CM3, a highly cross-linked cellulose in capsule form, expands in the stomach to a size several fold of its original volume. It is purported to induce a prolonged feeling of satiation and a delay in gastric emptying, thus promoting weight loss. We examined whether CM3 delays gastric emptying (using the stable isotope 13C-octanoic breath test) and whether it influences subjective feelings of appetite sensations (using visual analog scales, VASs). We performed a double-blind randomized placebo-controlled crossover trial in 19 moderately obese but otherwise healthy subjects (mean age 55 ± 9 years, BMI 31.1 ± 4.6 kg/m2). The subjects were treated with six capsules of CM3 or matching placebo 30 min before a standardized solid meal. Breath collection and VASs were performed over 4 h every 15 min and 30 min, respectively. Half-excretion time of 13CO2 in breath, indicating gastric emptying half time, was the primary outcome parameter. The study was powered to detect a change in gastric emptying of 20–30 min. Mean 13CO2 half-excretion time changed from 2.3 ± 0.4 to 2.4 ± 0.33 h (mean difference +6 min, 95% confidence interval (CI) −3 to +15 min; P = 0.17). Appetite sensations (hunger, satiation, fullness, prospective food consumption, desire to eat something sweet, salty, savory, or fatty) changed over time during the course of the postprandial phase but were not influenced by CM3 (repeated measures ANOVA). In obese subjects, acute administration of the weight-loss supplement CM3 does not delay gastric emptying and does not influence subjective appetite sensations.
With prevalence approaching one-third of the population, obesity is reaching epidemic proportions in the United States and other industrialized nations (1,2). In the United States, morbid obesity (BMI ≥ 40 kg/m2) affects >15 million Americans and causes an estimated 300,000 deaths per year (3). Dietary and lifestyle modifications, the basis of weight-loss interventions, are often ineffective (4). Moreover, drug treatment of obesity is limited in effect size, is often associated with adverse drug reactions, and has not been shown to be safe and effective in terms of clinical outcomes in long-term investigations (5,6,7,8,9,10). Therefore, there is a distinct need (and market) for alternative measures to support weight loss. These include a broad range of options—among them means to decrease gastric emptying and to increase the feelings of satiation, such as dietary fibers.
CM3 is a weight-loss supplement manufactured in Germany and available worldwide, which is sold in pharmacies over-the-counter (nonprescription, nonreimbursable) and via the Internet. It is advertised as of “herbal” origin, because it is derived from cotton wool and bark. CM3 contains highly cross-linked cellulose, compressed in small blocks (“comprimate”) and administered in the form of hard gelatin capsules, which, after dissolution of the capsule in the stomach, expand into a soft sponge-like cuboid. The recommended dose is two to three capsules taken 30 min before major meals. The manufacturer claims that CM3 stays for ∼6–8 h in the stomach where it stimulates “satiation sensors,” therefore causing a prolonged feeling of satiation, the sensation that arises during meal ingestion that contributes to meal termination, so that weight loss will “occur naturally.” There is only one randomized trial showing weight loss of 3–4 kg more than placebo (11) but high-quality confirmatory randomized trials showing the effectiveness of the product on body weight are lacking. Moreover, there are no published studies examining the postulated mechanisms of action of the product, namely its effect on gastric emptying and feelings of satiation.
The objectives of the present study were therefore to investigate whether CM3 (i) delays gastric emptying and (ii) affects subjective appetite sensations, such as hunger, satiation, and fullness, in moderately obese but otherwise healthy subjects. To investigate gastric emptying the 13C-octanoic acid breath test was used, an established, reliable and safe marker of gastric emptying (12). To examine whether CM3 affects gastric sensory functions, we used visual analog scales (VASs) ratings, which record subjective sensations, such as hunger, satiation, fullness, prospective food consumption, and the desire to eat something fatty, salty, sweet, or savory. VASs have been established as a reliable method in nutrition research for the assessment of appetite sensations both in younger and older subjects (13,14,15).
Methods and Procedures
Study design and 13C-octanoic acid breath test
This trial was a mono-center placebo-controlled, randomized, double-blind crossover study in the outpatient metabolic unit of a large German Heart Center at Rotenburg an der Fulda. After a screening visit, the subjects underwent the breath test protocol at two occasions, 1 week apart, according to a predetermined randomization crossover schedule. After an overnight fast of >10 h, they were treated once with six capsules of CM3 or six matching placebo capsules in a double-blinded fashion. The capsules were ingested 30 min before the test meal with 400 ml of water. Another 200 ml of water were ingested 20 min before the test meal. The test meal consisted of a pancake containing 200 mg 13C-octanoic acid (Eurisotop, St. Aubin, France). The pancake was prepared mixing water with a prepacked powdered premix (Gastro-Pan, Ovofood NV, Herk-de-Stad, Belgium, Lot No. B5702EG) and baking the resulting batter into a solid meal under standardized conditions. The composition of the pancake was as follows: 8.6 g egg yolk, 3.75 g egg white, 17 g wheat flour, 3 g milk powder, 5 g sunflower oil, and 5 g beet sugar, giving a total of 225 kcal. The macronutrient composition of the test meal was ∼48% of calories from carbohydrate, 37% from fat, and 15% from protein. The meal was ingested within <5 min together with 150 ml of water. Breath samples for 13CO2 measurements were collected into 10-ml tubes (Exetainer; Labco, Buckinghamshire, UK) at baseline (two samples for pooled baseline) and at 15-min intervals thereafter for a total of 4 h. No more food or drinks were allowed until the end of the study period. Ghoos et al. (12) were the first to report the technique for measuring gastric emptying based on the use of 13C-octanoic acid, a medium chain fatty acid that is rapidly absorbed in the duodenum and metabolized in the liver (16). After oxidation, the resulting 13CO2 is excreted into breath at a level which can be measured by isotope ratio mass spectrometry (IRMS). Venous blood was drawn for plasma glucose measurements at baseline and 2 h after the test meal.
The trial product (CM3 Kapseln, batch no. 114101; A&G Lifescience GmbH, Cologne, Germany) was purchased in a public pharmacy. In order to produce the placebo, hard gelatin capsules were opened, the cellulose comprimate removed, and the capsule resealed. Swallowing of the capsules was done swiftly under supervision to ensure that the trial remained blinded. The hard gelatin capsules have a light blue appearance and measure 23 mm in length and 8 mm in diameter. Each capsule contains a cylindrical block of cellulose (18 × 7 × 5 mm3). When put in water, the comprimate expands to a size of ∼18 × 7 × 33 mm3 (4.1 ml volume per capsule, 24.9 ml per study dose of six capsules).
Measurement of appetite sensations
In order to assess subjective appetite sensations, VASs were used as described by Flint et al. (13) VASs were composed of lines of 100-mm length with words anchored at each end, describing the extremes (i.e., “I have never been more hungry”/“I am not hungry at all”). Subjects were asked to make a mark across the line corresponding to their feelings. Quantification of the measurement was done by measuring the distance from the left end of the line to the mark. VASs were used to record eight qualities of appetite, namely hunger, satiety, fullness, prospective food consumption, desire to eat something sweet, salty, savory (hearty), and fatty. VASs were filled in at baseline (40 and 10 min before the test meal for pooled baseline), and at 30-min intervals over the 4-h period of the study.
Overweight or obese adult subjects, otherwise healthy, were recruited from the community by word of mouth. Inclusion criteria were a BMI >25 kg/m2 and the ability to understand and comply with the study protocol. Female volunteers, when not after menopause, had a pregnancy test, which had to be negative. Major exclusion criteria were: current dieting, severe gastrointestinal diseases, abdominal surgery (except for appendectomy), metabolic diseases including diabetes mellitus, thyroid diseases or other endocrine diseases, liver or kidney dysfunction, severe cardiovascular diseases, rheumatologic, oncologic and psychiatric diseases, and use of any type of medication. The trial was performed according to the current version of the Declaration of Helsinki. The nature and risks of the experimental procedure were explained to the subjects, and written informed consent was obtained from all participants.
A total of 19 subjects participated in the study, 7 men and 12 women. The demographic characteristics of the study population and the variables measured at the screening visit are shown in Table 1. Mean BMI was 31.1 kg/m2. One subject had a BMI <25 kg/m2, seven subjects had a BMI between 25 and 30 kg/m2 (overweight), seven had a BMI between 30 and 35 kg/m2 (grade 1 obesity), three had a BMI between 35 and 40 kg/m2 (grade 2 obesity) and one subject had a BMI >40 kg/m2 (grade 3 obesity). Body composition was determined at baseline by bioimpedance analysis at 50 kHz using a Multi Frequency Analyzer (Data Input GmbH, Hofheim, Germany). All subjects had the wish to loose weight. Laboratory measurements at baseline included serum creatinine, liver function tests, amylase, lipase, fasting blood glucose, red and white blood count, and thrombocytes and were all in the normal range (data not shown). The aforementioned parameters were analyzed using routine methods in the Heart Center's core laboratory.
Table 1. Demographic characteristics of the study population and variables measured at the screening visit (n = 19; 7 men and 12 women)
Stable isotope 13C breath test data analysis
The 13C isotopic enrichment of the breath samples was measured using continuous-flow IRMS (ABCA 20-20; Europa Scientific, Crewe, UK). Enrichment analysis using IRMS results in a δ13 value, expressing the 13C enrichment of the sample. The δ value is defined in the formula δ13 = (Rs/Rr − 1) × 1,000, where Rs is the 13C/12C ratio of the CO2 in the sample and Rr is the 13C/12C ratio of the CO2 in the internationally agreed standard, Pee Dee belemnite. The δ value is then converted to percentage 13C recovery per hour of the amount administered in the test meal using previously published equations (17).
Breath test modeling
We used a linear model as described previously to analyze the breath 13CO2 enrichment data (18,19). Percentage-dose recovery over time data were fitted to Equation (1) and cumulative breath 13CO2 recovery data to Equation (2).
Figure 1 shows a representative example of one breath test. We used calculations initially described by Elashoff to determine gastric emptying parameters (20). Equation (1) is the first derivative of the power exponential formula modified by Siegel et al. (21) with a correction factor m. Overall gastric emptying is assessed by four parameters that are derived from these equations, (i) the 13CO2 half-excretion time (t1/2b; Eq. 3), (ii) the time of the maximal 13CO2 excretion rate (tmax; Eq. 4), (iii) parameter β, and (iv) parameter k. The latter two parameters allow the analysis of the initial or subsequent phases of gastric emptying, respectively. Larger values for β indicate an initially slower rise, whereas smaller values an initially faster one. Larger k values indicate a steeper rise of the cumulative excretion curve, whereas smaller values a more gradual one (21,22). The half-excretion time (t1/2b) was calculated using Equation (3), and time to maximal excretion (tmax) using Equation (4) (ref. 12). Parameter m describes the total cumulative 13C recovery when time is infinite. Cumulative 13CO2 excretion cannot reach 100% because of fixation of a substantial amount of the orally administered dose in the bicarbonate pool in the body. The gastric half-emptying time and lag phase, as measured by the breath test, are—by the nature of their mathematical definition—completely determined by the shape of the curve and therefore independent of the endogenous CO2 production. Curve fitting was performed using SAAM II (Simulation, Analysis, and Modeling Software for Kinetic Analysis), Numerical Module, Version 1.1.2 (SAAM Institute, University of Washington, Seattle, WA).
Half-excretion time (t1/2b) was chosen as the primary outcome parameter because it best describes the hypothesized delay in gastric emptying by CM3. The a priori statistical working hypothesis was that CM3 delays gastric emptying. Despite expecting a delay, we decided to test two-sided. Sample size calculation for the primary outcome parameter was calculated assuming an α of 0.05, a power of 0.8, and a s.d. of the difference of 0.5 h. Using the two-sided Student's t-test, to detect a mean change of 0.5 h would require 10 subjects; to detect a mean change of 0.33 h, 18 subjects would be required. Sample size calculation was performed using StudySize Version 1.0.9 (CreoStat HB, Västra Frölunda, Sweden). Secondary outcome parameters were the time of the maximal 13CO2 excretion rate (tmax), parameters β and k and the results of the VASs. Before the code was broken, all measurements and data evaluations were completed and a blinded review was conducted. Safety and tolerability were assessed for all subjects. Data are given as means ± s.d. unless otherwise indicated. Descriptive and explorative tests were used for the secondary outcome parameters.
VASs were analyzed using repeated measures ANOVA. Because sphericity could not be assumed throughout (Mauchly's test), we used Huynh-Feldt epsilon corrections to adjust the degrees of freedom. In order to assess which parameters predicted response to the dietary supplement, we performed post hoc multivariate analyses dividing the subjects into responders and nonresponders. We used SPSS Version 14.01.1 (SPSS, Chicago, IL).
Characteristics of the subjects
Demographic characteristics of the subjects as well as the main results of bioimpedance analysis can be seen in Table 1. All 19 subjects included in the study completed both protocols successfully without protocol violations. Tolerability of the interventional product as well as of the procedures was good.
In the placebo phase there was a trend toward a decrease in blood glucose 2 h postprandially from 82 ± 12 to 75 ± 15 mg/dl (mean difference −7 mg/dl, 95% confidence interval (CI) −14 to −0.3 mg/dl; P = 0.059). During the active intervention (six capsules of CM3), there were no differences between the baseline and the 2-h postprandial blood glucose concentrations; 82 ± 14 mg/dl vs. 78 ± 19 mg/dl (mean difference −4 mg/dl, 95% CI −9 to +1 mg/dl; P = 0.13). Repeated measures ANOVA indicated that glucose decreased significantly (P = 0.014) but treatment had no influence (P = 0.76). There was also no significant interaction between glucose and treatment (P = 0.50).
The results of the 13C-octanoic breath test are shown in Table 2. There was a small increase in 13CO2 half-excretion time (t1/2b), the primary outcome parameter, by CM3, which did not reach significance (mean difference +5.9 min, 95% CI −2.8 to +14.7 min, P = 0.17). There was also a small increase in the time of maximal 13CO2 excretion (tmax) (mean difference, +7.2 min, 95% CI −0.48 to 30 min; P = 0.065). The regression-estimated constant β also tended to increase (P = 0.088), indicating an initially slower rise of the curve. The regression-estimated constant k as well as the parameter m (total cumulative 13C recovery) remained unchanged. Figure 2 depicts the mean values (±s.d.) of 13CO2 breath curves. Figure 3 shows graphical illustrations of all parameters of the individual subjects.
Table 2. Comparison of 13C-octanoic breath test parameters between control (placebo) and CM3 treatment
Because in a negative study it is always relevant to assess the possibility of a type II error, we conducted a post hoc analysis based on the mean (0.098 h) and the s.d. of difference of 0.29 h between the two treatments. The effect size of ∼12 min in t1/2b that could have been detected with a power of 0.8 based on the s.d. actually observed shows that the sample size was sufficient to detect a clinically meaningful effect of the dietary intervention.
Multivariate regression analyses
We performed post hoc multivariate analyses to determine which parameters predicted response to the dietary supplement, dividing the subjects into the ones with a shorter or longer t1/2b, the latter ones considered “responders.” In a sex-adjusted logistic regression model we found waist-to-hip ratio to be a significant predictor. Waist-to-hip ratio was higher in responders compared to nonresponders (0.90 ± 0.07 vs. 0.84 ± 0.08; P = 0.0496). There was no significant association with sex, age, body weight, BMI, or fasting blood glucose. Sex-adjusted waist circumference tended to be significantly higher in responders (104 ± 10 cm vs. 92 ± 11 cm; P = 0.084). In another model including age, sex, waist-to-hip ratio, and fasting glucose, waist-to-hip ratio tended to remain a significant predictor (P = 0.067).
Tables 3 and 4 show the results of the VASs. Figure 4 shows the mean (±s.d.) VASs data over time. Repeated measures ANOVA revealed that CM3 had no effect on any of the eight parameters examined. Highly significant effects were seen when the influence of time on the parameters was analyzed, especially in the parameters hunger, satiation, fullness, prospective food consumption, and desire to eat something savory. The desire to eat something salty or fatty was also significantly different over time, whereas there was no effect on the desire to eat something sweet. Analyses involving sequence were mostly nonsignificant. The calculated areas under the curve (Table 4) were not different between placebo and CM3. Borderline significant between-subjects effects could be observed in the parameters desire to eat something sweet, salty, and fatty.
Table 3. Results of the anchored visual analog scales (time)
Table 4. Results of the anchored visual analog scales (area under the curve, AUC)
There was no association between bioimpedance analysis measurements and any of the primary or secondary outcome parameters (data not shown).
Obesity and overweight are currently affecting over 1.1 billion individuals worldwide and are associated with chronic morbidity and premature mortality (23,24). Achieving just a 5–10% loss of body weight is associated with improvements in cardiovascular risk profiles and reduced incidence of type 2 diabetes (24,25). The marked rise in the prevalence of obesity over the past two decades is best explained by behavioral and environmental changes; therefore control of food intake is of paramount importance for weight control. Feelings of hunger and satiety have long been associated with gastric motor and sensory functions and the role of the stomach in the control of food intake has been of interest for at least a century (26). The development and use of various devices and interventions to promote satiety and/or to decelerate gastric emptying has been receiving increased attention over the past decade among researchers and clinicians.
CM3, a plant biopolymer (β−1,4-glucose polymer) consisting of >10,000 cross-linked monomers, is marketed as a product supposedly to expand in the stomach to a size ∼15-fold of its original and is purported to induce weight loss. Purpose of the present study was to examine the suggested mechanisms of CM3 action, namely whether it delays gastric emptying and affects subjective appetite sensations. Our results show that CM3 has no effect on gastric emptying and on gastric sensory functions.
Gastric emptying in obesity has been reported as being more rapid (27,28), normal (29,30), or even slower (31,32) compared to normal-weight individuals. It has been shown, however, that in obese individuals there is a significant negative association between gastric emptying rates and fullness score (33), therefore suggesting that substances that would delay gastric emptying might decrease nutrient intake, and subsequently induce weight loss. Our data show that CM3 does not significantly alter gastric emptying rates. From the four parameters assessing gastric emptying only one, the time of maximal 13CO2 excretion rate, was borderline increased (mean difference +7.2 min, P = 0.065). The other three, namely the 13CO2 half-excretion time and parameters β and k remained unchanged. Regression-estimated constant β, however, tended to be increased (P = 0.088), which indicates a slower rise of the early 13CO2 excretion curve; this change tended to reach statistical significance but does not appear to be clinically relevant. Furthermore, it should be pointed out that there is an intra-individual coefficient of variation in gastric emptying of ∼13% (34,35), which is larger than the mean difference seen in the present trial. Mean half-excretion time of solids measured with the octanoic acid breath test is usually ∼120 min and we considered a change of >20–30 min to be clinically relevant—the study was accordingly powered. The results of our multivariate analyses point out that an increased waist-to-hip ratio could be a predictor for “response” to the supplement, but these findings are based on post hoc analyses and need to be confirmed in further studies.
It could be postulated, that the putative weight-lowering effect of CM3 is independent from its effects on gastric emptying rates and mediated through a CM3-induced increase in the feeling of satiation (either because of its fiber content or because of its volume). Under normal circumstances, the presence of gastric contents and subsequent distention stimulates mechanoreceptors in the musculature of the stomach wall and inhibits hunger sensations (32). This in turn induces satiation and reduces food intake. However, we found no effect of CM3 on hunger or any other appetite sensations. Is this finding surprising? Current recommendations suggest that adults consume 20–35 g of dietary fiber per day, depending on age and gender (36,37). It has been suggested that doubling the dietary fiber intake may increase satiation and induce a reduction in energy intake (38). However, there are only very few randomized controlled trials using fiber in order to achieve weight loss—with conflicting results. Two randomized controlled trials showed that ∼7 g/day of fiber can reduce body weight significantly more than placebo (39,40). Another trial compared fermentable and nonfermentable fiber at higher doses (∼27 g/day) and concluded that there is no role for short-term use (4 weeks) of either supplement in promoting weight loss (41). Salas-Salvado et al. recently showed in a 16-week randomized controlled trial that fiber supplements (8 and 12 g/day) do not induce weight loss greater than those that received placebo (42). Interestingly, regarding satiation, fiber supplements marginally increased in postlunch but not postdinner.
Another possible mechanism for an effect of CM3 could be through an increase in the feeling of satiation, because of a sheer volume effect. It has been reported that gastric distention induces feelings of fullness and satiation (26). In specific, chemoreceptors and mechanoreceptors in the stomach wall signal satiation through vagal and splanchnic nerves (43). However, CM3 at the doses given with an approximate total volume in the stomach of 25 ml could hardly cause such an effect, considering that the mean maximal tolerated volume in normal-weight subjects is 1 l and in obese individuals is up to 2 l (26,44). We were unable to confirm the manufacturer's claim that CM3 expands ∼15-fold—the products we used expanded only 6–7-fold.
Indirect effects on satiation through hormonal signaling cannot be excluded. We reasoned a difference in the 2-h response in blood glucose could be an indicator for such effects. However, glucose tended to be decreased to the same degree during placebo and CM3 and no significant effect of the treatment could be substantiated.
Our study has several strengths and weaknesses. A definite asset of the study is its crossover design which eliminates bias of interindividual differences in gastric emptying and other confounders. We used double of the manufacturer's recommended dosage to be certain not to miss a possible effect due to underdosing. Our sample size and power calculation renders this rigorously designed study clear conclusions in terms of clinical relevance. The method of the octanoic acid breath test is accepted standard for measuring gastric emptying. Although scintigraphy is usually considered to be the reference method for measuring gastric emptying in humans, several drawbacks (e.g., equipment, irradiation) limit its application in routine practice. On the other hand, breath tests correlate very well with scintigraphy (35,45,46,47,48). We used a formula method to determine the outcome parameters that reflects the biphasic nature of gastric emptying and allows calculation of the lag phase and rate of emptying for both solids and liquids (20,21). As a possible weakness of the study that cannot be excluded is the effects of CM3 on gastric emptying occur only after chronic administration and were thus not observed after acute single administration. In specific, it remains possible that other processes that contribute to food intake could be influenced by the fiber supplement. For example, a second meal or ileal-break effect may occur, in which case desire to eat at the next and subsequent gastric emptying may have been influenced. However, it has never been demonstrated in vivo that CM3 has any physiologic effects in humans.
Safety and tolerability of the acute administration of the product were excellent. It should be mentioned however, that in the literature there is one report of CM3 causing partial intestinal obstruction requiring surgery in a female patient after taking four CM3 capsules 2 weeks earlier (49) and there are several reports in spontaneous adverse reaction databases (50).
In conclusion, the present study shows that CM3, marketed as a weight-loss supplement, does not delay gastric emptying and does not increase subjective feelings of satiety, fullness or any other appetite qualities, associated with food intake. Randomized controlled trials with weight loss as an endpoint are needed in order to examine whether CM3 might induce weight loss through gastric motor and sensory function-independent pathways.
This trial was funded by the Institute for Clinical Research, Center for Cardiovascular Diseases, Rotenburg an der Fulda, Germany. The manufacturer of the trial product had been asked for financial support but declined. IRMS for breath 13CO2 enrichment was done at the Department of Medicine, Division of Gastroenterology and Gastrointestinal Research Centre, University Hospital Gasthuisberg, Leuven, Belgium. We gratefully acknowledge the contribution of Dr Geypens in supervising these measurements. The authors' responsibilities were as follows—H.K.B. conceived of and designed the study, directed and supervised the clinical performance of the study, performed the statistical analyses, and wrote the manuscript; S.U. and R.D. recruited the participants, conducted the clinical and procedurals parts of the study and collected and handled the samples; M.U. and I.G.B. reviewed the manuscript and contributed important intellectual content.