Clinical Care/Education/Nutrition ORIGINAL ARTICLE

Beneficial Effects of Soy Intake in Postmenopausal Women With Type 2 Diabetes

1 1,4 VIJAY JAYAGOPAL, MRCP PAUL E. JENNINGS, FRCP very similar degrees of hyperglycemia, the 1 1 PAULA ALBERTAZZI, MRCOG DAVID A. HEPBURN, FRCP Japanese-Americans with type 2 diabetes 2 1 ERIC S. KILPATRICK, MRCPATH STEPHEN L. ATKIN, FRCP 3 showed significantly higher levels of ELAINE M. HOWARTH, MSC plasma insulin after a 75-g oral glucose tolerance test (OGTT) than Japanese with diabetes (6,8), and BMI correlated with insulin levels only for the Japanese- OBJECTIVE — Phytoestrogen consumption has been shown to reduce risk factors for car- American men (8). This observation sug- diovascular disease. Type 2 diabetes confers an adverse cardiovascular risk profile particularly in gested a greater degree of insulin women after menopause. The aim of this study was to determine whether a resistance among the Japanese-Americans with soy protein and isoflavones affected insulin resistance, glycemic control, and cardiovascular and that factors other than BMI were re- risk markers in postmenopausal women with type 2 diabetes. sponsible for the difference in plasma in- sulin levels between the two groups (9). RESEARCH DESIGN AND METHODS — A total of 32 postmenopausal women with Soy is a staple in the diet of the Japanese diet-controlled type 2 diabetes completed a randomized, double blind, cross-over trial of dietary supplementation with (soy protein 30 g/day, isoflavones 132 mg/day) versus population, and consumption of soy has placebo (cellulose 30 g/day) for 12 weeks, separated by a 2-week washout period. been shown to have an inverse relation- ship with mortality from CVD. These le- RESULTS — Compliance with the dietary supplementation was Ͼ90% for both treatment gumes contain complex carbohydrates, phases. When compared with the mean percentage change from baseline seen after 12 weeks of vegetable protein, soluble fibers, oligosac- placebo, phytoestrogen supplementation demonstrated significantly lower mean values for fast- charides, minerals, and phytoestrogens, ing insulin (mean Ϯ SD 8.09 Ϯ 21.9%, P ϭ 0.006), insulin resistance (6.47 Ϯ 27.7%, P ϭ Ϯ ϭ Ϯ ϭ particularly the isoflavones and 0.003), HbA1c (0.64 3.19%, P 0.048), total (4.07 8.13%, P 0.004), LDL cholesterol (7.09 Ϯ 12.7%, P ϭ 0.001), cholesterol/HDL cholesterol ratio (3.89 Ϯ 11.7%, P ϭ , that may be beneficial in the 0.015), and free thyroxine (2.50 Ϯ 8.47%, P ϭ 0.004). No significant change occurred in HDL management of diabetes (10). In vitro cholesterol, triglycerides, weight, blood pressure, creatinine, sulfate, studies have shown isoflavones to have androstenedione, and the hypothalamic-pituitary-ovarian axis hormones. antidiabetic properties such as inhibiting intestinal brush border uptake of glucose, CONCLUSIONS — These results show that dietary supplementation with soy phytoestro- having ␣-glucosidase inhibitor actions, gens favorably alters insulin resistance, glycemic control, and serum lipoproteins in postmeno- pausal women with type 2 diabetes, thereby improving their cardiovascular risk profile. and also demonstrating tyrosine kinase inhibitory properties (11–13). Diets con- Diabetes Care 25:1709–1714, 2002 taining soy protein rich in isoflavones have been shown to improve insulin re- sistance in ovariectomized cynomolgus ardiovascular diseases (CVDs), es- menopausal estrogen depletion (4) and monkeys (14) and to reduce insulin levels pecially coronary heart disease and increased insulin resistance (5) may con- in healthy postmenopausal women C cerebrovascular disease, are the tribute to the high risk of accelerated CVD (15,16). Previous studies on the effects of leading causes of death in women (1). in women with type 2 diabetes. using a soy-based diet in individuals with Type 2 diabetes increases the risk of death Epidemiological data suggest that in diabetes (17–20) have been performed on from CVD by two- to fourfold (2), and Japanese-Americans in Seattle, WA, the a heterogenous population using differing women with diabetes are four times more prevalence of type 2 diabetes is four times soy preparations, and there are few data likely to die from CVD than men (3). Post- that in Japanese in Tokyo (6,7). Despite focusing specifically on the effect of soy ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● phytoestrogens on postmenopausal From the 1Department of Medicine, University of Hull, Hull, U.K.; the 2Department of Clinical Biochemistry women with type 2 diabetes. This double- and Immunology, Hull Royal Infirmary, Hull, U.K.; the 3Department of Applied Statistics, University of Hull, blind, cross-over, placebo-controlled 4 Hull, U.K.; and the Department of Medicine, York District General Hospital, York, U.K. study was undertaken to determine the Address correspondence and reprint requests to Dr. Vijay Jayagopal, Michael White Centre for Diabetes and Endocrinology, Brocklehurst Building, Hull Royal Infirmary, 220-236, Anlaby Road, Hull, HU3 2RW, effect of dietary supplementation with U.K. E-mail: [email protected]. 30 g of soy proteins containing 132 mg of Received for publication 18 March 2002 and accepted in revised form 1 July 2002. phytoestrogen on indexes of glycemic Abbreviations: CVD, cardiovasular disease; HOMA-IR, insulin resistance measured by Homeostasis control, insulin resistance, and cardiovas- Model Assessment; OGTT, oral glucose tolerance test. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion cular risk in postmenopausal women with factors for many substances. diet-controlled type 2 diabetes.

DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002 1709 Soy phytoestrogen intake in postmenopausal women with type 2 diabetes

RESEARCH DESIGN AND during the placebo phase. The blood Synchron LX20 analyzer (Beckman- METHODS pressure readings for these women were Coulter, High Wycombe, U.K.). LDL cho- excluded from the analysis. lesterol was measured using the Subjects Friedewald equation. Plasma glucose was A total of 40 postmenopausal women Study design measured using a Synchron LX20 ana- with type 2 diabetes were screened for A randomized, double-blind, placebo- lyzer (Beckman-Coulter), and serum in- inclusion in the study. Women were con- controlled cross over study was under- sulin was assayed using a competitive sidered postmenopausal if menstrual pe- taken. A total of 17 women were initially chemiluminescent immunoassay per- riods had been absent for Ͼ1 year and assigned to placebo and 16 women were formed using the DPC Immulite 2000 an- follicle-stimulating hormone level was el- assigned to soy for 12 weeks. After a alyzer (Euro/DPC, Llanberis, U.K.). The evated. Women with a fasting venous 2-week washout period, the participants coefficient of variation of this method was plasma glucose concentration Ͼ7.0 received the alternative treatment for an 8%, calculated using duplicate study sam- mmol/l or a 2-h postprandial concentra- additional 12 weeks. The soy preparation ples. The analytical sensitivity was 2 ␮U/ tion Ͼ11.1 mmol/l after a 75-g OGTT (Essential Nutrition, Brough, U.K.) con- ml, and there was no stated cross- were recruited for the study. Exclusion tained 30 g of isolated soy protein with reactivity with proinsulin. The insulin criteria included any secondary cause of 132 mg of isoflavones. The proportions of resistance was calculated using the Ho- hyperglycemia, current or previous (in genistein, daidzein, and aver- meostasis Model Assessment method the preceding 6 months) use of estrogen aged 53, 37, and 10%, respectively, with (HOMA-IR ϭ (Insulin ϫ glucose)/22.5) therapy, treatment with insulin or oral hy- 95% of the isoflavones present as their (21). Serum luteinizing hormone, follicle- poglycemic agents, untreated hypothy- glucoside conjugates. The product was stimulating hormone, testosterone, estra- roidism, history of drug or abuse, completely devoid of soluble fiber and diol, triiodothyronine, free thyroxine, and history of breast or uterine cancer. each sachet provided 243 kcal. The con- and thyroid-stimulating hormone were A total of 33 subjects were eligible for trol supplement consisted of an identical measured on an Architect analyzer (Ab- inclusion in the study: age (mean Ϯ SD) sachet containing 30 g of pure microcrys- bott Laboratories, Maidenhead, U.K.). 62.5 Ϯ 6.77 years, BMI (kg/m2) 32.2 Ϯ talline cellulose (Emcocel; Penwest Phar- Sex hormone–binding globulin was mea- 5.0, time from diagnosis of type 2 diabetes maceuticals, Patterson, NY) of no sured using the DPC Immulite 2000 Ϯ 2.6 2.7 years. All subjects received ad- significant calorific content. Patients were analyzer (Euro/DPC) and HbA1c was vice on a diabetes diet from a registered instructed to maintain an isocaloric diet. measured on a Diabetes Control and dietitian before randomization. At the The randomization was performed by the Complications Trial (DCCT)-aligned HA- randomization visit, all subjects were ad- supplying company with a random num- 8140 analyzer (A. Menarini Diagnostics, vised to maintain their diabetes diet and ber generating table. A copy of the ran- Berkshire, U.K.) using the manufacturer’s level of physical activity throughout the domization code was kept secure by the recommended protocol. Routine bio- study and were also instructed by inter- hospital pharmacy and was available to chemistry and liver function tests were view and through an information leaflet the investigator only in emergency. All performed at each visit using standard to minimize ingestion of soy and soy- subjects gave their written informed con- methods. based products for the entire duration of sent and the protocol was approved by the study. This advice was reinforced at all the Hull and East Riding Local Research Statistical analysis and sample size subsequent visits. One subject was with- Ethics Committee. calculation drawn from the study after 5 months fol- Results from subjects who completed lowing the diagnosis of a myocardial Study measurements both treatment phases were included in infarction, and these data were excluded After an overnight fast, weight and blood the analysis. Mean percentage changes because there was less than the minimum pressure were measured and blood sam- obtained at the end of the soy treatment of two paired data for analysis. None of ples were collected at screening, baseline, phase were compared with the results at the patients had complications of diabe- and then every 6 weeks for each phase of the end of the placebo phase, using the tes. One subject was started on 10 mg/day the trial. Compliance checks and dietary paired Student’s t test for biochemical of simvastatin during the placebo arm of reinforcement were performed at each data and the Wilcoxon’s signed-rank test the trial. Her data were omitted from the visit after randomization. Compliance for clinical observations. The Wilcoxon’s lipid analysis. In three subjects, antihy- was monitored based on counting re- signed-rank test was applied to biochem- pertensive medication was altered during turned medication. Fasting venous blood ical data that violated the assumptions of the course of the study. In one subject, samples were collected into serum gel, normality when tested using the Kolmog- bendroflumethiazide 2.5 mg/day was EDTA, and fluoride oxalate tubes. Sam- orov-Smirnov test. The period and the added to her medication regimen during ples were separated by centrifugation at carryover effect that may have occurred the phytoestrogen treatment phase. In an- 2000g for 15 min at 4°C, and the aliquots from the cross-over design were tested us- other subject, amlodipine 10 mg/day was stored at –20°C within1hofcollection. ing the appropriate Student’s t test. changed to felodipine 10 mg/day during Sitting blood pressure was measured after Using two-sided 5% significance the placebo phase. In the third subject, 10 min of rest using an automated device level, a sample of 33 patients per group the dose of atenolol was increased from (NPB-3900; Nellcor Puritan Bennett, was needed, assuming a 10% drop-out 50 to 100 mg/day during the phytoestro- Pleasanton, CA). Total cholesterol, tri- rate. This gave the study 80% power to gen treatment phase and doxazosin 2 mg/ glyceride, and HDL cholesterol levels detect a 12% difference in LDL between day changed to diltiazem 180 mg/day were measured enzymatically using a treatments (assuming a common SD of

1710 DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002 Jayagopal and Associates

16) and an 18% difference in fasting in- phases (Table 1). There were no changes sulin between treatments, given a com- in routine biochemistry, including renal mon standard deviation of 25 (22). function (as measured by serum creati- The results were considered signifi- nine), and liver function tests. Serum free cant if the two-tailed P value was Ͻ0.05. thyroxine decreased by 2.5% during the nQuery version 4 was used for sample soy phase compared with baseline, and size determination. Statistical analysis this difference was significant when com- was performed using SPSS for Windows pared with placebo (P ϭ 0.004). However, NT version 9.0 (SPSS, Chicago, IL) soy did not affect either serum thyroid- stimulating hormone or triiodothyronine. RESULTS — A total of 32 postmeno- No effects were seen for estradiol, follicle- pausal women with diet-controlled dia- stimulating hormone, luteinizing hor- betes completed the entire cross-over mone, sex hormone– binding globulin, study. No period effect or carryover ef- testosterone, dehydroepiandrosterone fects were identified for any of the ob- sulfate, and androstenedione (Table 2). served results. Both the soy preparation and placebo were well tolerated, and the CONCLUSIONS — The phytoestro- overall compliance was 93% during the gen preparation used in this study re- soy phase and 91% during the placebo duced insulin resistance, improved phase. Side effects were similar during glycemic control, and reduced total cho- both treatment phases and were predom- lesterol primarily through a reduction in inantly gastrointestinal. Heartburn devel- LDL cholesterol, leading to an improve- oped in three subjects during the study, ment in the ratio of total cholesterol to two in the soy phase and one while on HDL cholesterol. These data show that placebo. Six subjects complained of feel- soy supplementation has a therapeutic ef- ing bloated on both soy and placebo. One fect for both glycemic control and for car- subject had self-limiting mouth ulcers diovascular risk markers, at least in the during the placebo phase and one subject short term, for postmenopausal women had a myocardial infarction during the with type 2 diabetes. Based on these data, soy phase and was withdrawn from the high soy intake may explain the epidemi- study. ological observation of lower insulin lev- Figure 1—A schematic representation of the els in Japanese compared with Japanese- Effects on glycemic control changes seen in the indexes of glycemic control Americans (6–9). The therapeutic over both placebo and phytoestrogen during the Serum insulin, HOMA-IR, and HbA1c de- two treatment phases. Group A represents the potential of soy for diabetes was first sug- creased significantly during the soy phase study subjects who received placebo first fol- gested in 1910 (23). The few small studies of the trial and returned to baseline dur- lowed by phytoestrogens, and group B repre- of the effect of soy on glycemic control in ing the placebo phase. This was the same sents the study subjects who received diabetes have shown inconsistent results for both group A, which received placebo phytoestrogen first followed by placebo. The that have been primarily attributed to the for the first 12 weeks and then soy for an periods between visits 3 and 4 represent the soluble fiber content of soybean prepara- additional 12 weeks, and group B, which washout period. tion (18–20). In a small study of 14 received soy first and then placebo (Fig. women and 6 men, 6 weeks of treatment 1). Compared with baseline at the start of with soy protein (50 g/day), isoflavones the treatment phase, 12 weeks of soy sup- vs. 5.78 Ϯ 1.02 mmol/l) and was signifi- (165 mg/day), and cotyledon fiber (20 plementation reduced fasting insulin by cantly lower compared with placebo (P ϭ g/day) compared with placebo (casein 50 8.09% (14.5 Ϯ 6.58 vs. 16.7 Ϯ 8.13 ␮IU/ 0.004). LDL cholesterol decreased by g/day and cellulose 20 g/day) showed an ml), HOMA-IR by 6.47% (4.94 Ϯ 2.94 vs. 7.09% and the total cholesterol–to–HDL improvement in lipid parameters but no Ϯ 5.54 3.40 units), and HbA1c by 0.64% cholesterol ratio decreased by 3.89% after difference in glucose, insulin, or HbA1c (6.78 Ϯ 0.61 vs. 6.83 Ϯ 0.64%). Com- the soy phase compared with baseline (17). A modest improvement in blood pared with the change seen during the (3.35 Ϯ 0.85 vs. 3.63 Ϯ 0.91 and 4.43 Ϯ glucose attributed to fiber intake from placebo phase, insulin, HOMA-IR, and 1.26 vs. 4.64 Ϯ 1.37 mmol/l, respective- soybeans has been reported previously in HbA1c were significantly lower with soy ly). These changes were also significantly individuals with type 2 diabetes, in both (P ϭ 0.006, 0.003, and 0.048, respective- lower with soy than with placebo treat- acute and some longer studies (18,19), ly). The mean fasting venous glucose did ment (P ϭ 0.001 and 0.015, respective- although this was not found by others not change with either soy or placebo ly). No changes were seen in either HDL (20). In vitro studies suggested several treatment (Table 1). or triglyceride levels during and between mechanisms for a direct pharmacological the two treatment phases (Table 1). action of soy on glycemic control, includ- Effects on lipids ing a tyrosine kinase inhibitory action, Compared with baseline at the start of the Other effects changes in insulin receptor numbers and treatment phase, total cholesterol de- There was no difference in either weight affinity, intracellular phosphorylation, creased with soy by 4.07% (5.52 Ϯ 0.92 or blood pressure between treatment and alterations in glucose transport

DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002 1711 Soy phytoestrogen intake in postmenopausal women with type 2 diabetes

Table 1—Subject characteristics at the start of and after 12 weeks of treatment

Soy Placebo Treatment Baseline 12 weeks Baseline 12 weeks difference (n ϭ 32) (n ϭ 32) % Change (n ϭ 32) (n ϭ 32) % Change (P value) Weight (kg) 81.2 Ϯ 13.1 81.7 Ϯ 12.7 ϩ0.77 Ϯ 1.81 81.6 Ϯ 13.1 81.7 Ϯ 13.1 ϩ0.19 Ϯ 1.91 0.369 Fasting glucose (mmol/l) 7.29 Ϯ 1.49 7.37 Ϯ 1.63 ϩ1.14 Ϯ 10.8 7.23 Ϯ 1.37 7.57 Ϯ 1.93 ϩ4.31 Ϯ 12.7 0.340 Ϯ Ϯ Ϫ Ϯ Ϯ Ϯ ϩ Ϯ HbA1c (%) 6.83 0.64 6.78 0.61 0.64 3.19 6.82 0.66 6.88 0.59 1.08 3.90 0.048 Fasting insulin (␮lU/ml) 16.7 Ϯ 8.13 14.5 Ϯ 6.58 Ϫ8.09 Ϯ 21.9 15.5 Ϯ 7.26 17.3 Ϯ 10.7 ϩ9.92 Ϯ 29.1 0.006 HOMA-IR 5.54 Ϯ 3.40 4.94 Ϯ 2.94 Ϫ6.47 Ϯ 27.7 5.14 Ϯ 3.04 6.09 Ϯ 4.82 ϩ14.7 Ϯ 32.1 0.003 Cholesterol (mmol/l), n ϭ 31* 5.78 Ϯ 1.02 5.52 Ϯ 0.92 Ϫ4.07 Ϯ 8.13 5.62 Ϯ 1.05 5.75 Ϯ 1.07 ϩ2.83 Ϯ 8.70 0.004 LDL cholesterol (mmol/l), n ϭ 31 3.63 Ϯ 0.91 3.35 Ϯ 0.85 Ϫ7.09 Ϯ 12.7 3.47 Ϯ 1.06 3.62 Ϯ 1.07 ϩ5.35 Ϯ 15.2 0.001 HDL cholesterol (mmol/l) n ϭ 31* 1.32 Ϯ 0.33 1.32 Ϯ 0.34 ϩ0.69 Ϯ 10.1 1.30 Ϯ 0.30 1.31 Ϯ 0.30 ϩ1.29 Ϯ 8.78 0.787 Cholesterol/HDL ratio, n ϭ 31* 4.64 Ϯ 1.37 4.43 Ϯ 1.26 Ϫ3.89 Ϯ 11.7 4.54 Ϯ 1.35 4.61 Ϯ 1.35 ϩ2.23 Ϯ 12.1 0.015 Triglycerides (mmol/l), n ϭ 31* 2.20 Ϯ 1.21 2.01 Ϯ 1.03 Ϫ1.58 Ϯ 31.2 2.18 Ϯ 1.62 2.19 Ϯ 1.51 ϩ2.81 Ϯ 24.9 0.624 Systolic blood pressure, n ϭ 29† 146.9 Ϯ 18.8 144.7 Ϯ 14.6 Ϫ0.76 Ϯ 9.63 147.2 Ϯ 15.2 149.5 Ϯ 17.2 ϩ1.65 Ϯ 7.46 0.325 Diastolic blood pressure, n ϭ 29† 81.5 Ϯ 8.78 80.7 Ϯ 8.78 Ϫ0.41 Ϯ 10.4 82.6 Ϯ 7.8 82.2 Ϯ 8.73 Ϫ0.23 Ϯ 8.83 0.991 Data are means Ϯ SD. All serum results are obtained from fasting variables. % change, percent difference compared with baseline. P value was calculated using Wilcoxon’s signed-rank test. *One subject was removed from the analysis of the lipid effects because they were started on a statin drug during the trial period; †three subjects were removed from the analysis of the effects on blood pressure because their medications were changed during the trial period.

(11,13,24,25). In this study, the improve- glycemic agents (26). This action of soy women was reported after soy treatment ment in insulin concentration and insulin isoflavones in reducing insulin resistance (27), and a primate study showed that soy resistance noted after the active treatment without a change in weight seems, there- improved vascular reactivity in female but phase was not only significant when com- fore, to be independent of the action of not in male monkeys (28). Whether a sex pared with the change seen after the pla- soy fiber that enhances satiation and thus effect is evident for glycemic control and cebo phase but also when compared with would cause an improvement in insulin insulin resistance is unclear. However, an baseline (P ϭ 0.005 and 0.032, respec- resistance secondary to weight loss (10). improvement in insulin sensitivity and tively). The 8% decrease in fasting insulin It seems that the washout period between glucose effectiveness occurred in ovariec- and the corresponding decrease in insulin the two treatment phases was adequate, tomized healthy female cynomolgus resistance without a change in weight because the insulin levels and insulin re- monkeys given a soy-based diet but not in seems to support a direct pharmacologi- sistance in the group receiving the active male diabetic monkeys in whom only the cal effect of soy in these patients, although medication first returned to baseline be- lipid parameters were improved (14,29). information on the effect of soy on body fore the start of the placebo phase. Determination of response to soy by fat distribution is required to be more cer- Some of the metabolic effects of soy menopausal status is suggested by studies tain of this. Nevertheless, the changes seem to be specific to women and influ- showing that soy protein (containing 132 seen here are comparable to the reported enced by menopausal status. A worsening mg of isoflavones per day) decreased se- insulin-lowering effect of some oral hypo- of endothelial function in men but not in rum insulin significantly in healthy post-

Table 2—Effects of soy supplementation versus placebo on plasma hormone concentrations and creatinine

Soy Placebo Treatment Baseline 12 weeks Baseline 12 weeks difference (n ϭ 32) (n ϭ 32) % Change (n ϭ 32) (n ϭ 32) % Change (P value) LH (IU/l) 28.4 Ϯ 14.8 25.9 Ϯ 12.6 Ϫ4.28 Ϯ 20.6 28.8 Ϯ 14.5 28.3 Ϯ 13.8 ϩ0.30 Ϯ 21.8 0.866 FSH (IU/l) 48.7 Ϯ 18.7 44.6 Ϯ 16.1 Ϫ5.23 Ϯ 16.8 49.2 Ϯ 18.6 47.9 Ϯ 18.3 Ϫ2.29 Ϯ 13.1 0.443 Catradiol (pmol/l) 103.7 Ϯ 51.2 137.4 Ϯ 60.4 ϩ42.7 Ϯ 55.6 117.6 Ϯ 58.1 140.0 Ϯ 78.4 ϩ29.2 Ϯ 76.8 0.079 Testosterone (nmol/L), n ϭ 31* 2.16 Ϯ 0.85 1.77 Ϯ 0.77 Ϫ15.8 Ϯ 32.8 2.19 Ϯ 0.84 1.90 Ϯ 0.97 Ϫ11.4 Ϯ 35.9 0.970 SHBG (nmol/l) 39.3 Ϯ 16.4 36.4 Ϯ 14.7 Ϫ4.98 Ϯ 16.9 39.1 Ϯ 17.1 36.0 Ϯ 16.0 Ϫ6.88 Ϯ 14.9 0.985 Androstenedione (nmol/l) 4.25 Ϯ 2.25 3.97 Ϯ 1.60 ϩ3.65 Ϯ 36.3 4.73 Ϯ 3.59 4.08 Ϯ 1.88 ϩ8.31 Ϯ 52.3 0.722 DHEAS (␮mol/l) 2.09 Ϯ 1.17 2.01 Ϯ 1.03 Ϫ0.35 Ϯ 21.3 2.02 Ϯ 1.04 2.07 Ϯ 1.28 ϩ0.40 Ϯ 22.5 0.889 Creatinine (␮mol/l) 77.6 Ϯ 15.8 75.3 Ϯ 15.9 Ϫ2.53 Ϯ 9.45 76.2 Ϯ 16.1 78.2 Ϯ 18.6 ϩ2.95 Ϯ 11.7 0.072 TSH (mU/l) 2.15 Ϯ 1.08 2.27 Ϯ 1.35 ϩ9.32 Ϯ 36.4 2.18 Ϯ 1.08 2.26 Ϯ 1.18 ϩ9.48 Ϯ 46.1 0.739 Free thyroxine (pmol/l) 14.1 Ϯ 1.53 13.7 Ϯ 1.65 Ϫ2.50 Ϯ 8.47 13.8 Ϯ 1.75 14.4 Ϯ 1.87 ϩ4.26 Ϯ 9.22 0.004 Free triiodothyronine (pmol/l) 3.18 Ϯ 0.46 3.23 Ϯ 0.50 ϩ2.55 Ϯ 15.3 3.26 Ϯ 0.63 3.29 Ϯ 0.58 ϩ3.69 Ϯ 25.7 0.970 Data are means Ϯ SD. P value was calculated using Wilcoxon’s signed-rank test. *One patient was removed because the testosterone result was an extreme outlier (Ͼ3 SD from the mean of the group). % Change, percent difference compared with baseline; LH, luteinizing hormone, FSH, follicle-stimulating hormone; SHBG, sex hormone binding globulin; DHEAS, dehydroepiandrosterone sulfate; TSH, thyroid-stimulating hormone.

1712 DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002 Jayagopal and Associates menopausal women (15) but not in studies (14) and in healthy postmeno- 482, 2001 healthy premenopausal women (30). To pausal women (15,16). It remains unclear 6. Fujimoto WY, Leonetti DL, Kinyoun JL, circumvent the possible confounding ef- which of these two components of soy are Newell-Morris L, Shuman WP, Stolov fects of sex and menopausal status, the primarily responsible for the benefits seen WC, Wahl PW: Prevalence of diabetes current study was designed to include an or indeed whether both need to act syn- mellitus and impaired glucose tolerance among second-generation Japanese-Ameri- all-female postmenopausal population. ergistically for optimum benefit. The can men. Diabetes 36:721–729, 1987 Antihypertensive medication is known to quantity of isoflavones needed to achieve 7. Fujimoto WY, Leonetti DL, Bergstrom affect carbohydrate metabolism. In three a therapeutic benefit also remains un- RW, Kinyoun JL, Stolov WC, Wahl PW: women, blood pressure medication was clear. The dose of isoflavones (132 mg/ Glucose intolerance and diabetic compli- changed during the study period. How- day) used in this study was based on the cations among Japanese-American women. ever, excluding these subjects from the dose previously shown to reduce insulin Diabetes Res Clin Pract 13:119–129, 1991 analysis of the indexes of carbohydrate levels in healthy postmenopausal women. 8. Fujimoto WY, Akanuma Y, Kanazawa Y, metabolism did not significantly alter the This dose is higher than the isoflavone Mashiko S, Leonetti D, Wahl P: Plasma result. content in diets commonly consumed in insulin levels in Japanese and Japanese- The most consistently reported bene- some Asian countries (20–80 mg/day), American men with type 2 diabetes may be related to the occurrence of cardiovas- ficial effect of soy has been on lipids, and where soy is a staple (32). Currently, cular disease. Diabetes Res Clin Pract 6: a meta-analysis of 38 controlled human foods such as bread enriched with soy 121–127, 1989 clinical trials using on average 47 g of soy isoflavones are available and can be used 9. Fujimoto WY: The importance of insulin protein daily showed significant reduc- to supplement other natural sources. resistance in the pathogenesis of type 2 tions in total cholesterol (9%), LDL cho- However, before definite dietary recom- diabetes mellitus. Am J Med 108 (Suppl. lesterol (13%), and triglycerides (11%) mendations can be suggested, further in- 6a):9S–14S, 2000 (22). In the present study, we observed formation is needed from studies 10. Anderson JW, Smith BM, Washnock CS: significant improvements in total choles- assessing the effects of lower doses of Cardiovascular and renal benefits of dry terol, LDL cholesterol, and cholesterol-to- isoflavones on carbohydrate metabolism. bean and soybean intake. Am J Clin Nutr HDL ratio but no change in triglyceride In summary, these data show that 70:464S–474S, 1999 11. Vedavanam K, Srijayanta S, O’Reilly J, Ra- levels. In accord with others (22), HDL short-term dietary phytoestrogen supple- man A, Wiseman H: Antioxidant action cholesterol was not affected by soy treat- mentation reduces insulin resistance and and potential antidiabetic properties of an ment. The mechanism of the lipid- improves glycemic control in postmeno- isoflavonoid-containing soyabean phyto- lowering effect of soy is unclear. It has pausal women with type 2 diabetes, while chemical extract (SPE). Phytother Res 13: been suggested that soy may induce a hy- also reducing their cardiovascular risk by 601–608, 1999 perthyroid state (31), but our data did not lowering LDL cholesterol. Studies of 12. Lee DS, Lee SH: Genistein, a soy isofla- support this because the mean free thy- longer duration are needed to determine vone, is a potent alpha-glucosidase inhib- roxine levels decreased significantly with whether these effects are sustained and itor. FEBS Lett 501:84–86, 2001 soy and serum thyroid-stimulating hor- have a beneficial effect on reducing car- 13. Sorenson RL, Brelje TC, Roth C: Effect of mone and free triiodothyronine levels diovascular events. tyrosine kinase inhibitors on islets of Langerhans: evidence for tyrosine kinases were unchanged. Obesity is a known im- in the regulation of insulin secretion. En- portant variable in carbohydrate and lipid docrinology 134:1975–1978, 1994 metabolism, and it is important to deter- References 14. Wagner JD, Cefalu WT, Anthony MS, Lit- mine whether the effect of phytoestrogens 1. Mosca L, Manson JE, Sutherland SE, wak KN, Zhang L, Clarkson TB: Dietary is influenced by the degree of obesity. In Langer RD, Manolio T, Barrett-Connor E: soy protein and estrogen replacement this study, however, only two subjects Cardiovascular disease in women: a state- therapy improve cardiovascular risk fac- had BMI Ͻ25 kg/m2; therefore, no mean- ment for healthcare professionals from tors and decrease aortic cholesteryl ester ingful comparison could be made be- the American Heart Association Writing content in ovariectomized cynomolgus tween the lean and obese groups. Group. Circulation 96:2468–2482, 1997 monkeys. Metabolism 46:698–705, 1997 However, exclusion of these two lean sub- 2. Eastman RC, Keen H: The impact of car- 15. Duncan AM, Underhill KE, Xu X, Laval- diovascular disease on people with diabe- leur J, Phipps WR, Kurzer MS: Modest jects from the analysis did not signifi- tes: the potential for prevention. Lancet hormonal effects of soy isoflavones in cantly alter the results. 350 (Suppl. 1):SI29–SI32, 1997 postmenopausal women. J Clin Endocrinol In this study, the soy preparation 3. Greenland P, Reicher-Reiss H, Goldbourt Metab 84:3479–3484, 1999 used was a combination of soy protein U, Behar S: In-hospital and 1-year mortal- 16. Goodman-Gruen D, Kritz-Silverstein D: and isoflavones. In previous in vitro stud- ity in 1,524 women after myocardial in- Usual dietary isoflavone intake is associ- ies assessing the beneficial effect of soy in farction: comparison with 4,315 men. ated with cardiovascular disease risk fac- diabetes (11–13), isoflavones have been Circulation 83:484–491, 1991 tors in postmenopausal women. J Nutr shown to act as ␣-glucosidase inhibitors, 4. Sowers JR: Diabetes mellitus and cardio- 131:1202–1206, 2001 to inhibit intestinal brush border uptake vascular disease in women. Arch Intern 17. Hermansen K, Sondergaard M, Hoie L, of glucose, and to have tyrosine kinase Med 158:617–621, 1998 Carstensen M, Brock B: Beneficial effects 5. Stoney RM, O’Dea K, Herbert KE, Dragi- of a soy-based dietary supplement on inhibitory properties. Furthermore, the cevic G, Giles GG, Cumpston GN, Best lipid levels and cardiovascular risk mark- benefit of soy protein enriched with JD: Insulin resistance as a major determi- ers in type 2 diabetic subjects. Diabetes isoflavones (‘whole’ soy protein) on in- nant of increased coronary heart disease Care 24:228–233, 2001 dexes of carbohydrate metabolism has risk in postmenopausal women with type 18. Tsai AC, Vinik AI, Lasichak A, Lo GS: Ef- also been previously proven in animal 2 diabetes mellitus. Diabet Med 18:476– fects of soy polysaccharide on postpran-

DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002 1713 Soy phytoestrogen intake in postmenopausal women with type 2 diabetes

dial plasma glucose, insulin, glucagon, soy bean as a food in diabetes. Am J Med 3053–3060, 2001 pancreatic polypeptide, somatostatin, and Sci 140:793–803, 1910 28. Honore EK, Williams JK, Anthony MS, triglyceride in obese diabetic patients. 24. Jonas JC, Plant TD, Gilon P, Detimary P, Clarkson TB: Soy isoflavones enhance Am J Clin Nutr 45:596–601, 1987 Nenquin M, Henquin JC: Multiple effects coronary vascular reactivity in atheroscle- 19. Mahalko JR, Sandstead HH, Johnson LK, and stimulation of insulin secretion by the rotic female macaques. Fertil Steril 67: Inman LF, Milne DB, Warner RC, Haunz tyrosine kinase inhibitor genistein in nor- 148–154, 1997 EA: Effect of consuming fiber from corn mal mouse islets. Br J Pharmacol 114:872– 29. Wagner JD, Zhang L, Greaves KA, bran, soy hulls, or apple powder on glu- 880, 1995 Shadoan MK, Schwenke DC: Soy protein cose tolerance and plasma lipids in type II 25. Abler A, Smith JA, Randazzo PA, Rothen- reduces the arterial low-density lipopro- diabetes. Am J Clin Nutr 39:25–34, 1984 berg PL, Jarett L: Genistein differentially tein (LDL) concentration and delivery 20. Thomas BL, Laine DC, Goetz FC: Glucose inhibits postreceptor effects of insulin in of LDL cholesterol to the arteries of di- and insulin response in diabetic subjects: rat adipocytes without inhibiting the in- abetic and nondiabetic male cynomol- acute effect of carbohydrate level and the sulin receptor kinase. J Biol Chem 267: gus monkeys. Metabolism 49:1188–1196, addition of soy polysaccharide in defined- 3946–3951, 1992 2000 formula diets. Am J Clin Nutr 48:1048– 26. Kipnes MS, Krosnick A, Rendell MS, Egan 30. Duncan AM, Merz BE, Xu X, Nagel TC, 1052, 1988 JW, Mathisen AL, Schneider RL: Pioglita- 21. Matthews DR, Hosker JP, Rudenski AS, zone hydrochloride in combination with Phipps WR, Kurzer MS: Soy isoflavones Naylor BA, Treacher DF, Turner RC: Ho- sulfonylurea therapy improves glycemic exert modest hormonal effects in pre- meostasis model assessment: insulin control in patients with type 2 diabetes menopausal women. J Clin Endocrinol resistance and beta-cell function from mellitus: a randomized, placebo-con- Metab 84:192–197, 1999 fasting plasma glucose and insulin con- trolled study. Am J Med 111:10–17, 31. Forsythe WA 3rd: Soy protein, thyroid centrations in man. Diabetologia 28:412– 2001 regulation and cholesterol metabolism. 419, 1985 27. Teede HJ, Dalais FS, Kotsopoulos D, Li- J Nutr 125:619S–623S, 1995 22. Anderson JW, Johnstone BM, Cook- ang YL, Davis S, McGrath BP: Dietary soy 32. Barnes S, Peterson TG, Coward L: Ratio- Newell ME: Meta-analysis of the effects of has both beneficial and potentially ad- nale for the use of genistein-containing soy protein intake on serum lipids. N Engl verse cardiovascular effects: a placebo- soy matrices in chemoprevention trials for J Med 333:276–282, 1995 controlled study in men and postmeno- breast and prostate cancer. J Cell Biochem 23. Friedenwald J, Ruhrah J: The use of the pausal women. J Clin Endocrinol Metab 86: Suppl 22:181–187, 1995

1714 DIABETES CARE, VOLUME 25, NUMBER 10, OCTOBER 2002