ORIGINAL COMMUNICATION a Biochanin-Enriched Isoflavone from Red Clover Lowers LDL Cholesterol in Men

ORIGINAL COMMUNICATION A biochanin-enriched isoflavone from red clover lowers LDL cholesterol in men

P Nestel1*, M Cehun1, A Chronopoulos1, L DaSilva2, H Teede2 and B McGrath2

1Baker Medical Research Institute Wynn Domain, Monash University, Melbourne, Australia; 2Department of Vascular Medicine, Monash University, Melbourne, Australia

Objective: To determine whether the two major isoflavones in red clover differ in their effect on low-density lipoprotein cholesterol (LDL-C). Design: A randomised, placebo-controlled, double-blind trial; two parallel groups taking one of the two isoflavones within which treatment and placebo were administered in a crossover design. Setting: Free-living volunteers. Subjects: A total of 46 middle-aged men and 34 postmenopausal women. Intervention: Two mixtures of red clover isoflavones enriched in either biochanin (n ¼ 40) or formononetin (n ¼ 40) were compared. Placebo and active treatment (40 mg/day) were administered for 6 weeks each in a crossover design within the two parallel groups. Main outcome measures: Plasma lipids were measured twice at the end of each period. Results: Baseline LDL-C concentrations did not differ significantly between men (n ¼ 46) and women (n ¼ 34), nor between those randomised to biochanin or formononetin. Interaction between time and treatments, biochanin, formononetin and corresponding placebos (two-way ANOVA) on LDL-C showed a significant effect of biochanin treatment alone. The biochanin effect was confined to men; median LDL-C was 3.61 (3.05–4.14) mmol/l with biochanin and 3.99 (3.16–4.29) mmol/l with the corresponding placebo (RM ANOVA with Dunnett’s adjustment Po0.05). The difference between placebo and biochanin effects on LDL-C was 9.5%. No other lipid was affected and women failed to respond significantly to treatment. Conclusion: Isolated isoflavones from red clover enriched in biochanin (genistein precursor) but not in formononetin (daidzein precursor), lowered LDL-C in men. This may partly explain the previous failure to demonstrate cholesterol-lowering effects with mixed isoflavones studied predominantly in women. Sponsorship: Novogen Ltd, North Ryde NSW, Australia, provided partial support including provision of tablets and outside monitoring. European Journal of Clinical Nutrition (2004) 58, 403–408. doi:10.1038/sj.ejcn.1601796

Keywords: isoflavones; red clover; biochanin; formononetin; LDL cholesterol

Introduction when increasing amounts of isoflavones within soy protein Since the publication of the meta-analysis of the lipid- were compared (Crouse et al, 1999). However, trials of lowering effect of soy protein (Anderson et al, 1995), the purified isoflavones have to date been disappointing (Nestel precise constituent in soy responsible for reducing the et al, 1997, 1999; Hodgson et al, 1998; Howes et al, 2000; low-density lipoprotein cholesterol (LDL-C) concentration Simons et al, 2000). Further, there is large variability in remains uncertain. The isoflavone content is likely to response to soy as shown in the meta-analysis, since about contribute strongly, since isoflavone-depleted soy protein half of the 31 studies analysed by Anderson et al (1995) appears less effective (Gardner et al, 2001; Wangen et al, showed only minimal LDL-C lowering for reasons that have 2001) and one large study has shown a dose-related effect not been resolved. Among recent well-designed and exe- cuted trials, the reduction in non-high-density lipoprotein cholesterol (HDL-C) has been as low as 2.6% (Teixeira et al, *Correspondence: P Nestel, Baker Medical Research Institute, PO Box 2000) or that of LDL-C no greater than with the milk 6492, St Kilda Rd Central, Melbourne 8008, Australia. E-mail: [email protected] protein-based control diets (Teede et al, 2001; Meinertz et al, Received 5 March 2003; revised 23 April 2003; accepted 16 May 2003 2002). As is seen with other cholesterol-lowering nutritional Isoflavone lowers LDL cholesterol in men P Nestel et al 404 interventions, the best responses to soy protein were shown Table 1 Characteristics of subjects by gender and by treatment by individuals with the highest initial LDL-C levels (Crouse et al, 1999; Vigna et al, 2000). On the other hand, two recent Age Body BMI Subjects Number (y) mass (kg) (kg/m2) studies that compared two soy protein products with very

different isoflavone contents reported no specific LDL-C Men 46 58 (7) 80.5 (11.1)* 25.9 (2.7)** lowering attributable to isoflavones (Jenkins et al, 2002; Women 34 58 (6) 65.6 (12) 24.5 (3.7) Lichtenstein et al, 2002). Studies with intact soy protein have Formononetin 40 58 (7) 73.6 (11.5) been conducted in both men and women with no apparent Formononetin placebo 40 73.9 (11.1) Biochanin 40 57 (6) 75.1 (15.6) et al gender difference (Crouse , 1999). Adding to these Biochanin placebo 40 74.9 (15.7) variables is the large interindividual variation in isoflavone et al n absorption (Tsunoda , 2002). A puzzling result is the Significantly different between men and women (Po0.001); ** (Po0.02). apparent absence of a response to purified isoflavones when Differences in body weight between the two treated groups and their placebo taken in amounts that show a benefit within soy protein. periods were not significant. Values are means (s.d.) Alternative compounds in soy carry fewer experimental credentials and have been reviewed recently (Erdman, 2000). Only one of the trials with pure isoflavones included men (Hodgson et al, 1998), although it has been recognized Table 2 Baseline lipids measured after run-in, prerandomisation phase that women and men may respond differently to other dietary interventions such as changing the intakes of Men Women saturated fatty acids and cholesterol. Clifton and Nestel Lipid (mmol/l) Biochanin Formononetin Biochanin Formononetin (1992) in a study of 26 men and 25 women matched for age, LDL-C and BMI, found that men responded to dietary Total cholesterol 5.57 (0.85) 5.56 (1.05) 5.40 (1.05) 6.17 (0.87) saturated fat and cholesterol with a greater rise in LDL-C, LDL-C 3.73 (0.85) 3.67 (0.95) 3.37 (0.94) 3.85 (1.05) whereas women showed greater increments in HDL-C. Cobb HDL-C 1.21 (0.3) 1.30 (0.3) 1.62 (0.46) 1.78 (0.44) et al (1992) also found higher LDL-C/HDL-C responses Triacylglycerol 1.42 (0.93) 1.29 (0.56) 0.90 (0.48) 1.18 (0.47) among men than women as the saturation of the dietary fat was altered. Values are means (s.d.), n=80. The primary objective of the present study was therefore to evaluate the changes in LDL-C with supplements of the two main individual isoflavones (biochanin and formononetin) m2, medication for lowering lipids or blood pressure, present in red clover. This was studied in 46 men and 34 diabetes mellitus, a past or present history of cancer or women with preparations that comprised predominantly chronic bowel disease, allergy to soy, unusual dietary habits biochanin (B; the precursor of genistein) or predominantly and unwillingness to discontinue supplements of any kind formononetin (F; the precursor of daidzein) in a randomised, or adhere to a predetermined diet. For women, hormone double-blind, placebo-controlled trial. replacement therapy within the last 2 months and absence of a recent normal mammogram (42 y) were additional exclusions. Methods The project was approved by the Human Ethics Commit- Study population tee of the Alfred Group of Hospitals and signed consent was Middle-aged men and women were recruited by advertise- obtained from the volunteers. ment. The women had been postmenopausal for 41 y and shown to have FSH values above 40 mmol/l. In all, 82 subjects were recruited, but two withdrew prior to randomi- Experimental design sation. Men (n ¼ 46) and women (n ¼ 34) were of similar The 80 subjects were allocated by random numbers to receive mean age, but the men were heavier (means and standard either B or F with equal numbers receiving B and F. Each deviations): men 80.5 (11.0) kg and women 65.6 (11.9) kg isoflavone was compared against placebo pills of similar (Po0.001); respective BMI values for men and women were colour and consistency. Among the 46 men, 25 received B 25.9 (2.7) and 24.5 (3.7) kg/m2 (Po0.02) (Table 1). The 80 and 21 F; among the 34 women, 15 received B and 19 F. subjects were randomised to either B or F (80 mg daily in two Each study began with a 2-week run-in, during which tablets) irrespective of sex; the two treatment groups did not subjects became acquainted with procedures such as record- differ in average age or average body mass. Their baseline ing daily the taking of placebo pills and completing 3-day lipid values are shown in Table 2. food frequency questionnaires (simplified by the Anti- Inclusion criteria were: age 45–70 y; postmenopausal status Cancer Council of Victoria, Australia, that produced and for women; plasma total cholesterol 4–7.5 mmol/l; plasma analysed the food records). The questionnaires were com- triacylglycerolo3.5 mmol/l. Exclusion criteria were: smok- pleted during the final 3 days of the run-in and the final ing, alcohol intake 44 standard drinks daily, BMI434 kg/ intervention periods, respectively. Volunteers had been

European Journal of Clinical Nutrition Isoflavone lowers LDL cholesterol in men P Nestel et al 405 advised to maintain their entry patterns of food intake and 24 h at 371C with glucuronidase and the free isoflavones the macronutrients that may affect LDL-C as shown in were eluted with 3 ml ethanol from a C-18 solid-phase Table 3 demonstrated no significant differences during the extraction column (Waters, Sydney, Australia), and separated trial period. The average difference in body mass between a by HPLC. The HPLC system consisted of a C-18 stationary- treatment period and its corresponding placebo period of phase column and a gradient acetonitrile/water mobile o0.8 kg (Table 1) was not significant. phase. The limit of detection of the assay for each isoflavone The subjects were allocated randomly to receive placebo or was 5 ng/ml and the inter-assay CV was o15%. active treatment first, for a period of 6 weeks. This was followed by a 1-week washout (placebo) before the crossover to the alternate 6-week period. The total duration of the Data analysis study was 15 weeks. Blood was collected into EDTA-contain- Responses to isoflavones vs. placebo were analysed initially ing tubes twice after each 6-week period, mostly on for the 46 men and the 34 women for within-subject consecutive days and never more than 3 days apart. Subjects interactions and time or order effects. The analyses included had fasted overnight for not less than 10 h. both isoflavones and their respective placebo periods. The two isoflavone preparations were obtained from Interaction between time and treatment was tested by two- Novogen Ltd (North Ryde, NSW, Australia). The mixtures way ANOVA. Since this indicated an effect of biochanin were concentrated ethanolic extracts from red clover treatment on LDL-C, further analyses were carried out separated into two fractions. B comprised biochanin:formo- separately for men and women. Comparisons were analysed nonetin in a ratio of approximately 3.5:1 with 4% genistein by RM ANOVA with appropriate adjustments for multiple and o1% daidzein. F comprised formononetin:biochanin in comparisons. The distribution of LDL-C was not normal a ratio of 4.9:1 with o1% genistein and daidzein. among the men and Friedman repeated-measures analysis of Urinary excretion was measured over a 24 h period at the variance on ranks was substituted and adjusted by Dunnett’s end of the placebo and isoflavone periods. all pairwise multiple comparison procedure. In men, the differences between each treatment and its placebo were compared for B vs. F by one-way ANOVA adjusted by Tukey Test. Laboratory procedures Plasma cholesterol and triacylglycerol concentrations and HDL-C were measured in a specialised laboratory using Results standard enzyme-based methods and measurement in a Table 4 shows the trial data for plasma lipids in men and Cobas-Bio automated analyser (Roche, Basel, Switzerland). women, respectively. The distribution of LDL-C among men LDL-C was calculated. was not normal and hence both mean (standard deviation) Urinary isoflavones were measured as described by Tsuno- and medians (25–75% range) are given. da et al (2002). Briefly, aliquots of urine were incubated for ANOVA showed that there was no effect of order of therapy on any lipid variable. Following 6 weeks on placebo, total plasma cholesterol and LDL-C values did not differ Table 3 Macronutrient intakes at the end of run-in period and end of significantly between men and women. HDL-C concentra- study tion was significantly greater among the women (Po0.01, t- Daily intake End run-in period End of study test). Plasma triacylglcerol tended to be higher among men, but not significantly so. Analysis of differences according to Energy (kJ) 7953 (2994) 7458 (2454) isoflavone randomisation, especially for LDL-C, showed no Fat (g) 73.1 (33.1) 67.0 (28.9) significant differences during placebo phases, although the (%en) 33.0 32.3 two comparisons were marginally different. Men rando- Saturated fatty acids (g) 27.8 (12.9) 25.7 (11.6) mised to B tended to show higher initial LDL-C levels than (%en) 12.6 12.4 Monounsaturated FA (g) 25.4 (12.0) 23.7 (10.9) women (P ¼ 0.061, t-test). Among men, LDL-C levels did not (%en) 11.4 11.4 differ significantly between those drawn to B or F; among the Polyunsaturated FA (g) 13.1 (7.8) 11.3 (6.2) women, those drawn to F had marginally higher initial LDL- (%en) 5.9 5.5 C levels than those drawn to B (P ¼ 0.067, t-test). Protein (g) 88.9 (32.4) 83.8 (27.5) (%en) 17.9 18.0 Effect of treatment on LDL-C was significant for B Carbohydrate (g) 223 (87) 212 (75) treatment (P ¼ 0.026; two-way ANOVA). (%en) 45.0 45.5 The results were further analysed among men and women Dietary fibre (g) 26.1 (10.0) 24.1 (8.7) by type of treatment. Among men, the respective median Cholesterol (mg) 257 (117) 244 (100) LDL-C concentrations for placebo and B were 3.99 and 3.61 mmol/l (Friedman RM ANOVA on ranks: P ¼ 0.026; with None of the differences between run-in period and end of study was significant. Alcohol consumption was o4% energy. Calculated from 3-day Dunnett’s adjustment Po0.05). Median LDL-C tended to be food frequency data. Values are means (s.d.), n=80. lower with F than with placebo (3.62 vs. 3.87 mmol/l), but

European Journal of Clinical Nutrition Isoflavone lowers LDL cholesterol in men P Nestel et al 406 Table 4 Plasma lipid data for all treatments in men and women

Lipid (mmol/l) Biochanin Placebo Formononetin Placebo

Cholesterol Men 5.39 (0.75) 5.54 (0.75) 5.52 (1.27) 5.66 (1.23) Women 5.25 (1.01) 5.27 (1.06) 6.10 (0.94) 6.21 (1.11)

LDL-C Men Mean 3.64 (0.93) 3.82 (0.84) 3.65 (1.09) 3.68 (1.09) Median 3.61 (3.05–4.14) 3.99a (3.16–4.29) 3.62 (3.08–4.23) 3.87 (2.9–4.36) Women Mean 3.24 (0.86) 3.26 (0.95) 3.83 (0.92) 3.92 (0.98)

HDL-C Men 1.18 (0.28) 1.21 (0.29) 1.28 (0.28) 1.35 (0.31) Women 1.64 (0.38) 1.59 (0.33) 1.79 (0.53) 1.80 (0.49)

Triacylglycerol Men 1.36 (0.71) 1.21 (0.47) 1.29 (0.71) 1.37 (0.71) Women 0.92 (0.41) 0.90 (0.49) 1.09 (0.43) 1.18 (0.49)

aRM ANOVA with Dunnetts adjustment for multiple comparisons, Po0.05; comparison of median LDL-C concentrations at the end of biochanin and placebo periods. n=25 men biochanin; 21 men formononetin; 15 women biochanin; 19 women formononetin. Values are means (s.d.) except for medians (25–75%) for LDL-C in men.

the difference was not significant. The difference between flavones was 18.9 (23.6) mg/day, indicative of the large the LDL-C medians for B and placebo in men was variability among individuals. 0.38 mmol/l (9.5%). The difference between the two isoflavone effects was tested directly by comparing the mean differences in LDL-C between each isoflavone and its placebo. Among the 25 men Discussion receiving B, the mean difference for LDL-C was 0.18 mmol/l The results support the possibility that individual isoflavones and the corresponding value among the 21 men receiving F affect LDL-C differentially. Only in men was LDL-C lowered was 0.03 mmol/l. The difference for LDL-C between mean significantly with a mixture in which biochanin predomi- (placebo minus B) vs. mean (placebo minus F) was significant nated (B). Two issues arise therefore: the reasons for the (Po0.05; one-way ANOVA adjusted by Tukey Test for all different gender response and for the greater benefit from pairwise multiple comparisons). biochanin, the precursor of genistein. The present findings Isoflavones recovered in urine showed as expected that of a differential response between men and women and most, but by no means all, the primary isoflavones, between the responses to biochanin and formononetin may formononetin and biochanin were converted into daidzein help explain the failure by ourselves (Nestel et al, 1997, 1999) and genistein, respectively (Tsunoda et al, 2002). Further, the and others (Hodgson et al, 1998; Howes et al, 2000; Simons total excreted by individuals consuming B was less than that et al, 2000) to demonstrate an effect on plasma LDL-C with excreted with the consumption of F, as noted by others isolated isoflavones. In each of the previous studies, the (Busby et al, 2002). Of particular interest was whether a isoflavones contained a mixture of B and F that would have minority of individuals who excreted substantial amounts of diluted any effect attributable to B alone. equol, a metabolite of daidzein and hence of formononetin, The present findings are consistent with other evidence for showed a different LDL-C response than nonequol excretors, a differential effect of dietary interventions on LDL-C despite formononetin itself having no effect. In total, 15 between women and men. Gender differences were noted subjects were found to excrete substantial amounts of equol by Savolainen et al (1991), who found that men had a as defined by Rowland et al (2000). Eight had been significantly greater rise in LDL-C than women in response randomised to F and seven to B: the mean (s.d.) LDL-C to a high dietary fat intervention. Clifton and Nestel (1992) concentrations were 3.39 (0.74) and 3.29 (0.65) mmol/l with had shown that women responded to an increase in placebo and isoflavone treatments, respectively. This differ- saturated fat and cholesterol with a rise in HDL-C, whereas ence was not significant, nor was the difference in LDL-C older men, especially those with higher baseline LDL-C significant between equol excretors taking F or B, although levels, showed substantial increases in LDL-C. Cobb et al the numbers are small. Total excretion of recovered iso- (1992) also reported a gender difference in lipoprotein

European Journal of Clinical Nutrition Isoflavone lowers LDL cholesterol in men P Nestel et al 407 responses to changes in the saturation of fat. Among women, controversy over the need to have isoflavones retained HDL-C underwent the greater change so that the LDL- within soy protein is not resolved. However, there is a C:HDL-C ratio in men remained higher than in women. suggestion that isoflavones, like oestrogens, may upregulate However, Geil et al (1995) found that both men and women the LDL receptors (Lovati et al, 1987; Baum et al, 1998) and responded similarly to a fat-reduced diet. thereby accelerate clearance of LDL-C from the circulation. If In this context, the failure of isolated isoflavones, from that requires interaction with oestrogen receptors in liver both soy and red clover, to lower LDL-C might be viewed as and other tissues, then biochanin/genistein might be being partly due to the predominance of women in these expected to exert a greater effect than formononetin/ studies. Clearly, however, the effects of isoflavones differ daidzein. from those of dietary total fat, fatty acid type and Urinary excretion of isoflavones is a very approximate cholesterol. Changes in the consumption of the latter indication of absorption. The recovery during the interven- commonly give rise to changes in HDL-C as well as to LDL- tion with F of 18.9 mg/day or just under 25% of the ingested C and under appropriate circumstances, also in triacylglycer- amount is similar to amounts that we have reported (Nestel ol. With the exception of the large study by Howes et al et al, 1997; Tsunoda et al, 2002). Nonurinary routes of (2000), isolated red clover isoflavones have not lowered excretion and failure to measure minor metabolites (only triacylglycerol significantly. In that study, a significant equol and O-desmethylangloensin were measured) contri- inverse relationship was observed between triacylglycerol bute to underestimation. Out of 80 subjects, 15 excreted levels and the excretion of genistein and isoflavone meta- large amounts of equol and the others mostly none. It is bolites, suggesting that triacylglycerol fell in those women generally accepted that equol excretors are in the minority who absorbed isoflavones best. (Rowland et al, 2000). The interest in equol derives partly It should, however, be noted that studies of soy protein from the fact that equol binds to oestrogen receptors with have also not been uniformly effective in changing plasma greater affinity than its parent compound formononetin and lipid levels. Many of the trials included in the large meta- equally with that of genistein (Morito et al, 2001). analysis of soy protein (Anderson et al, 1995) failed to lower In summary, LDL-C was lowered significantly with a LDL-C. Other recent trials have also not found differences biochanin-rich isoflavone mixture isolated from red clover, between soy protein isolates and control proteins, mostly but this was seen only in men and not in women. It is a casein (Gardner et al, 2001; Teede et al, 2001). By contrast, novel finding that requires confirmation. A formononetin- Crouse et al (1999) showed a clear soy isoflavone-related enriched isoflavone mixture was ineffective in both sexes. effect on LDL-C. Despite the large number of subjects in that Neither HDL-C nor triacylglycerol concentrations were study, 156 women and men, there was no clear gender– affected by isoflavone supplementation. isoflavone interaction, since LDL-C lowering was of marginal significance in the postmenopausal women (P ¼ 0.07) and in the men; premenopausal women showed no change. The basal LDL-C concentration appeared to be the best predictor Acknowledgements of the isoflavone-mediated reduction in LDL-C, much as has Novogen Ltd (North Ryde, NSW, Australia) provided the been observed with other dietary interventions. isoflavone supplements and placebo tablets and arranged an To what extent baseline LDL-C might have been a factor in outside monitor to supervise the trial. All investigators the present study is uncertain, but we believe it to be minor contributed to the design and interpretation of the for the following reasons. LDL-C concentrations did not experiment. differ significantly between the men and women during the placebo phase. Among those randomised to B, men had slightly higher LDL-C levels than women but not signifi- References cantly so. Among men, in whom alone LDL-C concentra- Anderson JW, Johnstone BM & Cook-Newell MB (1995): Meta- tions were lowered with B, the placebo LDL-C levels were not analysis of the effects of soy protein intake on serum lipids. N. significantly different between those treated with B and F. Engl. J. Med. 333, 276–382. Baum JA, Teng H, Erdman Jr JW, Weigel RM, Klein BP, Persky VW, On the other hand, women randomised to receive F tended Freels S, Surya P, Bakhit RM, Ramos E, Shay NF & Potter SM (1998): to show, at the end of the placebo phase, higher LDL-C Long-term intake of soy protein improves blood lipid profiles and levels: 3.92 (0.98) mmol/l than those randomised to B: 3.26 increases mononuclear cell low-density-lipoprotein receptor mes- (0.95) mmol/l; P ¼ 0.067; yet there was no difference in LDL- senger RNA in hypercholesterolemic, postmenopausal women. Am. J. Clin. Nutr C lowering. . 68, 545–551. Busby MG, Jeffcoat AR, Bloedon LT, Koch MA, Black T, Dix KJ, Heizer The differential responses to B and F might have been due WD, Thomas BF, Hill JM, Crowell JA & Zeisel SH (2002): Clinical to the fact that genistein, which is derived from biochanin characteristics and pharmacokinetics of purified soy isoflavones: shows substantially greater affinity for the oestrogen recep- single-dose administration to healthy men. Am. J. Clin. Nutr. 75, 126–136. tor than does daidzein, which is derived from formononetin Clifton PM & Nestel PJ (1992): Influence of gender, body mass index, (Setchell, 1998). The precise mechanism for the cholesterol- and age on response of plasma lipids to dietary fat plus reducing effect of isoflavones is not known. Indeed, the cholesterol. Arterioscler. Thromb. 12, 955–962.

European Journal of Clinical Nutrition Isoflavone lowers LDL cholesterol in men P Nestel et al 408 Cobb MA, Teitlebaum H, Risch N, Jekel J & Ostfeld A (1992): Morito K, Hirose T, Kinjo J, Hirakawa T, Okawa M, Nohara T, Ogawa Influence of dietary fat, apolipoprotein E phenotype, and sex on S, Inoue S, Muramatsu S & Masamune Y (2001): Interaction of plasma lipoprotein levels. Circulation 86, 849–857. phytoestrogens with estrogen receptors a and b. Biol. Pharm. Bull. Crouse JR, Morgan T, Terry JG, Ellis J, Vitolins M & Burke GL (1999): 24, 351–356. A randomized trial comparing the effect of casein with that of soy Nestel PJ, Pomeroy S, Kay S, Komesaroff P, Behrsing J, Cameron JD & protein containing various amounts of isoflavones on plasma West L (1999): Isoflavones from red clover improve systemic concentrations of lipids and lipoproteins. Arch. Intern. Med. 159, arterial compliance but not plasma lipids in menopausal women. 2070–2076. J. Clin. Endocrinol. Metab. 84, 895–898. Erdman Jr JW (2000): Soy protein and cardiovascular disease. A Nestel PJ, Yamashita T, Sasahara T, Pomeroy S, Dart A, Komesaroff P, statement for healthcare professionals from the Nutrition Com- Owen A & Abbey M (1997): Soy isoflavones improve arterial mittee of the AHA. Circulation 102, 2555–2559. compliance but not plasma lipids in menopausal women. Gardner CD, Newell KA, Cherin R & Haskell WL (2001): The effect of Arterioscler. Thromb. Vasc. Biol. 17, 3392–3398. soy protein with or without isoflavones relative to milk protein on Rowland IR, Wiseman H, Sanders TA, Adlercreutz H & Bowey EA plasma lipids in hypercholesterolemic postmenopausal women. (2000): Interindividual variation in metabolism of soy isoflavones Am. J. Clin. Nutr. 73, 728–735. and lignans: influence of habitual diet on equol production by the Geil PB, Anderson JW & Gustafson NJ (1995): Women and men with gut flora. J. Nutr. 36, 27–32. hypercholesterolemia respond similarly to an American Heart Savolainen MJ, Rantala M, Kervinen K, Jarvi L, Suvanto K, Rantala T Association step 1 diet. J. Am. Diet. Assoc. 95, 436–441. & Kesaniemi YA (1991): Magnitudes of dietary effects on plasma Hodgson JM, Puddey IB, Beilin LJ, Mori TA & Croft KD (1998): concentration: role of sex and apoprotein E phenotype. Athero- Supplementation with isoflavonoid phytoestrogens does not alter sclerosis 86, 145–152. serum lipid concentrations: a randomized controlled trial in Setchell KD (1998): Phytoestrogens: the biochemistry, physiology humans. J. Nutr. 128, 728–732. and implications for human health of soy isoflavones. Am. J. Clin. Howes JB, Sullivan D, Lai N, Nestel P, Pomeroy S, West L, Eden JA & Nutr. 68(Suppl), 1333S–1346S. Howes LG (2000): The effects of dietary supplementation with Simons LA, von Konigsmark M, Simons J & Celermajer DS (2000): isoflavones from red clover on the lipoprotein profiles of post- Phytoestrogens do not influence lipoprotein levels or endothelial menopausal women with mild to moderate hypercholesterolae- function in healthy, postmenopausal women. Am. J. Cardiol. 85, mia. Atherosclerosis 152, 143–147. 1297–1301. Jenkins DA, Kendall CWC, Jackson C-J C, Connelly PW, Teede HJ, Dalais F, Kotsopoulos D, Liang YL & McGrath BP (2001): Parker T, Faulkner D, Vidgen E, Cunnane SC, Leiter LA & Soy protein dietary supplementation containing phytoestrogens Josse RG (2002): Effects of high- and low-isoflavone soyfoods improves lipid profiles and blood pressure: a double-blind, on blood lipids, oxidized LDL, homocysteine, and blood randomized, placebo-controlled study in men and post-menopau- pressure in hyperlipidemic men and women. Am. J. Clin. Nutr. sal women. J. Clin. Endocr. Metab. 86, 3053–3060. 76, 4365–4372. Teixeira SR, Potter SM, Weigel R, Hannum S, Erdman Jr JW & Hasler Lichtenstein AH, Jalbert SM, Adlercreutz H, Goldin BR, Rasmussen H, CM (2000): Effects of feeding 4 levels of soy protein for 3 and 6 wk Schaefer EJ & Ausman LA (2002): Lipoprotein response to diets on blood lipids and apolipoproteins in moderately hypercholes- high in soy or animal protein with and without isoflavones in terolemic men. Am. J. Clin. Nutr. 71, 1077–1084. moderately hypercholesterolemic subjects. Arterioscler. Thromb. Tsunoda N, Pomeroy S & Nestel P (2002): Absorption in humans Vasc. Biol. 22, 1852–1858. of isoflavones from soy and red clover is similar. J. Nutr. 132, Lovati MR, Manzoni C, Canavesi A, Sirtori M, Vaccarino V, 2199–2201. Marchi M, Gaddi A & Sirtori CR (1987): Soybean protein Vigna GB, Pansini F & Bonaccorsi G (2000): Plasma lipoproteins in diet increases low density lipoprotein receptor activity in mono- soy-treated postmenopausal women: a double-blind, placebo- nuclear cells from hypercholesterolemic patients. J. Clin. Invest. controlled trial. Nutr. Metab. Cardiovasc. Dis. 10, 315–322. 80, 1498–1502. Wangen KE, Duncan AM, Xu X & Kurzer MS (2001): Soy isoflavones Meinertz H, Nilausen K & Hilden J (2002): Alcohol-extracted, but not improve plasma lipids in normocholesterolemic and mildly intact dietary soy protein lowers lipoprotein(a) markedly. Arter- hypercholesterolemic postmenopausal women. Am. J. Clin. Nutr. ioscler. Thromb. Vasc. Biol. 22, 312–316. 73, 225–231.

European Journal of Clinical Nutrition