1222 Am J C/in Nuir 1991;53:1222-9.. Printed in USA. © 1991 American Society for Clinical Nutrition

Effects of antioxidant supplementation on platelet function:a randomized pair-matched, placebo-controlled, double-blindtrial in men with low antioxidant status�3

Jukka T Salonen, Riitta Salonen, Kari Seppanen, Soili Rinta-Kiikka, Maija Kuukka,Heikki Korpela, Georg Alfihan, Marjatta Kantola, and Wolfgang Schalch

ABSTRACT We investigated the effect on platelet function

ofsupplementing men with low antioxidant status with 600 mg

ascorbic acid, 300 mg a-tocopherol, 27 mg /3-carotene, and 75

�tg selenium in yeast daily. Eighty men were randomly assigned

in pairs (matched for smoking, baseline antioxidant status, and

time and day ofentry) by use ofa double-blind design to receive

supplement or placebo for 5 mo. Compared with 39 control

subjects, 39 antioxidant-supplemented men experienced the fob-

bowing net reductions during the double-blind period: 20% (P

= 0.012) in serum lipid peroxides, 24% (P = 0.035) in ADP-

induced platelet aggregation, 42% (P = 0.040) in the rate of ATP

release during aggregation, 5 1% (P = 0.018) in serum (platelet-

produced) thromboxane B2 , and 29% (P = 0.024) in plasma /3-

thromboglobulin concentration. The data support our hypothesis

that antioxidant supplementation of men with low antioxidantstatus and high fat intake reduces lipid peroxidation, the capacity

of platelets to aggregate and to produce thromboxane A2 , andin vivo platelet activation. Am J Clin Nutr 199 l;53:

1222-9.

KEY WORDS Antioxidants, clinical trials, lipid peroxida-tion, platelet aggregability

Introduction

One prospective population study (1) and other case-controlstudies (2-4) observed an association between low selenium sta-

tus and ischemic heart disease (IHD). Cross-population studies

showed an inverse relationship between plasma vitamin C (5),

E, and A (6) concentrations and coronary mortality, but thesefindings were not confirmed in prospective within-population

studies (7-9). Although not all studies supported the relationship

between low antioxidant status and IHD, the information aboutthe potential mechanisms whereby antioxidant deficiency could

increase the risk of IHD is essential (10).

Prostacyclin synthesis in the arterial endothelium is, in cell

cultures, irreversibly inhibited by lipid peroxides (1 1). These are

generated either enzymatically by lipoxygenases and cycbooxy-

genase or by the influence of free radicals on cell-membrane

lipids. Intracellular lipid-peroxide concentration is regulated by

glutathione peroxidase, superoxide dismutase, and catabase en-

zymes as well as by the chain-breaking antioxidants a-tocopherol,

ascorbic acid, and fl-carotene. Thus, theoretically, a deficiency

in these antioxidants leads to reduced prostacyclin synthesis (12),

which can promote platelet aggregability. In an in vitro study,

ascorbic acid stimulated prostaglandin 12 (PGI2) synthesis by

endothelial cells whereas a-tocopherol and selenium, an integral

part ofgbutathione peroxidase, inhibited the platelet productionof thromboxane A2 (13).

There is clinical evidence suggesting that administration ofvitamin E and calcium could prevent postoperative throm-

boembolism (14). Marked platelet hyperaggregability was ob-

served in children with vitamin E deficiency, with complete re-

versal after vitamin E supplementation (15). Previous clinical

trials of the effects of ct-tocopherob (16-21) and ascorbic acid(22, 23) on platelet function have, however, produced inconsis-

tent results. Indicators of platelet activation in vivo have notbeen measured in any of the previous studies. We observed

markedly enhanced platelet aggregability at low serum selenium

concentrations in two independent, random population samples

of middle-aged Eastern Finnish men (24, 25). This inverse as-

sociation persisted even after other major determinants of plateletaggregability were allowed for.

The purpose of the present study was to test the hypothesis

that antioxidant supplementation of men with low antioxidant

status and relatively high intake of saturated fatty acids will re-

duce serum lipid peroxides and platelet activity as expressed byplatelet aggregability, serum thromboxane B� concentration, and

plasma concentrations of platelet-specific proteins.

Subjects and methods

Study subjects

The subjects ofthis trial were screened in a survey ofa random

population sample of men aged 54 y from eastern Finland at

I From the Departments ofCommunity Health and General Practice,and Chemistry and the Research Institute of Public Health, University

of Kuopio, Kuopio, Finland; the Kuopio Research Institute of ExerciseMedicine, Kuopio, Finland; the Department of Biochemistry, The Na-tional Public Health Institute of Finland, Helsinki, Finland; and theDepartment of Human Nutrition and Health, F Hoffman-La RocheLtd, Basle, Switzerland.

2 Supported by the Finnish Academy (grant for the KIHD study), the

University of Kuopio, and F Hoffman-La Roche Ltd.3 Address reprint requests to JT Salonen, University of Kuopio, P0

Box 6, 7021 1 Kuopio, Finland.

Received May 17, 1990.

Accepted for publication August 19, 1990.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

ANTIOXIDANTS AND PLATELET FUNCTION 1223

examination in the Kuopio Ischemic Heart Disease Risk Factor

Study (KIHD) (24-27). A total of 1 327 men was screened be-

tween March 1984 and June 1987. Plasma a-tocopherob was

not used as an inclusion criterion, but all subjects had relatively

low plasma a-tocopherol values (22. 1 ± 6.4 �tmol/L, I ± SD;

range 9.1-39.9 �mol/L) at the beginning ofthe study. Of these,

176 men with plasma ascorbic acid < 41.5 �tmol/L and serum

selenium < 1 .7 �tmol/L were invited to a reassessment of an-

tioxidant status in June and July 1987. Blood samples were ob-

tamed from 150 ofthese men; 26 did not show up, the majority

because of deaths and emmigration. Of the 150 men, 1 12 were

eligible for the study. Twelve men refused to participate in the

trial and 20 were excluded because ofillness or medications that

could have affected platelet function. The remaining 80 men

were entered into the trial. None ofthese men took antioxidants,

vitamin supplements, or prostaglandin-inhibiting drugs (eg,

acetosalicylic acid) before the trial. One subject randomly as-

signed to antioxidant supplementation discontinued the trial af-

ter 2 1 d of supplementation. The pairwise control was also ex-

cluded from the study population. Seventy-eight subjects com-

pleted the trial and underwent all assessments.

The mean baseline daily dietary intake offats was 122 g (40%

of total energy intake). There were no statistically significant

differences between the supplemented and the placebo men in

the changes in smoking, alcohol consumption, physical activity,

or any of the 60 nutrients assessed.

The study protocol was approved by the ethical committee

of the University of Kuopio, Kuopio, Finland.

Study design and intervention

The subjects were stratified by smoking status. There were 54

nonsmokers and 26 smokers. Subjects were paired in these strata

by their baseline values ofplasma ascorbic acid and a-tocopherol

and serum selenium. The members ofthe 40 pairs were randomly

assigned, one to antioxidant supplementation and the other to

placebo. To avoid confounding effects of seasonal variability,

both members ofeach pair were entered in the trial on the same

day and time. Two pairs were entered on each Monday and

Friday between August 1987 and January 1988. The purpose

ofthe extended subject entry over a period of6 mo was to balance

out the seasonal variation in vitamin intakes.

Each subject started the trial with a 1-mo placebo period.

Baseline refers to the end ofthis 1-mo lead-in period. The subjects

were instructed to keep diet, physical activity, and the amount

ofsmoking constant during the trial and not to use any vitamin

or mineral supplements. Four-day dietary recording was carried

out at the end of the trial as described in detail previously (27).

The blood antioxidant status was assessed at the beginning and

at the end of the lead-in period. After the 1-mo lead-in the sub-

jects receiving supplementation were treated with the antioxidant

preparation for 5 mo and those receiving placebo were given

visually identical placebo capsules for 5 mo.

The dosage was three capsules daily in both groups. An an-

tioxidant capsule contained 200 mg ascorbic acid, 100 mg a-

tocopherol, 9 mg /3-carotene, and 25 jzg selenium in selenium-

enriched yeast. The filling material was peroxide-free arachis oil.

Antioxidant status was reassessed at the end of the trial.

Weight, serum lipids, serum lipid peroxides, blood glutathione

peroxidase, platelet aggregability, serum thromboxane B2, and

platelet-specific proteins were measured at the end of the bead-

in period and at the end of the trial.

An 8-wk bioavailability study was carried out in 10 volunteers

who were not subjects ofthe present trial. The same dosing was

used as was planned for the trial. In this pilot study, mean plasma

ascorbic acid rose from 63.6 to 79.5 .tmol/L (25% increase), a-

tocopherob from 24.4 to 47. 1 zmol/L (93% increase), and /3-

carotene from 0.7 to 3.5 �mol/L (400% increase). The concen-

trations ofplasma ascorbic acid and a-tocopherol reached a pla-

teau in 1 wk, whereas plasma /3-carotene rose linearly for 3 wk

and more slowly for another 2 wk.

Blood collection

The subjects were asked not to take any antiinflammatory

analgesics during the 2 wk before blood sampling and to abstain

from alcohol for 1 wk before blood sampling. The subjects were

also asked to fast and not to smoke for 12 h before blood sam-

pling. Blood sampling took place after the subjects had been

supine for 30 mm. Blood was collected without a tourniquet by

a flawless venipuncture. Terumo Venoject VT-050X88 vacuum

tubes (Terumo Corporation, Tokyo, Japan) and Terumo 18 G

needles (Terumo Europe, Leuven, Belgium) were used in drawing

blood samples. The tubes contained 1 vol of 3. 1% trisodium

citrate as the anticoagulant to 9 vol of blood.

Biochemical determinations

Plasma concentrations of a-tocopherob and /3-carotene were

determined with HPLC as described by Vuilleumier et al (28)

and plasma ascorbic acid was determined by using a fluorometric

method (29). Serum selenium concentrations were determined

with the graphic-furnace atomic-absorption spectrophotometric

technique, Zeeman background correction, and pyrolyticabby

coated graphite tubes with a platform (Zeeman 5000 atomic

absorbtion spectrophotometer, HGA 500, Perkin-Elmer, Nor-

walk, CT) (1).

The main serum lipoprotein fractions (very-low-density,

(VLDL)bow-density (LDL)and high-density (HDL) lipoproteins)

were separated within 3 d of blood sampling by a combination

of ultracentrifugation and precipitation (30). The cholesterol

content (mmol/L) of all lipoprotein fractions was determined

enzymatically (cholesterol CHOD-pap method, Boehringer

Mannheim, Mannheim, FRG).

Serum lipid peroxides were measured fluorometrically by using

the thiobarbituric acid (TBA) reaction (31). In this method TBA-reacting substances other than lipid peroxides were eliminated

by precipitating lipid peroxides along with serum proteins in the

phosphotungstic acid-sulfuric acid system. The determinants

were done from deep-frozen (at -70 #{176}C)serum samples within

24 mo of blood sampling.

Blood glutathione peroxidase activity was determined from

prediluted (1:9) frozen (at -70 #{176}C)whole-blood samples with

an automated modification ofthe method described by Levander

et al (32) within 3 mo of blood sampling. Catalase activity of

red blood cells was inhibited with Drabkins’ solution (33).

For measuring platelet aggregation adenosine 5’-diphosphatesodium salt (ADP) and adenosine 5’-triphosphate disodium salt(ATP; Sigma Chemical Company, St Louis) were dissolved in9 g NaC1 in 1 L distilled water. Silicon-treated SMI capillary

tubes and an SMI micropipettor (Scientific Manufacturing In-

dustries, Miami) were used for adding ADP and ATP to the

reaction mixture. All plasma samples were handled with plastic

Pasteur pipettes (Orioba Oy Probab, Helsinki, Finland). Coulter

thrombocyte control CTC-3 (Coubter Electronics Ltd, Luton,

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

1224 SALONEN ET AL

TABLE 1

Baseline and posttrial values and change during the trial in the plasma antioxidative vitamins, serum selenium, lipids, and indicators of lipid

peroxidation in the 39 p airs of supplem ented and place bo-treated mena

Net percent P forSupplemented Placebo change in the

supplementedmen

intrapairdifference

in changeBaseline Posttrial Change Baseline Posttrial Change

Plasma ascorbic acid

(�mol/L) 28.39 ± 2.32 75.52 ± 2.50 47.13 ± 3.S2t 31.22 ± 2.27 27.25 ± 2.89 -3.97 ± 3.18 +274 ± 52 <0.001Plasma a-tocopherol

(�mol/L) 23.22 ± 0.95 47.37 ± 0.87 24.15 ± l.S5t 20.90 ± 0.74 18.58 ± 0.79 -2.32 ± l.08f +73 ± 8 <0.001a-Tocopherol:(VLDL

+ LDL cholesterol)(�mol/mmol) 5.38 ± 0.16 10.96 ± 0.34 5.58 ± 0.28t 5.15 ± 0.18 5.18 ± 0.23 0.03 ± 0.18 +101 ± 9 <0.001

Plasma fl-carotene

(�mol/L) 0.57 ± 0.05 3.71 ± 0.37 3.14 ± 0.35t 0.67 ± 0.05 0.50 ± 0.05 -0.17 ± 0.03t +656 ± 75 <0.001Serum selenium

(�tmol/L) 1.58 ± 0.03 1.80 ± 0.03 0.22 ± 0.03t 1.60 ± 0.03 1.51 ± 0.03 0.09 ± 0.02t +19.5 ± 2.5 <0.001

Serum total cholesterol(mmol/L) 5.6 ± 0.16 5.5 ± 0.15 -0.1 ± 0.10 5.5 ± 0.15 5.3 ± 0.15 -0.2 ± 0.08 - NS

Serum LDL cholesterol(mmol/L) 3.8 ± 0.15 3.7 ± 0.13 -0.1 ± 0.1 1 3.7 ± 0.14 3.5 ± 0.13 -0.2 ± 0.08 - NS

Serum HDL

cholesterol (mmol/L) 1.28 ± 0.05 1.21 ± 0.06 -0.07 ± 0.05 1.33 ± 0.04 1.31 ± 0.06 -0.02 ± 0.06 - NSBody weight (kg) 80.72 ± 2.05 81.51 ± 2.13 0.80 ± 0.33f 81.29 ± 1.58 81.69 ± 1.64 0.40 ± 0.31 - NS

Serum lipid peroxidesLI

thrombocytes) 1.74±0.12 1.31 ±0.08 -0.43±0.lOt 1.46±0.08 1.36±0.07 -0.10±0.08 -19.9±7.2 0.012Blood glutathione

peroxidase(mU/g Hb) 1.64 ± 0.04 1.73 ± 0.05 0.09 ± 0.03� 1.62 ± 0.05 1.59 ± 0.05 -0.03 ± 0.05 +7.3 ± 3.6 0.032

a � � SEM. VLDL, very-low-density lipoprotein; LDL, low-density lipoprotein; HDL, high-density lipoprotein.

tf� Significantly different from zero: tP < 0.001, fP < 0.05, §P < 0.01.

UK) was used as a control when platelets in plasma samples

were counted.

The blood tubes were mixed well by gentle inversion and werekept at room temperature for 30 mm. The blood samples wereagain mixed gently and centrifuged at +20 #{176}Cfor 10 mm at 250x g. Platelet-rich plasma (PRP) was collected to a polypropylene

tube. For platelet-poor plasma (PPP) preparation the same blood

tubes were centrifuged once more at room temperature for 10

mm at 4000 X g and PPP was handled in the same way as PRP.

Thrombocytes in PRP and PPP were counted with a Throm-

bocounter C (Coulter Electronics Ltd).Thrombocyte aggregation in samples of PRP was measured

as a function of time by the Whole Blood Aggro-Meter Model

500 (Chronobog Corporation, Havertown, PA). Changes in both

optical density, as measured with changes in electrical impedance

(mV), and ATP secretion (luminescence) were recorded simul-

taneously. The measurement of aggregation was described in

detail previously (34). Luciferase-luciferin reagent (100 �tL;

Chronolog Corporation) was added to the sample of PRP (1.0

mL) and the PPP sample was adjusted with 100 �tL 9 g NaCb/

L to the same basic optical density. The aggregating agent, 5 �tL1.0 mmol ADP/L, was injected into the PRP samples to start

the reaction. Final concentration ofADP was 4.52 �tmob/L. Each

subject’s samples before and after the double-blind period were

always handled with the same dose ofaggregating stimulus. After

a 10-mm run the luminescence channel was calibrated by adding

4 �iL 2.004 mmol ATP/L into the PRP samples (final ATP con-

centration 7.26 zmol/L). The measurement of platelet aggreg-

ability was carried out within 3 h of blood drawing.

For analysis ofthromboxane B2 in serum (produced by plate-

lets during in vitro incubation), 1 mL blood was taken and al-

bowed to clot at 37 #{176}Cfor 30 mm. After centrifugation at 500

x g for 15 mm at 4 #{176}C,the serum was frozen at -20 #{176}Cuntil

assayed. The determinations were carried out after 20-26 mo

ofstorage. The extraction ofthromboxane B2 was performed by

Powell’s method (35). Thromboxane B2 concentration in the

extract was determined by using reverse-phase HPLC. In our

laboratory this method gives > 50% lower values of serum

thromboxane B2 than do the commercially available radioim-

munoassay (RIA) kits and thus may be more specific.Plasma /3-thromboglobulin (/3-TG) was determined by a com-

mercial RIA (Amersham, Amersham, UK). The method depends

on competition between sample fl-TG and ‘251-labeled /3-TG for

a limited number of binding sites on a fl-TG-specific antibody.

Plasma samples (25 ML), standard (25 .tL), ‘251-labeled /3-TG

(100 giL; radioactivity 7.4 MBqJL), and anti-fl-TG rabbit serum

(100 �.tL) were pipetted into the polystyrene tubes. The reaction

solutions were mixed and incubated for 1 h at room temperature.

After incubation 250 �iL saturated ammomum sulphate solution

were added into the tubes. The tubes were mixed and centrifuged

10 mm at 1 500 X g. Supernatants were removed and radioac-

tivity in precipitates was measured by a MiniGamma �y counter

(LKB Wallac, Turku, Finland). All tests were run in duplicate

and the results were calculated by using the spline function (36).

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

ANTIOXIDANTS AND PLATELET FUNCTION 1225

TABLE 2Baseline and posttrial values and change during the trial in the indicators of in vitro and in vivo platelet function in 39 pairs of supplemented and

placebo-treated mena

Supplemented Placebo

Net percent

change in the

supplemented

men

P for

intrapair

difference

in changeBaseline Posttrial Change Baseline Posttrial Change

Extent (amplitude) of

secondary aggregationphase (mV) 19.8 ± 1.5 15.6 ± 1.8 -4.2 ± 1St 16.5 ± 1.8 16.9 ± 2.2 0.5 ± 2.2 -23.7 ± 13.3 0.035

Velocity (slope) ofsecondary aggregationphase(mV/s) 0.177±0.016 0.130±0.021 -0.048±0.Ol6t 0.143±0.017 0.132±0.019 -0.011 ±0.019 -20.9± 14.0 0.028

Maximum ATP release(�imol/L) 3.10 ± 0.38 2.22 ± 0.33 -0.88 ± 0.40� 2.36 ± 0.30 2.49 ± 0.42 0.13 ± 0.38 -32.6 ± 17.9 0.040

Velocity (slope) of ATP

release

(�mol. L’ .s’) 0.053 ± 0.009 0.034 ± 0.007 -0.020 ± 0.009� 0.038 ± 0.006 0.040 ± 0.007 0.002 ± 0.007 -41.5 ± 21.5 0.034

Serum thromboxane B2

Platelet-count adjustedI

thrombocytes) 121 ± 13 76 ± 18 -45 ± 21t 138 ± 18 155 ± 18 17 ± 26 -51.1 ± 27.2 0.018

Unadjusted (nmol/L) 287 ± 32 152 ± 27 -135 ± 361 372 ± 43 407 ± 42 36 ± 56 -59.6 ± 23.2 0.003

Plasma �3-thromboglobulin

Platelet-count adjusted(�ig/10”thrombocytes) 27.0 ± 3.1 19.5 ± 1.8 -7.5 ± 3.2t 20.5 ± 1.9 20.9 ± 4.0 0.3 ± 4.4 -29.1 ± 20.2 0.024

Unadjusted (�igIL) 64.3 ± 8.5 43.9 ± 4.4 -20.4 ± 7.8t 55.6 ± 4.3 54.6 ± 1 1.3 -1.0 ± 12.2 -30.2 ± 22.5 0.048

Plasma platelet factor 4

Platelet-count adjusted

(�g/l0”

thrombocytes) 6.97 ± 2.7 5.03 ± 0.7 -1.93 ± 2.7 3.79 ± 0.4 4.85 ± 0.9 1.06 ± 1.0 -43.0 ± 41.3 0.132

Unadjusted (�tgIL) 18.2 ± 8.2 1 1.6 ± 1.7 -6.6 ± 8.1 10.1 ± 0.9 12.4 ± 2.3 2.3 ± 2.6 -48.9 ± 46.7 0.134

aI± SEM.

ff1 Significantly different from zero: tP < 0.01, tP < 0.05, §P < 0.001.

Plasma platelet factor 4 (PF4) was measured by using an RIA

(Abbott Laboratories, Wiesbaden-Delkenheim, FRG). The

method is a competitive RIA in which nonradioactive PF4 in

plasma competes with a constant amount of ‘25I-PF4 reagent

solution (125 giL, radioactivity 13.0 MBCIJL). Anti-PF4 goat

serum (125 jzL) was pipetted into the polystyrene tubes. Tube

contents were mixed and the tubes were incubated at room tem-

perature for 2 h. After incubation 500 �tL ammonium sulphatesolution (73% saturated) was added into the tubes. Tube contents

were mixed and then allowed to stand 10 mm before centrifu-

gation. All tubes except total-count tubes were centrifuged at

room temperature at 1 500 X g for 20 mm. Supernatants were

removed by decanting and radioactivity in precipitates was

measured by a ‘y counter. All tests were run in duplicate and

the results were calculated by using the spline function (36).

Determinations of plasma /3-TO and PF4 from deep-frozen

(at -70 #{176}C)plasma samples were carried out within 3 mo of

the end of the trial.

Statistical methods

The association ofantioxidant status with indicators of platelet

activation at the baseline was tested with two-way SPSS-X anal-

ysis ofcovariance (37). Because subjects were matched pairwise,

statistical analysis was based on the t test for paired samples.

The a priori power to detect a difference of 5 mV (‘-25%) in

the ADP-induced platelet aggregation between the supplemented

and the placebo subjects in the change from end of bead in to

end of trial was > 0.8 at one-tailed a of 0.05, assuming an SD

of 10 mV for this platelet-aggregation variable (based on KIHD

data, ref 24). The net change was computed by subtracting the

percentage change in the control subject from the percentage

change in the corresponding paired supplemented subject. The

statistical significance levels for differences in change between

the supplemented and the placebo-treated men are based on

one-sided tests. The matching of subject pairs with regard to

smoking was done to enable the comparison of the treatment

effects between smokers and nonsmokers, even with suboptimal

study power.

Serum thromboxane B2 and plasma /3-TO and PF4 concen-

trations were adjusted for blood platelet count by computing

ratios. Both adjusted and unadjusted concentrations are pre-

sented for main results.

The statistical significance of linear trends in changes in in-

dicators of platelet activation over quartiles of changes in the

supplemented vitamins and selenium was tested with one-way

analysis of variance. The pairing ofsubjects was broken for this

analysis.

Results

To analyze the cross-sectional association of antioxidant vi-

tamin and selenium status with in vivo platelet activation, the

mean values of serum thromboxane B2 and plasma /3-TO con-

centrations at entry to the double-blind period were computed

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

1226 SALONEN ET AL

end of the washout period. Plasma ascorbic acid concentration change in the placebo group (Table 2).

TABLE 3

Baseline values and change during the trial in the antioxidative vitamins and serum lipid peroxides, and indications of platelet activation in the

supplemented and placebo-treated men, by smoking statusa

Nonsmokers (2 6 pairs) Smokers (13 pairs)

Net change Net change

Supplemented Placebo Percent Pt Supplemented Placebo Percent Pt

Plasma ascorbic acidBaseline (�mol/L) 31.2 ± 3.1 31.5 ± 2.6 22.8 ± 2.7 30.7 ± 4.5Change (%) 273 ± 7 lf - 1 ± 13 274 ± 72 0.001 256 ± 44j - 18 ± 14 274 ± 46 <0.00 1

Plasma a-tocopherol: (VLDL

+ LDL cholesterol)Baseline (�tmo1/L) 5.60 ± 0.19 5.20 ± 0.22 4.99 ± 0.27 5.06 ± 0.30Change (%) 102 ± 9� 2 ± 5 100 ± 10 <0.001 105 ± 18t 1 ± 3 104 ± 19 <0.001

Plasma fl-caroteneBaseline (�mol/L) 0.60 ± 0.07 0.77 ± 0.07 0.50 ± 0.09 0.48 ± 0.08Change (%) 668 ± 82� -25 ± 5� 693 ± 82 <0.001 560 ± 159t -23 ± 8� 583 ± 160 0.002

Serum seleniumBaseline (�imol/L) 1.59 ± 0.04 1.59 ± 0.04 1.57 ± 0.05 1.62 ± 0.06Change (%) 1 1 ± 3j -6 ± 2� 17 ± 3 <0.001 19 ± 2f -6 ± 2� 24 ± 3 <0.001

Serum lipid peroxidesBaseline (zmol/l0�

thrombocytes) 1.75 ± 0.18 1.48 ± 0.10 1.74 ± 0.10 1.43 ± 0.13Change(%) -19±8t -9±7 -11 ± 11 0.157 -35±8� -2±7 -35± 11 0.006

Serum thromboxane B2Baseline (�imol/l0�

thrombocytes) 105 ± 15 132 ± 22 151 ± 23 150 ± 34Change(%) -20±26 5± 19 -25±32 0.188 -60± l6� 24±39 -84±42 0.019

Plasma fl-thromboglobulinBaseline(�g/10� thrombocytes) 25.1 ± 2.9 21.6 ± 2.0 30.7 ± 7.4 18.6 ± 3.9Change(%) -20± 14 15±29 -35±32 0.040 -40±2111 -28±22 -23±30 0.178

aj± SEM.

t One tailed, derived from paired t tests. All P values were consistent (at the same significance level) with those from Wilcoxon matched-pairssigned-rank tests.

t�II Significantly different from zero: �P < 0.001, §P < 0.01, lIP < 0.05.

for all subjects, grouped by plasma vitamin and selenium con-

centrations (broken down at medians), with an additional

breakdown by smoking status (no vs yes) in two-way covariance

analysis. Both serum thromboxane B2 (by 46%) and plasma /3-

TO (by 12%) were nonsignificantly higher in smokers than innonsmokers. When adjustments were made for body mass index

(expressed in kg/m2), history of or presence of current IHD (no

vs yes), serum LDL cholesterol (mmol), and coffee consumption

(g/wk) as covariates and for smoking status as a binary factor,

the mean serum thromboxane B2 concentration was higher in

the lower half of plasma ascorbic acid distribution (385 vs 271

nmol/L, two-tailed P = 0.039). There were similar nonsignificant

trends for the ratio ofplasma a-tocopherob to cholesterol (VLDL

plus LDL), /3-carotene, and serum selenium. Identically adjusted

plasma /3-TO concentration was higher (75.4 �g/L) in the bower

half than in the upper half(53.7 �g/L) of the ascorbic acid dis-

tribution (two-tailed P = 0.041) in a two-way covariance analysis.

This difference was greater in smokers than in nonsmokers (P

= 0.030 for interaction).

The baseline and posttriab means ofthe antioxidative vitamin

and selenium concentrations, serum lipoproteins, serum lipid

peroxides, and blood glutathione peroxidase are presented in

Table 1. The supplemented and the placebo-treated subjects had

comparable plasma vitamin and selenium concentrations at the

rose in the supplemented subjects by 166% (range -4% to

+ 1 358%), a-tocopherol rose by 104% (range 28-2 16%), /3-car-

otene rose by 549% (range 11-1938%), and serum selenium

concentration rose by 13% (range - 13% to +57%) (Table 1). In

the placebo-treated subjects neither plasma ascorbic acid nor

the ratio ofplasma a-tocopherob to cholesterol (VLDL plus LDL)

changed significantly whereas plasma /3-carotene decreased by

26%, plasma a-tocopherob decreased by 1 1%, and serum sebe-

nium decreased by 6%.

There were no statistically significant changes in the serum

total-, LDL-, or HDL-cholesterob concentrations in either

the supplemented or the placebo-treated men during the

supplementation (Table 1). Mean body weight increased by 1%

in the supplemented subjects. Platelet-count-adjusted serum

lipid peroxides were reduced by 25% in the supplemented

and by 7% in the placebo-treated subjects. Blood glutathi-

one peroxidase increased by 6% in the supplemented subjects

and did not change significantly in the placebo-treated men

(Table 1).

The extent of ADP-induced platelet aggregation was reduced

during the antioxidant supplementation by 21% (P < 0.01) (Ta-

ble 2) whereas there was practically no change in the placebogroup. The average velocity ofADP-induced platelet aggregation

declined 27% (P < 0.01) in the supplemented subjects, with no

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

ANTIOXIDANTS AND PLATELET FUNCTION 1227

TABLE 4

Change of serum lipid peroxides, serum thromboxane B2, and plasma fl-thrombogbobulin in concentration quartiles (I-IV) ofchange in vitamins,selenium, and lipid peroxidesa

Quartile of I II III IV rtP for

linear trend

�mo//1O” thrombocytes

Serum /ipid peroxide change

Plasma ascorbic acid change

Plasma a-tocopherol change

Plasma fl-carotene change

Serum selenium changeSerum lipid peroxide change

0.01 ± 0.1 1

-0.26 ± 0.09

-0. 18 ± 0. 10

-0.17 ± 0.10-

-0.42 ± 0. 17 -0.22 ± 0.09

0.07 ± 0. 10 -0.37 ± 0. 10

-0.09 ± 0.09 -0.43 ± 0. 19

-0.01 ± 0.12 -0.54 ± 0.15- -

-0.43 ± 0. 13-0.50 ± 0. 18

-0.35 ± 0. 1 1

-0.33 ± 0.12-

-0.28-0.3 1

-0.06

-0.27

-

0.0290.035

0.102

0.003

-

Serum thromboxane B2 changePlasma ascorbic acid changePlasma a-tocopherol change

Plasma fl-carotenechange

Serum selenium changeSerum lipid peroxide change

3 1 ± 4375 ± 35-1 ± 34

25 ± 38

63 ± 22

2 1 ± 2 1 -7 1 ± 2 1

-4 1 ± 3 1 -69 ± 28

15 ± 40 -7 ± 35

-17 ± 34 -35 ± 24

30 ± 29 -8 ± 29

-29 ± 39

-20 ± 30

-62 ± 17

-17 ± 36

-8 ± 29

-0.28

-0.24

-0.23

-0.18

-0.09

0.025

0.002

0.041

0.084

0.330

Plasma fl-thromboglobulin changePlasma ascorbic acid changePlasma ct-tocopherol change

Plasma fl-carotene change

Serum selenium changeSerum lipid peroxide change

0.6 ± 5.9-4. 1 ± 1.8

-8.4 ± 3.3

6.0 ± 8.1-12.7 ± 4.8

- 1 .9 ± 7.0 - 1.8 ± 4. 14.6 ± 8.6 -9.5 ± 5.0

3. 1 ± 8.7 -8.3 ± 3.0

-3.5 ± 4.7 -12.8 ± 4.7

-6.2 ± 2.3 1.1 ± 6.6

- 1 1.6 ± 4. 1-5.6 ± 4.0

-0.9 ± 4.0

-3.5 ± 2.44.7 ± 7.0

-0. 18-0. 14

0.06

-0.150.24

0.071

0.228

0.3360.0610.008

ai+SEM

t Simple coefficient of correlation between changes in antioxidants and changes in lipid peroxides, serum thromboxane B2 , and plasma fl-throm-boglobulin.

The maximum ATP release, as measured by the maximal

concentration during ADP induction, decreased by 28% (P

< 0.05) and the average velocity of ATP release decreased by

38% (P < 0.05) in the supplemented men whereas no changes

occurred in the placebo group (Table 2). The differences in

changes in all four indicators of in vitro platelet aggregability

between the supplemented and placebo subjects were statistically

significant (P < 0.05).

Platelet-count-adjusted serum thromboxane B2 concentra-

tions were reduced by 37% P < 0.05; unadjusted by 47%, (P

< 0.001) in the supplemented men and did not change signifi-

cantly in the placebo-treated men. The difference in thrombox-

ane B2 change between the supplemented and placebo-treated

men was statistically significant (P = 0.018 for adjusted and P= 0.003 for unadjusted). The plasma concentrations of both

platelet-specific proteins, which indicate the platelet activity in

vivo, decreased during supplementation: platelet-count-adjusted

/3-TO by 28% (P < 0.05; unadjusted by 32%, (P < 0.01) andPF4 by 28%. The latter change did not reach statistical signifi-

cance because of the large variability of baseline values among

the supplemented men (Table 2). The difference in change be-

tween the supplemented and placebo-treated men was statisti-

cally significant (P < 0.05) for /3-TO but not for PF4.

In the baseline measurements for one supplemented subject

and for none of the control subjects, the ratio of plasma /3-TO

to PF4 was < 2, indicating a possibility of cx vivo platelet ac-

tivation. In the end-of-trial measurement this was the case for

three supplemented and three control subjects. For 32 pairs the

ratio was > 2 in both assessments. In this subgroup the mean

thromboxane B2 fell in the supplemented subjects from 276 ± 32

(1 ± SEM) to 1 64 ± 32 nmol/L (by 40.6%; P < 0.0 1 for change)

and increased in the control subjects from 403 ± 49 to 411

± 49 nmol/L, giving a net reduction of 43.6% (P = 0.032). In

the same subgroup plasma /3-TO was reduced among the sup-

plemented from 60.6 ± 6.5 to 45.5 ± 4.3 �sg/L (by 26.9%; P

< 0.05 for change) and was increased among the control subjects

from 56.8 ± 4.9 to 59.9 ± 13.1 �sg/L, the net reduction being

32.0% (P = 0.071).

The reduction in platelet-count-adjusted serum thromboxane

B2 in the supplemented smokers was 60% (P < 0.01) and in the

supplemented nonsmokers, 20% (Table 3). The net reduction

ofserum thromboxane B2 was 84% (P = 0.019) in smokers and

25% (P = 0. 188) in nonsmokers. Platelet-count-adjusted plasma

/3-TO concentration decreased among the supplemented menby 40% (P < 0.05) in smokers and by 20% in nonsmokers. The

net reduction of plasma /3-TO was 23% (NS) in smokers and

35% (NS) in nonsmokers (Table 3).

To analyze the contributions of changes in different antioxi-

dants to changes in platelet activation, the distributions of

changes in platelet-count-adjusted serum lipid peroxides, serum

thromboxane B2 , and plasma /3-TO concentrations were com-

puted in quartiles ofchanges in the supplemented vitamins and

selenium in all 78 subjects who completed the trial (Table 4).A great increase in plasma ascorbic acid, a-tocopherol, and serum

selenium was related (P < 0.05) to a large reduction in serum

lipid peroxides. Similarly, the change in plasma ascorbic acid,

a-tocopherol, and /3-carotene (P < 0.05 for each) and, though

nonsignificantly, serum selenium was inversely associated with

the change in serum thromboxane B2 . The changes in plasmaascorbic acid and serum selenium had a nonsignificant associ-

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

z

0.�

0

I-

-Ja.z

z

U

1228 SALONEN ET AL

ation with the change in plasma /3-TO (Table 4). The greater

the reduction in serum lipid peroxides, the greater was the de-

crease in plasma fl-TO (P = 0.008 for linear trend for adjusted

and P = 0.019 for unadjusted), as shown in Figure 1. The lipidperoxide change had no significant association with the change

in serum thromboxane B2.

Discussion

It is well-established that increased concentrations of platelet-

specific proteins in plasma reflect enhanced platelet stimulation

in vivo (38-40), except possibly when there is deficient metab-

olism of these proteins in certain disease conditions. Several

studies suggested that /3-TO binds to endothelial cells and blocks

prostacyclin formation (4 1 ). A reduction in plasma concentra-

tions of/3-TG should thus indicate a reduction ofplatelet activity

in vivo.

The assessment ofin vivo platelet activity is tedious and com-

plicated because the platelets tend to get activated during blood

sampling. Because of its shorter half-life FF4 is proportionately

more elevated in ex vivo platelet activation than is fl-TO (39).

Thus, the ratio of these platelet-specific proteins can be used to

rule out spurious platelet activation. The exclusion of pairs of

subjects with possible ex vivo platelet activation did not influence

the results ofthis study decisively. Because plasma concentrations

ofplatelet products and peroxides are dependent on the platelet

count, we presented the main findings with an adjustment for

the platelet count. This adjustment had, however, a minor effect

on the results.

Serum thromboxane B2, as measured in this study, mainly

describes the production of thromboxane A2 by the platelets

during the 30-mm in vitro incubation ofblood at 37 #{176}C.About

one thousandth ofthe thromboxane B2 measured originates from

the circulating blood. For this reason we have also used the term

platelet-produced thromboxane for the same measurement (34).

This assessment of serum thromboxane B2 is an indicator of the

capacity of platelets to produce thromboxane A2.

As an index of lipid peroxidation, serum lipid peroxide con-

centration assayed by the TBA method was used because of its

sensitivity (3 1 ). This method has been criticized for lack of spec-

ificity (42) but this problem is eliminated or found to have a

negligible effect when Yagi’s method is used (3 1, 43). Moreover,

there is a good correlation between the results obtained by Yagi’s

method and those achieved with HPLC (44).

The lower lipid-peroxide concentrations in the end-of-trial

samples ofthe placebo-treated group as compared with the base-

line samples suggest that some lipid peroxidation took place

during the storage of serum samples at -70 #{176}C.If the higher

antioxidant concentrations of the supplemented subjects inhib-

ited this, some of the observed net reduction of lipid peroxides

during the trial would have been due to this ex vivo inhibition

of lipid peroxidation.

The findings of the present study disagree with those of a

recent small but well-controlled trial by Stampfer et al (2 1), who

observed no significant effects on either in vitro platelet aggreg-

ability or serum thromboxane B2 concentration in 20 healthy

university-hospital staff members who were supplemented with

727 mg vitamin E/d for 5 wk. In addition to study power and

measurement method, the most likely explanation for the dis-

crepancy is the difference in the baseline antioxidant status and

dietary fat intake between the two study populations. Five weeks

QUARTILE OF LIPID PEROXIDE CHANGE

FIG 1. Mean change in platelet-count-adjusted plasma fi-thrombo-

globulin Lg,/” thrombocytes) in the quartiles of platelet-count-adjusted

serum lipid peroxide change. r = 0.24, P = 0.008.

may also be too short a period to establish lipid peroxidation-

mediated changes in platelet function.

We also used our data to analyze the cross-sectional associ-

ations between antioxidants and platelet activity and the rela-

tionship between changes in antioxidants and changes in mdi-

cators of platelet activity. The large range of changes in plasma

antioxidant concentrations between subjects enabled the batter

analysis. Both nonexperimental analyses supported the main

finding. Considering the small sample size and the conservative

statistical approach taken (45), the presence of the associations

between antioxidant changes and change in lipid peroxides on

one hand and the associations of lipid peroxide change with

change in plasma fl-TO on the other are in accordance with the

hypothesis that antioxidants reduce platelet activity through the

elimination of circulating lipid peroxides.

This study demonstrates that long-term supplementation witha combination of ascorbic acid, a-tocopherol, fl-carotene, and

selenium of men with low antioxidative vitamin status and rel-

atively high intake of saturated fatty acids reduces serum lipid

peroxide content, the activation of platelets in vivo, and the

aggregabibity and thromboxane production of the platelets in

vitro. Antioxidant supplementation had no effect on the cho-

lesterol concentrations in the main lipoprotein fractions. Al-

though the sample size was insufficient for a formal analysis

comparing smokers with nonsmokers, the data suggest that the

effect of antioxidant supplementation may be greater in smokers

than in nonsmokers. These observations provide additional in-

formation of possible mechanisms through which low antioxi-

dant status can increase the risk of IHD and promote the pro-

gression of atherosclerosis. U

We are deeply grateful to Hannele Kastarinen, public health nurse,

for subject management.

References

I . Salonen iT, Alfthan 0, Pikkarainen J, Huttunen JK, Puska P. As-

sociation between cardiovascular death and myocardial infarctionand serum selenium in a matched-pair longitudinal study. Lancet1982:2:175-9.

2. Wang YX, B#{246}ckerK, Reuter H, et al. Selenium and myocardial

infarction: glutathione peroxidase in platelets. Klin Wochenschr

198 l;59:817-8.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from

ANTIOXIDANTS AND PLATELET FUNCTION

3. Oster 0. Drexler M, Schenk J, et al. The serum selenium concen-

tration of patients with acute myocardial infarction. Ann Clin Res

1986:18:36-42.

4. Kok FJ, Hofman A, Witteman JCM. Decreased selenium levels in

acute myocardial infarction. JAMA 1989;261:1161-4.5. Gey KF. St#{228}helinHB, Puska P. Evans A. Relationship of plasma

level of vitamin C to mortality from ischemic heart disease. Ann

NY Acad Sci l987;498: 110-23.

6. Gey F, Puska P. Plasma Vitamins E and A inversely correlated tomortality from ischemic heart disease in cross-cultural epidemiology.

Ann NY Acad Sci l989;570:268-82.

7. Salonen iT, Salonen R, Penttil#{228}I, et al. Serum fatty acids, apoli-

poproteins, selenium and vitamin antioxidants and the risk of deathfrom coronary artery disease. Am J Cardiol 1985;56:226-3I.

8. Kok FJ, de Bruijn AM, Vermeeren R, et al. Serum selenium, vitaminantioxidants, and cardiovascular mortality: a 9-year follow-up study

in the Netherlands. Am J Clin Nutr l987;45:462-8.

9. Gey KF, Brubacher GB, Staehelin HB. Plasma levels of antioxidantvitamins in relation to ischemic heart disease and cancer. Am J Clin

Nutr 1987;45(suppl 5):l368-77.

10. Salonen JT. Selenium in ischaemic heart disease. Int J Epidemiol

1987; 16:323-7.

I 1. Warso MA, Lands WEM. Lipid peroxidation in relation to pros-tacyclin and thromboxane physiology and pathophysiology. Br Med

BUll l983;39:75-83.

I 2. Salonen JT. Selenium in cardiovascular diseases and cancer. Epi-

demiologic findings from Finland. In: Bostrdm H, Ljungstedt N,

eds. Trace elements in health and disease. Scandia InternationalSymposia, The Scandia Group, Stockholm: Almqvist & Wiksell In-

ternational,1985:172-86.

1 3. Srivastava KC. Ascorbic acid enhances the formation of prostaglan-

din E1 in washed human platelets and prostacyclin in rat aortic rings.

Prostaglandins Leukot Med l985;l8:227-33.

14. Kanofsky JD, Kanofsky PB. Prevention ofthromboembolic disease

by vitamin E. N Engl J Med 1981:305:173-4.15. Lake AM, Stuart MJ, Oski FA. Vitamin E deficiency and enhanced

platelet function: reversal following E supplementation. J Pediatr

l977;90:722-5.

16. Steiner M. Effect of alpha-tocopherol administration on platelet

function in man. Thromb Haemost l983;49:73-7.17. Srivastava KC. Vitamin E exerts antiaggregatory effects without in-

hibiting the enzymes of the arachidonic acid cascade in platelets.

Prostaglandins Leukot Med l986;2l:177-85.

18. Szczeklik A, Gryglewski Ri, Domagala B, Dworski R, Basista M.

Dietary supplementation with vitamin E in hyperlipoproteinemias:

effects on plasma lipid peroxides, antioxidant activity, prostacyclingeneration and platelet aggregability. Thromb Haemost 1985;54:

425-30.

19. Colette C, Monnier LH, Cartry E. Platelet function in type I diabetes:

effects ofsupplementation with large doses ofvitamin E. Am J Clin

Nutr 1988;47:256-61.

20. Swartz SL, Willet WC, Hennekens CH. A randomized trial of the

effect of vitamin E on plasma prostacyclin (6-keto-PGF1,,) levels inhealthy adults. Prostaglandins Leukot Med l985;20:439-48.

21. Stampfer MJ, Jakubowski JA. Faigel D, Vaillancourt R, Deykin D.

Vitamin E supplementation effect on human platelet function, ar-

achidonic acid metabolism and plasma prostacyclin levels. Am J

Clin Nutr 1988;47:700-6.22. Bordia A, Verma SK. Effect of vitamin C on platelet adhesiveness

and platelet aggregation in coronary artery disease patients. Clin

Cardiol 1985;8:552-4.

1229

23. Gordova C, Musca A, Violi F, Perrone A, Alessandri C. Influence

of ascorbic acid on platelet aggregation in vitro and in vivo. Ath-

erosclerosis 1982;41 :15-9.24. Salonen JT, Salonen R, Sepp#{228}nen K, et al. Relationship of serum

selenium and antioxidants to plasma lipoproteins, platelet aggreg-

ability and prevalent ischaemic heart disease in Eastern Finnish men.

Atherosclerosis l988;48: 1226-32.25. Salonen iT. Antioxidants and platelets. Ann Med l989;2 1:59-62.

26. Salonen JT. Is there a continuing need for longitudinal epidemiologic

research?-the Kuopio Ischaemic Heart Disease Risk Factor Study.Ann Clin Res 1988;20:46-50.

27. Ihanainen M, Salonen R, Sepp#{228}nenR, Salonen JT. Nutrition data

collection in the Kuopio Ischaemic Heart Disease Risk Factor Study:nutrient intake of middle-aged Eastern Finnish men. Nutr Res

l989;9:597-604.

28. Vuilleumier J-P, Keller HE, Gysel D, Hunziker F. Clinical chemical

methods for the routine assessment of the vitamin status in human

populations. Int J Vitam Nutr Res 1983;53:265-72.29. Brubacher G, Vuilleumier JP. Vitamin C. In: Curtius HC, Roth M,

eds. Clinical biochemistry, principles and methods. Vol II. Berlin:Walter de Gruyter, 1974:989-97.

30. Carlson K. Lipoprotein fractionation. J Clin Pathol l973;26(suppl

5):32-7.

31. Yagi K. Assay for serum lipid peroxide level and its clinical signif-

icance: In: Yagi K, ed. Lipid peroxides in biology and medicine.New York: Academic Press 1982:223-42.

32. Levander OA, DeLoach DP, Morris VC, Moser PB. Platelet gluta-

thione peroxidase activity as an index of selenium status in rats. JNutr 1983:1 13:55-63.

33. Paglia DE, Valentine WN. Studies ofthe quantitative and qualitativecharacterization of erythrocyte glutathione peroxidase. J Lab Clin

Med l967;70: I 58-69.34. Salonen R, Nikkari T, Sepp#{228}nen K, et al. Effect of omega-3 fatty

acid supplementation on platelet aggregability and platelet produced

thromboxane. Thromb Haemost 1987;57:269-72.

35. Powell WS. Rapid extraction ofarachidonic acid metabolites frombiological samples using octadecylsilyl silica. In: Lands WEM, SmithWL, eds. Methods in enzymology. Prostaglandins and arachidonate

metabolites. Vol 86. New York: Academic Press, I 982:467-77.36. Wold S. Spline functions in data analysis. Technometrics 1974;l6:

1-11.37. SPSS Inc. Users guide. SPSS-X. 3rd ed. Chicago: SPSS Inc. 1988.

38. Niewiarowski 5, Varma KG. Biochemistry and physiology of secreted

platelet proteins. In: Colman RW, Hirsh J, Marder VJ, SalzmanEW, eds. Hemostasis and thrombosis. Basic principles and clinicalpractice. Philadelphia: JB Lippincott, 1982:421-30.

39. Harker LA, Zimmerman TS. Measurements of platelet function.

London: Churchill Livingstone, Longman Group, 1983.

40. Williams CE, Pentwistle MBP, Short PE. Platelet function tests: a

critical review of methods. Med Lab Sci l985;42:262-74.

41. Hope W, Martin Ri, Chesterman CN, Morgan FJ. Human /3-

thromboglobulin inhibits P012 production and binds to a specificsite in bovine aortic endothelial cells. Nature 1979;282:2 10-2.

42. Halliwell B, Gutteridge JMC. Free radicals in biology and medicine.

Oxford, UK: Clarendon Press, 1985.43. Stringer MD, GorOg PG. Freeman A, Kakkar VV. Lipid peroxides

and atherosclerosis. Br Med J l989;298:28l-4.44. Therasse J, Lemonier F. Determination ofplasma lipoperoxides by

high performance liquid chromatography. J Chromatogr 1987;4l3:

237-41.

45. Norman OR. Issues in the use ofchange scores in randomized trials.

J Clin Epidemiol 1989;42:1097-105.

at PE

NN

SY

LVA

NIA

ST

AT

E U

NIV

PA

TE

RN

O LIB

RA

RY

on March 6, 2016

ajcn.nutrition.orgD

ownloaded from