table I Supplemental Table II Supplemental Table III Supplemental Table IV Supplemental Figures Supplemental Figure I Supplemental Figure II

Supplemental Figure IIISupplemental Figure IV

SUPPLEMENTAL METHODS AND RESULTS Effects of non-compliance to allocated statin treatment on risk associations Supplemental Table A1 shows the percentage of participants who were non-

compliant with their study statin allocation by three years after randomization,

subdivided by baseline quintile of the four main LDL-related lipid measures in this

report.

In the statin arm, there were no steady trends in the percentages of participants not

on statin with quintiles of the LDL-related measures.

In the placebo arm, statin use increased steadily with baseline quintiles of the LDL-

related measures. The difference in statin use between the top and bottom quintile

was greatest for quintiles of non-HDL-C (25.6% [SE1.1]) and least for quintiles of

LDL-P (21.3% [SE1.2]; p<0.0001 for the additional association of statin use with non-

HDL-C given LDL-P; Supplemental Table A1). A calculated LDL-C (based on total

cholesterol, HDL-C and triglycerides), total cholesterol or non-HDL-C are the

measures most likely to have been used in clinical practice in the UK at the time of

the Heart Protection Study, since other measures would not generally have been

available. The relative strength of the trends with apoB and LDL-P may largely reflect

the strength of their correlations with non-HDL-C (main text Table 2).

Use of non-study statin in the placebo arm would lower risk. Higher non-study statin

use at higher quintiles of LDL-related measures would lower risk to a greater extent,

thus attenuating the observed association of risk with LDL-related measures. As this

happened to varying extents with respect to the different LDL-related measures, the

extent to which their observed associations will have been attenuated will vary and

so introduce a bias in their comparison. Non-study statin use may depend not only

on LDL-related lipid levels but also on other aspects of existing disease, disease

progression or risk, and, hence, it is not possible to quantify reliably the impact of

differential non-compliance.

Supplemental Table A1: Variation in statin use by quintiles of baseline LDL-related measures

Statin arm Percentage (SE) of participants not on statin at 3 years*

by quintile of baseline on-statin measure (n=9294)

Placebo arm Percentage (SE) of participants on statin at 3 years*

by quintile of baseline off-statin measure (n=9079))

Quintile LDL-C Non-HDL-C ApoB LDL-P LDL-C Non-HDL-C ApoB LDL-P 1 17.1 (0.9) 15.9 (0.9) 16.3 (0.9) 17.0 (0.9) 3.8 (0.5) 3.3 (0.4) 3.6 (0.4) 5.5 (0.5) 2 14.2 (0.8) 15.6 (0.8) 15.7 (0.8) 15.1 (0.8) 10.8 (0.7) 10.3 (0.7) 9.6 (0.7) 10.4 (0.7) 3 14.5 (0.8) 14.2 (0.8) 14.3 (0.8) 15.3 (0.8) 16.6 (0.9) 15.3 (0.8) 17.1 (0.9) 15.6 (0.9) 4 14.4 (0.8) 14.4 (0.8) 13.8 (0.8) 14.3 (0.8) 20.2 (0.9) 22.0 (1.0) 21.7 (1.0) 21.3 (1.0) 5 19.1 (0.9) 19.4 (0.9) 19.4 (0.9) 17.8 (0.9) 28.2 (1.1) 28.9 (1.1) 27.8 (1.1) 26.8 (1.0)

Difference top minus bottom quintile 24.4 (1.1) 25.6 (1.1) 24.2 (1.1) 21.3 (1.2) χ1

2 for trend No steady trend

457 549 514 371 * Study or non-study statin use amongst those who had not had a major occlusive coronary event prior to this during the trial, and with imputation of missing compliance based on last known statin use from earlier follow-ups.

Supplemental Table I: Numbers (%) of participants suffering various events by baseline history of prior disease

All patients(20021)

CHD(13077)

CVD(3194)

PVD(6583)

Diabetes or hypertension

only(3041)

ACE inhibitor, diuretic or

digoxin used(7058)

N-BNP > 2000 pg/mL

(7065)

Major occlusive coronary event 1796 (9.0%) 1359 (10.4%) 333 (10.4%) 701 (10.6%) 153 (5.0%) 757 (10.7%) 905 (12.8%)Revascularization 2187 (10.9%) 1567 (12.0%) 333 (10.4%) 1080 (16.4%) 123 (4.0%) 747 (10.6%) 822 (11.6%)Other cardiac event 1043 (5.2%) 841 (6.4%) 179 (5.6%) 455 (6.9%) 62 (2.0%) 683 (9.7%) 737 (10.4%)Ischemic stroke (including stroke of unknown origin) 995 (5.0%) 591 (4.5%) 327 (10.2%) 406 (6.2%) 118 (3.9%) 415 (5.9%) 477 (6.8%)

Prior disease in the HPS trial inclusion criteria* Indicators of heart failure and arrhythmia

Type of vascular event

* CHD coronary heart disease; CVD cerebrovascular disease; PVD peripheral vascular disease; participants may be included in several of these categories; diabetes or hypertension only denotes no prior CHD, CVD or PVD.

Supplemental Table II: Mean and standard deviation (SD) of the baseline lipid measurements* across both treatment arms and self-correlation between baseline and follow-up measurements in participants on and off statin

Mean SD‡ on statin** off statin***

LDL-C 2989 734 0.63 0.74Non-HDL-C 3862 888 0.63 0.78Apo B 1.86 0.39 0.66 0.74LDL-P 1.25 0.34 0.67 0.75LDL-Psmall 0.68 0.36 0.70 0.74LDL-Plarge 0.44 0.24 0.63 0.73IDL-P 0.12 0.09 0.26 0.35LDL size 20.71 0.59 0.66 0.76

Total cholesterol 4914 883 0.62 0.76VLDL-P 0.07 0.03 0.64 0.68LDL-P+VLDL-P 1.32 0.35 0.67 0.76

HDL-C 1053 289 0.82 0.83ApoA1 43.04 6.47 0.63 0.67HDL-P 33.00 5.18 0.70 0.73HDL-Psmall 19.52 4.73 0.61 0.60HDL-Pmedium 9.09 4.79 0.60 0.59HDL-Plarge 4.39 2.46 0.83 0.83HDL size 8.93 0.44 0.80 0.78

†All concentrations have been expressed in molar units (μmol/l) as this allows the relative numbers of different types of particles or molecules to be readily seen. Thus, for example, the mean LDL-C of 2989 μmol/l and the mean LDL-P of 1.25 μmol/l indicates and average of 2989/1.25 cholesterol molecules per LDL particle. To convert concentrations to g/l, multiply cholesterol fractions by 3.87x10-4 g/μmol, apo B by 0.513 g/μmol and apo A1 by 0.028 g/µmol.‡Standard deviation across both treatment arms after adjustment for age, sex, simvastatin and vitamin allocation, smoking, prior disease, SBP, eGFR, medication and N-BNP group.**The self-correlation is the correlation coefficient between the randomization and follow-up measurements in samples taken 2-3 years apart in participants while taking statin, adjusted for age, sex, vitamin allocation and assay type.***The self-correlation is the correlation coefficient between the screening and follow-up measurements in samples taken 2-3 years apart in participants while not taking statin, adjusted for age, sex, vitamin allocation and assay type .

LDL-related factors

HDL-related factors

Particle concentrations (μmol/l†) and sizes (nm) Self correlation

*Baseline measurements are from the randomization sample (on statin) for participants in the statin arm and from the screening sample (off statin) for participants in the placebo arm.

§VLDL-P indicates very low density and chylomicron particle number (particles of diameter > 27nm).

§

§

Supplemental Table III: Association of other cardiac event with HDL-related factors given LDL-P, in all participants*

HR (95%CI) per 1SD† χ12 p value HR (95%CI) per 1SD† χ1

2 p value

0.94 (0.88-1.00) 3.5 0.06 0.96 (0.90-1.03) 1.4 0.240.92 (0.86-0.98) 6.6 0.01 0.90 (0.84-0.96) 10.4 0.001

0.84 (0.79-0.90) 25.7 <0.0001 0.79 (0.74-0.85) 47.8 <0.0001HDL size 1.07 (1.00-1.15) 3.5 0.06 1.16 (1.08-1.24) 17.5 <0.0001HDL size given HDL-P 1.13 (1.05-1.21) 10.9 0.001 1.24 (1.16-1.33) 36.3 <0.0001

HDL-Psmall 0.91 (0.85-0.98) 6.3 0.01 0.82 (0.76-0.88) 31.3 <0.0001HDL-Pmedium 0.89 (0.83-0.96) 9.3 0.002 0.90 (0.83-0.97) 7.8 0.005HDL-Plarge 0.97 (0.84-1.13) 0.1 0.71 1.10 (0.95-1.27) 1.7 0.19

HDL-Psmall 0.82 (0.76-0.89) χ3

2 0.72 (0.66-0.78) χ32

HDL-Pmedium 0.81 (0.74-0.89) 30.2 <0.0001 0.76 (0.70-0.83) 71.3 <0.0001HDL-Plarge 0.98 (0.84-1.14) 1.06 (0.91-1.22)

χ22 χ2

2

4.5 0.11 23.5 <0.0001

HDL-P

† All SDs are as in figure 2

Adjusted for N-BNP Not adjusted for N-BNP

§Calculated from the difference between the χ32 for the subclasses jointly and the χ1

2 for HDL-P alone

Additional predictive value of subclasses over HDL-P§

‡ Hazard ratios for all HDL subclasses are per 5.18 μmol/l (1SD of HDL-P)

* Analyses are adjusted for simvastatin and vitamin allocation, age, sex, smoking, prior disease, SBP, eGFR, medication and LDL-P

HDL subclasses ‡ jointly

HDL subclasses ‡ singly

HDL-CApoA1

Supplemental Table IV: Influence of adjustments on the estimated associations between major occlusive coronary event and selected lipid-related measures

LDL-related measuresStatin armAdjustments HR (95% CI) per 1SD χ1² HR (95% CI) per 1SD χ1²Age, sex, simvastatin and vitamin allocation 1.23 (1.14-1.32) 28.3 1.26 (1.17-1.36) 35.0+smoking 1.22 (1.13-1.31) 26.1 1.25 (1.16-1.35) 32.7+SBP 1.22 (1.13-1.31) 25.9 1.25 (1.16-1.35) 32.5+eGFR 1.21 (1.12-1.30) 24.1 1.25 (1.16-1.34) 31.0+prior disease* 1.20 (1.11-1.29) 21.6 1.21 (1.12-1.31) 23.6+medication 1.20 (1.12-1.29) 22.6 1.22 (1.13-1.32) 26.0+N-BNP 1.22 (1.14-1.32) 27.9 1.25 (1.16-1.35) 32.3

Placebo armAdjustments HR (95% CI) per 1SD χ1² HR (95% CI) per 1SD χ1²Age, sex, simvastatin and vitamin allocation 1.10 (1.04-1.16) 12.1 1.12 (1.06-1.18) 17.4+smoking 1.09 (1.04-1.16) 10.3 1.11 (1.06-1.17) 15.2+SBP 1.09 (1.03-1.15) 9.6 1.11 (1.05-1.17) 14.4+eGFR 1.07 (1.02-1.13) 6.2 1.10 (1.04-1.16) 11.4+prior disease* 1.07 (1.01-1.13) 5.3 1.09 (1.03-1.15) 9.1+medication 1.07 (1.01-1.13) 5.7 1.09 (1.03-1.15) 9.2+N-BNP 1.09 (1.03-1.15) 8.6 1.11 (1.05-1.17) 13.2

HDL-related measures, given LDL-PAll participantsAdjustments HR (95% CI) per 1SD χ1² HR (95% CI) per 1SD χ1²Age, sex, simvastatin and vitamin allocation 0.87 (0.83-0.92) 27.1 0.83 (0.79-0.87) 59.3+smoking 0.87 (0.83-0.92) 26.2 0.83 (0.79-0.87) 56.0+SBP 0.87 (0.83-0.92) 27.6 0.83 (0.79-0.87) 60.9+eGFR 0.88 (0.84-0.93) 22.4 0.84 (0.80-0.88) 52.2+prior disease* 0.93 (0.88-0.98) 8.5 0.87 (0.83-0.91) 31.5+medication 0.92 (0.87-0.97) 10.3 0.87 (0.83-0.91) 32.6+N-BNP 0.91 (0.86-0.96) 13.2 0.89 (0.85-0.93) 22.5

*Baseline history of CHD, PVD, CVD, hypertension or diabetes (see methods for details of terms included)

HDL-P given LDL-PHDL-C given LDL-P

Non-HDL-C LDL-P

LDL−CNon−HDL−CApo BLDL−PLDL size, given LDL−P

Total CholesterolVLDL−P*LDL−P+VLDL−P*


Major occlusive coronary eventor revascularization

Major coronary event

1580 events

946 events

0.6 0.8 1 1.2 1.4

Hazard ratio (95% CI)per 1SD† higher

χ1 2

Statin arm

1.20 (1.14−1.26) 43.41.19 (1.13−1.25) 42.01.19 (1.13−1.25) 41.31.19 (1.12−1.25) 38.50.97 (0.91−1.02) 1.5 1.14 (1.08−1.20) 23.01.09 (1.04−1.15) 11.51.19 (1.13−1.25) 39.9

1.23 (1.15−1.32) 34.21.21 (1.14−1.29) 31.41.22 (1.14−1.30) 32.31.23 (1.15−1.32) 35.11.01 (0.94−1.08) 0.0

2010 events

1244 events

0.6 0.8 1 1.2 1.4

Hazard ratio (95% CI)per 1SD† higher

χ1 2

Placebo arm

1.12 (1.08−1.17) 33.91.13 (1.08−1.17) 33.61.14 (1.09−1.19) 38.11.15 (1.10−1.20) 46.20.99 (0.94−1.03) 0.4 1.09 (1.05−1.14) 18.11.06 (1.02−1.10) 7.71.15 (1.10−1.20) 46.4

1.08 (1.03−1.14) 10.01.08 (1.03−1.14) 8.81.09 (1.04−1.15) 10.91.09 (1.03−1.14) 10.31.03 (0.98−1.09) 1.2

Supplemental Figure I: Comparison of the predictive strengths of LDL-related measures for major occlusive coronary event or revascularization and for major coronary event.

*VLDL-P indicates very low density and chylomicron particle number (particles of diameter > 27nm)

†SD: adjusted standard deviation calculated across both treatment arms (LDL-C: 0.73 mmol/l (0.28 g/l), non-HDL-C: 0.89 mmol/l (0.34 g/l),apoB: 0.20 g/l, total cholesterol: 0.88mmol/l (0.34g/l), LDL-P: 0.34 µmol/l, VLDLP: 0.03 µmol/l, LDL-P+VLDLP: 0.35 µmol/l and LDL size: 0.59 nm)

HDL−CApo A1

HDL−PHDL size, given HDL−P

HDL−CApo A1


HDL−CApo A1


HDL−CApo A1


Major occlusive coronary event

Revascularization

Other cardiac event

Ischemic stroke

757 events

970 events

505 events

430 events

0.6 0.8 1 1.2 1.4Hazard ratio (95% CI)

per 1SD* higher

χ12

Statin arm

0.88 (0.81−0.96) 9.20.89 (0.82−0.96) 9.60.89 (0.83−0.96) 10.31.02 (0.93−1.11) 0.1

0.91 (0.85−0.98) 6.40.99 (0.93−1.06) 0.10.97 (0.91−1.03) 0.90.91 (0.84−0.99) 4.6

0.92 (0.83−1.01) 2.90.88 (0.80−0.96) 7.70.80 (0.73−0.88) 23.41.19 (1.07−1.32) 10.8

0.95 (0.86−1.05) 1.00.94 (0.85−1.04) 1.50.92 (0.84−1.01) 2.91.10 (0.98−1.23) 2.6

1039 events

1217 events

538 events

565 events

0.6 0.8 1 1.2 1.4Hazard ratio (95% CI)

per 1SD* higher

χ12

Placebo arm

0.92 (0.86−0.99) 5.20.93 (0.87−0.99) 4.70.88 (0.82−0.94) 13.31.01 (0.94−1.09) 0.1

0.90 (0.84−0.96) 9.80.94 (0.89−1.01) 3.20.93 (0.87−0.99) 5.70.99 (0.92−1.06) 0.1

0.95 (0.86−1.05) 1.00.95 (0.87−1.05) 1.00.89 (0.81−0.98) 5.51.07 (0.97−1.18) 1.7

0.98 (0.90−1.07) 0.21.00 (0.92−1.09) 0.00.96 (0.88−1.05) 0.91.02 (0.92−1.12) 0.1

Supplemental Figure II: Comparison of the predictive strengths of HDL-related measures, given LDL-P, for vascular events *SD: adjusted standard deviation calculated across both treatment arms (HDL-C: 0.29 mmol/l (0.11 g/l), apoA1: 0.18 g/l, HDL-P: 5.18 µmol/l and HDL size: 0.44 nm)






Revascularization

Other cardiac event

Ischemic stroke

637 events

807 events

412 events

322 events

0.6 0.8 1 1.2 1.4

Hazard ratio (95% CI)per 1SD* higher

χ12

Statin arm: men

1.24 (1.14−1.35) 22.51.22 (1.12−1.32) 20.91.22 (1.12−1.33) 21.61.24 (1.14−1.35) 24.30.99 (0.91−1.09) 0.0

1.19 (1.10−1.28) 18.41.17 (1.09−1.26) 16.91.16 (1.08−1.25) 15.61.14 (1.06−1.23) 11.40.98 (0.90−1.06) 0.3

1.01 (0.90−1.13) 0.01.00 (0.89−1.11) 0.01.05 (0.94−1.17) 0.71.05 (0.94−1.17) 0.80.96 (0.86−1.07) 0.6

1.11 (0.98−1.26) 2.61.14 (1.01−1.28) 4.21.15 (1.03−1.30) 5.51.08 (0.95−1.22) 1.40.95 (0.83−1.07) 0.8

120 events

163 events

93 events

108 events

0.6 0.8 1 1.2 1.4

Hazard ratio (95% CI)per 1SD* higher

χ12

Statin arm: women

1.28 (1.08−1.52) 7.71.23 (1.04−1.44) 5.61.27 (1.08−1.49) 7.51.26 (1.06−1.49) 6.70.90 (0.74−1.09) 1.1

1.12 (0.96−1.30) 2.01.14 (0.99−1.32) 3.01.13 (0.97−1.31) 2.51.16 (1.00−1.35) 3.50.83 (0.71−0.98) 4.8

1.14 (0.92−1.41) 1.41.08 (0.88−1.33) 0.51.11 (0.90−1.36) 0.91.20 (0.98−1.47) 2.90.99 (0.79−1.23) 0.0

1.14 (0.95−1.37) 1.91.06 (0.88−1.27) 0.41.06 (0.88−1.28) 0.41.06 (0.88−1.29) 0.41.05 (0.86−1.29) 0.2

Supplemental Figure III: Comparison of the predictive strengths of LDL-related measures for vascular events in the statin allocated arm, subdivided by sex. *SD: adjusted standard deviation calculated across both treatment arms (LDL-C: 0.73 mmol/l (0.28 g/l), non-HDL-C: 0.89 mmol/l (0.34 g/l), apoB: 0.20 g/l, LDL-P: 0.34 µmol/l and LDL size: 0.59 nm)

HDL−CApo A1


HDL−CApo A1


HDL−CApo A1


HDL−CApo A1



Revascularization

Other cardiac event

Ischemic stroke

1508 events

1845 events

837 events

755 events

0.6 0.8 1 1.2 1.4 Hazard ratio (95% CI)

per 1SD* higher

χ12

Men

0.91 (0.85−0.96) 10.20.91 (0.86−0.96) 10.90.89 (0.85−0.94) 16.41.01 (0.95−1.07) 0.1

0.90 (0.85−0.95) 15.10.96 (0.91−1.01) 2.50.95 (0.90−0.99) 4.80.96 (0.91−1.02) 1.5

0.93 (0.85−1.00) 3.70.91 (0.84−0.98) 5.60.85 (0.79−0.92) 17.81.12 (1.04−1.21) 8.0

0.98 (0.91−1.07) 0.20.99 (0.91−1.06) 0.10.95 (0.89−1.03) 1.51.04 (0.95−1.13) 0.7

288 events

342 events

206 events

240 events

0.6 0.8 1 1.2 1.4 Hazard ratio (95% CI)

per 1SD* higher

χ12

Women

0.91 (0.82−1.02) 2.50.93 (0.84−1.04) 1.70.87 (0.77−0.98) 5.81.05 (0.92−1.20) 0.5

0.95 (0.85−1.05) 1.21.00 (0.92−1.10) 0.00.97 (0.88−1.08) 0.30.94 (0.83−1.06) 1.0

0.97 (0.85−1.11) 0.20.93 (0.81−1.06) 1.20.80 (0.69−0.93) 9.21.14 (0.98−1.33) 2.7

0.95 (0.85−1.07) 0.60.96 (0.86−1.08) 0.40.91 (0.80−1.03) 2.41.11 (0.96−1.28) 2.0

Supplemental Figure IV: Comparison of the predictive strengths of HDL-related measures, given LDL-P, for vascular events, subdivided by sex (all participants) *SD: adjusted standard deviation calculated across both treatment arms (HDL-C: 0.29 mmol/l (0.11 g/l), apoA1: 0.18 g/l, HDL-P: 5.18 µmol/l and HDL size: 0.44 nm)

166

summary

배경

Low density lipoprotein cholesterol(LDL-C)과 high

density lipoprotein cholesterol(HDL-C)은 죽상경화성

혈관질환 발생의 주요 위험인자이다. 하지만 지질단백

입자의 농도가 더 중요한 위험인자일 수 있다는 역학연

구 결과들도 있다.

방법 및 결과

기저 지질단백 수치와 apo B, apo A1의 농도 그리고 핵

자기공명(nuclear magnetic resonance)을 이용하여 평

가한 지질단백 입자의 자료를 이용하여 심혈관질환 발

생 위험도를 측정하였다. 20,000명의 환자를 심바스

타틴 투여군과 대조군으로 무작위 배정하여 5.3년 동

안 5,000건 이상의 심혈관사건 발생을 연구한 Heart

Protection Study의 자료를 연구 분석에 이용하였다. 폐

색성 관상동맥질환의 발생은 LDL-C 농도, 입자와 높은

연관성을 보였다. 표준편차 정도의 변화가 질환 발생에

미치는 연관 위험도를 분석한 결과, LDL-C은 1.25, non-

HDL-C은 1.22, apo B는 1.23, LDL 입자 개수는 1.25의

위험도를 보였다. 혈관 재개통의 필요성도 비슷한 연관

성을 보였으나, 허혈성 뇌졸중이나 심부전 등에 대해서

는 유의한 연관성을 보이지 않았다. LDL 입자 개수를 보

정한 후에 분석한 폐색성 관상동맥질환의 발생 위험도

를 보면 HDL-C은 0.91, HDL 입자 개수는 0.89였다. 심부

전을 포함한 기타 심장질환의 발생은 HDL-C 농도와 유

의한 연관성을 보이지 않았으나, 총 HDL 입자 개수의

경우 0.84, small HDL 입자 개수의 경우 0.82로 역의 상

관관계를 보였다.

결론

해마다 평균 2%의 관상동맥질환 발생 위험도를 지닌

환자군에서 콜레스테롤 농도 수치, 아포지질단백 그리

고 LDL 입자 개수는 폐색성 관상동맥질환 발생을 예측

하는 유의한 위험인자로서 비슷한 정도의 예측력을 보

였다. HDL 입자 개수와 심부전을 포함한 기타 심혈관질

환 발생 위험의 인과관계는 향후에 추가로 규명할 필요

가 있다.

지질단백 입자 정보는 심혈관질환 발생 예측에 중요하다: 향후 표준 예측인자로서의 가능성

김상현교수서울대학교 보라매병원 순환기내과

167Lipid

commentary

LDL-C과 HDL-C은 죽상경화성 심혈관질환 발생의 주요

위험인자이고, LDL-C을 낮추는 것은 심혈관질환 발생을

유의하게 감소시킨다고 알려져 있다. 하지만 세부 항목

을 들여다보면 결론은 그렇게 간단하지 않다.

우선 무작위배정 치료 연구들의 결과를 보면, 스타틴

을 투여하여 LDL-C을 낮추면 관상동맥질환 사망률, 심

근경색 그리고 혈관 재개통술의 필요성은 감소시키지

만, 급성 심장사나 부정맥 혹은 심부전 등 기타 심장질

환에 의한 사망률에 대해서는 효과가 뚜렷하지 않다. 또

한, 역학연구에서 허혈성 뇌졸중은 관상동맥질환에 비

해 LDL-C과 약한 연관성을 보이나, 무작위배정 치료 연

구에서 스타틴 투여에 따른 LDL-C 저하는 정도의 차이

는 있지만 허혈성 뇌경색을 유의하게 감소시킨다.1

HDL-C과 심혈관질환의 연관성은 더욱 복잡한 결론을

보인다. HDL-C은 역학연구에서 관상동맥질환의 발생과

역의 상관관계를 보이나, apo B, non-HDL-C과 역의 상

관관계로 연관되어 있기에 LDL-C의 영향을 받을 수밖

에 없다. 따라서 LDL-C, HDL-C 농도 수치 이외에 심혈관

질환 발생을 예측하는 데 유용한 추가적인 위험인자 혹

은 생화학적 지표의 필요성이 대두되었다.

LDL과 HDL의 입자에 포함된 콜레스테롤 함량은 개인

별로 차이가 크고, 이는 죽상경화 위험도에 차이를 가져

오므로, 아포지질단백을 측정하거나 입자 크기 혹은 입

자 개수를 핵자기공명이나 전기영동으로 측정하는 것

이 추가적인 위험인자로서 중요하다고 알려져 있다. 하

지만 HDL 입자의 경우 크기에 따른 위험도가 여러 연구

에서 상반된 결과를 보였다.2

이 연구에서는 LDL 입자 개수나 apoB, 그리고 non-

HDL-C이 위험도 예측력이 매우 뛰어남을 보여주고 있

으며, 특히 폐색성 관상동맥질환에서 예측력이 뛰어남

을 보여주고 있다. 뇌졸중이나 기타 심장질환의 경우 추

가적인 정보를 제공할 수 있는 다른 위험인자의 발견이

필요함을 보여준다. 따라서 중성지방이 높은 대사증후

군이나 당뇨병에서뿐만 아니라, 중간 위험군에서도 지

질단백 콜레스테롤 농도 수치 이외의 지질단백 입자 개

수나 크기가 중요한 위험인자로서 만족할만한 심혈관

질환 발생 예측력을 얻을 수 있다. 더구나 지질단백 입

자 개수나 아포지질단백의 측정은 생체 내의 지질대사

상태를 좀 더 기전적으로 잘 반영할 수 있다. 향후 지질

단백 입자 개수나 아포지질단백의 측정에 관한 기술이

더 발전되어 임상 진료에서 일반화될 수 있을 정도의 간

결성과 저렴한 검사 비용 등의 환경이 제공된다면, 진료

지침에서 일상적인 측정과 환자 진료에의 이용을 권유

할 수 있을 것이다.

ReferencesCholesterol Treatment Trialists’ (CTT) Collaboration, Baigent C, Blackwell L, 1. Emberson J, Holland LE, Reith C, Bhala N, Peto R, Barnes EH, Keech A, Simes J, Collins R. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet. 2010;376:1670-1681. Barter P, Kastelein J, Nunn A, Hobbs R; Future Forum Editorial Board. High density 2. lipoproteins (HDLs) and atherosclerosis; the unanswered questions. Atherosclerosis. 2003;168:195-211.

168

Vascular Medicine

Lipids and Lipoproteins and Risk of Different VascularEvents in the MRC/BHF Heart Protection Study

Sarah Parish, DPhil; Alison Offer, PhD; Robert Clarke, FRCP; Jemma C. Hopewell, PhD;Michael R. Hill, DPhil; James D. Otvos, PhD; Jane Armitage, FRCP; Rory Collins, FMedSci;

on behalf of the Heart Protection Study Collaborative Group

Background—Low-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol are establishedrisk factors for vascular disease, but lipoprotein particle concentrations may be stronger determinants of risk.

Methods and Results—Associations between vascular events and baseline concentrations of cholesterol fractions,apolipoproteins B and A1, and lipoprotein particles assessed by nuclear magnetic resonance were considered in the HeartProtection Study randomized trial of simvastatin versus placebo (�5000 vascular events during 5.3 years of follow-upamong 20 000 participants). Major occlusive coronary events were equally strongly associated with the cholesterol- andparticle-based total LDL measures; adjusted hazard ratios per 1-SD-higher level were 1.25 (95% confidence interval[CI], 1.16–1.34) for LDL cholesterol, 1.22 (95% CI, 1.14–1.32) for non–HDL cholesterol, 1.23 (95% CI, 1.15–1.33)for apolipoprotein B, and 1.25 (95% CI, 1.16–1.35) for LDL particle number. Given the total LDL particle number, thedistribution between small and large particles did not add predictive value. Associations of these different LDL-relatedmeasures were similar with arterial revascularization procedures but much weaker or nonexistent with ischemic strokeand other cardiac events (mainly heart failure). After adjustment for LDL particle number, the hazard ratios for majorocclusive coronary event per 1-SD-higher level were 0.91 (95% CI, 0.86–0.96) for HDL cholesterol and 0.89 (95% CI,0.85–0.93) for HDL particle number. Other cardiac events were inversely associated with total (hazard ratio, 0.84; 95%CI, 0.79–0.90) and small (0.82; 95% CI, 0.76–0.89) HDL particle number but only very weakly associated with HDLcholesterol (0.94; 95% CI, 0.88–1.00).

Conclusions—In a population at 2% average coronary event risk per year, cholesterol, apolipoprotein, and particlemeasures of LDL were strongly correlated and had similar predictive values for incident major occlusive vascularevents. It is unclear whether the associations between HDL particle numbers and other cardiac events represent a causalor reverse-causal effect.

Clinical Trial Registration—URL: http://www.isrctn.org/. Unique identifier: ISRCTN48489393.(Circulation. 2012;125:2469-2478.)

Key Words: coronary disease � lipids � lipoproteins � lipoproteins, HDL � lipoproteins, LDL � risk assessment

Large observational studies indicate strong positive asso-ciations between low-density lipoprotein (LDL) parti-

cles, which carry cholesterol, and the risk of coronary heartdisease (CHD).1,2 Randomized trials have demonstrated thatlowering LDL cholesterol (LDL-C) with statins reduces therisk of CHD death, nonfatal myocardial infarction (MI),ischemic stroke, and the need for revascularization proce-dures, with less benefit apparent on deaths resulting fromother cardiac disease (including sudden death, arrhythmia,and heart failure).3–5 Observational studies, however, havetypically reported weaker associations of LDL-C (or relatedmeasures) with ischemic stroke than with CHD,1,2 which is

discordant with the findings of randomized trials in which thebenefits of LDL lowering were similar for CHD and stroke.3,6

Clinical Perspective on p

High-density lipoprotein (HDL) cholesterol (HDL-C) isinversely associated with CHD in observational studies(largely among individuals without prior CHD).1,2 However,HDL-C is negatively correlated with apolipoprotein B (apoB)and with non–HDL-C2; hence, some of its predictive value isnot independent of these LDL-related measures. In the largeEmerging Risk Factors Collaboration (ERFC) meta-analysis

Received October 27, 2011; accepted March 19, 2012.From the Clinical Trial Service Unit and Epidemiological Studies Unit, University of Oxford, Oxford, UK (S.P., A.O., R.C., J.C.H., M.R.H., J.A., R.C.),

and LipoScience Inc, Raleigh, NC (J.D.O.).The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.

111.073684/-/DC1.Correspondence to Sarah Parish, DPhil, Heart Protection Study, Clinical Trial Service Unit and Epidemiological Studies Unit, Richard Doll Bldg, Old

Road Campus, Roosevelt Drive, Oxford OX3 7LF, UK. E-mail [email protected]© 2012 American Heart Association, Inc.

Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.111.073684

2469 by IMED Korea on July 31, 2012http://circ.ahajournals.org/Downloaded from

169

of prospective observational studies of individuals withoutprior vascular disease, the lower CHD risk associated with1-SD-higher HDL-C attenuated from 29% to 22% afteradjustment for non–HDL-C and was only about half the sizeof the 56% higher risk associated with 1-SD-highernon–HDL-C.2

The extent to which LDL and HDL particles influence riskthrough atherosclerosis, inflammation, and other postulatedmechanisms may vary according to subtypes of lipoproteinparticle.7–9 Small dense LDL particles are believed to beparticularly hazardous, but it is unclear whether these asso-ciations are independent of other lipid measures.8,10–12 Stud-ies comparing different HDL subtypes (based on size, den-sity, or apolipoprotein content)13 have yielded conflictingresults about the relevance of different subtypes.14 Discor-dance in results may be attributable to small study sizes andvariable allowance for associations with the LDL system.Preexisting disease in participants may also be a confoundingfactor because cardiac disease can reduce the functionality ofHDL particles.15

LDL and HDL are traditionally quantified by measuringthe cholesterol they contain (LDL-C and HDL-C), but inter-individual variations in the cholesterol content of LDL andHDL particles could mean that alternative analytic measuresare more informative.16–18 One approach is to measure thepredominant protein moiety on the particles: apoB for LDLand apolipoprotein A1 (apoA1) for HDL. LDL particles andtheir precursors each carry 1 molecule of apoB on theirsurface (with �90% of apoB being on LDL particles); thus,plasma apoB concentrations provide a good estimate of LDLparticle concentrations. Plasma apoA1 concentrations, incontrast, are not proportional to HDL particle concentrationsbecause HDL particles contain variable numbers (2–5) ofapoA1 molecules. Nuclear magnetic resonance (NMR) spec-troscopy offers another way to quantify lipoproteins, provid-ing particle concentrations of LDL (LDL-P) and HDL(HDL-P) and their subclasses.19 The present analyses inves-tigate the associations of cholesterol, apolipoprotein, andparticle measures of LDL and HDL, as well as subclasslevels, with �5000 vascular outcomes in 20 000 high-riskindividuals during 5 years of follow-up in the MRC/BHFHeart Protection Study (HPS).

MethodsRecruitment and Eligibility CriteriaDetails of the HPS have been reported previously.20,21 Between 1994and 1997 in the United Kingdom, 20 536 men and women 40 to 80years of age were recruited and assigned randomly to receive 40 mgsimvastatin daily or matching placebo (and separately, in a 2�2factorial design, to receive antioxidant vitamins or placebo capsules).To be eligible, participants had a nonfasting blood total cholesterolconcentration of at least 3.5 mmol/L (135 mg/dL) and either had aprevious diagnosis of CHD, cerebrovascular disease, other occlusivedisease of noncoronary arteries, or diabetes mellitus (type I or II) orwere men �65 years of age undergoing treatment for hypertension.Individuals were excluded if their doctor considered statin therapyclearly indicated or contraindicated. At initial screening, writtenconsent to participate was obtained, and a nonfasting blood samplewas taken (screening sample); participants then began a run-in phaseinvolving 4 weeks of placebo followed by 4 to 6 weeks of 40 mgsimvastatin daily plus vitamin supplementation. At the end of this

run-in period, a nonfasting blood sample was taken (randomizationsample), and compliant and eligible individuals were randomlyassigned treatment for �5 years.

Laboratory MeasurementsBlood samples taken into heparinized vacutainers at screening andrandomization visits were chilled to �4°C and couriered overnightto the coordinating central laboratory for separation, assay, andlong-term storage in liquid nitrogen (��80°C). Analyses for thisreport were based on the screening samples for participants subse-quently randomly allocated to placebo-simvastatin at the randomiza-tion visit and on the randomization samples (after taking thesimvastatin regimen) for participants who were subsequently ran-domly allocated to active simvastatin. Plasma lipid fractions (includ-ing HDL-C, apoB, and apoA1), triglycerides, creatinine, N-terminalpro-B-type natriuretic peptide (N-BNP), and C-reactive protein wereassayed with previously reported methods.4,21,22 Because methodsfor LDL-C have progressed, LDL-C was reassayed with theN-geneous method (Genzyme Diagnostics).23 Lipoprotein particleprofiles were measured by NMR spectroscopy with the LipoProfile-3algorithm at LipoScience, Inc (Raleigh, NC). LDL and HDL particlesubclasses (�mol/L) were quantified from the amplitudes of theirspectroscopically distinct lipid methyl group NMR signals, andweighted-average LDL and HDL sizes were derived from the sum ofthe diameter of each subclass multiplied by its relative masspercentage based on the amplitude of its methyl NMR signal.19

Diameter range estimates for the subclasses were as follows: smallLDL-P, 18 to 21.2 nm; large LDL-P, 21.2 to 23 nm; intermediate-density lipoprotein particles, 23 to 27 nm; small HDL-P, 7.3 to 8.2nm; medium HDL-P, 8.2 to 8.8 nm; and large HDL-P, 8.8 to 13 nm.LDL-P and HDL-P are the totals of the particle number concentra-tions of the LDL and HDL subclasses, respectively. The glomerularfiltration rate was estimated with the Modification of Diet in RenalDisease formula.24 During the study, blood was collected annuallyfrom a random sample of participants attending follow-up. RepeatNMR and chemistry measurements in the 4000 participants withavailable samples taken while compliant with their allocated statintreatment allowed assessment of the reproducibility of each lipidmeasure.

Follow-Up and Determination of EventsParticipants were followed up for the incidence of events, includingMI, stroke, vascular procedures, death, and hospital admissions forother cardiac events (ascertained as previously reported in detail)20;this report includes a mean of 5.3 years of follow-up until November11, 2001, when study treatment ended. Because most deaths fromheart failure and sudden death were believed to have an underlyingcoronary cause, they had been included in the prespecified trialoutcome of major coronary event (nonfatal MI or coronary death).However, because some such deaths may have had nonocclusivecauses (eg, arrhythmia), the following outcome categories were usedin the present analyses: major occlusive coronary event, defined asnonfatal MI or coronary death other than death from heart failure orsudden death (corresponding to the revised definition of majorcoronary event in the recent Cholesterol Treatment Trialists’ meta-analyses3), and other cardiac events, defined as hospitalization ordeath resulting from heart failure, sudden death, or noncoronarycardiac death.

ComplianceAmong those allocated placebo, noncompliance resulting from theuptake of statin therapy increased during the course of the study andwas greater among those with higher baseline cholesterol levels (3%versus 29% at 3 years in the bottom versus top quintile of baselinenon–HDL-C). This bias would tend to attenuate the observedassociations of risk with LDL-related measures but not affect theassociations with HDL-related measures. Because a determinant ofsuch statin use may have been a high LDL-C or non–HDL-C level,use may be more directly related to cholesterol levels than to thelevels of other LDL-related measures (such as LDL-P or apoB) and

2470 Circulation May 22, 2012

by IMED Korea on July 31, 2012http://circ.ahajournals.org/Downloaded from

170

so lead to greater attenuation of associations with LDL-C andnon–HDL-C than of associations with LDL-P and apoB (see theonline-only Data Supplement). When considering comparisons be-tween LDL-related measures, we have therefore placed particularemphasis on the results in the statin arm to reduce the potential forsuch confounding.

Statistical AnalysisHazard ratios (HRs) were calculated for the first postrandomizationoccurrence of each outcome per 1-SD-higher level of the baselinelipid-related factor with Cox proportional hazards regression. Stan-dard adjustment in risk analyses included simvastatin and vitaminallocation, baseline 5-year age group by sex, prior history of vasculardisease or diabetes mellitus (as a 3-way variable for CHD [MI, otherCHD, no CHD] and a 5-way variable for cerebrovascular disease,peripheral vascular disease, and diabetes mellitus [each alone,multiple, none]), treated hypertension, medication use (angiotensin-converting enzyme inhibitors, anticoagulants, �-blockers, broncho-dilators, calcium antagonists, digoxin, diuretics, hormone replace-ment therapy, hypoglycemics, inhaled steroids, insulin), cigarettesmoking status (current, ex-smoker, never), systolic blood pressure,and estimated glomerular filtration rate as continuous variables, andN-BNP as a categorical variable with 5 groups (cut points: 400,1000, 2000, and 5000 pg/mL based around the ESC guidelines thatgrade N-BNP �400 pg/mL as heart failure unlikely and N-BNP�2000 pg/mL as heart failure likely25). The median value wassubstituted for the few missing values for any covariate (�0.7%).The SD of a lipid-related factor was estimated by the root meansquare error in a regression of the factor on the standard adjustmentterms. For LDL subclasses, the SD of the total particle concentrationwas used so that all the HRs were per the same unit of particlenumber, facilitating comparison of whether the hazard per particlevaried by subclass; likewise, hazard ratios for HDL subclasses werecalculated per SD of HDL-P. Likelihood ratio tests, yielding �2

statistics and corresponding P values, were used to assess evidenceof association after progressive inclusion of terms. Pearson correla-tion coefficients were used except in correlations involving N-BNPwhen Spearman were used. All analyses were based on the 20 021participants with complete information on concentrations of NMRparticles, cholesterol fractions, and apolipoproteins.

ResultsDuring a mean follow-up of 5.3 years, 1796 participants hada major occlusive coronary event, 2187 had a revasculariza-tion, 1043 had an “other” cardiac event, and 995 had apresumed ischemic stroke (Table 1). History of occlusivevascular disease and indicators of heart failure25,26 were majordeterminants of the incidence of vascular events (Table I inthe online-only Data Supplement), highlighting the impor-tance of adjusting for these factors in all analyses. The meanconcentrations of the directly measured chemistries and theNMR lipoprotein subclasses are shown in Table II in theonline-only Data Supplement, together with self-correlationsthat indicate the reproducibility of measurements in samplescollected from the same individual a few years apart. All ofthe LDL-related factors except intermediate-density lipopro-tein particle number showed similar self-correlations of�0.65 on statin and 0.75 off statin. The self-correlations ofthe different HDL-related factors showed greater variationbut were similar on and off statin (eg, off-statin self-correlations: 0.83, 0.67, 0.73, and 0.60 for HDL-C, apoA1,HDL-P, and small HDL-P, respectively).

Correlations Between LDL- andHDL-Related MeasuresThe 4 total LDL-related measures were strongly correlatedwith each other in measurements made in the same samples

(Table 2): non–HDL-C and apoB were the most stronglycorrelated (0.93), whereas LDL-P was correlated moststrongly with apoB (0.84) and least strongly with LDL-C(0.79). The 3 total HDL-related measures were less stronglycorrelated with each other (pairwise correlation coefficientsbetween 0.61 and 0.81; Table 2). The LDL- and HDL-relatedmeasures were considerably intercorrelated. LDL size andHDL size were positively correlated with each other (0.53);both were negatively correlated with LDL-P and triglyceridesand positively correlated with HDL-P. HDL-C was morestrongly negatively correlated with several LDL-related mea-sures than was HDL-P (eg, correlations of �0.34 and �0.10,respectively, with LDL-P and of �0.46 and 0.06 with logtriglycerides).

LDL-Related AssociationsFigure 1 compares the hazard ratios and �2 statistics for thestrengths of associations of LDL-C, non–HDL-C, apoB, andLDL-P for several different outcomes after adjustment for

Table 1. Numbers of Participants Suffering VariousVascular Events

Type of Vascular EventStatin Arm

(n�10 033), nPlacebo Arm(n�9988), n

Major occlusive coronary event*†

Nonfatal MI 388 598

Death resulting from MI 148 197

Death resulting from CHD (excludingheart failure and sudden death)

245 286

Subtotal 757 1039

Revascularization

Coronary revascularization 542 743

Noncoronary revascularization, aortic 67 62

Noncoronary revascularization, carotid 44 83

Noncoronary revascularization,peripheral

421 474

Subtotal 970 1217

Other cardiac event

Nonfatal heart failure 305 337

Death resulting from heart failure(with underlying CHD cause)†

64 81

Death resulting from heart failure(without underlying CHD cause)

6 10

Sudden death† 150 154

Other noncoronary cardiac death 20 16

Subtotal 505 538

Ischemic stroke (including stroke ofunknown origin)

Nonfatal ischemic stroke 371 488

Fatal ischemic stroke 76 101

Subtotal 430 565

MI indicates myocardial infarction; CHD, coronary heart disease.*Major occlusive coronary event corresponds to the revised definition of

major coronary event used in the latest cycle of analyses from the CholesterolTreatment Trialists collaboration.1

†Major coronary event (as defined in the protocol of the Heart ProtectionStudy clinical trial) is composed of these events.

Parish et al Lipids, Lipoproteins, and Vascular Events 2471


171

baseline covariates. Major occlusive coronary event andrevascularization showed clear associations with all 4 LDL-related measures in both the statin and placebo arms, whereasother cardiac event and ischemic stroke showed only weakassociations with these measures. In the statin arm, the LDLparticle measures (apoB and LDL-P) and the cholesterolmeasures (LDL-C and non–HDL-C) showed similar strengthsof association with major occlusive coronary event: HR of1.23 (95% confidence interval [CI], 1.15–1.33) and 1.25(95% CI, 1.16–1.35) for apoB and LDL-P, respectively, and1.25 (95% CI, 1.16–1.34) and 1.22 (95% CI, 1.14–1.32) forLDL-C and non–HDL-C, respectively (Figure 1). Findingswere similar for revascularization (Figure 1) and for thecombined outcome of major occlusive coronary event orrevascularization (HRs between 1.19 and 1.20; Figure I in theonline-only Data Supplement). In the placebo arm, theassociations of all of the measures with the combinedoutcome were weaker than in the statin arm, but (given thatsome bias might arise from compliance associations) therewas no clear evidence of any differences between the mea-sures. The separate HRs for major occlusive coronary eventand revascularization (Figure 1) were statistically compatiblewith the results for the combined outcome (Figure I in theonline-only Data Supplement). Including lower density par-ticles with LDL-P made little difference, but total cholesterol

was a weaker predictor than LDL-C or non–HDL-C (FigureI in the online-only Data Supplement).

Table 3 shows the associations of LDL subclasses withmajor occlusive coronary event in the statin arm. Consideredsingly, the association with small LDL-P was stronger thanthat with large LDL-P, but both were weaker than theassociation with LDL-P. The �2

3 for the strength of theassociation with the 3 LDL subclasses jointly was the same asthat for LDL-P alone (32.4 versus 32.3), indicating that theLDL subclasses did not contribute any additional predictivevalue over LDL-P. A similar pattern was seen for revascu-larization. The strengths of the associations with the mainLDL-P subclasses individually corresponded to their respec-tive correlations with LDL-P (correlation coefficient: 0.76 forsmall LDL-P, 0.16 for large LDL-P; Table 2).

HDL-Related AssociationsFigure 2 contrasts the associations of different HDL-relatedmeasures, given LDL-P, with various vascular outcomes inall participants. (Associations in the statin and placebo armsseparately were similar and are given in Figure II in theonline-only Data Supplement.) For major occlusive coronaryevent, the HRs were �0.90, with HDL-P showing a slightlystronger association than HDL-C and apoA1 (�2

1: 22.5, 13.2,and 13.1, respectively), whereas for revascularization, only

Table 2. Correlation Coefficients* Between Measurements of Different Lipid-Related Factors Within The Same Sample in 20 021 Participants

LDL-C Non–HDL-C ApoB LDL-P LDL-Psmall LDL-Plarge IDL-PLDLSize HDL-C ApoA1 HDL-P HDL-Psmall HDL-Pmedium HDL-Plarge

HDLSize

LogTriglycerides

LDL-related factors

LDL-C 0.87 0.87 0.79 0.39 0.42 0.36 0.06 �0.07 0.05 �0.02 0.17 �0.08 �0.22 �0.30 0.20

Non–HDL-C 0.87 0.93 0.80 0.52 0.19 0.44 �0.15 �0.18 0.05 0.03 0.24 �0.06 �0.28 �0.33 0.48

ApoB 0.87 0.93 0.84 0.54 0.24 0.38 �0.13 �0.18 0.03 �0.01 0.24 �0.10 �0.29 �0.34 0.39

LDL-P 0.79 0.80 0.84 0.76 0.16 0.34 �0.29 �0.34 �0.14 �0.10 0.30 �0.18 �0.44 �0.52 0.37

LDL-Psmall 0.39 0.52 0.54 0.76 �0.48 0.13 �0.76 �0.62 �0.36 �0.18 0.40 �0.26 �0.66 �0.65 0.58

LDL-Plarge 0.42 0.19 0.24 0.16 �0.48 �0.11 0.76 0.44 0.28 0.07 �0.22 0.09 0.39 0.27 �0.41

IDL-P 0.36 0.44 0.38 0.34 0.13 �0.11 �0.08 0.02 0.13 0.16 0.09 0.12 �0.07 �0.08 0.15

LDL size 0.06 �0.15 �0.13 �0.29 �0.76 0.76 �0.08 0.59 0.35 0.16 �0.38 0.26 0.56 0.53 �0.55

HDL-related factors

HDL-C �0.07 �0.18 �0.18 �0.34 �0.62 0.44 0.02 0.59 0.81 0.61 �0.20 0.43 0.84 0.70 �0.46

ApoA1 0.05 0.05 0.03 �0.14 �0.36 0.28 0.13 0.35 0.81 0.77 0.07 0.42 0.67 0.48 �0.08

HDL-P �0.02 0.03 �0.01 �0.10 �0.18 0.07 0.16 0.16 0.61 0.77 0.28 0.58 0.45 0.28 0.06

HDL-Psmall 0.17 0.24 0.24 0.30 0.40 �0.22 0.09 �0.38 �0.20 0.07 0.28 �0.52 �0.33 �0.50 0.31

HDL-Pmedium �0.08 �0.06 �0.10 �0.18 �0.26 0.09 0.12 0.26 0.43 0.42 0.58 �0.52 0.29 0.34 �0.04

HDL-Plarge �0.22 �0.28 �0.29 �0.44 �0.66 0.39 �0.07 0.56 0.84 0.67 0.45 �0.33 0.29 0.89 �0.39

HDL size �0.30 �0.33 �0.34 �0.52 �0.65 0.27 �0.08 0.53 0.70 0.48 0.28 �0.50 0.34 0.89 �0.38

Other factors

Log triglycerides 0.20 0.48 0.39 0.37 0.58 �0.41 0.15 �0.55 �0.46 �0.08 0.06 0.31 �0.04 �0.39 �0.38

Log N-BNP† �0.07 �0.11 �0.10 �0.11 �0.15 0.09 �0.04 0.13 0.08 �0.03 �0.10 �0.19 0.01 0.15 0.15 �0.17

Log CRP† 0.04 0.06 0.08 0.12 0.16 �0.06 �0.01 �0.12 �0.15 �0.13 �0.09 �0.01 0.00 �0.16 �0.11 0.08

BMI§ 0.04 0.09 0.08 0.13 0.22 �0.16 0.04 �0.20 �0.23 �0.14 �0.04 0.15 �0.07 �0.24 �0.24 0.21

eGFR†‡ �0.08 �0.12 �0.09 �0.08 �0.10 0.05 �0.01 0.06 0.10 0.05 0.09 �0.06 0.09 0.11 0.10 �0.14

LDL-C indicates low-density lipoprotein cholesterol; HDL-C, high-density lipoprotein cholesterol; apo, apolipoprotein; LDL-P, particle concentration of LDL; IDL-P,particle concentrations of intermediate-density lipoprotein; HDL-P, particle concentration of HDL; N-BNP, N-terminal pro-B-type natriuretic peptide; CRP, C-reactiveprotein; BMI, body mass index; and eGFR, estimated glomerular filtration rate.

*Adjusted for age, sex, simvastatin and vitamin allocation, prior disease, systolic blood pressure, smoking, eGFR, log N-BNP, and medication.†Available in screening samples only (n�9988).‡Not adjusted for eGFR.§Available for 19 857 participants.



172

HDL-C showed much association (�21, 15.9). For both

outcomes, however, the CIs for all 3 measures overlapped.Table 4 shows the associations of the HDL subclasses with

major occlusive coronary event in all participants both withand without adjustment for LDL-P. Without such adjustment,HDL-C and HDL-P showed similar strengths of association

with major occlusive coronary event risk. However, theseassociations derived partly from negative correlations withLDL-related measures (correlations of large HDL-P andHDL-C with LDL-P, �0.44 and �0.34, respectively; Table2). Hence, given LDL-P, the strengths of association withmajor occlusive coronary event were reduced, becoming

Figure 1. Comparison of the predictivestrengths of low-density lipoprotein(LDL)–related measures for vascularevents. *Adjusted standard deviation cal-culated across both treatment arms: LDLcholesterol (LDL-C), 0.73 mmol/L (0.28g/L); non–high-density lipoprotein cho-lesterol (non–HDL-C), 0.89 mmol/L (0.34g/L); apolipoprotein B (Apo B), 0.20 g/L;particle concentration of LDL (LDL-P),0.34 �mol/L; and LDL size, 0.59 nm.

Table 3. Association of Major Occlusive Coronary Event and Revascularization With Low-DensityLipoprotein–Related Factors* in the Statin Arm

Major Occlusive Coronary Event Revascularization

HR (95% CI)per 1 SD† �2

1 PHR (95% CI)per 1 SD† �2

1 P

LDL-P 1.25 (1.16–1.35) 32.3 �0.0001 1.15 (1.07–1.23) 15.3 �0.0001

LDL size 0.92 (0.85–1.00) 4.1 0.04 0.92 (0.86–0.99) 5.2 0.02

LDL size given LDL-P 0.97 (0.89–1.05) 0.5 0.47 0.95 (0.88–1.02) 2.1 0.15

LDL subclasses‡ singly

LDL-Psmall 1.20 (1.11–1.30) 20.0 �0.0001 1.14 (1.06–1.22) 12.8 0.0003

LDL-Plarge 1.08 (0.96–1.23) 1.6 0.21 1.02 (0.91–1.14) 0.1 0.71

IDL-P 1.53 (1.08–2.16) 5.5 0.02 1.13 (0.82–1.54) 0.5 0.47

LDL subclasses‡ jointly, �23

LDL-Psmall 1.25 (1.14–1.36) 1.17 (1.08–1.26)

LDL-Plarge 1.23 (1.08–1.41) 32.4 �0.0001 1.12 (0.99–1.26) 16.3 0.001

IDL-P 1.32 (0.94–1.86) 1.03 (0.75–1.42)

Additional predictive value ofsubclasses over LDL-P, �2

2§0.1 0.93 0.9 0.62

HR indicates hazard ratio; CI, confidence interval; LDL-P, particle concentration of low-density lipoprotein; and IDL-P, particleconcentration of intermediate-density lipoprotein.

*Analyses are adjusted for age, sex, simvastatin and vitamin allocation, smoking, prior disease, systolic blood pressure, estimatedglomerular filtration rate, medication, and N-terminal pro-B-type natriuretic peptide.

†All standard deviations are as in Figure 1.‡HRs for all LDL subclasses are per 0.34 �mol/L (1 SD of LDL-P).§Calculated from the difference between the �2

3 for the subclasses jointly and the �21 for LDL-P alone.



173

slightly weaker for HDL-C than for HDL-P. The �23 for the

strength of association with the 3 HDL subclasses jointly wasminimally greater than that for HDL-P alone (24.0 versus22.5), indicating that, given LDL-P, the HDL subclasses didnot contribute additional predictive value over HDL-P formajor occlusive coronary event.

Ischemic stroke showed no significant association with anyof the HDL-related measures. Other cardiac events werestrongly associated with HDL-P and with mean HDL particlesize, given HDL-P (HR, 0.84 [95% CI, 0.79–0.90] and 1.13[95% CI, 1.05–1.21], respectively; Figure 2). Table III in theonline-only Data Supplement shows that these associationsderived from inverse associations of risk with small HDL-Pand medium HDL-P (HR, 0.82 [95% CI, 0.76–0.89] and 0.81[95% CI, 0.74–0.89], respectively, in analyses of the sub-classes jointly). This pattern of association was most apparentin the absence of adjustment for N-BNP, suggesting thatpreexisting cardiac insufficiency (as indicated by higherN-BNP) was also associated with this HDL pattern. Lack of

adjustment for N-BNP also affected the relative strengths ofthe associations of the HDL-related measures with majorocclusive coronary event, causing the HDL-P association toappear stronger (Figure 3). The associations of HDL-relatedmeasures with any major coronary event (ie, including 449heart failure or sudden deaths believed to have an underlyingcoronary cause) also weakly reflected the pattern seen forother cardiac event.

Influence of Prior DiseaseThe adjustments for prior disease (included in all analyses)had little net effect on the LDL-related associations. Incontrast, HDL-C and HDL-P associations were considerablyattenuated by the adjustment for prior disease-related factors.For example, �2 values for the associations with majorocclusive coronary event were reduced by 50% to 60% afterfurther adjustment beyond age, sex, allocated treatment arm,smoking, and blood pressure (Table IV in the online-onlyData Supplement). The attenuation was somewhat greater for

Figure 2. Comparison of the predictivestrengths of high-density lipoprotein(HDL)–related measures, given particleconcentration of low-density lipoprotein,for vascular events (all participants).*Adjusted standard deviation calculatedacross both treatment arms: HDL cho-lesterol (HDL-C), 0.29 mmol/L (0.11 g/L);apolipoprotein A1 (Apo A1), 0.18 g/L;particle concentration of HDL (HDL-P),5.18 �mol/L; and HDL size, 0.44 nm.



174

HDL-P than for HDL-C, largely through the additionalassociation of HDL-P with N-BNP (Table IV in the online-only Data Supplement and Table 2).

Findings in this investigation did not differ by sex (FiguresIII and IV in the online-only Data Supplement), and almostall participants (97%) were white.

DiscussionIn this large study, there was no clear evidence of a differencein the strengths of the associations of any of the particle orcholesterol-based LDL-related measures with major occlu-sive coronary events or revascularization (1580 such eventsin the statin arm, 2010 in the placebo arm). The lack of

Table 4. Association of Major Occlusive Coronary Event With High-Density Lipoprotein–Related Factors inAll Participants*

Not Adjusted for LDL-P Adjusted for LDL-P

HR (95% CI)per 1 SD† �2

1 PHR (95% CI)per 1 SD† �2

1 P

HDL-C 0.87 (0.83–0.92) 30.1 �0.0001 0.91 (0.86–0.96) 13.2 0.0003

ApoA1 0.89 (0.85–0.94) 19.4 �0.0001 0.91 (0.87–0.96) 13.1 0.0003

HDL-P 0.88 (0.83–0.92) 28.7 �0.0001 0.89 (0.85–0.93) 22.5 �0.0001

HDL size 0.91 (0.87–0.95) 16.0 �0.0001 0.97 (0.92–1.03) 0.8 0.36

HDL size given HDL-P 0.94 (0.89–0.99) 6.3 0.01 1.01 (0.96–1.07) 0.2 0.66

HDL subclasses‡ singly

HDL-Psmall 1.03 (0.98–1.09) 1.3 0.26 0.98 (0.93–1.03) 0.6 0.44

HDL-Pmedium 0.88 (0.83–0.93) 22.7 �0.0001 0.90 (0.85–0.95) 13.6 0.0002

HDL-Plarge 0.79 (0.71–0.88) 20.3 �0.0001 0.89 (0.79–1.00) 3.8 0.05

HDL subclasses‡ jointly, �23

HDL-Psmall 0.93 (0.87–0.99) 0.90 (0.85–0.96)

HDL-Pmedium 0.87 (0.82–0.93) 38.0 �0.0001 0.87 (0.81–0.93) 24.0 �0.0001

HDL-Plarge 0.82 (0.73–0.91) 0.92 (0.81–1.03)

Additional predictive value ofsubclasses over HDL-P, �2

2§9.3 0.01 1.6 0.45

LDL-P indicates particle concentration of low-density lipoprotein; HR, hazard ratio; CI, confidence interval; HDL-C, high-densitylipoprotein cholesterol; ApoA1, apolipoprotein A1; and HDL-P, particle concentration of HDL.

*Analyses are adjusted for age, sex, simvastatin and vitamin allocation, smoking, prior disease, systolic blood pressure, estimatedglomerular filtration rate, medication, and N-terminal pro-B-type natriuretic peptide.

†All standard deviations are as in Figure 2.‡HRs for all HDL subclasses are per 5.18 �mol/L (1 SD of HDL-P).§Calculated from the difference between the �2

3 for the subclasses jointly and the �21 for HDL-P alone.

Figure 3. Comparison of the predictivestrengths of high-density lipoprotein(HDL)–related measures, given particleconcentration of low-density lipoprotein,for vascular events with and withoutadjustment for N-terminal pro-B-typenatriuretic peptide (N-BNP; all partici-pants). Major coronary event is the com-bination of nonfatal myocardial infarctionor coronary heart disease death, deathcaused by heart failure (with underlyingcause CHD), and sudden death. HDL-Cindicates HDL cholesterol; Apo A1, apoli-poprotein A1; and HDL-P, particle con-centration of HDL.



175

additional predictive value from LDL subclasses or LDLmean size among people with prior vascular disease in thisstudy is consistent with the results from the most informativeof the previous studies, which were in populations withoutprior cardiovascular disease (European Prospective Investi-gation into Cancer [EPIC]–Norfolk with 1003 coronary arterydisease cases,18 Women’s Health Study [WHS] with 1015cardiovascular cases17). Other studies have reported mixedresults,10 but the variety of approaches to measurement andstatistical analysis makes a consensus summary of themdifficult. Furthermore, improvements in assay methods andstandardization23,27 may have altered the relative predictivevalues of different measures.

The strengths of the associations of the LDL-P subclasseswith occlusive events in the present study broadly reflectedtheir correlations with LDL-P. Small LDL-P was the subclassmost strongly correlated with LDL-P, but its association withmajor occlusive coronary event risk was weaker than that ofLDL-P. The lack of added predictive value from the sub-classes cannot be attributed to differences in reproducibilitybecause the reproducibility of the LDL subclasses (in samplescollected a few years apart) was similar to that of the 4 totalLDL-related measures. These 4 measures were stronglycorrelated with each other (�0.8 or higher) in the presentstudy, whereas the correlations between cholesterol andparticle measures in the EPIC-Norfolk and WHS studies weresomewhat weaker (eg, �0.63 between LDL-P and LDL-C).

The ERFC meta-analysis of �4000 CHD events in peoplewithout prior vascular disease also found that non–HDL-Cand apoB (and LDL-C, which was available for �2000 cases)had similar strengths of association with CHD.2 This finding,however, differs from those of earlier large retrospectivecase-control studies in acute MI, which reported that apoli-poproteins were superior predictors of MI than cholesterolfractions,28,29 but this difference may be explained by distor-tions of HDL-C levels in the hours after an acute MI whenapolipoproteins may be more robust measures. The HR formajor occlusive coronary event per unit of non–HDL-C (1.22per 0.34 mg/dL in the statin arm) in the present study was lowcompared with that found in the ERFC meta-analysis (1.56per 0.29 mg/dL after correction for within-person variation).In the present study, within-person variation, together withsome variation in statin use (even in the statin arm), resultedin correlations of only �0.55 between baseline and midstudynon–HDL-C values (data not shown), which would attenuateassociations with risk. The weaker HRs observed in thishigh-risk population would also be consistent with the weakerHRs seen at older ages and higher systolic blood pressure inprevious large meta-analyses.1,2

The weaker and indefinite associations with ischemicstroke in the present study are also consistent with the relativestrengths of the associations of non–HDL-C with coronaryevents and with ischemic stroke in the ERFC (in which, basedon 1800 strokes, the log hazard ratios were only a quarter asgreat for ischemic stroke as for CHD). NMR-measuredlipoprotein subclass measurements did not provide any in-sight into the discordance between the weak observationalfindings and the effect of LDL-lowering therapy on ischemicstroke.

HDL-C and HDL size were more strongly correlated thanapoA1 or HDL-P were with the total LDL-related measures.Thus, adjustment for LDL-P attenuated the risk associationwith HDL-C more than it attenuated the risk associations withapoA1 and HDL-P. The weak levels of association of all HDLmeasures with CHD, however, limited the power to discrim-inate between these measures. The correlation of baselineHDL-C with repeat measurements obtained during follow-upwas strong (0.82), so allowance for within-person variation30

would only increase the strength of association modestly (eg,from 0.90 to 0.88). The modest HRs for the HDL-relatedmeasures found in this high-risk population (compared withERFC) may indicate a tendency to weaker associations in lesshealthy individuals, as suggested by the marked trends withage and systolic blood pressure in the ERFC analyses.2 Suchattenuation may be a manifestation of HDL becoming dys-functional in individuals with existing cardiovascular dis-ease.15,31–34 Adjustment for prior disease, including forN-BNP as a marker of heart failure, considerably attenuatedassociations with HDL-related measures, highlighting thepotential for confounding by preexisting disease and forconflicting observations from studies with differing degreesof adjustment.

A strong inverse association was seen between smallHDL-P and both N-BNP and other cardiac event risk (Table2 and Table III in the online-only Data Supplement). Becausecardiac events in this study occurred predominantly in par-ticipants who had preexisting cardiac disease at entry, it is notpossible to determine whether the association represents acausal or reverse-causal effect (ie, whether higher smallHDL-P protects against the risk of a cardiac event7 or whetherlower small HDL-P is a consequence of the preexistingdisease). Small and large mean HDL sizes were associatedwith different forms of prior disease: Small mean HDL sizewas associated with high triglycerides and LDL-P and withother features of the metabolic syndrome (Table 2), aspreviously noted in the EPIC-Norfolk study,35 and large meanHDL size was associated with high N-BNP and other cardiacevent risk.

The present large study was prospective in design and, incontrast to previous studies, was able to look separately atassociations of lipoprotein subclasses with coronary events,other cardiac events, ischemic strokes, and revascularizations.Another strength of this study was that it included detailedadjustment for prior vascular disease and other establishedrisk factors. It extends previous results, showing similarstrengths of risk association with different LDL and HDLmeasures in a population at higher risk of occlusive coronaryevent (average, 2%/y). Published studies have suggested thatthe magnitude of some associations of lipid levels with riskmay attenuate with increasing levels of risk factors (such asage and systolic blood pressure); thus, it is a limitation of thisstudy that the observed associations may be representative ofonly this type of high-risk population.

ConclusionsIn a wide range of individuals at varying levels of cardiovas-cular risk, lipoprotein particles, apolipoproteins, and choles-terol fractions of LDL are of similar predictive value for



176

major occlusive coronary events, and LDL subclasses providelittle additional information. Lipoprotein particle numbermeasurements could be considered an alternative to LDL andHDL cholesterol measurements for risk prediction if theybecame available at a similar cost. Measurement of a particle-based measure for LDL such as apoB or LDL-P also makesmore biological sense because these measurements are ofparticles attached to the arterial wall.36 Associations ofHDL-related measures with risk are considerably influencedby prior disease; in particular, interpretation of the relation-ship between HDL particle measures, prior disease, and riskof other cardiac events requires further study.

AcknowledgmentsThe most important acknowledgement is of the participants in thestudy and of the steering committee and collaborators listed in aprevious report.20 We also gratefully acknowledge the staff of theWolfson laboratories in the Clinical Trial Service Unit for their helpwith processing, storage, and retrieval of blood samples.

Sources of FundingThe HPS was funded by the UK Medical Research Council, BritishHeart Foundation, Merck & Co (manufacturers of simvastatin), andRoche Vitamins Ltd (manufacturers of vitamins). NMR lipoproteinmeasurements were funded by LipoScience, Inc. Dr Hopewellacknowledges support from the Oxford British Heart FoundationCentre of Research Excellence.

DisclosuresThe Clinical Trial Service Unit has a policy of not acceptinghonoraria or other payments from the pharmaceutical industry exceptfor the reimbursement of costs to participate in scientific meetings.Dr Otvos is an employee and shareholder of LipoScience, Inc. Theother authors report no conflicts.

References1. Prospective Studies Collaboration. Blood cholesterol and vascular mor-

tality by age, sex, and blood pressure: a meta-analysis of individual datafrom 61 prospective studies with 55 000 vascular deaths. Lancet. 2007;370:1829–1839.

2. Di Angelantonio E, Sarwar N, Perry P, Kaptoge S, Ray KK, ThompsonA, Wood AM, Lewington S, Sattar N, Packard CJ, Collins R, ThompsonSG, Danesh J. Major lipids, apolipoproteins, and risk of vascular disease.JAMA. 2009;302:1993–2000.

3. Baigent C, Blackwell L, Emberson J, Holland LE, Reith C, Bhala N, PetoR, Barnes EH, Keech A, Simes J, Collins R. Efficacy and safety of moreintensive lowering of LDL cholesterol: a meta-analysis of data from170,000 participants in 26 randomised trials. Lancet. 2010;376:1670–1681.

4. Emberson JR, Ng LL, Armitage J, Bowman L, Parish S, Collins R.N-terminal pro-B-type natriuretic peptide, vascular disease risk, and cho-lesterol reduction among 20,536 patients in the MRC/BHF heart pro-tection study. J Am Coll Cardiol. 2007;49:311–319.

5. Cleland JGF, McMurray JJV, Kjekshus J, Cornel JH, Dunselman P,Fonseca C, Hjalmarson A, Korewicki J, Lindberg M, Ranjith N, vanVeldhuisen DJ, Waagstein F, Wedel H, Wikstrand J; CORONA StudyGroup. Plasma concentration of amino-terminal pro-brain natriureticpeptide in chronic heart failure: prediction of cardiovascular events andinteraction with the effects of rosuvastatin: a report from CORONA(Controlled Rosuvastatin Multinational Trial in Heart Failure). J Am CollCardiol. 2009;54:1850–1859.

6. Amarenco P, Goldstein LB, Szarek M, Sillesen H, Rudolph AE, CallahanA 3rd, Hennerici M, Simunovic L, Zivin JA, Welch KM. Effects ofintense low-density lipoprotein cholesterol reduction in patients withstroke or transient ischemic attack: the Stroke Prevention by AggressiveReduction in Cholesterol Levels (SPARCL) trial. Stroke. 2007;38:3198–3204.

7. Kontush A, Chapman MJ. Antiatherogenic small, dense HDL–guardianangel of the arterial wall? Nat Clin Pract Cardiovasc Med. 2006;3:144–153.

8. Rizzo M, Berneis K. Low-density lipoprotein size and cardiovascular riskassessment. QJM. 2006;99:1–14.

9. Francis GA. The complexity of HDL. Biochim Biophys Acta. 2010;1801:1286–1293.

10. Ip S, Lichtenstein AH, Chung M, Lau J, Balk EM. Systematic review:association of low-density lipoprotein subfractions with cardiovascularoutcomes. Ann Intern Med. 2009;150:474–484.

11. Third report of the National Cholesterol Education Program (NCEP)Expert Panel on Detection, Evaluation, and Treatment of High Blood Cho-lesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002;106:3143.

12. Mora S, Szklo M, Otvos JD, Greenland P, Psaty BM, Goff DC Jr,O’Leary DH, Saad MF, Tsai MY, Sharrett AR. LDL particle subclasses,LDL particle size, and carotid atherosclerosis in the Multi-Ethnic Study ofAtherosclerosis (MESA). Atherosclerosis. 2007;192:211–217.

13. Rosenson RS, Brewer HBJ, Chapman MJ, Fazio S, Hussain MM,Kontush A, Krauss RM, Otvos JD, Remaley AT, Schaefer EJ. HDLmeasures, particle heterogeneity, proposed nomenclature, and relation toatherosclerotic cardiovascular events. Clin Chem. 2011;57:392–410.

14. Barter P, Kastelein J, Nunn A, Hobbs R. High density lipoproteins(HDLs) and atherosclerosis: the unanswered questions. Atherosclerosis.2003;168:195–211.

15. Dodani S, Grice DG, Joshi S. Is HDL function as important as HDLquantity in the coronary artery disease risk assessment? J Clin Lipidol.2009;3:70–77.

16. Otvos JD, Mora S, Shalaurova I, Greenland P, Mackey RH, Goff DC.Clinical implications of discordance between low-density lipoproteincholesterol and particle number. J Clin Lipidol. 2011;5:105–113.

17. Mora S, Otvos JD, Rifai N, Rosenson RS, Buring JE, Ridker PM.Lipoprotein particle profiles by nuclear magnetic resonance comparedwith standard lipids and apolipoproteins in predicting incident cardiovas-cular disease in women. Circulation. 2009;119:931–939.

18. El Harchaoui K, van der Steeg WA, Stroes ES, Kuivenhoven JA, OtvosJD, Wareham NJ, Hutten BA, Kastelein JJ, Khaw KT, Boekholdt SM.Value of low-density lipoprotein particle number and size as predictors ofcoronary artery disease in apparently healthy men and women: the EPIC-Norfolk Prospective Population Study. J Am Coll Cardiol. 2007;49:547–553.

19. Jeyarajah EJ, Cromwell WC, Otvos JD. Lipoprotein particle analysis bynuclear magnetic resonance spectroscopy. Clin Lab Med. 2006;26:847–870.

20. Heart Protection Study Collaborative Group. MRC/BHF Heart ProtectionStudy of cholesterol lowering with simvastatin in 20,536 high-risk indi-viduals: a randomised placebo-controlled trial. Lancet. 2002;360:7–22.

21. Heart Protection Study Collaborative Group. MRC/BHF Heart ProtectionStudy of cholesterol-lowering therapy and of antioxidant vitamin supple-mentation in a wide range of patients at increased risk of coronary heartdisease death: early safety and efficacy experience. Eur Heart J. 1999;20:725–741.

22. Heart Protection Study Collaborative Group. C-reactive protein concen-tration and the vascular benefits of statin therapy: an analysis of 20 536patients in the Heart Protection Study. Lancet. 2011;377:469–476.

23. Nauck M, Warnick GR, Rifai N. Methods for measurement of LDL-cho-lesterol: a critical assessment of direct measurement by homogeneousassays versus calculation. Clin Chem. 2002;48:236–254.

24. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A moreaccurate method to estimate glomerular filtration rate from serum creat-inine: a new prediction equation: Modification of Diet in Renal DiseaseStudy Group. Ann Intern Med. 1999;130:461–470.

25. Dickstein K, Cohen-Solal A, Filippatos G, McMurray JJ, Ponikowski P,Poole-Wilson PA, Stromberg A, van Veldhuisen DJ, Atar D, Hoes AW,Keren A, Mebazaa A, Nieminen M, Priori SG, Swedberg K, Vahanian A,Camm J, De Caterina R, Dean V, Dickstein K, Filippatos G, Funck-Brentano C, Hellemans I, Kristensen SD, McGregor K, Sechtem U, SilberS, Tendera M, Widimsky P, Zamorano JL. ESC guidelines for thediagnosis and treatment of acute and chronic heart failure 2008: the TaskForce for the Diagnosis and Treatment of Acute and Chronic HeartFailure 2008 of the European Society of Cardiology: developed in col-laboration with the Heart Failure Association of the ESC (HFA) and



177

endorsed by the European Society of Intensive Care Medicine (ESICM).Eur Heart J. 2008;29:2388–2442.

26. Struthers A, Lang C. The potential to improve primary prevention in thefuture by using BNP/N-BNP as an indicator of silent “pancardiac” targetorgan damage: BNP/N-BNP could become for the heart what microalbu-minuria is for the kidney. Eur Heart J. 2007;28:1678–1682.

27. Contois JH, McConnell JP, Sethi AA, Csako G, Devaraj S, Hoefner DM,Warnick GR. Apolipoprotein B and cardiovascular disease risk: positionstatement from the AACC Lipoproteins and Vascular Diseases DivisionWorking Group on Best Practices. Clin Chem. 2009;55:407–419.

28. Parish S, Peto R, Palmer A, Clarke R, Lewington S, Offer A, Whitlock G,Clark S, Youngman L, Sleight P, Collins R. The joint effects of apoli-poprotein B, apolipoprotein A1, LDL cholesterol, and HDL cholesterolon risk: 3510 cases of acute myocardial infarction and 9805 controls. EurHeart J. 2009;30:2137–2146.

29. McQueen MJ, Hawken S, Wang X, Ounpuu S, Sniderman A, ProbstfieldJ, Steyn K, Sanderson JE, Hasani M, Volkova E, Kazmi K, Yusuf S.Lipids, lipoproteins, and apolipoproteins as risk markers of myocardialinfarction in 52 countries (the INTERHEART study): a case-controlstudy. Lancet. 2008;372:224–233.

30. Clarke R, Shipley M, Lewington S, Youngman L, Collins R, Marmot M,Peto R. Underestimation of risk associations due to regression dilution inlong-term follow-up of prospective studies. Am J Epidemiol. 1999;150:341–353.

31. Sviridov D, Mukhamedova N, Remaley AT, Chin-Dusting J, Nestel P.Antiatherogenic functionality of high density lipoprotein: how muchversus how good. J Atheroscler Thromb. 2008;15:52–62.

32. Feng H, Li XA. Dysfunctional high-density lipoprotein. Curr Opin Endo-crinol Diabetes Obes. 2009;16:156–162.

33. Nicholls SJ, Zheng L, Hazen SL. Formation of dysfunctional high-densitylipoprotein by myeloperoxidase. Trends Cardiovasc Med. 2005;15:212–219.

34. Fogelman AM. When good cholesterol goes bad. Nat Med. 2004;10:902–903.

35. El Harchaoui K, Arsenault BJ, Franssen R, Despres JP, Hovingh GK,Stroes ES, Otvos JD, Wareham NJ, Kastelein JJ, Khaw KT, BoekholdtSM. High-density lipoprotein particle size and concentration andcoronary risk. Ann Intern Med. 2009;150:84–93.

36. Arsenault BJ, Boekholdt SM, Kastelein JJ. Lipid parameters for mea-suring risk of cardiovascular disease. Nat Rev Cardiol. 2011;8:197–206.

CLINICAL PERSPECTIVELow-density lipoprotein (LDL) cholesterol and high-density lipoprotein (HDL) cholesterol are established risk factors forvascular disease, but there is uncertainty as to whether measures of the numbers of LDL and HDL particles, rather thantheir cholesterol content, might be better predictors of risk. Small LDL particles are widely believed to be particularlyhazardous, but the independent predictive value of lipoprotein subclasses is also unresolved. The present investigationconsiders the associations between baseline concentrations of cholesterol fractions, apolipoproteins B and A1, andlipoprotein particles assessed by nuclear magnetic resonance with vascular events during 5.3 years of follow-up among20 000 high-risk individuals (2% average coronary event risk per year). Cholesterol and particle measures of LDL werestrongly correlated and had similar predictive value for incident major occlusive coronary events and for arterialrevascularization procedures. Given the total LDL particle number, the distribution between small and large particles didnot add predictive value. Associations of these different LDL-related measures were much weaker or nonexistent withischemic stroke and other cardiac events (mainly heart failure). After adjustment for LDL particle number, the predictivevalues of the cholesterol and particle measures of HDL for incident major occlusive coronary events were similar. Incontrast, other cardiac events were strongly associated with fewer total and small HDL particles and larger mean HDL sizebut only very weakly associated with HDL cholesterol. The present results indicate that LDL lipoprotein particlemeasurements provide little additional predictive value over traditional measures for occlusive vascular events in ahigh-risk population. It is unclear whether the associations between HDL particle numbers and other cardiac eventsrepresent a causal or reverse-causal effect.



Lipids and Lipoproteins and Risk of Different Vascular Events in the ...

Documents

Transcript of Lipids and Lipoproteins and Risk of Different Vascular Events in the ...