CHD is the leading cause of morbidity and mortality in Western countries. An elevated concentration of LDL-cholesterol (LDL-C) is a well-established independent risk factor for atherosclerosisReference Eisenberg1. In addition, LDL oxidation has been implicated by several studies as one of the initial steps of atherogenesis, and therefore associated with higher risk of CHDReference Ross2, Reference Witztum and Steinberg3. Elevated concentrations of oxidised LDL have recently been identified as a strong predictor for subsequent acute CHD events in healthy menReference Meisinger, Baumert, Khuseyinova, Loewel and Koenig4.
Plant sterols (PS) have long been reported to have significant cholesterol-lowering propertiesReference Pollak5. Studies conducted throughout the past decades suggested that an intake of 1·5–2 g PS/d is needed to achieve an optimal LDL-C lowering effectReference Katan, Grundy, Jones, Law, Miettinen and Paoletti6. It is generally accepted that PS decrease circulating cholesterol concentrations by suppressing intestinal absorption of cholesterol due to the higher affinity of PS to micelles compared to cholesterol, resulting in less cholesterol being incorporated in chylomicronsReference Ikeda and Sugano7, Reference Heinemann, Kullak-Ublick, Pietruck and von Bergmann8. Several reports have demonstrated that the solubility of PS may play an important role in the process of PS incorporation into micellesReference Ostlund9. Indeed, low intestinal bioavailability of purified phytosterols was shown to be elevated by esterification to fatty acids, dissolving in dietary diacylglycerol (DAG) oil, or by emulsifying with lecithin micellesReference Gremaud, Dalan, Piguet, Baumgartner, Ballabeni, Decarli, Leser, Berger and Fay10–Reference Spilburg, Goldberg, McGill, Stenson, Racette, Bateman, McPherson and Ostlund13. To date, the most common process used to enhance the solubility of PS is by esterifying PS with n-6 PUFA, such as soyabean oil and sunflower oil (SO) fatty acids. PS can therefore be incorporated into fatty foods, such as margarines and spreadsReference Weststrate and Meijer14. Emerging new approaches consist of esterifying PS to fatty acids associated with additional health benefits, such as fish oil fatty acidsReference Demonty, Chan, Pelled and Jones15.
Early epidemiological evidence showed a lower incidence of CHD in Mediterranean countriesReference Aravanis, Corcondilas, Dontas, Lekos and Keys16 where olive oil (OO) is the primary source of fatReference Stark and Madar17. The consumption of OO, which contains high levels of oleic acid (a MUFA), was inversely associated with IHD, presumably due to hypolipidaemic effectsReference Mattson and Grundy18. This notion has been further confirmed in several clinical intervention trialsReference Kratz, Cullen, Kannenberg, Kassner, Fobker, Abuja, Assmann and Wahrburg19–Reference Puiggros, Chacon, Armadans, Clapes and Planas21. MUFA supplementation was also shown to have protective effects against lipid peroxidationReference Kratz, Cullen, Kannenberg, Kassner, Fobker, Abuja, Assmann and Wahrburg19, Reference Reaven and Witztum22. Likewise, PS were also reported to have antioxidant propertiesReference Berger, Jones and Abumweis23.
The objective of this study was to assess whether a novel formulation of PS that had been esterified with OO fatty acids (PS-OO) would exert the effects of its components on the blood lipid profile and lipid peroxidation. We tested this hypothesis in mildly overweight, hypercholesterolaemic subjects who consumed an OO-rich diet that was further supplemented either with PS-OO or with PS esterified to SO fatty acids (PS-SO).
Twenty-four volunteers (eleven males, thirteen postmenopausal females) were recruited from the Montreal area by an advertisement posted in local newspapers. The inclusion criteria were as follows: baseline LDL-C >2·6 mmol/l (100 mg/dl), BMI ranging from 24 to 30 kg/m2 and aged 30–65 years. Subjects were excluded if they had taken medications known to affect lipid metabolism, such as cholestyramine, colestipol, gemfibrozil, probucol, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase inhibitors, fish oil capsules and supplements containing PS, during the previous 3 months. Subjects who had been diagnosed with diabetes mellitus, kidney disease or liver disease, and those who smoked, consumed more than two glasses per day of alcoholic beverages and/or took two or more doses per week of laxatives or concentrated sources of fibre were also excluded. Subjects with thyroid disease (n 2) and with high blood pressure (n 4) were included in the study since they had been stable in response to thyroid and blood pressure treatments, and their medications were maintained throughout the trial. One subject on hormone replacement therapy was also included in the study and her regimen was maintained at the same dose throughout the study period.
Fasting blood samples were obtained for screening purposes. Before enrolment in the study, the candidates provided a medical history and underwent a complete physical examination conducted by the study physician. The study protocol was reviewed and approved by the Human Ethical Review Committee of the Faculty of Agriculture and Environmental Sciences for the School of Dietetics and Human Nutrition at McGill University (protocol number REB# 808-0403). All subjects received explanations about the protocol and written consent forms were obtained from each participant.
Experimental design, protocol and diets
The study was a semi-randomized, crossover, double blind, clinical intervention trial using a Latin square sequence. It consisted of three 28 d phases separated by a 4-week washout interval. During each dietary treatment phase, the subjects were provided with an OO-based, weight-maintaining, North-American diet. During the washout period, the subjects consumed their own habitual diets. All subjects received the control OO diet during the first phase, after which they were randomly assigned to the two other dietary treatments.
The basic diet contained approximately 15 % energy as protein, 55 % energy as carbohydrates, 30 % energy as fat, of which approximately 70 % was provided by OO, 80 mg cholesterol per 4186·8 kJ (1000 kcal); and 12 g fibre per 4186·8 kJ (1000 kcal). The OO of the basic diet provided 0·02 g/d naturally occurring free PS. The basic diet composition is shown in Table 1. Table 2 lists the composition of the treatment oils. During the two randomized phases, an isoenergetic amount of OO was replaced by either (i) 21·4 g low-fat PS-SO margarine providing the equivalent of 1·7 g soyabean sterols (Take control®, Unilever Bestfoods NA, Baltimore, MD) or (ii) 9·1 g PS-OO containing the equivalent of 1·7 g soyabean sterols enzymatically esterified to OO fatty acids, as well as 1·4 g DAG (Enzymotec Ltd, Migdal HaEmeq, Israel). No antioxidants were supplemented to the PS-OO matrix or other tested matrices.
OO, control olive oil; PS-SO, plant sterols esterified with sunflower oil fatty acids; PS-OO, plant sterols esterified with olive oil fatty acids; ND, not detected.
* Typical values.
The diets were provided to the subjects by the Mary Emily Clinical Nutrition Research Unit at McGill University. These diets consisted of three isoenergetic meals, prepared according to a 3-d cycle menu. All meals (breakfast, lunch and supper) were prepared in the metabolic kitchen of the clinic where the foods were weighed precisely to 0·5 g during meal preparation. Treatment oils were ingested at breakfast and under supervision to monitor compliance. The subjects were instructed to consume only foods and beverages provided by the clinic.
All three experimental diets were isoenergetic. Individual basal energy requirements were calculated by the Mifflin equationReference Mifflin, St Jeor, Hill, Scott, Daugherty and Koh24, and were then multiplied by a physical activity factor of 1·7 to meet the total energy requirements of mildly to moderately active healthy adults. Energy intake was adjusted during the first 2 weeks of the control phase to maintain a constant body weight. A comparable energy level was maintained during the two subsequent phases. Subjects were encouraged to keep a constant exercise level throughout the study to ensure that body weights remained unchanged.
Fasting blood samples were collected from the subjects on days 1, 2, 28 and 29 of each phase. On day 28 of each phase, postprandial plasma TAG concentrations were measured 4 h after breakfast. A complete blood count was done on day 29 for monitoring purposes.
Blood samples were collected in vacutainer tubes and centrifuged for 15 min at 1000 g at 4°C, within 30 min, to separate plasma from erythrocytes. Plasma and erythrocytes were immediately stored at − 80°C until lipid analysis. Total cholesterol (TC), HDL-cholesterol (HDL-C), and TAG concentrations were measured by automated methods on the multianalyzer Dimension RxL Max utilizing enzymatic reagents Flex (Dade Behring Diagnostic, Marburg, Germany). LDL-C was calculated by the Friedewald equationReference Friedewald, Levy and Fredrickson25 for individuals with TAG levels < 4·5 mmol/l, while LDL-C was measured directly by the abovementioned enzymatic methods when TAG levels were >4·5 mmol/l. Apo A-I, apo B-100 and lipoprotein (a) (Lp(a)) levels were measured on the BN ProSpec Nephelometer (Dade, Behring Diagnostics, Marburg, Germany) utilizing the Dade Behring N Antisera assays to determine apo A-I, apo B-100, and Lp(a), respectively (Behring Diagnostics).
Plasma cholesterol precursor and plant sterols
Plasma PS concentrations were determined from the non-saponifiable material of plasma lipids as reported previouslyReference Ntanios and Jones26. Briefly, an internal standard, 5α-cholestane, was added to 0·5 ml plasma sample. Plasma samples were saponified with 0·5 m methanolic KOH for 1 h at 100°C and the non-saponifiable materials were extracted with petroleum diethyl ether. After extraction, the samples were derivatized with 1·3 ml TMS reagent (pyridine–hexamethyldisilazan–trimethylchlorosilane; 9:3:1, by volume); Sigma-Aldrich Canada Ltd, Oakville, ON, Canada). The samples were then injected into GLC (HP 5890 Series II, Hewlett Packard, Palo Alto, CA, USA), equipped with a flame ionization detection and auto-injector system and with a 30 m capillary column. The column temperature was 285°C and the isothermal running conditions were maintained for 30 minutes. Detector and injector temperatures were set to 310°C and 300°C, respectively. Lathosterol, campesterol, stigmasterol and β-sitosterol peaks were identified by comparison with authenticated standards (Sigma-Aldrich Canada Ltd).
Plasma lipid peroxidation
Apo B-containing lipoproteins were precipitated with manganese chloride–heparin by ultracentrifugationReference Gidez, Miller, Burstein, Slagle and Eder27. The LDL fraction was re-suspended in normal saline after centrifugation. The thiobarbituric acidreactive substance (TBARS) assayReference Lefevre, Beljean-Leymarie, Beyerle, Bonnefont-Rousselot, Cristol, Therond and Torreilles28 was used to measure lipid peroxidation in the plasma LDL subfraction (OXItek, ZeptoMetrix Corporation, Buffalo, NY, USA).
Based on previous publications with a comparable designReference Jones, Ntanios, Raeini-Sarjaz and Vanstone29, twenty subjects would be required in order to detect a clinically significant difference ( − 0·48 mmol/l) in LDL-C levels, the primary outcome, using 0·84 mmol/l as a standard deviation, at the 0·05 level of significance and 80 % power. A total enrolment of twenty-four patients was originally estimated as being required to enable a study dropout rate of 15 %.
All data were expressed as means and their standard errors. Statistical significance was set at P < 0·05 for all analyses. Variables that were not normally distributed were log transformed before analysis. Differences in plasma variables were tested by repeated-measures ANOVA with the type of dietary matrix in each intervention arm as the within-subject factor and with endpoint values as the dependent variable. Baseline values were inserted into the model as covariates if their interaction with dietary matrices was found to be statistically significant. Subsequently, contrast analyses were used to identify differences between pairs of diets. When variables failed to demonstrate any treatment effect, the two-tailed paired Student's t test was used to compare baseline and endpoint values within each diet phase. Furthermore, a modified Cohen's effect size was calculated for endpoint values to evaluate changes from the baseline OO diet. Data were analyzed with the use of SAS software (version 8.0; SAS Institute Inc, Cary, NC, USA).
Twenty-four subjects (eleven males, thirteen females) were recruited and twenty-one subjects (eleven males, ten females) completed the entire trial. The three females who dropped out during the first phase reported difficulties with the transportation to the clinic (n 1) or with daily clinic visits (n 1) and personal affairs (n 1). The baseline characteristics of study subjects who completed the trial are displayed in Table 3. No side effects were reported after consuming the treatment oils. Results from the complete blood count at the end of each phase were within the normal range for all subjects (data not shown). The mean baseline bodyweight for OO, PS-SO, and PS-OO (73·9 (se 2·7) kg, 74·7 (se 2·8) kg, and 74·4 (se 2·8) kg, respectively) and the percentage changes in bodyweight values ( − 0·7 (se 0·3) %, − 0·4 (se 0·2) %, and − 0·7 (se 0·2) %, respectively) did not differ between treatments. Baseline values (following washout periods) of all the characteristics presented in Table 3 were not statistically different between dietary phases (data not shown).
Plasma lipid concentrations
Table 4 lists plasma lipid concentrations at the end of each treatment phase. Supplementation of an OO-based diet with either PS-SO or PS-OO resulted in reduced (P = 0·0218 and 0·0185, respectively) LDL-C levels compared with control OO, but there was only a mild tendency towards a reduction (P = 0·0839) in TC levels. TC:HDL-C ratios following PS-SO treatment were lower (P = 0·0018) compared with control OO, but they were not significantly different from those observed with PS-OO. Consumption of the PS-matrices did not influence plasma HDL-C or TAG (fasting and postprandial) concentrations (Table 4).
OO, control olive oil; PS-SO, plant sterols esterified with sunflower oil fatty acids; PS-OO, plant sterols esterified with olive oil fatty acids.
* P values obtained by repeated-measures ANOVA, with †baseline concentrations included in the model as covariates. Values of plasma lipid concentrations were normalized using a log transformation.
a,b Values not sharing a common superscript letter are significantly different at P < 0·05.
The effects of the dietary fats and PS matrices on apo concentrations are also presented in Table 4. Apo A-I concentrations were higher (P = 0·0052 and P < 0·0001) following PS-OO administration relative to control OO-diet and PS-SO, respectively, presumably due to differences in baseline values (P < 0·0001). PS-containing diets had a strong tendency to induce (P = 0·0577) lower (10–11%, change from baseline) apo B-100 concentrations. This was associated with a mild tendency towards decreased (P = 0·1030) LDL-C:apo B ratios, especially following the PS-OO treatment (an effect size of 0·47). Apo B-100:apo A-I ratios were lower (P = 0·0052) with PS-SO (6%, relative to baseline) and, to a lesser degree (P = 0·2057), with PS-OO (4%, relative to baseline) administrations compared with the control OO-diet. The effect of PS-SO consumption (an effect size of 0·35) on apo B-100:apo A-I ratios did not differ (P = 0·8698) from the effect of PS-OO supplementation (an effect size of 0·31).
Plasma Lp(a) concentrations were not altered (P = 0·1182) by the dietary treatments at the end of the feeding phases. When endpoint values were compared to baseline concentrations, however, consumption of PS-OO did not influence Lp(a) levels, while the control OO-based diet and PS-SO treatments resulted in an increase (P = 0·0050 and P = 0·0421, respectively) of Lp(a) concentrations (Fig. 1).
Plasma lipid peroxide concentrations
The dietary treatments did not impact (P = 0·1295) endpoint plasma TBARS concentrations. However, when baselines and endpoints were compared (Fig. 2), the consumption of PS-OO was shown to have resulted in reduced ( − 13 %; P = 0·0097) TBARS levels, whereas the OO-based diet ( − 10 %; P = 0·0993) and the PS-SO ( − 3 %; P = 0·1640) treatments failed to produce a comparable effect.
Plasma neutral sterol concentrations
Consumption of PS-SO and PS-OO resulted in a statistically significant increase in plasma campesterol, stigmasterol, and β-sitosterol concentrations compared with the control OO-based diet (Table 5). Similar observations were obtained when plasma PS concentrations were normalized to cholesterol levels, although to a somewhat larger extent. The concentration of cholesterol precursor, lathosterol and its ratio to cholesterol were elevated (P = 0·0031 and 0·0268, P = 0·0564 and 0·0047) by PS-SO and PS-OO treatments, respectively, compared with the control OO-diet.
OO, control olive oil; PS-SO, plant sterols esterified with sunflower oil fatty acids; PS-OO, plant sterols esterified with olive oil fatty acids.
* P values obtained by repeated-measures ANOVA, with †baseline concentrations included in the model as covariates. Values of plasma lipids concentrations were normalized using a log transformation.
a,b Values not sharing a common superscript letter are significantly different at P < 0·05.
Our results suggest that supplementation of an OO-based diet with OO fatty acids esterified to PS and mixed with dietary DAG reduces LDL-C levels and could lower LDL susceptibility to oxidation compared with an OO-based diet in hypercholesterolaemic mildly-overweight subjects. Consumption of both PS-OO and PS-SO diets favoured comparable beneficial reductions of CHD-related risk factors. However, in the context of an OO background diet, PS-OO feeding improved plasma antioxidant properties and protected against the increase in Lp(a) levels over the study period, while PS-SO supplementation did not show such an action.
In the current study, PS-OO and PS-SO supplementation to the base OO-diet tended to reduce TC concentrations by − 8 and − 6 %, relative to baseline respectively. Importantly, we have shown that ingestion of 1·7 g/d PS esterified to OO or SO fatty acids significantly decreased LDL-C concentrations by 6–9 % compared to the base OO-diet. These decreases are consistent with results from a recent meta-analysisReference Katan, Grundy, Jones, Law, Miettinen and Paoletti6, in which an intake of 1·5–1·9 g PS /d was associated with a 7·0–10·1 % reduction in LDL-C levels. Moreover, plasma levels of apo B-100 tended to lower with consumption of PS diets, but to a somewhat lesser extent than LDL-C concentrations. These observations suggest that the PS treatments may have affected the cholesterol content of the LDL particles more than their number. On the other hand, there was a significant reducing effect of PS-containing treatments on apo B-100:apo A-I ratio, suggesting that, in fact, these diets favour a beneficial suppression of apo B-100 levels compared with the control OO-diet. Taken together, our findings suggest that supplementation with PS provides superior protection against coronary artery disease risk factors than a healthy OO-based diet.
In this study, the PS concentrations in plasma increased following consumption of both PS-SO and PS-OO as compared with the base OO-diet. Plasma PS concentrations are naturally low since they are poorly absorbed. Nevertheless, high PS intake has been consistently shown to substantially increase circulating PS concentrationsReference Jones, Vanstone, Raeini-Sarjaz and St-Onge30, Reference Vanstone, Raeini-Sarjaz, Parsons and Jones31. In the current study, however, the degree of elevation in plasma PS concentrations and their ratios to cholesterol due to PS consumptions were 2–3-fold lower than what we had previously observed, when PS were administered equally across two to three daily mealsReference Vanstone, Raeini-Sarjaz, Parsons and Jones31. Taken together with the somewhat limited LDL-C lowering effect, these findings suggest that a single morning dose of PS may result in a lower PS bioavailability and therefore lesser efficacy. The link between PS bioavailability and treatment efficacy warrants further investigation.
Compared with the baseline values, the serum lipid peroxides associated with LDL particles were significantly reduced following PS-OO treatment. Early publications suggested that OO consumption resulted in LDL enrichment with oleic acid and, consequently, in a greater resistance to oxidationReference Aviram and Eias32. Furthermore, a diet enriched in MUFA, rather than PUFA, was shown to inhibit LDL oxidationReference Kratz, Cullen, Kannenberg, Kassner, Fobker, Abuja, Assmann and Wahrburg19, Reference Puiggros, Chacon, Armadans, Clapes and Planas21, Reference Berry, Eisenberg, Haratz, Friedlander, Norman, Kaufmann and Stein33. The presence of high levels of PS in dietReference Homma, Ikeda, Ishikawa, Tateno, Sugano and Nakamura34 or in vitro Reference van Rensburg, Daniels, van Zyl and Taljaard35 has been associated with decreased lipid peroxidation. Interestingly, similar observations were recently noted following administration of MUFA-enriched PS mixed with DAG to atherosclerotic apo E deficient miceReference Fuhrman, Plat, Herzog and Aviram36. Therefore, the antioxidant properties of MUFA-diets and PS could have contributed, at least in part, to the reduction in TBARS concentrations observed in our study on PS-OO.
In the current study, OO-diet and PS-SO supplementation resulted in increased Lp(a) levels over the study period, whereas no such deleterious effect was observed following PS-OO consumption. Elevated concentrations of Lp(a) have been suggested to be a risk factor for a variety of atherosclerotic and thrombotic disordersReference Boffa, Marcovina and Koschinsky37, Reference Tsimikas, Lau, Han, Shortal, Miller, Segev, Curtiss, Witztum and Strauss38. The atherogenicity of Lp(a) may be mediated, at least in part, by associated proinflammatory oxidized phospholipidsReference Tsimikas, Lau, Han, Shortal, Miller, Segev, Curtiss, Witztum and Strauss38, Reference Tsimikas, Bergmark, Beyer, Patel, Pattison, Miller, Juliano and Witztum39, but the link between Lp(a) and LDL oxidation in response to dietary modifications remains unclear. Although OO was shown in some studies to protect LDL from oxidationReference Aviram40, Reference Fito, Covas, Lamuela-Raventos, Vila, Torrents, de la Torre and Marrugat41, there are several reports in which its consumption resulted in elevated Lp(a) concentrationsReference Mensink and Katan42, Reference Vessby, Unsitupa and Hermansen43. Likewise, PUFA-enriched diets have been shown to increase Lp(a) levelsReference Silaste, Rantala, Alfthan, Aro, Witztum, Kesaniemi and Horkko44, presumably via interacting with several transcription factors, such as NF-κB and PPAR αReference Maziere, Conte, Degonville, Ali and Maziere45, Reference Tai, Corella, Demissie, Cupples, Coltell, Schaefer, Tucker and Ordovas46. In contrast, dietary DAG ingestionReference Teramoto, Watanabe, Ito, Omata, Furukawa, Shimoda, Hoshino, Nagao and Naito47 was recently associated with a slight but significant reduction in Lp(a) concentrations, possibly as a result of decreased hepatic fat contents. In our current study, the PS-OO treatment, which contained equivalent amounts of OO fatty acids as the control OO-diet, in addition with DAG maintained low Lp(a) levels. Although the levels of DAG administered in the PS-OO dietary matrix were considerably lower than the dose of DAG that had been shown to be effective in human clinical studiesReference Mensink and Katan42, Reference Rudkowska, Roynette, Demonty, Vanstone, Jew and Jones48, its presence could have contributed to maintaining Lp(a) levels. Taken together with the reduction in plasma TBARS concentrations, these results suggest that PS-OO may exert protective actions against oxidative stress.
In conclusion, our findings demonstrate that providing PS-containing matrices to hypercholesterolaemic, mildly overweight subjects fed with an OO-based diet results in optimized plasma lipid concentrations. Furthermore, consumption of PS-OO, but not PS-SO, may reduce the susceptibility of LDL to oxidative stress, which, in turn, could protect against increases in Lp(a) concentrations. Therefore, in the context of an OO-based diet, supplementation with PS-OO matrixed with DAG in OO may offer a greater level of protection against CHD than the traditional PS-SO formula to hypercholesterolaemic individuals.
We thank Dr Joel Lavoie who performed the lipid analyses at the Montreal Cardiology Institute and Dr William Parsons who was the study physician. We also acknowledge the staff of the Mary Emily Clinical Nutrition Research Unit. We thank Esther Shabtai from the Statistics Services Unit, Tel Aviv Sourasky Medical Centre, Israel for helping with the statistical analyses and Esther Eshkol for editorial assistance. We acknowledge Dr Tzafra Cohen and Dr Yael Herzog from Enzymotec Ltd. for their helpful comments on the manuscript. The sunflower oil esters of plant sterols were kindly provided by Unilever (USA). This study was funded by Enzymotec Ltd, Israel. Except for D. Pelled, who is the Director of Clinical Studies at Enzymotec Ltd, none of the authors had any personal or financial conflict of interest.