University of Southern California Dissertations and Theses

Antioxidants and risk of myocardial infarction and cancer in a cohort of middle-aged finnish men: the kuopio ischemic heart disease risk factor study

(USC Thesis Other)

Antioxidants and risk of myocardial infarction and cancer in a cohort of middle-aged finnish men: the kuopio ischemic heart disease risk factor study

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600

ANTIOXIDANTS AND RISK OF MYOCARDIAL INFARCTION AND CANCER
IN A COHORT OF MIDDLE-AGED FINNISH MEN
The Kuopio Ischemic Heart Disease Risk Factor Study
by
Li Li
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
in Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Preventive Medicine - Cardiovascular Epidemiology)
May 1996
Copyright 1996 Li Li
UMI Number: 9636351
UMI Microform 9636351
Copyright 1996, by UMI Company. AH rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90007
This dissertation, written by
LI. LI ..................................... .
under the direction of Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
DISSERTATION COMMITTEE
Chairperson
Li Li James H. Dwyer
ABSTRACT
Antioxidants and Risks of Myocardial Infarction and Cancer in a Cohort of Middle-aged Finnish
Men
Free radical mediated damage has been implicated in many degenerative diseases,
including atherosclerosis and cancer. Antioxidants, being capable of neutralizing free radicals,
are hypothesized to play an important role in the prevention of coronary heart disease (CHD)
and cancer. However, results from previously reported epidemiologic studies have been largely
inconclusive.
The focus of the present study is to investigate the associations between dietary and
plasma antioxidants (vitamin E, vitamin C, carotenoid and selenium) and risks of acute
myocardial infarction (Ml) and cancer in a cohort of middle-aged Finnish men, participating the
Kuopio Ischemic Heart Disease Risk Factor study. In multivariate regression analysis, no
association was found between dietary intakes of vitamin E (RR = 1.41, 95% Cl =0.75-2.64),
vitamin C (RR = 1.33,95% Cl =0.72-2.47) or carotenoid (RR = 0 .9 0 ,9 5 % Cl =0.49-1.65) and
risk of Ml. Interestingly, the association between plasma ascorbic acid and alpha-tocopherol
and Ml were found to be modified by alcohol drinking: no association was found between
alpha-tocopherol and Ml among non-drinkers (RR = 1.08, 95% Cl =0.84-1.40), while a
marginal inverse association was found for ascorbic acid (RR = 0.79, 95% Cl =0.59-1.07); in
contrast, a positive association was found for both alpha-tocopherol and ascorbic acid among
drinkers, RR = 1.50 (95% Cl = 0.98-2.28) and RR = 1.43 (95% Cl = 0.96-2.10) respectively.
While no association was found between serum selenium and risk of Ml, an inverse,
albeit non-significant, association was found between serum selenium and cancer. Cigarette
smoking was found to be an effect modifier of the selenium-cancer association: the protective
effect of selenium on cancer was found to be confined to smokers (RR = 0.32, 95% Cl =0.12-
0.89), with little effect among non-smokers (RR = 1.10, 95% Cl =0.43-2.85).
Results from our current analysis suggest that the effects of antioxidants on risk of
coronary heart disease and cancer may be largely dependent on individual's risk profile, such
as smoking and drinking behavior. The findings from our analysis warrants further well-
controlled study targeting at smokers and alcohol drinkers.
To my parents and wife
ii
Acknowledgements
I am grateful to my people who have helped me to write this thesis. I especially
wish to thank my mentor Dr. James H. Dwyer, for leading me to this field of research,
and for teaching me the art of statistics and epidemiology. I am indebted to Dr. Jukka
T. Salonen, director of the Kuopio Ischemic Heart Disease Risk Factor Study. Thank
you to all colleagues in Kuopio, Finland for providing me with much need
encouragement.
I want to also acknowledge Dr. Chih-Ping Zhou, Dr. Alex Sevanian and Dr. Dan
Sharp for being instrumental in my Quals, and Dr. Andy Johnson, Dr. Steve Stohlman
and Dr. Mark Menski for serving so generously in my Dissertation Committee. I also
appreciate many suggestions from my classmates and friends Lisa Nicholson, Ping
Sun, Wei Sun and Raymond Palmer at the Institute for Health Promotion and Disease
Prevention Research of the University of Southern California.
My wife Qin Yao owns my special thanks for her patience, love, and support
in matters of daily living while I sat at the computer. Most of all, I thank my family,
who accepted and supported me choosing a career path away from home.
Table of Contents
Page
Dedication ii
Acknowledgements iii
List of Tables v-vi
List of Figures vii
Chapter 1. Antioxidant Vitamins and Coronary Heart Disease:
A Literature Review.
Chapter 2. Antioxidant Vitamin Intakes and Risk of Myocardial
Infarction: The Kuopio Ischemic Heart Disease Risk
Factor Study.
Chapter 3. Association of Plasma Alpha-Tocopherol and Ascorbic
Acid with Myocardial Infarction: Multivariate
Correction for Bias due to Measurement Errors.
Chapter 4. Serum Selenium and Risk of Cancer in a Cohort of
Middle-aged Finnish Men: The Kuopio Ischemic Heart
Disease Risk Factor Study.
42
65
90
iv
List of Tables
Chapter 1.
Table 1.
Table 2.
Chapter 2.
Table 1.
Table 2.
Table 3.
Table 4.
Table 5.
Chapter 3.
Table 1.
Page
Review of Studies of Antioxidant Vitamins 36-37
and Atherosclerosis in Animal Models
Epidemiological Studies Correlating Antioxidants 38
and Prooxidants to Ischemic Heart Disease
Descriptive Characteristics of Myocardial Infarction 60
Cases and Non-cases: The Kuopio Ischemic Heart
Disease Risk Factor Study
Relative Risk Estimates of Myocardial Infarction 61
according to Tertile Groups of Dietary Vitamin E
Intake: The Kuopio Ischemic Heart Disease Risk
Factor Study (n = 1,609)
Relative Risk Estimates of Myocardial Infarction 62
according to Tertile Groups of Dietary Vitamin C
Intake: The Kuopio Ischemic Heart Disease Risk
Factor Study (n = 1,609)
Relative Risk Estimates of Myocardial Infarction 63
according to Tertile Groups of Dietary Carotenoid
Intake: The Kuopio Ischemic Heart Disease Risk
Factor Study (n = 1,609)
Relative Risk Estimates of Myocardial Infarction 64
according to Tertile Groups of Dietary Vitamin E,
Vitamin C and Carotenoid Intake: The Kuopio
Ischemic Heart Disease Risk Factor Study (n = 1,609)
Baseline Characteristics of Incident Myocardial 84
Infarction Cases and Non-cases: The Kuopio
Ischemic Heart Disease Risk Factor Study
v
Page
Table 2.
Table 3.
Table 4.
Table 5.
Table 6.
Chapter 4.
Table 1.
Table 2.
Table 3.
Table 4.
Table 5.
Age-adjusted Correlations among Repeated
Measurements of Risk Factors in the Validation
Subsample (n = 91): The Kuopio Ischemic Heart
Disease Risk Factor Study
Relative Risk (RR) Estimates of Myocardial
farction for Plasma Alpha-Tocopherol and
Ascorbic Acid: The Kuopio Ischemic Heart
Disease Risk Factor Study
Relative Risk (RR) Estimates of Myocardial
Infarction for Plasma Alpha-Tocopherol and
Ascorbic Acid according to Alcohol Consumption:
The Kuopio Ischemic Heart Disease Risk Factor
Study
Estimation of Validity, Error Variance and
Inter-individual Variance from a Measurement Model
Relative Risk Estimates from Proportional Hazard
Model: without and with Correction for Measurement
Errors
Baseline Characteristics of Analytic Sample:
The Kuopio Ischemic Heart Disease Risk Factor Study
Baseline Serum Selenium Level and Incidence
of Cancer by Exam Year and Smoking Status:
The Kuopio Ischemic Heart Disease Risk Factor Study
Estimated Relative Risk (RR) of Cancer for Serum
Selenium: The Kuopio Ischemic Heart Disease Risk
Factor Study
Multivariate Relative Risk (RR) Estimates of Cancer for
Serum Selenium, according to Exam Year: The Kuopio
Ischemic Heart Disease Risk Factor Study
Summary of Prospective Studies of Selenium Levels
in relation to Risk of Cancer, all Sites Combined
85
86
87
88
89
109
110
111
112
113
vi
List of Figures
Page
Chapter 1.
Figure 1.
Figure 2.
Figure 3.
Chapter 4.
Figure 1.
Angina Pectoris and Plasma Vitamins A, C, E
and Carotene.
Risk of Coronary Heart Disease and Vitamin E
Supplement Use (lU/day).
Dietary Vitamin E and Risk of CHD: Finnish
Men and Women (N = 5,133).
Estimated Odds Ratio (OR) of Cancer for
Tertiles of Serum Selenium by Smoking Status.
39
40
41
114
vii
Chapter 1. Antioxidant Vitamins and Coronary Heart Disease: A Literature Review
1. Introduction
A recent report of a large scale intervention trial of antioxidant vitamins in
Finland has raised great concern about the proposed beneficial effect of antioxidants,
such as vitamin E and /?-carotene, in the prevention of cancer and cardiovascular
diseases84. For years, it has been suggested that oxidative damage caused by free
radicals is involved in the pathogenesis of many degenerative diseases, including
cancer and atherosclerosis1'4. Recent studies of lipoprotein oxidation has contributed
to the formation of the hypothesis that lipoprotein oxidation may play a pivotal role in
atherogenesis and antioxidant vitamins may protect against or retard the process of
atherogenesis due to their free-radical scavenging or antioxidative properties2'6.
Although there is increasing body of evidence that antioxidant vitamins may protect
against the development of cancer and cardiovascular diseases, most of the published
studies are either from animals, which may have limited relevance to humans, or
observational. Neither of these prove causality in humans. Proof of a causal effect of
antioxidants against cardiovascular disease will have to await further results from
several ongoing intervention trials. Given the paucity of evidence from randomized
trials at this stage, and the unexpected findings of the Finnish ATBC Cancer Prevention
Study, the anti-atherogenic effect of antioxidant vitamins is still controversial and
inconclusive. The present paper reviews the hypothesis of oxidative modification of
low density lipoprotein (LDL) in the pathogenesis of atherosclerosis and the role of
antioxidants in the prevention of coronary heart disease (CHD), focusing on
epidemiologic studies and potential interactions among several antioxidants and
prooxidants and the implication for future intervention trials.
7
2. Lipoprotein oxidation in atherosclerosis
Elevated LDL is an established risk factor of cardiovascular disease, but the
possible mechanisms are only now being understood, in vitro and in vivo studies
suggest the atherogenecity of LDL is greatly enhanced by oxidative damage7'10.
Oxidative modification of LDL (Ox-LDL) results in the formation of fatty acid
hydroperoxides and subsequent fragmentation of the fatty acid, which leads to the
formation of short chain aldehyde, followed by fragmentation of apo B11'12. Ox-LDL
has been shown to be cytotoxic for fibroblasts, smooth muscle cells, and endothelial
cells7,9,10, and to be chemotactic for circulating monocytes and to inhibit the migration
of resident macrophages in the intima12,13. Ox-LDL is avidly taken up by macrophages
through the 'Scavenger Receptor Pathway'13. The scavenger receptor differs from the
native LDL receptor in that the expression of the scavenger receptor is not down-
regulated by the intracellular cholesterol content, thus the process is unchecked and
leads to accumulation of cholesterol in macrophages; the lipid-laden macrophages thus
become foam cells, the first patho-histological manifestation of early atherosclerosis.
The fact that antioxidants, such as vitamin E, C and yff-carotene, protect against
such damage has raised the possibility of their role in inhibiting atherogenesis and
preventing various manifestations of coronary heart diseases.
3. Antioxidative property o f vitamins
An antioxidant, in a broad sense, is any substance that, when present at low
concentration compared to that of an oxidation substrate, significantly delays or
inhibits oxidation of the substrate14. Free radical mediated oxidation of LDL can be
generally classified into initiation, propagation and degradation. Consequently,
2
antioxidative defense may be classified into prevention, interception and repair.
Interception is the domain of chain-breaking antioxidants, such as tocopherol, reacting
with peroxyl radicals, and often the term 'antioxidant' is used in a restricted sense to
denote just this activity14,15.
(1) Vitamin E
Vitamin E, a generic term for tocopherol and tocotrienols, is the major fat-
soluble chain-breaking antioxidant in LDL. Reactions with organic peroxyl radicals
account for its antioxidant activity and is believed to be the basis for the antioxidative
function of vitamin E in vivo, protecting LDL from free radical attack14,15. The reaction
of vitamin E with lipid peroxides produces lipid hydroperoxides and the relatively stable
tocopheroxyl radical, which does not propagate radical chains, thus breaking the free-
radical initiated chain propagation of LDL oxidation14.
Several in vitro studies have examined the effect of a-tocopherol on LDL
oxidation using either malondialdehyde (MDA) or thiobarbituric acid reactive
substances (TBARS) or conjugated diene as the index of LDL oxidation16'24. Esterbauer
et al. have shown that vitamin E depletion preceded the overt oxidation of LDL lipids16.
Jialal and Grundy et al. reported that the production of TBARS was partially inhibited
when LDL added with 40 /vMm a-tocopherol is subject to 24 hour oxidation with 2.5
IjM copper17. A similar inhibitory effect of a-tocopherol was observed when conjugated
diene was used as an index of LDL oxidation. Furthermore, this study also showed that
a-tocopherol decreased the uptake of oxidized LDL by macrophages by 45 percent. In
a series of experiments using the lag phase as an index of LDL oxidation19'22,
Esterbauer et al. have shown that supplementation of plasma with a-tocopherol prior
3
to isolation of LDL resulted in a fourfold enrichment of LDL with a-tocopherol.
Consequently, this enrichment of LDL resulted in a proportional increase in the lag
phase of oxidation and hence increased the oxidation resistance of the LDL. Gabriele
Bittolo-Bon et al. reported that in a group of volunteers treated with 900 mg a-
tocopherol per day for a week, the LDL vitamin E content doubled; no oxidation by
copper was observed as judged by time-dependent changes in LDL electrophoretic
mobility, MDA production was dramatically decreased from a mean of 1300 //mol/l
before supplementation to 25 /t/mol/l after15. Also, two studies reported by Dieber-
Rotheneder et al.22, and Jialal et al.17, respectively, showed that LDLs from subjects
fed with a-tocopherol were more resistant to oxidative modification as compared to
controls.
Although vitamin E has been shown to inhibit LDL oxidation when added to LDL
in vitro or when given as a supplement before isolation of LDL, the endogenous vitamin
E content of native LDL has been consistently shown not to correlate with the extent
of LDL oxidation19'23,24. In the report by Esterbauer, although the lag phase of LDL
was linearly correlated with the dose of vitamin E supplementation, it was not
correlated with the vitamin E content of LDL before supplementation19. In the study
by Jialal, the addition of 40 //M vitamin E to LDL only partially inhibited the oxidation
of LDL and addition of 40 /vM ascorbic acid has shown a greater inhibitory effect on
LDL oxidation17. These seemingly conflicting results may imply: (1) endogenous
vitamin E content of LDL may be insufficient to protect against LDL oxidation, and the
full antioxidative property of vitamin E may only exert at a very high dose as generally
used in experimental studies, or there may be a critical threshold of vitamin E level that
may only be achieved by high dose supplementation; (2) other antioxidants such as
4
ascorbic acid are probably of importance in the prevention of LDL oxidation; (3) a
synergistic effect may exist between vitamin E and vitamin C and other antioxidants.
(2) Vitamin C
Vitamin C (ascorbic acid) is the major aqueous antioxidant in the biological
system and is considered the most important antioxidant in extracellular
compartments25. Ascorbic acid has also shown a sparing effect of a-tocopherol27. The
oxidized a-tocopherol can be reduced to a-tocopherol by ascorbic acid. Ascorbate has
been shown to efficiently scavenge superoxide, hydrogen peroxide, hydroxyl radicals,
peroxyl radicals, and singlet oxygen25'28. Frei et al. have shown that in human plasma
lipids, ascorbate was far more effective in inhibiting lipid peroxidation initiated by a
peroxyl radical initiator than other plasma components, such as protein thiols, urate,
bilirubin, as well as a-tocopherol25. Frei et al. also showed that in plasma exposed to
the gas phase of cigarette smoke, endogenous ascorbic acid is the first antioxidant to
be totally consumed and the only one capable of preventing initiation of detectable lipid
peroxidation. Ubiquinol-10, protein thiols, and bilirubin are also oxidized at significant
rates, but oxidations of uric acid and a-tocopherol are very slow29. In the study by
Jialal, addition of 40//M ascorbic acid to plasma decreased the TBARS production by
about 90 percent, while addition of vitamin E only partially inhibited the oxidation of
LDL by about 30 percent. In plasma challenged by activated neutrophils, the order of
antioxidant consumption is ascorbic acid = ubiquinol-10 = protein thiols > uric acid,
without significant consumption of bilirubin and a-tocopherol. Detectable lipid
peroxidation is initiated immediately after completion of ascorbic acid consumption,
despite the presence of high concentrations of all the remaining plasma antioxidants,
5
including a-tocopherol. Frei et al. also showed that a-tocopherol is relatively ineffective
in plasma oxidation in the presence of aqueous peroxyl radicals26. These findings
indicate that a-tocopherol is relatively inefficient in scavenging aqueous peroxyl
radicals and neutrophil-derived oxidants. The antioxidative activity of vitamin E is thus
hypothesized to be restricted within LDL molecules, probably at the interface of LDL
and aqueous phase. Alpha-tocopherol thus may only act as a chain breaking
antioxidant, which breaks the propagation of lipid oxidation, but may not prevent the
initiation of free radical reactions in the aqueous phase. Ascorbic acid, on the other
hand, may work in the aqueous phase of physiologic system and prevent the initiation
of free radical reaction, and may also spare or recycle the oxidized a-tocopherol. In this
sense, a synergistic inhibitory effect between vitamin C and E, and other aqueous
antioxidants may exist. In epidemiologic studies, a statistical interaction between
vitamins E and C may be expected.
Results from these studies implicate that the integrity of plasma lipid and
membrane may only be maintained by the multi-level antioxidant system in the body,
which may consist of several antioxidants and the synergistic actions among them.
(3) 0-carotene
p-carotene and related carotenoids share many similarities with vitamin E,
including natural occurrence in plant foods, lipid solubility, deposition in fatty portions
of cells, such as membranes, and antioxidant properties15. The capacity of yff-carotene
to quench singlet oxygen is believed to be the basis of the antioxidant property of 0-
carotene. By quenching singlet oxygen, yff-carotene can thus reduce the concentration
of chain-carrying peroxyl radicals at low partial pressure of oxygen, a property that
6
complements the chain-breaking action of vitamin E at high oxygen concentration30,
yff-carotene has been reported to be inversely associated with lung and breast cancer
in humans, and it has also been found to have chemopreventive properties in a number
of animal tumor models31'33. However, results from several in vitro studies of the
effect of/ff-carotene on LDL oxidation have been inconsistent. Morel et al.34, and Van
Hinsberg et al35, failed to show that ^-carotene inhibits LDL oxidation. As pointed out
by some authors17, an important point that must be considered with in vitro testing
with ^-carotene is that it has varying solubility in different solvents. Although /?-
carotene, as a fat soluble antioxidant, is readily soluble in hexane and chloroform, it
is only sparingly soluble in ethanol. In some of the studies, ethanol was used as the
solvent for /?-carotene34. It is possible that the ^-carotene used in these studies was
not adequately prepared. This may account for the failure to show an inhibitory effect
of /ff-carotene on LDL oxidation in these studies. A study by Jialal et al., in which
hexane and ethanol were used to dissolve ^-carotene, reported that/?-carotene at the
concentration of 5.0 j j M clearly inhibits LDL oxidation18. Compared to LDL without
addition of ^-carotene, the absorbance of chromatography fractions of LDL at A = 4 5 0
nm decreased by 75 percent in the presence of /?-carotene. Also, /?-carotene inhibited
macrophage modification of LDL substantially. Navab et al. have shown that
preincubation of co-cultures of aortic endothelial cells and aortic smooth muscle cells
with /ff-carotene and a-tocopherol prevented LDL modification and its induction of
monocyte transmigration36. Evidence from these last two studies clearly indicates an
inhibitory effect of /?-carotene on LDL oxidation.
Although ^-carotene and a-tocopherol may synergistically inhibit LDL oxidation
due to their similarity in antioxidant properties, administration of high doses of /?-
7
carotene has been shown to decrease the plasma concentration of a-tocopherol in
animals and humans, and vice versa37'41. In one clinical trial, levels of a-tocopherol
were reduced by 40 percent when /ff-carotene was administered daily to normal
subjects for eight months37. Evidence also indicates that excessive vitamin E decreases
the levels of ^-carotene in the serum of humans and the plasma and liver of rats39,40.
In a study reported by Willett et al38 , 30 mg /ff-carotene and 728 mg vitamin E were
administered daily, with a ratio of /?-carotene to vitamin E of 1:24; the resulting serum
level of /ff-carotene was significantly reduced. In the Finnish ATBC Cancer Prevention
Study, the combined effect of ^-carotene (20 mg/day) and vitamin E (50 mg/day), /?-
carotene alone, or vitamin E alone, on the development of lung cancer in 29,133
subjects was examined. Results of the pilot study suggest a nonsignificant reduction
of serum ^-carotene in the group receiving the combination capsule41. Although /?-
carotene and vitamin E are nutrient antioxidants that have been shown to have mutual
and complementary effects in biological systems, excessive amounts of ^-carotene can
lead to reduced vitamin E levels and vice versa, especially when administered on a
long-term basis. Increasing tissue /?-carotene at the expense of vitamin E or increasing
vitamin E at the expense of ^-carotene may compromise the overall protection afforded
to the biological systems by dietary antioxidants.
4. Dietary antioxidants and coronary heart disease
(1) Animal studies
Over one dozen animal studies have been conducted to test the hypothesis that
antioxidants delay atherosclerosis. Table 1 summarizes results from selected studies.
Some of them are briefly reviewed here. Three trials have reported decreased
8
atheromatous lesion formation in vitamin E fed animals compared to controls.
Restricted ovulatory hens develop hyperlipidemia and subsequent aortic intimal
thickening, a demonstration of early atherosclerosis42'44. Smith et al reported that
compared to control hens, those fed 1000 mg of vitamin E per kilogram of feed had
reduced levels of plasma peroxides and less aortic intimal thickening42. Wojcicki et al
reported a statistically significant 25 percent reduction in aortic atherosclerotic lesions
in hypercholesterolemic mongel rabbits fed 10 milligrams per kilogram per day of
vitamin E compared to controls43. Verlangieri and colleague reported a 54 percent
lesion reduction in monkeys fed 108 IU of vitamin E per day compared to controls44.
Vitamin C deficiency has been shown to cause atherosclerotic lesions in guinea
pigs45. Subsequent studies in deficient guinea pigs reveal regression of early
atherosclerotic lesions after administration of vitamin C46. Altman et al reported
delayed progression and, in some cases, regression of atherosclerotic lesions among
hypercholesterolemic rats supplemented with vitamin C47. Bocan et al. reported that
supplement of vitamin E and C (500 mg/kg each) for eight weeks decreased diet-
induced thoracic aortic cholesterol ester (CE) content by 40 percent and lesion
coverage by 46 percent in hypercholesterolemic New Zealand White rabbits, without
effect on plasma cholesterol48. However, no effect was observed on injury-induced
iliac-femoral atherosclerotic lesions.
Probucol is a synthetic cholesterol-lowering agent, which also has been shown
to be a powerful antioxidant. Several animal trials have tested probucol in the
development of atherosclerotic lesions. Kita et al. reported reduced LDL oxidation
among rabbits administered probucol for six months compared to those given placebo,
and the percentage of surface area of total thoracic aorta with visible plaques was
9
reduced dramatically compared to controls (7.0% vs. 54.2%)49. Carew et al controlled
for probucol's cholesterol-lowering effect by using lovastatin in one of the two rabbit
control groups50. Compared to lovastatin-treated rabbits, probucol further reduced the
rate of formation of fatty streaks in Watanabe heritable hyperlipidemic rabbits,
suggesting that the benefit of probucol may be at least in part due to its antioxidant
effects. Administration of BHT, another synthetic antioxidant, and selenium, an
antioxidant enzyme cofactor, also result in reduced atherosclerotic lesions in
hypercholesterolemic rabbits51,52.
(2) Epidemiologic studies
Table 2 lists some of the epidemiologic studies that correlate dietary or blood
antioxidants and prooxidants to ischemic heart disease. Cross-cultural surveys or
ecologic studies have provided equivocal, but suggestive, results that dietary
antioxidants are associated with reduced risk of coronary heart disease. Two studies
in the United Kingdom reported a significant inverse relationship between consumption
of fresh fruits and vegetables and rates of atherosclerotic disease53,54. In the United
States, the steady decline in cardiovascular mortality has generally been attributed to
improvement in treatments, or secondary prevention, as well as primary prevention due
to decreased cigarette smoking, better control of hypertension and cholesterol in the
population. Descriptive studies have also shown, however, that the early decline in
cardiovascular disease (CVD) was preceded by significant changes in the intake of
antioxidant-rich foods. Verlangieri has raised the hypothesis that part of the secular
decline in CVD may be due to increased fresh fruit and vegetable intake resulting from
greater year round availability of these foods55. Ginter reported an inverse relationship
10
between vitamin C production and coronary disease mortality rates in the United
States over the last twenty years56, although changes in other risk factors may have
largely confounded these results due to the lack of adequate control of confounding
in ecologic studies.
More recently, Artaud-Wild and colleagues examined the relation between
coronary mortality and forty dietary variables among forty countries, including France
and Finland. They found that CHD mortality rate was strongly correlated with
Cholesterol-Saturated Fat Index (CSI)57. However, despite comparable intakes of
dietary cholesterol and saturated fat, France and Finland stood out at the two ends of
the extreme, France with a total coronary mortality rate of 193 per 100,000 men in
1961 and 227 per 100,000 men in the middle of 1970s, and Finland 837 and 1,037
per 100,000 men respectively. This paradox can at least partly be explained by a
higher intake of milk and butterfat and a lower intake of vegetables in Finland, and a
higher intake of vegetables and wines in France, which may be attributed to the rich
antioxidant content in the French diet, including vitamins C and E, carotenoids and
phenolics in vegetables and red wine57.
Prospective studies on dietary intakes have consistently found an inverse
relationship between consumption of fruits and vegetables high in antioxidant vitamins
and subsequent risk of CHD. Examples are reports by Carcia-Palmieri et al. from the
Puerto Rico Heart program58, Kromhout et al. from the Zutphen Study59, and Kushi et
al. from the Ireland-Boston Diet-Heart Study60. However, since nutrients derived from
the same food sources are in general highly intercorrelated, the protective effect of
fruit and vegetable intake may not be attributed specifically to the antioxidant vitamins
in these foods. Although findings from these studies have a valuable public health
11
implication in the prevention of CHD, their relevance to etiology inferences are
generally limited.
Recently, several studies have specifically examined the relation of common
dietary antioxidant vitamins and subsequent risk of CHD and lend support to the
antioxidant hypothesis of coronary heart disease. Gaziano et al reported that among
1,299 elderly Massachusetts residents, the relative risk of CVD mortality was 0.55
(0.34-0.87) among those in the highest quartile of/ff-carotene consumption, compared
to those in the lowest quartile after adjustment for other established risk factors61.
In a prospective cohort of Swedish women, estimates of vitamin C intake from a 24-
hour recall dietary history were inversely correlated with CVD event rate; however,
these findings did not persist after controlling for age62. Enstrom et al. examined the
relation between vitamin C intake and total mortality and cardiovascular disease
mortality in the First National Health and Nutrition Examination Survey(NHANES I)
Epidemiologic Follow-up study cohort of a representative sample of 11,348
noninstitutionalized U.S. adults aged 25-74 years, reporting that the index of vitamin
C intake was strongly inversely associated with risk of total mortality and risk of
cardiovascular disease for males and weakly inverse for females63. Compared to the
lowest vitamin C intake group, males with the highest vitamin C intake have a
standardized mortality ratio (SMR) of 0.65 (0.52-0.80) for all causes, and 0.58 (0.41 -
0.78) for all cardiovascular diseases; females have a SMR of 0.90 (0.74-1.090 for all
causes and 0.75 (0.55-0.99) for all cardiovascular diseases. In the Zutphen Study of
693 elderly men free of myocardial infarction at baseline, dietary flavonoid intake was
significantly inversely related to subsequent five year mortality from coronary heart
disease and moderately related to the incidence of a first fatal or nonfatal myocardial
12
infarctio (P<0.08)64. Relative risks for mortality from coronary heart disease and
incidence of a first myocardial infarction were about 50 percent lower in the highest
tertile group of flavonoid intake than in the lowest tertile group. The effect of flavonoid
intake on age-adjusted mortality from coronary heart disease became even more
pronounced after adjusting for other dietary variables, including vitamin E, C and /?-
carotene.
More recently, two large prospective studies of U.S. women (the Nurse's Health
Study) and men (the Health Professionals Follow-up Study), and one relatively smaller
study of Finnish men and women (n = 5,133) have all reported an inverse relation of
vitamin E, and a similar, but weaker, relation of vitamin C and carotenoid with
subsequent risk of CHD65'67. These studies offer the most compelling evidence
presented to date that vitamin E, and possibly vitamin C and /?-carotene, can reduce
the incidence and mortality from CHD.
The Nurses' Health Study comprised of a cohort of 87,245 female nurses aged
34-59 years65. During eight years of follow-up, 552 cases of major CHD were
documented. Dietary intake of nutrients was estimated at the beginning of the study
from a self-administered questionnaire that ascertained the frequency of intake of 61
foods. Supplement use was estimated from a questionnaire administered four years
after the start of the study, which documented the use of both multivitamins and other
specific vitamins, including vitamin E. The Health Professionals Follow-up Study
consisted of 39,910 male health professionals aged 40-75 years66. During four years
of follow-up, 667 cases of CHD were documented. Dietary intake was estimated at
baseline from food-frequency questionnaire containing 131 foods. Additional questions
also assessed how frequently vitamin supplements, including vitamin E, were taken.
13
Compared to persons in the lowest quintile of total vitamin E (including supplements),
the relative risk of CHD for persons in the upper quintile was 0.60 (0.50-0.87) for
women and 0.64 (0.49-0.83) for men. The relative risk for women and men in the
upper quintile of dietary vitamin E (excluding supplements) was, however, not
significantly different from 1.0, with 0.95 (0.72-1.23) for women after adjusting for
age and smoking status, and 0.79 (0.54-1.15) for men in the fully adjusted model. The
quintile analysis of supplement intake was also presented for men, for whom the
relative risk in the highest quintile was 0.70 (0.55-0.89); however, there was little
evidence of a linear dose-response relationship. For women, the relative risk for any
vitamin E supplement intake was 0.57 (0.41-0.78) with adjustments for age and
smoking. Additional analysis in which duration of supplement use was considered
show little evidence of an increased effect with increased duration. Little association
was observed between CHD and vitamin supplement use of less than two years'
duration in either cohort. In the cohort study of men, some evidence suggested an
inverse association between CHD and intake of carotene and vitamin C, as well as
vitamin E.
In the Finnish study, 5,133 men and women aged 30-69 years and initially free
from heart disease were followed for fourteen years67. Dietary intakes were assessed
by a dietary history method covering the total habitual diet during the previous year.
An inverse association was observed between dietary vitamin E intake and subsequent
coronary mortality in both men and women, with RR of 0.68 (0.42-1.11, p for
trend = 0.01) and 0.35 (0.14-0.88, p for trend < 0.01), respectively, between the
highest and lowest tertiles of the intake. A significant inverse association of vitamin
C intake and risk of CHD was also found in women, with a RR of 0.49 (0.24-0.98) for
14
the highest vitamin C intake tertile, but not in men. No significant inverse association
was found for carotenoid intake in either women (RR = 0.62, 0.30-1.29) or men
(RR = 1.02, 0.70-1.48). In this population, the percentage of vitamin supplement use
is extremely low, with an estimate of 3 percent. This study is the first that explicitly
showed a protective effect of dietary vitamin E (without supplement) against coronary
heart disease.
An unpublished prospective study found a strong protective relationship
between incident CHD and use of vitamin E supplements in Quebec, Canada68. Among
a cohort of 2,226 men aged 45-76 vitamin use was inversely associated with ischemic
heart disease (IHD) risk, with RR of 0.22 (0.05-0.90) for IHD death, and 0.42 (0.18-
0.98) for myocardial infarction (Ml).
Although the data from the above dietary intake studies are compatible with a
possible benefit of antioxidant vitamins in the prevention of CHD, especially vitamin
E, the available observational data are sparse and not all consistent. Caution should be
taken in the causal interpretation of these results. Firstly, dietary measurements are
generally flawed with tremendous measurement errors. Although confounding may be
controlled for statistically in multivariate analysis, conventional statistical methods are
inadequate for adjustment for measurement errors. Residual confounding thus may
have distorted the true relationship between interested antioxidants and risk of disease.
Secondly, dietary intakes of nutrients are, in general, highly intercorrelated; the
observed inverse association of dietary antioxidant and risk of CHD may not be due to
the effect of the interested antioxidant per se. It may be, for example, that greater
dietary intake of the interested antioxidant is only a marker for some dietary practice
or even non-dietary life-style variables that are truly protective. It is, in fact, plausible
15
that intake of antioxidant-rich foods is indeed protective, but that the benefit results
not from their antioxidant properties, but from some other components these foods
have in common. Thirdly, the bioavailability of antioxidants vary from person to
person, which may further introduce measurement errors when dietary intakes of
antioxidants are analyzed implicitly as surrogates for the underlying biologically active
antioxidants in human body. Fourthly, although multivariate adjustment is a common
practice in the analysis of dietary data, unknown or unmeasured confounding factors
cannot be controlled for. A 'third' factor confounding the relationship thus can not be
totally dismissed.
5. Plasma and serum antioxidants and coronary heart disease
Studies examining the relationship between plasma levels of antioxidants and
risk of CHD have yielded inconsistent results (Table 2). Riemersa et al examined the
relation between risk of angina pectoris and plasma concentrations of vitamins A, C,
E and carotene in a population case-control study of 110 cases of angina assessed by
a self-administered chest pain questionnaire, and 394 controls selected from a sample
of 6,000 men aged 35-64 years69. An inverse relationship between plasma levels of
vitamin E, vitamin C, and /?-carotene and angina was found. Compared to those in the
highest quintiles, the estimated unadjusted Odds Ratio (OR) for those in the lowest
quintiles was 2.51 (1.24-5.10) for vitamin E, 2.35 (1.16-4.78) for vitamin C and 2.64
(1.32-5.29) for /?-carotene. The relationship persisted for vitamin E (OR = 2.68, 1.07-
6.70) after controlling for smoking and other cardiovascular risk factors, despite the
fact that misclassification of angina from a chest pain questionnaire may have
underestimated the true effect of each antioxidants. The inverse association between
16
angina and low plasma carotene disappeared and the effect of plasma vitamin C was
substantially reduced after adjustment for smoking. Rameriz et al. reported decreased
leukocyte ascorbic acid levels in cases of angiographically proven coronary artery
disease (CAD) compared to controls without CAD referred for cardiac
catheterization70. Kardinaal etal. reported from the European Community Multicenter
Study (EURAMIC) on antioxidants, myocardial infarction, and breast cancer that /?-
carotene concentration in adipose-tissue was inversely associated with risk of
myocardial infarction71. The age-adjusted OR in the lowest quintile of ^-carotene as
compared to the highest was 2.62 (1.79-3.83). The relation persisted with additional
adjustment for other risk factors, with an estimated OR of 1.78 (1.17-2.71). A low a-
tocopherol concentration was not associated with risk of myocardial infarction.
Interestingly, however, a-tocopherol did seem to modify the effect of /?-carotene; at
high a-tocopherol concentrations the inverse association of /?-carotene with Ml was
more pronounced as compared to the lowest and middle a-tocopherol tertile groups.
In the last two studies, instead of using plasma or serum samples, ascorbic acid in
leukocyte or /ff-carotene and a-tocopherol in adipose-tissue were used. How relevant
those measurements are to the corresponding plasma or serum levels is not clear.
However, as argued by some investigators, the leukocyte or adipose-tissue
concentration of antioxidants, especially fat-soluble antioxidants, such as vitamin E and
P-carotene, may better correlate with long-time dietary intake of these nutrients, and
may better reflect the long-term steady state of these antioxidants in the human body.
To this extent, the use of these measurements in epidemiologic studies may offer the
most etiologic meaning.
Four nested case-control studies have reported conflicting results72'75. Because
17
blood samples for antioxidant determination were obtained before disease occurrence,
these studies are termed prospective blood-based studies. Although Street et al
demonstrated an inverse association between prediagnostic serum/ff-carotene level and
subsequent risk of Ml (RR = 0.42, 0.19-0.89), two others reported no consistent
association between serum antioxidants and subsequent CVD. Kok et al. reported no
association between serum vitamins A and E and selenium levels in frozen serum
samples and CVD mortality in a nested case-control study of residents in Netherlands.
A potential limitation of this study may be the instability of vitamin antioxidants in
serum samples after nine years of storage. In a nested case-control study of Finnish
men, Salonen et al. found no consistent association between serum vitamins A, E, or
selenium and death from CAD. In this study, samples were stored for seven years at -
20°C before analysis, so vitamin instability may also have affected these results. In
another nested case-control study of the population-based MONICA Augsburg cohort,
serum vitamin E concentration and subsequent seven year follow-up of Ml was
investigated. There were no marked differences between cases and their matched
controls in the means of vitamin E concentration (33.9 //mol/l vs. 32.8
//mol/l,p = 0.37), or in the mean vitamin E/total cholesterol ratios (4.89//mol/mmol vs.
4.82, p = 0.75). The adjusted relative risk in the lowest tertile of vitamin E relative to
the upper two tertiles was 0.72 (0.33-1.57); for the lowest tertile of the ratio, the RR
was 0.81 (0.42-1.56). Although the instability of vitamin E during long time storage
may to some extent affect the results, the lack of an association of vitamin E with risk
of Ml was probably more due to the high average levels of vitamin E in the study
population.
Three cross-cultural studies by Gey and colleagues reported strong inverse
18
associations between plasma vitamin E levels standardized to plasma cholesterol levels
and vitamin C levels and rates of ischemic heart disease (IHD) mortality in 16 European
populations76'78. The findings for vitamin A and selenium were less consistent.
Interestingly, out of these 16 populations, in a subgroup of 12 populations with the
total plasma cholesterol ranging in a narrow band of usual levels (220-240 mg/dl), total
cholesterol lacked any statistically significant direct correlation with IHD mortality
(r2 = 0.04) as was expected, but this was also the case for blood pressure (diastolic
r2 = 0.08, systolic r2 = 0.01), and for the combination of total cholesterol and diastolic
blood pressure in multivariate analysis (r2 = 0.10, p = 0.62). On the other hand, among
the essential antioxidants examined, vitamin E showed a very strong inverse correlation
with IHD. In univariate analysis, the correlation of both absolute level of vitamin E and
the vitamin E-cholesterol ratio with IHD were in the leading rank (r2 = 0.63, and
r2 = 0.73, respectively) followed by the moderately strong correlation for vitamin C
(r2 = 0.41) and marginal correlations for vitamin A (r2 = 0.22) and carotene (r2 = 0.21),
whereas selenium lacked any correlation. In a partial regression analysis of combined
vitamins with fixed effects of three other classical risk factors, the combination of
vitamin C and lipid-standardized vitamin E yielded a strong inverse correlation with IHD
(r2 = 0.87); the combination of vitamins A, C and E yielded a stronger correlation
(r2 = 0.90). As the authors concluded, in the major cluster of European populations
whose three classical risk factors (cholesterol, blood pressure and cigarette smoking)
were very similar, the sevenfold differences in IHD mortality could in univariate and
multivariate analysis be explained at least sixty percent by the differences in the
plasma levels of vitamin E and up to ninty percent by the combination of vitamins E,
A, and C.
19
In the Basel Prospective Study79, in which the plasma vitamins E and A are
exceptionally high, Gey et al. reported that poor plasma status of carotene (< 0 .2 3
//mol/l) and vitamin C (<22.7/vmol/l) were associated with an increased risk from IHD
and stroke in the subsequent 12 year follow-up, independent of vitamin E and of the
classical cardiovascular risk factors. More interestingly, an interaction between
carotene and vitamin C was found. Low levels of both carotene and vitamin C increase
the risk of IHD further, with RR of 1.53 (1.07-2.20) for low carotene, 1.25 (0.77-2.01)
for vitamin C, and 1.96 (1.10-3.50) for the combination of low carotene and vitamin
C. In the case of stroke, a significant over-multiplicative interaction was found, with
RR of 2.07 (0.78-5.460 for low carotene and normal vitamin C, 1.28 for normal
carotene and low vitamin C, and 4.17 (1.68-10.33) for low carotene and low vitamin
C, compared to normal carotene and vitamin C. The results from this study indicate
that antioxidants other than vitamin E may also play an important role in the prevention
of both cardiovascular and cerebrovascular diseases, despite the high vitamin E level
in this population, and an interactive or synergistic effect may be present among the
essential antioxidant vitamins. An interaction of serum copper, a prooxidant, and
selenium with LDL in the progression of carotid atherosclerosis as measured by the
maximal carotid intima media wall thickness was also reported by Salonen and
colleagues80. A series of reports, by Salonen and colleagues, relating serum
prooxidants, such as iron (ferritin), copper (ceuroplamin) and mercury, to inceased risks
of cardiovascular disease and total mortality provide indirect yet strong evidence of the
protective role of antioxidants in the etiology of heart disease80'83. The study of serum
mercury and risk of myocardial infarction offers some interesting insights into the
apparent paradoxical discrepancy of heart disease risk between Finland, France and
20
other European countries83. The paradoxical direct association between high fish intake
and increased risk of coronary heart disease in the Kuopio Ischemic Heart Disease Risk
Factor study is thought to be accounted for by a high level of mercury content in fresh
water fish due to environmental contamination in this country. Men in the highest
tertile (> 2 .0 jjq/q) of hair mercury content had a 2.0-fold (1.2-3.1, p = 0.05) risk of
acute myocardial infarction as compared with those with a lower hair mercury intake,
and high hair mercury content was associated with high fish consumption and high
titers of immune complexes containing oxidized LDL.
Although inconsistent results have been yielded from studies of plasma or
serum levels of antioxidants and risk of CHD, the failure to find an inverse association
in some of the studies reviewed above may at least partly be attributable to the
methodologic flaws in these studies.
The totality of the evidence from both the animal and epidemiologic studies, however,
supports the antioxidative role of vitamins in the prevention of CHD, presumably due
to the prevention of LDL oxidation.
6. Randomized trials
Although animal models and observational studies, in general, favor an
antioxidative and protective role of vitamins in atherosclerosis and consequent CHD,
etiologic inference or proof of causality can only emerge from large-scale randomized
trials.
Unfortunately, evidence from randomized trials that specifically aimed to test
the effect and efficacy of antioxidant vitamins on cardiovascular or cerebrovascular
disease is sparse. Some secondary evidence from trials that designed to test the effect
21
of antioxidant vitamins on cancers, however, supports the hypothesized antioxidant
role of vitamins in the prevention of development of CHD84,85.
A randomized, double-blind, placebo-controlled 6 year trial among 29,133 male
smokers in Finland failed to detect a significant protective effect of vitamin E (50
mg/day) on the primary endpoint (lung cancer incidence, RR = 0.98)84. Surprisingly, an
excess of 18 percent incidence of lung cancer was found among the men who
received /7-carotene (20 mg/day) compared to the placebo group. However, a point
worth noting is that among the placebo group, the baseline serum concentration of cr-
tocopherol and /?-carotene are inversely associated with the incidence of lung cancer;
also an inverse association between dietary intake of vitamin E and /?-carotene at
baseline and the risk of lung cancer during the trial was found. In secondary analyses
of cardiovascular and cerebrovascular mortality, there were, however, slight protective
trends for ischemic heart disease (RR = 0.95) and ischemic stroke (RR = 0.84) for
vitamin E; hemorrhagic stroke showed the opposite trend (RR = 1.5). For ^-carotene,
an increase in incidence was found for the major endpoints, with RR of 1.12 for
ischemic heart disease, RR of 1.23 for ischemic stroke, and RR of 1.17 for
hemorrhagic stroke. The authors of the study argued that the 'unexpected' findings
may be due to the extreme play of chance.
In another cancer prevention trial of similar sample size (n = 28,584) in Linxian,
China, a-tocopherol (30 mg/day), selenium (50 /yg/day) and /?-carotene (15 mg/day)
together with other trace micronutrients were randomized to eight conditions.
Significant reductions in total mortality (RR = 0.91) and cancer mortality (RR = 0.85)
were found for these antioxidants during five years of follow-up. A protective trend
was also found for cerebrovascular mortality (RR=0.90). No reductions in mortality
22
were found for the other supplements used in this study85.
Finally, the role of /?-carotene in the treatment of CVD was examined among
a subgroup of men with angina in the Physicians Health study, a randomized, double
blind, placebo-controlled trial of 22,071 male physicians aged 40-84 years at
baseline86. A preliminary analysis of a subgroup of 333 subjects with prior Ml, stroke,
or transient ischemic attack (TIA), found a significant reduction in combined
subsequent vascular events among those receiving ^-carotene, compared to those
receiving placebo (RR = 0.46,0.24-0.85). The /?-carotene component, which is testing
50 mg on alternate days, has proceeded and is scheduled to continue until 1995. This
trial should provide important and relevant information on the possible role of p-
carotene in the primary or secondary prevention of CVD.
Clearly, evidence from more well-designed intervention trials are needed before
a conclusive causative role of antioxidant vitamins in the etiology of CHD may be
ascertained.
7. Discussion and future directions for research
1) Threshold or deficiency effect o f antioxidant vitamins
Although evidence reviewed above in general supports a protective role of
vitamins in the development of ischemic heart disease, the apparent discrepancy
among studies, however, suggests that there may be a threshold or deficiency effect
of antioxidant vitamins on risk of ischemic heart disease. The strongest evidence
supporting this notion is provided by the cross-cultural ecologic study reported by Gey
et al.78 The cross-sectional surveys were based on the well-known North-South
gradient of IHD in the European regions. Groups of approximately 100 healthy males,
23
aged 40-49 years, from regions with different mortality rates from CHD were
compared with regard to the plasma level of the principal essential antioxidants. The
study populations were representative for regions with 1) high incidence (> 350 death
yearly from IHD/100,000 males 40-59 years of age): North Karelia, Finland; Southwest
Finland; and Edinburgh, Scotland; 2) medium incidence (< 250/100,000): Belfast,
North Ireland; 3) low incidence (< 130/100,000): Thun and Prealpine, Switzerland;
and Sapri, Southern Italy. In the areas with low and medium coronary mortality, the
plasma value of cholesterol-standardized vitamin E was between 26-28//M (11.2-12.0
mg/l). In contrast, in the areas with most frequent IHD, the medians was below 20-
2 1 .5 /vM (8.6-9.3 mg/l) and on average about 25 percent lower (p < 0 .0 1 ) than areas
with medium-to-low mortality; there was only a small overlap in the distribution. This
comparison suggests that the threshold of risk would be less than 25 jjM or equal to
the average of the three high risk regions, at or above 21 jjM. This suggested
threshold of vitamin E for risk of IHD would be twice as high as the classical vitamin
E deficiency level in men with regards to neuromuscular disorders. The medians of
plasma vitamin C from areas with low CHD mortality were fairly high, between 35-37
fjM (0.62-0.65 mg/dl), and their lowest quartiles were practically out of the critical
range of accepted marginal vitamin C deficiency level, below 22.7 /jM or 0.4 mg/dl.
In contrast, regions with medium-to-high coronary mortality had a significantly lower
vitamin C status. Their medians were on average 33 percent lower (p<0.01), and
were very close to the borderline of marginal vitamin C deficiency level, 26 jjM . Data
on vitamin A, yff-carotene, and selenium, however, did not reveal significant
differences.
Corroborating with the observation by Gey et al., evidence from some within-
24
population studies also supports the notion of a threshold effect of vitamin E and
vitamin C. The failure to find an inverse association between vitamin E and risk of CHD
in the Augsburg cohort and the Basel cohort may be attributed to the already high
levels of plasma vitamin E in these two populations75,79. In the Basel Prospective
Study, all plasma vitamin E quartiles (median 35 //M/I) were above the presumed
critical threshold for IHD risk, approximately 26-28 //M/I. In the Augsburgh cohort,
both cases and their matched controls had a fairly high mean serum vitamin E
concentration, 33.9 //M/I and 32.8 //M/I (p = 0.37), respectively. In the study reported
by Riemersma et al. an apparent threshold effect was found for plasma /7-carotene and
vitamin A, as well as vitamins E and C69. Comparison of the RR estimates for quintiles
of the four antioxidants was graphically depicted in Figure 1. The critical threshold
values for plasma vitamin E and vitamin C indicated in this study were very similar to
those suggested by Gey et al. from the ecologic study.
Prospective studies relating dietary antioxidant vitamins to subsequent risk of
coronary heart disease also provide support for this concept. In the Nurse's Health
Study and the Health Professionals Study65,66, no increased protective effect was
observed for vitamin E intake beyond a supplement dose of 250 lU/day for both men
and women (Figure 2). In the Mobile Clinic of the Social Insurance Institute Study from
Finland67, an apparent threshold dietary intake of vitamin E was found for men, but not
for women (Figure 3). In men, a dietary intake of vitamin E in the second tertile (6.9-
8.9 mg/day) was found to have an almost identical protective effect (RR = 0.97, 0.67-
1.40) to that (RR = 1.0) of the top tertile intake (> 8.9 mg/day) when both were
compared to the bottom tertile group of intake (RR = 0.68, 0.42-1.11).
In summary, the presently available data suggest that in order to achieve the
25
optimal protective effect of essential antioxidants against ischemic heart disease, an
optimal daily intake may be necessary. It is not prudent for the public, however, to
take daily mega-dose of supplements as advocated by some over-zealous believers, for
the obvious reason that no extra protective effect may be expected when the essential
antioxidants reach the saturation or threshold levels. Also, a high level of one
antioxidant achieved by overdose supplementation is frequently at the expense of other
essential antioxidants. An imbalance of essential antioxidants may likely to do more
harm than good to the optimum health. Before the biological mechanisms and health
effects of essential antioxidants are fully elucidated, it is necessary for the beneficial
and potential adverse effects of high-dose vitamin supplementation to be carefully
weighed.
2) Future directions for research
The totality of available epidemiologic evidence supports the possibility that
antioxidant vitamins may have a protective effect in CHD, and basic research supports
a plausible mechanism for the involvement of oxidative stress in atherogenesis. At
present, antioxidants represent a promising but clearly unproven means for the
prevention of CHD.
Whether benefits of antioxidants might be expected solely from increased intake
of foods alone is still not clear. Findings from some of the studies reviewed here imply
that there may be a critical or threshold level of vitamin E beyond which no extra
benefit will be gained. The elevated risk of CHD in some populations, including the
U.S., may be due merely to the effect of vitamin E deficiency, rather than the vitamin's
antioxidant properties. The advocation of high dose supplement use of antioxidant
26
vitamins to the public is still premature at this stage.
Antioxidative defense against oxidative stress may involve an array of many
antioxidants that exert multilevel protection. Many synergistic and interactive links of
essential antioxidants and prooxidants may be present in the human body, as shown
by experimental and epidemiologic data. Overload of any one single antioxidant by high
dose supplementation, at the expense of others, may eventually compromise the
body's antioxidative defense system. The simultaneous optimization of all principal
antioxidants would well fit the established principle of multiple risk factor intervention
in the multifactorial process of CHD.
Several avenues should be pursued in future research on antioxidants and CHD.
Further basic research and better animal models are needed to explore the mechanism
of oxidation and the extent to which this occurs in vivo, and the optimal level of
antioxidant vitamins that should be maintained in the human body for the prevention
of CHD. Although additional descriptive and observational study findings may provide
further support for a possible beneficial effect of dietary antioxidants, the only way to
determine definitively whether antioxidants offer a benefit in both treatment or
prevention of atherosclerotic disease is through randomized trials. Because of the long
latency period of atherogenesis, future intervention trials aimed at the prevention of
CHD events will inevitably have to keep a large number of subjects in compliance with
the trial condition for a very long time before any definitive results can be obtained.
Given the high percentage of vitamin supplement use in the U.S population, this kind
of trial is clearly not feasible. The ultrasound imaging technique that has now being
used in some epidemiologic studies to measure the intima media wall thickness (IMT)
is of promise in this aspect. Intervention trials, in which IMT is measured as an early
27
indication of atherosclerosis, and in which the combination of several essential
antioxidants is simultaneously tested in a factorial manner, should provide reliable
evidence on the role of antioxidants in the etiology and the primary prevention of
cardiovascular disease.
28
Reference
1. Peto R, Doll R, Buckley JD, Sporn MB. Can dietary yff-carotene materially reduce
human cancer rates? Nature 1981;290:201-8.
2. Steinberg D, Parthasarathy S, Carew TE, Khoo JC, Witztum JL. Beyond
cholesterol: modification of low-density lipoprotein that increase its
atherogenecity. N Engl J M ed 1989;320:915-24.
3. Steinberg D. Antioxidants in the prevention of human atherosclerosis. Summary
of the proceedings of a National Heart, Lung, and Blood Institute Workshop:
September 5-6, 1991, Bethesda, Maryland. Circulation 1992;85(6):2337-44.
4. Packer L. Protective role of vitamin E in biological systems. Am J Clin Nutr
1991 ;53:Suppl:1050s-1055s.
5. Steinbrecher UP, Zhang HF, Lougheed M. Role of oxidatively modified LDL in
atherosclerosis. Free Radic Bio Med 1990;9(2):155-168.
6. Steiberg D, Witztum J. lipoproteins and atherogenesis: current concepts. JAMA
1990;264:3047-52.
7. Quinn MT, Parthasarathy S, Fong LG, Steinberg D. Oxidatively modified low
density lipoproteins:a potential role in recruitment and retention of
monocyte/macrophages during atherogenesis. Proc Natl Acad Sci USA
1987;84:2995-8.
8. Palinski W, Rosenfeld ME, Yla-Herttuala S. Low-density lipoprotein undergoes
oxidative modification in vivo. Proc Nat! Acad Sci USA 1989;86:1372-6.
9. Evensen SA, Gaidai KS, Nilsen E. LDL-induced cytotoxicity and its inhibition by
anti-oxydant treatment in cultured human endothelial cells and fibroblasts.
Atherosclerosis 1983; 49:23-30.
10. Morel DW, Hessler JR, Chisolm GM. Low density lipoprotein cytotoxicity
induced by free radical peroxidation of lipid. J Lipid Res 1983; 24:1070-6.
11. Hessler JR, Morel DW, Lewis LJ, Chisolm GM. Lipoprotein oxidation and
lipoprotein-induced cytotoxicity. Arteriosclerosis 1983; 3:215-22.
12. Mangin EL, Kerns S, Nguy J, Eisenberg SB, Li Z, Henry PD. Endothelial toxicity
of oxidized LDL is suppressed by treatment of the lipoprotein with a
lysolecithinase. Circulation 1990;82(Suppl. Ill):abstract 675.
13. Sparrow CP, Parathasarathy S, Steinberg D. A macrophage receptor that
recognizes oxidized low density but not acetylated low density lipoprotein. J
Biol Chem 1989; 264:2599-604.
29
14. Halliwell B, Gutteridge JMC. Free radicals in biology and medicine. 2nd ed. New
York: Oxford University Press, 1989.
15. Packer L, Fuchs J. Vitamin E in health and disease. Macel Dekker, Inc. 1993.
16. Esterbauer H, Jurgens G, Koller E. Autoxidation of human low density
lipoprotein: loss of polyunsaturated fatty acids and vitamin E and generation of
aldehydes. J Lipid Res 1987;28:495-509.
17. Jialal I, Grundy Sm. Combined supplementation with alpha-tocopherol, ascorbic
and beta-carotene inhibits LDL oxidation. Circulation 1992;86{4):l-462.
18. Jialal I, Norkus EP, Cristol L, Grundy SM. Beta-carotene inhibits the oxidative
modification of low-density lipoprotein. Biochem Biophys A cta
1991; 1086(1 ):134-138.
19. Esterbauer H, Dieber-Rotheneder M, Striegl G, Waeg G. Role of vitamin E in
preventing the oxidation of low-density lipoprotein. Am J Clin Nutr
1991 ;53:Suppl:314s-321 s.
20. Esterbauer H, Puhl H, Dieber RM, Waeg G, Rabl H. Effect of antioxidants on
oxidative modification of LDL. Ann Med 1991;23(5):573-81.
21. Esterbauer H, Dieber-Rotheneder M, Striegl G, Waeg G. Vitamin E and other
lipophilic antioxidants protect LDL against oxidation. Fat Sci Technol
1989;91:316-24.
22. Dieber-Rotheneder M, Puhl H, Waeg G, Striegl G, Esterbauer H. J Lipid Res
1991;32:1325-1332.
23. Babiy AV. Gebicki JM, Sullivan DR. Vitamin E and low density lipoprotein
oxidizability induced by free radicals. Atherosclerosis 1990;81:1175-82.
24. Jessup W. Rankin SM, De Walley CV. Alpha-tocopherol consumption during
low-density lipoprotein oxidation. Biochem J 1990;265:399-405.
25. Frei B, England L, Ames BN. Ascorbate is an outstanding antioxidant in human
blood plasma. Proc Natl Acad Sci USA 1989;86:6377-81.
26. Doba T, Burton GW, Ingold KU. Antioxidant and co-antioxidant activity of
vitamin C. The effect of vitamin C, either alone or in the presence of vitamin E
or a water-soluble vitamin E analogue, upon the peroxidation of aqueous
multilamellar phospholipid liposomes. Biochem Biophys Acta 1985;835:298-
303.
27. Vatassery GT, Smith WE, Quach HT. Ascorbic acid, glutathione and synthetic
antioxidants prevent the oxidation of vitamin E in platelets, lipids
1989;24:1043-7.
30
28. Frei B, Stocker R, Ames BN. Antioxidant defenses and lipid peroxidation in
human blood plasma. Proc Natl Acad Sci USA 1988;85:9748-52.
29. Frei B, Forte TM, Ames BN, Cross CE. Gas phase oxidants of cigarette smoke
induced lipid peroxidation and changes in lipoprotein properties in human blood
plasma: protective effects of ascorbic acid. Biochem J 1991 ;277:133-38.
30. Burton GW, Ingold KU. /?-carotene: an unusual type of lipid antioxidant. Science
1984;224:569-73.
31. Menkes MS, Comstock GW. Helsing KJ. Serum beta-carotene, vitamins A and
E, selenium, and the risk of lung cancer. N Engl J M ed 1986;315:1250-4.
32. Wald NJ, Boreham J, Hayward JL, Bulbrook RD. Plasma retinol, beta-carotene,
and vitamin E in relation to the future risk of breast cancer. Br J Cancer
1984;49:321-4.
33. Santamaria L, Bianchi A, Arnaboldi A. Chemoprevention of indirect and direct
chemical carcinogenesis by carotenoids and oxygen radical quenchers. Ann N
Y Acad Sci 1988;534:584-96.
34. Morel D, Hessler J, Chisolm M. J Lipid Res 1983;24:1070-1076.
35. Van Hinsberg VW, Scheffer M, Havekes L, Kempen H. Biochem Biophys Acta
1986;878:49-64.
36. Navab M, Imes S, Hama S, Hough G, Ross L, Bork R, Valente A, Berliner J,
Drinkwater D, Laks H. J Clin Invest 1991;88:2039-2046.
37. Xu MJ, Peng YM, Liu Y, Alberts DC, Plezia PM, Sayers SM. Effects of chronic
oral administration of /?-carotene on plasma cr-tocopherol concentrations in
normal subjects. Proc Am Assoc Cancer Res 1990;31:126.
38. Willett WC, Stampfer MJ, Underwood BA, Taylor JO. Vitamins A, E, and
carotene: effect of supplementation on their plasma levels. Am J Clin Nutr
1983;38:559-66.
39. Arnrich L, Arthur VA. Interactions of fat-soluble vitamins in hypervitaminoses.
Ann N Y Acad Sci 1980;355:109-18.
40. Alam BS, Brown LR, Alam SQ. Influence of dietary fats and vitamin E on
plasma and hepatic vitamin A and /?-carotene levels in rats fed excess 1 3 -
carotene. Nutr Cancer 1990; 14:111 -6.
41. Albanes D, Virtamo J, Rautalahti M. Pilot study: the U.S.-Finland Lung Cancer
Prevention Trial. J Nutr Growth Cancer 1986;3:207-14.
42. Smith TL, Kummerow FA. Effect of dietary vitamin E on plasma lipids and
31
atherogenesis in restricted ovulatory chickens. Atherosclerosis 1989;75:105-
109.
43. Wojcicki J. Rozewicka L, Barcew-Wiszniewska B. Effect of selenium and
vitamin E on the development of experimental atherosclerosis in rabbits.
A therosclerosis 1991;87:9-19.
44. Verlangieri AJ, Bush M. Prevention and regression of atherosclerosis by alpha-
tocopherol. J Am Coll Nutr 1992; 11(2): 131-138.
45. Willis GC. An experimental study of the intimal ground substances in
atherosclerosis. Can Med Assoc J 1953;69:17-22.
46. Willis GC. The reversibility of atherosclerosis. Can M ed Assoc J 1957;77:106-
109.
47. Altman RFA, Schaeffer MV, Salles CA. Phospholipids associated with vitamin
C in experimental atherosclerosis. Drug Res 1980;30:627-630.
48. Bocan TM, Mueller SB, Brown EQ, Uhlendorf PD, Mazur MJ, Newton RS.
Antiatherosclerotic effects of antioxidants are lesion-specific when evaluated
in hypercholesterolemic New Zealand white rabbits. Exp Mol Pathol
1992;57(1 ):70-83.
49. Kita T, Nagano Y, Yokode M, Ishi K. Probucol prevents the progression of
atherosclerosis in Watanabe heritable hyperlipidemic rabbit, an animal model for
familial hypercholesterolemia. Proc Natl Acad Sci USA 1987;84:5928-5931.
50. Carew T, Schwenke D, Steinberg D. Antiatherogenic effect of probucol
unrelated to its hypocholesterolemic effect: evidence that antioxidants in vivo
can selectively inhibit low density lipoprotein degradation in macrophage-rich
fatty streaks and slow the progression of atherosclerosis in the Watanable
heritable hyperlipidemic rabbits. Proc Natl Acad Sci USA 1987;84:7725-7729.
51. Bjokhem I, Breuer 0, Diczfalusy U. The antioxidant butylated hydroxytoluene
protects against atherosclerosis. Arteriosclerosis Thrombosis 1991; 11:15-22.
52. Fong LG, Witztum L, Steinberg D. Nonenzymatic oxidative cleavage of peptide
bonds in apoprotein B-100. J Lipid Res 1987;28:1466-1477.
53. Acheson RM, Williams DR. Does consumption of fruit and vegetables protect
against stroke? Lancet 1983; 1:1191-1193.
54. Armstrong BK, Mann JL, Adelstein AM, Eskin F. Commodity consumption and
ischemic heart disease mortality, with special reference to dietary practices. J
Chronic Dis 1975;36:673-677.
55. Verlangieri AJ, Kapeghian JC, El-Dean S, Bush M. Fruit and vegetable
32
consumption and cardiovascular disease mortality. Med Hypotheses 1985; 16:7-
15.
56. Ginter E. Decline of coronary mortality in the United States and vitamin C. Am
J Clin Nutr 1979;32:511-512.
57. Artaud-Wild SM, Connor SL, Sexton G, Connor WE. Differences in coronary
mortality can be explained by differences in cholesterol and saturated fat
intakes in 40 countries but not in France and Finland: A Paradox. Circulation
1993;88(6):2771-2779.
58. Garcia-Palmieri MR, Sorlie P, Tillotson J. Relationship of dietary intake to
subsequent coronary heart disease incidence. The Puerto Rico Heart Health
Program. Am J Clin Nutr 1980;33:1818-27.
59. Kromhout D, De Lezenne Coulander C. Diet, prevalence and 10-year mortality
from coronary heart disease in 871 middle-aged men. The Zutphen Study. Am
J Epidemiol 1984; 119:733-41.
60. Kushi LH, Lew RA, Stare FJ. Diet and 20-year mortality from coronary heart
disease. The Ireland-Boston Diet-Heart Study. N Engl J Med 1985;312:811-18.
61. Gaziano JM, Manson JE, Branch LG, LaMott GA, Golditz GA, Hennekens CH.
Dietary beta carotene and decreased cardiovascular mortality in an elderly
cohort. J Am Coll Cardiol, (abstract); 1992; 19(3 Suppl. A):377A.
62. Lapidus L, Anderson H, Bengtson C, Bosceus I. Dietary habits in relation to
incidence of cardiovascular disease and death in women; a 12 year follow-up
of participants in the study of women in Gothengerg, Sweden. Am J Clin Nutr
1986;44:444-448.
63. Estrom JE, Kanim LE, Klein MA. Vitamin C intake and mortality among a sample
of the United States population. Epidemiology 1992;3:194-202.
64. Hertog MGL, Feskens EJM, Hollman PCH, Katan MB, Kromhout D. Dietary
antioxidant flavonoids and the risk of coronary heart disease: the Zutphen
Elderly Study. Lancet 1993;342:1007-11.
65. Stampfer M, Hennekins C, Manson J, Colditz G, Rosner B, Willett W. Vitamin
E consumption and the risk of coronary disease in women. N Engl J Med
1993;328:1444-9.
66. Rimm E, Stampfer M, Ascherio A, Giovannucci E, Colditz G, Willett W. Vitamin
E consumption and the risk of coronary disease in men. N Engl J Med
1993;328:1450-6.
67. Knekt P, Reunanen A, Jarvinen R, Seppanen R, Heliovaara M, Aromaa A.
Antioxidant vitamin intake and coronary mortality in longitudinal population
33
study. Am J Epidemiol 1994; 139:1180-9.
68. Meyer F, Bairati I, Dagenais GR. Ischemic heart disease incidence and mortality
in relation to vitamin supplement use in a cohort of 2,226 men (abstract).
Circulation 1994;89(2):932.
69. Riemersma RA, Wood DA, Macintyre CCA, Elton RA, Gey KF, Olive MF. Risk
of angina pectoris and plasma concentrations of vitamins A, C, and E and
carotene. Lancet 1991;337:1-5.
70. Ramirez J, Flowers NC. Leukocyte ascorbic acid and its relationship to coronary
artery disease in man. Am J d in Nutr 1980;33:2079-87.
71. Kardinaal AFM, Kok FJ, Ringstad J, Gomez-Aracena J, Mazaev VP, Kohlmeier
L, Martin BC. Antioxidants in adipose tissue and risk of myocardial infarction:
the EURAMIC study. Lancet 1993;342:1379-84.
72. Street DA, Comstock GW, Salkeld RM. A population-based case-control study
of association of serum antioxidants and myocardial infarction. (Abstract). Am
J Epidemiol 1991; 134:719-20.
73. Kok FJ, De Bruijn AM, Hofman A, Vanlaar A. Serum selenium, vitamin
antioxidants and cardiovascular mortality: a 9 year follow-up study in the
Netherlands. Am J d in Nutr 1987;45:462-468.
74. Salonen JT, Salonen R, Penttila I, Herranen J. Serum fatty acids,
apolipoproteins, selenium and vitamin antioxidants and risk of death from
coronary artery disease. Am J Cardiol 1985;56:226-231.
75. Hense HW, Stender M, Bors W, Keil U. Lack of an association between serum
vitamin E and myocardial infarction in population with high vitamin E levels.
Atherosclerosis 1993;103:21-28.
76. Gey FK, Puska P, Jordan P, Moser UK. Inverse correlation between plasma
vitamin E and mortality from ischemic heart disease in cross-cultural
epidemiology. Am J d in Nutr 1991;53:326s-331s.
77. Gey KF. The antioxidant hypothesis of cardiovascular disease: epidemiology and
mechanisms. Biochem Soc Trans 1990; 18(6): 1041-5.
78. Gey KF, Brubacher GB, Stahelin HB. Plasma levels of antioxidant vitamins in
relation to ischemic heart disease and cancer. Am J d in Nutr 1987;45:1368-
1377.
79. Gey KF, Stahelin HB, Eichholzer M. Poor plasma status of carotene and vitamin
C is associated with higher mortality from ischemic heart disease and stroke:
Basel Prospective Study, d in invest 1993;71:3-10.
34
80. Salonen JT, Salonen R, Seppanen K, Kantola M, Suntioinen S, Korpela H.
Interactions of serum copper, selenium, and low density lipoprotein cholesterol
in atherogenesis. BMJ 1991;302:756-60.
81. Salonen JT, Nyyssonen K, Korpela H, Tuomilehto J, Seppanen R, Salonen R.
High stored iron levels are associated with excess risk of myocardial infarction
in Eastern Finnish men. Circulation 1992;86:803-811.
82. Salonen JT, Salonen R, Korpela H, Suntioinen S, Tuomilehto J. Serum copper
and the risk of acute myocardial infarction: a prospective population study in
men in Eastern Finland. Am J Epidemiol 1991;134:268-276.
83. Salonen JT, Seppanen K, Nyyssonen K, Korpela H, Kauhanen J, Tuomilehto J,
Esterbauer H, Tatzber F, Salonen R. Intake of mercury from fish, lipid
peroxidation, and the risk of myocardial infarction and coronary, cardiovascular,
and any death in Eastern Finnish men. Circulation 1995;91:645-655.
84. Alpha-tocopherol_beta_carotene_cancer_prevention_study_group.Theeffectof
vitamin E and beta carotene on the incidence of lung cancer and other cancers
in male smokers. N Engl J M ed 1994;330(15): 1029-1035.
85. Blot WJ, L i J-V, Taylor PR, et al. Nutrition intervention trials in Linxin, China:
Supplementation with specific vitamin/mineral combinations, cancer incidence,
and disease-specific mortality in the general population. J Natl Cancer Inst
1993;85:1483-1492.
86. Gaziano JM, Manson JE, Ridker PM, Buring JE, Hennekens CH. Beta carotene
therapy for chronic stable angina. Circulation 1990;82(4, Suppl lll):lll-202.
35
Talile 1. Review uf Studies of Antioxiduut Vituinins and Atherosclerosis In Animal Models
Source Model Antioxidant Supplementation Results
Smith TL, et al.
Atherosclerosis, 75 (1989)
Chicken 1000 1U vitamin E per kg diet With vs. Without Vitamin E
1. TBARS:
7.7 ±1.4 vs. 12.2±2.8 (am/ml)
2. Intimal Thickness:
15±3 vs. 18±1 (pm)
Verlangieri AJ, et al.
J Am Coll Nutr, 11(2), 1992
Monkeys 108 lU/day d-a-tocopherol Treatment vs. Placebo (36 months):
1. % Ultrasound stenosis:
18% vs. 87%
2. Plasma tocopherol:
negatively p with % stenosis
Wilson RB, ei al.
J Nutr. 108 (1978)
Rabbits 1.0% dl-atocoplierol Tocopherol vs. Butter (12 months)
1. Mean Aortic scores
0.2 vs. 1.1
2. Incidence of atherosclerosis
0/12 vs. 6/12
Donaldson WE
Poultry Science 61 (1982)
Japanese
Quail
100 nig d-a-tocopherol acetate
per kg diet
Tocopherol vs. No Tocopherol
1. % Aortic stenosis:
43 vs. 42 (%)
2. Average aortic lesion score:
0.6 vs. 1.0
3. Severity score:
1.4 vs. 2.5
Bocan TMA, et al. New Zealand
Rabbits
500 mg/kg vitamin E and C each 1. Thoracic aortic CE content:
decreased by 40 (%)
2. Thoracic lesion coverage:
decreased by 46 (%)
(Table 1. continued)
Moses C, et al. New Zealand
Rabbits
IM 25 mg a-locopherol daily Tocopherol vs. No Tocopherol
1. 8 wk total cholesterol (mg %):
985 vs. 763
2. Atherosclerotic score:
1.5+ vs. 1.0+
Kleinveld HA, et al.
Aiterioscler Thromb.(1994)
Rabbits
(W111IL)
0.025 % (wt/wl) vitamin E Tocopherol vs. No Tocopherol
1. Lag phase (mins):
180.0±25 vs. 116.4 ±13.3
2. % Aortic arch surface with lesion:
92.6 ±2.8 vs. 93.5 ±2.1
3. % Thoracic aorta lesion:
50.7±18.9 vs. 65.1±18.0
4. % Abdominal aorta lesion:
19.1 ±17.7 vs. 23.3 ±10.2
Andersson TLG, el al.
Atherosclerosis H I (1994)
Rabbits
(WIIL)
0.2% vitamin E Endothelial Dysfunction
1. Response to nitroprusside:
no difference between groups
2. Response to acetylcholine:
t in vitamin E group
Godfried BS, et al.
BrJ Nutr. 61 (1989)
!
Rabbits (1) 2000 mg vitamin E/kg diet
(2)10,000 mg vitamin E/kg diet
Vitamin E vs. Control
1. Extent of Aortic Plaque Formation:
49±6 vs. 58±5
2. Endothelial Loss (%):
54 ±76 vs. 22±5
3. Plaque Area (%):
73 ±9 vs. 44 ±6
Table 2. Epidemiological Studies Correlating Antioxidants and Prooxidants to Ischemic Heart Disease
Study Sample Size and Design Antioxidant/Prooxidant
WHO European Study 1,600 men Plasma Vitamins E, C, A and /3-
Cross Cultural Carotene
Basel Study (Switzerland) 2,974 men Plasma Vitamins E, C, A and /3-
Prospective Carotene
WHO Edinburgh Study (UK) 6,000 men Plasma Vitamin E, Vitamin C and
Cross-sectional V itam in A
Baltimore Study (US) 25,802 men &women Serum /3-carotene, Lutein and
Prospective Vitamin E
Nurses’ Health Study (USA) 87,245 women Dietary Vitamin E, Vitamin C and
prospective /3-Carotene
Health Professionals Follow-up 39,910 men Dietary Vitamin E, Vitamin C and
Study (USA) prospective /3-Carotene
First National Health and 11,348 men St women Dietary Vitamin C
Nutrition Examination Study
(USA)
prospective
Physicians’ s Health Study (USA) 333 men
prospective
Dietary /3-carotene (supplement)
Mobile Clinic o f the Social 5,133 men & women Dietary Vitamin E, C and /3-
Insurance Institution Study
(Finland)
prospective Carotene
Quebec Study (Canada) 2,226 men
prospective
Dietary Vitamin E (supplement)
USC Cholesterol Lowering 162 men Dietary Vitamin E
Atherosclerosis Study (CLAS) prospective
Kuopio Ischemic Heart Disease 1,931 men Serum Iron (Ferritin)
Risk Factor Study (Finland) prospective (Prooxidant)
Kuopio Ischemic Heart Disease 1,666 men Serum Copper
Risk Factor Study (Finland) Prospective (Prooxidant)
Kuopio Ischeminc Heart Disease 1,833 men Serum Mercury
Risk Factor Study (Finland) Prospective (prooxidant)
38
Figure 1. Angina Pectoris and Plasma Vitamins A, C, E and Carotene
O dda Ratio
7
e
6
4
3
2
1
0
O dda Ratio
1*t 2nd 3nd 4th
Plasma Vitamin E Qulntlles
6th
7
6
6
4
3
2
1
0
tat 2nd 3nd 4 th 6 lh
•euroas R laaaraa DA a t at.
Plasma Vitamin A Qulntlles
a o u f c t j R U m t r a a BA « | * 1
O dda Ratio
9.6
2.6
1.6
0.6
la l 2nd 3nd 4th 6th
O d d a R a tio
Plasma Carotene Qulntlles
S.6
2.6
1.6
0.6
ta t 2nd 3nd 4th 6th
Plasm a Vitamin C Qulntlles
aouteai R ltm araa RA a t *L
Figure 2. Risk of Coronary Heart Disease and
Vitamin E Supplement Use (lU/day)
Relative Risk
Women (n-87,245)
0.8
0.6
0.4
0.2
0 <100 100-250 250 +
Vitamin E Supplementation (lU /day)
Rlmm E B / S t a m p f e r M J e t al, N E JM , 1 9 9 3
Figure 3. Dietary Vitamin E and Risk of
CHD: Finnish Men & Women (N=5,133)
Relative Risk
Men
0.8
Women
0.6
0.4
0.2
1st tertile 2nd tertile 3rd tertile
Dietary Vitamin E Tertiles
Knekt P. et al, AJE, 1994
Chapter 2. Antioxidant Vitamin intakes and Risk o f Myocardial Infarction: The Kuopio
ischemic Heart Disease Risk Factor Study
Abstract
Objective: To assess the relationship of dietary antioxidant vitamin intake and risk of
myocardial infarction.
Design: Longitudinal follow-up study, initial examination 1984-1989, follow-up until
December, 1992.
Participants: 1,609 middle-aged men free of coronary heart disease and cancer living
in Eastern Finland.
Results: After an average of 5.5 years of follow-up, 74 incident myocardial infarction
(Ml) cases were documented. In multivariate Proportional Hazard regression analyses,
no association was found between dietary intake of vitamin E, vitamin C or carotenoid
with Ml. Compared with men in the lowest tertiles of dietary intakes, the relative risk
estimates for the top tertile groups of dietary intakes were 1.41 (0.75-2.64) for
vitamin E, 1.33 (0.72-2.47) for vitamin C and 0.90 (0.49-1.65) for carotenoids,
respectively. Estimation of the association between the sum of dietary intake of the
three antioxidants and Ml yielded a RR of 1.35 (0.74-2.43). Furthermore, no
association was found between the major food determinants of antioxidants and Ml.
Conclusion: Our results do not support the antioxidant hypothesis of cardiovascular
disease. The lack of associations of dietary vitamin intakes, however, may be
attributed to the relatively low dietary intake of antioxidants in this population.
Key words: Diet, Vitamin E, Vitamin C, Carotenoid, Myocardial Infarction.
42
Introduction
Free radical mediated oxidation of low density lipoprotein (Ox-LDL) is thought
to play a pivotal role in the development of atherosclerosis1"3. Oxidized LDL has been
implicated in many steps of atherogenesis. Ox-LDL has been shown to be cytotoxic to
endothelial cells4,5, chemotactic to circulating monocytes and promoting the adhesion
of monocytes to the endothelium and migration into the subendothelium6. Oxidative
modification of LDL particles in the arterial subendothelium results in structural
changes4,5. Uptake of Ox-LDL by macrophage through the unchecked scavenger
receptor pathway leads to the formation of foam cells and initiates the pathological
process of atherogenesis2,3. Antioxidants such vitamin E, vitamin C and carotenoid are
hypothesized to help prevent atherosclerosis by blocking the oxidative modification of
LDL.
Accumulating data from animal models and in vitro experimental studies support
a link between dietary antioxidants and a reduced risk of atherosclerosis7'11; however
epidemiologic studies have yielded inconsistent results. Ecologic correlation analysis
demonstrated an inverse association of dietary consumption of fruits and vegetables
and coronary mortality, presumably due to a high antioxidant content in fruits and
vegetables12"13. Many studies directed specially at dietary antioxidants intake on an
individual level have yielded ambiguous results. Recently, two large cohort studies in
the U.S. have shown that a high supplemental use of vitamin E was associated with
a low incidence of coronary heart disease (CHD) in both men and women, but no
inverse association was found for dietary vitamin E intake alone14,15. In both studies,
dietary vitamin C intake alone or supplement use failed to show a significant inverse
association with risk of CHD.
43
Finland is one of the nations with the highest incidence and mortality from
CHD. Deficiency of antioxidants in the Finnish diet has long been hypothesized to be
one contributing factor to the excess risk of CHD16. We thus examined the association
between dietary intake of vitamin E, vitamin C and carotenoid and subsequent
incidence of myocardial infarction (Ml) in a cohort of middle-aged Eastern Finnish men
participating the Kuopio Ischemic Heart Disease Risk Factor study.
Materials and Methods
The Kuopio Ischemic Heart Disease Risk Factor Study is a population-based
study designed to investigate previously unestablished risk factors for CHD and extra
coronary atherosclerosis in eastern Finnish men, a population with one of the highest
incidence and mortality rates of CHD17. The details of the study design and
implementation has been described in detailed elsewhere18,19. Briefly, baseline exams
were carried out between March 1984 and December 1989. The study sample
included 3,235 men aged 42, 48, 54, or 60 years at the baseline examination. Of
these men, 2,682 (82.9 percent) participated. Men with either prevalent CHD (n = 667)
or cancer (n = 46) or having a total energy intake less than 500 kcal or over 5,000 kcal
(n = 12) were excluded from the present analysis. Prevalent ischemic heart disease was
defined as either a history of myocardial infarction (Ml) or angina pectoris, positive
angina pectoris on effort in the London School of Hygiene interview20, or use of
nitroglycerin tablets once a week or more frequently. Of the remaining 1,957 men,
complete data on the interested variables were available for 1,604 men, and was used
for subsequent analyses.
The consumption of foods was assessed at baseline exam with an instructed
44
and interview-checked 4-day dietary recording by household measures21. The
instructions were given and the completed food records were checked by a nutritionist.
The intake of nutrients, including vitamin E, vitamin C and carotene was estimated
using NUTRICA software. The data bank of NUTRICA is complied using mainly Finnish
values for the nutrient composition of foods19. In the present study, the average intake
of nutrients from the 4-day dietary records were used. Total energy intake was
calculated based on the amounts of fat, protein, and carbohydrates consumed.
Reproducibility of dietary intakes was estimated by repeating the 4-day dietary
recording on a random subsample of 50 men approximately 12 months after the
baseline examination. For vitamin E, vitamin C and carotenoid, the intraclass
correlations were 0.56, 0.62, and 0.14 respectively; and it was 0.52 for alcohol
consumption, and 0.67 for total energy intake.
A self-administered questionnaire concerning socio-demographic factors,
individual and family medical history, smoking, and physical activity was performed by
all cohort members. Repeat interviews to obtain medical history were conducted by
a physician. A subject was defined as a smoker if he had ever smoked on a regular
basis and smoked cigarettes, cigar, or pipe within the past 30 days. The life-long
exposure to smoking ("cigarette year") was estimated as the product of years smoked
and the number of tobacco products smoked daily at the time of examination. Years
smoked were defined as the sum of years of smoking regardless of when smoking had
started, whether the subject had stopped smoking, and whether smoking had occurred
continuously or during several periods. The socioeconomic status was measured with
the summary index that combined measures of income, education, occupation,
occupational prestige, material standard of living, and housing condition.
45
Leisure time physical activity was assessed using a 12-Month Leisure Time
Physical Activity Questionnaire. The questionnaire included the most common leisure
time physical activities of middle-aged Finnish men, selected on the basis of a previous
population study conducted in Finland. The intensity of physical activity was expressed
in metabolic units (METs). The MET is the ratio of metabolic rate during physical
activity to the metabolic rate at rest. One MET corresponds to an energy expenditure
of approximately 1 kcal/kg/hour and an oxygen uptake of 3.5 ml/kg/minute.
Resting blood pressure was measured on the first examination day by one nurse
with a random-zero mercury sphygmomanometer. The measuring protocol included
three measurements in supine, and one in standing, and two in sitting position. The
mean of all six systolic blood pressure values was used in the present analysis.
The respiratory gas exchange was measured breath-by-breath with an MGC
2001 system (Medical Graphics Corp., Minneapolis, MN) during a symptom-limited
exercise test. The testing protocol comprised a linear increase of work load by 20
W/min. Highest oxygen uptake during the test was defined as V02max. Exercise ECGs
were coded manually by one cardiologist. The criteria for ischemia were 1) ischemic
ECG defined as horizontal or downsloping ST depression > 0.5 mm or upsloping ST
depression > 1.0 mm; 2)typical angina pectoris pain leading to discontinuation of
exercise; 3) maximal heart rate during exercise < 130 beats per minute.
Venous blood was taken from each subject after having abstained from alcohol
for 3 days, smoking for 12 hours, and fasting for 8-12 hours. The main lipoproteins
were separated from fresh serum samples using ultracentrifugation and precipitation.
The HDL2and HDL3 subfractions were separated using ultracentrifugation at 108,000g
for 62 hours against a density of 1.125 g/cm3. Total serum cholesterol was measured
46
enzymatically.
As a part of the multinational MONICA project, an acute myocardial infarction
(AMI) registry was established in the province of Kuopio in 1982. The registry collects
detailed diagnostic information of all heart attacks in the population in a prospective
manner. Heart attacks were classified as definite AMI, possible AMI, no AMI, or
insufficient data according to explicitly defined, uniform diagnostic criteria as described
elsewhere22. Coverage of the AMI registry was checked against the national
computerized death certificate registry. Diagnostic information and the date of onset
of all heart attacks in the present study cohort was obtained by record linkage based
on the uniform Finnish personal identification code.
Statistical Analysis
Estimation of the age-adjusted mean dietary intake of nutrients and total energy
was based on generalized linear models. The associations between dietary vitamin E,
C, and carotenoid and Ml were estimated using Cox regression model. First, the
associations of dietary intakes of vitamin E, vitamin C and carotenoid were estimated
in base models including baseline age and year of examination as covariates. Then, the
associations were estimated with further adjustment for classical risk factors for CHD,
i.e. serum total cholesterol, HDL, systolic blood pressure, body mass index (BMI),
smoking status, exercise ECG abnormality and total energy. Finally, the associations
were estimated in full models with potential life style confounders being added into the
models, including leisure time physical activity, alcohol consumption, saturated fat
acids intake, socioeconomic status, maximal oxygen uptake and living areas. To assess
the associations of major food determinants of antioxidant vitamins and risk of Ml,
47
several foods were added one at a time to the fully adjusted models. The distributions
of dietary intake variables and leisure time physical activity are skewed in the cohort.
To normalize the distributions of these variables, vitamin E, vitamin C and carotenoid
intakes were categorized into tertiles in all models; total energy intake and leisure time
physical activity were categorized into quartiles; alcohol consumption was categorized
into 0, 0.1 to 15.0, 15.1 to 30.0, or >30.1 gram per day. Relative risks (RR) of Ml
were estimated with the lowest tertile groups as the references. Tests for trends were
assessed by assigning each participant the median value of vitamins for the categories
and modeling these values as continuous variables. All P values are two-sided.
Results
During an average of 5.5 years of follow-up, 74 incident cases of myocardial
infarction were ascertained. The age-adjusted mean dietary intakes and risk factors
according to disease status are presented in Table 1. The mean intakes of vitamins E,
C and carotenoid among incident Ml cases were all lower than those among Ml free
subjects, but none of them was statistically significant (p>0.05). Instead, the Ml
cases were found to have a significantly higher intakes of saturated fatty acids, total
energy and lower daily alcohol consumption than Ml free subjects (p<0.05).
The age-adjusted and multivariate RR's of Ml according to tertile groups of
vitamin E intake are shown in Table 2. As compared with men in the lowest tertile
group, men in the top tertile group had an age-adjusted RR estimate of 1.11 (95%
Cl =0.64-1.94). Further adjustments for other risk factors had little effect on the
results, with RR estimates all above 1.0, and all P values for trend were nonsignificant.
48
Similar results were found for dietary intake of vitamin C. The age-adjusted RR
of Ml for men in the top tertile vitamin C group was 0.94 (95% Cl =0.55-1.61).
Further adjustments for other risk factors yielded RR estimates above 1.0, but P values
for trend were not significant (Table 3).
The age-adjusted RR of Ml for men in the top tertile carotenoid group was 0.70
(95% Cl =0.40-1.21). Further adjustment for other risk factors reduced the inverse
association between carotenoid intake and risk of Ml, and none of the RR estimates
from the multivariate models were significant (Table 4).
To estimate association between overall dietary intake of antioxidant vitamins
and risk of Ml, the cohort was then categorized into tertile groups for the sum of
vitamin E, vitamin C and carotenoid. The RR's of Ml for the composite intake of
antioxidant vitamins from different multivariate Proportional Hazard models were
presented in Table 5. No association of the sum of antioxidant vitamin intake with risk
of myocardial infarction was found in multivariate models. The estimate from the fully
adjusted model yielded a RR of 1.35 (95% Cl = 0.74-2.43). Analyses of the interaction
between intakes of antioxidant vitamins, vitamins and smoking, vitamins and serum
cholesterol did not reveal any significant results (data not shown).
The associations between the major food determinants of antioxidant vitamins
and Ml were also estimated, including intake of cereal, grain, margarine, sum of
vegetables, potato, fruits, milk products, meat, eggs, linoleic fatty acids, and fiber.
Foods were added into the multivariate model one at a time. Both models with and
without adjustment for antioxidant vitamins were estimated. None of the foods
analyzed was found to be associated with risk of Ml (data not shown).
49
Discussion
In the present study, we examined the relationship between dietary intake of
vitamin E, vitamin C and carotenoid and subsequent incidence of Ml in a cohort of
middle-aged Finnish men living in Eastern Finland. Results failed to reveal any
significant association between antioxidant vitamins or their major food determinants
and risk of Ml. Our results did not support the antioxidant hypothesis of CHD.
Evidence from an increasing number of studies supports the notion that
antioxidant vitamins may play a protective role in the prevention of coronary heart
disease. Some indirect evidence in favor of the antioxidant hypothesis has been
derived from ecologic correlation analyses and epidemiologic studies on an individual
level, demonstrating an inverse association between the consumption of fruits and
vegetables and the risk of cardiovascular disease12,13,17,23-2S. More direct evidence
comes from studies based on plasma or serum concentrations of antioxidant vitamins.
An inverse correlation between serum levels of antioxidant vitamins and coronary heart
mortality has also been demonstrated in ecologic studies29,30. Prospective and cross-
sectional studies based on the blood levels of antioxidant vitamins presented to date
largely support an inverse association, although not all studies unambiguously support
the antioxidant hypothesis31-33. In a separate analysis of the relationship between
plasma levels of a-tocopherol and ascorbic acid and risk of Ml in the current cohort,
we also failed to find an inverse association between plasma a-tocopherol. An inverse
association between plasma ascorbic acid and Ml was found after multivariate
correction for measurement errors, but the relative risk estimate failed to reach
significance, presumably due to the relatively small sample size in our study (see
Chapter 3).
50
Studies investigating dietary intake of antioxidant vitamins have yielded
inconsistent results, with many studies reporting a null association between dietary
antioxidant intake and risk of coronary heart disease. Two prospective studies in the
U.S., the Nurse's Health Study and the Health Professional Study, had extensively
investigated the relationship between dietary intake of and supplement use of
antioxidant vitamins and subsequent incidence of CHD in tw o large cohorts of women
and men respectively14,15. Both studies reported an inverse association between total
vitamin E intake (supplements plus dietary intake) or vitamin E supplement use alone
and risk of CHD, but failed to reveal an inverse association for dietary vitamin E intake
alone. In both studies, total vitamin C intake or vitamin C supplement use alone or
dietary vitamin C intake failed to show any association with CHD. Carotenoid intake
was found only to be inversely associated with CHD among current and former
smokers, but not among non-smokers. These results raise the concern that subjects
took vitamin supplements might differ substantially from those who did not take
supplements. Indeed, in both studies, subjects with a higher intake of vitamin E or
vitamin C have a relatively healthier risk profile, with a lower prevalence of cigarette
smoking and a higher level of physical activity. Although these variables were
controlled for in multivariate analyses, residual confounding could still exist to some
extent. More seriously, serum cholesterol was not available and was not controlled for
in both studies. Since vitamin E is a fat-soluble antioxidant and has been shown to be
highly correlated with serum cholesterol concentration, failure to adjust for serum
cholesterol could provide a serious source of confounding of the observed
association34. Nevertheless, results from these two large cohort studies offer the most
compelling evidence to date supporting the antioxidant hypothesis.
51
Findings from these two studies also indicate that a threshold effect may exist
for vitamin E intake. For both men and women, persons with a daily total vitamin E
intake over 250 IU did not show any more beneficial effect against CHD as compared
with those who took 100-250 IU vitamin E per day. These results indicate that the
maximal anti-atherogenic effect of vitamin E may be achieved at a daily dose of 100-
250 IU. This level of vitamin E apparently cannot be reached by conventional diet.
Cross-cultural comparison of mean serum levels of vitamin E, vitamin C and ^-carotene
and CHD mortality in different regions in European also support this notion30. In
contrast to the U.S. population with a high prevalence of vitamin supplements use,
supplement use is rare among Eastern Finnish men. The mean daily dietary intake of
vitamin E in our samples was only 9.4 g, about 7.0 IU per day with the highest intake
of 31 IU (42 g) per day. Therefore, the daily dietary intake of vitamin E in our sample
w as apparently far lower than the critical value of 100-250 IU as suggested by the
two U.S. studies. This relatively low and homogenous dietary intake of vitamin E may
offer one possible explanation for the lack of association in our study.
The single one study reported so far that explicitly demonstrated an inverse
association between dietary intake of vitamins and risk of CHD was reported from
Finland, the Mobile Clinic of the Social Insurance Institution Study35. A higher dietary
intake of vitamin E was found to be associated with a lower risk of coronary death in
both men and women, while an inverse association for dietary vitamin C intake was
found only in women. No association between dietary carotenoid intake and coronary
mortality was found in either men or women. The apparent discrepancy between this
study and our analysis may be partly attributed to the relatively smaller sample size
and shorter duration of our study. The mean duration of follow-up in the cited study
52
was 14 years, with 186 documented death from CHD among 2,748 men at risk,
representing 6.8 percent of CHD mortality. In contrast, only 74 incident Ml cases were
documented in our sample, with an incidence rate of 4.6 percent. Another factor that
may explain the discrepancy was the different end points used in our study. Instead
of coronary mortality, incidence of Ml was used in our analyses. It is likely that due to
the long latency and chronicity nature of cardiovascular disease and the short duration
of follow-up in our study, dietary antioxidant intakes determined at baseline may not
represent the true long-term dietary antioxidant intake of the cohort members. Twelve
month reproducibility correlation based on a random sample of 50 men in our study
was 0.56 for vitamin E, 0.62 for vitamin C and only 0.14 for carotenoid. Intra
individual variations of dietary intake could have attenuated the true relationship
between antioxidant vitamins and Ml. Use of different methodology in dietary
assessment might also partly contribute to the discrepancy. In the Mobile Clinic of the
Social Insurance Institution Study, dietary intakes were assessed from a dietary history
interview covering the total habitual diet of the subjects during the previous 12
months. A 4-8 month reproducibility study revealed intra-individual correlations of
0.78, 0.53, and 0.56 for vitamin E, vitamin C, and carotene, respectively. There are
considerable seasonal differences in food consumption in Eastern Finland. Goods
imported and exported affect availability of foods in this country and may further
contribute to the variation of dietary intake of essential nutrients in this population. It
is conceivable that 4-day dietary records may have underestimated the seasonal and
day-to-day variation of nutrient intake in our samples. Food frequency questionnaire
may be a superior method to dietary record in the context of Finnish diet. The lower
reproducibility correlations of antioxidant intakes in our sample as compared to those
53
from the Mobile Clinic Study, especially that of carotenoid, support this argument.
In vitro studies have demonstrated synergistic effects among vitamin E, vitamin
C and /?-carotene36. However, to date, no epidemiologic studies have reported
synergism among these antioxidant vitamins. In our study, analyses failed to find any
interactions among the antioxidant vitamins and other potential effector modifiers. It
is likely that the biological interactions among those antioxidants demonstrated under
experimental conditions may have little relevance to their in vivo activities or that the
synergistic effects may be so small that only studies of very large sample size may
possibly reveal any meaningful interactions in the population.
A recently published paper based on the current cohort data reported that
environmental contamination of fresh water fish by heavy metal mercury might be
responsible for the observed direct association between fish consumption and risk of
Ml in this cohort37. Although adjustment for dietary or hair mercury level did not
materially alter our results, this report raised the possibility that other unknown
environmental contaminants in Eastern Finland may have confounded the association
between antioxidant vitamin intake and risk of Ml through some unknown
mechanisms. Unfortunately, data available at this point limits efforts to define potential
confounders.
Differences in coronary mortality rates the world over could largely be explained
by differences in dietary cholesterol and saturated fat intake12. However, the
discrepancy between France and Finland can not be explained by such differences,
although the two countries have similar intakes of cholesterol and saturated fat.
Dietary intakes other than cholesterol and saturated fatty acid, such as vegetables and
fiber, were suggested to partly account for this paradox12. A number of cohort studies
54
revealed that a high consumption of vegetables and fruits is associated with a low risk
of coronary heart disease25'28,38. The protective effect of high fruit and vegetable
consumption is commonly believed to be due to the high antioxidant vitamin contents
in these foods. However, analyses of the major food sources of antioxidant vitamins
in the present study failed to reveal any association with Ml.
Although no significant association was found between dietary intake of
antioxidant vitamins and risk of Ml in the present analysis, it cannot be excluded that
dietary antioxidants may have a potential role in the prevention of CHD. The failure to
find an inverse association in our study could at least be partly attributed to
methodological limitations and relatively low intake of antioxidants among our study
subjects. Earlier reports based on this study data revealed that high dietary and blood
iron, copper and mercury are associated with an excess risk of Ml, presumably due to
their lipid oxidation promoting effects, which offer indirect evidence supporting the
antioxidant hypothesis18'37,39,40. Further studies directed at populations with more
diverse dietary intake of antioxidant vitamins are needed to fully elucidate the potential
role of antioxidant vitamins in the prevention of CHD. An intervention trial of
antioxidant vitamins with the carotid artery intima-media wall thickness as the end
point is being conducted in Eastern Finland, involving some subjects from the current
cohort. The results from this trial may provide more insightful information about the
role of dietary antioxidants in the prevention of CHD. Based on our current knowledge
of antioxidants and evidence from other studies, it is advisable at this point that
moderate daily use of antioxidant supplements may help to reduce the high risk of CHD
in this population.
5 5
Reference
1. Steinberg D, Parthasarathy S, Carew TE, Khoo JC, Witztum JL. Beyond
cholesterol: modification of low-density lipoprotein that increases its
atherogenecity. N Engl J M ed 1989;320:915-24.
2. Steinberg D, Witztum J. Lipoproteins and atherogenesis: current concepts.
JAMA 1990;264:3047-52.
3. Steinbrecher UP, Zhang HF, Lougheed M. Role of oxidatively modified LDL in
atherosclerosis. Free Radio Bio Med 1990;9(2):155-168.
4. Morel DW, Hessler JR, Chisolm GM. Low density lipoprotein cytotoxicity
induced by free radical peroxidation of lipid. J Lipid Res 1983;24:1070-6.
5. Hessler JR, Morel DW, Lewis LJ, Chisolm GM. Lipoprotein oxidation and
lipoprotein-induced cytotoxicity. Arteriosclerosis 1983;3:215-22.
6. Quinn MT, Parthasarathy S, Fong LG, Steinberg D. Oxidatively modified low
density lipoproteins: a potential role in recruitment and retention of
monocyte/macrophage during atherogenesis. Proc Nat/ Acad Sci USA
1987;84:2995-8.
7. Jialal I, Grundy SM. Combined supplementation with alpha-tocopherol, ascorbic
and beta-carotene inhibits LDL oxidation. Circulation 1992;86(4):l-462.
8. Esterbauer H, Puhl H, Dieber RM, Waeg G, Rabl H. Effect of antioxidants on
oxidative modification of LDL. Ann Med 1991 ;23(5):573-81.
9. Smith TL, Kummerow FA. Effect of dietary vitamin E on plasma lipids and
atherogenesis in restricted ovulatory chickens. Atherosclerosis 1989;75:105-
109.
10. Wojcicki J. Rozewicka L, Barcew-Wiszniewska B. Effect of selenium and
vitamin E on the development of experimental atherosclerosis in rabbits.
Atherosclerosis 1991;87:9-19.
11. Verlangieri AJ, Bush M. Prevention and regression of atherosclerosis by alpha-
tocopherol. J Am Coll Nutr 1992; 11 (2): 131-138.
12. Artaud-Wild S, Connor SL, Sexton G, Connor WE. Differences in coronary
mortality can be explained by differences in cholesterol and saturated fat
intakes in 40 countries but not in France and Finland. Circulation
1993;88:2771-2779.
13. Armstrong BK, Mann JL, Adelstein AM, Eskin F. Commodity consumption and
ischemic heart disease mortality, with special reference to dietary practices. J
Chronic Dis 1975;36:673-677.
56
14. Rimm EB, Stampfer MJ, Ascherio A, Giovannucci E, Colditz GA, Willett WC.
Vitamin E consumption and the risk of coronary heart disease in men. N Engl
J M ed 1993;328:1450-6.
15. Stampfer MJ, Hennekens CH, Manson JE, Colditz GA, Rosner B, Willett WC.
Vitamin E consumption and the risk of coronary heart disease in women. NEngl
J M ed 1993;328:1444-9.
16. Salonen JT. Is there a continuing need for longitudinal epidemiologic research?
The Kuopio Ischemic Heart Disease Risk Factor Study. Ann Clin Res
1988;20:46-50.
17. Keys A. Seven Countries: A Multivariate Analysis of Death and Coronary Heart
Disease. Cambridge/London, Harvard University Press, 1980
18. Salonen JT, Nyyssonen K, Korpela H, Tuomilehto J, Seppanen R, Salonen R.
High stored iron levels are associated with excess risk of myocardial infarction
in Eastern Finnish Men. Circulation 1992;86:803-811.
19. Ihanainen M, Salonen R, Seppanen R, Salonen JT. Nutrition data collection in
the Kuopio Ischemic Heart Disease Risk Factor Study: nutrient intake of middle-
aged Eastern Finnish Men. Nutr Res. 1989;9:597-604.
20. Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Survey Methods.
Geneva, World Health Organization, 1982.
21. Knuts LR, Rastas M, Seppanen K, Rauramaa R, Karvetti RL, Aro A. Nutrica: A
computer program for calculation of food and nutrient intake, in Third Eurofoods
Meeting, Warsaw, May 24-27, 1987.
22. Tuomillehto J, Arstila M, Kaarsalo E, Kankaanpaa J, Ketonen M, Kuulasmaa K,
Lehto S, Miettinen H, Mustaniemi H et al. Acute myocardial infarction in
Finland: baseline data from the FINMONICA AMI registry in 1983-1985. Eur
Heart J 1992;13:577-587.
23. Verlangieri AJ, Kapeghian JC, El-Dean S, Bush M. Fruit and vegetable
consumption and cardiovasculardisease mortality. Med Hypotheses 1985; 16:7-
15.
24. Ginter E. Decline of coronary mortality in the United States and vitamin C. Am
J Clin Nutr 1979;32:511-51 2.
25. Morris JN, Marr JW, Clayton DG. Diet and heart: a postscript. Br Med J
1977;2:1307-14.
26. Garcia-palmieri MR, Sorlie P, Tillotson J. Relationship of dietary intake to
subsequent coronary heart disease incidence. The Puerto Rico Heart Health
Program. Am J Clin Nutr 1980;33:1818-27.
5 7
27. Kromhout D, De Lezenne Coulander C. Diet, prevalence and 10-year mortality
from coronary heart disease in 871 middle-aged men: The Zutphen Study. Am
J Epidemiol 1984; 119:733-41.
28. Kushi LH, Lew RA, Stare FJ. Diet and 20-year mortality from coronary heat
disease: The Ireland-Boston Diet-Heart Study. N Engl J Med 1985;312:811-18.
29. Gey KF, Brubacher GB, Stahelin HB. Plasma levels of antioxidant vitamins in
relation to ischemic heart disease and cancer. Am J C/In Nutr 1987;45:1368-
1377.
30. Gey KF, Stahelin HB, Eichholzer M. Poor plasma status of carotene and vitamin
C is associated with higher mortality from ischemic heart disease and stroke:
Basel Prospective Study. Clin invest 1993;71:3-10.
31. Riemersma RA, Wood DA, Macintyre CCA, Elton RA, Gey KF, Olive MF. Risk
of angina pectoris and plasma concentrations of vitamins A, C, and E and
carotene. Lancet 1991;337:1-5.
32. Ramirez J, Flowers NC. Leukocyte ascorbic acid and its relationship to coronary
artery disease in man. Am J Clin Nutr 1980;33:2079-87.
33. Hense HW, Stender M, Bors W, Keil U. Lack of an association between serum
vitamin E and myocardial infarction in population with high vitamin E levels.
Atherosclerosis 1993;103:21-28.
34. Halliwell B. Current status review. Free radicals, reactive oxygen species and
human disease: a critical evaluation with special reference to atherosclerosis.
B rJ Exp Pathol 1989;70:737-57.
35. Knekt P, Reunanen A, Jarrinen R, Seppanen R, Heliovaara M, Aromaa A.
Antioxidant vitamin intake and coronary mortality in a longitudinal population
study. Am J Epidemiol 1994; 139:1180-9.
36. Frei B, Stocker R, Ames BN. Antioxidant defenses and lipid peroxidation in
human blood plasma. Proc Natl Acad Sci USA 1988;85:9748-52.
37. Salonen JT, Seppanen R, Nyyssonen K, Korpela H, Kauhanen J, Tuomilehto J,
Esterbauer H, Tatzer F, Salonen R. Intake of mercury from fish, lipid
peroxidation, and the risk of myocardial infarction and coronary, cardiovascular,
and any death in Eastern Finnish Men. Circulation 1995;91:645-655.
38. Anderson JW, Chen W, Sieling B. Hypolipidemic effect of high carbohydrate,
high fiber diet. Metabolism 1980;29:551-558.
39. Salonen JT, Salonen R, Korpela H, Suntioinen S, Tuomilehto J. Serum copper
and the risk of acute myocardial infarction: a prospective population study in
men in Eastern Finland. Am J Epidemiol 1991;134:268-276.
58
40. Salonen JT. The role of iron as cardiovascular risk factor. Curr Opin Lipid
1993;42:277-282.
59
Table 1. Descriptive characteristics of Myocardial Infarction Cases and Non-cases
The Kuopio Ischemic Heart Disease Risk Factor Study
Variables
MI Cases In =741
mean (SD)
Non-cases (n = 1.535)
mean (SD)
Age (years) 53.9 (3.0) 52.2 (5.2)
Vitamin E intake (mg/d) 9.1 (3.4) 9.4 (4.1)
Vitamin C intake (mg/d) 92.8 (47.5) 97.1 (57.5)
Carotenoid intake (mg/d) 1.9 (1.7) 2.7 (3.4)
Saturated fatty acid intake (g/d) 60.8 (21.7) 56.1 (20.0)
Total energy intake (kcal/d) 2717.5 (612.8) 2595.4 (624.6)
Alcohol consumption (g/d) 7.3 (14.6) 9.6 (18.8)
Serum cholesterol (mmol/1) 6.3 (1.1) 5.9 (1.0)
High density lipoprotein (mmol/1) 1.2 (0.3) 1.3 (0.3)
Systolic blood pressure (mmHg) 140.6 (17.3) 133.8 (16.3)
Body mass index (kg/m3 ) 27.2 (3.4) 26.7 (3.4)
Maximal oxygen uptake (ml/kg/min) 28.0 (6.5) 32.3 (7.6)
Physical activity (MET h/y) 443.8 (686.3) 635.5 (760.2)
Socioeconomic status 12.3 (5.3) 10.7 (4.7)
Smoking status (%) 54.5 28.6
exercise ECG abnormality (%) 32.5 17.6
Family history of IHD (%) 55.4 46.6
Living in rural area (%) 42.9 26.0
Note: All means for continous variables were age-adjusted.
60
T abic 2. Relative Risk Estim ates of M yocardial Infarction, According to T ertile G roups for D ietary V itam in E Intake
'(lie Kuopio Iscliemic H eart Disease Risk Factor Study (11=1,609)
Tertile G roup
V ariables 1 2 3 P fo r T rend
Vitamin E-median intake (mg/d) 5.75 8.65 13.86
Number of Myocardial Infarction 28 23 23
Number at Risk 536 539 534
Relative risk
Base Model* 1.0 1.02 1.11 0.79
95% Cl - (0.58-1.77) (0.64-1.94)
Multivariate Model lb 1.0 1.04 1.33 0 4 8
95% Cl - (0.58-1.85) (0.73-2.43)
Multivariate Model 2C 1.0 1.18 1.46 0.43
95% Cl - (0.66-2.13) (0.78-2.72)
Multivariate Model 3J 1.0 1.13 1.41 0.41
95% Cl - (0.63-2.04) (0.75-2.64)
Note:
a. The base model included baseline age and years ol baseline examination as covariates.
b. Model 1 included baseline age, baseline examination years, setum cholesterol, UDL, systolic blood pressure, smoking status, BM1,
exercise liCG abnormality and total energy intake as covariates.
c. Model 2: with further adjustment for maximal oxygen uptake, saturated fatty acid intake, living areas, socioecunomic status, alcohul
drinking, family history of ischemic heart disease, and leisure time physical activity.
d. Model 3: dietary vitamin C and carotenoid were added into model 2.
Table 3. Relative Risk Intim ates of M yocardial Infarction, According’ (o T ertile G roups for D ietary V itam in C Intake
'th e Kuopio Ischemic H eart Disease Risk Factor Study ( n = 1,609)
Tertile G roup
V ariables 1 2 3 P fo r T read
Vitamin C-median intake (mg/d) 47.1 86.2 157.6
Number of Myocardial Infarction 25 20 29
Number at Risk 536 538 535
Relative risk
Base Model* 1.0 0.73 0.94 0.67
95% Cl - (0.41-1.32) (0.55-1.61)
Multivariate Model lb 1.0 0.87 1.19 0.63
95% Cl - (0.47-1.59) (0.67-2.12)
Multivariate Model 2C 1.0 0.88 1.32 0.71
95% Cl - (0.48-1.62) (0.73-2.39)
Multivariate Model 3d 1.0 0.90 1.33 0.64
95% Cl - (0.48-1.69) (0.72-2.47)
Note:
a. Base model ineluded baseline age and baseline examination years as covariates.
b. Model I included baseline age, baseline examination yeats, seiuni cltolesletol, 111)1., systolic blood pressure, smoking status, OMI,
exercise ECG abnormality and total energy intake us covariates.
c. Model 2: with further adjustment fur maximal oxygen uptake, saturated fatty acid intake, living areas, socioeconomic status, alcohol
drinking, family history of ischemmic heart disease, and leisure lime physical activity.
d. Model 3: dietary vitamin E and carotenoid were added into model 2.
Table 4. Relative Risk Estim ates of M yocardial Infarction, According to Tertile G roups for D ietary C aroteuoid intake
T he Kuopio Ischemic H eart Disease Risk Factor Study (11= 1,609)
T ertile G roup
V ariables I 2 3 P for T ren d
Caritenoids-median intake (mg/d) 0.62 1.71 5.62
Number of Myocardial Infarction 32 21 21
Number at Risk 536 538 535
Relative risk
Base Model* 1.0 0.60 0.70 0.22
95% Cl - (0.35-1.04) (0.40-1.21)
Multivariate Model lb 1.0 0.70 0.95 0.43
95% Cl - (0.40-1.23) (0.54-1.67)
Multivariate Model T 1.0 0.71 0.98 0.47
95% Cl - (0.41-1.25) (0.55-1.74)
Multivariate Model 3J 1.0 0.70 0.90 0.38
95% Cl - (0.39-1.23) (0.49-1.65)
Note:
a.
b.
d.
Uase model included baseline age and baseline examination yeais as envariates.
Model I included;baseline age, baseline examination yeais, seium cbolesteiol, IIDL, systolic blood pressure, smoking status, DM1,
exercise BCG abnormality and total energy intake us envariates.
Model 2: with further adjustment for maximal oxygen uptake, saturated fatty acid intake, living ureas, socioeconomic status, alcohul
drinking, family history o f ischetnic heart disease, and leisure time physical activity.
Model 3: dietary vitamin B and vitamin C were added into model 2.
Table 5. Relative Risk Estim ates of M yocardial Infarction, A ccording to Tertile G roups of Sum of Dietary V itam in E
Vitam in C and C arotcnoid Intake: The Kuopio Ischemic H eart Disease Risk Factor Study (n = 1,609)
Tertile G roup
Variables 1 2 3 P fo r T ren d
Sum of Vitamins-median intake 57.1 98.3 171.6
(mg/d)
Number of Myocardial Infarction 26 19 29
Number at Risk 536 537 536
Relative risk
Base Model* 1.0 0.68 0.93 0.51
95% Cl - (0.37-1.22) (0.55-1.58)
Multivariate Model lb 1.0 0.80 1.18 0.58
95% Cl - (0.43-1.48) (0.66-2.09)
Multivariate Model 2 ' 1.0 0.85 1.35 0.53
95% Cl ” (0.46-1.58) (0.74-2.43)
Note:
a. Bast: model included baseline age and baseline examination yeais as cnvariales.
b. Mode) 1 included baseline age, baseline examination yeais, seium clmlcsleiul, HD!., systolic blood pressure, smoking status, BMI,
exercise liCG abnormality and total energy intake as envariates.
c. Model 2: with further adjustment fur maximal oxygen uptake, saturated fatty acid intake, living areas, socioeconomic status, alcohol
drinking, family history o f ischemic heat I disease, and leisure lime physical activity.
Chapter 3. Association of P/asma Alpha-Tocopherol and Ascorbic Acid with Myocardial
Infarction: Multivariate Correction for Bias due to Measurement Errors
Abstract
Objective: To assess the relationship of plasma alpha-tocopherol and ascorbic acid and
subsequent risk of myocardial infarction (Ml).
Design: Longitudinal follow-up study, initial examination 1984-1989, follow-up until
December, 1992.
Participants: 1,547 middle-aged men free of coronary heart disease and cancer living
in Eastern Finland.
Results: After an average of 5.5 years of follow-up, 74 incident myocardial infarction
(Ml) cases were documented. In multivariate Proportional Hazard regression analyses,
no association was found between plasma alpha-tocopherol and Ml among nondrinker
(< 0 .5 g/day), with a relative risk (RR) estimate of 1.08 (0.84-1.40), while a marginal
inverse association was found for plasma ascorbic acid, with a RR estimate of 0.79
(0.59-0.13). In contrast, positive associations between plasma alpha-tocopherol and
ascorbic acid and Ml were found among alcohol drinkers (> 0 .5 g/day), the RR
estimates were 1.50 (0.98-2.28), 1.43 (0.96-2.12) for plasma alpha-tocopherol and
ascorbic acid, respectively. Further, multivariate correction for measurement errors
(ME) yielded a RR of 1.07 (0.79-1.45) for plasma alpha-tocopherol, and 0.55 (0.10-
3.06) for ascorbic acid.
Conclusion: Our results suggest that alcohol consumption is an effect modifier of the
associations between plasma alpha-tocopherol and ascorbic acid and risk of Ml. The
inverse association between ascorbic acid and Ml is seriously attenuated by ME, while
a spuriously direct association between plasma alpha-tocopherol was created due to
bias of ME.
65
Key words: Alpha-tocopherol, Ascorbic Acid, Measurement Error, Effect Modification,
Myocardial Infarction.
Introduction
There is growing interest in the possible role of free radicals in the development
of atherosclerosis, and its sequela coronary heart disease (CHD). Oxidative
modification of low density lipoprotein (LDL) has been postulated as one of the major
initial steps in atherosclerosis.1,2 A variety of antioxidant defense systems in the
human body are able to eliminate prooxidants and scavenge free radicals. The most
readily available naturally occurring antioxidants in food are tocopherol and ascorbic
acid, commonly referred to as vitamin E and vitamin C, respectively. Vitamin E is the
major antioxidant in the lipid phase and protects polyunsaturated fatty acids from
peroxidation. Vitamin C acts in the water-soluble compartment and has a sparing effect
on vitamin E.3
Findings in prevalent cases of angina pectoris and results from ecological
studies relating CHD mortality and average serum vitamin E and vitamin C levels across
populations appear to support the hypothesis that vitamin E, of all the antioxidant
systems, has the strongest protective effect on CHD.4,5 Study results are, however,
still conflicting. No association with vitamin E was found in studies of men with
prevalent ischemic heart disease in Finland and across four populations with marked
differences in CHD mortality.6,7 Similarly, some nested case-control studies were
unable to confirm protective effects of high serum vitamin E concentration against
death due to CHD.8,9
A common problem in epidemiologic studies is that measured risk factors are
66
only surrogate indicators of the individual's true level of exposure, and frequently
flawed with measurement errors (ME). A well known consequence of the error in
measurement is the distortion, frequently attenuation, of the underlying relationship
between the risk factors and disease.10,11 Plasma alpha-tocopherol and ascorbic acid
concentrations are subject to intra-individual fluctuations as well as to laboratory
errors. It is thus likely that the associations of plasma alpha-tocopherol and ascorbic
acid with CHD observed in previously reported epidemiologic studies are biased
without appropriate control over ME, which may partly explain the discrepancies of
results from different studies.
The aim of the present study is to present an empirical example of a method for
correcting ME bias using multivariate Cox regression model. The method is applied to
data from a cohort of middle-aged Eastern Finnish men to investigate the relationship
between plasma alpha-tocopherol and ascorbic acid with risk of subsequent acute
myocardial infarction (Ml). This method is an extension of a logistic procedure
described by Rosner et al.12 We hypothesized that the crude (without correction for
ME) estimates of the associations of plasma alpha-tocopherol and ascorbic acid with
Ml are substantially biased, and ME is a likely explanation for the excess risk of Ml
associated with high concentrations of plasma alpha-tocopherol in this cohort.
Materials and Methods
The Kuopio Ischemic Heart Disease Risk Factor Study is a population-based
study designed to investigate previously unestablished risk factors for CHD and extra
coronary atherosclerosis in eastern Finnish men, a population with one of the highest
incidences and mortality rates of CHD. The details of the study design and
67
implementation has been described in detailed elsewhere13,14. Briefly, baseline exams
were carried out between March 1984 and December 1989. The study sample
included 3,235 men aged 42, 48, 54, or 60 years at the baseline examination. Of
these men, 2,682 (82.9 percent) participated. Men with either prevalent CHD (n = 667)
or cancer (n = 46) or having a total energy intake less than 500 kcal or over 5,000 kcal
(n = 12) were excluded from the present analysis. Prevalent ischemic heart disease was
defined as either a history of Ml or angina pectoris, positive angina pectoris on effort
in the London School of Hygiene interview15, or use of nitroglycerin tablets once a
week or more frequently. Of the remaining 1,957 men, complete data on the interested
variables were available for 1,547 men, and was used for subsequent analyses.
A self-administered questionnaire concerning socio-demographic factors,
individual and family medical history, smoking, and alcohol consumption was
performed on all cohort members. Repeat interviews to obtain medical history were
conducted by a physician. A subject was defined a smoker if had ever smoked on a
regular basis and smoked cigarettes, cigar, or pipe within the past 30 days. The life
long exposure to smoking ("cigarette year") was estimated as the product of years
smoked and the number of tobacco products smoked daily at the time of examination.
Years smoked were defined as the sum of years of smoking regardless of when
smoking had started, whether the subject had stopped smoking, and whether smoking
had occurred continuously or during several periods. The consumption of alcohol in the
previous 12 months was assessed with a quantity-frequency method by using the
Nordic Alcohol Consumption Inventory, which contains 15 items. The socioeconomic
status was measured with the summary index that combined measures of income,
education, occupation, occupational prestige, material standard of living, and housing
68
condition. Alcohol consumption was also assessed at baseline exam with an instructed
and interview-checked 4-day dietary recording by household measures.
Resting blood pressure was measured on the first examination day by one nurse
with a random-zero mercury sphygmomanometer. The measuring protocol included
three measurements in supine, and one in standing, and two in sitting position. The
mean of all six systolic blood pressure values was used in the present analysis.
The respiratory gas exchange was measured breath-by-breath with an MGC
2001 system (Medical Graphics Corp., Minneapolis, MN) during a symptom-limited
exercise test. The testing protocol comprised a linear increase of work load by 20
W/min. Highest oxygen uptake during the test was defined as V02max. Exercise ECGs
were coded manually by one cardiologist. The criteria for ischemia were 1) ischemic
ECG defined as horizontal or downsloping ST depression > 0 .5 mm or upsloping ST
depression > 1 .Omm; 2)typical angina pectoris pain leading to discontinuation of
exercise; 3) maximal heart rate during exercise < 130 beats per minute.
Venous blood was taken from each subject after having abstained from alcohol
for 3 days, smoking for 12 hours, and fasting for 8-12 hours. The main lipoprotein
were separated from fresh serum samples using ultracentrifugation and precipitation.
The HDL2and HDL3 subfractions were separated using ultracentrifugation at 108,000g
for 62 hours against a density of 1.125 g/cm3. Total serum cholesterol was measured
enzymatically. Plasma alpha-tocopherol and ascorbic acid were measured by a HPLC
method.
As a part of the multinational MONICA project, an acute myocardial infarction
(AMI) registry was established in the province of Kuopio in 1982. The registry collects
detailed diagnostic information of all heart attacks in the population in a prospective
69
manner. Heart attacks were classified as definite AMI, possible AMI, no AMI, or
insufficient data according to explicitly defined, uniform diagnostic criteria described
earlier elsewhere16. The coverage of the AMI registry was checked against the national
computerized death certificate registry. Diagnostic information and the date of onset
of all heart attacks in the present study cohort was obtained by record linkage based
on the uniform Finnish personal identification code.
Statistical Analysis
Baseline characteristics of incident Ml cases and Ml free subjects were first
computed. Correlations among repeated measures of risk factors were also presented.
For standardization, plasma alpha-tocopherol to serum total cholesterol ratio was used
throughout the current analyses.
Crude associations of plasma alpha-tocopherol and ascorbic acid with Ml were
first estimated using multivariate Proportional Hazards models with adjustment for
baseline age, serum total cholesterol, high density lipoprotein (HDL), systolic blood
pressure, body mass index (BMI), smoking status, alcohol consumption, maximal
oxygen uptake, family history of ischemic heart disease, ECG abnormality in exercise,
socioeconomic status, living areas and baseline exam years.
To estimate the magnitudes of measurement errors in risk factors, a
measurement model was assessed based on repeated measurements in a sub-sample
of 91 subjects. The inter- and intra-individual (errors) variances, and validity of
measurements were estimated.
To correct for the effects of MEs in plasma alpha-tocopherol and ascorbic acid,
we employed a multivariate method described by Rosner et al12 and a matrix algorithm
70
developed by Dwyer and Sun (University of Southern California, unpublished). The
dwyer-Sun program is a generalization of the Rosner's method that applies to different
forms of regressions, including Cox Proportional Hazards model. In the current
analysis, repeated measurements of plasma alpha-tocopherol, ascorbic acid, serum
total cholesterol, HDL, systolic blood pressure, BMI (determined at baseline and in
1990) on 91 subjects were used for correction for MEs. Alcohol consumption assessed
from 4-day dietary records and a questionnaire, both administered at baseline, were
used. Smoking status, maximal oxygen uptake, family history of IHD, exercise ECG
abnormality, socioeconomic status, living areas and baseline exam years were also
included in the multivariate models, but no attempt was made to correct for possible
errors in these variables.
The method proposed by Rosner and associates for ME correction involves four
steps. First, uncorrected or crude Cox regression coefficients (/?) are estimated using
the main study data on the whole cohort (baseline exam). Second, a multivariate linear
regression model is performed on the validation sample, with the baseline
measurements being regressed on the repeated measurements. Third, the corrected
regression estimates (/?*)are obtained by weighing the uncorrected Cox regression
coefficients by the variance-covariance matrix relating the main study variables to the
validation study variables. Finally, the variances of the corrected /? coefficients are
obtained using the multivariate delta method, and asymptotic confidence limits for the
corrected regression coefficients are derived as/?* ± Z 1.0/2[var(/?*)]1/2. Var(/?*) includes
a component attributable to the estimation of the repeated correlations, and is obtained
by expanding the risk function in a truncated Taylor series.17 Comparison of the
uncorrected and the corresponding corrected /? coefficients and associated relative risk
71
(RR) estimates indicates the biasing effect of ME and is presented as percentage of
changes.
Results
Table 1 presents descriptive statistics of all the predictor variables for the 74
incident Ml cases and Ml free subjects. The mean plasma alpha-tocopherol-to-serum
cholesterol ratio was slightly higher for the Ml cases than that of Ml free subjects, but
the difference was not significant (P>0.05). In contrast, a significantly lower mean
plasma ascorbic acid concentration was found for Ml cases (P<0.05). The Ml cases
also had a lower level of HDL and maximal oxygen uptake, and a higher level of serum
total cholesterol, systolic blood pressure, and BMI. There were higher proportions of
smokers, exercise ECG abnormality at baseline, living in rural areas, and family history
of ischemic heart disease among Ml cases.
The age-adjusted correlations among repeated measurements of plasma alpha-
tocopherol-to-serum cholesterol ratio, plasma ascorbic acid and other known risk
factors for CHD on the validation subsample is presented in Table 2. A relatively low
correlation was found for plasma alpha-tocopherol-to-serum cholesterol ratio, plasma
ascorbic acid and serum total cholesterol; while a high correlation was found for HDL,
systolic blood pressure and BMI. Table 2 also presents the intercorrelations among
baseline measurements and repeated measurements of risk factors. Except for the
correlation between baseline serum total cholesterol and repeated measurement of HDL
(r = 0.34), all other correlations were small.
The uncorrected multivariate RR estimates for alpha-tocopherol and ascorbic
acid are presented in Table 3. With adjustment for age and baseline exam years, a
72
marginally direct association was found between plasma alpha-tocopherol and risk of
Ml (P = 0.08); while an inverse association was found for plasma ascorbic acid
(P = 0.05). Further adjustment for other risk factors substantially reduced the
association between plasma ascorbic acid and Ml, but the direct association between
plasma alpha-tocopherol and Ml was only slightly decreased, with a P value of 0.10.
Recent studies suggest that heavy alcohol consumption may be associated with
a low level of plasma alpha-tocopherol by influencing the hepatic alpha-tocopherol
regulatory function.18,19 The subjects were thus categorized into non-drinkers (< 0.5
g/day) and drinkers 0 0 . 5 g/day) based on the 4 day average alcohol consumption at
baseline. Multivariate Cox regression analyses were then performed separately for the
two groups and the results are summarized in Table 4. In the base model, a
significantly inverse association between plasma ascorbic acid and Ml was found
among non-drinker (P = 0.01); further adjustment for other confounders slightly
reduced the association, resulting a RR of 0.79 (P = 0.13). In contrast, among alcohol
drinkers, no association was found between plasma ascorbic acid and Ml in base
model; further adjustment for confounders resulted in a RR of 1.43, and this estimate
was marginally significant (P = 0.08). While no associations were found between
plasma alpha-tocopherol and risk of Ml among non-drinkers, a marginal direct
association was found among alcohol drinkers, with a RR estimate of 1.5 (P = 0.06)
from the fully adjusted model.
Table 5 summarizes the estimates from the measurement model performed on
the validation samples. The model included estimations of measurement error for
plasma alpha-tocopherol-to-cholesterol ratio, plasma ascorbic acid, serum total
cholesterol, HDL, systolic blood pressure and BMI, with baseline age included as a
73
error-free variable. A high validity of the baseline measurement was found for plasma
alpha-tocopherol, serum cholesterol, HDL, systolic blood pressure and BMI, as reflected
by the corresponding factor loadings M1 ). Baseline measurement of plasma ascorbic
acid, however, showed a poor validity. The variance of each observed variable was
partitioned into inter-individual and error variance, Var(F) and e1( respectively. The
estimated error variance for baseline ascorbic acid accounted for as high as 75 percent
of the observed variance. The corresponding figures for plasma alpha-tocopherol,
serum total cholesterol, HDL, systolic blood pressure and BMI were 31.0, 36.0, 29.0,
40.0 and 6.0, respectively. These results intuitively indicated that severe bias may
exist in the estimations of RR without correction for MEs, especially for plasma
ascorbic acid.
The comparison of RR estimates with and without correction for ME is
presented in Table 6. Corroborating with the results presented in Table 5, correction
for MEs resulted in substantial changes of RR estimates for plasma ascorbic acid and
plasma alpha-tocopherol. The RR for plasma alpha-tocopherol decreased from 1.18 to
1.07 (13.0 percent decrease), indicating little increase of Ml risk with an increase of
plasma alpha-tocopherol. In contrast, the inverse association between plasma ascorbic
acid and Ml was augmented, with the RR decreased from 0.95 to 0.55 (42.1 percent
decrease), indicating that the inverse association between plasma ascorbic acid and
Ml risk was substantially attenuated by ME. Considerable changes of RR estimates
were also observed for serum cholesterol, HDL, and systolic blood pressure; while only
slight change was found for BMI. It is interesting to found out that correction for ME
also resulted in substantial changes of RR for exercise ECG abnormality, smoking,
maximal oxygen uptake and living areas, although these variables were assumed to be
74
error free. It is noteworthy that correction for MEs is inevitably accompanied by
inflation of the standard errors and thus widening of the 95 percent confidence
intervals. Disproportional increase of standard error as compared to the change of point
estimate of RR may thus compromise the power of the analysis when correction for
measurement errors is attempted.
Discussion
A high levels of plasma alpha-tocopherol and ascorbic acid have been reported
to be associated with a decreased risk of CHD in epidemiologic studies.20'22 The
protective effects of alpha-tocopherol and ascorbic acid are generally ascribed to their
antioxidative properties.3 Finland is among the countries of the highest mortality and
incidence of cardiovascular disease in the world. The excess risk of heart disease in
Finland is believed to be associated with a poor antioxidant status in this
population23,24. It is thus hypothesized that low plasma levels of alpha-tocopherol and
ascorbic acid are associated with high risk of CHD and its sequelae, myocardial
infarction. Surprisingly, preliminary analysis of the current cohort data found that a
high concentration of plasma alpha-tocopherol was associated with an increased risk
of Ml, while no association was found for plasma ascorbic acid. The aim of the present
analysis was to investigate potential confounders and the biasing effects of ME on the
associations between plasma alpha-tocopherol, ascorbic acid and risk of Ml in this
cohort.
Recent development in the understanding of the in vivo metabolism of alpha-
tocopherol suggests that the plasma level of alpha-tocopherol is regulated by hepatic
alpha-tocopherol binding (transferring) proteins, which preferentially incorporate alpha-
7 5
tocopherol into the circulating lipoproteins.18 Hepatic dysfunction is thus likely to
compromise the preferential incorporation of alpha-tocopherol into circulating
lipoproteins, resulting in a decrease of plasma concentration of alpha-tocopherol.
Alcohol consumption may compromise the normal hepatic function, and consequently
reduce the plasma level of alpha-tocopherol. Because alcohol consumption, especially
light or moderate alcohol drinking, has been shown to be inversely associated with risk
of CHD,25,26 it is thus likely that alcohol consumption may confound the association
between plasma alpha-tocopherol and risk of Ml. In the present study, the cohort
members were classified into non-drinkers and drinkers, and multivariate regression
models were estimated separately for the two groups. It is interesting to find out that
there were considerable differences between the two groups in the associations
between plasma alpha-tocopherol and ascorbic acid with risk of Ml. While no
association was found between plasma alpha-tocopherol and risk of Ml among the non
drinkers, a positive association was found among those who drank 0.5 g or more
alcohol per day (RR = 1.50, P = 0.06). For plasma ascorbic acid, while an inverse
association with Ml was found among non-drinkers, a positive association was
observed among those who drank 0.5 g or more per day (RR = 1.43, P = 0.08). The
lack of statistical significance in the RR estimates was presumably due to the small
sample size in our study. To our knowledge, this is the first report that showed a
differential effects, depending on alcohol consumption, of plasma antioxidant vitamins
on risk of Ml. These results raised the possibility that failure to find an inverse
association between plasma alpha-tocopherol and ascorbic acid with risk of CHD in
some previously reported studies may be due to the lack of adequate control of alcohol
consumption,27,28 which may have masked the protective effects of plasma
76
antioxidants against CHD among nondrinkers.
Measurement error or intra-individual variation of risk exposure represents a
special source of bias in observational study. Failure to correct for ME in multivariate
models may seriously distort the true underlying relationship between interested risk
factors and disease. Only until recently, correction for ME in risk exposures has been
applied to epidemiologic studies of coronary heart disease.29" 31 In the present study,
a method proposed by Rosner et al. was used to correct for ME in multivariate RR
estimates for plasma alpha-tocopherol and ascorbic acid. Consistent with our
hypothesis, simultaneous multivariate correction for MEs in several risk factors of Ml
yielded a RR of 1.07 (0.79-1.45) for an increment of 0.3 //g alpha-tocopherol per mmol
serum total cholesterol, representing 13 percent decrease of risk as compared to the
uncorrected RR estimate. This result indicated that the crude (uncorrected) estimate
of RR for plasma alpha-tocopherol was substantially exaggerated by ME and spuriously
created a direct association between alpha-tocopherol and risk of Ml. In contrast,
correction for ME yielded a RR of 0.55 (0.10-3.06) for ascorbic acid as compared to
0.95 (0.75-1.20) from the uncorrected model, representing approximately 42 percent
underestimation of the inverse association between plasma ascorbic acid and risk of
Ml by the uncorrected RR estimate. This result indicated that the protective effect of
ascorbic acid against Ml was seriously masked by ME. It is noteworthy, however, that
correction for ME was associated with inflation of the standard error estimates which
hampered hypothesis testing in our analysis. These results imply that ME correction
entails an adequately large size of the validation sample to ensure enough power to
detect any potential effect. One major determinant of the size of the validation sample
is the reliability of the measurements. A quite large sample size is needed when the
77
reliability of measurement is low. A reliability coefficient of 0.60 for plasma alpha-
tocopherol was reported by Willett et al. and Gey et al.32 Compared with these
reports, the reliability correlations in our sample were relatively low for both plasma
alpha-tocopherol and ascorbic acid, which limited the statistical power of our analysis.
Our results also empirically exemplified that ME may virtually inflate, as well as
attenuate as commonly believed, the RR estimates in multivariate analysis. It should
be realized that the biasing effect of ME is not always unidirectional. Quantitative
statistical methods to correct for ME is necessary in the investigation of the
relationship between risk exposures and disease.1 1
A synergistic effect between alpha-tocopherol and ascorbic acid has been
demonstrated in experimental studies,33,34 but has not been reported in epidemiologic
studies. Under experimental conditions, ascorbic acid has been shown to reduce the
oxidized alpha-tocopherol, tocopheroxyl radicals, thus may preserve the body alpha-
tocopherol pool. Ascorbic acid has also been shown to be a major antioxidant by
scavenging free radicals in the aqueous phase, thus may have a sparing effect on
alpha-tocopherol.33 In our analyses, interactive models were also estimated with and
without correction for MEs in alpha-tocopherol, ascorbic acid. However, no evidence
was found suggestive of an interaction between plasma alpha-tocopherol and ascorbic
acid. Our analysis also failed to reveal any interactions between alpha-tocopherol and
ascorbic acid and cigarette smoking.
Several methodological limitations of the current analysis should be pointed out.
First, the method used in this study assumes that the intra-individual variability in
plasma alpha-tocopherol, ascorbic acid and the covariates is solely due to random
fluctuation. In reality, however, this assumption may not hold. It is likely that there
78
may be some unknown sources of non-random errors in the measurements and some
real changes of plasma alpha-tocopherol, ascorbic acid, as well as the covariates,
might have occurred since the repeated measurements were obtained one to five years
after baseline exam. It is also likely that due to the increasing knowledge of CHD and
self-consciousness of health, individuals may have changed their life-styles or diet,
which may affect the individual's plasma alpha-tocopherol and ascorbic acid levels. A
major concern is the supplementation use of vitamins. It is estimated that the
prevalence of vitamin supplements use is approximately 50 percent in the U.S.
population. Although supplements use in our sample was rare at the time of baseline
examination, the use rate was probably higher in 1990 when the repeated
measurements were determined. Unfortunately, no information is available for us at
this point. The relatively low reliability correlations of alpha-tocopherol (r = 0.43),
ascorbic acid (r = 0.31) and serum cholesterol (r = 0.35) in the validation sample were
suggestive of some real changes. To this point, over-correction for ME may present to
some extent in our analysis. However, given the chronicity nature of heart disease, the
exposures of concern here are the long-term levels of plasma alpha-tocopherol and
ascorbic acid and their relationships to Ml. It is thus unlikely that changes of plasma
alpha-tocopherol and ascorbic acid during a relatively short period of time in middle-
aged persons would substantially alter the risk of Ml. It is thus plausible to treat the
real change as one source of ME. Another limitation of the method is the normality
assumption of the predictor variables. This assumption w as apparently violated for the
categorical variables included in the our analyses. Since no direct correction for MEs
were attempted for these variables, inclusion of them in the analyses should, however,
not materially affect the results.
79
Nevertheless, the issue of ME may have important implications for conducting
epidemiologic research. It underscores the need for reproducibility and validation
studies, therefore establishing some potential new priorities when planning and
implementing studies. From an etiological point of view, ME correction may help clarify
causal mechanisms of disease and interpretation of findings from observational studies.
Therefore, there is a growing consensus that the design, implementation, and
analytical phases of future studies should consider the possibility and consequences
of exposure measurement errors.
In summary, the example presented here illustrated the feasibility and the
precautions that one should bear in mind when implementing procedures for ME
correction. Our data also suggest that alcohol consumption may be an effect modifier
of the associations between plasma alpha-tocopherol, ascorbic acid and risk of Ml.
High levels of both plasma alpha-tocopherol and ascorbic acid were found to be
associated with an excess risk of Ml among alcohol drinkers, while a protective effect
of ascorbic acid was found among non-drinkers. These results, if replicated in other
studies, can be of important public health implications: use of vitamin supplements,
while may be beneficial for the protection against heart disease among non-drinkers,
the combination of high doses of supplements and alcohol drinking may virtually
potentiate the risk of CHD. The findings from our analysis warrant further investigation
in well designed study.
80
Reference
1. Ross R. The pathogenesis of atherosclerosis-an update. N Engl J Med
1986;314:488-493.
2. Steinberg D, Parthasarathy S, Carew TE, Khoo JC, Witztum JL. Beyond
cholesterol: modification of low-density lipoprotein that increase its
atherogenecity. N Engl J Med 1989;320:915-24.
3. Halliwell B, Gutteridge JMC. Free radicals in biology and medicine. 2nd ed. New
York: Oxford University Press, 1989.
4. Riemersma RA, Wood DA, Macintyre CA, Elton RA, Gey KF, Olive MF. Risk of
angina pectoris and plasma concentrations of vitamins A, C and E and carotene.
Lancet 1991;337:1-5.
5. Gey FK, Puska P, Jordan P, Moser UK. Inverse correlation between plasma
vitamin E and mortality from ischemic heart disease in cross-cultural
epidemiology. Am J Clin Nutr 1991;53:326s-331s.
6. Salonen JT, Salonen R, Seppanen K, Kantola M, Parviainen M, Alfthan G,
Maenpaa PH, Taskinen E, Rauramaa R. Relationship of serum selenium and
antioxidants to plasma lipoproteins, platelet aggregability and prevalent
ischemic heart disease in Eastern Finnish men. Atherosclerosis 1988;70:155-
61.
7. Riemersma RA, Oliver M, Elton RA, Alfthan G, Vartiainen E, Salo M, Rubba P,
Mancini M, Georgi H, Gey KF. Plasma antioxidants and coronary heart disease:
vitamins C and E, and selenium. Eur J Clin Nutr 1990;44:143-8.
8. Kok FJ, de Bruijn AM, Vermeeren R, Hofman A, van Laar A, de Bruin M,
Hermus RH, Valkenburg HA. Serum selenium, vitamin antioxidants and
cardiovascular mortality: a 9-year follow-up study in the Netherlands. Am J Clin
Nutr 1987;45:462-8.
9. Salonen JT, Salonen R, Penttila I, Herranen J, Jauhiainen M, Kantola M,
Lappetelainen R, Aflthan G, Puska P. Serum fatty acids, apolipoproteins,
selenium and vitamin antioxidants and the risk of death from coronary artery
disease. Am J Cardiol 1985;56:226-33.
10. Michalek JE, Tripathi RC. The Effects of errors in diagnosis and measurement
on the estimation of the probability of an event. JASA 1980;75:713-21.
11. Thomas D, Stram D, Dwyer JH. Exposure measurement error: influence on
exposure-disease relationships and methods of correction. Ann Rev Publ Health
1993;14:69-93.
12. Rosner B, Willett WC, Spiegelman D. Correction of logistic regression relative
81
risk estimates and confidence intervals for measurement error: the case of
multiple covariates measured with error. Am J Epidemiol 1990;132:734-45.
13. Salonen JT. Is there a continuous need for longitudinal epidemiologic research?
The Kuopio Ischemic Heart Disease Risk Factor Study. Ann Clin Res
1988;20:46-50.
14. Salonen JT, Nyyssonen K, Korpela H, Tuomilehto J, Seppanen R, Salonen R.
High stored iron levels are associated with excess risk of myocardial infarction
in Eastern Finnish men. Nutr Res 1989;9:597-604.
15. Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Survey Methods.
Geneva, World Health Organization, 1982.
16. Tuomillehto J, Arstila M, Kaarsalo E, Kankaanpaa J, Ketonen M, Kuulasmaa K,
Lehto S, Miettinen H. Acute myocardial infarction in Finland: baseline data from
the FINMONICA AMI registry in 1983-1985. Eur Heart J 1992;13:577-587.
17. Fuller WA. Measurement error models. John Wiley & Sons, 1987, Chapter 1,
Appendix 1A, pp85-88.
18. Traber MG. Determinants of plasma vitamin E concentrations. Free Rad Biol &
Med 1994;10:229-239.
19. Stryker WS, Kaplan LA, Stein EA, Stampfer MJ, Sober A, Willett WC. The
relation of diet, cigarette smoking, and alcohol consumption to plasma beta-
carotene and alpha-tocopherol levels. Am J Epidemiol 1988;127:283-296.
20. Gey KF, Stahelin HB, Eichholzer M. Poor plasma status of carotene and vitamin
C is associated with higher mortality from ischemic heart disease and stroke:
Basel Prospective Study. Clin Invest 1993;71:3-10.
21. Gey FK, Puksa P, Jordan P, Moser UK. Inverse correlation between plasma
vitamin E and mortality from ischemic heart disease in cross-cultural
epidemiology. Am J Clin Nutr 1991;53:326s-331s.
22. Street DA, Comstock GW, Salkeld RM. A population based case-control study
of association of serum antioxidants and myocardial infarction. (Abstract). Am
J Epidemiol 1991;134:719-20.
23. Salonen JT, Salonen R, Korpela H, Suntioinen S, Tuomilehto J, Esterbauer H,
Tatzber F, Salonen R. Intake of mercury from fish, lipid peroxidation, and the
risk of myocardial infarction and coronary, cardiovascular, and any death in
Eastern Finnish men. Circulation 1995;91:645-655.
24. Artaud-Wild SM, Connor SL, Sexton G, Connor WE. Differences in coronary
mortality can be explained by differences in cholesterol and saturated fat
intakes in 40 countries but not France and Finland: A Paradox. Circulation
82
1993;88(6):2771 -2779.
25. Garcia-Palmieri MR, Sorlie P, Tillotson J. Relationship of dietary intake to
subsequent coronary heart disease incidence. The Puerto Rico Heart Health
Program. Am J Clin Nutr 1980;33:1818-27.
26. Kushi LH, Lew RA, Stare FJ. Diet and 20-year mortality from coronary heart
disease. The Ireland-Boston Diet-Heart Study. N Engl J Med 1985;312:811-18.
27. Hense HW, Stender M, Bors W, Keil U. Lack of an association between serum
vitamin E and myocardial infarction in population with high vitamin E levels.
Atherosclerosis 1993;103:21-28.
28. Kardinaal AFM, Kok FJ, Ringstad J et al. Antioxidants in adipose tissue and risk
of myocardial infarction: the EURAMIC study. Lancet 1993;342:1379-84.
29. MacMahon S, Peto R, Cutler J. Blood pressure, stroke and coronary heart
disease. I: prolonged differences in blood pressure: prospective observational
studies corrected for the regression dilution bias. Lancet 1990;335:765-774.
30. Davis CE, Rifkind BM, Brenner H, Gordon DJ. A single cholesterol measurement
underestimates the risk of coronary heart disease. An empirical example from
the Lipid Research Clinics Mortality Follow-up Study. JAMA 1990;264:3044-
3046.
31. Law MR, Wald NJ, Wu T, Hackshaw A, Bailey A. Systematic underestimation
of association between serum cholesterol concentration and ischemic heart
disease in observational studies: data from the BUPA study. BMJ
1994;308:363-6.
32. Willett WC. Nutritional Epidemiology. Oxford University Press, 1991.
33. Frei B, England L, Ames BN. Ascorbate is an outstanding antioxidant in human
blood plasma. Proc Natl Acad Sci USA 1989;86:6377-81.
34. Vatassery GT, Smith WE, Quach HT. Ascorbic acid, glutathione and synthetic
antioxidants prevent the oxidation of vitamin E in platelet. Lipids
1989;24:1043-7.
83
T able 1. Baseline C haracteristics o f Incident M yocardial Infarction Cases an d Non-cases
T he K uopio Ischem ic H e a rt Disease R isk F acto r S tudy __________
V ariables Non-cases ( n = 1,473) Cases (n = 7 4 )
Plasma a-tocopherol/cholesterol (g/mmol) 1.44(0.30)' 1.51 (0.36)
Plasma ascorbic acid (mg/1) 8.43 (4.02) 7.05 (4.33)
Serum cholesterol (mmol/1) 5.86 (1.03) 6 .2 6 (1 .0 6 )
High density lipoprotein (mmol/1) 1.32 (0.30) 1.22 (0.30)
Systolic blood pressure (mmHg) 133.8 (16.3) 140.3 (17.1)
Body mass index (kg/mJ) 26.7 (3.4) 27.3 (3.7)
Maximal oxygen uptake (ml/kg/min) 32.3 (7.6) 28.0 (6.4)
Alcohol consumption (g/d) 9.3 (18.8) 6.6 (13.0)
Socioeconomic status 10.7 (4.7) 12.3 (5.4)
Smoking status (% )2 28.7 53.7
Family history o f IHD (%) 46.5 54.7
Living in rural area (% )3 26.3 42.5
Exercise EKG abnormality (%)* 17.5 31.3
Age (years) 52.2 (5.3) 53.8 (3.3)
1. Standard deviation o f continuous variables included in parentheses.
2. Percentage o f smokers as defined.
3. Percentage o f subjects living in rural areas at baseline exam.
4. Percentage ill subjects with abnormal EKG at haseline exam.
84
Table 2. Age-adjusted C orrelations am ong Repealed M easurem ents of Risk f actors in tbe V alidation Subsam ple (n= 91)
The Kuopio Ischemic H eart Disease Risk Factor Study
Risk Factors
Plasma a-tocopherol/cholesterol(g/rrimol) 0.43' -o.or- 0.16 0.03 0.12 0.09
Plasma ascorbic acid (mg/l) 0.043 0.31 0.16 -0.01 0.01 -0.08
Serum cholesterol (mmol/1) 0.04 0.03 0.35 0.34 -0.04 -0.15
High density lipoprotein (mmol/1) -0.07 0.07 0.24 0.68 0.10 -0.15
Systolic blood pressure (mmHg) 0.22 -0.04 -0.03 0.08 0.62 0.19
Body mass index (kg/m3 ) 0.12 -0.11 -0.10 -0.17 0.25 0.90
1. On-diagnnul (Bold) indicate correlations between repeated measures o f the same risk factors.
2. Upper triangle (Italic) indicate correlations among baseline measurements and the repeated measurements o f other risk factors.
3. Lower triangle indicate correlations among repeated measurements and the baseline measurements o f other risk factors.
I'al)If 3. Relative Risk (RR) Intim ates of M yocardial Infarction for Plasma A lpha-tocopherol and Ascorbic Acid
The Kuopio Ischemic H eart Disease Risk Factor Study
Model Plasma A lpha-tocopherol3 Plasm a Ascorbic Acid4
Base M odel1
Relative Risk 1.21 0.79
95% Cl 0.98-1.50 0.62-1.00
P Value 0.08 0.05
M ultivariate Model2
Relative Risk 1.18 0.95
95% Cl 0.97-1.45 0.75-1.20
P Value 0.10 0.67
Note:
1. Uase model included baseline nee, plasma alpha-tocopherol to cholesterol ratio, ascorbic acid and baseline exam years.
2. Covariates added to (be base model: total cholesterol, high density lipoptolein, systolic blood pressure, smoking status, body mass index,
exercise HCG abnormality, family history of ischemic heart disease, socio-economic status, living areas and maximal oxygen uptake.
3. Relative risk estimated for 0.3 increment o f plasma alpha-tocopherol to serum cholesterol ratio.
4. Relative risk estimated for 4.0 mg/l increment o f plasma ascorbic acid.
Table 4. Relative Risk (RR) Intim ates of M yocardial Infarction for Plasma A lpha-tocopherol and Ascorbic Acid,
according to Alcohol Consum ption: The Kuopio Ischemic H eart Disease Risk Factor Study
Model RR 95% C l P value
Light D rinkers (:S0.5 g/day, n=783)
D m e m o d el1
Alpha-tocopherol5 1.19 0.91-1.56 0.22
Ascorbic acid4 0.67 0.51-0.93 0 . 0 1
M ultivariate m odel2
Alpha-tocopherol 1.08 0.84-1.40 0.54
Ascorbic acid 0.79 0.59-1.07 0.13
Heavy D rinkers (> 0 .5 g/day, n=765)
Dm e m odel
Alpha-tocopherol 1.26 0.87-1.83 0.22
Ascorbic acid 1.04 0.69-1.55 0.86
M ultivariate m odel
Alpha-tocopherol 1.50 0.98-2.28 0.06
Ascorbic acid
■ - ------------------------------------------- ------ - , - 4 - --------- — - ■ ■ ■ ■ ,
1.43 0.96-2.12 0.08
Nnie:
1. Duse model included baseline aye, plasma nlplia-locoplieiol, asciuhic acid and baseline exam yeais.
2. Covariales added In hase model: lolal clmlesleiol, high deusiiy lipopiotein, systolic blood pressure, smoking status, body mass index,
exercise ECG abnormality, family history o f ischemic heart disease, living areas, socioeconomic status and maximal oxygen uptake.
3. Estimated for 0.3 ( ) increment of plasma alpha-tocopherol to total cholesterol ratio.
4. Estimated for 4.0 mg/l increment of plasma ascorbic acid.
Table 5. Estimations of Validity, E rro r V ariance and Inter-individual V ariance from a M easurem ent M odel
Risk Factor x ,1 V
V ar ( F f
e,4 £z*
Plasma a-tocopherol/cholesterol (g/mmol) 0.83 0.46 0.049 0.022 0.17
Plasma ascorbic acid (mg/1) 0.49 0.57 3.17 9.96 9.97
Serum Cholesterol (mmol/1) 0.80 0.48 0.27 0.14 0.91
High density lipoprotein (mmol/1) 0.84 0.81 0.042 0.018 0.021
Systolic blood pressure (mmHg) 0.77 0.83 141.4 94.7 63.2
Body mass index (kg/m7 ) 0.97 0.93 5.55 0.40 0.84
1. X , denotes factor loading or validity of (lie baseline measurements.
2. X 2 denotes factor loading or validity o f the repeated measurements.
3. Var(F) denotes estimates of true variance of risk factors corrected for measurement errors.
4. e, denotes error variance of the baseline measurements.
5. (2 denotes error variance of the repeated measurements.
Table (i. Relative Hislc Estimates from Proportional H azard M ade!1 : w ithout and witli C orrection for M easurem ent E rro r
'The Kuopio Ischemic Heart Disease Ilisk Factor Study (N = 1,537)
Variables ( + AJ)
UncoiTcdctl
RR (95% Cl)
C orrected
RR (95% Cl) % C hange3
Plasma a-tocopherol* (+ 0 .3 pg/mmol) 1.18 (0.97-1.44) 1.07 (0.79-1.45) -13.0
Plasma ascorbic acid (+ 4 mg/l) 0.95 (0.75-1.20) 0.55 (0.10-3.06) -42.1
Serum cholesterol (+1 mmol/l) 1.21 (0.98-1.49) 1.48 (0.70-3.09) + 22.3
Serum HDL (+ 0.3 mmol/l) 0.78 (0.59-1.03) 0.56(0.23-1.36) -28.2
Systolic blood pressure (+ 1 6 mmllg) 1.26 (1.03-1.54) 1.42 (0.78-2.58) + 12.7
Body mass index (+ 3 .4 kg/mJ) 1.08 (0.84-1.39) 1.00(0.66-1.53) -7.4
Exercise EKG abnormality 1.55 (0.94-2.60) 2.72 (0.60-12.3) + 75.5
Maximal oxygen uptake (+ 8 ml/kg/min) 0.71 (0.51-0.99) 0.78 (0.68-1.46) + 9 .8
Cigarette smoking
(smokers vs. non-smokers)
2.69 (1.62-4.46) 2.17 (0.99-4.72) -19.0
Family history of IHD
(yes vs. no)
1.35 (0.85-2.15) 1.30 (0.68-2.50) -3.7
Living area \
(rural vs. urban)
1.57 (0.98-2.52) 1.70 (0.79-3.65) + 8 .3
Age (+ 5 years) 1.13 (0.77-1.66) 1.15 (Q.74-1.78) + 2 .3
Note:
1. Baseline exain years were included in the model.
2. Increment in continuous variables associated with UR estimates;
3. % Change estimated as [(Corrected - Uncorrected)/Uncurrecled]*100;
4. Plasma a-tocopherol to serum cholesterol ratio.
Chapter 4. Serum Selenium and Risk of Cancer in a Cohort of Middle-aged Finnish
Men: The Kuopio ischemic Heart Disease Risk Factor Study
Abstract
Objective: To assess the relationship between serum selenium and subsequent risk of
cancer.
Design-. Longitudinal follow-up study, initial examination in 1984-1989, follow-up until
December, 1992.
Participants-. 1,962 middle-aged men initially free of cancer and coronary heart disease
living in Eastern Finland.
Results: After an average of 5.5 years of follow-up, sixty three incident cancer cases
were documented. An inverse association between serum selenium and risk of cancer
was observed. After controlling for age, and years of baseline examination, the
estimated relative risk (RR) of cancer associated with an increment of 23 jt/g/l of serum
selenium was 0.38 (95 percent confidence interval, 0.16-0.91; p = 0.03). Further
adjustment for potential confounders yielded a RR estimate of 0.62 (95 percent
confidence interval, 0.26-1.48; p = 0.28). In multivariate analysis, cigarette smoking
was found to be an effect modifier of the association between serum selenium and risk
of cancer: the RR estimate was 1.10 (95 percent confidence interval, 0.43-2.85) for
non-smokers; and it was 0.32 (95 percent confidence interval, 0.12-0.89) for smokers.
This difference was highly statistically significant (p<0.01).
Conclusion: These findings support the hypothesis of an inverse relationship between
serum selenium and cancer. The protective effect of selenium on risk of cancer is
mainly confined to smokers, with little effect on non-smokers.
Key words: Selenium, Cancer, Smoking, Cohort Study, Effect Modification.
90
Introduction
It has been proposed that selenium has an anticarcinogenic effect due to its
antioxidant activity, and effects on carcinogen metabolism.1,2 Evidence from animal
studies supports the hypothesis that low intake of selenium is associated with an
increased risk of cancer. Experimental studies indicate that selenium inhibits the
mutagenic effects of different carcinogens, as measured by in vitro mutagen assay
procedure.3,4 In animal models, selenium supplementation has been shown to reduce
the frequency of chemically induced tumors on several sites.5"7
In ecologic studies, an inverse association has been observed between average
selenium intake or blood levels and both total and site-specific cancer mortality in the
United States and internationally.8,9 Shamberger and Frost first proposed that the
geographic distribution of the essential nutrient selenium in the United States was
inversely associated with cancer mortality.10 Subsequently, study of cancer mortality
and forage selenium levels in U.S. counties have reported a consistent pattern of
excess cancer mortality within the regions of the United States that have lower
selenium compared with those regions that have variable or adequate selenium
levels.11 This association is similar for both males and females and for each of the
major cancer sites examined.
Prospective studies of selenium and cancer, however, have yielded inconsistent
results. Whereas the results of some prospective studies have suggested an inverse
relationship between selenium and cancer at all sites combined,12"14 other prospective
studies of this relationship have yielded null results.15"17 One suggested explanation
for this apparent discrepancy of results among prospective studies previously reported
is the relatively adequate intake of selenium and the lack of variation of selenium levels
91
in the U.S. population.16 Effect modification of the relation between selenium and
cancer by several factors, e.g. sex, age, and smoking may also to some extent explain
finding of an inverse association between selenium and cancer in some studies, but not
in others.18,19
Selenium in foods and forage crops in Finland has been known to be noticeably
low as compared to other European countries and the United States.20 In the present
study, we prospectively investigated the association between serum selenium level and
subsequent risk of total cancer in a cohort of middle-aged men living in Eastern
Finland, an area with a notably low selenium content in soil and crops .
Methods and Materials
The Kuopio Ischemic Heart Disease Risk Factor study is a population based
study to investigate previously unestablished risk factors for myocardial infarction and
carotid atherosclerosis in men in Eastern Finland. Details of the study design have been
described elsewhere.21 Briefly, the baseline examinations were carried out between
March 1984 and December 1989. The study sample was composed of 3,235 men in
Eastern Finland aged 42, 48, 54 or 60 years at the baseline examination. Of these,
2682 (82.9%) participated. Men with either prevalent coronary heart disease (n = 667)
or cancer (n = 46) were excluded from the present analysis, as these diseases could
have influenced the dietary habits and concentration of selenium in the blood. Of the
remaining 1969 men, complete data on the primary variables were available for 1,962
men.
Measured risk factors included anthropometric measures, blood pressure,
fasting serum lipids, blood chemistry, and a self-administered questionnaire concerning
92
sociodemographic factors, individual and family medical history, smoking, alcohol
consumption and physical activity. Repeat interviews to obtain medical history were
conducted by a physician. A subject was defined a smoker if he had ever smoked on
a regular basis and smoked cigarettes, cigar, or pipe within the past 30 days. The life
long exposure to smoking ("cigarette year") was estimated as the product of years
smoked and the number of tobacco products smoked daily at the time of examination.
Years smoked were defined as the sum of years of smoking regardless of when
smoking had started, whether the subject had stopped smoking, and whether smoking
had occurred continuously or during several periods. The consumption of alcohol in the
previous 12 months was assessed with the quantity-frequency method by using the
Nordic Alcohol Consumption Inventory, which contains 15 item. The socioeconomic
status was measured with the summary index that combined measures of income,
education, occupation, occupational prestige, material standard of living, and housing
condition.
Leisure time physical activity was assessed using a 12-Month Leisure Time
Physical Activity Questionnaire. The questionnaire included the most common leisure
time physical activities of middle-aged Finnish men, selected on the basis of a previous
population study conducted in Finland.22 The intensity of physical activity was
expressed in metabolic units (METs). The MET is the ratio of metabolic rate during
physical activity to the metabolic rate at rest. One MET corresponds to an energy
expenditure of approximately 1 kcal/kg/hour and an oxygen uptake of 3.5
ml/kg/minute.
The respiratory gas exchange was measured breath-by-breath with an MGC
2001 system (Medical Graphics Corp., Minneapolis, MN) during a symptom-limited
93
exercise test. The testing protocol comprised a linear increase of work load by 20
W/min. Highest oxygen uptake during the test was defined as V02max. Exercise ECGs
were coded manually by one cardiologist. The criteria for ischemia were 1) ischemic
ECG defined as horizontal or downsloping ST depression ^ 0 .5 mm or upsloping ST
depression > 1.0mm; 2)typical angina pectoris pain leading to discontinuation of
exercise; 3) maximal heart rate during exercise ^ 1 3 0 beats per minute.
Venous blood was taken from each subject after having abstained from alcohol
for 3 days, smoking for 12 hours, and fasting for 8-12 hours. A sample of serum was
stored at -20°C in closed plastic tubes for 1-2 years before the determination of
selenium concentration. Serum selenium concentration was measured by an atomic
absorption spectrometric method using furnace, Zeeman background correction, and
pyrolytically coated graphite tubes with a platform.23
Data on death in the cohort were obtained from the National Death Certificate
Register, which records all deaths of Finnish citizens. Underlying cause of death was
taken as that assigned at the Central Statistical Office of Finland. Data on nonfatal
cancers were derived from the National Hospital Discharge Data Register.
Between March 1984 and December 1992, a newly diagnosed cancer was
registered in 63 of the 1,962 men at risk. The longest follow-up period for cancer
events was 8.75 years, and the median follow-up time was approximately 5.5 years.
Statistical Analysis
The descriptive characteristics of cancer cases and cancer free subjects were
first computed. Mean serum selenium concentrations according to baseline exam year
were also computed.
94
The association of serum selenium and risk of total cancer was analyzed using
Cox Proportional Hazards regression models. A base model including serum selenium,
age and the year of baseline examination was first estimated. The rationale for the
inclusion of baseline exam years in all models was based on the fact that fertilizers
fortified with selenium has been widely used in Finland since 1986. To further control
for confounding, covariates then were added into the base model, including leisure
time physical activity, maximal oxygen uptake, body mass index (BMI), cigarette-years,
alcohol consumption, family history of ischemic heart disease, ischemic ECG in
exercise, social economic status, and place of residence. To explore any threshold
effect of selenium on risk of cancer, serum selenium concentrations were categorized
into tertile groups, and dummy variables corresponding to tertile groups were then
used.
Previous studies suggest that cigarette smoking may be an effect modifier of
the association between selenium and risk of cancer.24 To assess the interaction
between serum selenium and cigarette smoking, two interactive models were
estimated. First, the product of the deviation score of serum selenium concentration
(subtracting mean) and smoking status (0 = non-smoker, 1 = smoker), was added to the
full regression model. In a second interactive model, tertile groups of selenium were
used and the products of dummy variables of selenium tertiles and smoking status
were added to the full regression model.
To explore possible heterogeneity of the selenium-cancer association between
persons examined during 1984-1985 and those examined after 1986-1989, separate
models were also estimated for the two subgroups defined by baseline exam years.
Tests of interaction between serum selenium and the dummy variable corresponding
95
to the year of exam (EXAMY = 0: 1984-1985; EXAMY = 1: 1986-1989) were also
estimated.
The RR estimates of cancer for selenium were calculated based on an increment
of 23 (fjgl\) serum selenium. In categorical analyses, the RR were estimated based on
comparison of the top tertile group to the bottom tertile group of serum selenium. All
tests of significance were two-sided at a =0.05.
Results
The baseline mean concentration of serum selenium for the 63 cancer cases
was significantly lower than that of cancer free subjects, 90.4 fjg/\ and 102.1 /vg/l,
respectively (p<0.05). Baseline characteristics of cases and cancer-free subjects
assumed to be associated with risk of cancer are presented in Table 1. Notable
differences were found between cancer cases and non-cases in cigarette-years,
amount of weekly alcohol consumption, and leisure time physical activity. A higher
proportion of cancer cases was found to have a family history of ischemic heart
disease (IHD) and exercise ECG indicating IHD as compared to cancer free subjects.
Previous studies indicate that low serum selenium is associated with an elevated risk
of heart disease. To control for potential confounding by underlying ischemic heart
disease, family history of IHD and exercise ECG IHD were included in subsequent
multivariate regression analyses. Since social economic status and residential areas
(rural versus metropolitan) may influence the accessibility to foods that may affect the
long-term intake of and serum concentration of selenium, these two variables were
also included in multivariate regression analyses.
In Finland, the use of selenium fortified fertilizer has been widely used since
96
1986. Because the subjects of the cohort were recruited into the study during a period
of 5 years spanning 1984 to 1989, an upward drift of baseline serum selenium
concentration was expected for subjects entering the study after 1986 as compared
to those entering the study before 1986. This phenomenon was reflected by the
steady increase of mean baseline serum selenium concentration according to year of
baseline examination (Table 2). The mean serum selenium levels for persons recruited
during 1984-1985 and those during 1986-1989 were 78.8 (//g/l) and 115.6 (//g/l)
respectively, and the corresponding accumulative incidence of cancer were 4.6 and 2.4
percent, respectively. Also indicated in Table 2 are the higher incidence of total cancer
and the lower mean serum concentration of selenium in smokers as compared with
non-smokers.
Table 3 summarizes the estimated RR and 95 percent confidence intervals from
multivariate regression models. The base model included age and year of baseline
examination as covariates. The estimated RR associated with an increment of 23 //g/l
selenium was 0.38 (p = 0.03). A moderate inverse association remained with further
adjustment for other confounders, the estimated RR was 0.62 (0.26-1.48), although
this estimate was not statistically significant. The change of the RR estimate compared
to that from the base model indicated substantial confounding by other risk factors.
Maximal oxygen uptake (£ = -0.059, SE = 0.022) and cigarette-years (£ = 0.0011,
SE = 0.0003) were found to be significantly associated with risk of cancer.
A significant interaction between serum selenium and cigarette smoking was
found in both models with continuous and categorical variables of serum selenium. The
RR estimates for each 23 //g/l increment of selenium according to smoking status are
presented in Table 3. While no reduction of cancer risk was found for non-smokers
97
(RR = 1.10, 95% Cl =0.43-.85), a significant decrease of risk was found for smokers
(RR = 0.32, 95% Cl =0.12-0.89), and the difference of RR between smokers and non-
smokers was statistically significant.
To explore potential threshold effect of serum selenium, models with serum
selenium categorized into tertiles were estimated. In base model, a significant decrease
of cancer risk (RR = 0.36, 95% Cl =0.14-0.97) was found for persons in the middle
tertile group as compared with those in the bottom tertile group; while a smaller and
non-significant decrease of risk (RR = 0.55, 95% Cl =0.19-1.60) was found for the top
tertile group. Further adjustment for confounders diminished the effect of selenium,
with a RR of 0.51 (95% Cl =0.19-1.33) and 0.86 (95% Cl =0.30-2.50) respectively
for the middle and top tertile groups. Furthermore, interactive models testing effect
modification by smoking status indicated that a threshold effect of selenium existed
among smokers, as demonstrated by the similar RR estimates for the second and top
tertile groups of serum selenium among the smokers, with RR estimates of 0.54 (95%
Cl =0.18-1.67) and 0.57 (95% Cl = 0.14-2.22), respectively. Interestingly, among non-
smokers, an inverse association was found only in the second tertile group (RR = 0.50,
95% Cl =0.15-1.60), with little beneficial effect observed in the top tertile group
(RR = 1.09, 95% Cl =0.35-3.43) (Figure 1). The lack of significance in any of the
above RR estimates may presumably be due to the small sample size and number of
cases in our data. When the subjects were categorized into tertile groups according to
cut-off values within year of baseline examination, the same pattern was observed,
results were thus not presented here.
98
Discussion
In the present study based on 5.5 year follow-up of a cohort of middle-aged
men in Eastern Finland, an inverse association between serum selenium concentration
and subsequent risk of cancer all site combined was observed; a 60 percent reduction
of cancer risk with each 23 //g/l increment of serum selenium was found with
adjustment for age and year of baseline examination; and a 40 percent reduction was
observed with full adjustment for confounders, but it was not statistically significant
(p = 0.28). Presumably, the small number of cases and relatively short period of follow-
up in the study has limited the power of the present analysis. Three out of four
previously reported prospective studies from Finland found a significant inverse
association between serum selenium concentration and risk of cancer all site
combined,13,14,19 while a non-significant RR of 0.9 was reported in the other study
(Table 5)25. The totality of evidence from these studies, together with the results from
the present analysis, supports a protective effect of serum selenium on the
development of cancer all site combined in Finland, in which a relative deficiency of
selenium in soil has been well known as compared to other European countries. These
results are suggestive of a role of selenium deficiency in the etiology of cancer
development.
More interestingly, a differential effect of serum selenium on risk of cancer,
dependent on cigarette smoking, was found in the present study. A 70 percent
reduction of cancer risk was found among smokers with each increment of 23 //g/l
serum selenium (p<0.05), while no beneficial effect was observed among non-
smokers in the cohort. This difference in the selenium-cancer association between
smokers and non-smokers is unlikely to be explained by confounding, since many
99
known or suspected risk factors for cancers were simultaneously adjusted in the
multivariate models. It is also unlikely that bias may explain this differential effect
between smokers and non-smokers due to the prospective nature of the present study.
A plausible explanation for this differential effect between smokers and non-smokers
is the antioxidative property of selenium as the prosthetic factor of glutathione
peroxidase, a major antioxidative endogenous enzyme in the human free radical
defense system. Selenium is essential for the antioxidative activity of glutathione
peroxidase, whose major role is to detoxify hydrogen and hydroxyl free radicals.26
Furthermore, experimental studies indicate that the activity of glutathione peroxidase
can be induced by both oxidative stress and selenium.28'30 Increased activity of
glutathione peroxidase has been demonstrated in animal models when challenged with
high oxidative stress or supplemented with high dose of selenium. Given the high
oxidative stress posed by smoking and the presence of high level of selenium, a higher
activity of glutathione peroxidase may thus be expected in smokers as compared to
non-smokers. The induced higher activity of glutathione peroxidase then may offer
greater protection against free radical damage to DNA, thus the development of cancer
among smokers.
Our results also suggest a threshold effect of serum selenium on risk of cancer.
Compared with subjects in the bottom tertile of selenium concentration, a protective
effect of selenium was observed in the middle tertile with a reduction of 49 percent
of cancer risk, but only 14 percent in the upper selenium tertile group, although both
RR estimates were not statistically significant. Further analyses of interaction between
smoking and selenium revealed that a threshold effect was mainly confined to
smokers, with a reduction of 46 percent of cancer risk for both the middle and top
100
tertile groups; while a comparable protective effect was observed in the middle tertile
group among the non-smokers, with a 50 percent reduction of cancer risk, no
protection by selenium was found in the top tertile group among the non-smokers
(Figure 1). This pattern of observation may be explained by the mixed biological effects
of selenium. While selenium is essential for the activity of glutathione peroxidase, thus
protection of free radical mediated damage, excessive selenium, however, has been
shown to be cytotoxic.31,32 An optimal level of selenium is thus crucial for its maximal
beneficial effect against cancer. This paradoxical effect of selenium, together with the
inducible nature of glutathione peroxidase activity, may account for the observed
results in the present study. While in smokers, presumably with higher oxidative stress,
high selenium levels are probably necessary for the activity of glutathione peroxidase,
the same selenium level may exert cytotoxic effects in non-smokers due to a relatively
lower oxidative stress. This cytotoxic effect may thus offset the protective effect of
glutathione peroxidase against risk of cancer.
A major limitation in the present study is that the serum concentration in the
samples obtained at the baseline examination may not be necessarily representative
of the long term selenium level. This is especially a concern due to fact that the cohort
members were recruited into the study during a five year span, while selenium fortified
fertilizers were introduced in this country in 1986, two years after the study started.
For approximately 62 percent of the cohort members, serum selenium concentrations
were determined based on blood samples collected after 1986; for those subjects, it
is likely that the observed selenium levels may actually be higher than their long-term
selenium levels and might have distorted the association between selenium and risk
of cancer. A repeated measurement of serum selenium on 523 subjects selected from
101
those examined after 1985 was performed in 1990. Of those, 362 persons were
cancer and CHD free both at baseline and follow-up. Age-adjusted correlation between
repeated measures of selenium based on those subjects was 0.27 (p < 0 .0 1 ). This low
correlation indicated that the serum selenium level was relatively unstable over time
partly due to the fact that the dietary intake of selenium in Finland varies from year to
year depending on the ratio of imported (selenium content 100-200/yg/kg) to domestic
(selenium content 10-20 >ug/kg) grain in the diet, and the level of serum selenium only
reflects short-term (days to weeks) selenium balance.13 To explore the influence of
selenium change over time on the selenium-cancer association, separate analyses were
performed on subjects (n = 740) recruited during 1984 to 1985, and those (n = 1,222)
recruited during 1986 to 1989, and the results are summarized in Table 4. An inverse
yet non-significant association was only found among persons examined during 1984-
1985, but no evidence of an inverse association was found among persons examined
during 1986-1989. The lack of significance of the RR estimates among the early
samples was probably due to the small size of the sample. To statistically control for
bias due to the secular change of selenium levels, years of baseline examination were
included in all regression models in the pooled analyses. Although it can not be totally
discarded that change of selenium over time has biased the results even with
adjustment for baseline examination year, the change of selenium over time, however,
would tend to bias the observed association toward the null so that the actual relative
risk would be expected to be larger than those observed. The finding of an interaction
between smoking and serum selenium in the pooled analysis also argues against this
alternative hypothesis.
Cancer generally develops during a very long period of time, with a long
102
subclinical latency. The alternative hypothesis of reversal causation that latent or
subclinical malignancy may have actually lowered the selenium level can not be
excluded based the present study. Previous studies indicated that pre-cancer cases had
a lower serum selenium levels than that of their matched controls.12 Clarification of
the cause-effect role of selenium in the etiology of cancer clearly entail evidence from
randomized intervention study with a very long duration of follow-up, probably
decades, and a very large sample size.
Previous epidemiologic studies of selenium and specific site of cancer indicated
that selenium is protective against development of cancer at some sites, but not
others.33 The majority of cases in the present study was lung cancer (n = 24), followed
by large bowel cancer (n = 17). The small number of cancer cases at specific sites
precludes the attempt to investigate the effect of selenium on risk of specific site
cancers. Follow-up data of the current cohort up to December 1994 is being collected,
it would be very interesting to repeat the analyses with site-specific cancers when the
data become available.
Results from the present study and previous prospective studies of the
relationship between selenium levels and cancer at all sites combined are shown in
Table 5. One suggested explanation for the apparent inconsistency in these various
studies is that a number of these studies suggesting an inverse association were
conducted in Finland, including the present analyses, where the levels of selenium
were substantial lower than that in the United States. In earlier studies from Finland
conducted in the early 1970's, the mean selenium levels were within the range of 50-
60 //g/l, as compared to 136 //g/l estimated from 14 areas in the United States.12,19
In the present study, the mean selenium level was about 102 //g/l for non-cancer
103
subjects. It is interesting to find that even in a Finnish population with relatively high
levels of selenium as compared to that in the early 1970's, an inverse association
between selenium and risk of cancer all sites combined was observed. The categorical
analyses in the our analyses indeed suggested a threshold effect of selenium and may
explain the failure to find a significant association in some of the studies conducted in
the United States and other countries with affluent selenium status.15'17,33-35
The results from our analyses and previous prospective studies conducted in
Finland are suggestive of a protective role of selenium in risk of cancer. In an
intervention trial in Linxian China, a population with a poor status of selenium, a 13
percent reduction in cancer mortality was observed in adults receiving a combination
of selenium, vitamin E, and beta carotene compared with subjects not receiving this
combination of micronutrients.38 However, it can not be concluded that the observed
beneficial effect was due to selenium rather than to vitamin E or beta carotene.
Currently, selenium is the most promising of these three nutrients, in part due to the
recently published null results for the primary end point of lung cancer in the Finnish
cancer prevention trial using beta carotene and alpha tocopherol.39 Intervention trial
involving longer duration of treatment and follow-up may eventually elucidate the role
of selenium in the prevention of cancer.
In summary, the results of our study continue to suggest a potential protective
role of selenium in the prevention of cancer, and the protective effect may be even
more pronounced in smokers. However, the results from this study should not be the
sole basis upon which individual or public health decisions are made. The potential
benefits and adverse effects have to be carefully weighted in the context of cancer
prevention. Our study demonstrates the complexity of the underlying association
104
between selenium and cancer and yet the importance of carefully designed cancer
prevention trials using selenium.
105
Reference
1. Chow CK. Nutritional influence on cellular antioxidant defense system. Am J
Clin Nutr 32:1066-1081, 1979
2. Medina D, Morrison DG. Current ideas on selenium as a chemopreventive
agent. Pathol Immunopathol Res 7:187-199, 1988
3. Arciszewska LK, Martin SE, Milner JA. The antimutagenic effect of selenium
on 7,12-dimethylbena[a]anthracene and metabolites in the Ames salmonella
microsome system. Biol Trace Elem Res 1982;4:259-267.
4. Shamberger RJ, Baughman FF, Kalchert SL. Carcinogen-induced chromosomal
breakage decreased by antioxidants. Proc Natl Acad Sci USA 1973;2:319-322.
5. Jacobs MM, Forst CF, Beams FA. Biochemical and clinical effects of selenium
on dimethylhydrazine-induced colon cancer in rats. Cancer Res 1981 ;41:4458-
4465.
6. Thompson HJ, Becci PJ. Selenium inhibition of N-methyl-N-nitrosourea-induced
mammary carcinogenesis in the rat. J Natl Cancer Inst 1980;65:1299-1301.
7. Marshall MV, Arnott MS, Jacobs MM. Selenium effects on carcinogenicity and
metabolism of 2-acetylaminofluorene. Cancer Lett 1979; 7:331-338
8. Shamberger RJ, Tytko SA, Willis CE. Antioxidants and cancer. Part IV.
Selenium and age-adjusted human cancer mortality. Arch Environ Health
31:231-235, 1976.
9. Schrauzer GN, White DA, Schneider CJ. Cancer mortality correlation studies-lll:
statistical associations with dietary selenium intakes. Bioinorg Chem 7:23-31,
1977.
10. Shamberger R J, Frost DV. Possible protective effect of selenium against human
cancer. Can M ed Assoc J 100:682, 1969.
11. Clark LC, Cantor K, Allaway WH. Selenium in forage crops and cancer mortality
in U.S. counties. Arch Environ Health 46:37-42, 1991.
12. Willett WC, Polk BF, Morris JS. Prediagnostic serum selenium and risk of
cancer. Lancet 2:130-134, 1983.
13. Salonen JT, Alfthan G, Huttunen JK. Association between serum selenium and
the risk of cancer. Am J Epidemiol 120:342-349, 1984.
14. Salonen JT, Salonen R, Lappetelainen R. Risk of cancer in relation to serum
concentrations of selenium and vitamins A and E: matched case-control
analysis of prospective data. Br M ed J (Clin Res Ed) 290:417-420, 1985.
106
15. Peleg I, Morris S, Hames CG. Is serum selenium a risk factor for cancer? Med
Oncol Tumor Pharmacother 2:157-163, 1987.
16. Coates RJ, Weiss IM S, Daling JR. Serum levels of selenium and retinol and the
subsequent risk of cancer. Am J Epidemiol 128:515-523, 1988.
17. Garland M, Morris JS, Stampfer MJ, Colditz GA, Spate VL, Baskett CK, Rosner
B, Speizer FE, Willett WC, Hunter DJ. Prospective study of toenail selenium
levels and cancer among women. J Natl Cancer Inst 87:497-505, 1995.
18. Swanson CA, Longnecker MP, Veillon C. Selenium intake, age, gender, and
smoking in relation to indices of selenium status of adults residing in a
seleniferous area. Am J Clin Nutr 52:858-862, 1990.
19. Knekt P, Aromaa A, Maatela J. Serum selenium and subsequent risk of cancer
among Finnish men and women. J Natl Cancer Inst 82:864-868, 1990.
20. Varo P, Nuurtamo M, Saari E. Mineral element composition of Finnish foods. III.
Annual variations in the mineral element composition of cereal grains. A ctaA gr
Scand 1980 (suppl);22:27-35.
21. Salonen JT. Is there a continuing need for longitudinal epidemiologic research?
The Kuopio Ischemic Heart Disease Risk Factor Study. Ann Clin Res
1988;20:46-50.
22. Malkia E, Impivaara 0 , Maatela J. Physical activity of Finnish adults. (In Finnish
with English summary). Turku, Finland: Social Insurance Institute of Finland,
1988.
23. Alfthan G, Kumpulainen J. Determination of selenium in small volumes of blood
plasma and serum by electrothermal atomic absorption spectrometry. Anal
Chim Acta 1982;140:221-7.
24. Clark LC, Alberts DS. Selenium and cancer: risk or protection? J Natl Cancer
Inst 1995;87(7):473-5.
25. Virtamo J, Valkeila E, Alfthan G. Serum selenium and the risk of cancer. A
prospective follow-up of nine years. Cancer 60:145-148, 1987.
26. Griffin AC. Role of selenium in the chemoprevention of cancer. Adv Cancer Res
1979;29:419-42.
27. Alfthan G, Aro A, Arvilommi H. Selenium metabolism and platelet glutathione
peroxidase activity in healthy Finnish men: effects of selenium yeast, selenite,
and selenate. Am J Clin Nutr 53:120-125, 1991.
28. Luoma P, Korpela H, Sotaniemi EA, Kumpulainen J. Serum selenium,
glutathione peroxidase, lipids and human liver microsomal enzyme activity. A
107
double-blind controlled trial of selenium supplementation. Biol Trace Elem Res
1985;8:13-7.
29. Forman HJ, Rotman El, Fisher AB. Roles of selenium and surfur-containing
amino acids in protection against oxygen toxicity. Lab Invest 49:148-153,
1983.
30. Thomson CD, Robinson MF, Campbell DR, Rea HM. Effect of prolonged
supplementation with daily supplements of selenomethionine and sodium
selenite on glutathione peroxidase activity in blood of New Zealand residents.
Am J Clin Nutr 1982;36:24-31.
31. Levander OA. Selenium: Biochemical action, interactions and some human
health implications. Clinical, biochemical and nutritional aspects of trace
elements. Prasad AS, Ed. AR Liss, NY. 345-368, 1982.
32. Yang GQ, Wang S, Zhou R, Sun S. Endemic selenium intoxication of humans
in China. Am J Clin Nutr 1988;37:382-388.
33. Nomura A, Heilbrun LK, Morris JS. Serum selenium and the risk of cancer, by
specific sites: case-control analysis of prospective data. J N atl Cancer Inst
79:103-108, 1987.
34. Ringstad J, Jacobsen BK, Tretli S. Serum selenium concentration associated
with risk of cancer. J Clin Pathol 4 1 :454-457, 1988.
35. Criqui MH, Bangdiwala S, Goodman DS. Selenium, retinol, retinol-binding
protein, and uric acid. Associations with cancer mortality in a population-based
prospective case-control study. Ann Epidemiol 1:385-393, 1991.
36. Fex G, Pettersson B, Akesson B. Low plasma selenium as a risk factor for
cancer death in middle-aged men. Nutr Cancer 10:221-229, 1987.
37. Kok FJ, de Bruijn AM, Hofman A. Is serum selenium a risk factor for cancer in
men only? [published erratum appears in Am J Epidemiol 126:559, 1981] Am
J Epidemiol 125:12-16, 1987.
38. Blot WJ, Li JY, Taylor PR, et al. Nutrition intervention trials in Linxian, China:
supplementation with specific vitamin/mineral combinations, cancer incidence,
and disease-specific mortality in the general population. J Natl Cancer Inst
85:1483-1492, 1993
39. Alpha-tocopherol, Beta-Carotene Cancer Prevention Study Group. The effect of
vitamin E and beta carotene on the incidence of lung cancer and other cancers
in male smokers. N Engl J Med 1994;330(151:1029-1035.
108
1 0 9
T ab le I. B aseline C h a ra c te ristic s o f A n aly tic S am p le: The Kuopio Ischem ic H eart D isease Study
V ariab les
C ases (n=63>
M ean (SD)
N on-cases (n = 1 8 9 9 )
M ean (SD)
Age (years) 54.4 (3.6) 52.7 (5.0)
Serum selenium (/xg/l) 90.4 (28.2) 10 2 .1 (23.2)
Leisure tim e activity (M E T lir/y) 5 4 9 .1 (7 9 1.6) 630.6 (772.2)
M axim al oxygen uptake (m l/kg/m in) 2 7 .I ( 8 .I) 31.3 (7.7)
Body m ass index (kg/m 2 ) 2 6 .6 (4.2) 26 .9 (3.5)
C igarette sm oking (cig-yr) 372.2 (484.2) 144.9 (301.4)
Alcohol consum ption (g/w eek) 44.5 (119 .9) 97.5 (243.4)
Social econom ic status I2.9 (5.2) 11.1 (4.7)
Fam ily history o f IH D (% ) 49.2 48 .2
Exercise IH D (% ) 3 8 .1 2 9 .8
Living in rural (% ) 3 1.7 28.1
110
T a b le 2. B aseline S e ru m S elen iu m Level a n d In cid en ce o r C a n c e r b y E x am Y e a r a n d S m o k in g S ta tu s
77le K uopio Ischem ic Ile a il Disease Risk F actor Study
B aseline Y ear* M ean + S D (/xg/l) N u m b e r of C ases A t risk %
1984 72.1 + 12.5 22 324 6.79
1985 8 4 .0 ± 16.6 12 416 2 .8 8
1986b 108.9 ± 14.5 9 348 2.5 8
1987 116.6 ± 13.8 10 378 2.65
1988 118.7 + 12.6 3 290 1.03
1989 1 2 1 .0 + 1 5 .6 7 206 3 .4 0
S m o k in g S ta tu s
N o n -sm o k e rs 1 0 3 .4 + 2 3 .2 28 1402 1.99
S m o k e rs 9 7 .5 ± 16.0 35 560 5 .25
Note:
a: Baseline expm was performed from 1984 through 1989.
b: Fertilizer fortified by selenium started in Finland in 1986.
T a b le 3. E stim a te d O d d s R a tio (O K ) o r C a n c e r Tor S c ru m S elen iu m in L ogistic R eg ressio n M odel
The K uopio Ischem ic H eart D isease liisk F actor Study
Serum selenium OR 95% Cl P value
Base model
Age, selenium, exam y e a f 0.38 0.16-0.91 0.03
M ultivariate model
Age, selenium, exam year, covariates1 ’ 0.62 0.26-1.48 0.28
Interaction in multivariate m odcr
Non-smokers 1.10 0.43-2.85
0.01
Smokers 0.32 0.12-0.89
Note:
a:
b:
c:
Baseline exam year was coded us five dummy variables.
Covariates include leisure time physical activity, maximal oxygen uptake, BMI, cigarette smoking,
family history of ischemic heart disease, ischemic 1-CG in exercise, living area and social economic status.
Serum selenium and smoking status interaction estimated in multivariate model.
T a b le 4. M u ltiv a ria te L ogistic O d d s R a tio (O R ) E stim ates o f C a n c e r fo r S eru m S elen iu m , a c c o rd in g lo E xam Y e a r
77te Kuopio Ischemic ! I earl Disease Risk Factor Study (n = l,9 6 2 )
Serum Selenium (23 /ig/l) OR* 95% C l P value
Baseline Exam: 1984-1985 (n=740)
Base model
Age, selenium 0.32 0.16-0.63 0.001
M ultivariate model
Age, selenium, covariatesb 0.57 0.27-1.20 0.13
Interaction in multivariate model
Non-smokers 0.75 0.27-2.13
0.46
Smokers 0.46 0.18-1.19
Baseline Exam: 1985-1989 (n = 1,222)
Base model
Age, selenium 1.11 0.64-1.95 0.70
M ultivariate model
Age, selenium, covariates 1.15 0.66-2.01 0.62
Interaction in multivariate model
Non-smokers 1.17 0.58-2.33
0.96
Smokers j 1.13 0.53-2.18
N ote: i
a: O dds R atio (O R ) estim ated for an increm ent o f 23 (pg/1) o f serum selenium .
b: Covariates included leisure time physical activity, maximal oxygen uptake, BMI, cigarette smoking, alcohol consumption, family history
of ischemic heart disease, ischemic ECG in exercise, living area and social economic status.
113
Table 5. Summary of prospective studies ill' selenium levels in relation to risk of cancer, all sites combined1
Study Population No. of cases Exposure Comparison RR (95% Cl)
Willett et al. (12) U.S. 111 Serum Highest three quintile vs. lowest 0.5 (0.3-0.9)
Salonen et al. (13) Finland 128 Serum Highest (wo tertiles vs. lowest 0.3 (0.1-0.7)
Peleg el al. (15) Geogia 130 Serum Highest vs. lowest quartiles 1.0
Salonen et al. (14) Finland' 51 Serum Highest two tertiles vs. lowest 0.2 (0.03-0.8)
Fex et ul. (36) Sweden (males)2 35 Plasma Highest vs. lowest quintile 0.3’
Kok et al. (37) Holland2
Males
Females
Both
40
29
69
Serum Highest four quintiles vs. lowest
0.4 (0.2-0.8)
0.7 (0.2-2.0)
0.5 (0.3-1.0)
Nomura el al. (33) Japanese men in
Hawaii
280 Serum Highest vs. lowest quintiles 0.84
Virtamo et al. (25) Finland (male) 109 Serum Highest vs. lowest tertiles 0.9 (0.5-1.5)
Coates et al. (16) Washington State 154 Serum Highest vs. lowest quintile 1.0 (0.5-1.8)
Ringstad et al. (34) Norway 60 Serum Highest three quartiles vs. lowest 0.7 (0.3-1.7)
Knekt et al. (19) Finland
Males
Females
597
499
Serum Highest vs. lowest quintile
0.45
0.9*
Criijui el al. (35) North America2 123 Plasma Not given
♦♦
Garland et al. (17) U.S. (females) 934 Toenails Highest vs. lowest quintile
1.2 (0.9-1.7)
Note:
1. Results are for both sexes combined, unless noted.
2. Cancer mortality was the end point.
3. P for trend<0.05.
4. P for trend=0.6l.
5. P for trend<0.001.
6. P for trend=0.60.
** Not given; the authors state that the coefficient was positive and P=0.43.
F i g l . E s t i m a t e d O d d s R a t i o ( O R ) o f C a n c e r f o r
T e r tile s o f S e r u m S e le n iu m b y S m o k i n g S t a t u s
4.0 -i
3.5 -
- - Non-smokers — Q—
3.0 -
Smokers 2.5 -
0.5 -
0.0 -
1st Tertile 2ndTertile 3rdTertile

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A prospective cohort study of morbidity and mortality among middle-aged men in Shanghai, China

PDF

Breast cancer in the multiethnic cohort study: Genetic (prolactin pathway genes) and environmental (hormone therapy) factors

PDF

Genetic risk factors in breast cancer susceptibility: The multiethnic cohort

PDF

Artery wall injury and LDL-cholesterol in early atherosclerosis: The Los Angeles atherosclerosis study

PDF

Family history, hormone replacement therapy and breast cancer risk on Hispanic and non-Hispanic women, The New Mexico Women's Health Study

PDF

Risk factors for breast cancer according to estrogen and progesterone receptor status

PDF

Dietary fiber intake and atherosclerosis progression: The Los Angeles Atherosclerosis Study

PDF

A large-scale genetic association study of prostate cancer in a multi-ethnic population

PDF

Colorectal cancer risks in Singapore Chinese: Polymorphisms in the insulin-like growth factor-1 and the vitamin D receptor

PDF

Colorectal cancer: genomic variations in insulin-like growth factor-1 and -2

PDF

Polymorphism of CYP2E1 gene and the risk of lung cancer among African-Americans and Caucasians in Los Angeles County

PDF

Risk factors associated with moderate and serious injuries attributable to the 1994 Northridge earthquake, Los Angeles County

PDF

A case/parental/sibling control study of Ewing's sarcoma/peripheral primitive neuroectodermal tumor (pPNET)

PDF

Dietary fats, fat metabolizing genes, and the risk of breast cancer

PDF

Descriptive epidemiology of thyroid cancer in Los Angeles County, 1972-1995

PDF

Androgens and breast cancer

PDF

BRCA1 mutations and polymorphisms in African American women with a family history of breast cancer identified through high throughput sequencing

PDF

A case-control study of passive smoking and bladder cancer risk in Los Angeles

PDF

Risk factors, sequellae and intervention studies for the treatment of diabetes mellitus in Native American populations

PDF

Determinants of mammographic density in African-American, non-Hispanic white and Hispanic white women before and after the diagnosis with breast cancer

Asset Metadata

Creator Li, Li (author)

Core Title Antioxidants and risk of myocardial infarction and cancer in a cohort of middle-aged finnish men: the kuopio ischemic heart disease risk factor study

Degree Doctor of Philosophy

Degree Program Preventive Medicine - Cardiovascular Epidemiology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag health sciences, nutrition,health sciences, oncology,health sciences, public health,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Dwyer, James H. (committee chair), Johnson, Andy (committee member), Menski, Mark (committee member), Stohlman, Steve (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-91279

Unique identifier UC11354775

Identifier 9636351.pdf (filename),usctheses-c17-91279 (legacy record id)

Legacy Identifier 9636351.pdf

Dmrecord 91279

Document Type Dissertation

Rights Li, Li

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA