|
|
||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
RAPID COMMUNICATION |
Departments of1 , Nutrition and Dietetics2 Home Economics and Ecology, Harokopio University, Athens 17671, Greece3 Division of Endocrinology, Diabetes and Metabolism, ST 816, Department of Medicine, Harvard Medical School, Beth Israel Deaconess Medical Center (BIDMC), 330 Brookline Avenue, Boston, Massachusetts 02215, USA
(Correspondence should be addressed to C S Mantzoros; Email: cmantzor{at}bidmc.harvard.edu)
Objective: Our aim was to investigate associations between dietary factors and high molecular weight (HMW) as well as total adiponectin in a sample of apparently healthy adult Mediterranean women.
Design and methods: Two hundred and twenty women were enrolled in this study. Anthropometric and body composition measurements were performed in all subjects. Fasting blood samples were taken; HMW and total adiponectin concentrations were measured. Food intake was evaluated using 3-day food records. The frequency of consumption of several food groups was approximately quantified in terms of number of servings per day. Furthermore, dietary intakes of betaine, choline, and free choline were estimated.
Results: Women in the highest HMW adiponectin tertile had higher fruit intake compared with those with lower levels, after adjusting for potential confounders (P=0.04). On the contrary, dietary betaine and choline intakes were not different among HMW adiponectin tertiles. In linear models, fruit consumption, controlling for biological and lifestyle variables, was significantly related to HMW adiponectin (partial r=0.15, P=0.04), but the association with total adiponectin did not reach statistical significance (partial r=0.11, P=0.12). A significant negative correlation between total adiponectin and refined cereals was also observed (partial r=–0.16, P=0.03).
Discussion: This is the first study that evaluates associations between dietary factors and HMW adiponectin levels. The associations found are moderate and indicate that, after multivariate adjustment, fruit consumption is related to HMW adiponectin in both linear and nonlinear models.
| Introduction |
|---|
|
|
|---|
, after multivariate adjustment for various demographic, clinical, and lifestyle characteristics in a sample of apparently healthy Greek adults (9). Although adiponectin counteracts the proinflammatory effects of tumor necrosis factor-
and IL-6 inhibits adiponectin expression and secretion in adipocytes (10), associations between the above dietary components and total or HMW adiponectin levels have not been evaluated. Thus, our aim was to investigate associations between several dietary factors and total as well as HMW adiponectin in a sample of apparently healthy adult Mediterranean women. | Subjects and methods |
|---|
|
|
|---|
Two hundred and twenty Greek women (mean age, 48.3±12.3 years; age range 18–84 years) were consecutively enrolled in this study. Women were recruited through local advertisement. None of them was referred directly by a physician. Although study subjects reported no known history of diabetes, we found two of them with fasting glucose levels higher than 126 mg/dl. Furthermore, 7 subjects were taking medication (corticosteroids, lipid-lowering drugs, or other drugs) and 15 were on a weight-reducing diet. The study protocol was approved by the Ethics Committee of Harokopio University; all subjects were informed on the purpose and procedures of the study and signed a consent form. They provided demographic, medical, and lifestyle information, including dietary, physical activity, and smoking habits. A fasting blood sample was taken from all participants who also underwent anthropometric and body composition measurements. With regard to menstruation status, they were classified as premenopausal if they had regular menses, perimenopausal if they were suffering from irregular menses, and postmenopausal if they had ceased menstruating for at least 12 months.
Body composition
Body composition was evaluated in all subjects using dual-energy X-ray absorptiometry whole-body scanner (Model DPX+; Lunar Corp., Madison, WI, USA). Anthropometric measurements were also performed, namely body weight, height, waist, and hip circumferences. Body mass index (BMI) was calculated as kg/m2.
Dietary assessment
Dietary intake of the 220 participants was assessed using 3-day food records. Subjects were asked to record type and amount of food and beverage consumed for two specific consecutive weekdays and one weekend day during the same week. Clear instructions were given to them on how to record the quantity of food eaten using standard household and other measures. The frequency of consumption of several food groups was approximately quantified in terms of number of servings per day. The food groups assessed were the core food groups of the traditional Greek diet, as depicted in the Mediterranean diet pyramid and the dietary guidelines for the Greek population (11). For the present analysis, however, only food groups assumed to be consumed on a daily basis were used, namely fruits and vegetables, dairy products (low and full fat), and cereals (nonrefined and refined). Dietary intakes of betaine, choline, and free choline were calculated based on the food values provided by the US Department of Agriculture. Low-energy reporting was evaluated for each subject using the ratio of the energy intake/basal metabolic rate (EI/BMR). BMR was estimated by the Schofield equations that have been adopted by the 2004 Food and Agriculture Organization (FAO)/WHO/United Nations University (UNU) report (12). Participants with EI/BMR<1.04 were classified as low-energy reporters, whereas those with EI/BMR
1.04 normal energy reporters or non-low-energy reporters (13).
Physical activity assessment
Assessment of physical activity was performed through a brief self-reported questionnaire (the Harokopio Physical Activity Questionnaire (HAPAQ)) that collects the previous week's self-reported physical activity (14). HAPAQ examines the time spent in light, moderate, and high-intensity activities and also requires sleeping to be recorded. The questionnaire is based on the metabolic equivalents of all activities of the previous week, including activities at work, leisure time, and rest or sleep, thus allowing the prediction of mean physical activity level (PAL).
Biochemical analysis
Blood samples were drawn after a 12-h fast between 0830 and 1030 h, and plasma was immediately frozen in –80 °C until biochemical analysis. Plasma glucose concentrations and lipid profile were determined using commercially available enzymatic colorimetric assays (Alfa Wassermann BV, Woerden, The Netherlands) on an automated ACE analyzer (Schiapparelli Biosystems, Inc, Fairfield, NJ, USA). Plasma adiponectin and insulin concentrations were measured by RIA (adiponectin (Linco Research, St Charles, MO, USA): sensitivity 1 ng/ml, intraassay coefficient variation 1.78–6.21%; insulin (Diagnostics Systems Laboratory, Webster, TX, USA): sensitivity 1.3 µIU/ml, intraassay coefficient variation 4.5–8.3%). In addition, plasma levels of HMW adiponectin were determined using a novel ELISA test (Human Multimeric Adiponectin ELISA; ALPCO Diagnostics, Salem, NH, USA). The sensitivity of this assay was 0.05 ng/ml. IR was estimated using the homeostasis model assessment (HOMA-IR) with the following formula: HOMA-IR=fasting insulin (µIU/ml)xfasting glucose (mmol/l)/22.5 (15).
Statistical analysis
Continuous variables are presented as mean±S.D. and categorical variables as absolute frequencies. ANOVA was used for evaluating differences in the investigated outcomes across tertiles of HMW or total adiponectin. We examined the potential role of dietary factors as predictors of HMW or total adiponectin concentrations using bivariate and multivariate models. Pearson and partial correlation coefficients were also calculated. All reported P values were based on two-tailed tests at a significance level of 5%. Statistical package for Social Sciences software, version 13.0 (SPSS Inc. 2003, Chicago, IL, USA), was used for all the statistical calculations.
| Results |
|---|
|
|
|---|
|
When women were categorized according to their total adiponectin tertiles, associations observed with anthropometric variables were similar to those observed in the HMW adiponectin analysis. Furthermore, women in the lowest tertile of adiponectin levels had a higher percentage of energy from carbohydrates (P=0.03) and lower from fat (P=0.02), as well as higher total cereal intake (P=0.02; data not shown). These correlations became nonsignificant after multivariate adjustment.
Associations between plasma levels of HMW or total adiponectin and selected food groups were also evaluated in linear models, before and after adjustment for the above-mentioned confounders (Table 2). In multivariate analysis, fruit consumption was significantly related to HMW adiponectin (partial r=0.15, P=0.04), but not significantly to total adiponectin (partial r=0.11, P=0.12), whereas a statistically significant negative correlation between total adiponectin and refined cereals was observed (partial r=–0.16, P=0.03). No statistically significant associations were found between HMW adiponectin and nonrefined cereal consumption (partial r=–0.10, P=0.18) or between total adiponectin and low-fat dairy consumption (partial r=0.11, P=0.13). The results did not change when the analysis was restricted to women not on a weight-reducing diet or those not taking medication and having normal blood glucose levels. After further exclusion of low-energy reporters, the associations remain of similar strength.
|
| Discussion |
|---|
|
|
|---|
Furthermore, the weak, negative association between refined cereals and total adiponectin levels confirms previous findings on the negative association between glycemic index and adiponectin levels (8). No association was found between either dietary betaine or choline intake and HMW adiponectin, suggesting that the potential mechanisms linking these compounds with inflammation (9) may not be mediated by adiponectin levels.
Among the strengths of this study is the use of food records for evaluating subjects' dietary intake, including nutrient intake and food group consumption. Three-day food records provide quantitatively accurate information on food consumed during the investigated period: by recording food while it is consumed, the problem of reporting bias or omission is lessened, whereas subjects are not restricted to select from a predetermined list of foods included in the food frequency questionnaires. Furthermore, this method might provide much more meaningful information regarding short-term regulation of circulating hormone levels. Although plasma adiponectin levels do not vary markedly over longer periods of time in the absence of significant body weight changes (20), there may be a short-term biovariability. Short-term biovariability has been found to be twofold greater in healthy subjects than in those with the metabolic syndrome (21), and the present study has focused on healthy subjects.
The percentage of low-energy reporting in our sample, around 40%, may evoke some concern regarding potential bias. However, this proportion of low-energy reporters could be expected, as our sample consisted of a high percentage of middle-aged, overweight women; the literature points to a higher proportion of low-energy reporting among female, older and overweight, subjects (22). On the other hand, existing evidence indicates that there are no differences in bread, potatoes, meat, or vegetable and fruit consumption between underreporters and other subjects (23), and that sweets and snacks are the mostly underreported food items (24, 25). Our study findings did not include these socially undesirable foods; therefore, we believe that the associations reported herein were not influenced by the misreporting of dietary intake. Finally, to rule out conclusively any possible bias from this source, low-energy reporting was considered as a potential confounder in the additional multivariate models. We also stratified the analysis for low-energy reporters only and the associations found remained of similar strength.
Since this is an observational study evaluating a moderate number of participants, it cannot elucidate mechanisms or determine the direction of causality. Although confounding was appropriately controlled for through standard statistical procedures, residual confounding by other serum hormones or unmeasured factors remains a possibility. Dietary intake was evaluated in terms of individual nutrients or food groups; dietary pattern analysis, assessing intercorrelations between dietary variables, would be another alternative for studying interactions between biological indexes and diet. Measurement error in terms of diet and/or determination of hormone levels would result in random misclassification and could have suppressed the relevant effect estimates, but could not have resulted in demonstrating statistical significance. Confirmation of our results by future studies in other population samples, including also men, and assessment of the potential more prolonged effects of diet on both HMW and total adiponectin are warranted.
| Declaration of interest |
|---|
|
|
|---|
| Funding |
|---|
| References |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
P. Perez-Martinez, D. Corella, J. Shen, D. K Arnett, N. Yiannakouris, E S. Tai, M. Orho-Melander, K. L Tucker, M. Tsai, R. J Straka, et al. Association between glucokinase regulatory protein (GCKR) and apolipoprotein A5 (APOA5) gene polymorphisms and triacylglycerol concentrations in fasting, postprandial, and fenofibrate-treated states Am. J. Clinical Nutrition, January 1, 2009; 89(1): 391 - 399. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |