A Weight-Loss Diet Including Coffee-Derived Mannooligosaccharides Enhances Adipose Tissue Loss in Overweight Men but Not Women

Abstract

Mannooligosaccharides (MOS), extracted from coffee, have been shown to promote a decrease in body fat when consumed as part of free-living, weight-maintaining diets. Our objective was to determine if MOS consumption (4 g/day), in conjunction with a weight-loss diet, would lead to greater reductions in adipose tissue compartments than placebo. We conducted a double-blind, placebo-controlled weight-loss study in which 60 overweight men and women consumed study beverages and received weekly group counseling for 12 weeks. Weight and blood pressure were measured weekly, and adipose tissue distribution was assessed at baseline and at end point using magnetic resonance imaging. A total of 54 subjects completed the study. Men consuming the MOS beverage had greater loss of body weight than men consuming the Placebo beverage (−6.0 ± 0.6% vs. −2.3 ± 0.5%, respectively, P < 0.05). Men consuming the MOS beverage also had reductions in total body volume (P < 0.0001), total (P < 0.0001), subcutaneous (P < 0.0001), and visceral (P < 0.05) adipose tissue that were greater than changes observed in those consuming the Placebo beverage. In women, changes in body weight and adipose tissue compartments were not different between groups. Adding coffee-derived MOS to a weight-loss diet enhanced both weight and adipose tissue losses in men, suggesting a potential functional use of MOS for weight management and improvement in adipose tissue distribution. More studies are needed to investigate the apparent gender difference in response to MOS consumption.

INTRODUCTION

With the high prevalence of obesity in much of the world, identifying functional ingredients that may influence weight reduction is of increasing interest. The risk for all-cause mortality increases across the spectrum of overweight to obese (1), and moderate weight reduction (≥10% initial weight) and maintenance have been recommended to improve long-term health (2). Body composition provides dimension to the obesity paradox, with the understanding that specific fat compartments are differentially associated with metabolic risk (3,4). It has been suggested that even small reductions in visceral adipose tissue may lead to improved lipid profiles, blood pressure, and insulin sensitivity (5). Therefore, coupling weight-loss dietary strategies with bioactive compounds proposed to modulate body fat distribution may lead to greater improvements in metabolic health than with moderate weight reduction alone.

A large body of research has examined the effects of various functional foods on overall body composition and selective fat depots. For example, green tea catechins have been proposed as therapeutic agents for body fat reduction (6,7). The proposed mechanism by which catechins are thought to improve body composition is via increases in 24-h energy expenditure and promotion of fat oxidation. Similarly, medium chain triglycerides may also have antiobesity effects, through promoting fat oxidation and thermogenesis (8,9). Their intake as part of a weight-loss program also improves weight loss and adipose tissue distribution (10).

Coffee-derived mannooligosaccharides (MOS), with a degree of polymerization distribution of 2-6 (DP 2-6), have been shown to function as bioactive agents in several studies conducted in the United Kingdom and Japan (11–15). Studies carried out in humans have found that MOS favorably alters body composition by decreasing both visceral and subcutaneous fat (11,12,16,17). One proposed mechanism for body fat reduction with MOS consumption is increased fat excretion. Several studies have shown that intake of 1–3 g/day of MOS can increase laxation frequency (15,18–20), as well as total fecal fat content (14,21). Further, in vitro (22) and animal studies (23,24) have indicated that MOS intestinal fermentation may increase the production of short-chain fatty acids (SCFA) which may have an impact on lipid metabolism and subsequently improve insulin sensitivity. However, effects of MOS intake on thermogenesis and beta oxidation have not been examined.

We recently showed that MOS consumption improved body composition in men, when consumed as part of a weight-maintaining diet (25). The objective of this study was to determine whether consuming a MOS-containing beverage in conjunction with dietary weight-loss counseling could enhance weight and adipose tissue reduction in healthy, overweight American men and women, compared to dietary weight-loss counseling alone. We hypothesized that the consumption of coffee-derived MOS would lead to greater reductions in adipose tissue compartments relative to Placebo.

METHODS AND PROCEDURES

Subjects and study design

Overweight and obese adults, ages 19–65 years with a BMI of 27–33 kg/m², were recruited through local newspaper advertisements and posted announcements around St. Luke’s/Roosevelt Hospital Center and Columbia University (New York, NY). All subjects were required to be weight stable for at least 3 months prior to enrollment in the study. Potential participants were excluded if they had been diagnosed with diabetes, hypertension, or were on medications known to affect body weight, plasma lipids, and blood pressure. Additional exclusion criteria included an abnormal score on the Brief Symptom Inventory questionnaire (26), a maximal body width (46 cm) at either the hips or shoulders (due to size restrictions of the magnetic resonance imaging (MRI) scanner bore), history of eating disorders, and alcohol or substance abuse. Subjects were given the opportunity to discuss the study protocol prior to signing an informed consent. The study was approved by the St. Luke’s/Roosevelt Hospital Center institutional review board.

Upon enrollment, subjects were randomly assigned (using a random digit table) to consume one of two instant coffee beverages, MOS or Placebo, during the 12-week weight-loss study. Approximately 75% of the subjects enrolled (n = 47) had participated in a previous MOS study (25) in which they consumed the same dose of MOS or Placebo as part of their free-living, weight maintenance diet. For those subjects, the weight-loss study was conducted 7–10 weeks (average 8.4 ± 0.1 week) after completion of the previous supplementation study to ensure that there was no residual effect of previous MOS or Placebo consumption. A study of the effects of whole grains on SCFA production used 7-day washout periods between 8 different test days and found significant differences between test meals on SCFA the following morning (27). Subjects who participated in the previous supplementation study were re-randomized to the study beverage. The beverages were assigned study-specific three-digit beverage codes by the supplier to ensure blinding of subjects and investigators. Thirteen subjects had never participated in a MOS study.

Dietary counseling

Group dietary counseling was conducted weekly (60-min sessions) by the same registered dietitian. At baseline, each subject’s resting metabolic rate (RMR) was determined using the Robertson and Reid equation as suggested by Heshka et al. (28), and total energy expenditure was calculated as RMR/0.7. The weight-loss diet involved moderate calorie restriction (−500 kcal/day) from calculated total energy expenditure and emphasized the consumption of low-calorie foods such as fruits and vegetables and reduction of high-fat, energy-dense foods. The dietitian instructed patients on weight-loss strategies through the use of the LEARN Manual for Weight Management (29). Adherence to study protocol and beverage consumption was reinforced during each counseling session.

Beverage consumption

The Placebo beverage was a commercially available instant coffee (Kraft Foods, Banbury, UK) containing trace amounts of MOS. The MOS beverage was identical, except that it contained an additional 2 g of coffee-derived MOS per serving, for a total MOS intake of 4 g/day (details on MOS extraction and preparation and study beverages are provided elsewhere (25)). At screening, subjects were exposed to samples of both beverages to ensure that there were no subjective differences with regard to palatability or tolerance. Participants were instructed to empty an entire sachet into a mug, add hot water, and mix thoroughly. Subjects were required to consume their study beverage twice daily, with meals. Subjects were provided a noncaloric sweetener (Splenda, McNeil Nutritionals, Ft. Washington, PA) and instructed not to add anything else to the beverage such as milk or other sweeteners. Beverages were dispensed at the weekly counseling session. At the first visit, subjects were given a 2-week supply of beverages to ensure they would have sufficient beverages in the event they missed a weekly counseling session.

Measurements

Anthropometrics

At screening, body weight was measured to the nearest 0.1 kg and height to the nearest 0.1 cm using a calibrated scale (Tanita BWB-800A Class III scale, Tanita Corporation, Arlington Heights, IL) and stadiometer (Detecto Industrial Scales of New York, NY), respectively. Waist and hip circumferences were obtained at baseline, week 6, and week 12. Waist circumference was measured at the level of the umbilicus (30), and hip circumference was measured at the level of widest circumference over the buttocks. Waist and hip measurements were taken by the same investigator.

Magnetic resonance imaging

Whole-body MRI scans were performed as previously reported by Gallagher et al. (31) to assess total and regional adiposity at baseline and end point. Each subject’s pre- and post-MRI scans were segmented and analyzed by the same trained technician using image analysis software (Tomovision, Montreal, QC, Canada) at the New York Obesity Research Center Image Reading Center (New York, NY). The analytical procedures are reported in more detail elsewhere (25).

Subject diaries and questionnaires

During the study, participants were asked to fill out daily diaries to record product consumption, medication use, and physical activity. Participants were asked to maintain their baseline level of physical activity throughout the 12-week study period. Physical activity was recorded on weekly logs.

Statistical analyses

The data analyses included anthropometric measurements and MRI data. Our analyses of anthropometric measurements included data collected at weeks 0, 6, and 12. For subjects missing week 6 measurements, weeks 5 and 7 data were averaged and used as the midpoint value. Similarly, week 11 was used as an end point measurement for subjects missing week 12 data. All data were analyzed on a completers’ basis. Completers of the study were defined as those who completed all assessments and physical measurements through week 11 of the study and had end point MRI data.

Mixed measures ANOVA was used to analyze the data. Using ANOVA, we tested the effects of time, beverage, and beverage × time interactions on body composition and diastolic and systolic blood pressure. Age, gender, and race were also entered in the model to control for any demographic effects. Three-way interactions with race, beverage, and time as well as sex, beverage, and time were also tested. The three-way interaction between race, beverage, and time was not significant. In the model assessing the effect of our intervention on body weight, race alone was not significant; therefore, this variable was removed from the models. This model included baseline body weight, age, beverage, time, and beverage × time and sex × beverage × time interactions. Because the three-way interaction between sex, beverage, and time was significant for body weight, statistical analyses for body weight and body composition variables were conducted on the whole study sample as well as separately by sex. Separate mixed effects analysis of variance was performed using BY statement for sex in SAS for Windows (version 9.1, SAS Institute, Cary, NC). Time, beverage, and the time × beverage interaction always remained in the model since these were our main variables of interest.

In models assessing the effect of our intervention on body composition variables assessed by MRI baseline value for the dependent variable and total body volume (no lungs) at each time point were included. The models for all participants included total body volume (no lungs), baseline value for the dependent variable, sex, time, beverage, and beverage × time and sex × beverage × time interactions. The models by sex included total body volume (no lungs), baseline value for the dependent variable, time, beverage, and the beverage × time interaction. Waist circumference data were assessed similarly except that body weight at each time point was included in the model. Pearson correlations were done to examine the correlation between changes in body composition during the weight maintenance study and those during the present weight-loss study for those participants who completed both studies and consumed the MOS beverage for both studies.

Statistical analyses were performed using SAS. Statistical significance was set at P < 0.05. Data are presented as means ± s.e.m.

Results

A total of 60 men and women started the weight-loss study. Of the total sample, six participants were lost to follow-up and 54 completed the 12-week study. However, only 49 subjects were included in the MRI analysis (MOS, n = 22; Placebo, n = 27). Two women were not included in the MRI analysis because they had missing data at baseline, but were included in the secondary outcome analyses, and three other participants (two men, one woman) were excluded from all analyses for failure to disclose the presence of exclusionary behaviors at baseline. This decision was made prior to data analysis and unblinding of the study investigators. Table 1 shows the characteristics of the 51 subjects included in the secondary outcome analyses. There were no differences between groups with regard to baseline characteristics or withdrawal of study participation, and no adverse events were reported.

Table 1.

Baseline characteristics of subjects

Characteristic	Placebo		MOS
	All subjects (n = 35)	Completers (n = 29)	All subjects (n = 25)	Completers (n = 22)
Age (y)	46.1 ± 1.9	47.9 ± 1.9	42.6 ± 2.5	42.9 ± 2.8
Sex (female/male)	20/15	18/11	16/9	15/7
Height (cm)	168.3 ± 1.6	168.0 ± 1.9	165.9 ± 2.2	165.0 ± 2.4
Weight (kg)	84.9 ± 1.9	84.8 ± 2.1	83.6 ± 2.5	82.7 ± 2.8
BMI (kg/m2)	29.9 ± 0.4	30.0 ± 0.4	30.2 ± 0.4	30.3 ± 0.5
Ethnicity (W/AA/H/O)	15/13/4/3	13/9/4/3	11/7/6/1	10/6/5/4

AA, African American; W, white; H, Hispanic; O, other (includes Asians, multiple ethnicity, East Indians).

Anthropometric data

There was a sex × beverage × time interaction on body weight (P = 0.033) as well as a beverage × time interaction (P = 0.012). When data were analyzed separately by sex, there was a significant beverage × time interaction in men (P < 0.0001) but not in women. Data, adjusted for baseline body weight and age, showed that men who consumed the MOS beverage as part of their weight-loss program had a 5.6 ± 0.5 kg reduction in body weight compared to 2.9 ± 0.4 kg for men consuming the Placebo. Women consuming the MOS and Placebo beverages had comparable weight loss: −3.4 ± 0.5 kg and −3.3 ± 0.5 kg, respectively.

Due to the known issues related to missing data in randomized clinical studies, we have done permutation tests for anthropometric variables, as part of a sensitivity analysis. The P values from the permutations test were similar or less than the observed P values, which is an evidence for the robustness of our analysis. The permutation P value for body weight was 0.048 (whole data) and <0.0001 (for men).

Whole-body MRI analyses

There was no significant sex × beverage × time interaction on total body volume and visceral adipose tissue compartments assessed by MRI but the three-way interaction was significant for subcutaneous adipose tissue (P = 0.0049) and total adipose tissue (P = 0.007). In men and women together, there was a significant beverage × time interaction on total body volume (P = 0.0257), total adipose tissue (P = 0.0035), subcutaneous adipose tissue (P = 0.0048), visceral adipose tissue (P = 0.0030), and intermuscular adipose tissue (P = 0.0143). Subjects consuming the MOS beverage tended to have greater reductions in each of the body compartments analyzed compared to subjects consuming the Placebo beverage. When analyses were done separately by sex, the time × beverage interaction was not significant for total body volume (P = 0.37), total adipose tissue (P = 0.17), and subcutaneous adipose tissue (P = 0.312) but there was a trend for visceral adipose tissue (P = 0.078) whereas the interaction was significant in all body compartments in men (all P < 0.01).

Because of the sex effect in the interaction between beverage and time in the anthropometric data, MRI data were also analyzed separately for men and women (Table 2). For men, there was a significant time × beverage interaction with greater reductions in total body volume (P = 0.005), total adipose tissue (P < 0.001), subcutaneous adipose tissue (P < 0.001), and visceral adipose tissue (P = 0.009; after adjustment for baseline values and total body volume) with MOS consumption compared to Placebo. In women, we found a beverage × time interaction indicating a reduction in intermuscular adipose tissue (P = 0.021) and a trend for slightly greater decrease in visceral adipose tissue (P = 0.079) with MOS consumption. For women, the change in muscle volume was −0.16 ± 0.15 L with MOS (P = 0.288, adjusted P = 0.703) and −0.51 ± 0.15 l with Placebo (P = 0.0017, adjusted P = 0.0089) and for men, the change was −1.18 ± 0.34 L with MOS (P = 0.0027, adjusted P = 0.0149) and −0.84 ± 0.26 L with Placebo (P = 0.0054, adjusted P = 0.0248).

Table 2.

MRI adipose tissue distribution (by gender) at baseline and after 12 weeks of a weight-loss diet with a MOS-containing coffee beverage or placebo coffee beverage

Coffee ID	Gender	Total adipose tissue		Subcutaneous adipose tissue		Visceral adipose tissue		Intermuscular adipose tissue		Total body volume (no lungs)
		Baseline	12 weeks	Baseline	12 weeks	Baseline	12 weeks	Baseline	12 weeks	Baseline	12 weeks
Placebo	Female	35.33 ± 0.43	34.69 ± 0.42	31.55 ± 0.36	30.49 ± 0.36	2.59 ± 0.07	2.65 ± 0.07**	1.31 ± 0.04	1.37 ± 0.04**	71.84 ± 0.57	69.68 ± 0.57
MOS	Female	35.46 ± 0.44	33.92 ± 0.45	31.88 ± 0.36	30.22 ± 0.37	2.51 ± 0.07	2.32 ± 0.07	1.32 ± 0.04	1.19 ± 0.04	72.17 ± 0.59	69.08 ± 0.59
Placebo	Male	29.87 ± 0.29	29.56 ± 0.29***	24.21 ± 0.24	23.80 ± 0.24***	4.09 ± 0.09	4.22 ± 0.09***	1.56 ± 0.03	1.53 ± 0.03	83.07 ± 0.54	81.31 ± 0.54**
MOS	Male	29.79 ± 0.37	26.44 ± 0.37	24.21 ± 0.30	21.43 ± 0.31	4.04 ± 0.11	3.58 ± 0.12	1.52 ± 0.04	1.46 ± 0.03	83.05 ± 0.67	77.81 ± 0.67

Data are means ± s.e.m., in liters (l), adjusted for baseline value and total body volume. Women: Placebo, n = 16; MOS, n = 15. Men: Placebo, n = 11; MOS, n = 7. Significantly different from end point MOS.

^*P < 0.05,

^**P < 0.005,

^***P < 0.0001.

We found a significant sex × beverage-by-time interaction on waist circumference, with women showing a trend (P = 0.059) for greater reduction in waist circumference with consumption of the Placebo beverage (−3.8 ± 1.0 cm) compared to the MOS beverage (−0.1 ± 1.1 cm). Although not significant, the reverse was found when data were analyzed with men alone: −3.6 ± 1.6 cm MOS and −0.3 ± 1.0 cm Placebo (beverage × time interaction, P = 0.249).

Of those individuals who also participated in our previous weight maintenance study, seven were assigned to MOS group for both studies and nine were in the Placebo group for both studies. The rest consumed different test products during the weight maintenance and the weight-loss study. When we looked at the correlation between percent change in various body composition variables in those who were in the MOS group for both studies, we found correlation coefficients of 0.48 for percent change in body weight, 0.59 for subcutaneous adipose tissue, 0.30 for visceral adipose tissue, 0.62 for total adipose tissue, and 0.52 for total body volume.

Discussion

This is the first study to show that incorporating coffee-derived MOS in a weight-loss diet can enhance adipose tissue loss in overweight individuals. In addition, our data are among the first to show the effects of MOS consumption on body composition in an American population. Other randomized controlled studies investigating the effects of similarly derived MOS on body composition have mostly been conducted in overweight and obese Japanese subjects consuming weight-maintaining diets (11,12,16,17) with the exception of our previous weight maintenance study in an American population (25). Moreover, our data provide additional information regarding a possible sex effect of MOS consumption on body composition since our anthropometric data suggest greater benefits of MOS consumption in men compared to women.

Our data, showing a slight improvement in weight loss with MOS consumption, are similar to those observed by Kumao et al. (16) over 12 weeks of MOS consumption within a weight maintenance diet. However, the major purported benefit of MOS consumption is an improvement in body composition with reductions in visceral adipose tissue. The magnitude of the differences in percent change in visceral adipose tissue between the MOS and the Placebo groups in our study is similar (11,16,17) or greater (12,25) than those observed in other free-living, but not weight-loss, studies. However, except for our previous study (25), all other studies used computed tomography to assess visceral fat area, rather than total visceral adipose tissue.

Previous studies have not reported on data separately by sex; therefore, we cannot determine if the gender differences suggested by our data have been observed by others. In our previous weight maintenance study, we found that MOS consumption improved body composition in men but not women (25). Because of the small number of men in the present study (n = 18), we cannot fully ascertain a sex difference in response to MOS consumption. However, similar observations have been made with other bioactive ingredients, such as medium chain triglycerides (32) and green tea extracts (7,33). There are also data to support a sex difference in basal fat metabolism (34). Women store fat more efficiently than men (35), and men have a higher rate of fat oxidation (36,37). These findings would suggest that women may be more resistant to fat store mobilization than men, and as a result, may not be as responsive to a thermogenic agent as men. If we assume that MOS functions by promoting fat oxidation, then these differences may explain the sex differences in body composition changes observed here.

MOS may be involved in energy balance regulation and body composition changes via effects on SCFA. Studies have suggested that MOS consumption may increase intestinal SCFA production (22–24). It has been shown that incubating adipose tissue with propionic acid increases leptin mRNA and protein levels and down-regulates resistin mRNA expression (38). In C57BL/6J mice fed a high-fat diet, butyrate consumption increased energy expenditure and reduced weight gain and fat mass, preventing diet-induced obesity (39). These studies suggest that SCFA can influence energy balance via effects on adipokines and/or energy expenditure, which may explain the role of MOS in weight management.

The major strength of this study is its double-blind, randomized, Placebo-controlled design. There was no detectible difference in color or palatability of beverages and neither the investigators nor the subjects knew which coffee beverage contained MOS, preventing any bias during the dietary counseling sessions or data analyses. The use of MRI scans provided precise measurements of adipose tissue distribution on a wholebody level, something that cannot be achieved with other imaging methods. Finally, we had a very low drop-out rate in this study, ~10%, and excellent compliance with study beverage consumption.

Although our drop-out rate was low, our sample size was small when data were analyzed separately by sex. Our final sample size was considered to be adequate based on our initial calculations for the whole group. However, we had not anticipated sex differences in responses to MOS consumption and the need to analyze our data separately for men and women. Future studies should power on sex separately to obtain an adequate sample size. In addition, our data were presented without adjusting for multiple comparisons. However, we had few preplanned comparisons and making such adjustments would not change the conclusions of this study. Also, a substantial number of participants in this study had been involved in a previous weight maintenance study with MOS. However, the weight maintenance study was double-blind and unblinding for that study did not occur until after all participants had been enrolled in the weight-loss study. In addition, study products were re-coded and participants re-randomized to study product. As a result, participants and investigators were unaware of the assigned test product for both studies. Also, the study participants who were involved in the weight maintenance study of MOS underwent a washout period of ~8 weeks. It is highly unlikely that the metabolic effects of MOS would be maintained over that period of time without sustained consumption. Nevertheless, one cannot exclude the possibility of residual confounding due to prior exposure to MOS. Because of the limited amount of research on MOS as a bioactive agent in adipose tissue reduction, there is real potential for future studies to advance our knowledge in this field. First, future research should further explore the sex differences in the body composition changes with MOS consumption. Second, potential mechanisms of action for MOS should be thoroughly examined. Inclusion of fecal samples as well as measures of SCFA production and lipogenesis in conjunction with body composition data would be beneficial. Third, studying the thermogenic effects of MOS through measurements of resting and postprandial energy expenditure could help determine whether this food component works through increasing metabolic rate. Finally, studies could be conducted to understand potential ethnic differences in the responsiveness to MOS.

The data shown here begin to describe the effects of MOS on body composition when consumed as a component of a weight-loss diet and extend previous findings in other population groups. Bioactive ingredients that support weight loss could be very beneficial in the fight against obesity. However, we need to better understand how sex may influence responsiveness to weight-loss adjuncts before conclusions can be drawn regarding the efficacy of any potential ingredient.