Low-Density Lipoproteins Containing Apolipoprotein C-III

and the Risk of Coronary Heart Disease

Carlos O. Mendivil, MD, DSc; Eric B. Rimm, DSc; Jeremy Furtado, DSc;
Stephanie E. Chiuve, DSc; Frank M. Sacks, MD
BackgroundLow-density lipoprotein (LDL) that contains apolipoprotein (apo) C-III makes up only 10% to 20% of
plasma LDL but has a markedly altered metabolism and proatherogenic effects on vascular cells.
Methods and ResultsWe examined the association between plasma LDL with apoC-III and coronary heart disease in 320
women and 419 men initially free of cardiovascular disease who developed a fatal or nonfatal myocardial infarction
during 10 to 14 years of follow-up and matched controls who remained free of coronary heart disease. Concentrations
of LDL with apoC-III (measured as apoB in this fraction) were associated with risk of coronary heart disease in
multivariable analysis that included the ratio of total cholesterol to high-density lipoprotein cholesterol, LDL
cholesterol, apoB, triglycerides, or high-density lipoprotein cholesterol and other risk factors. In all models, the relative
risks for the top versus bottom quintile of LDL with apoC-III were greater than those for LDL without apoC-III. When
included in the same multivariable-adjusted model, the risk associated with LDL with apoC-III (relative risk for top
versus bottom quintile, 2.38; 95% confidence interval, 1.54 3.68; P for trend 0.001) was significantly greater than that
associated with LDL without apoC-III (relative risk for top versus bottom quintile, 1.25; 95% confidence interval,
0.76 2.05; P for trend0.97; P for interaction 0.001). This divergence in association with coronary heart disease
persisted even after adjustment for plasma triglycerides.
ConclusionsThe risk of coronary heart disease contributed by LDL appeared to result to a large extent from LDL that
contains apoC-III. (Circulation. 2011;124:2065-2072.)
Key Words: apolipoproteins cholesterol metabolism myocardial infarction risk factors

ery low-density lipoprotein (VLDL) and low-density

lipoprotein (LDL) are heterogeneous lipoprotein classes
that vary in size and content of lipid and protein.1 Apolipoprotein (apo) B is the required structural apolipoprotein of
VLDL and LDL. Each VLDL and LDL has only 1 molecule
of apoB but may have no or many molecules of apoC-III
attached to its surface.13 About 40% to 60% of VLDL and
10% to 20% of LDL have apoC-III.2,3 ApoC-III has deleterious effects on the metabolism of VLDL and LDL4 6 and on
functions of cells that participate in atherosclerosis.7 ApoC-III
interferes with the binding of VLDL to receptors in the liver,
which inhibits the removal of VLDL from plasma.5,6,8 Nearly all
VLDL that has apoC-III undergoes intravascular lipolysis of its
triglyceride content, producing LDL with apoC-III, and large
LDL with apoC-III is metabolized to small LDL.5,6,9

Clinical Perspective on p 2072

ApoC-III containing VLDL and LDL prepared from human plasma activate monocytes that circulate in blood to

adhere to vascular endothelial cells, an early step in atherosclerosis.10,11 ApoC-III also activates vascular endothelial
cells to produce adhesion molecules and induces insulin
resistance in these cells, reducing their secretion of nitric
oxide.12 These actions are not shared by VLDL or LDL that
does not have apoC-III. The presence of apoC-III on LDL is
also associated with compositional changes that favor LDL
adhesion to the subendothelial extracellular matrix.13 Thus,
apoC-III may trap its associated lipoproteins in the arterial
wall and bring in blood monocytes, crucial steps in the initiation
and progression of atherosclerosis. ApoC-III concentrations in
VLDL and LDL are positively associated with the progression
of atherosclerosis or risk of coronary heart disease (CHD),14 17
and we had reported that LDL with apoC-III is associated with
recurrent cardiovascular disease among patients with a prior
myocardial infarction and type 2 diabetes mellitus.3
To evaluate whether plasma concentrations of VLDL and
LDL that have apoC-III, as measured by apoB in each of

Received January 21, 2011; accepted August 26, 2011.

From the Harvard School of Public Health, Boston, MA (C.O.M., E.B.R., J.F., S.E.C., F.M.S.); Universidad de los Andes Medical School, Bogota,
Colombia (C.O.M.); and Channing Laboratory (E.B.R., F.M.S.) and Division of Preventive Medicine, Department of Medicine (E.B.R., S.E.C.), Brigham
and Womens Hospital and Harvard Medical School, Boston, MA.
Guest Editor for this article was Mary Cushman, MD, MSc.
The online-only Data Supplement is available with this article at http://circ.ahajournals.org/lookup/suppl/doi:10.1161/CIRCULATIONAHA.
111.056986/-/DC1.
Correspondence to Frank M. Sacks, MD, Harvard School of Public Health, Department of Nutrition, 665 Huntington Ave, Bldg 1, Room 202, Boston,
MA 02115. E-mail fsacks@hsph.harvard.edu
2011 American Heart Association, Inc.
Circulation is available at http://circ.ahajournals.org

DOI: 10.1161/CIRCULATIONAHA.111.056986

Downloaded from http://circ.ahajournals.org/

by guest on March 22, 2016
2065

2066

Circulation

November 8, 2011

them, are more strongly associated with CHD risk than those
of the same lipoproteins without apoC-III, we prospectively
studied 2 US populations initially free of CHD: the Nurses
Health Study (NHS; women) and the Health Professionals
Follow-up Study (HPFS; men).

Methods
Study Population
The NHS and the HPFS are prospective cohort investigations
involving 121 700 female US registered nurses who were 30 to 55
years of age at baseline in 1976 and 51 529 US male health
professionals who were 40 to 75 years of age at baseline in 1986,
respectively. Information about health and disease is assessed
biennially.18 21
From 1989 through 1990, a blood sample was requested from all
participants in the NHS, which 32 826 women provided. Similarly,
between 1993 and 1995, a blood sample was provided as requested
by 18 225 men in the HPFS. Participants who provided blood
samples were similar to those who did not. In the NHS, among
women without cardiovascular disease or cancer before 1990, we
identified 350 women who had a nonfatal myocardial infarction or
fatal CHD between the date of the blood draw and June 2004. In the
HPFS, we identified 425 men initially free of cardiovascular disease
who had a nonfatal myocardial infarction or fatal CHD between the
date of the blood draw and the return of the 2004 questionnaire.
Using risk-set sampling (a technique in which, in a prospective
cohort, control subjects are selected among the group of individuals
exposed to the studied factor at the time the event occurs in the
case22), we randomly selected controls in a 1:1 ratio who were
matched for age (1 year), smoking status (never, past, or current),
and date of blood sampling (2 months) from the participants who
were free of cardiovascular disease at the time CHD was diagnosed
in the case patients. Within the NHS cohort, an additional matching
criterion was fasting status at the time of blood sampling.

Assessment of CHD
Study physicians who were unaware of the participants exposure
status confirmed the diagnosis of myocardial infarction on the basis
of the criteria of the World Health Organization. Deaths were
identified from state vital records and the National Death Index or
reported by the participants next of kin or the postal system. Fatal
CHD was confirmed by an examination of hospital or autopsy
records, by the listing of CHD as the cause of death on the death
certificate, if CHD was the underlying and most plausible cause, and
if evidence of previous CHD was available.

Assessment of Other Factors

Medical history, lifestyle, and demographic information was derived
from the questionnaire administered in 1990 to women and 1994 to
men, with missing information substituted from previous questionnaires. Physical activity was expressed in terms of metabolic
equivalent hours. The questionnaires and the validity and reproducibility of measurements have been described previously.18 21

Measurement of Lipids and

Apolipoproteins Levels
Blood samples from women were collected in heparin-treated tubes;
samples from men, in EDTA-treated tubes. The tubes were placed on
ice packs, shipped in Styrofoam containers by overnight courier,
centrifuged, divided into aliquots, and stored in liquid-nitrogen
freezers (130C or colder). One vial containing 0.6 mL (NHS) or
0.5 mL (HPFS) frozen plasma was shipped to the lipoprotein
laboratory at the Harvard School of Public Health for analysis of
lipoprotein types. Study samples were sent to the laboratory for
analysis in randomly ordered batches, and the laboratory personnel
were unaware of a samples case-control status. The entire plasma
sample was thawed, filtered, and incubated overnight with affinitypurified polyclonal goat antihuman apoC-III antibodies (Academy

Biomedical, Houston, TX) bound to Sepharose 4B resin, as previously described and validated.2,23 The unbound lipoproteins that did
not contain apoC-III were collected by gravity flow, and the bound
lipoproteins that contained apoC-III were eluted with 3 mol/L
sodium thiocyanate. The efficiency of the apoC-III immunoaffinity
separation (percentage of ligand removed from plasma by the resin)
was 99%. Subsequently, apoC-III bound and unbound fractions
were ultracentrifuged to isolate the VLDL (d1.006 g/mL), LDL
(1.006d1.063 g/mL), and high-density lipoprotein (HDL;
d1.063 g/mL).24 ApoB and apoC-III were measured in whole
plasma and in VLDL and LDL with and without apoC-III by ELISA,
and cholesterol and triglycerides were measured by enzymatic
methods (Infinity Kit, Thermo Fischer Scientific, Waltham, MA).
To test the repeatability of the assays, we ran blinded controls that
were sent by the Channing Laboratory disguised as samples and
included in every batch with their own identifications. These controls
were included and run in every batch of real samples so that each
control went through the whole process at least 8 times. The
coefficients of variation for NHS were 9% for apoB in LDL without
apoC-III and 17% for apoB in LDL with apoC-III; for HPFS, they
were 17% for apoB in LDL without apoC-III and 13% for apoB in
LDL with apoC-III.
The study protocol was approved by the institutional review board
of the Brigham and Womens Hospital and the Human Subjects
Committee Review Board of the Harvard School of Public Health.
After the exclusion of participants with insufficient sample or
missing data for apoB in at least one of the LDL and VLDL
fractions, the data set consisted of 1476 participants (320 female and
419 male cases, 320 female and 419 male controls). The cases were
identified and the controls selected in 2-year cycles. Two men
selected as controls in 1 cycle later developed CHD and were
selected in a subsequent cycle, this time as cases. For this reason, the
study sample contains 1476, not 1478, individuals. For these 2
participants, we used the same baseline measurements in the pair in
which they were cases and in the pair in which they were controls.
Because there is only 1 apoB molecule per lipoprotein particle, the
apoB concentration represents the concentration of lipoproteins of a
given type. Our main exposure, specified in the protocol and
according to our previous result,3 was the concentration of LDL with
apoC-III, as reflected by the concentration of apoB in this fraction,
ie, apoB in LDL with apoC-III.

Statistical Analysis
Data from the 2 cohorts were analyzed at the individual-participant
level (as opposed to separate analysis in each cohort plus meta-analysis). Apolipoprotein measurements were divided into quintiles,
from the lowest to highest levels, based on the sex-specific distributions among the controls. We analyzed the association between
apolipoprotein levels and the risk of CHD using conditional logistic
regression. With risk-set sampling, the odds ratio derived from the
logistic regression directly estimates the hazard ratio and thus the
relative risk.22 We analyzed our exposures in 4 logistic models:
model 1 conditioned on matching factors only; model 2 with
additional adjustment for parental history of CHD before 60 years of
age, alcohol intake, and personal history of hypertension at baseline;
model 3 with adjustment for all the variables in model 2 plus body
mass index and personal history of diabetes mellitus; and model 4
with the addition of triglyceride concentration at baseline. Baseline
was defined as the year blood was drawn. To test for linear trend, we
used the median levels of apolipoprotein measurements in the control
categories as a continuous variable.
To test for an interaction between LDL types, we ran a model that
included LDL quintile, presence or absence of apoC-III, and an
interaction term.
All P values are 2 tailed, and P values 0.05 were considered to
indicate statistical significance. All analyses were performed with
the use of SAS software, version 9.1 (SAS Institute).

Results
LDL with apoC-III represented on average 12% of total LDL.
Participants of both sexes who developed CHD during

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

Mendivil et al
Table 1.

LDL With apoC-III and CHD

2067

Baseline Characteristics of the Study Sample

Women*

Characteristic

Men*

Cases (n320)

Controls (n320)

Cases (n419)

Controls (n419)

606

606

...

648

648

...

25

25

...

26.85.6

25.44.2

0.001

26.13

25.53

Parental history of CHD before 60 y of age, %

20

14

0.001

15

12

Postmenopausal, %

91

88

0.001

...

...

...

Postmenopausal hormone therapy among

postmenopausal women, %

33

35

0.001

...

...

...

Age, y
Current smoker, %
Body mass index, kg/m2

...
0.017
0.001

Aspirin use, %

26

30

0.001

40

33

0.001

History of hypertension, %

53

31

0.013

37

28

0.001

History of diabetes mellitus, %

17

0.001

0.001

0.31

Alcohol consumption, g/d

Median

0.9

1.1

Q1Q3

04.1

06.4

4.4
015.0

7.5

0.009

1.018.2

Total cholesterol, mmol/L

6.111.19

6.011.22

0.20

5.621.01

5.410.93

0.001

LDL cholesterol, mmol/L

3.811.04

3.651.01

0.059

3.470.85

3.260.83

0.001

HDL cholesterol, mmol/L

1.350.36

1.530.44

0.001

1.090.28

1.220.34

0.001

3.71.5

3.41.6

3.71.1

3.41.4

0.001

1.591.11

1.230.65

0.001

1.630.92

1.340.76

0.001

0.34

0.34

0.26

0.83

0.131.07

0.080.87

Ratio of total to HDL cholesterol

Triglycerides, mmol/L

0.014

C-reactive protein, mg/L

Median

0.52

0.38

Q1Q3

0.201.45

0.141.02

Total plasma apoB, g/L

0.001

0.920.28

0.880.31

0.07

0.970.28

0.90.26

ApoB in VLDL with apoC-III, g/L

0.0120.011

0.0090.008

0.001

0.0230.016

0.020.015

0.003

ApoB in VLDL without apoC-III, g/L

0.001

0.0360.028

0.030.024

0.001

0.0380.027

0.0310.024

ApoB in LDL with apoC-III, g/L

0.090.084

0.0750.056

0.003

0.1350.085

0.1170.08

0.001

ApoB in LDL without apoC-III, g/L

0.780.24

0.770.28

0.51

0.770.23

0.730.22

0.001

Total plasma apoC-III, g/L

0.180.11

0.160.1

0.30

0.170.09

0.160.09

0.16

0.020.019

ApoC-III in VLDL, g/L

0.0170.016

0.028

0.0270.018

0.0240.016

0.019

0.0340.03

0.0280.022

0.001

0.0320.023

0.0280.019

0.003

Molar ratio of apoC-III to apoB in VLDL

131117

161240

0.18

8048

8562

0.12

Molar ratio of apoC-III to apoB in LDL

2720

2823

0.35

167

167

0.35

ApoC-III in LDL, g/L

CHD denotes coronary heart disease; Q1, quartile 1; Q3, quartile 3; LDL, low-density lipoprotein; HDL, high-density lipoprotein; apo, apolipoprotein; and VLDL, very
low-density lipoprotein. Plus-minus values are meanSD. To convert values for cholesterol to milligrams per deciliter, multiply by 38.6. To convert values for
triglycerides to milligrams per deciliter, multiply by 88.57. The body mass index is weight in kilograms divided by the square of the height in meters.
*Data on women are from the Nurses Health Study; data on men are from the Health Professionals Follow-up Study. Matching criteria were age, smoking status,
and date of blood sampling; among women, additional matching criteria included fasting status at the time of blood sampling.
P values for the difference between cases and controls (unadjusted) were determined by paired Student t test for variables expressed as meanSD, by the
signed-rank test for variables expressed as medians, and by the McNemar 2 test for variables expressed as percentages.
Current aspirin use was defined as 1 to 4 d/wk for women and as 2 times per week for men.

follow-up had significantly higher concentrations of LDL

with apoC-III. Male but not female cases had higher concentrations of LDL without apoC-III than controls (Table 1).
Compared with participants in the lowest quintile, participants in the highest quintile of LDL with apoC-III had a
significantly increased risk of CHD (relative risk for highest
versus lowest quintile, 2.58; 95% confidence interval, 1.78
3.74; P for trend 0.001), conditioning in matching factors
(Table 2). LDL without apoC-III was associated with CHD
but with a lower relative risk than LDL with apoC-III (1.72;
95% confidence interval, 1.14 2.61). Adjustment for other
risk factors in models 2 through 4, including personal history of

diabetes mellitus and triglyceride concentration, had little effect

on the relative risks (Table 2). The risk associated with LDL
with apoC-III was similar in NHS and HPFS participants, with
no significant sex interaction, but the risk associated with LDL
without apoC-III was higher in HPFS (Table I in the online-only
Data Supplement). It was not possible to perform multivariate
logistic analyses in the diabetic subgroup (n126) because of its
small size (model did not converge), but among nondiabetic
participants (n1350), the multivariate-adjusted results were
consistent with those for the complete study sample (relative risk
for top versus bottom quintile of LDL with apoC-III, 2.39; 95%
confidence interval, 1.56 3.67).

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

2068

Circulation

November 8, 2011

Table 2. Relative Risks of Coronary Heart Disease During Follow-Up in the Complete Study Sample, According to the Quintile of
Low-Density Lipoprotein Types (as Measured by the Apolipoprotein B Concentration in Each Fraction) or Apolipoprotein
Concentrations at Baseline
P for
Trend

P for
Difference
Between Sexes

Quintile Categories
1

LDL with apoC-III

n

240

270

297

318

351

Model 1*

1.00 (Ref)

1.38 (0.961.98)

1.71 (1.22.44)

2.12 (1.463.08)

2.58 (1.783.74)

0.001

0.74

Model 2

1.00 (Ref)

1.32 (0.891.98)

1.65 (1.112.45)

1.99 (1.323.00)

2.41 (1.603.64)

0.001

0.97

Model 3

1.00 (Ref)

1.28 (0.861.91)

1.57 (1.062.32)

1.87 (1.242.81)

2.23 (1.483.37)

0.001

0.72

Model 4

1.00 (Ref)

1.16 (0.771.74)

1.35 (0.902.03)

1.50 (0.972.31)

1.64 (1.032.60)

0.001

0.56

265

295

317

288

311

LDL without apoC-III

n
Model 1

1.00 (Ref)

1.4 (0.982.00)

1.60 (1.112.32)

1.38 (0.942.03)

1.72 (1.142.61)

0.06

0.09

Model 2

1.00 (Ref)

1.58 (1.052.38)

1.58 (1.042.40)

1.46 (0.952.25)

1.78 (1.112.84)

0.12

0.053

Model 3

1.00 (Ref)

1.58 (1.062.36)

1.57 (1.042.38)

1.49 (0.972.28)

1.72 (1.082.75)

0.13

0.039

Model 4

1.00 (Ref)

1.53 (1.022.31)

1.45 (0.962.21)

1.36 (0.882.10)

1.44 (0.892.32)

0.29

0.032

ApoC-III in LDL
n

274

270

282

308

342

Model 1

1.00 (Ref)

1.03 (0.721.47)

1.16 (0.811.67)

1.56 (1.072.29)

2.10 (1.413.14)

0.001

0.88

Model 2

1.00 (Ref)

1.1 (0.751.63)

1.25 (0.841.84)

1.53 (1.012.32)

2.19 (1.423.38)

0.001

0.65

Model 3

1.00 (Ref)

1.03 (0.691.53)

1.15 (0.771.71)

1.45 (0.952.21)

1.97 (1.263.06)

0.002

0.50

Model 4

1.00 (Ref)

0.98 (0.661.46)

0.96 (0.641.46)

1.15 (0.741.80)

1.33 (0.812.20)

0.004

0.43

279

296

285

286

330

Model 1

1.00 (Ref)

1.21 (0.851.72)

1.11 (0.781.58)

1.15 (0.791.67)

1.60 (1.092.37)

0.016

0.23

Model 2

1.00 (Ref)

1.17 (0.801.71)

1.11 (0.761.64)

1.12 (0.751.68)

1.62 (1.062.48)

0.023

0.36

Model 3

1.00 (Ref)

1.18 (0.801.75)

1.12 (0.761.66)

1.11 (0.731.68)

1.48 (0.962.28)

0.085

0.40

Model 4

1.00 (Ref)

1.01 (0.681.51)

0.90 (0.61.36)

0.79 (0.501.24)

0.96 (0.601.56)

0.45

0.18

Total plasma apoC-III

n

LDL indicates low-density lipoprotein; apo, apolipoproteins. Relative risks and 95% confidence intervals are given for each quintile compared with the lowest quintile
of each apolipoprotein measurement. The group of women included 320 cases and 320 controls with 14 years of follow-up. The group of men included 419 cases
and 419 controls with 10 years of follow-up. Quintiles and median values of apolipoprotein levels are based on values in controls. For each relative risk, quintile 1
served as the reference group. Matching factors were age, smoking status, and the month of blood sampling. Among women, data were also adjusted for fasting
status at the time of blood sampling.
*Model 1 is conditioned only on matching factors. Model 2 is also adjusted for the presence or absence of a parental history of coronary heart disease before 60
years of age, alcohol intake, and personal history of hypertension. Model 3 is adjusted for all variables in model 2 plus body mass index and personal history of diabetes
mellitus. Model 4 is adjusted for all variables in model 3 plus plasma triglycerides.
P values for trend are based on the median levels of apolipoproteins in quintiles of the controls.
Calculated as the P for the interaction between sex and median apolipoprotein levels in quintiles of the controls.

Independence of Association of LDL With

apoC-III From Other Lipid Risk Factors
We explored whether the association between LDL with
apoC-III and CHD was independent of other established lipid
risk factors. In the background of model 2, we added LDL
cholesterol, HDL cholesterol, plasma apoB, plasma triglycerides, ratio of total cholesterol to HDL cholesterol, or
non-HDL cholesterol, one by one; in all models, LDL with
apoC-III remained significantly associated with CHD (Figure
1). In contrast, LDL without apoC-III was not significantly
independent of LDL cholesterol, plasma triglycerides, ratio of
total cholesterol to HDL cholesterol, or non-HDL cholesterol
(Figure I in the online-only Data Supplement).

No Effect Modification by Hypertriglyceridemia

To address the strength of the relationship between LDL with
apoC-III and CHD at different triglyceride levels, we first

added a quintile interaction term of triglyceridesLDL with

apoC-III to the model, and its coefficient was not significant
(P0.46). Second, we computed the relative risk for LDL
with apoC-III in the participants who had normal or high
triglycerides according to the standard cut point, 150 mg/dL.
The point estimates of the relative risks were similar: 1.61 for
participants with triglycerides 150 mg/dL and 1.77 for
participants with triglycerides 150 mg/dL (P for interaction0.63). Thus, our finding regarding LDL with apoC-III
as an independent predictor of CHD was not significantly
modified by hypertriglyceridemia.

Comparison of LDL With and Without apoC-III

When quintiles of LDL with apoC-III and quintiles of LDL
without apoC-III were included in the same multivariableadjusted model (Figure 2A), only LDL with apoC-III was

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

Mendivil et al

Figure 1. Relative risk of coronary heart disease during

follow-up in the complete study sample, according to levels of
apolipoprotein (apo) B in low-density lipoprotein (LDL) with
apoC-III, in models including other major lipid risk factors. Each
part represents a separate model in which apoB and another
lipid risk factor were mutually adjusted. Solid bars represent relative risks for quintile 5 vs 1 of each risk factor; error bars represent 95% confidence intervals. All models were adjusted for
matching factors, presence or absence of a parental history of
coronary heart disease before 60 years of age, alcohol intake,
and personal history of hypertension. HDL indicates highdensity lipoprotein.

associated with CHD (relative risk, 2.38; 95% confidence

interval, 1.54 3.68; versus relative risk, 1.25; 95% confidence interval, 0.76 2.05 for LDL without apoC-III; P for
difference in slopes 0.001). Given that apoC-III stimulates
hepatic secretion of triglycerides, we performed additional
analyses adding plasma triglycerides to this model (Figure
2B). The association with CHD was still significantly higher
for LDL with apoC-III (P for difference in slopes0.001).
We obtained similar results in a model adjusted for the
same covariates in which cholesterol substituted for apoB.
Cholesterol in LDL with apoC-III (relative risk for highest
versus lowest quintile, 2.01; 95% confidence interval, 1.35
2.99), but not cholesterol in LDL without apoC-III (relative
risk for highest versus lowest quintile, 1.30; 95% confidence
interval, 0.86 1.95), was significantly associated with CHD
(Figure II in the online-only Data Supplement).

Analyses for apoC-III Concentrations

The concentration of apoC-III in LDL was associated with
CHD after adjustment for risk factors (Table 2), but the

LDL With apoC-III and CHD

2069

Figure 2. Relative risk of coronary heart disease during

follow-up in the complete study sample, mutually adjusting for
apolipoprotein (apo) B in low-density lipoprotein (LDL) with and
without apoC-III. Relative risks and 95% confidence intervals
are given for each quintile vs the lowest quintile. A, The model
was also adjusted for matching factors, presence or absence of
a parental history of coronary heart disease before 60 years of
age, alcohol intake, and personal history of hypertension. B, The
model was adjusted for all variables in A plus personal history
of diabetes mellitus and plasma triglycerides. Dark diamonds
represent LDL with apoC-III; P for trend 0.001 in A and P for
trend0.07 in B, Light squares represent LDL without apoC-III;
P for trend0.97 in A and P for trend0.22 in B. P0.001 for
difference in slopes in A and P0.001 for difference in slopes in B.

association was less consistent than for the apoB concentration of LDL with apoC-III (concentration of LDL particles
with any apoC-III; Figure 3). Total plasma apoC-III was
associated with increased risk of CHD (Table 2), but the
association lost significance in model 3 or after addition of
any of the major known lipid risk factors to model 2.

Analyses of VLDL
Concentrations of VLDL with apoC-III and VLDL without
apoC-III were associated with CHD similarly and significantly. Adjustment for body mass index and diabetes mellitus
in model 3 attenuated the relative risks for both VLDL types,
but both were still significant (Table II in the online-only
Data Supplement).

Relative Size of VLDL and LDL With and

Without apoC-III
The mean molar ratio of triglycerides to apoB, an indicator of
VLDL particle size, was higher for VLDL with apoC-III than
for VLDL without apoC-III (11 724 versus 7988) in the study
sample. The mean molar ratio of cholesterol to apoB, an
indicator of LDL particle size, was higher for LDL with
apoC-III than for LDL without apoC-III (2934 versus 2373).

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

elative risk o
of coronary
Re
heart dis
sease

2070

Circulation

November 8, 2011

Tertile 1

3.0
2.5

ApoC-III in LDL

Tertile 2
Tertile 3

2.0
1.5
1.0
0.5
n= 268 108 34
0.0

140 231 121

45 138 389

T til off apoB

Tertile
B in
i LDL with
ith apoC
C III

Figure 3. Relative risk of coronary heart disease during followup, according to the tertile of apolipoprotein (apo) B in lowdensity lipoprotein (LDL) with apoC-III and the tertile of apoC-III
in LDL at baseline. Subjects in tertile 1 of apoB in LDL with
apoC-III and tertile 1 of apoC-III in LDL served as the reference
group. The model was also adjusted for matching factors, presence or absence of a parental history of coronary heart disease
before 60 years of age, alcohol intake, personal history of
hypertension, personal history of diabetes mellitus, and plasma
triglycerides.

Characteristics Associated With a High

Concentration of LDL With apoC-III
Higher levels of LDL with apoC-III were associated with a
markedly higher prevalence of diabetes mellitus, higher
plasma triglycerides, lower HDL cholesterol levels, slightly
higher body mass indexes, and a higher prevalence of
hypertension (Table III in the online-only Data Supplement).
Contrastingly, higher levels of LDL without apoC-III were
not associated with any of these conditions but were associated with higher LDL cholesterol and plasma apoB as was
LDL with apoC-III.

Analyses Excluding Statin Users

In analyses restricted to the 704 participants who never
received cholesterol-lowering medications during follow-up
and adjusted for all variables in model 2, the relative risk of
CHD for the top versus bottom quintile of LDL with apoC-III
was 2.6 (95% confidence interval, 0.877.83; P for
trend0.049).

Use of Postmenopausal Hormone Replacement

Therapy and LDL Types
To explore whether hormone replacement therapy might
confound the association between LDL with apoC-III and
CHD, we compared levels of LDL with and without apoC-III
among women who used or did not use hormone replacement.
Levels of LDL with apoC-III did not differ between users and
nonusers (0.85 mg/dL for users versus 0.81 mg/dL for
nonusers; P0.53). The same was observed for LDL without
apoC-III (77.6 mg/dL for users versus 77.5 mg/dL for
nonusers; P0.96).

Discussion
In this prospective study of individuals initially free of
cardiovascular disease, the concentration of LDL that contains apoC-III was robustly associated with CHD, significantly more so than LDL that does not contain apoC-III.

The association between LDL with apoC-III and CHD was

linear and persisted after adjustment for diverse risk factors.
In fact, the association was only slightly affected by adjustment for alcohol use, family history of CHD, diabetes
mellitus, hypertension, or hormone use, although the study
had limited power to address subtle modification. When
concentrations of LDL with apoC-III and LDL without
apoC-III were included in the same model, only LDL with
apoC-III was associated with CHD, suggesting that a considerable proportion of the CHD risk commonly attributed to
LDL concentrations is in fact due to one of its subfractions
that contains apoC-III. This result extends our previous
finding in patients with preexisting CHD and type 2 diabetes
mellitus.3 The sample size of our study, 739 cases, was able
to identify a minimum increased relative risk of 1.4 for either
type of LDL. However, it is possible that a larger sample size
would identify a mild relation between LDL without apoC-III
and CHD.
Besides its effects that impair lipoprotein metabolism and
stimulate monocyte adhesion to vascular endothelial cells,
apoC-III may promote the inflammatory process that fuels
atherosclerosis through activation of Toll-like receptor 2 in
monocytes25,26 and through induction of insulin resistance
and inflammatory signaling pathways governed by nuclear
factor-B in endothelial cells.10,12 Finally, heterozygote carriers of a nonsense mutation in the apoC-III gene have
significantly less coronary artery calcification and CHD than
noncarrier subjects from the same population.27 All this
evidence provides biological plausibility for a direct involvement of LDL with apoC-III in atherosclerosis.
There are on average 10 to 20 molecules of apoC-III but
only 1 molecule of apoB in each LDL with apoC-III.2 The
apoB concentration of LDL with apoC-III was more strongly
associated with CHD than the apoC-III concentration of the
same LDL. Thus, the number of apoC-III molecules per LDL
may be less important than the concentration of this type of
LDL. This was not what we found for VLDL, in which
particles with or without apoC-III were similarly associated
with CHD risk. Thus, the biological effects of apoC-III may
be lipoprotein specific. In this respect, enhancement of LDL
adhesion to proteoglycans by apoC-III depends on a critical
site in apoB.13 It is conceivable that this site may be exposed
in LDL only after most triglycerides in VLDL have been
lipolyzed.
Our results suggest only a modest association between total
plasma levels of apoC-III and CHD that was attenuated after
adjustment for other lipid risk factors. Plasma apoC-III is a
relatively simple determination in a clinical laboratory,
whereas the measurement of LDL with apoC-III is more
complex and time-consuming but seems to provide more
meaningful information.
The association between LDL with apoC-III and CHD was
independent of LDL cholesterol, in agreement with previous
evidence showing that plasma VLDL and LDL with apoC-III
predict the progression of atherosclerosis even among patients whose LDL cholesterol is markedly reduced with
lovastatin.28 Thus, LDL with apoC-III may explain part of the
residual CHD risk among individuals without elevated LDL
cholesterol.

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

Mendivil et al
Our analyses excluding users of cholesterol-lowering drugs
at any point during follow-up showed that, among never
users, the relative risk estimate for LDL with apoC-III was
quite similar to that in the complete study sample, ruling out
confounding by indication of statins. This result is in accordance with a published analysis of the impact of pravastatin
on concentrations of lipoproteins with and without apoC-III
in patients with diabetes mellitus from the Cholesterol and
Recurrent Events (CARE) trial29: Statin treatment reduced
LDL with apoC-III by 29% and LDL without apoC-III by
30%. In addition, all traditional lipid risk factors were
associated with CHD in our study sample (Table IV in the
online-only Data Supplement), so the results for LDL with
apoC-III do not seem to be the consequence of an unusual
population.
Our indirect estimation of LDL size by the molar ratio of
cholesterol to apoB showed that LDL with apoC-III was on
average larger than LDL without apoC-III, so it seems
unlikely that LDL with apoC-III is just acting as a proxy for
small, dense LDL. ApoC-III has been reported to transfer
from VLDL to HDL during freezing or storage.30 In this
regard, we performed a small experiment in samples from 4
volunteers measuring apoC-III in VLDL, LDL, and HDL
immediately after plasma separation and after 5 months of
storage at 80C. The distribution of apoC-III changed only
very slightly and nonsignificantly (from 14% in VLDL, 16%
in LDL, and 70% in HDL when analyzed fresh to 12% in
VLDL, 15% in LDL, and 73% in HDL after freezing/storage/
thawing), so we believe this effect is unlikely to have had an
important influence on our results under our sample handling
and preserving procedures. Another potential concern is
whether a high concentration of LDL with apoC-III might
just be identifying subjects with hypertriglyceridemia and
mild insulin resistance. Although there was a significant
correlation between LDL with apoC-III and plasma triglycerides (r0.42), the association between CHD and LDL with
apoC-III persisted after adjustment for triglycerides, suggesting that both characteristics may have common origins but
are independently associated with CHD.
A limitation of our study is that we did not have baseline
measurements of plasma glucose and insulin that would have
permitted a better adjustment for carbohydrate metabolism,
so we relied on self-reported diagnosis of diabetes mellitus.
Another relevant limitation is that we performed multiple
statistical tests to explore multiple factors of interest and
multiple models per factor. This may have inflated the rate of
type I error above the nominal 5% used per test.
We cannot be certain whether apoC-III is causally responsible for CHD or a marker for other properties of certain LDL
particles. Human LDL with apoC-III has been found to have
a different lipid composition than LDL without apoC-III2,13
that may influence the effect of LDL with apoC-III on atherosclerosis. Nevertheless, when our results are combined with the
evidence of reduced atherosclerosis in humans with genetically
reduced apoC-III27 and with the abundant documentation of
direct proatherogenic effects of apoC-III,10 13,25,26 they encourage further research into apoC-III as a potentially causative
factor in atherosclerosis and cardiovascular disease and as a
target for therapy. Ultimately, even if the presence of apoC-III is

LDL With apoC-III and CHD

2071

just a label for a type of LDL very strongly associated with

CHD, the existence of a marker for such particles can prove
useful in the targeting and optimization of preventive
interventions.

Conclusions
Our results suggest that the concentration of LDL with
apoC-III is independently associated with the risk of CHD
and that much of the risk ordinarily attributed to LDL may in
fact be due to LDL particles that contain apoC-III. These
findings are consistent with prior in vitro, animal, and
epidemiological evidence suggesting enhanced atherogenicity of lipoproteins containing apoC-III. More than contributing to the improvement of already excellent prediction
models,3133 our results may endorse LDL with apoC-III as a
highly attractive target for the prevention or treatment of
atherosclerotic diseases.

Sources of Funding
This work was supported by National Institutes of Health/National
Heart, Lung, and Blood Institute grant HL070159.

Disclosures
Dr Sacks has received research funding from ISIS Pharmaceuticals
to investigate the effect of a biotechnology product on apoC-III
levels. The other authors report no conflicts.

References
1. Alaupovic P. Significance of apolipoproteins for structure, function, and
classification of plasma lipoproteins. Methods Enzymol. 1996;263:32 60.
2. Campos H, Perlov D, Khoo C, Sacks FM. Distinct patterns of lipoproteins
with apoB defined by presence of apoE or apoC-III in hypercholesterolemia and hypertriglyceridemia. J Lipid Res. 2001;42:1239 1249.
3. Lee SJ, Campos H, Moye LA, Sacks FM. LDL containing apolipoprotein
CIII is an independent risk factor for coronary events in diabetic patients.
Arterioscler Thromb Vasc Biol. 2003;23:853 858.
4. Ooi EM, Barrett HM, Chan D, Watts GF. Apolipoprotein C-III: understanding an emerging cardiovascular risk factor. Clin Sci. 2008;114:
611 624.
5. Zheng C, Khoo C, Ikewaki K, Sacks FM. Rapid turnover of apolipoprotein C-III-containing triglyceride-rich lipoproteins contributing to the
formation of LDL subfractions. J Lipid Res. 2007;48:1190 1203.
6. Mendivil CO, Zheng C, Furtado J, Lel J, Sacks FM. Metabolism of
very-low-density lipoprotein and low-density lipoprotein containing apolipoprotein C-III and not other small apolipoproteins. Arterioscler
Thromb Vasc Biol. 2010;30:239 245.
7. Kawakami A, Yoshida M. Apolipoprotein CIII links dyslipidemia with
atherosclerosis. J Atheroscler Thromb. 2009;16:6 11.
8. Sehayek E, Eisenberg S. Mechanisms of inhibition by apolipoprotein C of
apolipoprotein E dependent cellular metabolism of human triglyceride
rich lipoproteins through the low density lipoprotein receptor pathway.
J Biol Chem. 1991;266:18259 18267.
9. Zheng C, Khoo C, Furtado J, Sacks FM. Apolipoprotein C-III and the
metabolic basis for hypertriglyceridemia and the dense low-density lipoprotein phenotype. Circulation. 2010;121:17221734.
10. Kawakami A, Aikawa M, Alcaide P, Luscinskas FW, Libby P, Sacks FM.
Apolipoprotein CIII induces expression of vascular cell adhesion
molecule-1 in vascular endothelial cells and increases adhesion of
monocytic cells. Circulation. 2006;114:681 687.
11. Kawakami A, Aikawa M, Libby P, Alcaide P, Luscinskas FW, Sacks FM.
Apolipoprotein CIII in apolipoprotein B lipoproteins enhances the
adhesion of human monocytic cells to endothelial cells. Circulation.
2006;113:691700.
12. Kawakami A, Osaka M, Tani M, Azuma H, Sacks FM, Shimokado K,
Yoshida M. Apolipoprotein CIII links hyperlipidemia with vascular endothelial function. Circulation. 2008;118:731742.
13. Hiukka A, Sthlman M, Pettersson C, Levin M, Adiels M, Teneberg S,
Leinonen ES, Hulten LM, Wiklund O, Oresic M, Olofsson SO, Taskinen
MR, Ekroos K, Boren J. ApoCIII-enriched LDL in type 2 diabetes

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

2072

14.

15.

16.

17.

18.

19.

20.

21.

22.

Circulation

November 8, 2011

displays altered lipid composition, increased susceptibility for sphingomyelinase, and increased binding to biglycan. Diabetes. 2009;58:
2018 2026.
Blankenhorn DH, Alaupovic P, Wickham E, Chin HP, Azen SP. Prediction of angiographic change in native human coronary arteries and
aortocoronary bypass grafts: lipid and nonlipid factors. Circulation. 1990;
81:470 476.
Hodis HN, Mack WJ, Azen SP, Alaupovic P, Pogoda JM, LaBree L,
Hemphill LC, Kramsch DM, Blankenhorn DH. Triglyceride- and cholesterol-rich lipoproteins have a differential effect on mild/moderate and
severe lesion progression as assessed by quantitative coronary
angiography in a controlled trial of lovastatin. Circulation. 1994;90:
42 49.
Luc G, Fievet C, Arveiler D, Evans AE, Bard JM, Cambien F, Fruchart
JC, Ducimetiere P. Apolipoproteins C-III and E in apoB- and non-apoB
containing lipoproteins in two populations at contrasting risk for myocardial infarction: the ECTIM study: Etude Cas Temoins sur Infarctus du
Myocarde. J Lipid Res. 1996;37:508 517.
Sacks FM, Alaupovic P, Moye LA, Cole TG, Sussex B, Stampfer MJ,
Pfeffer MA, Braunwald E. VLDL, apolipoproteins B, CIII and E and risk
of recurrent coronary events in the Cholesterol and Recurrent Events
(CARE) trial. Circulation. 2000;102:1886 1892.
Rimm EB, Giovannucci EL, Stampfer MJ, Colditz GA, Litin LB, Willett
WC. Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals. Am J Epidemiol. 1992;135:1114 1126.
Colditz GA, Manson JE, Hankinson SE. The Nurses Health Study:
20-year contribution to the understanding of health among women.
J Womens Health. 1997;6:49 62.
Giovannucci E, Colditz G, Stampfer MJ, Rimm EB, Litin L, Sampson L,
Willett WC. The assessment of alcohol consumption by a simple selfadministered questionnaire. Am J Epidemiol. 1991;133:810 817.
Chasan-Taber S, Rimm EB, Stampfer MJ, Spiegelman D, Colditz GA,
Giovannucci E, Ascherio A, Willett WC. Reproducibility and validity of
a self-administered physical activity questionnaire for male health professionals. Epidemiology. 1996;7:81 86.
Prentice RL, Breslow NE. Retrospective studies and failure time models.
Biometrika. 1978;65:153158.

23. Khoo C, Judge H, Sacks FM. Effects of estrogenic oral contraceptives on

the lipoprotein B particle system defined by apolipoproteins E and C-III
content. J Lipid Res. 1999;40:202212.
24. Pietzsch J, Subat S, Nitzsche S, Leonhardt W, Schentke KU, Hanefeld M.
Very fast ultracentrifugation of serum lipoproteins: influence on lipoprotein separation and composition. Biochim Biophys Acta. 1995;1254:
77 88.
25. Kawakami A, Osaka M, Aikawa M, Uematsu S, Akira S, Libby P,
Shimokado K, Sacks FM, Yoshida M. Toll-like receptor 2 mediates
apolipoprotein CIII-induced monocyte activation. Circ Res. 2008;103:
14021409.
26. Kawakami A, Aikawa M, Nitta N, Yoshida M, Libby P, Sacks FM.
Apolipoprotein CIII-induced THP-1 cell adhesion to endothelial cells
involves pertussis toxin-sensitive G protein- and protein kinase C alphamediated nuclear factor-kappaB activation. Arterioscler Thromb Vasc
Biol. 2007;27:219 225.
27. Pollin TI, Damcott CM, Shen H, Ott SH, Shelton J, Horenstein RB, Post
W, McLenithan JC, Bielak LF, Peyser PA, Mitchell BD, Miller M,
OConnell JR, Shuldiner AR. A null mutation in human APOC3 confers
a favorable plasma lipid profile and apparent cardioprotection. Science.
2008;322:17021705.
28. Alaupovic P, Mack WJ, Knight-Gibson C, Hodis HN. The role of triglyceride-rich lipoprotein families in the progression of atherosclerotic
lesions as determined by sequential coronary angiography from a controlled clinical trial. Arterioscler Thromb Vasc Biol. 1997;17:715722.
29. Lee SJ, Sacks FM. Effect of pravastatin on intermediate-density and
low-density lipoproteins containing apolipoprotein CIII in patients with
diabetes mellitus. Am J Cardiol. 2003;92:121124.
30. Cohn JS, Rodriguez C, Jacques H, Tremblay M, Davignon J. Storage of
human plasma samples leads to alterations in the lipoprotein distribution
of apoC-III and apoE. J Lipid Res. 2004;45:15721579.
31. Wilson PW, DAgostino RB, Levy D, Belanger AM, Silbershatz H,
Kannel WB. Prediction of coronary heart disease using risk factor categories. Circulation. 1998;97:18371847.
32. Assmann G, Cullen P, Schulte H. Simple scoring scheme for calculating
the risk of acute coronary events based on the 10-year follow-up of the
prospective cardiovascular Munster (PROCAM) study. Circulation.
2002;105:310 315.
33. Ridker PM, Buring JE, Rifai N, Cook NR. Development and validation of
improved algorithms for the assessment of global cardiovascular risk in
women: the Reynolds Risk Score. JAMA. 2007;297:611 619.

CLINICAL PERSPECTIVE
During the last few years, evidence has accumulated indicating that the protein composition of lipoproteins may be more
relevant to atherogenesis and cardiovascular risk than the absolute concentrations of plasma lipids. One of these
compositional factors is apolipoprotein (apo) C-III, a small interchangeable apolipoprotein that impairs hepatic uptake of
circulating lipoproteins and has various direct proatherogenic effects on the arterial wall. In this study, we prospectively
analyzed the association between plasma concentrations of low-density lipoprotein (LDL) that contains or does not contain
apoC-III and the appearance of coronary heart disease (CHD) events in 1400 men and women. Our key finding was that
in a multivariate-adjusted model that includes concentrations of both LDL with apoC-III and LDL without apoC-III, only
LDL with apoC-III was consistently and significantly associated with risk, indicating that an important fraction of the
cardiovascular risk traditionally adjudicated to LDL may in fact be due to LDL with apoC-III. The findings were robust
to adjustment for levels of plasma lipids, demographic cardiovascular risk factors, and diabetes status. The identification
of LDL with apoC-III as a lipoprotein type disproportionately associated with the development of CHD adds to the
enlarging body of evidence suggesting the involvement of apoC-III in atherogenesis and metabolic derangements and
should stimulate further research exploring apoC-III as an interventional target in the prevention of CHD.

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

Low-Density Lipoproteins Containing Apolipoprotein C-III and the Risk of Coronary

Heart Disease
Carlos O. Mendivil, Eric B. Rimm, Jeremy Furtado, Stephanie E. Chiuve and Frank M. Sacks
Circulation. 2011;124:2065-2072; originally published online October 10, 2011;
doi: 10.1161/CIRCULATIONAHA.111.056986
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright 2011 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539

The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circ.ahajournals.org/content/124/19/2065

Data Supplement (unedited) at:

http://circ.ahajournals.org/content/suppl/2011/10/07/CIRCULATIONAHA.111.056986.DC1.html

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial
Office. Once the online version of the published article for which permission is being requested is located,
click Request Permissions in the middle column of the Web page under Services. Further information about
this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation is online at:
http://circ.ahajournals.org//subscriptions/

Downloaded from http://circ.ahajournals.org/ by guest on March 22, 2016

Low density lipoproteins containing apolipoprotein C-III and the risk of

coronary heart disease
Mendivil CO, Rimm EB, Furtado J, Chiuve S, Sacks FM

Supplemental material

Supplemental Table 1.
Men
Quintile categories
3
4
Variable*
n=143
n=156
n=153
n=183
Apolipoprotein B in LDL with apolipoprotein CIII
Range - g/L
<0.060
0.060-0.086
0.086-0.115
0.115-0.161
1

5
n=201

P for
trend**

>=0.161

Median - g/L

0.046

0.072

0.10

0.13

0.20

Model 1

1.00 (ref)

1.28 (0.81-2.02)

1.25 (0.78-2.03)

2.05 (1.24-3.41)

2.42 (1.48-3.96)

<0.001

Model 2

1.00 (ref)

1.52 (0.87-2.68)

1.23 (0.68-2.20)

1.97 (1.07-3.63)

2.47 (1.37-4.44)

0.002

Model 3
1.00 (ref) 1.38 (0.77-2.46) 1.13 (0.62-2.04)
Apolipoprotein B in LDL without apolipoprotein CIII
Range - g/L
<0.551
0.551-0.671
0.671-0.766

1.77 (0.95-3.31)

2.22 (1.22-4.04)

0.007

0.766-0.909

>=0.909

Median - g/L

0.47

0.61

0.83

1.00

Model 1

1.00 (ref)

1.51 (0.94-2.4)

1.32 (0.8-2.17)

1.59 (0.96-2.62)

2.21 (1.26-3.87)

0.012

Model 2

1.00 (ref)

2.07 (1.15-3.70)

1.45 (0.79-2.67)

1.98 (1.06-3.71)

2.31 (1.16-4.61)

0.041

Model 3
1.00 (ref) 2.24 (1.23-4.08)
Apolipoprotein C-III in LDL
Range - g/L
<0.014
0.014-0.021

1.48 (0.79-2.78)

2.09 (1.1-3.96)

2.45 (1.21-4.97)

0.035

0.021-0.027

0.027-0.038

>=0.038

0.024

0.031

0.050

0.018

0.71

Median - g/L

0.010

Model 1

1.00 (ref)

0.8 (0.5-1.27)

0.94 (0.6-1.48)

1.33 (0.83-2.13)

1.76 (1.08-2.86)

0.002

Model 2

1.00 (ref)

0.8 (0.45-1.44)

0.97 (0.56-1.67)

1.34 (0.76-2.37)

2.18 (1.2-3.97)

0.001

Model 3
1.00 (ref) 0.7 (0.38-1.26)
Total plasma apolipoprotein C-III
Range - g/L
<0.099
0.099-0.130

0.91 (0.51-1.61)

1.22 (0.68-2.2)

1.94 (1.04-3.6)

0.003

0.130-0.161

0.161-0.206

>=0.206

Median - g/L

0.079

0.12

0.14

0.19

0.27

Model 1

1.00 (ref)

1.05 (0.65-1.70)

0.86 (0.53-1.41)

1.22 (0.75-2.00)

1.72 (1.03-2.87)

0.009

Model 2

1.00 (ref)

0.83 (0.45-1.5)

0.64 (0.35-1.19)

1.44 (0.79-2.61)

1.85 (0.99-3.44)

0.005

Model 3

1.00 (ref)

0.84 (0.45-1.56)

0.56 (0.29-1.09)

1.19 (0.63-2.26)

1.53 (0.78-2.98)

0.022

Supplemental Table 1, continued.

Women
Quintile categories
1
2
3
4
Variable*
n=97
n=114
n=144
n=135
Apolipoprotein B in LDL with apolipoprotein CIII
Range g/L
<0.034
0.034-0.050
0.050-0.072
0.072-0.106

5
n=150

P for
trend

>=0.106

Median - g/L

0.028

0.042

0.06

0.08

0.14

Model 1

1.00 (ref)

1.60 (0.88-2.89)

2.46 (1.42-4.25)

2.28 (1.30-3.99)

2.87 (1.61-5.11)

0.002

Model 2

1.00 (ref)

1.54 (0.71-3.37)

2.23 (1.08-4.61)

1.97 (0.93-4.17)

2.70 (1.27-5.72)

0.025

Model 3

1.00 (ref)

1.37 (0.61-3.07)

2.02 (0.95-4.27)

1.83 (0.85-3.95)

2.43 (1.11-5.3)

0.054

Apolipoprotein B in LDL without apolipoprotein CIII

Range - g/L

<0.55

0.55-0.68

0.68-0.82

0.82-0.96

>=0.96

Median - g/L

0.47

0.62

0.75

0.88

1.13

Model 1

1.00 (ref)

1.27 (0.73-2.24)

1.91 (1.09-3.35)

1.15 (0.63-2.08)

1.28 (0.68-2.41)

0.80

Model 2

1.00 (ref)

1.14 (0.54-2.44)

1.30 (0.60-2.80)

1.07 (0.48-2.40)

1.11 (0.47-2.64)

0.97

Model 3
1.00 (ref) 1.06 (0.47-2.36)
Apolipoprotein C-III in LDL
Range - g/L
<0.011
0.011-0.018

1.08 (0.48-2.44)

1.04 (0.44-2.46)

1.07 (0.41-2.78)

0.92

0.018-0.027

0.027-0.044

>=0.044

Median - g/L

0.008

0.014

0.021

0.033

0.060

Model 1

1.00 (ref)

1.58 (0.88-2.84)

1.75 (0.94-3.27)

2.15 (1.11-4.17)

2.92 (1.41-6.03)

0.012

Model 2

1.00 (ref)

1.89 (0.9-3.98)

2.14 (0.98-4.67)

2.4 (1.06-5.44)

2.36 (0.97-5.75)

0.25

Model 3

1.00 (ref)

1.97 (0.9-4.3)

2.13 (0.95-4.76)

2.38 (1.01-5.64)

2.26 (0.89-5.75)

0.37

0.13-0.17

0.17-0.21

>=0.21

Total plasma apolipoprotein C-III

Range g/L
<0.093
0.093-0.13
Median - g/L

0.068

0.11

0.15

0.18

0.25

Model 1

1.00 (ref)

1.39 (0.82-2.34)

1.43 (0.85-2.41)

1.00 (0.56-1.79)

1.35 (0.73-2.48)

0.62

Model 2

1.00 (ref)

1.95 (1.01-3.79)

1.48 (0.75-2.90)

1.23 (0.58-2.61)

1.44 (0.67-3.13)

0.74

Model 3

1.00 (ref)

2.23 (1.10-4.52)

1.39 (0.68-2.83)

1.24 (0.57-2.73)

1.22 (0.53-2.79)

0.82

Supplemental Table 1. Relative risks of coronary heart disease during follow-up, according to the
quintile of apolipoprotein concentrations at baseline, by sex. * Model 1 was conditioned on matching
factors (age, smoking status, and the month of blood sampling). Among women, data were also
conditioned on fasting status at the time of blood sampling. Model 2 was additionally adjusted for
presence or absence of a parental history of coronary heart disease before the age of 60 years, alcohol
intake and personal history of hypertension. Model 3 was adjusted for all the variables in model 2 plus
body-mass index and personal history of diabetes. P values for trend are based on the median
apolipoprotein levels in quintiles of the controls.

Supplemental Table 2.
Total sample

Variable*

Relative risk for

quintile 5
compared to
quintile 1

ApoB in VLDL with apoC-III

Range-(g/L)
Median-(g/L)
Model 1
2.07 (1.45-2.95)

Men

Women

P for
trend

Relative risk for

quintile 5 compared to
quintile 1

P for
trend

P for
difference
between
sexes

0.001

0.09

P for
trend

Relative risk for

quintile 5 compared
to quintile 1

<0.001

Q1<0.010, Q5>0.028
Q1: 0.006, Q5: 0.036
1.85 (1.17-2.94)

0.002

Q1<0.004, Q5>0.014
Q1: 0.002, Q5: 0.018
2.43 (1.39-4.26)

Model 2

1.96 (1.34-2.88)

<0.001

1.97 (1.12-3.47)

0.005

2.22 (1.05-4.68)

0.08

0.30

Model 3

1.73 (1.17-2.57)

0.002

1.79 (1.00-3.20)

0.011

1.85 (0.83-4.11)

0.27

0.53

ApoB in VLDL without apoC-III

Range-g/L

Q1<0.011, Q5>0.047

Q1<0.011, Q5>0.046

Median

Q1: 0.006, Q5: 0.062

Q1: 0.007, Q5: 0.064

Model 1

2.55 (1.72-3.79)

<0.001

3.06 (1.78-5.26)

<0.001

2.08 (1.15-3.73)

0.009

0.44

Model 2

2.26 (1.47-3.46)

<0.001

2.9 (1.49-5.65)

0.001

1.78 (0.81-3.87)

0.17

0.34

Model 3

1.98 (1.27-3.08)

0.006

2.56 (1.28-5.12)

0.006

1.55 (0.69-3.49)

0.26

0.39

ApoC-III in VLDL
Range-g/L

Q1<0.011, Q5>0.035

Q1<0.006, Q5>0.025

Median

Q1: 0.008, Q5: 0.043

Q1: 0.004, Q5: 0.035

Model 1

1.44 (1.01-2.05)

0.015

1.25 (0.78-2.01)

0.25

1.79 (1.04-3.11)

0.012

0.17

Model 2

1.47 (1.00-2.16)

0.03

1.64 (0.93-2.90)

0.08

1.26 (0.61-2.60)

0.32

0.52

Model 3

1.32 (0.88-1.96)

0.11

1.49 (0.83-2.68)

0.15

1.15 (0.54-2.46)

0.62

0.71

Supplemental Table 2. Association between apolipoprotein concentrations in VLDL and coronary heart
disease in the complete study sample and by sex. * Model 1 was conditioned on matching factors. Model
2 was additionally adjusted for presence or absence of a parental history of coronary heart disease before
the age of 60 years, alcohol intake and personal history of hypertension. Model 3 was adjusted for
everything in model 2 plus body-mass index and personal history of diabetes. P values for trend are
based on the median apolipoprotein levels in quintiles of the controls. Calculated as the P value for the
interaction between sex and median apolipoprotein levels in quintiles of the controls.

Supplemental Table 3.
A.

Median apoB in LDL with

apoC-III (g/L) *
Age (years)
BMI (Kg/m2)
Alcohol intake (g/day)
Physical activity
(met/hr/week)
Triglycerides (mmol /L)
Total cholesterol (mmol /L)
LDL cholesterol (mmol /L)
HDL cholesterol (mmol /L)
Apolipoprotein B (g/L)
CRP (median) (mg/L)
Male sex (%)
Current smokers (%)
Fasting (%)
Postmenopausal women (%)
Parental history of CHD (%)
Diabetes (%)
Hypertension (%)
B.

Median apoB in LDL without

apoC-III (g/L) *
Age (years)
BMI (Kg/m2)
Alcohol intake (g/day)
Physical activity
(met/hr/week)
Triglycerides (mmol /L)
Total cholesterol (mmol/L)
LDL cholesterol (mmol /L)
HDL cholesterol (mmol /L)
Apolipoprotein B (g/L)
CRP (median) (mg/L)
Male sex (%)
Current smokers (%)
Fasting (%)
Postmenopausal women (%)
Parental history of CHD (%)
Diabetes (%)
Hypertension (%)

Quintile of apoB in LDL with apoC-III

Q1
Q2
Q3
Q4
Q5
n=240 n=270 n=297 n=318
n=351
0.038
63.0
25.0
10.2

0.059
62.1
25.5
7.7

0.082
63.2
26.1
7.4

0.11
63.4
26.0
8.3

0.17
61.5
26.7
10.1

28.1
30.2
27.9
30.1
29.1
0.97
1.12
1.26
1.44
2.02
5.24
5.45
5.54
5.95
6.35
3.11
3.26
3.40
3.74
3.92
1.41
1.35
1.27
1.25
1.18
0.70
0.81
0.87
0.99
1.12
0.36
0.36
0.39
0.38
0.39
59.6
57.8
51.5
57.6
57.3
18.8
12.6
14.1
16.4
17.4
62.9
70.7
71.4
70.1
72.4
85.1
86.2
90.9
93.1
90.7
16.7
14.8
13.1
15.7
16.2
2.5
8.5
8.1
6.3
15.1
28.8
37.0
36.0
39.3
41.0
Quintile of apoB in LDL without apoC-III
Q1
Q2
Q3
Q4
Q5
n=265 n=295 n=317 n=288
n=311

p-trend
0.58
0.007
0.96
0.68
0.017
0.006
0.002
0.001
0.001
0.13
0.68
0.92
0.13
0.07
0.99
0.11
0.04

0.47
63.4
26.0
8.1

0.62
62.5
25.6
9.2

0.73
62.4
25.9
7.8

0.85
62.7
26.2
9.4

1.05
62.3
26.0
9.3

p-trend
0.15
0.41
0.37

27.7
1.42
5.08
2.94
1.28
0.58
0.36
55.9
15.1
61.9
92.2
10.9
9.8
37.4

31.5
1.36
5.51
3.28
1.32
0.76
0.28
58.0
16.3
69.2
81.6
15.9
9.2
39.3

24.2
1.37
5.70
3.47
1.31
0.88
0.44
50.5
17.7
71.3
87.5
16.4
9.8
42.0

32.0
1.41
5.99
3.69
1.26
1.02
0.42
59.4
13.2
74.3
94.6
16.7
8.0
34.4

30.3
1.47
6.40
4.16
1.23
1.30
0.38
59.8
16.7
71.7
92.7
16.1
6.1
31.5

0.64
0.29
0.001
0.002
0.19
0.002
0.45
0.52
0.98
0.08
0.47
0.17
0.06
0.24

Supplemental Table 3. Association between baseline characteristics (complete study sample) and plasma
levels of apoB in LDL. A. Baseline covariates across quintiles of apoB in LDL with apoC-III. B.
Baseline covariates across quintiles of apoB in LDL without apoC-III. Data are means unless specified
otherwise. * Quintiles were calculated using sex-specific apoB levels among controls, the reported
medians correspond to a weighted average of values in men and women.

Supplemental Table 4.

Variable

Relative risk for

quintile 5
compared to
quintile 1

Total cholesterol
Relative risk
1.80 (1.22-2.64)
LDL cholesterol
Relative risk
1.79 (1.23-2.64)
Plasma triglycerides
Relative risk
2.17 (1.51-3.11)
Plasma apolipoprotein B
Relative risk
3.11 (1.95-4.97)
Non HDL colesterol
Relative risk
2.53 (1.72-3.72)
HDL cholesterol
Relative risk
0.49 (0.29-0.82)

P for
trend*

<0.001
<0.001
<0.001
<0.001
<0.001
<0.001

Supplemental Table 4. Association between traditional lipid risk factors and coronary heart disease in
the complete study sample. Model was conditioned on matching factors, and additionally adjusted for the
presence or absence of a parental history of coronary heart disease before the age of 60 years, alcohol
intake and personal history of hypertension. * P values for trend are based on the median levels of each
variable in quintiles of the controls.

Supplemental Figure 1.

4,0

10,0
9,0
8,0
7,0

3,0

6,0
5,0
4,0
3,0
2,0
1,0
0,0

2,0
1,0

Risk of coronary heart disease

in quintile 5 relative to quintile 1

0,0

ApoB
in LDL without
apoC-III

ApoB
in LDL without
apoC-III

LDL
cholesterol

2,0

2,0

1,0

1,0
ApoB
in LDL without
apoC-III

Plasma
triglycerides

0,0

7,0

6,0

6,0

5,0

5,0

ApoB
in LDL without
apoC-III

HDL
cholesterol

4,0

4,0

3,0

3,0

2,0

2,0

1,0

1,0
0,0

Plasma
apoB

3,0

3,0

0,0

ApoB
in LDL without
apoC-III

Total to HDL
cholesterol
ratio

0,0

ApoB
in LDL without
apoC-III

Non-HDL
cholesterol

Supplemental Figure 1. Relative risk of coronary heart disease during follow up in the complete study
sample, according to levels of apoB in LDL without apoC-III, in models including other major lipid risk
factors. Each panel represents a separate model in which apoB and another lipid risk factor were mutually
adjusted. Solid bars represent relative risks for quintile 5 compared to quintile 1 of each risk factor, and
error bars represent 95% confidence intervals. All models were conditioned on matching factors, and
additionally adjusted for the presence or absence of a parental history of coronary heart disease before the
age of 60 years, alcohol intake and personal history of hypertension. * To be interpreted with caution,
because the linear correlation coefficient between the two variables was 0.92.

Relative risk of coronary

heart disease

Supplemental Figure 2.

4.0

2.0

1.0

0.5

Cholesterol
in LDL with
apoC-III

Cholesterol
in LDL without
apoC-III

Supplemental Figure 2. Relative risk of CHD during follow-up in the complete study sample, mutually
adjusting levels of cholesterol in LDL with and without apoC-III. Relative risks and 95% confidence
intervals are given for quintile 5 compared to quintile 1. The model was conditioned on matching
factors, and additionally adjusted for the presence or absence of a parental history of coronary heart
disease before the age of 60 years, alcohol intake and personal history of hypertension.