Hypertriglyceridemia as a
Cardiovascular Risk Factor
Melissa A. Austin,
PhD,
John E. Hokanson,
MPH, PhC,
and Karen L. Edwards,
PhD
To determine the relation between plasma triglyceride
levels and the risk of incident cardiovascular disease, the
semiquantitative techniques of meta-analysis were applied to 17 population-based prospective studies of triglyceride and cardiovascular disease. Sixteen of these
studies represented 2,445 events among 46,413 Caucasian men followed for an average period of 8.4 years,
and 5 studies represented 439 events among 10,864
Caucasian women followed for an average period of
11.4 years. Univariate relative risk (RR) estimates for
incident cardiovascular disease associated with a
1-mmol/L increase in triglyceride was 1.07–1.98 in
men, with a summary RR of 1.32 (95% confidence interval [CI]: 1.26 –1.39), indicating a 32% increase in
disease risk associated with increased triglyceride. In the
studies involving women, individual RR estimates for
triglyceride were 1.69 –2.05, with a summary RR of
1.76 (95% CI: 1.50 –2.07), indicating a 76% increase in
disease risk associated with increased triglyceride. After
adjustment for high-density lipoprotein cholesterol and
other risk factors, these risks were decreased to 14% in
men and 37% in women but remained statistically significant. Three recent prospective epidemiologic studies
have also shown that plasma triglyceride and low-density lipoprotein particle size predict subsequent coronary
artery disease in Caucasian populations. Taken together, these studies demonstrate the importance of triglyceride levels as a risk factor for cardiovascular
disease. Q1998 by Excerpta Medica, Inc.
Am J Cardiol 1998;81(4A):7B–12B
levated levels of plasma triglyceride have long
been associated with an increased risk of cardioE
vascular disease. In 1959, a case-control study by
some studies, the evidence for a causal relation (with
coronary artery disease) is still incomplete.”4 Similarly, data from the Lipid Research Clinics (LRC)
Follow-up Study demonstrated that triglyceride was
related to 12-year coronary artery disease mortality in
both men and women5 but this relation was not statistically significant after adjustment for covariates.
Thus, the role of triglyceride as a risk factor for
coronary artery disease remains to be fully elucidated.
This article will focus on the relation between
triglyceride and cardiovascular disease. First, the
semiquantitative techniques of meta-analysis will be
applied to population-based, prospective studies of
triglyceride and cardiovascular disease to estimate the
strength of this association in the general population
and to evaluate the effect of other risk factors, especially HDL cholesterol.6,7 Then, 3 recent prospective
studies that examined both plasma triglyceride and
low-density lipoprotein (LDL) particle diameter (size)
will be summarized.
Albrink and Man1 showed that fasting triglyceride
levels were increased among patients with coronary
artery disease compared with control subjects. The
earliest prospective study of triglyceride and ischemic
heart disease demonstrated an increased incidence of
ischemic heart disease among men with elevated triglyceride levels at baseline compared with men with
lower levels.2 However, even in this early study, the
investigators speculated that the triglyceride association may not have been independent of other plasma
lipid levels.
In the decades after these investigations, an extensive literature has examined the role of triglyceride as
a risk factor for cardiovascular disease. In a review of
these studies, Austin noted that most analyses showed
a relation between triglyceride and coronary artery
disease.3 However, a number of studies found that this
association did not remain statistically significant after
controlling for other lipid risk factors, especially highdensity lipoprotein (HDL) cholesterol. In reviewing
this and other evidence, the National Institutes of
Health Consensus Development Panel on Triglyceride, High-Density Lipoprotein, and Coronary Heart
Disease concluded that “for triglyceride, the data are
mixed. Although strong associations are found in
From the Department of Epidemiology, School of Public Health and
Community Medicine, and the Division of Metabolism, Endocrinology,
and Nutrition, Department of Medicine, School of Medicine, University of Washington, Seattle, Washington.
This work was supported by NIH R01 Grant HL45913 and was
performed during Dr. Austin’s tenure as an Established Investigator of
the American Heart Association.
Address for reprints: Melissa A. Austin, PhD, Department of Epidemiology, Box 357236, 1959 NE Pacific Avenue, University of
Washington, Seattle, Washington 98195-7236.
©1998 by Excerpta Medica, Inc.
All rights reserved.
META-ANALYSIS OF TRIGLYCERIDE
AND CARDIOVASCULAR DISEASE
Published studies were selected for the meta-analysis of triglyceride and cardiovascular disease based
on several criteria.6 First, only investigations that used
a prospective study design were chosen, ensuring that
elevations in plasma triglyceride preceded the onset of
disease. Second, only studies that used populationbased samples of subjects were selected, so that relative risk (RR) estimates were as applicable as possible
to the general population. Because the results from the
LRC Follow-up Study5 were based on a sample enriched with hyperlipidemic subjects, the RR values for
this analysis were adjusted for ascertainment (hyperlipidemic vs random sample) based on results kindly
provided by Dr. M. H. Criqui (personal communication). To exclude the possibility of postprandial ef0002-9149/98/$19.00
PII S0002-9149(98)00031-9
7B
fects, only studies that evaluated fasting triglyceride
levels were included. Each study cohort was included
only once in the analysis, using the publication that
reported the longest follow-up period. Both fatal and
nonfatal cardiovascular endpoints were included, although most studies focused on myocardial infarction
or coronary artery disease death. Finally, only Caucasian subjects were included in the analysis, since few
data were available on other ethnic groups.
A total of 17 studies2,5,8 –24 conforming to these
selection criteria were identified, including 16 studies
representing 2,445 events among 46,413 men followed for an average period of 8.4 years,2,5,8 –13,15–24
and 5 studies representing 439 events among 10,864
women followed for an average period of 11.4
years.5,10,14,16,18 Subjects were 15– 81 years of age in
all of the studies. Among men, 6 studies were from the
United States, 6 from Scandinavian countries, and 1
each from France, Germany, Italy, and the United
Kingdom. Among women, 3 studies were from the
United States and 2 from Scandinavia.
The meta-analysis was performed separately for
men and women using the techniques described by
Greenland.7 Briefly, RR estimates were determined
for each study by calculating b, the estimated slope,
from logistic regression analysis, standardized to a
1-mmol/L increase in triglyceride. To determine the
statistical significance of the association and to calculate confidence intervals (CIs), the variance of b was
next computed. The b value for each study was then
weighted by the inverse of the variance, reflecting
both the sample size and follow-up period of the
study. Finally, the weighted b values were averaged
and converted to the summary RR value, so that the
larger the study and the longer the follow-up period,
the greater the contribution to the summary RR. These
procedures resulted in the univariate summary RR for
the association between triglyceride and cardiovascular disease. To determine the effect of covariates on
this estimate, the same procedures were used to estimate the multivariate summary RR using the 6 studies
that included HDL cholesterol as a covariate.5,8,16,20,22,23 Of these studies, 4 included adjustment
for age, 4 for total cholesterol, 2 for LDL cholesterol,
4 for smoking, 6 for body mass index, and 5 for blood
pressure. To maximize the potential effects on the
multivariate summary RR, all of these covariate adjustments were included in the analysis.
As shown in Figure 1, univariate RR estimates for
cardiovascular disease associated with a 1-mmol/L
increase in triglyceride was 1.07–1.98 for men, with a
summary RR of 1.32 (95% CI: 1.26 –1.39).2,5,8 –13,15–24
This indicates a 32% increase in disease risk associated with triglyceride. Among the 5 prospective studies in women,5,10,14,16,18 individual RR estimates for
triglyceride was 1.69 –2.05 (Figure 1), all of which
were statistically significant (p ,0.05). The summary
univariate RR was higher for women than for men,
1.76 (95% CI: 1.50 –2.07), although the CI was somewhat wider because of the smaller sample size (approximately 10,800 women vs approximately 46,400
men). Thus, a 1-mmol/L increase in triglyceride was
8B THE AMERICAN JOURNAL OF CARDIOLOGYT
associated with a 76% increase in risk of incident
cardiovascular disease in women. Because only population-based studies were included, the summary RR
estimates derived from this analysis provide the best
available overall estimates for a relation between triglyceride levels and cardiovascular disease.
It has long been noted that triglyceride is a more
potent risk factor for cardiovascular disease in Scandinavian countries than in other countries,3 and metaanalysis allows a quantitative comparison of RR risk
values in different geographic locations. Among men,
RR values were 2.49, 1.25, and 1.34 for studies conducted in Scandinavia, in other European countries,
and in the United States, respectively (all p ,0.05).
Among studies in women, RR values were also higher
in Scandinavian countries than in the United States
(2.02 vs 1.71, respectively). Thus, in both men and
women, univariate RR values were indeed higher for
studies conducted in Scandinavian countries compared with studies conducted elsewhere.
As expected, multivariate RR estimates with adjustment for HDL cholesterol were attenuated (Figure
2). For men, the multivariate RRs for triglyceride were
0.98 –1.39, with a summary RR of 1.14 (95% CI:
1.05–1.28), which was statistically significant
(p ,0.05). For women, the summary multivariate RR
was higher than that for men—1.37 (95% CI: 1.13–
1.66). Thus, even after adjustment for HDL cholesterol, a statistically significant increase in the risk of
incident cardiovascular disease was associated with
triglyceride for both men and women. Importantly,
studies conducted in Scandinavia in which univariate
RR values are highest are not included in these multivariate RRs since HDL cholesterol adjustments were
not reported. Thus, the multivariate RR values reported here for triglyceride are likely to be conservative. Furthermore, when an external adjustment procedure was used to evaluate the confounding effect of
HDL cholesterol among those not reporting adjustment for this variable,7 the multivariate summary RR
was 1.22 (95% CI: 1.15–1.29) for men and 1.44 (95%
CI: 1.23–1.69) for women.
In summary, based on all of the available data from
population-based prospective studies, meta-analysis
has shown that increases in plasma triglyceride are
associated with a significant increase in risk of incident cardiovascular disease among both men and
women. In men, an increase of 1 mmol/L was associated with a 32% increase in risk of disease. A greater
increase in risk (76%) was found in women. Based on
data from studies that reported adjustments for HDL
cholesterol and other risk factors, multivariate RR
estimates were attenuated but were still statistically
significant, representing a 14% increase in risk for
men and a 37% increase in risk for women. Importantly, because only population-based studies were
included in the analysis, these results provide the best
available estimates of triglyceride risk in the general
population. Therefore, triglyceride is a risk factor for
cardiovascular disease, independent of HDL cholesterol.
VOL. 81 (4A)
FEBRUARY 26, 1998
FIGURE 1. Univariate relative risk (RR) estimates and 95% confidence intervals (CIs) for the
association between incident cardiovascular disease and a 1-mmol/L increase in triglyceride,
by gender. RR values are given on the x axis on a natural logarithm scale. The y axis lists
each study included in the meta-analysis, ordered by sample size, and the summary RR. An
RR of 1.0 (vertical dotted line) represents no association, and CIs that do not cover 1.0 indicate RRs that are statistically significant at the p 5 0.05 level. CHS 5 Cardiovascular Health
Center2; LRC 5 Lipid Research Clinics Follow-up Study5; WCGS 5 Western Collaborative
Health Study8,9; SPS 5 Stockholm Prospective Study10; GM 5 Men Born in 1913 Study11;
S-SLIC 5 Suomi-Salama Life Insurance Cohort12; PPS 5 Paris Prospective Study13;
GW 5 Study of Women in Göteburg14; UPPS 5 Uppsala Primary Preventive Study15;
FHS 5 Framingham Heart Study16; NKP 5 North Karelia Project17; CES 5 Cardiovascular
Epidemiology Study18; NAS 5 Normative Aging Study19; CSCHDS 5 Caerphilly and Speedwell Collaborative Heart Disease Studies20; RS(IV) 5 Reykjavik Study, Stage IV21;
PROCAM 5 Prospective Cardiovascular Munster Study22; and ROG 5 Rome Occupational
Groups.23,24 (Adapted from J Cardiovasc Risk.6)
PROSPECTIVE STUDIES OF
TRIGLYCERIDE AND LOW-DENSITY
LIPOPROTEIN PARTICLE SIZE
Since the completion of the meta-analysis described above, 3 new studies—the Physicians’ Health
Study,25 the Stanford Five-City Project,26 and the
Quebec Cardiovascular Study27— have examined the
association of baseline plasma triglyceride levels and
LDL particle size with incident coronary artery disease (Table I). Previous studies showed that triglycA SYMPOSIUM: STATINS AND HYPERTRIGLYCERIDEMIA
9B
FIGURE 2. Multivariate-adjusted relative risk (RR) estimates and 95% confidence intervals
(CIs) for the association between incident cardiovascular disease and a 1-mmol/L increase in
triglyceride, by gender, for those studies that adjusted for HDL cholesterol. RR values are
given on the x axis on a natural logarithm scale. The y axis lists each study included in the
meta-analysis, ordered by sample size, and the summary RR. An RR of 1.0 (vertical dotted
line) represents no association, and CIs that do not cover 1.0 indicate RRs that are statistically significant at the p 5 0.05 level. See Figure 1 for key to study abbreviations. (Adapted
from J Cardiovasc Risk.6) Note: In a recent report from the PROCAM study in which the follow-up period was extended to 8 years, the multivariate RR reached statistical significance.29
eride and LDL particle size, as determined by gradient
gel electrophoresis, are closely correlated risk factors
that are characteristic components of the “atherogenic
lipoprotein phenotype.”28 However, until the new
studies were published, prospective epidemiologic evidence examining LDL particle size as a predictor of
disease was lacking. The findings from the Physicians’ Health Study25 were based on a 7-year follow-up of 266 men with incident nonfatal myocardial
infarction or fatal coronary artery disease. Controls
were matched for age, smoking, and time of randomization. Importantly, study participants were “not specifically instructed to provide fasting specimens.” In
this study, mean baseline triglyceride values were
10B THE AMERICAN JOURNAL OF CARDIOLOGYT
higher in patients compared with controls (2.29
mmol/L vs 1.75 mmol/L, respectively; p 5 0.001).
Similarly, LDL diameter values were smaller in patients compared with controls (256 Å vs 259 Å, respectively; p ,0.001). As shown in Table I, the RR
for triglyceride was 1.43 (95% CI: 1.22–1.68) for a
1.13-mmol/L increase in triglyceride (p ,0.01), and
the RR for LDL particle size was 1.38 (95% CI:
1.18 –1.62) for an 8-Å decrease in size (p ,0.01).
Thus, both triglyceride and LDL particle size were
significant predictors of future coronary artery disease
in this study. However, the triglyceride association
was independent of other lipids (RR 5 1.33;
p 5 0.009), whereas the association with LDL particle
VOL. 81 (4A)
FEBRUARY 26, 1998
TABLE I Prospective Studies of Triglyceride and LDL Size as Risk Factors for CHD
Relative Risk
Study
Physicians’ Health Study25
Univariate
Adjusted
Quebec Cardiovascular Study27
Univariate
Adjusted
Triglyceride
LDL Particle Size
1.4*
1.3*
(100-mg/dL increase)
1.4*
1.1
(8-Å decrease)
—
1.3
(1-SD increase)
3.6*
3.0†
(Lowest vs highest tertile)
Mean Case-Control Difference
Stanford Five-City Study26
Univariate
Adjusted
—
—
25.1 Å*
22.8 Å*
*p ,0.01; †p ,0.05.
CHD 5 coronary heart disease; LDL 5 low-density lipoprotein; SD 5 standard deviation.
size was not (RR 5 1.09; p 5 0.46). This was due at
least in part to the strong inverse relation between
triglyceride and LDL particle size, with an estimated
Spearman correlation coefficient of 20.71 in this
study.
The Stanford Five-City Study26 was based on
incident cases of coronary artery disease, both fatal
and nonfatal, identified through community surveillance from 1979 –1992. Controls were selected
based on matching criteria for age, sex, ethnicity,
treatment or control city, and time of surveillance.
A total of 124 case-control pairs (90 male and 34
female pairs) were identified, and the analysis was
based on the case-control difference in LDL particle
size (Table I). For all subjects, a highly significant
case-control mean difference was found (25.1 Å;
p ,0.01), demonstrating that smaller LDL particle
size predicted incident coronary artery disease in
this population. The difference in LDL particle size
was independent of triglyceride, HDL cholesterol,
smoking, systolic blood pressure, and body mass
index in multivariate analyses, in contrast to the
finding from the Physicians’ Health Study25 described above.
Finally, the results of the Quebec Cardiovascular
Study27 were based on findings in 2,103 men, 114 of
whom developed ischemic heart disease during the
5-year follow-up period. Using 103 controls matched
for age, smoking, body mass index, and alcohol intake, the RR for the lowest tertile of LDL particle size
was 3.6 (p ,0.01; Table I), again using a nested
case-control approach. Furthermore, this relation remained statistically significant after adjustment for
triglyceride (RR 5 3.0; p ,0.05). However, in this
study triglyceride was not a significant predictor after
adjustment for LDL particle size.
Taken together, the data from these 3 studies unequivocally demonstrate the importance of both triglyceride and small, dense LDL particles as predictors
of coronary artery disease in middle-aged Caucasian
men and women.
SUMMARY AND CONCLUSIONS
Meta-analysis of data from population-based prospective studies has demonstrated that increased
plasma triglyceride is associated with a 32% increase
in risk of cardiovascular disease in men and a 76%
increase in risk of cardiovascular disease in women.
After adjustment for HDL cholesterol and other risk
factors, these risks were decreased to a 14% increase
in men and a 37% increase in women but remained
statistically significant. These results, based on more
than 46,400 men and 10,800 women, show that
plasma triglyceride is an independent risk factor for
cardiovascular disease. More recent prospective epidemiologic studies have also shown that plasma triglyceride and LDL particle size, 2 highly interrelated
risk factors, predicted subsequent coronary artery disease in 3 different Caucasian populations.
These results demonstrate the growing recognition
of the importance of triglyceride as a risk factor for
cardiovascular disease and the urgent need for clinical
trials to determine whether lowering plasma triglyceride levels will decrease disease risk.
1. Albrink MJ, Man EB. Serum triglycerides in coronary artery disease. Arch
Intern Med 1959;103:4 – 8.
2. Brown DF, Kinch SH, Doyle JT. Serum triglycerides in health and in ischemic
heart disease. N Engl J Med 1965;273:947–952.
3. Austin MA. Plasma triglyceride and coronary heart disease. Arterioscler
Thromb 1991;11:2–14.
4. NIH Consensus Development Panel on Triglyceride, High Density Lipoprotein, and Coronary Heart Disease. JAMA 1993;269:505–510.
5. Criqui MH, Heiss G, Cohn R, Cowan LD, Suchindran CM, Bangdiwala S,
Kritchevsky S, Jacobs DR Jr, O’Grady HK, Davis CE. Plasma triglyceride level
and mortality from coronary heart disease. N Engl J Med 1993;328:1220 –1225.
6. Hokanson JE, Austin MA. Plasma triglyceride is a risk factor for cardiovascular disease independent of high-density lipoprotein cholesterol: a meta-analysis
of population-based prospective studies. J Cardiovasc Risk 1996;3:213–219.
7. Greenland S. Quantitative methods in the review of epidemiol literature
Epidemiol Rev 1987;9:1–30.
8. Hulley SB, Rosenman RH, Bawol RD, Brand RJ. Epidemiology as a guide to
clinical decisions: the association between triglyceride and coronary heart disease. N Engl J Med 1980;302:1383–1389.
9. Rosenman RH, Brand RJ, Jenkins CD, Friedman M, Straus R, Wurm M.
Coronary heart disease in the Western Collaborative Group Study: final follow-up
experience of 81⁄2 years. JAMA 1975;233:872– 877.
10. Bottinger LE, Carlson LA. Risk factors for death for males and females: a
A SYMPOSIUM: STATINS AND HYPERTRIGLYCERIDEMIA
11B
study of death pattern in the Stockholm Prospective Study. Acta Med Scand
1982;211:437– 442.
11. Wilhelmsen L, Wedel H, Tibblin G. Multivariate analysis of risk factors for
coronary heart disease. Circulation 1973;48:950 –958.
12. Pelkonen R, Nikkilä EA, Koskinen S, Penttinen K, Sarna S. Association of
serum lipids and obesity with cardiovascular mortality. BMJ 1977;2:1185–1187.
13. Cambien F, Jacqueson A, Richard JL, Warnet JM, Ducimetiere P, Claude JR.
Is the level of serum triglyceride a significant predictor of coronary death in
“normocholesterolemic” subjects? The Paris Prospective Study. Am J Epidemiol
1986;124:624 – 632.
14. Bengtsson C, Bjorkelund C, Lapidus L, Lissner L. Association of serum lipid
concentrations and obesity with mortality in women: 20-year follow-up of participants in prospective population study in Gothenburg, Sweden. BMJ 1993;307:
1385–1388.
15. Åberg H, Lithell H, Selinus I, Hedstrand H. Serum triglycerides are a risk
factor for myocardial infarction but not for angina pectoris: results from a 10-year
follow-up of Uppsala Primary Preventive Study. Atherosclerosis 1985;54:89 –97.
16. Wilson P, Larson G, Castelli W. Triglycerides, HDL-cholesterol, and coronary artery disease: a Framingham update on the interactions. Can J Cardiol
1994;10(suppl B):5–9.
17. Salonen JT, Puska P. Relation of serum cholesterol and triglycerides to the
risk of acute myocardial infarction, cerebral stroke and death in eastern Finnish
male population. Int J Epidemiol 1983;12:26 –31.
18. Heyden S, Heiss G, Hames CG, Bartel AG. Fasting triglycerides as predictors
of total and CHD mortality in Evans County, Georgia. J Chron Dis 1980;33:
275–282.
19. Glynn RJ, Rosner B, Silbert JE. Changes in cholesterol and triglyceride as
predictors of ischemic heart disease. Circulation 1982;66:724 –731.
20. Bainton D, Miller NE, Bolton CH, Yarnell JWG, Sweetnam PM, Baker IA,
12B THE AMERICAN JOURNAL OF CARDIOLOGYT
Lewis B, Elwood PC. Plasma triglyceride and high density lipoprotein cholesterol
as predictors of ischaemic heart disease in British men: the Caerphilly and
Speedwell Collaborative Heart Disease Studies. Br Heart J 1992;68:60 – 66.
21. Sigurdsson G, Baldursdottir, Sigvaldason H, Agnarsson U, Thorgeirsson G,
Sigfusson N. Predictive value of apolipoproteins in a prospective survey of
coronary artery disease in men. Am J Cardiol 1992;69:1251–1254.
22. Assmann G, Schulte H. Relation of high-density lipoprotein cholesterol and
triglyceride to incidence of atherosclerotic coronary artery disease (the PROCAM
experience). Am J Cardiol 1992;70:733–737.
23. Menotti A, Spagnolo A, Scanga M, Dima F. Multivariate prediction of
coronary disease deaths in a 10-year follow-up of an Italian occupational male
cohort. Acta Cardiologica 1992;47:311–320.
24. Menotti A, Scanga M, Morisi G. Serum triglycerides in the prediction of
coronary artery disease: an Italian experience. Am J Cardiol 1994;73:29 –32.
25. Stampfer MJ, Krauss RM, Ma J, Blanche PJ, Holl LG, Sacks FM, Hennekens
CH. A prospective study of triglyceride level, low-density lipoprotein particle
diameter, and risk of myocardial infarction. JAMA 1996;276:882– 888.
26. Gardner CD, Fortmann SP, Krauss RM. Association of small low-density
lipoprotein particles with the incidence of coronary artery disease in men and
women. JAMA 1996;276:875– 881.
27. Lamarche B, Tchernof A, Moorjani S, Catin B, Dagenais GR, Lupien PJ,
Després J-P. Small, dense low-density lipoprotein particles as a predictor of the
risk of ischemic heart disease in men. Circulation 1997;95:69 –75.
28. Austin MA, King M-C, Vranizan KM, Krauss RM. Atherogenic lipoprotein
phenotype: a proposed genetic marker for coronary heart disease risk. Circulation
1990;82:495–506.
29. Assmann G, Schulte H, von Eckardstein A. Hypertriglyceridemia and elevated lipoprotein(a) are risk factors for major coronary events in middle-aged
men. Am J Cardiol 1996;77:1179 –1184.
VOL. 81 (4A)
FEBRUARY 26, 1998