¹Preventive Medicine and Nutrition Clinic, Department of Social Medicine, School of Medicine, University of Crete, Heraklion, Crete, Greece

²Department of Human Nutrition and Epidemiology, Wageningen University, Wageningen, The Netherlands

³Unilever Research Vlaardingen, Vlaardingen, The Netherlands

Correspondence to: A Kafatos, Preventive Medicine and Nutrition Clinic, Department of Social Medicine, School of Medicine, University of Crete, PO Box 1393, Heraklion, Crete, Greece. E-mail: kafatos@med.uoc.gr

^aGuarantor: A Kafatos.

^bContributors: MV, SW, KvP and AK contributed to the study design. AK, MV, SW and JM were involved in fieldwork organization and JM analyzed the data. JM wrote the paper with a substantial contribution from MV and contributions also from AK and SW.

Objective: To examine the effect of a low phenol olive oil and high phenol olive oil on markers of oxidation and plasma susceptibility to oxidation in normolipaemic smokers.

Design: Randomized single-blind cross-over trial with two intervention periods.

Setting: The Medical School and University Hospital of the University of Crete, Heraklion, Crete, Greece.

Subjects: Twenty-five healthy males and females completed the study.

Interventions: Each intervention was of three weeks duration and intervention periods were separated by a two week washout. Seventy grams of extra virgin olive oil was supplied to each subject per day in the intervention periods. The olive oils supplied differed in their phenol content by 18.6 mg/day. Two fasting venous blood samples were taken at the end of each intervention period.

Results: The markers of antioxidant capacity measured in fasting plasma samples (total plasma resistance to oxidation, concentrations of protein carbonyl as a marker of protein oxidation, malondialdehyde and lipid hydroperoxides as markers of lipid oxidation and the ferric reducing ability of plasma) did not differ significantly between the low and high phenol olive oil diets.

Conclusions: No effect of olive oil phenols on markers of oxidation in smokers was detected. It may be that the natural concentrations of phenols in olive oil are too low to produce an effect in the post-absorptive phase. Possible reasons for period effects and interactions between diet and administration period need attention to aid further cross-over trials of this kind.

Sponsorship: Unilever Research Vlaardingen, The Netherlands.

European Journal of Clinical Nutrition (2002) 56, 1024-1029. doi:10.1038/sj.ejcn.1601444

phenols; olive oil; cross-over trial; rate of oxidizability; lipid hydroperoxides; malondialdehyde; protein carbonyls; FRAP; smokers

Introduction

It is the classical Mediterranean diet which has been found to be associated with a reduced incidence of coronary artery disease (Serra-Majem et al, 1995; Keys et al, 1986). It has been suggested that a diet rich in monounsaturated fatty acids such as those from olive oil lowers the risk of coronary heart disease by decreasing cholesterol levels, if compared with saturated fatty acids, and by producing oleic acid enriched LDL particles, which are more resistant to oxidative modification (Bonanome et al, 1992; Reaven et al, 1991; Aviram & Elias, 1993 as cited in Vissers et al, 2001b). In addition to oleic acid, however, some types of extra virgin olive oil contain phenols with antioxidative properties. The major dietary sources of phenolic antioxidants are tea, fruit and vegetables, grape skins and extra virgin olive oil. Phenols from tea and fruit, however, are water-soluble, whereas those from olive oil are partly lipid-soluble. It has been shown in animal studies that phenols from olive oil are able to protect LDL from oxidation and there is an indication that olive oil phenols accumulate within LDL (Wiseman et al, 1996; Scaccini et al, 1992). Lipid-soluble phenols may be taken up by LDL particles in humans, where they can protect the particles from oxidation and thus inhibit the oxidation of LDL.

In Greece, as in other Mediterranean countries, the main source of dietary fat is olive oil (Mamalakis & Kafatos, 2001). Smokers are believed to have more oxidative stress than non-smokers, ie oxidative damage either due to the endogenous production of reactive oxygen species by mitochondria or due to a decrease in antioxidant levels. Therefore, it is thought that antioxidants may have a greater effect in smokers (Hulea et al, 1995; MacNee, 2000; Zhou et al, 2000). The aim of the present study was to determine whether dietary consumption of phenols from virgin olive oil affects markers of oxidation and plasma susceptibility to oxidation in normolipaemic subjects who are smokers and live in the Mediterranean region of Crete, Greece.

Methods

Subjects

Participants in the trial were recruited from the medical student population and from employees at the Medical School of the University of Crete. The study began with a screening procedure for suitable candidates, during which questions were asked of potential study subjects either by telephone or direct interview. Subsequently, blood samples were taken. Exclusion criteria were total plasma cholesterol >7.0 mmol/l, plasma fasting triglycerides >2.3 mmol/l, serum values of liver enzymes >54 U/l for alanine aminotransferase (ALT) or >30 U/l for aspartate aminotransferase (AST), use of medication known to affect concentrations of serum lipids, being pregnant or lactating and being on a prescribed diet. Those with a known history of gastrointestinal, liver or kidney disease were also excluded, as were those with glucosuria, proteinuria and anaemia. A mandatory inclusion criterion was that the subjects be smokers, ie smoke more than one cigarette every day of the week. The screening process continued until 28 healthy volunteers had been recruited. A physician independently reviewed the completed medical questionnaire. Written informed consent was obtained from the subjects.

Three male subjects (aged 21, 22 and 30 y) dropped out for personal reasons unconnected to the study (two at the start of the second week of the first period and the third during the third week in the first period) and their data have not been included in the following analysis. The baseline characteristics of subjects completing the study are provided in Table 1. Baseline dietary data were obtained using a modified 28 day dietary history containing a checklist of 84 common Greek foods.

Study design

The cross-over study was single-blind and randomized, consisting of two 3-week treatment periods. Prior to the first treatment period, there was a run-in period of 2 weeks. The two treatment periods were separated by a washout of 2 weeks. During the run-in and washout periods, subjects were requested to consume diets without olives and olive oil products. At the end of the pre-experimental period (just prior to the run-in period), a one-month modified dietary history questionnaire was used to estimate the energy intake of individual subjects. Stratification by age and sex was used in the allocation of subjects to each dietary group. The study protocol was approved by the Medical Ethics Committee of the University of Crete Medical School. An outline of the study design is provided in Figure 1.

Diet, instruction and compliance

The olive oils given to the subjects were Koroneiki extra virgin olive oil (phenol content 43 mg/kg) and Tsounati olive oil (phenol content 308 mg/kg). The phenol concentration was determined as described by Montedoro et al (1993). The percentages of the various phenols in the two oils were as follows: the high phenol oil contained 2% tyrosol, 1% hydroxytyrosol, 72% oleuropein-aglcones and 25% ligstroside-aglycones, whereas the corresponding percentages for the low phenol oil were 16, 2, 13 and 69%, respectively. Vitamin E was added to the high phenol oil so that the oils had the same vitamin E content (232 mg/kg). The subjects consumed 70 g of olive oil per day, ie 21.6 mg/day of phenols vs 3.0 mg/day of phenols for the high and low phenol oils, respectively. The olive oil was poured over the food consumed, 35 g at each of two meals. The subjects were blind to the type of oil they were given. On weekdays, meals were eaten under supervision at the University Hospital. On Fridays, the subjects were given the olive oil they had to consume during the weekend, with detailed instructions as to how the food should be prepared without oil.

Subjects were requested to maintain their usual food, drink and smoking habits as far as possible during the study. They were asked not to take vitamin supplements or aspirin, the latter being a radical scavenger. They were requested not to consume olives and other oil-containing products at any point during the study. In daily diaries, subjects recorded lapses of compliance with the protocol in addition to any signs of illness, medications taken, alcohol and foods eaten other than those provided.

Blood sampling and analysis

Two venous blood samples (about 40 ml) were taken at the end of each intervention period, after an overnight fast (on days 18 and 21). The various markers of oxidation in plasma were total plasma resistance to oxidation, concentrations of protein carbonyl as a marker of protein oxidation, malondialdehyde and lipid hydroperoxides as markers of lipid oxidation, and FRAP (ferric reducing ability of plasma) as a marker of the antioxidant capacity in plasma. Serum vitamin concentrations and plasma uric acid were also measured.

The susceptibility of total plasma to copper-mediated oxidation was determined by monitoring the formation of conjugated dienes (Princen et al, 1992). Malondialdehyde in plasma was determined as described by Wong et al (1987) with minor modifications (for further details see Vissers et al, 2001b). Lipid hydroperoxides in plasma were determined by the K-Assay LPO-CC kit (Kamiya Biomedical Co, Seattle, WA, USA) and an ELISA method was used in determining protein carbonyls (Buss et al, 1997). FRAP was determined using the Benzie and Strain method (Benzie & Strain, 1996). Vitamin concentrations in the serum were determined by high performance liquid chromatography (HPLC, Waters Instruments, Milfort, USA); for further details see Vissers et al (2001b). Plasma uric acid was determined using the UA plus kit (Boehringer, Mannheim, Germany).

Anthropometric measurements

At the end of the run-in period, the height and weight of the subjects were measured. Weight was also recorded at lunchtimes twice a week during the experimental periods to check stability of body weight during the study. Body mass index (BMI) was calculated as body weight divided by height squared (kg/m²).

Statistical analysis

The study had 80% power (with n=12 in each arm) to demonstrate a difference of 10% in lag time of LDL oxidation. An estimated standard deviation of 13.5% for the mean difference in lag time between the two olive oils was used in sample size calculations (Snedecor & Cochran, 1989). There was no evidence of major skewness in the distributions of the variables under consideration. Standard tests for two period cross-over trials were applied (Armitage & Berry, 1987; Pocock, 1983; Hills & Armitage, 1979). Two-sample t-tests were applied to assess the following effects, for each of the biochemical variables: (a) treatment effects; (b) period (time) effects, and (c) treatment-period interaction effects. In each case a pooled estimate of variance was used. Further statistical details of the estimates of effects can be found in the footnote to Table 2.

Results

The average lag time of copper-induced formation of conjugated dienes in total plasma was 0.83 mins longer after the high phenol diet (95% CI -6.5 to 8.1 min), as can be seen in Table 2. The average maximum rate of diene formation was equal after both treatments. In neither case, was there any statistical evidence of a dietary phenol effect. There was also no effect on the other markers of oxidation: the observed concentrations of malondialdehyde and lipid hydroperoxides in the plasma were on average 0.03 and 0.24 µmol/l lower after the high phenol diet, but again these differences were not statistically significant (with 95% CIs -0.06, 0.01 and -0.61, 0.13 µmol/l, respectively, see Table 2). Both FRAP and protein carbonyls displayed an average zero difference in concentration between high and low phenol diets.

For six of the 11 biochemical variables considered (results presented in Tables 2 and 3) there was evidence of a time trend (ie a period effect) between periods of intervention. The period effects were greatest in the vitamin concentrations but these measurements were not considered to be part of the intervention, as they were determined mainly as a check of equality between the two diets (see Table 3).

The body weights of the individuals remained stable over the entire period of the study: the mean weight in the first period was 70.9 (s.d. 13.3) kg and in the second period 71.0 (s.d. 13.0) kg, the weight of each individual being averaged over the six measurements in each period. The difference in weight between the two periods ranged from -2.7 to maximum 1.5 kg.

Discussion

Overall, the results of the present study do not provide any evidence that in vitro plasma susceptibility to oxidation and other markers of oxidation may be reduced by consumption of a diet rich in phenols from olive oil for 3 weeks, compared with a low-phenol diet. Markers of antioxidant status were found to be at similar levels, in the main, in a study performed in parallel to the present study at the University of Wageningen, The Netherlands, using the same olive oils as given in the present study with 46 non-smoking participants. With respect to study design, there were only three main differences between the Cretan and Dutch studies. Firstly, in the Dutch study, olive oil intake was adjusted for body weight and ranged from 55 to 102 g per day (with mean 69 g/day). Secondly, the Dutch subjects were instructed for the entire study period to have a background diet low in vitamin E (because of its antioxidant activity). Thirdly, in the Dutch study, oil was incorporated in sauces, mayonnaise, raisin rolls and cookies while in the Cretan study olive oil was poured over the meals. The findings of the Dutch study have been published separately (Vissers et al, 2001b). The only substantial difference found in the common measurements of the two studies was the average lipid hydroperoxide levels in the low phenol diet, which were 0.36 (s.d. 0.52) µmol/l in the Dutch study as compared with 0.71 (s.d. 0.87) µmol/l in the present study. The Dutch study also investigated oxidizability of HDL and LDL separately, but found no significant differences between low and high phenol olive oils (Vissers et al, 2001b).

The findings of the Cretan and Dutch studies tie in with two other human studies addressing the effect of minor components in olive oil on the susceptibility of LDL to oxidation in fasting blood (Nicolaïew et al, 1998; Bonanome et al, 2000). One reason for the lack of significant differences between the two diets might be that neither study addressed postprandial effects, as the testing was of bloods following an overnight fast. If the clearance of phenols from plasma is fast, as indicated by the findings of Visioli et al (2000), phenol concentrations may fall to levels too low to be detected after 12 h. However, a further recent study found that olive oil phenols did also not decrease the susceptibility of LDL to oxidation in postprandial samples if compared with a control group (Vissers et al, 2001a). Another explanation is that higher quantities of olive oil phenols may be necessary to detect a significant result, something that would be very difficult to impose on human subjects (Vissers et al, 2001b). The mean difference in phenol intake between the two diets was approximately 19 mg/day, which may not be a large enough difference for effects to be detected. The natural concentration of phenols in olive oil might be too low to produce an effect on markers of oxidation in the post-absorptive phase (Vissers et al, 2001b). Also, the background diet of the Cretan population is such that the intake of phenols from olive oil is already relatively high compared to intakes in non-Mediterranean countries, and the amount of antioxidants contained may be too high to allow for additional antioxidant effects to occur.

In our study it was not possible to establish the effectivity of the run-in and wash-out periods, as fat sources other than olives and oil-containing products were not controlled for in these periods. However, there was no a priori reason to believe that the within-subject sources of fat intake differed between run-in and wash-out periods. Also, no differences were evident from the foods that the subjects recorded in their daily diaries. Blood was not collected after the wash-out period because the design of the study was such that it was not necessary to take baseline measurements before the start of the second period and taking such measurements would have resulted in a loss of power. Another point to consider is that the estimates obtained were, in some cases, found to differ significantly between experimental periods. Possible reasons for this period effect need to be examined in more detail. From Table 1 it can be seen that the baseline characteristics of the two groups were similar (no evidence was found of a difference between the two groups for any variable, using the non-parametric Mann-Whitney test). As the external environment did not change (meals provided at the hospital, weights measured twice a week etc), it may be that during the washout period the subjects may have unknowingly consumed foods that affected the results in the second period. This explanation seems somewhat unlikely, however, as the blood samples were taken at the end of the experimental period so the markers would need to be affected after 3 weeks for the explanation to be valid.

It should be noted that, as the trial was relatively small in size, the treatment-period interaction test may not be very sensitive. In general, when an interaction is found in crossover studies, the best policy is to leave the within-patient analysis and confine the analysis to the first period using a two-sample t-test (Pocock, 1983). Using only data from the first period is, however, less precise than the full cross-over approach as it is subject to between-subject variation. For the plasma lipid hydroperoxide measurements in which evidence of an interaction was detected, the single-period approach was taken (using the Mann-Whitney test), but no evidence of significant diet differences were found.

It is only within the last few years that studies have focused on the antioxidant properties of olive oil in humans. Although no statistically significant findings resulted from the present study, the authors believe that these findings should be presented and discussed to aid further studies in taking account of possible limiting factors which were not known a priori to affect outcome in this relatively uninvestigated area of human research. The present study does not provide definitive evidence that the phenol content of olive oils does not have an effect on oxidation resistance in humans, but rather that no evidence is provided of such an effect under the conditions of the study.

Acknowledgements

We thank Unilever Research Vlaardingen, The Netherlands for financial support of the project. Many thanks also go to Professor MB Katan for his helpful advice. We are indebted to our research team at the time of the study; in particular thanks go to Dr Christos Hatzis for unfailing organizational and technical support, Giorgia Martimianaki for vitamin analyses, to the dieticians Eva Mouka, Ioanna Apostolaki and Foteini Cheiladaki and to our secretarial administrator, Gianna Kelesi for her assistance in coordination of the study. Finally, many thanks go to the participating staff at Wageningen University and of course to all the volunteers in Heraklion for their time and enthusiastic compliance.

Armitage P, Berry G. (1987). Statistical Methods in Medical Research 2nd edn; pp 223-225, Oxford: Blackwell Scientific.

Aviram M, Eias K. (1993). Dietary olive oil reduces low-density lipoprotein uptake by macrophages and decreases the susceptibility of the lipoprotein to undergo lipid peroxidation. Ann. Nutr. Metab., 37: 75-84. MEDLINE

Benzie IF, Strain JJ. (1996). The ferric reducing ability of plasma (FRAP) as a measure of 'antioxidant power': the FRAP assay. Anal. Biochem., 239: 70-76. MEDLINE

Bonanome A, Pagnan A, Biffanti S, Opportuno A, Sorgato F, Dorella M, Maiorino M, Ursini F. (1992). Effect of dietary monounsaturated and polyunsaturated fatty acids on the susceptibility of plasma low density lipoproteins to oxidative modification. Arterioscler. Thromb., 12: 529-533. MEDLINE

Bonanome A, Pagnan A, Caruso D, Toia A, Xamin A, Fedeli E, Berra B, Zamburlini A, Ursini F, Galli G. (2000). Evidence of postprandial absorption of olive oil phenols in humans. Nutr. Metab. Cardiovasc. Dis., 10: 111-120. MEDLINE

Buss H, Chan TP, Sluis KB, Domigan NM, Winterbourn CC. (1997). Protein carbonyl measurement by a sensitive ELISA method. Free Rad. Biol. Med., 23: 361-366. MEDLINE

Hills M, Armitage P. (1979). The two-period cross-over clinical trial. Br. J. Clin. Pharmac., 8: 7-20.

Hulea SA, Olinescu R, Nita S, Crocnan D, Kummerow FA. (1995). Cigarette smoking causes biochemical changes in blood that are suggestive of oxidative stress: a case-control study. J. Environ. Pathol. Toxicol. Oncol., 14: ((3-4)) 173-180. MEDLINE

Keys A, Menotti A, Karvonen MJ, Aravanis C, Blackburn H, Buzina R, Djordjevic BS, Dontas AS, Fidanza F, Keys MH, Kromhout D, Nedeljkovic S, Punsar S, Seccareccia F, Toshima H. (1986). The diet and 15-year death rate in the seven countries study. Am. J. Epidemiol., 124: 903-915. MEDLINE

MacNee W. (2000). Oxidants/antioxidants and COPD. Chest, 117: ((5 Suppl 1)) 303S-317S. MEDLINE

Mamalakis G, Kafatos A. (2001). Mediterranean diet and longetivity. In: The Mediterranean Diet: Constituents and Health Promotion A Matalas, A Zampelas, V Stavrinos & I Wolinsky, pp 205-223, Boca Raton, FL: CRC Press.

Montedoro G, Servili M, Baldioli M, Selvaggini R, Miniati E, Macchioni A. (1993). Simple and hydrolyzable compounds in virgin olive oil. III. Spectroscopic characterizations of the secoiridoid derivatives. J. Agric. Food Chem., 41: 2228-2234.

Nicolaïew N, Lemort N, Adorni L, Berra B, Montorfano G, Rapelli S, Cortesi N, Jacotot B. (1998). Comparison between extra virgin olive oil and oleic acid rich sunflower oil: effects on postprandial lipemia and LDL susceptibility to oxidation. Ann. Nutr. Metab., 42: 251-260. MEDLINE

Pocock SJ. (1983). Clinical Trials: a Practical Approach pp 114-119, Chichester: John Wiley.

Princen HM, van Poppel G, Vogelezang C, Buytenhek R, Kok FJ. (1992). Supplementation with vitamin E but not beta-carotene in vivo protects low density lipoprotein from lipid peroxidation in vitro. Effect of cigarette smoking. Arterioscler. Thromb., 12: 554-562. MEDLINE

Reaven P, Parthasarathy S, Grasse BJ, Miller E, Almazan F, Mattson FH, Khoo JC, Steinberg D, Witztum JL. (1991). Feasibility of using an oleate-rich diet to reduce the susceptibility of low-density lipoprotein to oxidative modification in humans. Am. J. Clin. Nutr., 54: 701-706. MEDLINE

Scaccini S, Nardini M, D'Aquino M, Gentili V, Di Felice M, Tomassi G. (1992). Effect of dietary oils on lipid peroxidation and on antioxidant parameters of rat plasma and lipoprotein fractions. J. Lipid Res., 33: 627-633. MEDLINE

Serra-Majem L, Ribas L, Treserras R, Ngo J, Salleras L. (1995). How could changes in diet explain changes in coronary heart disease mortality in Spain? The Spanish paradox. Am. J. Clin. Nutr., 61: ((Suppl)) 1351S-1359S. MEDLINE

Snedecor GW, Cochran WG. (1989). Statistical Methods 8th edn, pp 102-106, Ames, IA: The Iowa State University Press.

Visioli F, Galli C, Bornet F, Mattei A, Patelli R, Galli G, Caruso D. (2000). Olive oil phenolics are dose-dependently absorbed in humans. FEBS Lett., 468: 159-160. MEDLINE

Vissers MN, Zock PL, Leenen R, Roodenburg AJC , van Putte KPAM, Katan MB. (2001a). Effect of consumption of phenols from olives and extra virgin olive oil on LDL oxidizability in healthy humans. Free Rad. Res., 35: 619-629.

Vissers MN, Zock PL, Wiseman SA, Meyboom S, Katan MB. (2001b). Effect of phenol-rich extra virgin olive oil on markers of oxidation in healthy volunteers. Eur. J. Clin. Nutr., 55: 334-341.

Wiseman SA, Mathot JN, de Fouw N, Tijburg LB. (1996). Dietary non-tocopherol antioxidants present in extra virgin olive oil increase the resistance of low density lipoproteins to oxidation in rabbits. Atherosclerosis, 120: 15-23. Article MEDLINE

Wong SH, Knight JA, Hopfer SM, Zaharia O, Leach-CN J, Sunderman-FW J. (1987). Lipoperoxides in plasma as measured by liquid-chromatographic separation of malondialdehyde-thiobarbituric acid adduct. Clin. Chem., 33: 214-220. MEDLINE

Zhou JF, Yan XF, Guo FZ, Sun NY, Qian ZJ, Ding DY. (2000). Effects of cigarette smoking and smoking cessation on plasma constituents and enzyme activities related to oxidative stress. Biomed. Environ. Sci., 13: 44-55. MEDLINE

Figure 1 Outline of crossover trial.

Table 1 Baseline characteristics of the 25 study subjects

Table 2 Oxidizability of plasma, and markers of antioxidant status at the end of the high and the low phenol diet

Table 3 Levels of lycopene, -carotene, retinol, -tocopherol, and uric acid at the end of the high and the low phenol diet