lable at ScienceDirect

Nutrition 28 (2012) e29–e35

Contents lists avai

Nutrition

journal homepage: www.nutr i t ionjrnl .com

Basic nutritional investigation

Dietary patterns and C-peptide concentrations in a Japanese workingpopulation

Shamima Akter Ph.D. a,*, Akiko Nanri Ph.D. a, Siyan Yi Ph.D. a, Ngoc Minh Pham Ph.D. a,Kayo Kurotani Ph.D. a, Yasumi Kimura Ph.D. b, Yumi Matsushita Ph.D. a, Tetsuya Mizoue Ph.D. a

aDepartment of Epidemiology and Prevention, International Clinical Research Center, National Center for Global Health and Medicine, Tokyo, JapanbDepartment of Nutrition and Life Science, Fukuyama University, Fukuyama, Japan

a r t i c l e i n f o

Article history:Received 4 August 2011Accepted 25 January 2012

Keywords:Cross-sectional studyC-peptideInsulin resistanceDietary patternJapanese

This study was supported by a Grant-in-Aid for18590601) and (B) 21390213 from the Japan Society fo* Corresponding author. Tel.: þ81-3-3202-7181; fax

E-mail address: samimarub@yahoo.com (S. Akter)

0899-9007/$ - see front matter � 2012 Elsevier Inc. Adoi:10.1016/j.nut.2012.01.018

a b s t r a c t

Objective: It remains unsettled whether dietary patterns play a role in insulin resistance. Weassessed the association of major dietary patterns with C-peptide concentrations in a Japaneseworking population.Methods: A cross-sectional study was conducted in 456 municipal employees (270 men and 186women) 21 to 67 y old who participated in a health survey at the time of their periodic checkup.The dietary patterns were derived by using the principal component analysis of the consumptionof 52 food and beverage items, which was assessed by a validated brief dietary history question-naire. Multiple regression analysis was used to estimate the means of C-peptide concentrationsacross tertiles of each dietary pattern score with the adjustment of potential confounders,including age, body mass index, physical activity, smoking, alcohol drinking, and energy intake.Results: We identified three dietary patterns: healthy, animal food, and Westernized breakfastpatterns. The Westernized breakfast pattern was characterized by high intakes of bread, confec-tionaries, and milk and yogurt but low intakes of rice and alcohol and was inversely associatedwith C-peptide concentrations in women but not in men. The multivariable-adjusted means ofC-peptide concentrations were 1.03 ng/mL (95% confidence interval 0.95–1.12), 0.95 ng/mL (95%confidence interval 0.88–1.03), and 0.89 ng/mL (95% confidence interval 0.82–0.97) for the lowestthrough the highest tertiles of the Westernized breakfast pattern score (P for trend ¼ 0.015) inwomen. Other dietary patterns were not appreciably associated with C-peptide concentrations. Ina subgroup, similar associations were observed between dietary patterns and the homeostasismodel assessment of insulin resistance.Conclusion: The Westernized breakfast pattern may be associated with a lower insulin resistance inJapanese women.

� 2012 Elsevier Inc. All rights reserved.

Introduction

Type 2 diabetes is increasing globally and in the previousdecade has become one of the main health threats [1]. In Japan,there are approximately 8.9 million people with strongly sus-pected diabetes and another 13.2 million potential people withdiabetes [2]. Such an increasemight be ascribed to changes in thesocioeconomic environment that could determine personallifestyles, in particular diet [3,4]. The identification of specific

Scientific Research ((C)r the Promotion of Science.: þ81-3-3202-7364..

ll rights reserved.

dietary factors that prevent type 2 diabetes is important for theformulation of new dietary recommendations. However, nutri-ents and foods are consumed in combination and their combinedeffects may be synergic. The analysis of dietary patterns hasfocused on a combination of several foods and overcome theproblems related to the close correlation among numerous foodsand nutrients [5]. Accumulating evidence has suggested thatsome dietary patterns are associated with a lower risk of type 2diabetes [6,7]. However, the role of dietary patterns in thepathogenesis of type 2 diabetes remains unclear.

Insulin resistance is an important underlying condition oftype 2 diabetes and a predictor of many chronic diseases [8,9].Blood insulin levels and C-peptide concentrations in the fastingstatus correlate with the degree of insulin resistance [10].

S. Akter et al. / Nutrition 28 (2012) e29–e35e30

Because circulating C-peptide is formed from proinsulin inequivalent amounts with insulin and is not extracted by the liver[11], C-peptide is a better surrogate than insulin for estimatinginsulin resistance [10] and thus represents an additionaladvantage in epidemiologic investigations.

Several studies have reported inverse associations ofC-peptide concentrations with specific foods and the intake ofnutrients such as carbohydrate [12], dietary fiber [13], calcium[14], and coffee [15]. To the best of our knowledge, however, onlytwo studies have reported the association between the dietarypattern and serum C-peptide in a healthy US population [16,17].According to these studies, a Western pattern (characterized byhigh intakes of processed meats, eggs, red meats, and high-fatdairy products) was associated with increased serum C-peptideconcentrations. However, none has addressed this issue in Asiansincluding the Japanese who differ from Western populations innot only dietary habits but also C-peptide concentrations owingto a lower insulin secretion [18]. The aim of the present studywas to investigate the relation between dietary patterns andC-peptide levels in a Japanese working population. We alsoexamined the association of the dietary pattern with thehomeostasis model assessment of insulin resistance (HOMA-IR)in a subgroup of the study population.

Materials and methods

Study procedure

This cross-sectional study was conducted in 2006 in municipal employees oftwo municipal offices in northeastern Kyushu, Japan [19]. At the time of theperiodic health examination, all full-time workers (n ¼ 601) except those ona long sick or maternity leave were invited to participate in this study. Partici-pants were asked to fill in the survey questionnaires beforehand, which werechecked by the research staff for completeness and, where necessary, clarified byasking the subjects during the examination. We also obtained data that wereroutinely collected in the health examination, including anthropometricmeasurements, biochemical data, and information about medical history,smoking, and alcohol drinking. Blood and urine specimens were also collected.The study was approved by the ethics committee of the National Center forGlobal Health and Medicine, Japan, and written informed consent was obtainedfrom all subjects before the survey.

Study subjects

Of 601 eligible employees, 547 subjects (323 men and 224 women) 21 to 67 yold participated in the survey (response rate 91%). After the exclusion of 17subjects with a history of cardiovascular disease or cancer and 5 subjects withmissing information on diet, 525 workers (310 men and 215 women) wereincluded for the analysis of dietary patterns [20]. We further excluded 11 subjectswith a history of diabetes, 58 subjects with missing information on C-peptidelevels, and 4 subjects with missing information for any of the covariates used inthe main analysis. Some subjects had overlapping data and met more than oneexclusion criteria. Therefore, 456 workers (270 men and 186 women) remainedfor the analysis of the association between dietary patterns and C-peptideconcentrations.

Dietary assessment

Dietary habits during the preceding month were assessed using a validatedbrief self-administered diet history questionnaire (BDHQ), which consisted of thefollowing five sections: 1) the intake frequency of 46 food and non-alcoholicbeverage items; 2) the daily intake of rice and miso soup; 3) the frequency ofdrinking alcoholic beverages and the amount per drink for five alcoholic bever-ages; 4) the usual cooking methods; and 5) the general dietary behavior. Dietaryintakes for 58 food and beverage itemswere estimated using an ad hoc computeralgorithm for the BDHQ according to the following procedures. For most items(46 items in section 1 and five alcoholic beverages in section 3), the dietary intakewas calculated based on the reported intake or drinking frequency. For rice andmiso soup, the dietary intake was calculated based on the reported number ofbowls of rice and cups of miso soup per day. The dietary intake of three fooditems usually added during cooking, namely salt, oil, and sugar, was estimatedaccording to the dietary history method using the qualitative information insections 4 and 5 together with the food intakes calculated as described earlier.

The intakes of table salt and salt-containing seasoning at the table and soupconsumed with noodles were estimated from answers to the correspondingqualitative questions in section 5. The dietary energy intake was estimated withreference to the standard tables of food composition in Japan [21]. According tothe validation study of the BDHQ using 16-d semi-weighted dietary records asthe gold standard, the Spearman correlation coefficients for 58 food and beverageitems in 92 women 31 to 69 y old and 92 men 32 to 76 y old were 0.14 to 0.82(median 0.44) and 0.22 to 0.83 (median 0.48), respectively [22].

C-peptide measurement

Venus blood (7 mL) was drawn in a tube and then taken in a cooler box toa laboratory, where the blood was centrifuged for 10 min and the serum wasdivided in a maximum of six tubes (0.5 mL each) within 4 h of blood drawing.Five of these tubes were stored at �80�C (four tubes) or �20�C (one tube for themeasurement of fatty acid composition) until analysis. Serum C-peptideconcentrations were measured by a chemiluminescent immunoassay method atan external laboratory (Mitsubishi Chemical Medicine, Tokyo, Japan). The coef-ficients of variation for C-peptide concentrations of 0.7, 3.0, and 13.2 ng/mL were3.3%, 3.8%, and 3.8%, respectively. The lower detection limit for C-peptide was 0.1ng/mL, and there was no upper detection limit.

Other variables

Body height was measured to the nearest 0.1 cm with the subjects standingwithout shoes. Body weight in light clothes was measured to the nearest 0.1 kg.Body mass index (BMI) was calculated as the body weight (kilograms) divided bythe square of the body height (meters). The smoking status, alcohol drinking,marital status, job title, job position, and non-job physical activity were self-reported in the lifestyle questionnaire. The job title was used to create cate-gories for occupational physical activity; sedentary work included managerialand clerical jobs, whereas physically activework included child care, school lunchcooking, and technical jobs. Non-job physical activity was expressed as the dailyminutes spent walking or cycling to or fromwork and the weekly hours engagedin each of five different activities (walking, low-, moderate-, and high-intensityactivities, and gardening). The sum of time spent for all the non-job activitieswere expressed in hours per week.

Statistical analyses

We performed principal component analysis based on the energy-adjustedintakes by using a density method of 52 food and beverage items excluding sixitems (sugar added to coffee and black tea, three items usually added duringcooking [salt, oil, and sugar], table salt and salt-containing seasoning at the table,and soup consumed with noodles) to derive the dietary patterns. The factorswere rotated by orthogonal transformation (varimax rotation) to maintain theuncorrelated factors and greater interpretability. We considered the eigenvalues,the scree test, and the interpretability of the factors to determine the number offactors to retain. Of several factors showing eigenvalues larger than 1, we decidedto retain three factors because the scree plots decreased substantially after thethird factor (from 2.45 to 2.05) and remained similar after the fourth factor (2.05for the fourth and 1.85 for the fifth factor) and because each of these factorsappeared to represent the major dietary characteristics of the Japanese. Thedietary patterns were named according to the food items showing high loading(absolute value) on three factors. The factor scores for each dietary pattern andfor each individual were calculated by summing the intakes of the food itemsweighted by their factor loadings. The factor scores were categorized into tertiles.

The trend association between a potential confounding variable and a dietarypattern score was tested by using the Mantel–Haenszel chi-square test forcategorical variables and linear regression analysis for continuous variables, withordinal numbers (1 to 3) assigned to the tertile categories of each dietary pattern.

We assessed the relation between the serum C-peptide concentrations anddietary patterns (sex-specific tertiles) by using multiple regression analysis.Before this analysis, we log-transformed C-peptide concentrations to betterapproximate the normal distributions. Back-transformed geometric means werecalculated and the results were presented as a geometric mean and its 95%confidence interval (CI). The first model was adjusted for age (year, continuous)and workplace (A or B) only, and the second model was further adjusted formarital status (married or unmarried), job position (low, middle, or high for men;low or middle and high for women because a small number of women was ina higher job position), occupational physical activity (sedentary or active work),current smoking (yes or no), alcohol drinking (0, ethanol<20 g/d, ethanol�20 g/d), non-job physical activity (0,>0 to<2, or�2 h/wk), BMI (kilograms per squaremeter, continuous), and total energy intake (kilocalories per day, continuous).The trend association was assessed by multiple regression analysis, with ordinalnumbers 1 to 3 assigned to the tertile categories of each dietary pattern. Becauseinsulin resistance is particularly pronounced in those who are overweight, werepeated these analyses stratified by BMI, with the sex-specific median BMI used

S. Akter et al. / Nutrition 28 (2012) e29–e35 e31

as the cutoff point (men <23.0 and �23.0 kg/m2, women <20.57 and �20.57 kg/m2). An interaction test was conducted between the dichotomized BMI variableand the tertile categories of each dietary pattern. P for interaction was tested byusing the likelihood ratio test. In 411 participants who had a measurement offasting glucose (Glucose CII Test Wako, Wako Pure Chemical Industries, Osaka,Japan), we repeated the main analysis for the relation of dietary patterns tohomeostasis model assessment of insulin resistance (HOMA2-IR), which wasestimated from the blood glucose and C-peptide concentrations witha computer-based modeling system provided by the Oxford Centre for DiabetesEndocrinology and Metabolism (http://www.dtu.ox.ac.uk/homa). A two-sided P< 0.05 was regarded as statistically significant. All analyses were performedusing STATA 11.0 (STATA Corp., College Station, TX, USA).

Results

We identified three major dietary patterns by principalcomponent analysis (Table 1). The first factor was nameda “healthy Japanese dietary pattern” because it represented highintakes of not only vegetables and fruits but also soy products,

Table 1Factor loading matrix for major dietary patterns identified by principal component an

Healthy Japanese dietary pattern

Carrots/pumpkin 0.78Mushrooms 0.73Green leaves vegetables 0.69Cabbage/Chinese cabbage 0.68Japanese radish/turnip 0.68Other root vegetables 0.67Tofu/atsuagey 0.50Seaweeds 0.48Potatoes 0.46Other fruit 0.37Persimmons/strawberries/kiwifruit 0.31Nattoz 0.30Citrus fruit 0.25Green tea 0.22100% fruit and vegetable juice �0.19Buckwheat noodles �0.25Cola drink/soft drink �0.29Chinese noodles �0.44Squid/octopus/shrimp/shellfish �0.16Canned tuna d

Lean fish d

Pork/beef d

Dried fish/salted fish d

Ham/sausage/bacon d

Chicken d

Oily fish d

Lettuces/cabbage (raw) 0.21Tomatoes 0.19Mayonnaise/dressing d

Small fish with bones 0.20Liver d

Egg d

Black tea/oolong tea d

Other pickles d

Rice d

Bread d

Western-type confectioneries d

Milk and yogurt d

Rice crackers/rice cake/okonomiyakix d

Japanese confectioneries d

Ice cream �0.19Miso soup d

Pickled green leaves vegetables 0.18Sake d

Beer �0.21Shochu �0.24

* A factor loading less than �0.15 is represented by a dash for simplicity. Omittedpatterns (coffee, Japanese wheat noodles, low-fat milk and yogurt, spaghetti and mac

y Deep-fried tofu.z Fermented soybeans.x Meat/fish and vegetable pancake.

mushrooms, and green tea. The second factor was characterizedby high intakes of fish and shellfish, meat, processed meat,mayonnaise, and egg and thus named the “animal food pattern.”The third factor represented high intakes of bread, confection-eries, milk and yogurt, mayonnaise, and egg and low intakes ofrice, alcohol, and fish, and the pattern was named the “West-ernized breakfast pattern.” The first to third dietary patternsaccounted for 9.9%, 5.0%, and 4.7%, respectively, of the variance inthe food intakes and totally explained 19.5% of the variability.

Table 2 presents the characteristics of the study subjectsaccording to the tertile of each dietary pattern score for eachmanand woman. Participants (men and women) with a higher scorefor the healthy Japanese dietary pattern were more likely to bemarried and physically active during leisure time. In women, thescore of this dietary pattern was increased with an increase ofage or BMI and was lower in those in a low job position.Regarding the Westernized breakfast pattern, women with

alysis*

Animal food pattern Westernized breakfast pattern

d d

d d

d d

0.17 d

d d

d d

d �0.160.16 �0.16

d d

d 0.21d d

d d

d d

d d

d 0.170.18 d

d 0.15d d

0.47 �0.210.44 d

0.43 d

0.43 d

0.41 �0.230.37 d

0.37 d

0.33 �0.290.30 0.170.29 d

0.29 0.280.28 �0.230.27 d

0.21 0.210.19 d

0.18 d

�0.67 �0.34d 0.57d 0.56d 0.36d 0.36d 0.36d 0.32d �0.180.18 �0.24

d �0.26d �0.41d �0.45

from the table are food items with a factor loading less than 0.15 for all dietaryaroni, whisky, and wine).

Table 2Characteristics of study participants by tertile of each dietary pattern score in men and women

Healthy Japanese dietary pattern Animal food pattern Westernized breakfast pattern

T1 (low) T2 T3 (high) P for trend* T1 (low) T2 T3 (high) P for trend* T1 (low) T2 T3 (high) P for trend*

MenNumber of subjects 90 90 90 90 90 90 90 90 90Age (y), mean � SD 43.6 � 10.1 43.6 � 10.3 44.4 � 11.7 0.63 44.6 � 10.9 44.7 � 9.9 42.3 � 11.1 0.15 48.1 � 10.2 42.7 � 10.7 40.9 � 9.9 <0.01Workplace A (%) 37.8 32.2 20.2 0.02 17.8 36.7 37.8 <0.01 30.0 20.0 42.2 0.08Married (%) 66.7 74.4 81.1 0.03 74.4 78.9 68.9 0.40 85.6 70.0 66.7 <0.01Sedentary work (%) 86.7 88.9 92.2 0.23 84.4 93.3 90.0 0.23 93.3 88.9 85.6 0.10Job position (low, %) 51.1 48.9 53.3 0.77 50.0 48.9 54.4 0.55 36.7 50.0 66.7 <0.01Non-job physical activity (�2 h/wk, %)y 33.3 38.9 51.1 0.02 43.3 36.7 43.3 1.00 47.8 36.7 38.9 0.23BMI (kg/m2), mean � SD 23.2 � 2.9 23.8 � 3.2 23.5 � 3.8 0.49 23.4 � 3.3 23.8 � 3.5 23.3 � 3.5 0.90 23.9 � 3.2 23.6 � 3.5 23.0 � 3.3 0.052Current smoker (%) 50.0 48.9 35.6 0.051 47.8 45.6 41.1 0.37 55.6 51.1 27.8 <0.01Current drinker (%) 56.7 50.0 62.2 0.45 56.7 52.2 60.0 0.65 86.7 51.1 31.1 <0.01Energy intake (kcal/d) 1732 � 505 1887 � 539 1963 � 555 <0.01 1860 � 519 1838 � 507 1883 � 495 0.77 1867 � 544 1941 � 590 1772 � 472 0.24

WomenNumber of subjects 62 62 62 62 62 62 62 62 62Age (y), mean � SD 35.2 � 9.2 42.8 � 9.9 44.6 � 10.2 <0.01 40.8 � 10.9 41.5 � 10.2 40.4 � 10.6 0.84 44.5 � 10.2 39.4 � 11.1 38.8 � 9.5 <0.01Workplace A (%) 24.2 24.2 16.1 0.28 14.5 16.1 33.9 0.01 14.5 19.4 30.7 0.03Married (%) 46.7 66.1 70.9 0.01 58.1 69.4 56.5 0.85 62.9 64.5 56.5 0.46Sedentary work (%) 79.0 59.7 61.3 0.04 61.3 69.4 69.4 0.34 64.5 62.9 72.6 0.34Job position (low, %) 85.5 56.5 48.4 <0.01 59.7 61.3 69.4 0.26 54.8 58.1 77.4 0.01Non-job physical activity (�2 h/wk, %)y 12.9 25.8 30.7 0.02 22.6 16.1 30.7 0.28 32.3 19.4 17.7 0.02BMI (kg/m2), mean � SD 20.3 � 2.6 20.9 � 2.9 21.9 � 3.6 <0.01 20.9 � 2.8 20.6 � 2.6 21.8 � 3.8 0.11 21.0 � 3.8 21.2 � 2.8 21.1 � 2.7 0.97Current smoker (%) 1.6 3.2 1.6 1.00 0.0 3.2 3.2 0.22 3.2 1.6 1.6 0.54Current drinker (%) 24.2 22.6 25.8 0.92 27.4 24.2 20.9 0.40 37.1 20.9 14.5 <0.01Energy intake (kcal/d) 1484 � 327 1671 � 404 1633 � 430 0.04 1613 � 435 1548 � 313 1626 � 429 0.85 1620 � 379 1612 � 404 1556 � 408 0.36

BMI, body mass index; T1 to T3, tertiles 1 to 3* Based on the Mantel–Haenszel chi-square test for categorical variables and linear regression analysis for continuous variables, with ordinal numbers 1 to 3 assigned to tertile categories of each dietary pattern.y Physical activity was expressed as the sum of weekly hours spent in sports activity and in walking and cycling to and from work.

S.Akter

etal./

Nutrition

28(2012)

e29–e35

e32

S. Akter et al. / Nutrition 28 (2012) e29–e35 e33

a higher score compared with those with a lower score wereyounger and less likely to drink alcohol and engage in non-jobphysical activity, whereas men with a higher score tended tobe younger and in low job position but were less likely to bemarried, an alcohol drinker, and a smoker.

Table 3 lists geometric means of C-peptide concentrationsaccording to each dietary pattern score separately for men andwomen. In women, a statistically significant inverse associationwas observed for the Westernized breakfast pattern. Themultivariable-adjusted geometric means of the C-peptideconcentrations were 1.03, 0.95, and 0.89 ng/mL for the lowestthrough the highest tertiles of theWesternized breakfast patternscore (P for trend ¼ 0.015). Neither the healthy Japanese dietarypattern nor the animal food pattern was related to the C-peptideconcentrations in women overall. Dietary patterns were notsignificantly associated with C-peptide concentrations in men.

In analyses stratified by BMI, a marginally significant, inverseassociation was observed between the healthy Japanese dietarypattern andC-peptide concentrations inwomenwith a lower BMI(<20.57 kg/m2) in an age- and workplace-adjusted model (P fortrend ¼ 0.07), whose association was slightly attenuated afterfurther adjustment for other variables (P for trend ¼ 0.09). Themultivariable-adjusted geometric means were 0.93 ng/mL (95%CI 0.82–1.06), 0.88 ng/mL (95%CI 0.78–0.99), and 0.79ng/mL (95%CI 0.69–0.90), respectively, for the lowest through the highesttertiles of the healthy Japanese dietary pattern. In women witha higher BMI (�20.6 kg/m2), no such associationwas observed (Pfor trend ¼ 0.48, P for interaction ¼ 0.24). An inverse associationbetween the Westernized breakfast pattern and C-peptideconcentrations was similarly observed, although not statisticallysignificant, in the two groups of women stratified by BMI.

A similar inverse association was observed between theWesternized breakfast pattern and the HOMA2-IR index inwomen (Appendix). The multivariable-adjusted geometric

Table 3Serum C-peptide concentrations (nanograms per milliliter) by tertile of dietary patter

Tertiles of dietary pattern score

T1 (low)

MenNumber of subjects 90Healthy Japanese dietary patternAge and workplace adjusted 1.13 (1.03–1.22)z

Multivariable adjustedy 1.14 (1.06–1.23)Animal food patternAge and workplace adjusted 1.16 (1.06–1.26)Multivariable adjustedy 1.16 (1.08–1.25)

Westernized breakfast patternAge and workplace adjusted 1.17 (1.07–1.28)Multivariable adjustedy 1.12 (1.03–1.21)

WomenNumber of subjects 62Healthy Japanese dietary patternAge and workplace adjusted 1.05 (0.96–1.16)Multivariable adjustedy 1.07 (0.98–1.16)

Animal food patternAge and workplace adjusted 0.96 (0.88–1.06)Multivariable adjustedy 0.98 (0.90–1.07)

Westernized breakfast patternAge and workplace adjusted 1.00 (0.91–1.10)Multivariable adjustedy 1.03 (0.95–1.12)

T1 to T3, tertiles 1 to 3* Based on multiple regression analysis, with ordinal numbers 1 to 3 assigned to thy Adjusted for age (year, continuous), workplace (A or B), sedentary work (yes or

women), marital status (married or unmarried), non-job physical activity (0, >0 to <2g/d, or ethanol �20 g/d), total energy intake (kilocalories per day, continuous), and b

z Geometric mean (95% confidence interval) for all such values.

means of the HOMA2-IR index were 0.77 (95% CI 0.70–0.84), 0.67(95% CI 0.61–0.73), and 0.65 (95% CI 0.69–0.90) for the lowestthrough the highest tertiles of theWesternized breakfast patternscore (P for trend ¼ 0.012). Other dietary patterns in women andall dietary patterns in menwere not significantly associated withthe HOMA2-IR.

Discussion

In this cross-sectional study of Japanese men and women, weidentified three major dietary patterns. Of these, the Western-ized breakfast pattern, characterized by high intakes of bread,confectioneries, milk and yogurt, mayonnaise, and eggs and lowintakes of rice, alcohol, and fish, was inversely associated withC-peptide concentrations in women. Similar associations wereobserved between the dietary patterns and the HOMA2-IR. Toour knowledge, this is the first investigation in Japanese adultsthat addresses the relation of dietary patterns to these markersof insulin resistance.

Dietary patterns similar to the Westernized breakfast patternin this study have been observed in Japanese populations [7,23–25]. In addition, the present finding of an inverse associa-tion between the Westernized breakfast pattern and markers ofinsulin resistance in women are in line with those of Japanesestudies in which similar dietary patterns were identified [7,23].In a previous study in middle-aged men, a dietary patterncharacterized by higher intakes of dairy products, confection-aries, bread, fruits, and vegetables and lower alcohol consump-tion was inversely associated with a lower prevalence of glucosetolerance abnormalities, including type 2 diabetes [7]. Ina large-scale cross-sectional study in Japanese adults, a West-ernized breakfast pattern characterized by the frequentconsumption of bread, margarine, and coffee and the infrequentconsumption of rice andmiso soupwas inversely associatedwith

n score

s

T2 T3 (high) P for trend*

90 90

1.22 (1.12–1.32) 1.18 (1.09–1.29) 0.411.18 (1.10–1.27) 1.20 (1.11–1.29) 0.18

1.17 (1.07–1.27) 1.20 (1.10–1.30) 0.611.15 (1.07–1.24) 1.21 (1.13–1.30) 0.42

1.20 (1.10–1.31) 1.15 (1.06–1.26) 0.801.17 (1.09–1.26) 1.23 (1.14–1.33) 0.14

62 62

0.84 (0.77–0.92) 0.98 (0.90–1.07) 0.390.86 (0.79–0.93) 0.95 (0.88–1.03) 0.12

0.91 (0.83–0.99) 1.00 (0.91–1.09) 0.610.94 (0.86–1.02) 0.95 (0.87–1.03) 0.50

0.96 (0.88–1.05) 0.91 (0.83–1.00) 0.160.95 (0.88–1.03) 0.89 (0.82–0.97) 0.015

e tertile categories of each dietary pattern.no), job position (low, medium, or high for men; low or medium and high forh/w, or �2 h/wk), current smoking (yes or no), alcohol drinking (no, ethanol <20ody mass index (kilograms per meter squared, continuous).

S. Akter et al. / Nutrition 28 (2012) e29–e35e34

an impaired glucose status as measured by hemoglobin A1c [23].The present and previous findings in Japanese populationscontradict the common belief that Westernization of the diet hasincreased chronic diseases including type 2 diabetes. In factsome foods characterizing the Westernized breakfast patternmay have favorable effects on glucose metabolism. For instance,given that fat decreases the glucose response [26], the intake offatty foods such as milk and yogurt may contribute to themoderation of the glycemic response after a meal. Moreover,there is epidemiologic evidence supporting a protective role ofdairy products against the insulin-resistance syndrome [27].Besides the independent effect of foods constituting the West-ernized breakfast pattern, the complex interactions among themmight have contributed to the observed protective association.

Contrary to general expectations, the healthy Japanese die-tary pattern, characterized by high intakes of vegetables, fruit,mushrooms, and soy products, was not associated with lowerC-peptide concentrations in men and women. In some [28,29]but not all [7,30] studies, the prudent or healthy dietarypattern has been related to a lower risk of type 2 diabetes. Inaddition, in an Iranian study, the prudent dietary patternwas notassociated with plasma glucose [31]. It should be noted thatthere was an important difference in the food loading patternfrom the dietary pattern among some of these studies and thepresent study. For instance, the healthy Japanese dietary patternin our study was not characterized by fish intake, which wasa feature of this pattern in another study [29]. In a meta-analysisof cohort studies [32], a larger intake of total vegetables, fruit, orfruit and vegetables combined was not associated with a lowerrisk of type 2 diabetes. The lack of an association between thehealthy Japanese dietary pattern and the C-peptide concentra-tions in the present study appears to be consistent with thefindings of those studies.

The animal food pattern, characterized by high intakes of fishand shellfish, meat, processed meat, mayonnaise, and eggs, wasnot appreciably associated with increased C-peptide concentra-tions in the present study. In two studies of a US population, theanimal food pattern (named the “Western pattern”) was positivelyassociated with serum C-peptide concentrations [16,17]. In thatstudy, the Western pattern was characterized by high intakesof meat and high-fat dairy products but was not associated withfish intake. In some studies [33,34], fish and shellfish intakeswere associated with a lower risk of type 2 diabetes. Comparedwith Western populations, the Japanese consumed much largeramounts of fish; the fish and shellfish supplies are 61 and 20 to 25kg/capita annually in Japan and Western countries, respectively[35]. In contrast, fat-derived energy intakes account for 20% to 25%of the total energy intake in Japanese adults [2], which is muchlower than those in Western populations [35]. Therefore, theanimal food dietary pattern, featured by amodestmeat intake anda high fish intake, may not adversely influence insulin resistancein the Japanese population.

In the present study, we observed no association between thedietary patterns and the C-peptide concentrations in men overallor in any subgroup of men. One possible explanation for thisfinding is that smoking, which was much more prevalent in menthan inwomen in our study, may play a dominant role in glucosemetabolism over dietary factors in men. In a previous study, thedietary patterns were associatedwith risk of type 2 diabetes onlyin never-smokers [28].

The major strengths of the present study include the highstudy participation rate (91%), the use of a validated question-naire (BDHQ), and the adjustment for strong potentialcofounding variables. In addition, because this study was

conducted in selected municipal offices at the time of thenon-selective recruitment for the annual health checkup witha high study participation rate, the possibility of selection biaswas low. Our study has several limitations that warrant mention.First, an association derived from a cross-sectional study doesnot necessarily indicate causality. We excluded participants whoreported a history of diabetes and severe diseases that mightaffect C-peptide concentrations or dietary habits to minimize thepossibility of reverse causality. However, some participantsmight have had undiagnosed diseases that affected insulinresistance. Second, we relied on a single time-point measure-ment of serum C-peptide, which decreased our ability to eval-uate an association between long-term circulating levels.However, Kaaks et al. [36] showed that the intraindividualvariation of the C-peptide level across 1 y is moderate, suggest-ing that a single measurement may be adequate to predict long-term circulating levels. The dietary analysis based on theprincipal component analysis involves subjective decisions indetermining and labeling the dietary patterns, which makes itdifficult to replicate the procedure. Fourth, because of the rela-tively small sample, the statistical power might not have beensufficient to detect a moderate association. Fifth, because somemeasurement errors are unavoidable in the assessment of con-founding variables, the possibility of residual confoundingcannot be excluded. Sixth, our study subjects do not representa random sample of the Japanese population, and caution isrequired in generalizing the present results to the entire Japa-nese population.

In conclusion, the Westernized breakfast pattern, character-ized by the high consumption of bread, confectionaries, and milkand yogurt but the low consumption of rice and alcohol, wasassociated with decreased C-peptide concentrations and theHOMA2-IR index in Japanese women. The present finding andthose of some other studies in Japan support that a Westernizedbreakfast pattern or a similar pattern is associated with lowerinsulin resistance, an important predictor of type 2 diabetes. Ourfinding needs to be confirmed by prospective studies.

Acknowledgments

The authors thank the study subjects for their cooperationand participation. The authors also thank Tamami Hatano, Aki-hiro Tanaka, and Yuko Ejima (Kyushu University, Fukuoka,Japan); Mio Ozawa (Fukuoka Women’s University, Fukuoka,Japan); and Akiko Hayashi and Kie Nagao (National Center forGlobal Health and Medicine) for their help in data collection.

References

[1] Shaw JE, Sicree RA, Zimmet PZ. Global estimates of the prevalence of dia-betes for 2010 and 2030. Diabetes Res Clin Pract 2010;87:4–14.

[2] Japan Ministry of Health, Labour and Welfare. Annual reports of theNational Health and Nutrition Survey in Japan, 2007. Tokyo: Daiichi-shuppan; 2010 (in Japanese).

[3] Hu FB. Globalization of diabetes: the role of diet, lifestyle, and genes.Diabetes Care 2011;34:1249–57.

[4] Murakami K, Sasaki S, Takahashi Y, Uenishi K, Japan Dietetic Students’Study for Nutrition and Biomarkers Group. Neighborhood socioeconomicstatus in relation to dietary intake and insulin resistance syndrome infemale Japanese dietetic students. Nutrition 2010;26:508–14.

[5] Newby PK, Tucker KL. Empirically derived eating patterns using factor orcluster analysis: a review. Nutr Rev 2004;62:177–203.

[6] Nettleton JA, Steffen LM, Ni H, Liu K, Jacobs DR Jr. Dietary patterns and riskof incident type 2 diabetes in the Multi-Ethnic Study of Atherosclerosis(MESA). Diabetes Care 2008;31:1777–82.

[7] Mizoue T, Yamaji T, Tabata S, Yamaguchi K, Ogawa S, Mineshita M, et al.Dietary patterns and glucose tolerance abnormalities in Japanese men. JNutr 2006;136:1352–8.

S. Akter et al. / Nutrition 28 (2012) e29–e35 e35

[8] Bloomgarden ZT. Developments in diabetes and insulin resistance. DiabetesCare 2006;29:161–7.

[9] Reaven GM. Insulin resistance/compensatory hyperinsulinemia, essentialhypertension, and cardiovascular disease. J Clin Endocrinol Metab2003;88:2399–403.

[10] Chen CH, Tsai ST, Chou P. Correlation of fasting serum C-peptide and insulinwith markers of metabolic syndrome-X in a homogenous Chinese pop-ulation with normal glucose tolerance. Int J Cardiol 1999;68:179–86.

[11] Kim BY, Jung CH, Mok JO, Kang SK, Kim CH. Association between serumC-peptide levels and chronic micro vascular complications in Korean type 2diabetic patients. Acta Diabetol 2012;49:9–15.

[12] Wu T, Giovannucci E, Pischon T, Hankinson SE, Ma J, Rifai N, et al. Fructose,glycemic load, and quantity and quality of carbohydrate in relation to plasmaC-peptide concentrations in US women. Am J Clin Nutr 2004;80:1043–9.

[13] Wayne SJ, Neuhouser ML, Ulrich CM, Koprowski C, Baumgartner KB,Baumgartner RN, et al. Dietary fiber is associated with serum sex hormonesand insulin-related peptides in postmenopausal breast cancer survivors.Breast Cancer Res Treat 2008;112:149–58.

[14] Wu T, Willett WC, Giovannucci E. Plasma C-peptide is inversely associatedwith calcium intake in women and with plasma 25-hydroxy vitamin D inmen. J Nutr 2009;139:547–54.

[15] Wu T, Willett WC, Hankinson SE, Giovannucci E. Caffeinated coffee, decaf-feinated coffee, and caffeine in relation to plasma C-peptide levels, a markerof insulin secretion, in U.S. women. Diabetes Care 2005;28:1390–6.

[16] Kerver JM, Yang EJ, Bianchi L, Song WO. Dietary patterns associated withrisk factors for cardiovascular disease in healthy US adults. Am J Clin Nutr2003;78:1103–10.

[17] Fung TT, Rimm EB, Spiegelman D, Rifai N, Tofler GH, Willett WC, et al.Association between dietary patterns and plasma biomarkers of obesityand cardiovascular disease risk. Am J Clin Nutr 2001;73:61–7.

[18] Yazaki Y, Kadowaki T. Combating diabetes and obesity in Japan. Nat Med2006;12:73–4.

[19] Murakami K, Mizoue T, Sasaki S, Ohta M, Sato M, Matsushita Y, Mishima N.Dietary intake of folate, other B vitamins, and omega-3 polyunsaturatedfatty acids in relation to depressive symptoms in Japanese adults. Nutrition2008;24:140–7.

[20] Nanri A, Kimura Y, Matsushita Y, Sato M, Mishima N, Sasaki S, et al. Dietarypattern and depressive symptoms among Japanese men and women. Eur JClin Nutr 2010;64:832–9.

[21] Science and Technology Agency. Standard tables of food composition inJapan (in Japanese). 5th rev enlarged ed. Tokyo: Printing Bureau of theMinistry of Finance; 2005.

[22] Kobayashi S, Murakami K, Sasaki S, Okubo H, Hirota N, Notsu A, et al.Comparison of relative validity for food group intake estimated by

Appendix

Homeostasis model assessment of insulin resistance by tertile of dietary pattern

Tertiles of dietary pattern score

T1 (low)

MenNumber of subjects 84Healthy Japanese dietary patternMultivariable adjustedy 0.84 (0.78–0.91)z

Animal food patternMultivariable adjustedy 0.85 (0.79–0.92)

Westernized breakfast patternMultivariable adjustedy 0.83 (0.77–0.91)

WomenNumber of subjects 53Healthy Japanese dietary patternMultivariable adjustedy 0.74 (0.68–0.82)

Animal food patternMultivariable adjustedy 0.70 (0.64–0.77)

Westernized breakfast patternMultivariable adjustedy 0.77 (0.70–0.84)

T1 to T3, tertiles 1 to 3* Based on multiple regression analysis, with ordinal numbers 1 to 3 assigned to thy Adjusted for age (year, continuous), workplace (A or B), sedentary work (yes o

for women), marital status (married or unmarried), non-job physical activity (0, >0 to<20 g/d, or ethanol �20 g/d), total energy intake (kilocalories per day, continuous), a

z Geometric mean (95% confidence interval) for all such values.

comprehensive and brief-type self-administered diet history questionnairesagainst 16-day dietary records in Japanese adults. Public Health Nutr2011;14:1200–11.

[23] Nanri A, Mizoue T, Yoshida D, Takahashi R, Takayanagi R. Dietary patternsand A1C in Japanese men and women. Diabetes Care 2008;31:1568–73.

[24] Shimazu T, Kuriyama S, Hozawa A, Ohmori K, Sato Y, Nakaya N, et al.Dietary patterns and cardiovascular disease mortality in Japan: a prospec-tive cohort study. Int J Epidemiol 2007;36:600–9.

[25] Sadakane A, Tsutsumi A, Gotoh T, Ishikawa S, Ojima T, Kario K, et al. Dietarypatterns and levels of blood pressure and serum lipids in a Japanesepopulation. J Epidemiol 2008;18:58–67.

[26] Owen B, Wolever TM. Effect of fat on glycaemic responses in normalsubjects: a dose–response study. Nutr Res 2003;23:1341–7.

[27] Lutsey PL, Steffen LM, Stevens J. Dietary intake and the development of themetabolic syndrome: the Atherosclerosis Risk in Communities study.Circulation 2008;117:754–61.

[28] Odegaard AO, Koh WP, Butler LM, Duval S, Gross MD, Yu MC, et al. Dietarypatterns and incident type 2 diabetes in Chinese men and women: theSingapore Chinese Health Study. Diabetes Care 2011;34:880–5.

[29] Yu R, Woo J, Chan R, Sham A, Ho S, Tso A, et al. Relationship betweendietary intake and the development of type 2 diabetes in a Chinesepopulation: the Hong Kong Dietary Survey. Public Health Nutr 2011;14:1133–41.

[30] Kim HS, Park SY, Grandinetti A, Holck PS, Waslien C. Major dietary patterns,ethnicity, and prevalence of type 2 diabetes in rural Hawaii. Nutrition2008;24:1065–72.

[31] Amini M, Esmaillzadeh A, Shafaeizadeh S, Behrooz J, Zare M. Relationshipbetween major dietary patterns and metabolic syndrome among individ-uals with impaired glucose tolerance. Nutrition 2010;26:986–92.

[32] Carter P, Gray LJ, Troughton J, Khunti K, Davies MJ. Fruit and vegetableintake and incidence of type 2 diabetes mellitus: systematic review andmeta-analysis. BMJ 2010;341:c4229.

[33] Villegas R, Xiang YB, Elasy T, Li HL, Yang G, Cai H, et al. Fish, shellfish, andlong-chain n-3 fatty acid consumption and risk of incident type 2 diabetesin middle-aged Chinese men and women. Am J Clin Nutr 2011;94:543–51.

[34] Nanri A, Mizoue T, Noda M, Takahashi Y, Matsushita Y, Poudel-Tandukar K,et al. Fish intake and type 2 diabetes in Japanese men and women: theJapan Public Health Center–based prospective study. Am J Clin Nutr2011;94:884–91.

[35] FAO Statistics Division 2010. Available at: http://faostat.fao.org/site/610/default.aspx#ancor. Accessed October 13, 2011.

[36] Kaaks R, Toniolo P, Akhmedkhanov A, Lukanova A, Biessy C, Dechaud H, et al.Serum C-peptide, insulin-like growth factor (IGF)-I, IGF-binding proteins,and colorectal cancer risk in women. J Natl Cancer Inst 2000;92:1592–600.

score (n [ 411)

s

T2 T3 (high) P for trend*

84 84

0.87 (0.80–0.94) 0.92 (0.85–0.99) 0.15

0.89 (0.82–0.96) 0.89 (0.82–0.96) 0.43

0.87 (0.81–0.94) 0.92 (0.85–1.00) 0.14

53 53

0.64 (0.59–0.70) 0.70 (0.64–0.77) 0.39

0.67 (0.61–0.73) 0.72 (0.65–0.79) 0.78

0.67 (0.61–0.73) 0.65 (0.59–0.71) 0.012

e tertile categories of each dietary pattern.r no), job position (low, medium, or high for men; low or medium and high<2 h/wk, or �2 h/wk), current smoking (yes or no), alcohol drinking (no, ethanolnd body mass index (kilograms per meter squared, continuous).