Introduction

Coffee and tea are the two most popular beverages worldwide, and their health effects have attracted increasing attention [1, 2]. Accumulating evidence suggests that coffee consumption affords protection against type 2 diabetes (T2D) [3] and cancer [4]. Likewise, habitual green tea consumption, a widely consumed tea among Asians, is reportedly associated with lower risk of some types of cancer, such as colorectal cancer [5]. Attempts have been made to decipher pathways underlying potentially beneficial roles of these two beverages against risks of T2D [6, 7] and cancer [8, 9]. Studies have linked circulating adipokines, which are hormones and cytokines produced by adipose tissues, to metabolic diseases [10]. Among the adipokines, adiponectin and leptin are major hormones secreted exclusively by the adipocytes and of considerable interest. It has been shown that adiponectin was inversely, whereas leptin was positively, associated with T2D [11, 12] and cancer risk [13]. Other adipokines, namely resistin, visfatin and plasminogen activator inhibitor-1 (PAI-1) have also been shown to be unfavorably associated with cancer risk [14–16]. It remains unclear whether a beneficial effect of coffee and green tea consumption against T2D and cancer is operated through improved adipokine profile.

Evidence for the association between coffee and green tea consumption and adipokines is inconsistent, scarce or lacking. Several studies showed a higher level of circulating adiponectin with increasing coffee consumption [17–23], while others did not [24, 25]. Few studies have examined the association between coffee consumption and circulating leptin, showing an inverse association [22, 26] or no change in serum leptin at higher levels of coffee consumption [18]. As for other adipocytokines including resistin, visfatin and PAI-1, we are aware of only one clinical study reporting an increase in plasma PAI-1 in heavy coffee drinkers than in light coffee drinkers [27]. Concerning green tea, no association with adiponectin [19, 25] and no effect on leptin [28] were reported. Coffee and green tea are a major source of caffeine, and studies of caffeine consumption and adipokines are limited, with one showing a positive association with adiponectin in US diabetic, but not non-diabetic women [17] and another one reporting an inverse association between caffeine consumption and plasma leptin [29].

Moreover, there is a sex difference in distribution of two major adipokines, leptin and adiponectin [30], which may lead to differential associations between them and consumption of coffee and green tea. However, this issue has received less attention. We thus conducted the present study to examine cross-sectionally the association of consumption of coffee, green tea and caffeine with serum adipokine among a Japanese working population. We hypothesized that higher consumption of these beverages is associated with higher adiponectin concentrations, but is associated with lower concentrations of leptin, resistin, visfatin and PAI-1.

Subjects and methods

Study procedure and subjects

The present study was based on surveys among municipal employees of two municipal offices in northeastern Kyushu, Japan (conducted in July 2009 in one office and November 2009 in another). Details of the survey have been described elsewhere [31]. Briefly, of all workers in these municipal offices (n = 605, mainly administrators, nurses and school lunch preparers) who were invited to the survey, 567 (325 men and 242 women aged 20–68 years) agreed to take part in this study (response rate 94 %). Participants were invited to fill in a survey questionnaire before the health checkup. The survey questionnaire was checked by research staff for completeness, and where necessary, clarifications were made with the subjects on the day of health examination. We also obtained data that were routinely collected in the health examination, including anthropometric measurements, biochemical data and information about medical history, smoking and alcohol drinking. We excluded participants who reported having a history of diabetes (n = 8), cancer (n = 13), cardiovascular disease (n = 11), nephritis (n = 1) and those receiving medical care for chronic hepatitis (n = 3). Additionally, we excluded pregnant participants (n = 8). Those who had missing data on serum adipokines (n = 17) were further excluded. Some of the excluded subjects had two or more conditions for exclusion, leaving a total of 509 subjects (296 men and 213 women). Of these, we retained 506 subjects for the analysis of visfatin after excluding those with visfatin levels below (n = 2) or above (n = 1) the detection limits. Data for the analysis of leptin included subjects under fasting condition (n = 486). The study protocol was reviewed and approved by the Ethics Committee of the National Center for Global Health and Medicine, Japan. Written informed consent was obtained from all subjects prior to the survey.

Laboratory procedures

Blood samples were obtained on the day of the health checkup. Venous blood (7 mL) was drawn into vacuum tubes and transported in a cooler box to the laboratory. Serum samples for measurement of adipokines were stored at −80 °C until biochemical assay. To quantify serum concentrations of adiponectin, leptin, resistin, visfatin and PAI-1, a Luminex suspension bead-based multiplexed array was performed using Bio-Plex 3D suspension array system and Bio-Plex Pro human diabetes assay panel (Bio-Rad Laboratories, Hercules, CA); the intra-assay coefficients of variation were 12 % for adiponectin, 11 % for leptin, 8 % for resistin, 19 % for visfatin and 21 % for PAI-1. Luminex technology, which enables simultaneous assessment of multiple biomarkers, has been widely used. The system requires only a small amount of clinical samples, and good correlation with ELISA and the usefulness for epidemiological studies have been well demonstrated [32].

Dietary assessment

Information about dietary intake during the preceding month was obtained using a validated brief self-administered diet history questionnaire (BDHQ) [33], which ascertained frequency of 46 food and non-alcoholic beverage items, including coffee and green tea. The response options for coffee or green tea consumption were never, <1 cup/week, 1 cup/week, 2–3 cups/week, 4–6 cups/week, 1 cup/day, 2–3 cups/day and ≥4 cups/day. Dietary energy and intakes of selected nutrients were estimated using an ad hoc computer algorithm with reference to the Standard Tables of Food Composition in Japan [34]. Correlations between consumptions of coffee and green tea according to the above-mentioned BDHQ and those from 16-day dietary records were high (Spearman’s r = 0.83 and 0.77 for coffee in men and women, respectively; Spearman’s r = 0.68 and 0.64 for green tea in men and women, respectively) [33]. To estimate caffeine consumption from coffee and green tea consumption, a value (cups/week) was, respectively, assigned to the above-mentioned consumption frequency of these beverages, namely 0, 0.5, 1, 2.5, 5, 7, 17.5 and 28. A cup size of green tea or coffee was specified as 173 ml for men and 150 ml for women [33]. According to the Food Composition Table in Japan [34], 100 ml of coffee and green tea contains approximately 60 and 20 mg of caffeine, respectively.

Other variables

Marital status, job position, alcohol and tobacco use, occupational and types of occupation were elicited in the questionnaire. Occupational physical activity was classified as sedentary work and active work. Questions on non-occupational physical activity ascertained daily minutes spent for walking or cycling for commuting to or from work, and five recreational activities (walking, low-, moderate-, and high-intensity activities and gardening). Non-occupational physical activity was estimated in metabolic equivalent value [35] and expressed as sum of MET multiplied by the time (in hours) spent performing each activity. Body height was measured to the nearest 0.1 cm with subjects standing without shoes. Body weight in light clothes was measured to the nearest 0.1 kg. Body mass index (BMI) was calculated by dividing weight by squared height (kg/m2).

Statistical analysis

The descriptive data were presented as the mean (standard deviation), median (interquartile range) or percentages. Coffee drinkers were divided into four groups: consuming <1, 1, 2–3 or ≥4 cups/day; green tea drinkers were categorized into three groups: consuming ≤1, 2–3 or ≥4 cups/day. Caffeine consumption was classified into quartiles (≤100, 101–159, 160–291 or ≥292 mg/day). Characteristics according to categories of coffee and green tea consumption were evaluated using ANOVA and Kruskal–Wallis tests for normal and non-normal continuous variables, respectively, and χ 2 test for trend for categorical variables. Analysis of covariance was performed to estimate geometric means and 95 % confidence intervals of adipokine concentrations after log transformation. Multiple linear regression was used to test the trend of association between coffee, green tea and caffeine consumption and serum adipokines by assigning ordinal numbers to the consumption categories of coffee, green tea and caffeine. We first adjusted for age (continuous), sex and workplace (site A or B) (model 1). Second, we adjusted for job position (low or middle and high), occupational physical activity (active or sedentary), non-occupational physical activity (<5 or ≥5 MET-hr/week), smoking status (never smokers, ex-smokers, current smokers consuming 1–19 cigarettes/day or current smokers consuming ≥20 cigarettes/day), alcohol drinking (non-drinker and current drinker consuming alcohol <1 day/week or current drinker consuming alcohol ≥1 days/week), total energy intake (log-transformed, continuous), mutual adjustment for green tea consumption (≤1, 2–3 or ≥4 cups/day) and coffee consumption (<1, 1 or ≥2 cups/day), hypertension (yes or no) defined as systolic blood pressure of ≥140 mmHg or diastolic blood pressure of ≥90 mmHg or current use of antihypertensive drugs, hyperlipidemia (yes or no) defined as serum low-density lipoprotein cholesterol ≥160 mg/dl [36] or current use of antihyperlipidemic drugs, BMI (continuous) and use of non-steroid anti-inflammatory drugs (yes or no) (model 2). In addition, we performed stratified analyses of coffee consumption and serum adipokines according to sex, BMI (<23 or ≥23 kg/m2) and smoking status (in male only: non-smoker or current smoker). Statistical test for interaction was conducted by likelihood ratio test, comparing multiple linear models with and without an interaction term with full adjustment for covariates. The interaction term was generated by multiplying each dichotomized variable mentioned above by categories of coffee, green tea or caffeine consumption, with beverage consumption categories being treated as continuous variables. Two-sided P values of less than 0.05 were considered statistically significant. All analyses were performed using Stata version 12.1 (StataCorp, College Station, TX, USA).

Results

Subject characteristics according to coffee and green tea consumption are summarized in Table 1. Individuals with a higher consumption of coffee were older, less likely to be engaged in sedentary work and more likely to be current smokers. Those with a higher consumption of green tea were more likely to be female and tended to use anti-inflammatory drugs. They were less likely to be current smokers and alcohol drinkers. Coffee consumption was inversely, significantly associated with serum leptin and PAI-1 (P for trend = 0.007 and 0.02, respectively) (Table 2). There was a borderline significant positive association between coffee consumption and serum resistin (P for trend = 0.053). Green tea consumption was unrelated to any serum adipokine. As with coffee drinkers, subjects with higher caffeine consumption had significantly lower serum concentrations of leptin and PAI-1 (P for trend = 0.02 for each adipokine) (Table 3). There was a suggestion of positive association between caffeine consumption and adiponectin (P for trend = 0.11) and resistin (P for trend = 0.09), although levels of these adipokines tended to decrease at the highest quartile of caffeine consumption (≥292 mg/day).

Table 1 Characteristics of the study subjects according to coffee and green tea consumption
Full size table
Table 2 Geometric means (95 % CI) of serum concentrations of adipokines according to coffee and green tea consumption
Full size table
Table 3 Geometric means (95 % CI) of serum concentrations of adipokines according to quartiles of caffeine consumption
Full size table

In subgroup analyses, we found that serum adiponectin concentrations were progressively higher with increasing coffee consumption in men (P for trend = 0.046), but not in women (P for trend = 0.43), although the interaction by sex was not statistically significant (P for interaction = 0.11) (Table 4). The inverse association of coffee consumption and serum leptin was evident in men only (P for trend = 0.03), albeit the interaction by sex did not attain statistical significance (P for interaction = 0.67). In addition, we observed a positive association between coffee consumption and serum resistin in men (P for trend = 0.004), but not in women (P for trend = 0.73), and such an association was modified by sex (P for interaction = 0.03). In men, there was a positive association between coffee consumption and resistin in current smokers (P for trend = 0.01), but not in non-smokers (P for trend = 0.35, P for interaction = 0.11). Similar associations were observed for caffeine consumption with regard to analyses stratified by sex, BMI and smoking status in men (Supplemental table). Of note, there was a significant interaction between smoking and coffee for serum resistin concentration in men (P for interaction = 0.04).

Table 4 Geometric means (95 % CI)a of serum concentrations of adipokines according to coffee consumption stratified by sex, BMI and smoking status (in men only)
Full size table

Discussion

In the present study, among a working Japanese population, we found inverse associations of coffee and caffeine consumption with serum concentrations of leptin and PAI-1. Although we found no significant association of coffee and caffeine consumption with serum adiponectin in all subjects, we noted a positive association of coffee and caffeine consumption with adiponectin in men, but not in women. Moreover, there was a suggestion of positive association between coffee consumption and resistin in male current smokers. Green tea consumption was not associated with any adipokine. This is one of the few studies addressing the association of coffee and caffeine consumption with adiponectin, leptin and PAI-1 among Asians and is the first investigation to report a positive association with resistin in smoking individuals.

Our findings on the association of coffee and caffeine consumption with serum leptin are consistent with those of some previous studies. A Japanese cross-sectional study of 3317 adults found a significant inverse association between coffee consumption and serum leptin [22]. A small study (n = 48) among healthy elderly Greek men also showed a lower leptin concentration in coffee drinkers consuming ≥2 cups/day than in those drinking <2 cups/day, albeit the difference was not statistically significant [26]. Additionally, a Swiss study including 76 subjects during the weight maintenance period disclosed an inverse association between caffeine consumption and circulating leptin [29]. Moreover, regular coffee consumption has consistently been associated with a decreased risk of T2D [3], in which circulating leptin is a good predictor [12]. Contrarily, a study of 5,556 participants in Switzerland found no association between caffeine consumption and plasma leptin [37]. This null association might be ascribed to the effect of potential confounding factors, including smoking, energy intake, hypercholesterolemia and hypertension, all of which were not adjusted in that study. Null effect of coffee ingestion on serum leptin was also shown in a non-randomized clinical trial [18]. A limited number of subjects (n = 47) and insufficient intervention period (8 weeks) may overshadow a potentially beneficial effect of coffee on leptin. Our data coupled with previous studies [22, 26, 29] suggest that habitual coffee consumption may be associated with lower blood leptin concentration.

We found a significant positive association of coffee and caffeine consumption with serum adiponectin in men, but not in women. Consistent with our results in men, two previous Japanese studies among men reported a positive association between coffee consumption and serum adiponectin [19, 23]. A favorable association of coffee consumption with serum adiponectin was also reported in one Japanese study, with a male preponderance (77 %) [22]. Similarly, a Swiss study observed a positive association between caffeine consumption and plasma adiponectin in men, but not in women [37]. However, in another Japanese study among a relatively small number of middle-aged and older men (n = 205), regular coffee consumption was not associated with serum adiponectin [24]. As with our null finding in women, two previous investigations found no association between coffee consumption and circulating adiponectin in apparently healthy Japanese women [24] or non-diabetic US female nurses [17]. The reason for the lack of an association in women needs to be addressed in future studies.

We found an inverse association of coffee and caffeine consumption with serum PAI-1. Contrary to our finding, a Finnish study among 60 hypertensive men showed a higher plasma concentration of PAI-1 in individuals consuming >4 cups of coffee/day than in those drinking <1 cup of coffee/day [27]. The result of Finnish study at clinical setting may be more prone to bias than that of the present investigation among apparently healthy participants. Further studies are needed to confirm the effect of high levels of coffee consumption on PAI-1.

Contrary to our hypothesis, coffee consumption was positively associated with serum resistin in male current smokers, but not in male non-smokers in the present study. This observation could be ascribed to chance, but tobacco smoke and some bioactive substances in coffee may work jointly to trigger low-grade inflammation, in which resistin is highly expressed [38], possibly causing resistin augmentation. Given resistin has been shown to predict cardiovascular events [39], the present data may partly explain an increased risk of major cardiovascular diseases associated with coffee consumption among smokers in some studies [40, 41]. Further investigation is required to elucidate the combined effects of smoking and coffee consumption on blood resistin.

Animal studies have shown that oral administration of green tea catechins increased plasma adiponectin levels [42] and improved hypoadiponectinemia [43] and that dietary green tea catechins decreased plasma leptin levels [44]. We thus expected that green tea consumption would be favorably associated with these two adipokines. However, we found no association of green tea consumption with either serum adiponectin or leptin. The null finding for adiponectin is consistent with two previous epidemiological studies reporting no association between green tea consumption and serum adiponectin [19, 25]. Clinical trials also demonstrated that consumptions of green tea and its extract did not change circulating adiponectin [28, 45, 46], neither did they affect plasma leptin [28]. In addition, we observed no association of green tea consumption with resistin, visfatin or PAI-1. In the present study, among a population with relatively high green tea consumption, we confirmed that habitual consumption of this beverage was not associated with blood adipokines. These findings do not support the hypothesis that green tea consumption may play a significant role in determining circulating levels of adipokines. But our study does not deny that green tea may decrease disease risk through mechanisms other than those related to adipokines.

There are several possible mechanisms underlying the association of coffee and caffeine consumption with serum leptin and PAI-1. Leptin is a hormone exclusively secreted by adipocytes, with 80 % total of leptin production come from subcutaneous fat [47]. Coffee is rich in caffeic acid and chlorogenic acid [1], which have been shown to decrease body fat mass and plasma leptin in mice [48]. Similarly, chronic caffeine consumption has been shown to decrease adipose pads weight and the number of adipocytes in rodents [49], thereby decreasing serum leptin concentrations [50]. In addition, regular coffee consumption has been inversely associated with low-grade inflammation [51] and markers of oxidative stress [18], both of which are related to increased leptin [10]. PAI-1 is produced by adipocytes [52], and its increase has been identified as an early marker of endothelial dysfunction [53]. An adiposity-decreasing effect of caffeinated coffee consumption as mentioned above and a beneficial role of decaffeinated coffee consumption in improved endothelial function [51, 54] may be contributed to lowering PAI-1 levels.

Major strengths of the present study include a high response rate (94 %), use of a validated dietary questionnaire, uniform measurements of serum adipokines and adjustment for important confounding factors. Several limitations deserve mention. First, the temporal relationship between coffee or caffeine consumption and adipokines cannot be established by a cross-sectional study. However, our study was conducted among an apparently healthy population, and we excluded underlying chronic diseases including cardiovascular disease, cancer and diabetes. Further, subjects were not aware of their serum adipokine levels when answering the questionnaire. Thus, it is unlikely that serum adipokine status could influence beverage consumption. Second, although coffee and green tea consumption was assessed using a validated, self-administered questionnaire showing a high correlation as compared to 16-d dietary records [33], the real amount of coffee and green tea consumed may be over- or underestimated. In addition, decaffeinated coffee and caffeinated coffee were not distinguished and coffee preparation methods were not specified; however, it is conceivable that decaffeinated or boiled coffee was rarely consumed in Japan. Third, a single-point measurement of serum adipokines may be subject to within-person variation and prone to the random measurement error. Nevertheless, there was a moderate to high level of intra-individual stability and a low level of seasonal variability of plasma adiponectin, leptin, resistin and PAI-1 [55]. Furthermore, serum visfatin and PAI-1 have shown relatively high CVs (19 and 21 %, respectively) among all of studied adipokines, but these values remain within or are only slightly higher than the desirable range (<20 %) of CVs in reproducibility studies [56]. Fourth, our relatively small sample size may have been underpowered to detect a moderate association, and a limited number of heavy coffee drinkers did not allow us to examine the association at a high level of coffee consumption. Fifth, despite adjustment for important potential confounders, the possibility of residual confounding cannot be ruled out. Finally, the present findings among a working population might not represent the entire adult population.

In conclusion, higher consumption of coffee and caffeine was associated with lower leptin and PAI-1 concentrations (both men and women) and with higher adiponectin concentrations (men only). The protective association of coffee intake with risk of T2D and cancer observed in epidemiologic studies may be partly through the modulation of these adipokines. Prospective investigations are required to confirm the present cross-sectional association, and mechanistic studies should clarify the underlying mechanism behind the association.