Type 2 diabetes is associated with severe complications including retinopathy, nephropathy, neuropathy, stroke and CVD, and has become a worldwide concern for public health. The number of patients with diabetes worldwide has been estimated to increase from 171 million in 2000 to 366 million by the year 2030( Reference Wild, Roglic and Green 1 ). An increased incidence rate of type 2 diabetes has also been reported in both adults and children during the past several decades( Reference Pinhas-Hamiel and Zeitler 2 , Reference Neville, Boye and Montgomery 3 ).
Diabetes is characterized by a hyperglycaemic condition due to insulin hyposecretion or insulin resistance at target organs, including adipose tissue, liver and muscle. Insulin resistance has been aetiologically linked to a pro-inflammatory state( Reference Molavi, Rassouli and Bagwe 4 ). Inflammatory pathways in insulin resistance can be initiated by extracellular mediators such as cytokines and NEFA, or by intracellular stresses such as the excessive production of reactive oxygen species( Reference Furukawa, Fujita and Shimabukuro 5 , Reference Xu, Barnes and Yang 6 ).
Soyabean is a member of the legume family and is a part of the traditional Asian diet. Soyabean and soya products are rich sources of various nutrients such as plant protein, fibre, vitamins, minerals and phyto-oestrogens (isoflavones)( Reference Ren, Kuhn and Wegner 7 ). Isoflavones may exert beneficial effects on glucose homeostasis because they structurally resemble oestradiol. Previous experimental studies have reported that soyabean phytochemical extracts inhibit glucose uptake in the intestine( Reference Vedavanam, Srijayanta and O’Reilly 8 ) and isoflavones have been shown to directly stimulate insulin production in pancreatic islet β cells through the cyclic adenosine monophosphate pathway( Reference Liu, Zhen and Yang 9 ). Furthermore, the administration of soya protein and isoflavones led to increased insulin production, improved glucose metabolism( Reference Lu, Wang and Song 10 ) and lower insulin requirements( Reference Ascencio, Torres and Isoard-Acosta 11 ) in animals.
An inverse association between soya food intake and type 2 diabetes has been reported in several, but not all human studies. A diet with soyanuts( Reference Azadbakht, Kimiagar and Mehrabi 12 ) and supplements with soya protein, fibre, isoflavone and other soya products improved glucose homeostasis in patients with type 2 diabetes( Reference Li, Hong and Saltsman 13 , Reference Jayagopal, Albertazzi and Kilpatrick 14 ). An inverse association between soya food intake and the prevalence of glycosuria was also reported in postmenopausal non-obese Chinese women( Reference Yang, Shu and Jin 15 ). Furthermore, a recent prospective study in Japan showed that higher intake of soya products and isoflavones was associated with a decreased odds ratio of diabetes among overweight or postmenopausal women( Reference Nanri, Mizoue and Takahashi 16 ). The consumption of some soya foods was inversely associated with the risk of type 2 diabetes among Asians in two prospective studies( Reference Villegas, Gao and Yang 17 , Reference Mueller, Odegaard and Gross 18 ). On the other hand, soya intake was positively associated with the risk of diabetes in another cohort study in Hawaii( Reference Morimoto, Steinbrecher and Kolonel 19 ).
To date, no studies have examined the association between soya intake and insulin resistance in Japan. Therefore, we investigated the association between soya product intake and insulin resistance levels in the Japanese population.
Methods
Study population
The present study population included participants in the Japan Multi-Institutional Collaborative Cohort (J-MICC) Study, which intends to examine associations between lifestyle and genetic factors and their interactions with lifestyle-related diseases. Details of the survey method have been reported elsewhere( Reference Hamajima 20 ). The study population consisted of two groups. The first group (group 1) comprised 577 men and women who had attended the Tokushima Prefectural General Health Check-up Center between 23 January 2008 and 24 November 2011 and agreed to participate in the J-MICC Study (response rate=14·8 %). The second group (group 2) consisted of 697 men and women living in Tokushima city. We distributed approximately 98 700 leaflets explaining the objective and the method of the J-MICC Study all over Tokushima city (total population=264 500) from July 2012 to February 2013. These 697 individuals read the leaflet and attended the health check-ups performed by our research team between 25 July 2012 and 27 February 2013. From the total of 1274 men and women (groups 1+2) aged 34–70 years, we excluded individuals with a previous history of stroke (n 14), IHD (n 29), medical history of or medical treatment for diabetes (n 68), those whose data on the health check-up (n 10) or soya products (n 2) were not available and those whose daily total energy intake was low (<4189 kJ/d) or high (>16 756 kJ/d; n 10)( Reference Hishida, Morita and Naito 21 ). The remaining 1148 individuals (565 men and 583 women) were used for analysis in the present study (Fig. 1). Participation was essentially voluntary and after explaining the details of the study, we obtained written informed consent from each participant. The study protocol was reviewed and approved by the Ethics Committees of the authors’ institutions.
An overview of the analysed individuals. Among 1274 men and women aged 34–70 years at baseline, living in Tokushima Prefecture between 2008 and 2013, who participated, 1148 eligible individuals (565 men and 583 women) were finally analysed
Questionnaires and calculation of the intake of soya protein and soya isoflavone
Study participants were requested to complete a structured self-administered questionnaire, including questions on physical activity during leisure time, frequency of intake of foods and beverages, smoking and drinking habits, and current and previous history of diseases.
Participants answered a validated, short FFQ on their dietary habits, which asked how often they had consumed forty-seven items of food/recipes and beverages over the past year( Reference Tokudome, Goto and Imaeda 22 – Reference Imaeda, Goto and Tokudome 25 ). The questionnaire included four items on soya product consumption: miso soup (=1); tofu (=2); fermented soyabeans and soyabeans (=3); and fried tofu mixed with vegetables, fried bean curd and thick deep-fried tofu (hereafter ‘deep-fried tofu’; =4). The frequency of soya product intake was classified into the following eight categories: ‘3 times/day’ (21/week), ‘twice/day’ (14/week), ‘once/day’ (7/week), ‘5–6 times/week’ (5·6/week), ‘3–4 times/week’ (3·5/week), ‘1–2 times/week’ (1·4/week), ‘1–3 times/month’ (0·7/week) and ‘almost never’ (0/week). The total frequency of non-fried soya product intake was calculated as the sum of miso soup, tofu and boiled or fermented soyabeans (1+2+3). The total frequency of soya product intake was calculated as the sum of miso soup, tofu, boiled or fermented soyabeans and deep-fried tofu (1+2+3+4).
Since the portion size of each soya product was not queried, the intake per meal from four 3 d diet records, which were collected within an interval of 4 months or shorter in a group of twenty-eight participants, was used to calculate the weekly intake of soya products. The portion sizes of miso soup (1), tofu (2), boiled or fermented soyabeans (3) and deep-fried tofu (4) were 9·9 (the amount used as miso), 53·8, 31·7 and 19·5 g/meal, respectively. Total soya protein intake was estimated by summing the soya protein contained in each specific soya food on the basis of the Standard Tables of Food Composition in Japan (Ministry of Education, Culture, Sports, Science and Technology, 2010). Total soya isoflavone intake was estimated by summing the soya isoflavone contained in each soya food according to estimates by the Ministry of Agriculture, Forestry and Fisheries of Japan.
The total energy intake of each participant was estimated using a program developed at the Department of Public Health, Nagoya City University School of Medicine( Reference Tokudome, Goto and Imaeda 22 , Reference Tokudome, Goto and Imaeda 23 ). Physical activity during leisure time was evaluated by multiplying the frequency and level of exercise-related metabolism, which consisted of three steps, summed and expressed as MET-h/week (where MET is metabolic equivalent of task). The three steps of exercise-related metabolism level were light (including walking and golf, 3·4 MET), moderate (including jogging and swimming, 7·0 MET) and heavy exercise (including marathon running and fighting sports, 10·0 MET).
Anthropometric and biochemical measurements
Body height, weight, waist circumference, blood pressure, fasting plasma glucose and serum levels of TAG and HDL cholesterol were obtained at the time of health check-ups. Participants were requested not to eat after 20.00 hours, and received a medical check-up from 08.00 to 11.30 hours the following day. Serum insulin was measured using a chemiluminescence immunoassay at BML Inc. (Tokyo, Japan). The homeostasis model assessment of insulin resistance (HOMA-IR) was calculated using the following equation: fasting insulin (mU/l)×fasting plasma glucose (g/dl)/405. Insulin resistance was defined as HOMA-IR≥2·5.
Statistical analysis
Continuous variables were expressed as mean and standard deviation, whereas those with skewed distribution were expressed as median and 25th–75th percentile. The one-way ANOVA, Kruskal–Wallis test or χ 2 test was used to determine significant differences between the characteristics of participants according to total soya products consumption. The associations of soya food consumption with the prevalence of insulin resistance were examined using multiple logistic regression analysis, after adjustment for gender, age (continuous), recruitment (binary), family history of type 2 diabetes mellitus (without a family history, unclear and with a family history) in father and mother, total energy intake (quartiles), physical activity (quartiles), smoking (current, no and ex-smokers) and drinking habits (current drinkers and others) in model 1, and additionally for BMI (continuous) in model 2. All multivariate models were further adjusted for total vegetable intake (quartiles) or total fibre intake (quartiles). Furthermore, in multiple logistic regression analysis, we adjusted for menopause status (men, premenopausal women, perimenopausal women and postmenopausal woman) in place of gender. The consumption of miso soup, tofu, total non-fried soya products and total soya products was divided into quartiles, with the first quartile being defined as the reference. Dummy variables were created for categorical variables and those except for reference categories were included in the models. Odds ratios and 95 % profile likelihood confidence intervals were calculated. We conducted tests for trend using the median value of each quartile for the consumption of each soya food and the likelihood ratio test in logistic models. All statistical tests were based on two-sided probabilities and performed using the SAS statistical software package version 8·2. All P values <0·05 were considered significant( 26 ).
Results
Characteristics of the participants
The total number of participants was 1148 (565 men and 583 women). The mean age was 52·6 (sd 9·9) years in men and 52·0 (sd 9·8) years in women. Table 1 shows the characteristics of the participants according to the frequency of total soya product intake. Age, physical activity, serum levels of HDL cholesterol and total energy intake increased as the consumption of total soya products (1+2+3+4) increased. On the other hand, fasting insulin, HOMA-IR and the proportion of current smokers decreased with increasing total soya product intake. Table 1 also shows the estimated intake of soya protein and isoflavone, presented for two types of soya products (1+2+3 and 1+2+3+4), according to frequency of total soya product intake.
Characteristics of the participants according to intake frequency of total soya products; 565 men and 583 women aged 34–70 years baseline, living in Tokushima Prefecture between 2008 and 2013, Japan Multi-Institutional Collaborative Cohort (J-MICC) Study
Q, quartile; MET, metabolic equivalent of task; HOMA-IR, homeostasis model assessment of insulin resistance.
* Values are presented as n and %
† Values are presented as mean and sd.
‡ Values are presented as median and 25th–75th percentile (P25–P75).
§ Total non-fried soya product intake included miso soup, tofu and boiled or fermented soyabeans.
|| Total soya intake included miso soup, tofu, boiled or fermented soyabeans and deep-fried tofu.
¶ Total soya product intake was estimated by summing the soya protein of each specific soya food on the basis of the Standard Tables of Food Composition in Japan (Ministry of Education, Culture, Sports, Science and Technology, 2010).
** Total soya product intake was estimated by summing the soya isoflavone of each specific soya food according to estimates of the Ministry of Agriculture, Forestry and Fisheries of Japan.
Associations of the intake of each soya product with insulin resistance
Gender-, age- and multivariate-adjusted associations between the consumption of each soya product and insulin resistance are presented in Table 2. After adjustments for age, gender, recruitment, family history of type 2 diabetes mellitus, total energy intake, physical activity, and smoking and drinking habits (model 1), the OR of insulin resistance decreased significantly as the consumption of miso soup, total non-fried soya products (1+2+3) and total soya products (1+2+3+4) increased (P for trend <0·05). On the other hand, the association between the intake of tofu and insulin resistance was rather U-shaped, although the P for trend was <0·05. To assess whether the association between soya product intake and insulin resistance was confounded or mediated by obesity, an analysis was performed with an additional adjustment for BMI (model 2). The relationships of miso soup, total non-fried soya products (1+2+3) and total soya products (1+2+3+4) with insulin resistance remained significant (Table 2, Fig. 2(a)).
Association between the intake of each soya product (1=miso soup; 2=tofu; 3=fermented soybeans and soybeans; 4=fried tofu mixed with vegetables, fried bean curd and thick deep-fried tofu) and insulin resistance among 565 men and 583 women aged 34–70 years baseline, living in Tokushima Prefecture between 2008 and 2013, Japan Multi-Institutional Collaborative Cohort (J-MICC) Study. Consumption was divided into quartiles (
, Q1; 
, Q2; 
, Q3; 
, Q4), with Q1 being defined as the reference. Results are presented as odds ratios with their 95 % profile likelihood confidence intervals represented by vertical bars, and are adjusted for gender, age, recruitment, family history of type 2 diabetes mellitus, total energy intake, physical activity, smoking and drinking habits, and BMI. (a) No significant linear trend was observed for the association between the frequency of intake of tofu and log (OR) of insulin resistance. However, the relationships of intake of miso soup, total non-fried soya products (1+2+3) and total soya products (1+2+3+4) with insulin resistance were significant. (b) The OR of insulin resistance decreased significantly as the estimated intake of soya protein from total non-fried soya products (1+2+3) and total soya products (1+2+3+4) increased (P for trend <0·05). However, the associations between the estimated intake of soya isoflavone from total non-fried soya products (1+2+3) and total soya products (1+2+3+4) and insulin resistance were slightly attenuated. *P for trend <0·05, **P for trend <0·01
Association between the intake frequency of each soya product and insulin resistanceFootnote * among 565 men and 583 women aged 34–70 years baseline, living in Tokushima Prefecture between 2008 and 2013, Japan Multi-Institutional Collaborative Cohort (J-MICC) Study
Q, quartile.
* Adjusted odds ratios and 95 % profile likelihood confidence intervals.
† Miso soup (times/week): Q1, 0–1·4; Q2, 1·4–3·5; Q3, 3·5–7·0; Q4, >7·0.
‡ Age- and gender-adjusted model: adjusted for age (continuous), gender and recruitment (binary).
§ Model 1: adjusted for age (continuous), gender, recruitment (binary), family history of type 2 diabetes mellitus (categorical), total energy intake (categorical), physical activity (categorical), smoking (categorical) and drinking habits (binary).
|| Model 2: adjusted for age (continuous), gender, recruitment (binary), family history of type 2 diabetes mellitus (categorical), total energy intake (categorical), physical activity (categorical), smoking (categorical), drinking habits (binary) and BMI (continuous).
¶ Tofu (times/week): Q1, 0–0·70; Q2, 0·70–1·4; Q3, 1·4–3·5; Q4, >3·5.
** Total unsweetened and non-fried soya products (times/week): Q1, 0–4·2; Q2, 4·2–7·0; Q3, 7·0–9·8; Q4,>9·8.
†† Total soya products (times/week): Q1, 0–4·9; Q2, 4·9–7·7; Q3, 7·7–11·2; Q4, >11·2.
Associations of the estimated intake of soya protein and soya isoflavone with insulin resistance
Table 3 shows the gender-, age-, and multivariate-adjusted associations between the estimated intake of soya protein and soya isoflavone from non-fried soya products (1+2+3) and total soya products (1+2+3+4) with insulin resistance. After adjustments for age, gender, recruitment, family history of type 2 diabetes mellitus, total energy intake, physical activity, and smoking and drinking habits (model 1), the OR of insulin resistance decreased significantly as the estimated intake of soya protein from non-fried soya products (1+2+3) and total soya products (1+2+3+4), and isoflavone from total soya products (1+2+3+4), increased (P for trend <0·05). On the other hand, the OR was lowest in the third quartile and the association between the intake of isoflavone contained in non-fried soya products (1+2+3) and insulin resistance was U-shaped, although the P for trend was <0·05. After an additional adjustment for BMI (model 2), the associations between the intake of soya protein from non-fried soya products (1+2+3) and total soya products (1+2+3+4) with insulin resistance were significant (Fig. 2(b)).
Associations of the estimated intake of soya protein and soya isoflavone with insulin resistanceFootnote * among 565 men and 583 women aged 34–70 years baseline, living in Tokushima Prefecture between 2008 and 2013, Japan Multi-Institutional Collaborative Cohort (J-MICC) Study
* Adjusted odds ratios and 95 % profile likelihood confidence intervals.
† Total unsweetened and non-fried soya products calculated on soya protein (g/week): Q1, 0–9·73448; Q2, 9·73448–15·39853; Q3, 15·39853–25·38757; Q4,>25·38757.
‡ Age- and gender-adjusted model: adjusted for age (continuous), gender and recruitment (binary).
§ Model 1: adjusted for age (continuous), gender, recruitment (binary), family history of type 2 diabetes mellitus (categorical), total energy intake (categorical), physical activity (categorical), smoking (categorical) and drinking habits (binary).
|| Model 2: adjusted for age (continuous), gender, recruitment (binary), family history of type 2 diabetes mellitus (categorical), total energy intake (categorical), physical activity (categorical), smoking (categorical), drinking habits (binary) and BMI (continuous).
¶ Total soya products calculated on soya protein (g/week): Q1, 0–11·93971; Q2, 11·93971–18·06136; Q3, 18·06136–28·57594; Q4, >28·57594.
** Total unsweetened and non-fried soya products calculated on soya isoflavone (mg/week): Q1, 0–39·89579; Q2, 39·89579–62·55006; Q3, 62·55006–99·25788; Q4, >99·25788.
†† Total soya products calculated on soya isoflavone (mg/week): Q1, 0–45·25961; Q2, 45·25961–67·91388; Q3, 67·91388–107·98101; Q4, >107·98101.
To assess whether the associations between the intake of soya products and insulin resistance were confounded by other factors, menopausal status, total vegetable intake or total fibre intake was further adjusted (data not shown). After adjustment for menopausal status, the results were essentially similar. When total vegetable intake was adjusted, the inverse association remained significant for miso soup, total non-fried soya products and total soya products. On the other hand, after adjustment for total fibre intake, the associations were significant for total non-fried soya product and total soya product intake, regardless of adjustment for BMI.
Discussion
In the present study we showed that a higher frequency of consumption of total non-fried soya products and total soya products was associated with reduced OR of insulin resistance among a Japanese population. When soya product intake was calculated as soya protein and soya isoflavone, the associations were slightly attenuated.
The association between the high intake of soya products and decreased OR of insulin resistance observed in the present study was not unexpected. The consumption of soya foods (tofu and other soya products) was previously associated with a significantly reduced prevalence of glycosuria in 39 385 Chinese women aged 40–70 years without diabetes, especially postmenopausal women with BMI<25 kg/m2 ( Reference Yang, Shu and Jin 15 ). Nanri et al. reported a significant correlation between soya products and daidzein intake and a reduced cumulative incidence of type 2 diabetes among women with BMI>25 kg/m2 and postmenopausal women( Reference Nanri, Mizoue and Takahashi 16 ). In a prospective study in China, Villegas et al. reported that the consumption of legumes and soya foods was associated with a decreased risk of type 2 diabetes in 64 227 women aged 40–70 years( Reference Villegas, Gao and Yang 17 ). Mueller et al. also observed that while unsweetened soya product intake was protective against diabetes, no significant correlation between the intake of soya-derived components and insulin resistance was observed in 43 176 Chinese Singaporeans( Reference Mueller, Odegaard and Gross 18 ). On the other hand, higher soya food intake was associated with a slightly elevated diabetes risk in three ethnic groups in a recent cohort study performed in the USA, and this increased risk was limited to overweight and obese subjects( Reference Morimoto, Steinbrecher and Kolonel 19 ). Some randomized controlled trials have reported that soya products( Reference Azadbakht, Kimiagar and Mehrabi 12 ), soya protein( Reference Azadbakht, Kimiagar and Mehrabi 12 – Reference Jayagopal, Albertazzi and Kilpatrick 14 ) and soya isoflavone( Reference Jayagopal, Albertazzi and Kilpatrick 14 ) improve glycaemic control and insulin sensitivity. However, others showed no significant effect of soya protein( Reference Liu, Chen and Ho 27 – Reference Sites, Cooper and Toth 29 ) and soya isoflavone( Reference Liu, Chen and Ho 27 , Reference Nikander, Tiitinen and Laitinen 30 , Reference Hall, Vafeiadou and Hallund 31 ).
The results of an experimental study suggested that soyabean protein has the potential to improve insulin resistance or ameliorate obesity by inhibiting lipogenesis and promoting lipolysis in the liver and adipose cells( Reference Bhathena and Velasquez 32 ). Furthermore, isoflavone is similar to endogenous oestrogen in structure and exhibits a weak oestrogen-like action by combining with the oestrogen receptor in various organs( Reference Bhathena and Velasquez 32 ). By adjusting the function of adipose cells and inhibiting lipoprotein lipase, oestrogen has been shown to adjust the number of adipose cells, adipose deposition and lipid production( Reference Hamosh and Hamosh 33 – Reference Naaz, Yellayi and Zakroczymski 35 ).
Therefore, we assessed the association between the intake of soya protein and isoflavone and insulin resistance. However, the frequency of the consumption of total soya products showed a clearer linear association with insulin resistance than the estimated intake of soya protein or isoflavones. One reason for this result may be that an in vivo metabolism of isoflavones was not considered in the present study. The soya isoflavones, such as daidzein and genistein, are phyto-oestrogens metabolized extensively by the intestinal microflora( Reference Wu, Oka and Ezaki 36 ). Some previous reports have suggested that 30–50 % of Japanese have isoflavone-producing ability depending on the intestinal bacteria( Reference Arai, Uehara and Sato 37 ) and this ability is increased by dietary habits such as higher consumption of fibre, green tea and fish oil( Reference Lampe, Karr and Hutchins 38 ). Therefore, there may be a discrepancy between the estimated intake of soya isoflavone and the actually metabolized amount of soya isoflavone. In addition, according to a recent meta-analysis, the purified or isolated components of soya (isoflavones or soya protein) were not as effective as whole soya foods in improving glycaemic control( Reference Liu, Chen and Ho 39 , Reference Ricci, Cipriani and Chiaffarino 40 ). The reasons for this may include the presence of other components of soya, such as fibre, saponin, polysaccharides, phytosterol and unsaturated fatty acids, or their interactions. Thus, various experimental and human studies have reported the beneficial effects of soya or nutrient components of soya on glucose metabolism. However, the effects and mechanisms of soya and soya foods on insulin resistance have not been fully clarified and warrant further study.
After adjustment for BMI, the OR and P for trend for the association between the intake of soya products, soya protein or soya isoflavone were not greatly altered (model 1 v. model 2 in Tables 2 and 3). These results suggest that the inverse association between soya product intake and insulin resistance was not strongly confounded or intermediated by obesity.
Our study has some limitations. First, the temporal relationship between soya product intake and insulin resistance remains obscure because our study was a cross-sectional design. Second, the sample size of the study was small. Third, information on soya product consumption was self-reported and our FFQ did not include information on portion size. However, we used information on soya product intake per meal, which was calculated from the four 3 d diet records of twenty-eight participants, in order to estimate the intake of soya protein and isoflavone. In consequence, Spearman’s rank-correlation coefficient of the estimated intake of soya protein from total non-fried soya products (1+2+3) and total soya products (1+2+3+4), and the estimated intake of soya isoflavone from total non-fried soya products (1+2+3) and total soya products (1+2+3+4), was 0·55 (P=0·002), 0·47 (P=0·011), 0·64 (P<0·001) and 0·59 (P=0·001), respectively (see online supplementary material, Supplemental Table 1). On the other hand, the intake of soya products, soya protein or isoflavone may have been underestimated. The median estimated soya isoflavone intake from soya foods based on the National Nutrition Survey of 2002 was 16–22 mg/d, which is equivalent to Q4 in the present study. Fourth, some potential confounders which have a great effect on insulin resistance may have not been completely adjusted for. However, the relationship between insulin resistance and soya products was robust after adjustments for menopause status, vegetable intake or total fibre intake. Finally, because all participants in the study were Japanese, our results may not be applicable to other ethnic groups.
Conclusion
In conclusion, our results showed that habitual high intake of soyabeans and soya products may be associated with a lower level of insulin resistance. The frequency of the intake of soya products, rather than the estimated intake of soya protein or isoflavone, showed a clearer relationship with insulin resistance. In the future, further studies are needed to clarify the relationship between insulin resistance and metabolites of soya products, using urine and blood samples.
Acknowledgements
Acknowledgements: The authors thank the following researchers for providing the usable FFQ and a program to calculate nutrient intake: Shinkan Tokudome at the National Institute of Health and Nutrition (formerly Nagoya City University); Chiho Goto at Nagoya Bunri University; Nahomi Imaeda at Nagoya Women's University; Yuko Tokudome at Nagoya University of Arts and Sciences; Masato Ikeda at the University of Occupational and Environmental Health; and Shinzo Maki at Aichi Prefectural Dietetic Association. Financial support: This study was supported in part by Grants-in-Aid for Scientific Research on Priority Areas of Cancer (number 17015018) and on Innovative Areas (number 221S0001) from the Japanese Ministry of Education, Culture, Sports, Science and Technology. The Japanese Ministry of Education, Culture, Sports, Science and Technology had no role in the design, analysis or writing of this article. Conflict of interest: None. Authorship: All authors (M.N., H.U., T.S., S.K.-K., M.Y., M.H. and K.A.) developed the idea for the study. M.N., H.U., S.K.-K., M.Y., M.H. and K.A. collected the data. Measurements and data analysis were completed by M.N. and K.A. H.U. and K.A. provided medical advice regarding interpretation of the data. M.N. drafted the manuscript with the help of K.A. All authors approved the final version of the manuscript. Ethics of human subject participation: The study protocol was reviewed and approved by the Ethics Committees of Nagoya University School of Medicine (affiliated with the former principal investigator, Nobuyuki Hamajima), Aichi Cancer Center (affiliated with the current principal investigator, Hideo Tanaka) and the University of Tokushima Graduate School.
Supplementary material
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S136898001400247X
