Chronic Ingestion of Flavan-3-Ols and Isoflavones Improves Insulin

Clinical Care/Education/Nutrition/Psychosocial Research ORIGINAL ARTICLE

Chronic Ingestion of Flavan-3-ols and Isoﬂavones Improves Insulin Sensitivity and Lipoprotein Status and Attenuates Estimated 10-Year CVD Risk in Medicated Postmenopausal Women With Type 2 Diabetes A 1-year, double-blind, randomized, controlled trial

1 2 PETER J. CURTIS, PHD KETAN DHATARIYA, MD he global prevalence of diabetes is in- 2 3 MIKE SAMPSON, MD PAUL A. KROON, PHD 1 1 creasing, with recent predictions sug- JOHN POTTER, MD AEDÍN CASSIDY, PHD T gesting that complications of diabetes accounts for 7% of all-cause mortality and 12% of healthcare costs worldwide (1). d fl OBJECTIVE To assess the effect of dietary avonoids on cardiovascular disease (CVD) risk in Pharmacological treatment is critical in de- postmenopausal women with type 2 diabetes on established statin and hypoglycemic therapy. laying the underlying progression of di- RESEARCH DESIGN AND METHODSdDespite being medicated, patients with type 2 abetes (2), yet it remains inadequate in diabetes have elevated CVD risk, particularly postmenopausal women. Although dietary flavo- preventing the increased risk of cardio- noids have been shown to reduce CVD risk factors in healthy participants, no long-term trials vascular disease (CVD) in patients with have examined the additional benefits of flavonoids to CVD risk in medicated postmenopausal type 2 diabetes, especially women (3). Gen- women with type 2 diabetes. We conducted a parallel-design, placebo-controlled trial with type der disparities in the treatment and control 2 diabetic patients randomized to consume 27 g/day (split dose) flavonoid-enriched chocolate fl fl of CVD risk factors have been reported (containing 850 mg avan-3-ols [90 mg epicatechin] and 100 mg iso avones [aglycone equiv- (3,4), and observational studies suggest alents)]/day) or matched placebo for 1 year. poorer metabolic control in women with RESULTSdNinety-three patients completed the trial, and adherence was high (flavonoid type 2 diabetes (4), suggesting potential 91.3%; placebo 91.6%). Compared with the placebo group, the combined flavonoid interven- gender differences in the mechanisms un- tion resulted in a significant reduction in estimated peripheral insulin resistance (homeostasis derlying disease progression. These data re- model assessment of insulin resistance [HOMA-IR] 20.3 6 0.2; P = 0.004) and improvement in inforce the need for effective preventative insulin sensitivity (quantitative insulin sensitivity index [QUICKI] 0.003 6 0.00; P = 0.04) as a fi 2 6 P fi and disease management strategies to re- result of a signi cant decrease in insulin levels ( 0.8 0.5 mU/L; = 0.02). Signi cant duce CVD risk associated with type 2 di- reductions in total cholesterol:HDL-cholesterol (HDL-C) ratio (20.2 6 0.1; P = 0.01) and LDL- 2 6 P abetes, particularly in women. cholesterol (LDL-C) ( 0.1 0.1 mmol/L; = 0.04) were also observed. Estimated 10-year total fl coronary heart disease risk (derived from UK Prospective Diabetes Study algorithm) was atten- Dietary avonoids are bioactive con- uated after flavonoid intervention (flavonoid +0.1 6 0.3 vs. placebo 1.1 6 0.3; P =0.02).No stituents, with observational data sup- effect on blood pressure, HbA1c, or glucose was observed. porting an association between high intakes of some flavonoid subclasses CONCLUSIONSdOne-year intervention with flavan-3-ols and isoflavones improved bio- fi fl and a reduced risk of CVD (5). In partic- markers of CVD risk, highlighting the additional bene tof avonoids to standard drug therapy in ular, there is growing evidence to support managing CVD risk in postmenopausal type 2 diabetic patients. the cardioprotective effects of two specific fl Diabetes Care 35:226–232, 2012 subclasses, the avan-3-ols (present in cocoa and tea) and isoflavones (present almost exclusively in soy). Observational studies support the association between high cocoa intake and reduced CVD risk ccccccccccccccccccccccccccccccccccccccccccccccccc and mortality (6), and improvements in From the 1Department of Nutrition, Norwich Medical School, University of East Anglia, Norwich, Norfolk, flow-mediated dilatation, hyperemic blood 2 U.K.; the Elsie Bertram Diabetes Centre, Norfolk and Norwich University Hospital NHS Trust, Norwich, flow, and blood pressure (BP) have fol- U.K.; and the 3Polyphenols and Health Group, Institute of Food Research, Norwich, Norfolk, U.K. Corresponding author: Aedín Cassidy, [email protected]. lowed short-term interventions with cocoa Received 1 August 2011 and accepted 25 October 2011. flavan-3-ols (7,8). A growing body of in DOI: 10.2337/dc11-1443. Clinical trial reg. no. NCT00677599, clinicaltrials.gov. vitro evidence also supports the potentially This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/suppl/doi:10 beneficial effects of flavan-3-ols on CVD .2337/dc11-1443/-/DC1. © 2012 by the American Diabetes Association. Readers may use this article as long as the work is properly risk, via effects on nitric oxide and glu- cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/ cose transporters (9,10). In Asian popula- licenses/by-nc-nd/3.0/ for details. tions, where soy is consumed as a staple,

226 DIABETES CARE, VOLUME 35, FEBRUARY 2012 care.diabetesjournals.org Curtis and Associates increased habitual intake of soy isoflavones effects of a 1-year combined interven- were conducted according to the principles has also been associated with a reduced risk tion of flavan-3-ols and isoflavones in expressed in the Declaration of Helsinki and of CVD and type 2 diabetes (11), and recent postmenopausal women with type 2 the study was approved by the Norfolk Re- meta-analyses suggest beneficial effects of diabetes. search Ethics Committee. All subjects pro- increased isoflavone intake on CVD risk vided written informed consent. biomarkers and glycemic control (12–14). RESEARCH DESIGN AND In patients with type 2 diabetes, no METHODS Study protocol long-term trials have been conducted to Subjects were randomized to placebo (n = examine the effects of either isoflavones or Subjects 59) or flavonoid-enriched (n =59)choc- flavan-3-ols on CVD risk biomarkers, One hundred and eighteen postmeno- olate daily for 1 year. Treatment allocation although the available data from short- pausal women (no menstruation for $12 was stratified according to age, BMI, years term studies (maximum duration 12 months), aged 51–74 years and receiving since menopause, insulin use, and patient weeks in both flavonoid subclasses) are standard U.K. care for type 2 diabetes (in- characteristics are shown in Table 1. suggestive of favorable effects (15–17). To cluding statin therapy, established for $12 The daily intake of 27 g flavonoid- our knowledge, no studies have exam- months and glycemic target of HbA1c of enriched chocolate (conceptualized by the ined the long-term effects of a combined 7.0%) were recruited through general prac- research team and formulated and pro- intervention of these bioactive flavonoids titioners, specialist diabetes clinics, and duced by Barry Callebaut [Lebbeke-Wieze, on CVD risk in medicated patients with local media advertisements (Fig. 1A). Ex- Belgium] specifically for the trial) pro- type 2 diabetes, and given our knowledge clusion criteria included use of hormone vided 90 mg epicatechin (850 mg total of their specific mechanisms of action, replacement therapy (within 6 months), flavan-3-ols) and 100 mg isoflavones there is the potential of synergistic effects. known allergy to the study foods, history (aglycone equivalents), the optimal dose Therefore, we conducted a randomized, of vascular disease or cancer, and poor di- for biological efficacy on the basis of double-blind, placebo–controlled, paral- abetes control or raised BP at screening (as published findings from previous short- lel trial to examine the vascular protective outlined in Fig. 1B). All study investigations term trials (18,19). Flavonoid extracts

Figure 1dEnrollment, randomization, and trial design. A: Enrollment and randomization. B: Inclusion/exclusion criteria. C: Assessment overview. MI, myocardial infarction; PE, physical exam.

care.diabetesjournals.org DIABETES CARE, VOLUME 35, FEBRUARY 2012 227 Type 2 diabetes and chronic ﬂavonoid intake

Table 1dBaseline characteristics of the 93 postmenopausal patients with type 2 diabetes and height) were taken in duplicate, ac- who completed the 1-year flavonoid intervention cording to standardized protocols. From banked plasma, insulin concentrations Flavonoid Placebo were assessed using routine methods (in- (n = 47) (n =46) sulin ELISA; Mercodia, Uppsala, Sweden). The intra-assay coefficients of variation 6 6 Age (years) 62.13 0.73 62.98 0.83 were 2.90% (HbA1c), 2.82% (glucose), BMI (kg/m2)32.696 1.09 31.85 6 0.87 1.24% (total cholesterol), 3.01% (HDL- Duration of diabetes (years) 7.04 6 1.05 5.76 6 0.70 C), 1.43% (triglycerides), and 3.84% (in- Years since menopause (years) 13.53 6 1.27 13.47 6 1.20 sulin). Insulin resistance and insulin Medicated with insulin (n [% group]) 9 [19%] 9 [20%] sensitivity were calculated using the ho- Medicated for hypertension (n [% group]) 28 [60%] 25 [54%] meostasis model assessment of insulin Past smoker (n [% group]) 14 [30%] 19 [41%] resistance (HOMA-IR) equation, HOMA- Insulin (mU/L) 7.82 6 0.77 8.79 6 0.83 IR = glucose 3 insulin/22.5 (23), and the Glucose (mmol/L) 7.58 6 0.29 7.67 6 0.29 quantitative insulin sensitivity index HOMA-IR 2.67 6 0.29 3.16 6 0.37 (QUICKI) (24) equation, QUICKI = 1/ QUICKI 0.343 6 0.005 0.336 6 0.005 (log [fasting insulin] + log [fasting glu- 6 6 HbA1c (%) 7.13 0.14 7.25 0.15 cose]), respectively. HOMA-IR, QUICKI, Cholesterol (mmol/L) 4.26 6 0.09 4.29 6 0.11 glucose, and insulin data are reported for LDL-C (mmol/L) 2.21 6 0.07 2.20 6 0.10 n = 45 (in both groups), triglycerides levels HDL-C (mmol/L) 1.40 6 0.05 1.34 6 0.05 for n =45flavonoid and n = 44 placebo Triglycerides (mmol/L) 1.44 6 0.09 1.69 6 0.12 subjects, and LDL-C levels for n =45fla- Mean 6 SEM or percentage. vonoid and n = 42 placebo subjects. Ten-year estimated risks for coronary heart disease (CHD), fatal CHD, stroke, were purchased (flavan-3-ols and Acticoa One week preceding and during the and fatal stroke were calculated using the cocoa [Barry Callebaut]; isoflavones and 1-year trial, habitual diet and exercise UK Prospective Diabetes Study (UKPDS) SoyLife40 [Frutarom, Veenendaal, the routines were maintained, with the excep- risk engine (version 2.0, www.dtu.ox.ac. Netherlands]), and intervention and pla- tion of restricting the intake of specific uk/ukpds_trial/index.php; Diabetes Tri- cebo chocolates were produced by Barry flavonoid-rich foods (a list of which was als Unit, University of Oxford, Oxford, Callebaut, which were matched for mac- provided). Four-day diet diaries were used U.K.). The UKPDS risk engine includes ronutrient content, appearance, and taste. to assess habitual dietary intake at base- age, diabetes duration, sex, presence of Batch-to-batch variability and effect of stor- line, 6 months, and 12 months (Supple- atrial fibrillation, ethnicity, smoking sta- age on flavonoid content were indepen- mentary Table 1), and adherence to study tus, HbA1c (%), systolic BP (mmHg), total dently determined using high-performance protocol was monitored by research staff cholesterol (mmol/L), and HDL-C liquid chromatography (HPLC) (20). between study assessments. Additionally, (mmol/L). The last observation carried Chocolate (13.5 g) was consumed for 24 h before study assessments, subjects forward was used to interpolate incom- n twice daily (lunchtime and evening) to were asked to refrain from caffeine, alcohol, plete data ( = 3 observations for HbA1c, maintain circulating levels of metabolites, and strenuous exercise. Each volunteer cholesterol, and HDL-C). given our knowledge of the half-life of consumed a low-flavonoid meal the even- the compounds (21). Subjects were ad- ing before each assessment followed by an Statistical analysis vised to exchange the study chocolate overnight fast (.10 h). Results are expressed as mean 6 SEM. At with foods of similar nutritional content At baseline and 12 months, vascular baseline (0 mol/L) and 12 months (12 (total daily intake provided 152 kcal, measures were made at the research fa- mol/L), differences in patient characteristics 10 g fat, and 13 g carbohydrate). Adher- cility, including 2-h ambulatory BP mon- were assessed using Student independent- ence to intervention was determined by itoring (Spacelabs Healthcare, Issaquah, sample t test (continuous data) and Pear- counting returned wrappers and objec- WA), with recordings every 10 min. Bi- son x2 test (categorical data). Data are tively assessed through quantification of ological samples (24-h urine and fasting presented for those completing the 1-year 24-h urinary total epicatechin (sum of blood) were also collected (Fig. 1C), with intervention per protocol, and the primary epicatechin, 39-methyl epicatechin, 49- levels of fasting glucose and lipoproteins test of effect was univariate ANCOVA methyl epicatechin, epicatechin-3-sulfate, (total cholesterol, HDL-cholesterol [HDL-C], analysis (intervention group as fixed fac- and epicatechin-methyl-sulfate) and iso- and triglycerides) measured photometri- tor [flavonoid/placebo] versus change in flavone (daidzein, genistein, and equol) cally using a clinical chemistry autoanalyzer variable [pre (0 mol/L)/post (12 mol/L)], excretion levels at 6 and 12 months deter- (ARCHITECT c Systems autoanalyzer; with baseline value of variable as covari- mined by HPLC with mass spectrometric Abbott Laboratories, Abbott Park, IL). ate). Further sensitivity analysis was con- (MS) detection after enzyme hydrolysis us- Fasting HbA1c was assessed by HPLC (A. ducted to exclude for low compliers ing established methods. Intra-assay coeffi- Menarini Diagnostics, Florence, Italy). (returning #70% wrappers), those with cients of variation (CVs) for total epicatechin, LDL-cholesterol (LDL-C) was determined annual change in BMI of greater than or daidzein, genistein, and equol levels were using the Friedewald equation (22). Fur- equal to 65kg/m2, and one volunteer with 7.7, 6.3, 8.0, and 8.1%, respectively, and thermore, plasma/urine aliquots were atrial fibrillation (n =5flavonoid; n =4 interassay CVs were 9.1, 11.0, 14.1, and stored at 2808C for subsequent analyses. placebo). Statistical analyses were per- 8.6%, respectively. Anthropometric measures (body weight formed using SPSS, release 18.0.0 (IBM

228 DIABETES CARE, VOLUME 35, FEBRUARY 2012 care.diabetesjournals.org Curtis and Associates HbA aaaemean are Data 2.21 2.67 (mmol/L) LDL-C HOMA-IR 2 Table D-:D- 0.66 0.343 HDL-C:LDL-C QUICKI lcs mo/)7.58 (mmol/L) Glucose nui m/)7.82 (mU/L) Insulin rgyeie mo/)1.44 (mmol/L) Triglycerides hlseo mo/)4.26 (mmol/L) Cholesterol D- mo/)1.40 (mmol/L) HDL-C oa hlseo:D- 3.19 cholesterol:HDL-C Total H ik(KD)()9.35 (%) (UKPDS) risk CHD aa H ik(KD)()6.14 (%) (UKPDS) risk CHD Fatal toers UPS % 7.16 (%) (UKPDS) risk Stroke aa toers UPS % 1.00 (%) (UKPDS) risk stroke Fatal Corp., New York, NY), and statistical signif- 133.68 (mmHg) BP Systolic icance was deﬁned as P , 0.05. 1c % 7.13 (%) d

RESULTSdNinety three patients com- with intervention combined 1-year of Effect pleted the trial, and subject characteristics 6 and key measures of diabetes control were * SEM. similar between groups at baseline (Table P

1); age ranged from 51 to 74 years, and , 2 BMI was 21.5–57.9 kg/m .Therewere ** 0.05. 25 withdrawals during the trial (21% of those randomized), with almost equal P numbers in both groups (Fig. 1A). There , mnh 2ots0 12months 0months were no differences in baseline character- 0.01. 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 istics of the 118 subjects randomized to 6 .3135.62 1.63 .72.10 0.07 2.34 0.29 .30.72 0.03 0.346 0.005 .97.48 0.29 .77.06 0.77 .91.46 0.09 .47.22 0.14 .94.22 0.09 .51.45 0.05 .13.02 0.11 .69.44 0.56 .96.53 0.49 .18.07 1.11 treatment (n = 118 in total; n =59infla- 1.15 0.15 vonoid and placebo group) (data not shown). In the 93 participants who completed the study, the mean 1-year adher- fl ence to intervention treatment was high, Placebo Flavonoid vn3osadiso and avan-3-ols 6 6 6 6 6 6 6 6 6 6 6 6 6 6 with 91.3 and 91.6% of chocolate bars 6 6 0.07 0.20 .30.06 0.03 0.003 0.005 0.33 0.60 .90.02 0.09 .50.09 0.15 0.10 .50.05 0.05 0.10 .70.09 0.57 .10.39 0.51 .60.91 1.26 consumed for the flavonoid and placebo 1.94 0.15 1.63 0.17 groups, respectively (determined by wrapper returns). Adherence in the flavo- fi 2 2 2 2 2 noid group was con rmed by the sus- 2 0.11 0.33 0.10 0.75 0.04 0.16 tained increases in total epicatechin and – fl 2mnh ots1 ots0 months 12 months 0 months 12 isoflavone levels in 24-h urine at 6 and 12 diabetes 2 type with patients postmenopausal medicated in risk CVD of biomarkers on avones 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 months (total epicatechin 148.5 6 12.3 6 .5 2.20 0.05* .17.67 0.31 .1 8.79 0.41* .51.69 0.05 .37.25 0.13 .54.29 0.05 .21.34 0.02 .5 3.37 0.05* .56.79 0.25 .76.52 0.17 .8137.38 1.00 1.28 0.03 .9*3.16 0.19** .2*0.68 0.02** 0.336 0.003* and 154.1 6 11.1 mmol/day; specifically 10.21 0.37* 3-methyl-epicatechin 6.4 6 0.6 and 7.0 6 0.6 mmol/day, 4-methyl-epicatechin 1.7 6 0.2 and 1.5 6 0.2 mmol/day, epicatechin-3-sulfate 52.7 6 4.8 and 6 m 6 6 6 6 6 6 6 6 6 6 6 6 6 6 58.6 5.1 mol/day, epicatechin-methyl- 6 6 .02.24 0.10 .73.60 0.37 .98.17 0.29 .39.40 0.83 .21.81 0.12 .57.44 0.15 .14.35 0.11 .51.35 0.05 .43.39 0.14 .50.66 0.05 0.332 0.005 .511.33 0.65 .67.90 0.56 .17.45 0.61 sulfate 75.5 6 7.1 and 73.1 6 5.6 mmol/ 138.91 1.13 1.85 0.10 day, daidzein 77.5 6 4.9 and 75.2 6 5.3 mmol/day, respectively, in hydrolyzed samples at 6 and 12 months). Thirty six percent of the flavonoid group were equol produc- ers, determined according to standard 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 .81.53 1.68 .31.13 0.73 .90.05 0.09 .30.45 0.43 .40.50 0.34 .30.61 0.93 .50.12 0.15 .70.19 0.17 .10.07 0.11 .60.01 0.06 .20.01 0.12 0.04 0.005 .51.12 0.65 .00.94 0.70 methods (25). 0.16 0.12 The intervention attenuated insulin resistance (20.3 6 0.2 HOMA-IR flavo- 6 P noid group, 0.4 0.2 placebo group; = 2 2 0.004

0.004) and improved insulin sensitivity, 0.03 – effects that were driven by a significant 0 months 12 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 decrease in insulin levels (Table 2). We 6 0.25** .004 0.59 0.41 1.50 0.07* .0*007 0.04 0.0076 0.003* 0.19 0.52* 0.11 0.11 0.09 .300 0.20 0.04 0.03 0.08* .2*00 0.003 0.09 0.02** 0.29* 0.26 0.11 also observed significant changes in lipo- 0.03 protein status: total-C:HDL-C ratio (flavonoid 20.2 6 0.1, placebo 0.0 6 0.6; P = 0.01), LDL-C levels (flavonoid 20.1 6 0.1 mmol/L, placebo 0.1 6 0.1 effect intervention – 2 2 2 2 2 2 2 2 2 2 2 P 2 mmol/L; = 0.04), and HDL-C:LDL-C 0 months 12 .60.04 0.16 .80.004 0.78 .00.09 0.60 .60.02 1.36 .00.29 0.10 .00.46 0.10 .10.25 0.11 .80.01 0.18 .40.02 1.04 .30.05 0.73 .20.33 0.02 .10.69 0.01 ratio (flavonoid 0.06 6 0.0, placebo Net 20.03 6 0.0; P = 0.003) (Table 2). The intervention also had a significant impact on estimated percent CHD risk, with risk attenuated in the intervention group rela- 6 tive to placebo for CHD (0.1 0.3 vs. – 1.1 6 0.3%, respectively; P = 0.02) months 12 and a trend toward a reduction in fatal P CHD (P = 0.05) (Table 2). The intervention had no effect on BP or glycemic care.diabetesjournals.org DIABETES CARE, VOLUME 35, FEBRUARY 2012 229 Type 2 diabetes and chronic flavonoid intake

an increase in LDL receptor activity (28). Likewise, isoflavones have been shown to decrease apolipoprotein B secretion in HepG2 cells through multiple mechanisms and may augment the efficacy of statins (29). We also observed a significant reduction in insulin resistance (HOMA-IR) and improvements in insulin sensitivity (QUICKI) that were driven by a significant decrease in insulin concentrations. These findings are clinically important, as insulin resistance is not only a key de- terminant of the metabolic syndrome but is also associated with increased arterial stiffening (30) and risk of cardiovascular events even in individuals with no diabetes (31). Our recent meta-analysis of short-term cocoa/chocolate trials (#4 months duration) also showed that intervention improved insulin resistance (HOMA-IR 20.67 [95% CI 20.98, 20.36]) due to significant reductions in serum insulin, although only 3 of the 42 studies were in patients with type 2 dia- Figure 2dThe effect of the 1-year flavonoid intervention on insulin resistance in compliant betes (A. Cassidy, personal communication). 6 , fl patients. A: Plasma insulin. B: HOMA-IR (mean SEM). *P 0.05, ** P =0.01;n =42 a- However another recent meta-analysis vonoid, n = 42 placebo. observed no effect of soy isoflavones on measures of glycemic control (32); al- control (HbA1c and glucose levels) (Table All patients were receiving statin ther- though only 3 of the 24 trials were in 2). At 12 months, mean BMI was similar in apy ($40 mg simvastatin or $10 mg type 2 diabetic patients, there was wide both groups. atorvastatin) as part of standard medical variability in source and dose of soy iso- In sensitivity analyses, a similar mag- care, but flavonoid intervention resulted flavones fed, and few studies reported nitude of effect of the flavonoid interven- in further improvements in lipoprotein changes in HOMA-IR. In another review, tion on HOMA-IR and insulin (Fig. 2) and status, specifically significant reductions only soy isoflavones and genistein ex- estimated 10-year risk of CHD (flavonoid in LDL-C and total-C:HDL-C ratio. The erted beneficial effects on HOMA-IR (14). 0.0 6 0.4%, placebo 1.2 6 0.4%; P = observed reduction of 0.16 mmol/L There are supporting underlying mecha- 0.02) was observed. Moreover, stronger LDL-C is consistent with data from the nisms for these effects; in vitro, flavan- statistical associations were observed for meta-analysis of short term (maximum 3-ols, isoflavones, and their metabolites the effects on estimated 10-year risk of 12 weeks) cocoa (20.15 mmol/L) (26) enhanced insulin-stimulated glucose up- fatal CHD (flavonoid 0.3 6 0.3%, placebo and soy isoflavone (20.13 mmol/L) (14) take in adipocytes through increased ex- 1.1 6 0.3%; P = 0.04) and lipoprotein sta- intervention trials in healthy participants pression of GLUT4, and improved insulin tus: total C:HDL-C ratio (flavonoid 20.2 6 and with the magnitude of change ob- secretory function of pancreatic b-cells 0.1, placebo 0.0 6 0.1; P = 0.004) and served in the few short-term trials on (10,33). LDL-C (flavonoid 20.1 6 0.1 mmol/L, soy isoflavones (16,17) or chocolate (15) Our combined intervention attenu- placebo 0.1 6 0.1 mmol/L; P = 0.03). in type 2 diabetic patients. Our findings ated the UKPDS-estimated, 10-year CHD do not suggest any additive effect of our risk as a result of beneficial effects on CONCLUSIONSdIn this 1-year, ran- combined intervention on lipoprotein lipoprotein status. The UKPDS algorithm domized, placebo-controlled intervention status in this medicated group. incorporates additional risk factors for trial, a combined intake of flavan-3-ols and The observed effect is of potential diabetes (i.e., HbA1c and duration of di- isoflavones significantly improved lipopro- clinical significance. A 1.0 mmol/L reduc- abetes) beyond traditional risk factors, tein status and markers of insulin sensitiv- tion in LDL-C has been associated with and, although validation studies have ity and attenuated the estimated 10-year a 21% decrease in vascular events in pa- identified a tendency to overestimate risk of CHD in postmenopausal women tients with type 2 diabetes (27), suggest- CHD risk (34), it is still considered a useful receiving standard therapy for type 2 di- ing that long-term flavonoid intervention tool (35) to compliment other CHD risk abetes. To our knowledge, this is the only may be associated with a 3.4% reduction assessments. Over half of our study pop- long-term flavonoid trial that has been in vascular events. These effects are also ulation was on antihypertensive medica- conducted in medicated postmenopausal supported by in vitro work, suggesting tion, and this may have influenced our patients with type 2 diabetes, and these that flavan-3-ols increase apolipoprotein lack of effect on BP. In the few previous data highlight the additional benefitofdi- A1 and decrease apolipoprotein B produc- short-term studies with type 2 diabetic etary flavonoids to standard drug therapy tion as a result of upregulation of sterol patients, flavan-3-ols alone did not im- in managing CVD risk in these patients. regulatory element binding proteins and prove BP (15,36), although in one study,

230 DIABETES CARE, VOLUME 35, FEBRUARY 2012 care.diabetesjournals.org Curtis and Associates isoflavones reduced BP in patients with or isoflavones is as effective when con- postmenopausal women. Am J Clin Nutr type 2 diabetes who were not on medica- sumed independently. 2007;85:895–909 tion for hypertension (37). This is in con- 6. Buijsse B, Feskens EJ, Kok FJ, Kromhout fl D. Cocoa intake, blood pressure, and trast to avonoid trials in participants d without type 2 diabetes, where systolic Acknowledgments This study was funded cardiovascular mortality: the Zutphen El- derly Study. Arch Intern Med 2006;166: BP and diastolic BP were reduced by short- by Diabetes UK (06/0003397, principal inves- tigator A.C.), with raw materials purchased 411–417 term flavan-3-ol (by 4.5 and 2.5 mmHg, fl fl from Frutarom Netherlands (iso avones) and 7. Hooper L, Kroon PA, Rimm EB, et al. Fla- respectively) (38) and iso avone intake Barry Callebaut Belgium (Acticoa cocoa) and vonoids, flavonoid-rich foods, and cardio- (by 2.5 and 1.5 mmHg, respectively) (39). trial chocolates manufactured by Barry Callebaut. vascular risk: a meta-analysis of randomized Our study has a number of limitations. P.A.K. and A.C. have received research controlled trials. Am J Clin Nutr 2008;88: The dropout rate during the intervention funding from Unilever Research to conduct an 38–50 was high (21%), with the most frequent anthocyanin trial and in vitro experimental 8. Wang-Polagruto JF, Villablanca AC, Polagruto reason for participant withdrawal (in both work (BBSRC-CASE PhD studentship). P.A.K. JA, et al. Chronic consumption of flavanol- groups) attributable to palatability (Fig. has been an independent member of the rich cocoa improves endothelial function 1A); however, these numbers are compa- Coressence Science Board since 2006 and is in and decreases vascular cell adhesion molecule in hypercholesterolemic post- rable to other nutrition trials in type 2 di- receipt of research funding from Danisco A/S. A.C. has received unrestricted funding from menopausal women. J Cardiovasc Pharmacol abetic patients (40), and as observed in the chocolate industry to conduct a systematic 2006;47(Suppl. 2):S177–S186; discussion other studies, a small number of patients review on chocolate and oxidative stress. No S206–S209 had changes in BP medication during the other potential conflicts of interest relevant to 9. Schewe T, Steffen Y, Sies H. How do di- 1-year trial period (n = 10). Our findings this article were reported. etary flavanols improve vascular function? also relate specifically to a combined fla- P.J.C. provided input for the study con- A position paper. Arch Biochem Biophys vonoid intervention given to postmeno- ception and design, coordinated the trial, 2008;476:102–106 pausal women receiving standard diabetes conducted laboratory and statistical analyses, 10. Ueda M, Nishiumi S, Nagayasu H, Fukuda therapy (including lipid-lowering medica- interpreted data, and drafted the manuscript. I, Yoshida K, Ashida H. Epigallocatechin tion), and further studies are now required M.S., J.P., and K.D. provided input for the gallate promotes GLUT4 translocation in to determine the relative influence of each study conception and design and read and ap- skeletal muscle. Biochem Biophys Res proved the final draft of the manuscript. P.A.K. Commun 2008;377:286–290 flavonoid subclass (flavan-3-ols and iso- fl provided input for the study conception and 11. Odegaard AO, Koh WP, Butler LM, et al. avones) on biomarkers of CVD risk in design, analyzed levels of flavonoids in foods Dietary patterns and incident type 2 di- this population and to examine whether and biological samples, and read and approved abetes in Chinese men and women: the similar effects are observed in male pa- the final draft of the manuscript. A.C. conceived Singapore Chinese Health Study. Diabetes tients or in other patient groups. Finally, and designed the study, interpreted data, draf- Care 2011;34:880–885 although the flavonoid intake in our trial ted the manuscript, and is the guarantor of this 12. Taku K, Umegaki K, Sato Y, Taki Y, Endoh K, could potentially be attained through article. Watanabe S. Soy isoflavones lower serum high intake of dietary flavonoid, an The authors thank Barry Callebaut Belgium total and LDL cholesterol in humans: a meta- fl and Frutarom Netherlands for their assistance analysis of 11 randomized controlled trials. emerging range of avonoid-rich func- – tional foods may offer additional oppor- in developing the trial foods and the Local Am J Clin Nutr 2007;85:1148 1156 fl Diabetes Research Network for invaluable 13. Li SH, Liu XX, Bai YY, et al. Effect of oral tunities to consume avonoids at optimal nursing support. isoflavone supplementation on vascular levels. Despite these limitations, our trial endothelial function in postmenopausal is the longest intervention to date to as- women: a meta-analysis of randomized sess the effects of flavonoids on CVD risk References placebo-controlled trials. Am J Clin Nutr factors in a population of medicated, 1. International Diabetes Federation. IDF 2010;91:480–486 postmenopausal patients with type 2 di- Diabetes Atlas, Fourth Edition.Brussels, 14. Ricci E, Cipriani S, Chiaffarino F, Malvezzi abetes and benefits from objective assess- Belgium, International Diabetes Federa- M, Parazzini F. Effects of soy isoflavones ment of compliance. tion, 2009 and genistein on glucose metabolism in This pragmatic 1-year trial provides 2. Stettler C, Allemann S, Jüni P, et al. Gly- perimenopausal and postmenopausal non- evidence to suggest that the intake of cemic control and macrovascular disease Asian women: a meta-analysis of random- fl in types 1 and 2 diabetes mellitus: meta- ized controlled trials. Menopause 2010;17: avonoids results in sustained improve- – fi analysis of randomized trials. Am Heart J 1080 1086 ments in lipid pro le and insulin sensi- 2006;152:27–38 15. Balzer J, Rassaf T, Heiss C, et al. Sustained tivity and an attenuation of estimated 3. Sarwar N, Gao P, Seshasai SR, et al.; benefits in vascular function through CHD risk, highlighting the additional Emerging Risk Factors Collaboration. flavanol-containing cocoa in medicated di- benefitofflavonoids to standard drug Diabetes mellitus, fasting blood glucose abetic patients a double-masked, random- therapy in managing CVD risk in patients concentration, and risk of vascular dis- ized, controlled trial. J Am Coll Cardiol with type 2 diabetes. Our study is the only ease: a collaborative meta-analysis of 102 2008;51:2141–2149 combined flavonoid trial and the longest prospective studies. Lancet 2010;375: 16. Pipe EA, Gobert CP, Capes SE, Darlington flavan-3-ol intervention to date, and our 2215–2222 GA, Lampe JW, Duncan AM. Soy protein fi 4. Kautzky-Willer A, Kamyar MR, Gerhat D, reduces serum LDL cholesterol and the ndings have potential clinical relevance. fi Long-term studies are now required to et al. Sex-speci c differences in metabolic LDL cholesterol:HDL cholesterol and control, cardiovascular risk, and inter- apolipoprotein B:apolipoprotein A-I ratios determine whether these effects are re- ventions in patients with type 2 diabetes in adults with type 2 diabetes. J Nutr 2009; stricted to populations of medicated mellitus. Gend Med 2010;7:571–583 139:1700–1706 postmenopausal women with estab- 5. Mink PJ, Scrafford CG, Barraj LM, et al. 17. Jayagopal V, Albertazzi P, Kilpatrick ES, lished type 2 diabetes and to determine Flavonoid intake and cardiovascular dis- et al. Beneficial effects of soy phytoes- whether chronic intake of flavan-3-ols ease mortality: a prospective study in trogen intake in postmenopausal women care.diabetesjournals.org DIABETES CARE, VOLUME 35, FEBRUARY 2012 231 Type 2 diabetes and chronic flavonoid intake

with type 2 diabetes. Diabetes Care 2002; 25. Setchell KD, Cole SJ. Method of defining pancreatic beta-cells. Eur J Pharmacol 25:1709–1714 equol-producer status and its frequency 2009;616:321–327 18. Schroeter H, Heiss C, Balzer J, et al. among vegetarians. J Nutr 2006;136: 34. van Dieren S, Peelen LM, Nöthlings U, (-)-Epicatechin mediates beneficial effects 2188–2193 et al. External validation of the UK Pro- of flavanol-rich cocoa on vascular func- 26. Jia L, Liu X, Bai YY, et al. Short-term effect spective Diabetes Study (UKPDS) risk tion in humans. Proc Natl Acad Sci USA of cocoa product consumption on lipid engine in patients with type 2 diabetes. 2006;103:1024–1029 profile: a meta-analysis of randomized Diabetologia 2011;54:264–270 19. Teede HJ, McGrath BP, DeSilva L, Cehun M, controlled trials. Am J Clin Nutr 2010;92: 35. van der Heijden AA, Ortegon MM, FassoulakisA,NestelPJ.Isoflavones reduce 218–225 Niessen LW, Nijpels G, Dekker JM. Pre- arterial stiffness: a placebo-controlled study 27. Kearney PM, Blackwell L, Collins R, et al.; diction of coronary heart disease risk in in men and postmenopausal women. Ar- Cholesterol Treatment Trialists’ (CTT) Col- a general, pre-diabetic, and diabetic pop- terioscler Thromb Vasc Biol 2003;23: laborators. Efficacy of cholesterol-lowering ulation during 10 years of follow-up: ac- 1066–1071 therapy in 18,686 people with diabetes in curacy of the Framingham, SCORE, and 20. Cooper KA, Campos-Giménez E, Jiménez 14 randomised trials of statins: a meta- UKPDS risk functions: the Hoorn Study. Alvarez D, Nagy K, Donovan JL, Williamson analysis. Lancet 2008;371:117–125 Diabetes Care 2009;32:2094–2098 G. Rapid reversed phase ultra-performance 28. Yasuda A, Natsume M, Osakabe N, 36. Mellor DD, Sathyapalan T, Kilpatrick ES, liquid chromatography analysis of the major Kawahata K, Koga J. Cacao polyphenols Beckett S, Atkin SL. High-cocoa polyphenol- cocoa polyphenols and inter-relationships influence the regulation of apolipoprotein rich chocolate improves HDL cholesterol of their concentrations in chocolate. J Agric in HepG2 and Caco2 cells. J Agric Food in type 2 diabetes patients. Diabet Med Food Chem 2007;55:2841–2847 Chem 2011;59:1470–1476 2010;27:1318–1321 21. Williamson G, Manach C. Bioavailabi- 29. Duncan RE, El-Sohemy A, Archer MC. 37. Howes JB, Tran D, Brillante D, Howes LG. lity and bioefficacy of polyphenols in Regulation of HMG-CoA reductase in Effects of dietary supplementation with humans. II. Review of 93 intervention MCF-7 cells by genistein, EPA, and DHA, isoflavones from red clover on ambu- studies. Am J Clin Nutr 2005;81(Suppl.): alone and in combination with mevastatin. latory blood pressure and endothelial 243S–255S Cancer Lett 2005;224:221–228 function in postmenopausal type 2 di- 22. Friedewald WT, Levy RI, Fredrickson DS. 30. Webb DR, Khunti K, Silverman R, et al. abetes. Diabetes Obes Metab 2003;5: Estimation of the concentration of low- Impact of metabolic indices on central 325–332 density lipoprotein cholesterol in plasma, artery stiffness: independent association 38. Desch S, Schmidt J, Kobler D, et al. Effect without use of the preparative ultracen- of insulin resistance and glucose with of cocoa products on blood pressure: trifuge. Clin Chem 1972;18:499–502 aortic pulse wave velocity. Diabetologia systematic review and meta-analysis. Am J 23. Matthews DR, Hosker JP, Rudenski AS, 2010;53:1190–1198 Hypertens 2010;23:97–103 Naylor BA, Treacher DF, Turner RC. Ho- 31. Nakamura K, Sakurai M, Miura K, et al. 39. Liu XX, Li SH, Chen JZ, Sun K, Wang XJ, meostasis model assessment: insulin resis- Homeostasis model assessment of insulin Wang XG, Hui RT. Effect of soy isoflavones tanceand beta-cell function from fasting resistance and the risk of cardiovascular on blood pressure: a meta-analysis of ran- plasma glucose and insulin concen- events in middle-aged non-diabetic Japanese domized controlled trials. Nutr Metab trations in man. Diabetologia 1985;28: men. Diabetologia 2010;53:1894–1902 Cardiovasc Dis. 8 February 2011 [Epub 412–419 32. Liu ZM, Chen YM, Ho SC. Effects of soy ahead of print] 24. Katz A, Nambi SS, Mather K, et al. Quan- intake on glycemic control: a meta-analysis 40. Jenkins DJ, Kendall CW, McKeown- titative insulin sensitivity check index: of randomized controlled trials. Am J Clin Eyssen G, et al. Effect of a low-glycemic a simple, accurate method for assessing in- Nutr 2011;93:1092–1101 index or a high-cereal fiber diet on type 2 sulin sensitivity in humans. J Clin Endo- 33. Fu Z, Liu D. Long-term exposure to genistein diabetes: a randomized trial. JAMA 2008; crinol Metab 2000;85:2402–2410 improves insulin secretory function of 300:2742–2753

232 DIABETES CARE, VOLUME 35, FEBRUARY 2012 care.diabetesjournals.org