Atherosclerosis 139 (1998) 179–187

Beneficial effect of gemfibrozil on the chemical composition andoxidative susceptibility of low density lipoprotein: a randomized,

double-blind, placebo-controlled study

Hiroshi Yoshida *, Toshitsugu Ishikawa, Makoto Ayaori, Hideki Shige, Toshimitsu Ito,Michio Suzukawa, Haruo Nakamura

First Department of Internal Medicine, National Defense Medical College, 3-2 Namiki, Tokorozawa, Saitama, 359, Japan

Received 26 August 1997; received in revised form 19 February 1998; accepted 2 March 1998

Abstract

Previous reports have shown that administration of fibrates can reduce coronary events and also improve plasma lipid levels.Oxidative modification of low density lipoprotein has been implicated in the pathogenesis of atherosclerosis, and the resistance oflow density lipoprotein (LDL) to in vitro oxidation has been found to be correlated with the extent of atherosclerosis. Weperformed a double-blind, placebo-controlled intervention trial to establish whether gemfibrozil could improve resistance of LDLto oxidation in patients with hyperlipidemia. Patients were randomly assigned to treatment with gemfibrozil (450 mg, twice a day,n=10) or placebo (n=9) for 8 weeks. Blood samples were obtained after an overnight (12 h) fast. Gemfibrozil administrationsignificantly reduced total plasma cholesterol and triglyceride levels and changed the LDL from small, dense particles (pattern B,525.5 nm) to larger, more buoyant particles (pattern A, \25.5 nm). Gemfibrozil significantly increased the lag time of LDLoxidation in vitro by 18.2% from 45.598.0 min at week 0 to 53.4911.4 min at week 8, but did not change LDL vitamin E andb-carotene concentrations. Surprisingly, gemfibrozil significantly decreased LDL lipid peroxides by −33.1% and increased theLDL vitamin E/lipid peroxide ratio by 67.6% from 1.390.5 at week 0 to 2.190.9 at week 8. These results demonstrate thatgemfibrozil treatment can render LDL less susceptible to oxidative modification while reducing plasma cholesterol and triglycerideand improving LDL subclass pattern. This antioxidative effect of gemfibrozil on LDL may be one of the factors which could delaythe progression of atherosclerosis. © 1998 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Gemfibrozil; LDL oxidation; Hyperlipidemia; Lipid peroxide; Vitamin E

1. Introduction

Many studies have demonstrated a strong correlationbetween plasma triglyceride (TG) levels and density andsize of low density lipoproteins (LDL) [1,2]. The plasmalipoprotein profile characterized by a predominance ofsmall, dense LDL particles, which has been designatedLDL subclass pattern B, is associated with an increasedrisk of coronary heart disease (CHD) [1,2].

The Helsinki Heart Study was a primary preventiontrial which demonstrated that gemfibrozil reducedCHD incidence in middle-aged dyslipidemic men [3].This clinical benefit by gemfibrozil was reflected byreduced plasma levels of total cholesterol (TC), LDL-cholesterol, and TG. Gemfibrozil treatment also re-sulted in increased levels of high density lipoprotein(HDL)-cholesterol and increased ratios of LDL-choles-terol to HDL-cholesterol. Previous reports havedemonstrated that gemfibrozil increased LDL particlesize along with a reduction in TG [4,5]. Therefore, theanti-atherosclerotic effect of gemfibrozil could be due inpart to altering LDL particle size.

* Corresponding author. Tel.: +81 429 921211 ext. 2366; fax:+81 429 965200.

0021-9150/98/$19.00 © 1998 Elsevier Science Ireland Ltd. All rights reserved.PII S0021-9150(98)00062-8

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187180

Small, dense LDL is more susceptible to oxidativemodification than large-sized, buoyant LDL [6,7]. Oxi-dized LDL plays an important role in atherogenesis,and oxidative modification of LDL could be a prerequi-site for macrophage uptake and cellular accumulationof cholesterol leading to the formation of atheroscle-rotic plaques [8]. Regnstrom et al. [9] reported that thesusceptibility of LDL to oxidation correlated signifi-cantly with the degree of coronary atherosclerosis. Fur-thermore, it is hypothesized that the protective effect ofgemfibrozil on atherosclerosis may be in part throughinhibiting LDL oxidation with the changes of LDLparticle size.

In the present study, we have evaluated the efficacyof gemfibrozil in reducing the oxidative susceptibility ofLDL in the treatment of patients with type II and typeIV hyperlipidemia. This trial involved a randomized,double-blind, placebo-controlled design in 19 patients.

2. Methods

2.1. Subjects

Men and postmenopausal women between the agesof 30–70 years were recruited using the following crite-ria: plasma TC levels more than 220 mg/dl, plasma TGlevels less than 400 mg/dl from baseline samples ob-tained after a 12-h fast. Those subjects who were obese,or were receiving any lipid lowering drugs or anyanti-oxidative vitamins were excluded. In addition,those subjects with diabetes mellitus, thyroid disorder,liver dysfunction or renal dysfunction were also ex-cluded. Before initiating the study, each patient wasfully informed of the details of the study and potentialrisks and benefits. All participating patients gave in-formed consent for the study, which was approved bythe Institutional Review Board (IRB) of National De-fense Medical College. The study was conducted in theoutpatient facilities of National Defense Medical Col-lege hospital.

2.2. Study design

Each subject was instructed on the therapeutic diet inan outpatient clinic to follow the step-one diet ofNational Cholesterol Education Program [10]. Patientswere advised to maintain the diet during the trial. Thisstudy was a randomized, double-blind, placebo-con-trolled study. After the initial screening period, all thesubjects were fed the placebo for 8 weeks, followed byeither gemfibrozil (450 mg, twice a day, n=10) orplacebo (n=9) for an additional 8 weeks. The clinicalcharacteristics of each group are shown in Table 1.

2.3. Measurements of plasma lipids and apolipoproteinsand LDL lipids and protein

Venous blood samples were obtained from each sub-ject after an overnight (12 h) fast with or withoutEDTA after the initial 8 week treatment period withplacebo (week 0) and at the end of the study (week 8).The blood samples were centrifuged at 2500 rpm for 25min, and serum or EDTA-plasma were separated im-mediately. All baseline plasma and serum samples werestored under nitrogen gas at −80°C for 9 weeks, andthe 8 week plasma and serum samples were storedunder nitrogen gas at −80°C for 1 week until analysis.Serum was used to measure serum lipids and serumapolipoproteins. EDTA-plasma was used to evaluatechemical composition, particle size and susceptibility ofLDL to oxidation. To measure LDL oxidation, LDLparticle size and the chemical composition of LDL,

Table 1Clinical characteristics of study subjects and changes in plasma lipidsand plasma lipoproteins in subjects with hyperlipidemia by treatmentwith gemfibrozil

PlaceboGemfibrozil

Gender 4/56/4(male/female)

53.3913.2Age (years) 56.396.225.292.5BMI (kg/ 24.691.5

m2)30.0Smoking 44.4

(%)

Plasmalipids

Week 0 Week 8 Week 0 Week 85.8290.8 6.7291.47 6.6791.19TC 6.5290.98

TG 1.5190.69*2.5691.21 2.3791.05 2.5491.61PL 5.3590.47*6.5090.58 6.6090.83 6.4890.80

4.8990.834.9791.163.8890.88LDL-C 4.3191.191.1990.31HDL-C 1.1090.281.1290.34 1.0790.27

5.6191.145.3890.83non HDL- 4.6090.80** 5.6191.37C

149921153918 144917155928Apo A-I37.296.6 39.294.6Apo A-II 35.495.9 34.894.5148927Apo B 132923 152938 1529337.392.0 5.891.2*Apo C-II 6.091.6 5.991.8

Apo C-III 15.493.6 10.791.8* 13.893.2 14.695.76.591.97.291.6 6.391.2Apo E 6.091.4**

3.690.53.390.9**4.190.6LPO 3.590.7

TC, indicates total cholesterol; TG, triglyceride; PL, phospholipids;LDL, low density lipoprotein; HDL, high density lipoprotein; Apo,apolipoprotein; LPO, lipid peroxides. Non HDL-C represents TCminus HDL-C. The lipid data are expressed as mmol/l except forLPO, which is expressed as nmol/ml. Apolipoprotein data are ex-pressed as mg/dl. Values represent mean 9S.D. LDL-cholesterolvalues were calculated according to the Friedewald equation [26].* pB0.01;* pB0.05 (comparisons between values at week 0 and those at week8).

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187 181

plasma was thawed on-ice in the dark at the time ofLDL preparation. Then LDL with a density between1.019 and 1.063 g/ml was isolated from the plasma bysequential ultracentrifugation following the method ofHavel et al. [11]. LDL was further separated into twosubclasses having densities between 1.019 and 1.040(LDL1) and between 1.040 and 1.063 (LDL2).

Serum and LDL lipids (total cholesterol, TC; freecholesterol, FC; triglyceride, TG; and phospholipids,PL) were measured enzymatically [12], and serumapolipoproteins (apo) A-I, A-II, B, C-II, C-III, E andLDL apo B were measured by using the immunotur-bidimetric method [13]. HDL-cholesterol was deter-mined in whole plasma after precipitation of apoB-containing lipoproteins with dextran sulfate andmagnesium chloride [14]. Lipoprotein protein contentwas measured by the method of Lowry et al. usingbovine serum albumin (BSA) as a standard [15]. Lipo-protein cholesteryl ester (CE) mass was calculated as(TC–FC)×1.68 [4]. Lipid peroxide (LPO) was mea-sured by using a commercially available kit (DeterminerLPO, Kyowa, Tokyo, Japan), which is based on acalorimetric assay using the reaction of a leu-comethylene blue derivative with lipid hydroperoxidesin the heme compounds [16]. For the measurements oflipid peroxides in LDL, 50 mg protein of LDL wasapplied. Coefficient of variation of the assay was lessthan 3%.

2.4. Measurements of 6itamin E, b-carotene and fattyacids of LDL

The concentration of vitamin E in LDL was quanti-tated by extraction and analysis using reversed-phaseHPLC as previously described [17,18]. Briefly, 200 m l ofLDL was precipitated using ethanol and the vitamin Ewas extracted with hexane. The hexane phase wasevaporated under N2 gas and the residue was dissolvedin ethanol. Vitamin E was separated by reversed-phaseHPLC using a C18 column (25×0.46 cm, 5 mm parti-cle size; TSK gel ODS-80Ts) eluted with ethanol/dis-tilled water (92:8, vol/vol) as the mobile phase at 1.0ml/min and measured in a UV detector (UV-8000,TOSOH, Tokyo) set at 295 nm. To determine theconcentration of b-carotene in LDL, 200 m l of LDLsample was mixed with 1 ml of distilled water and 1 mlof ethanol, vortexed and incubated at room tempera-ture for 2 min. Then 5 ml of n-hexane was added to themixture and shaken for 10 min at 4°C in the dark, andcentrifuged for 5 min at 2500 rpm at 4°C. The superna-tant was evaporated under N2 gas, and the residueresuspended in 100 m l of ethanol. An aliquot of 20 m lwas analyzed by HPLC using acetonitrile/methylenechloride dichloromethane (92/8, vol/vol) as the mobilephase at 1.5 ml/min and detection at 450 nm in a UVdetector.

To quantitate the fatty acids in LDL, total lipids ofLDL were extracted using a modification of the methoddescribed by Folch et al. [19]. The fatty acids were thentransmethylated and analyzed by gas chromatographyas previously described [20].

2.5. E6aluation of LDL oxidation

Each LDL sample was dialyzed against phosphate-buffered saline (PBS), pH 7.4 at 4°C for 24 h in thedark to remove the EDTA. Oxidative susceptibility ofLDL was determined by measuring the production ofconjugated diene according to the method of Ester-bauer et al. [21]. LDL, adjusted to 50 mg protein/mlwith PBS, was incubated with 2 mM CuSO4 in PBS (2ml final volume) at 37°C. Conjugated diene formationin LDL oxidation was monitored by absorbance at 234nm in a spectrophotometer (SHIMADZU 160A, Shi-madzu, Tokyo) equipped with a six-position automaticchanger. The absorbance at 234 nm was recorded every10 min for 4 h after initiating oxidation with copper.Lag time and propagation rate were determined aspreviously described [21]. The coefficient of variationfor the lag time assay in this study was 3 and 4.5% forthe intra-assay and inter-assay, respectively. For thepropagation rate assay, the coefficient of variation was4 and 5.6% for the intra-assay and inter-assay, respec-tively. Previously, we have reported that storage ofplasma at −80°C for 2 months before isolation ofLDL had no effect on in vitro LDL oxidation withcopper [22].

2.6. E6aluation of LDL particle size (LDL peakparticle diameter)

The particle diameters of predominant LDL peakswere determined by non-denaturing polyacrylamidegradient gel electrophoresis as previously described[22–25]. The electrophoresis was carried out at 10°C in2–16% polyacrylamide gradient gels (PAA 2–16, Phar-macia, Uppsala, Sweden) using Tris (90 mM)–boricacid (80 mM) buffer, pH 8.3, containing EDTA (3mM). Sample densities were increased by adding asolution consisting of 40% sucrose and 0.02% bro-mophenol blue, and an aliquot of 5–10 m l was added toeach lane. Applied voltage was set to 20 V for 15 min,followed by 70 V for 20 min, and finally 125 V for 24h. Gels were fixed with 10% sulfosalicylic acid for 1 hand stained in 50% methanol plus 10% acetic acidcontaining Coomassie Brilliant blue R-250. Afterdestaining in 20% methanol plus 9% acetic acid, gelswere scanned at 596 nm on a laser densitometer (LKBUltrascan XL). Migration distances were determinedfor each of the absorbance maximums, and correspond-ing molecular diameters were calculated from a calibra-tion curve using standards of known diameters. In

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187182

Fig. 1. Structural formula of gemfibrozil and its metabolites (M1 andM3).

3. Results

3.1. Characteristics of study subjects

At the beginning of the placebo period, all 19 hyper-lipidemic patients met the inclusion criteria. Therefore,all patients had hyperlipidemic phenotype IIa, IIb orIV. Table 1 lists the clinical characteristics of partici-pants in the study. The groups consisted of six men andfour women in the gemfibrozil group and four men andfive women in the placebo group. There were no differ-ences between the gemfibrozil group and the placebogroup in age (53.3913.2 vs. 56.396.2 years, respec-tively) or body mass index (BMI, 25.292.5 vs. 24.691.5, respectively). In the gemfibrozil group, there weresix patients with type IIb, two patients with type IIaand two patients with type IV hyperlipidemia. In theplacebo group, there were seven patients with type IIb,one patient with type IIa and one patient with type IVhyperlipidemia. There were no significant differences inserum lipid and apolipoprotein levels at baseline, ornumber of current smokers between the two treatmentgroups.

3.2. Effect of gemfibrozil on plasma lipid andapolipoprotein le6els

As shown in Table 1, gemfibrozil significantly de-creased plasma TG, PL, and non-HDL-cholesterol (TCminus HDL-cholesterol) from baseline values. Apo C-II, C-III and E were significantly decreased in thegemfibrozil group. In addition, plasma lipid peroxidelevels for patients in the gemfibrozil group were signifi-cantly decreased (−19.7%).

3.3. Effect of gemfibrozil on lipid and fatty acidcomposition and particle size of LDL

The results of LDL composition are presented inTable 2. Gemfibrozil treatment slightly increased theLDL FC/apo B ratio and decreased the LDL CE/FCratio, although the differences between baseline valuesand week 8 values did not reach significance (p=0.08and p=0.06, respectively). Other parameters of LDLcomposition were also not significantly different be-tween week 0 and week 8 values in either of the twogroups. Furthermore, chemical composition of eachLDL subfraction (LDL1 and LDL2) also did not differbetween week 0 and week 8. However, the CE/FC ratiowas slightly decreased in LDL 2 only, but again thedifference between week 0 and week 8 values did notreach significance (from 6.1191.13 to 5.6690.87, p=0.084).

There was no significant difference in fatty acidcomposition of LDL at week 0 between the gemfibroziland placebo groups. Major saturated, monounsatu-

accordance with previous reports using this method toevaluate LDL peak particle diameter (PPD) as de-scribed above, we defined that individuals with a pre-dominance of small LDL particles (LDL-PPD 525.5nm) are characterized as having LDL subclass patternB, while those with a predominance of larger, morebuoyant LDL particles (LDL-PPD \25.5 nm) arecharacterized as having LDL subclass pattern A [2,23–25].

2.7. Effect of in 6itro supplementation of gemfibroziland gemfibrozil metabolites on LDL oxidation

The structure of gemfibrozil and its metabolites (M1and M3), kind gifts from Sankyo and Warner–Lam-bert, Japan, are shown in Fig. 1. Plasma concentrationsof gemfibrozil, M1 and M3 at the clinical doses arearound 4 mM, 0.5 mM and 27 mM, respectively. Gemfi-brozil, M1 and M3, which were dissolved in ethanol,were added to LDL (50 mg/ml) in PBS (final volume: 2ml) just before adding 2 mM copper. The final concen-tration of ethanol in experiment mixture was 1% (v/v)in PBS, and LDL in PBS without gemfibrozil or itsmetabolites and LDL in PBS containing 1% ethanolwere set up as an experimental control. Conjugateddiene formation was monitored as described above.

2.8. Statistics

Values represent the mean 9S.D. The significance ofdifferences before and after treatment within the gemfi-brozil and placebo groups was analyzed by Wilcoxon’ssigned rank test. Wilcoxon’s rank sum test was used todetermine the significance of differences between thegemfibrozil and placebo groups. A value of pB0.05was taken as the criterion of significance. Correlationanalyses were carried out using Spearman correlationanalysis.

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187 183

Table 2Changes in LDL chemical composition

FC/apo B CE/apo B TG/apo B PL/apo B CE/FC ApoTC/apo B

GemfibrozilWeek 0 (baseline) 3.491.2 0.790.3 2.690.5 0.390.1 2.390.8 3.590.8 73.8925.2

3.990.3Week 8 1.090.3 2.990.3 0.390.1 2.590.3 3.290.5 72.4918.3

Placebo0.890.3 2.890.8 0.390.1 2.390.8 3.590.6Week 0 (baseline) 80.7928.63.691.00.890.3 2.690.6 0.390.1 2.390.5 3.590.53.490.8 89.1929.4Week 8

TC indicates total cholesterol; apo, apolipoprotein; FC, free cholesterol; CE, cholesteryl ester; TG, triglyceride; and PL, phospholipids. The dataare expressed as mmol/g except for CE/FC and apo B which are expressed as mol/mol and mg/dl, respectively. Values represent mean 9S.D.

rated and polyunsaturated fatty acids did not differbetween week 0 and week 8 in both groups (data notshown). In addition, linoleic acid/oleic acid ratios alsodid not differ (from 1.7290.41 to 1.8390.36 in thegemfibrozil group, from 2.090.37 to 2.0290.4 in theplacebo group).

Gemfibrozil treatment increased LDL particle sizebut not significantly (Table 3). However, there was asignificant difference in LDL particle size at week 8between the gemfibrozil group and the placebo group(pB0.05). In the gemfibrozil group, 60% changed frompattern B LDL (LDL peak particle diameter 525.5nm) to pattern A LDL (LDL peak particle diameter\25.5 nm) (Fig. 2). On the other hand, none of theplacebo group changed from pattern B to pattern A.The difference between the gemfibrozil and placebogroups was significant (pB0.01). Actually, gemfibroziltreatment slightly increased LDL particle size, but thesesize data ranged from only 25 to 26 nm. Although thisdifference was not significant, there was enough changein the mean particle size to cause a change of categoryfor LDL particle pattern.

3.4. Effect of gemfibrozil on oxidati6e susceptibility,lipid peroxide le6els and antioxidant 6itamin le6els ofLDL

As shown in Table 3, there were no significant differ-ences of LDL-vitamin E and LDL-b carotene levelsbetween week 0 and week 8 in both the gemfibrozil andplacebo groups. Gemfibrozil treatment increased LDLvitamin E but not significantly (p=0.08). However,gemfibrozil significantly reduced LDL–LPO and dra-matically increased the LDL vitamin E/LPO ratio by67.6% (1.390.5 and 2.191.0, at week 0 and week 8,respectively).

Gemfibrozil significantly increased the lag time ofLDL oxidation by 18.2%, but those in the placebogroup did not change (Table 3). On the other hand, thepropagation rate of LDL oxidation in vitro did notdiffer between week 0 and week 8 in both groups.

3.5. Correlation analyses between lag time of LDLoxidation and characteristics of LDL

There was a significant negative correlation betweenLDL–LPO and the oxidation in vitro from week 0 toweek 8 in the placebo group alone and in all datacombined (r= −0.73, p=0.025 and r= −0.52, p=0.022, respectively), but not correlate in the gemfibrozilgroup alone. The change in LDL vitamin E from week0 to week 8 did not correlate with the change in lagtime of LDL oxidation in either group. However, thechange in the LDL vitamin E/LPO ratio correlatedpositively with the change in lag time in the gemfibrozilgroup and in all data combined (r=0.70, p=0.025 andr=0.72, p=0.0005, respectively), but did not correlatein the placebo group. Likewise the change in LDLparticle size was correlated significantly with the changein lag time of LDL oxidation in the gemfibrozil groupand in all data together (r=0.821, p=0.0036, r=0.862, p=0.0001, respectively). There was no signifi-cant correlation between the changes in lag time andthe changes in LDL lipid composition.

3.6. In 6itro effect of gemfibrozil and gemfibrozilmetabolites on LDL oxidation

Neither gemfibrozil nor M3 changed the LDL oxida-tion at concentrations up to 100 mM (data not shown).However, M1 demonstrated an antioxidative effect oncopper-mediated LDL oxidation in a concentration-de-pendent manner (Fig. 3). The lag times for controlLDL, control with vehicle, LDL with 2.5 mM M1, LDLwith 5 mM M1 and LDL with 10 mM M1 were 55.2,60.4, 75, 115.9, and more than 200 min, respectively.The propagation rates for each were 13.5, 11.7, 10.2,and 7.6 nmol dienes/min per mg LDL protein, respec-tively. The propagation rate for the LDL with 10 mMM1 was not determined because of the long lag timeachieved at that case.

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187184

Table 3Changes in LDL oxidation, antioxidants, lipid peroxide and particle size

PlaceboGemfibrozil

Week 0 Week 8 Week 0 Week 8

LDL oxidizabilityLag time (min) 45.598.0 53.4911.4* 49.1910.2 47.8911.7

7.691.6Propagation rate (nmol/mg per min) 6.790.77.291.5 7.091.5

Antioxidants and lipids peroxide6.792.0Vitamin E (nmol/mg protein) 7.692.26.191.5 6.491.44.690.9 6.092.34.491.4 5.692.3Vitamin E (mmol/l)

1709190bcarotene (nmol/l) 134963 158967 1759613.791.7* 5.892.15.592.5 5.493.2Lipid peroxide (nmol/mg protein)

1.390.5Vitamin E/lipid peroxide (mol/mol) 2.190.9** 1.490.6 1.490.5

26.090.8LDL peak particle diameter (nm) 25.290.525.491.0 24.890.8

Values represent mean 9S.D.* pB0.05.** pB0.01 (week 0 vs. week 8).

4. Discussion

Gemfibrozil improves plasma lipids and apolipo-proteins in patients with dyslipidemia when adminis-tered at a total daily dose of 900 or 1200 mg [4,5].These effects result in a reduction in the risk of coro-nary heart disease (CHD) and also at least a 16%reduction in the incidence of CHD compared withplacebo [26]. Recent reports have demonstrated thatgemfibrozil improves haemostatic abnormalities and in-sulin resistance in patients with dyslipidemia [27,28].These beneficial effects of gemfibrozil may delay theprogression of atherosclerosis. In this report we de-scribed the effect of gemfibrozil on LDL oxidation in adouble-blind, placebo-controlled fashion in order toinvestigate another beneficial effect of gemfibrozil oncoronary risk reduction.

The present data show that gemfibrozil reduces theoxidative susceptibility of LDL. The measurement ofLDL oxidation was determined by lag time prior tooxidative propagation of conjugated diene formationinduced by incubating with copper. Therefore, the LDLoxidation is dependent on both the antioxidant levelsand lipid peroxide levels of LDL. Vitamin E is one ofthe most important lipophilic antioxidants in LDL, andconsequently it is thought that vitamin E content ofLDL determines to a large extent its resistance againstoxidative modification. In the present study, gemfibroziladministration resulted in a prolongation of lag time by18.2%. LDL-LPO was significantly decreased by gemfi-brozil (−33.1%, pB0.01), and thereby may be at-tributed to the elongation of the lag time of LDLoxidation in vitro. On the other hand, LDL vitamin Ewas slightly increased by gemfibrozil, but not signifi-cantly (+9.0%, p=0.08). However, gemfibrozil admin-istration dramatically increased the LDL vitamin

E/LPO ratio (+67.7%, pB0.01) which exceeded thechange of LDL–LPO. We have previously reportedthat the LDL vitamin E/LPO ratio appears to play apivotal role in modulating LDL oxidation, and thatreduced LDL vitamin E/LPO ratios could cause anincrease in LDL oxidation [22]. Bredie et al. [29] re-ported that gemfibrozil administration significantly de-creased LDL vitamin E content. The reason for thediscrepancy between that study and our study is notknown, but is most likely due to a difference in diet.Bredie et al. [29] also showed that gemfibrozil did notchange the lag time of LDL oxidation in spite ofdecreased plasma TG levels and increased LDL particlesize. However, that observation might be due to re-duced LDL vitamin E levels by gemfibrozil in thatexperiment.

The fatty acid composition of LDL is important instudying LDL–LPO and LDL oxidation becausepolyunsaturated fatty acids (linoleic acid, arachidonicacid) are highly susceptible to oxidative modification[20], but gemfibrozil administration did not change thefatty acid composition of LDL. Therefore, analyses offatty acid composition of LDL can not explain thereduction effect of gemfibrozil on LDL–LPO and LDLoxidation. The frozen storage of plasma might haveaffected the LDL–LPO, but this is unlikely becauseLDL–LPO and plasma lipid peroxide did not changein the placebo group. In any case, the mechanisms bywhich gemfibrozil decreases LDL–LPO remain to beelucidated.

Previous studies have reported that gemfibrozil treat-ment decreases plasma TG levels and alters the distri-bution of LDL subspecies with a shift from small denseLDL particles to large, less dense particles in hyper-triglyceridemic patients [4,5]. In the present study,gemfibrozil did not significantly change the LDL peak

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187 185

Fig. 2. Changes in LDL subclass pattern. (A) panel indicates changes in LDL subclass pattern between week 0 and week 8 in the gemfibrozilgroup. (B) panel indicates changes in the placebo group. Gemfibrozil significantly changed from LDL pattern B to LDL pattern A (pB0.01).

particle diameter, but did significantly change the LDLfrom pattern B to pattern A. Because it is known ingeneral that small dense LDL is more susceptible tooxidative modification than normal LDL [6], gemfi-brozil administration might protect LDL against invitro oxidation through the enlargement of LDL parti-cle size. The present study also showed that the changein LDL particle size positively correlated with thechange in the lag time of LDL oxidation in the gemfi-brozil group.

Tribble et al. [30] have shown that LDL with in-creased free cholesterol (FC) content is less susceptibleto oxidation and LDL with increased cholesteryl ester(CE) content is more susceptible to oxidation. In the

present study, gemfibrozil decreased the LDL CE/FCratio especially in LDL2, which is a denser LDL parti-cle subspecies and more susceptible to oxidation.Therefore, the decreased LDL CE/FC ratio might playa role in the reduced LDL oxidation with gemfibroziladministration.

Recent reports have indicated that metabolites ofdoxazosin, an alpha 1-adrenergic blocking antihyper-tensive agent, and a metabolite of gemfibrozil inhibitLDL oxidation [31,32]. As shown in Fig. 1, M-1metabolite is hydroxylated on a phenol ring. This phe-nolic hydroxy group has been reported to inhibit oxi-dative modification of LDL [31–35]. The present studydemonstrated that the M1 metabolite of gemfibrozilinhibits copper-mediated LDL oxidation, while gemfi-brozil itself and the M3 metabolite do not. However,the plasma concentration of the M1 metabolite is muchlower than that of gemfibrozil or the M3 metabolite.Actually, the plasma concentration of M1 metabolite,around 0.5 mM in the clinical dose, is lower than theconcentrations (2.5–10 mM) used in this in vitro study.Therefore, the direct inhibitory effect of the M1metabolite on LDL oxidation in vivo is likely to besmall, but still might in part play a role for antioxida-tive effect with gemfibrozil administration.

In conclusion, the present study showed that gemfi-brozil administration improves the resistance of LDLto oxidation with a reduction of LDL lipid peroxideand a favorable change of LDL particle pattern.Whether 18.2% prolongation of lag times of LDLoxidation in vitro is enough to be relevant to theprevention of the development of atherosclerosis inhumans is unknown. However, the risk reduction ofcoronary heart disease by gemfibrozil administration,as shown in Helsinki heart study, might be due in partto this protective effect of gemfibrozil against LDLoxidation.

Fig. 3. Inhibitory effect of M1, a metabolite of gemfibrozil, onconjugated diene formation during copper-mediated LDL oxidation.M1 metabolite, at the indicated concentrations, was incubated with50 mg/ml of LDL in 2 ml of PBS with 2 mM copper at 37°C for 190min, and conjugated diene formation was monitored by the changesin 234 nm wavelength absorbance as described in Methods. Opencircle shows control without M1 metabolite (LDL and PBS alone),and closed circle shows vehicle-control containing 1% ethanol. Opensquare, semi-closed square and closed square show 2.5 mM, 5 mMand 10 mM M1 metabolite addition, respectively.

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187186

Acknowledgements

We thank Sankyo company L.T.D. for financial sup-port and for serving as the study coordinator. We aredeeply indebted to Daniel Steinberg M.D., Ph.D. (Uni-versity of California, San Diego) for his valuable com-ments. We appreciate David A. Bird Ph.D. (Universityof California, San Diego) for helpful editorial work onour manuscript. We also thank Kaji D., Ph.D.,Tomiyasu K., M.D., Nakajima K., M.D., Higashi K.,M.D., Yonemura A., M.D., Sawada S., M.D., Ms.Miyajima E., Ms. Nomi M. and colleagues for helpfulassistance.

References

[1] Austin MA, King M-C, Vranizan KM, Krauss RM. Atherogeniclipoprotein phenotype: a proposed genetic marker for coronaryheart disease risk. Circulation 1990;82:495–506.

[2] Campos H, Genest JJ, Blijlevens E, McNamara JR, Jenner JL,Ordovas JM, Wilson PWF, Schaefer EJ. Low density lipoproteinparticle size and coronary heart disease. Arterioscler Thromb1992;12:187–95.

[3] Frick MH, Elo O, Haapa K, Heinonen OP, Heinsalmi P, HeloP, Huttunen JK, Kaitaniemi P, Koskinen P, Manninen V,Maenpaa H, Malkonen M, Manttari M, Norola S, PasternackA, Pikkarainen J, Romo M, Sjoblom T, Nikkila EA. HelsinkiHeart Study: primary-prevention trial with gemfibrozil in mid-dle-aged with dyslipidemia. New Engl J Med 1987;317:1237–45.

[4] Tsai MY, Yuan J, Hunninghake DB. Effect of gemfibrozil oncomposition of lipoprotein and distribution of LDL subspecies.Atherosclerosis 1992;95:35–42.

[5] Yuan J, Tsai MY, Hunninghake DB. Changes in compositionand distribution of LDL subspecies in hypertriglyceridemic andhypercholesterolemic patients during gemfibrozil therapy.Atherosclerosis 1994;110:1–11.

[6] de Graaf J, Hak-Lemmers HLM, Hectors MPC, DemackerPNM, Hendriks JCM, Stalenhoef AFH. Enhanced susceptibilityto in vitro oxidation of the dense low density lipoprotein sub-fraction in healthy subjects. Arterioscler Thromb 1991;11:298–306.

[7] Chait A, Brazg RL, Tribble DL, Krauss RM. Susceptibility ofsmall, dense, low-density lipoprotein to oxidative modification insubjects with the atherogenic lipoprotein phenotype pattern B.Am J Med 1993;94:350–6.

[8] Witztum JL, Steinberg D. Role of oxidized low density lipo-protein in atherogenesis. J Clin Invest 1991;88:1785–92.

[9] Regnstrom J, Nilsson J, Tornvall P, Landou C, Hamsten A.Susceptibility to low-density lipoprotein oxidation and coronaryatherosclerosis in man. Lancet 1992;339:1183–6.

[10] Report of the National Cholesterol Education Program ExpertPanel on detection, evaluation and treatment of high bloodcholesterol in adults. Arch Intern Med 1988; 148: 36–69.

[11] Havel RJ, Eder HA, Bragon JH. The distribution and chemicalcomposition of ultracentrifugally separated lipoprotein in humanserum. J Clin Invest 1995;34:1345–53.

[12] Warnick GR. Enzymatic methods for quantification of lipo-protein lipids. Methods Enzymol 1986;129:101–23.

[13] Noma A, Hata Y, Goto Y. Quantification of serum apolipo-protein A-I, A-II, B, C-II, C-III and E in healthy Japanese byturbidimetric immunoassay; reference values and age- and sex-related differences. Clin Chim Acta 1991;199:147–58.

[14] Warnick GR, Benderson J, Albers JJ. Dextran sulfate-Mg2+

precipitation for quantitation of high-density-lipoprotein choles-terol. Clin Chem 1986;28:1379–88.

[15] Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Proteinmeasurement with the folin phenol reagent. J Biol Chem1951;193:265–75.

[16] Ohishi N, Ohkawa H, Miike A, Tatano T, Yagi K. A new assaymethod for lipid peroxides using a methylene blue derivative.Biochem Int 1985;10:205–11.

[17] Bieri JG, Tolliver TJ, Catignani GL. Simultaneous determina-tion of a-tocopherol and retinol in plasma or red cells by HPLC.Am J Clin Nutr 1979;32:2143–9.

[18] Driskell WJ, Bashor MM, Neese JW. Beta-carotene determinedin serum by liquid chromatography with an internal standard.Clin Chem 1983;29:1042–4.

[19] Folch J, Lees JM, Stanley GHS. A simple method for theisolation and purification of total lipids from animal tissues. JBiol Chem 1957;226:497–509.

[20] Reaven P, Parthasarathy S, Grasse BJ, Miller E, Steinberg D,Witztum JL. Effects of oleate-rich and linoleate-rich diets on thesusceptibility of low density lipoprotein to oxidative modificationin mildly hypercholesterolemic subjects. J Clin Invest1993;91:668–76.

[21] Esterbauer H, Striegl G, Puhl H, Rotheneder M. Continuousmonitoring of in vitro oxidation of human low density lipo-protein. Free Radical Res Commun 1989;6:67–75.

[22] Yoshida H, Ishikawa T, Nakamura H. Vitamin E/lipid peroxideratio and susceptibility of LDL to oxidative modification innon-insulin-dependent diabetes mellitus. Arterioscler ThrombVasc Biol 1997;17:1438–46.

[23] Krauss RM, Burke DJ. Identification of multiple subclasses ofplasma low density lipoproteins in normal humans. J Lipid Res1982;23:97–104.

[24] Williams PT, Krauss RM, Nichols AV, Vranizan KM, WoodPDS. Identifying the predominant peak diameter of high-densitylipoprotein and low-density lipoprotein by electrophoresis. JLipid Res 1990;31:1131–9.

[25] Katzel LI, Krauss RM, Goldberg AP. Relations of plasma TGand HDL-C concentrations to body composition and plasmainsulin levels are altered in men with small LDL particles.Arterioscler Thromb 1994;14:1121–8.

[26] Friedewald WT, Levy RI, Fredrickson DS. Estimation of theconcentration of low-density lipoprotein cholesterol in plasma,without use of the preparative ultracentrifuge. Clin Chem1972;18:499–502.

[27] Andersen P, Smith P, Seljeflot I, Brataker S, Arnesen H. Effectsof gemfibrozil on lipid and haemostasis after myocardial infarc-tion. Thromb Haemost 1990;63:174–7.

[28] Tenkanen L, Manttari M, Manninen V. Some coronary riskfactors related to the insulin resistance syndrome and treatmentwith gemfibrozil. Experience from the Helsinki heart study.Circulation 1995;92:1779–85.

[29] Bredie SJH, de Bruin TWA, Demacker PNM, Kastelein JJP,Stalenhoef AFH. Comparison of gemfibrozil versus simvastatinin familial combined hyperlipidemia and effects on apolipo-protein-B-containing lipoproteins, low-density lipoprotein sub-fraction profile, and low-density lipoprotein oxidizability. Am JCardiol 1995;75:348–53.

[30] Tribble DL, Holl LG, Wood PD, Krauss RM. Variations inoxidative susceptibility among six low density lipoprotein sub-fractions of differing density and particle size. Atherosclerosis1992;93:189–99.

[31] Chait A, Gilmore M, Kawamura M. Inhibition of low densitylipoprotein oxidation in vitro by the 6- and 7-hydroxy-metabo-lites of doxazosin, an alpha 1-adrenergic antihypertensive agent.Am J Hypertens 1994;7:159–67.

H. Yoshida et al. / Atherosclerosis 139 (1998) 179–187 187

[32] Kawamura M, Hase K, Miyazaki S. In vitro inhibition oflow-density lipoprotein oxidation by a gemfibrozil metabolite.Clin Chem 1996;42:644–5.

[33] Miura S, Watanabe J, Sano M, Tomita T, Osawa T, Hara Y,Tomita I. Effects of various natural antioxidants on the Cu2+-mediated oxidative modification of low density lipoprotein. BiolPharm Bull 1995;18:1–4.

[34] Mangiapane H, Thomson J, Salter A, Brown S, Bell GD, White

DA. The inhibition of the oxidation of low density lipoproteinby (+ )-catechin, a naturally occurring flavonoid. Biochem Phar-macol 1992;43:445–50.

[35] Ishikawa T, Suzukawa M, Ito T, Yoshida H, Ayaori M, Nishi-waki M, Yonemura A, Hara Y, Nakamura H. Effect of teaflavonoid supplementation on the susceptibility of low-densitylipoprotein to oxidative modification. Am J Clin Nutr1997;66:261–6.

..