Despite the fact that various effective weight management strategies are available, overweight and obesity are increasing(Reference Farhat, Iannotti and Simons-Morton1), and a variety of weight-loss dietary supplements are currently being marketed as slimming aids. The efficacy of most of these supplements has not been proven. One such supplement is the bean extract, Phaseolus vulgaris.
The common bean P. vulgaris is a legume, which is predominantly found around Mexico and Central America(Reference Gentry2). It can be consumed by humans and has been described as belonging to the group of starch blockers, which have been postulated to have beneficial effects on body weight(Reference Obiro, Zhang and Jiang3, Reference Goldstein4). P. vulgaris has been reported to possess α-amylase inhibition activity and is believed to cause weight loss by promoting the mobilisation of the body's fat reserves as a result of energy restriction(Reference Obiro, Zhang and Jiang3). P. vulgaris is also purported to reduce body weight through appetite suppression(Reference Pusztai, Grant and Duguid5). In addition, it has been suggested that P. vulgaris may possess anti-diabetic properties by causing a reduction in postprandial hyperglycaemia, as well as a decrease in insulin secretion(Reference Boivin, Zinsmeister and Brown6).
Phaseolus vulgaris is marketed under different brand names such as Phaseolamin and Phase-2, and is available either as a single compound supplement or in combination with other dietary components. Animal studies have suggested that P. vulgaris causes weight loss(Reference Fantini, Cabras and Lobina7), and a number of clinical trials have been conducted to assess its efficacy in human subjects.
The aim of the present systematic review is to critically evaluate the evidence for or against the efficacy of P. vulgaris in reducing body weight.
Methods
Electronic searches were conducted in the following databases: Medline, Embase, Amed, Cinahl and The Cochrane Library. Each database was searched from inception up until July 2010. The search terms used included dietary supplement, food supplement, nutritional supplement, nutraceutical, anti-obesity agent, appetite suppressant, overweight, obesity, weight loss, slimming, body weight, body fat, BMI, starch blocker, α amylase inhibitor, kidney bean, common bean, P. vulgaris and derivatives of these. We also searched the Internet for relevant conference proceedings, and hand-searched relevant medical journals and our own files. The bibliographies of all located articles were also searched. No age, sex, time or language restrictions were imposed.
Only randomised, double-blind, placebo-controlled trials (RCT) were included in the present review. To be considered for inclusion, RCT had to test the efficacy of orally administered bean extract or refined P. vulgaris for body weight or fat reduction in overweight or obese human volunteers. Included studies also had to report body weight or body composition as an outcome measure. RCT were included irrespective of whether or not they incorporated adjustments in the participants' lifestyle (e.g. dietary restriction and exercise) or other co-interventions into the trial regimen. However, any such interventions had to be applied equally to both the P. vulgaris and placebo groups for studies to be considered for inclusion. Studies testing bean extract or P. vulgaris as part of a combination supplement, i.e. dietary interventions containing other supplements in addition to bean extract, were excluded from the review.
The eligibility of studies was assessed by two reviewers (I. O. and S. A.) independently. Data were extracted systematically by two independent reviewers (I. O. and S. A.) according to the patient characteristics, interventions and results. The methodological quality of all included studies was assessed by the use of a quality assessment checklist adapted from the Consolidated Standard of Reporting Trials guidelines(Reference Moher, Schulz and Altman8, Reference Altman, Schulz and Moher9). Disagreements were resolved through discussion.
Data are presented as means and standard deviations. Mean changes in body weight and body fat mass were used as common endpoints to assess the differences between the P. vulgaris and placebo groups. Using standard meta-analysis software(10), we calculated mean differences (MD) and 95 % CI for studies with adequate data for statistical pooling. The I 2 statistic was used to assess for statistical heterogeneity among studies.
Results
Our electronic searches returned 2512 non-duplicate citations, of which ten potentially relevant articles were identified, and the full texts of these were retrieved (Fig. 1). We also located one unpublished article via hand searching of bibliographies. We excluded one study because it was an open trial(Reference Koike, Koizumi and Tang11). Also, two studies were excluded because they involved the use of P. vulgaris or bean extract as part of a combination therapy(Reference Opala, Rzymski and Wilczak12, Reference Thom13) and another two were excluded because they did not report body weight or composition(Reference Udani, Singh and Barrett14, Reference Birketvedt, Travis and Langbakk15). Thus, six RCT(Reference Celleno, Toliani and D'Amore16–Reference Udani, Hardy and Madsen21) including a total of 247 participants met our inclusion criteria and were included. Key data are summarised in Tables 1 and 2.
Fig. 1 Flow chart showing the process for the inclusion of randomised clinical trials.
Table 1 Methodological characteristics of randomised clinical trials
M/F, males/females; ITT, intention-to-treat; N/R, not reported.
* Unpublished study.
Table 2 Main results of randomised clinical trials
(Mean values, standard deviations or standard errors)
PVE, Phaseolus vulgaris extract; PLA, placebo; AE, adverse events.
* In terms of percentage.
† Mean values were not significantly different (P = 0·4235).
‡ For all subjects.
§ Mean values were not significantly different (P = 0·35).
All RCT had one or more methodological weaknesses (Table 1). Only one reported an appropriate randomisation technique(Reference Udani, Hardy and Madsen21), and only one reported an appropriate allocation concealment procedure(Reference Celleno, Toliani and D'Amore16). Intention-to-treat analysis was included in only one RCT(Reference Udani, Hardy and Madsen21). All included RCT were of parallel design, and one trial was described as having a single-blinded run-in period to exclude non-adherent subjects, in order to improve compliance to the study protocol(Reference Celleno, Toliani and D'Amore16).
Most of the RCT included in the present review incorporated at least one form of lifestyle adjustment into their trials. There was a wide variation in the daily energy intake of participants in the different studies, with values of as low as 5020·8 kJ(Reference Birketvedt, Langbakk and Florholmen18) to as high as 9204·8 kJ(Reference Celleno, Toliani and D'Amore16). To monitor the energy intake of participants(Reference Celleno, Toliani and D'Amore16, Reference Udani and Singh17, Reference Díaz, Aguirre and Gotteland19), three studies enlisted the services of nutritionists, while two studies measured the dietary compliance of their participants using daily diet diaries(Reference Birketvedt, Langbakk and Florholmen18, Reference Udani, Hardy and Madsen21). There were two studies that described their diets as being constituted of complex carbohydrates(Reference Celleno, Toliani and D'Amore16, Reference Udani, Hardy and Madsen21), with one trial reporting its diet as a high-carbohydrate/low-fibre diet(Reference Udani, Hardy and Madsen21). In one RCT, the participants received their meals twice daily from the care providers to ensure compliance with daily energy requirements and also participated in behavioural therapy sessions to improve compliance to eating requirements(Reference Udani and Singh17). In one trial(Reference Udani and Singh17), the authors reported a significant difference in body-weight reduction in the P. vulgaris group compared with that in the placebo group, among subjects who had a high carbohydrate intake (P = 0·0412). In three RCT(Reference Udani and Singh17, Reference Birketvedt, Langbakk and Florholmen18, Reference Udani, Hardy and Madsen21), the authors mentioned exercise as part of lifestyle adjustment in their trials, with the authors of one trial reporting supervised exercises by a personal trainer(Reference Udani and Singh17). Participants in one RCT(Reference Celleno, Toliani and D'Amore16) were allowed to continue with their normal lifestyle during the intervention period. In two RCT(Reference Díaz, Aguirre and Gotteland19, Reference Meiss and Ballerini20), the authors did not report exercise as part of lifestyle adjustment in their trials. Body fat was measured using bioelectrical impedance in three RCT(Reference Celleno, Toliani and D'Amore16, Reference Udani and Singh17, Reference Díaz, Aguirre and Gotteland19), while body fat was estimated by a biodynamics fat analyser in one RCT(Reference Udani, Hardy and Madsen21). Another RCT(Reference Birketvedt, Langbakk and Florholmen18) calculated fat mass from percentage of body fat to body weight, while body fat measurement was not reported in one RCT(Reference Meiss and Ballerini20).
In three RCT, the authors did not provide adequate data for statistical pooling(Reference Udani and Singh17, Reference Meiss and Ballerini20, Reference Udani, Hardy and Madsen21). Of these three RCT, two reported non-significant differences in body-weight reduction between the P. vulgaris and placebo groups(Reference Udani and Singh17, Reference Udani, Hardy and Madsen21). The third trial(Reference Meiss and Ballerini20) reported a mean body-weight loss of 2·9 and 0·3 kg for the P. vulgaris and placebo groups, respectively; there was no report on inter-group differences, and there was no information on how many participants there were in the P. vulgaris and placebo groups.
A forest plot (random-effect model) for RCT with suitable data for statistical pooling (Fig. 2) reveals a statistically non-significant difference in body-weight reduction between the P. vulgaris and placebo groups (MD − 1·77 kg, 95 % CI − 3·33, 0·33). The I 2 statistic (75 %) suggests considerable heterogeneity. A further meta-analysis of these three RCT (Fig. 3) revealed a statistically significant reduction in body fat favouring P. vulgaris over placebo (MD − 1·86 kg, 95 % CI − 3·39, − 0·32). Heterogeneity was moderate in this analysis (I 2 = 53 %). A sensitivity analysis of two trials with similar dosages and duration of treatment(Reference Birketvedt, Langbakk and Florholmen18, Reference Díaz, Aguirre and Gotteland19) revealed a statistically non-significant difference in body-weight reduction between the P. vulgaris and placebo groups (MD − 1·45 kg, 95 % CI − 4·49, 1·59). Heterogeneity was considerable in this analysis (I 2 = 86 %).
Fig. 2 Forest plot showing the effect of Phaseolus vulgaris on body weight.
Fig. 3 Forest plot showing the effect of Phaseolus vulgaris on body fat.
Data on waist circumference were reported in four RCT, and two of these provided data for statistical pooling. A forest plot of the two studies with suitable data(Reference Birketvedt, Travis and Langbakk15, Reference Celleno, Toliani and D'Amore16) revealed a statistically significant decrease in favour of P. vulgaris over placebo (MD − 2·24 cm, 95 % CI − 3·84, − 0·63). Heterogeneity was not important in this analysis (I 2 = 0 %). The remaining two trials(Reference Udani and Singh17, Reference Udani, Hardy and Madsen21) reported a non-significant difference in waist-circumference reduction between the P. vulgaris and placebo groups (P = 0·8654 and P>0·05, respectively).
The dosages of P. vulgaris varied across the RCT. In two studies(Reference Celleno, Toliani and D'Amore16, Reference Meiss and Ballerini20), the participants had a daily dosage of 445 mg, and in other studies, the doses ranged from 1000 to 1500 mg daily. There was no significant relationship between dosage and body-weight loss (data not shown).
Adverse events reported in the RCT included headache, soft stool, flatulence and constipation. No serious adverse events and no significant differences in the frequency of adverse events between the P. vulgaris and placebo groups were observed. In total, thirty-one dropouts were reported: seventeen in the P. vulgaris group and fourteen in the placebo group. In one RCT, the reasons for dropouts were not reported(Reference Udani, Hardy and Madsen21).
Discussion
The aim of the present systematic review was to assess the efficacy of P. vulgaris as a weight-loss supplement. The overall meta-analysis results involving three studies with 133 participants indicate that P. vulgaris does not generate a statistically significant reduction in body weight when compared with placebo. This result is at variance with two other studies that did not provide adequate data for statistical pooling(Reference Udani and Singh17, Reference Udani, Hardy and Madsen21). Further meta-analysis suggests that P. vulgaris causes a statistically significant reduction in body fat when compared with placebo, but two studies without adequate data for meta-analysis reported non-significant differences in percentage of body fat between the P. vulgaris and placebo groups(Reference Udani and Singh17, Reference Udani, Hardy and Madsen21). However, the meta-analysis results should be interpreted with caution, given the high level of heterogeneity among the studies. The clinical relevance of the results is also uncertain, as the analyses fail to provide an indication that a clinically significant weight loss, defined as at least a 5 % reduction in body weight or fat from baseline was achieved(Reference Christian, Tsai and Bessesen22). The 5 % weight loss that is considered to be clinically significant is usually taken at the 6-month time point, and weight loss at 12 weeks is about two-thirds of the weight loss observed at the 6-month plateau. Thus, the estimated weight loss at plateau extrapolated from overall meta-analysis would be about 2·5 kg. The weight loss of 1·77 kg from the meta-analysis was below that expected at a weight-loss plateau. However, because of the heterogeneity evident in our meta-analysis, it is not possible to ascertain as to whether or not P. vulgaris supplementation results in weight loss ≥ 2·5 kg at 6 months. Though a meta-analysis of two studies indicates that P. vulgaris causes a significant reduction in waist circumference compared with the placebo, this result differs from the findings of two other studies that did not provide sufficient data for statistical pooling(Reference Udani and Singh17, Reference Udani, Hardy and Madsen21).
All the RCT included in the present systematic review had important methodological flaws, and the trial methodologies varied considerably. All RCT had small sample sizes, with the maximum number of participants in a single trial being sixty-two. Small sample sizes are prone to produce unreliable results(Reference Wittes23). Most of the RCT did not report carrying out a power calculation or performing intention-to-treat analysis. Majority of the studies were also of short duration, with some as short as 4 weeks. This seems too short for assessing the effects of P. vulgaris on body weight, and longer-term studies are required for this purpose.
The variety in study methodology (in relation to both quality and design), small sample sizes, variation in dosages and the generally short duration of the intervention period limit the extent to which efficacy or otherwise can be inferred, and lack of detailed reporting creates doubts regarding the internal and external validity of the included studies.
Lifestyle modification is regarded as a cornerstone in the management of obesity(Reference Wadden, Butryn and Wilson24). Though most of the RCT incorporated lifestyle modifications into their trial regimen, they differed in the amounts of average daily energy intake, as well as in the level of exercise undertaken by study participants. The degree to which the adjustment for these lifestyle factors influenced the outcome of the results in the studies is unclear.
Most of the studies suggested that the use of P. vulgaris appeared to be generally safe, with reported side effects being mostly mild gastrointestinal symptoms. Consumption of raw or undercooked P. vulgaris has been associated with a variety of serious adverse events, due to the presence of phytohaemagglutinin(Reference Celleno, Toliani and D'Amore16); phytohaemagglutinin is largely inactivated in the processing stage of this supplement. This does not, however, rule out the possibility of serious adverse events if the supplement is taken on the long term. It will be prudent in future investigations to incorporate surveillance time frames into trial designs; to date, investigators have tended to stop monitoring for adverse events once the study duration is completed(Reference Ioannidis, Evans and Gøstzsche25).
Most of the studies included in the review reported their source of funding. The majority of RCT involving the use of P. vulgaris have been commercially funded by the private industry. None of the RCT was funded exclusively by government.
The present review has several limitations. Though we searched both electronic and non-electronic sources, we may not have identified all RCT involving the use of P. vulgaris as a weight-loss supplement, in particular, those that remain unpublished. In addition, the methodological quality of all of the studies identified from our searches is poor, and most studies are of short duration. These factors prevent us from drawing firm conclusions about the effects of P. vulgaris on human body weight.
Conclusion
The evidence from RCT is not adequate enough to conclusively determine the effects of P. vulgaris supplementation on body weight. The methodological quality of all RCT is poor, and most are of short duration. Larger and more rigorous trials with longer duration are required to objectively assess the effects of this herbal supplement on body weight.
Acknowledgements
We would like to thank Marie-Laure Despres for help with translation of a Spanish article. I. O. receives funding from GlaxoSmithKline. The funder had no role in the preparation of the manuscript. S. A., R. T. and E. E. declare no potential competing interests. The contribution of the authors was as follows: I. O. was involved in the protocol development, literature search, study selection, data extraction, data presentation, co-drafting of the review. S. A. was responsible for the literature search, study selection, data extraction, co-drafting of the review. R. T. participated in the data presentation and co-drafting of the review. E. E. formulated the idea for the review, and supervised the protocol development and co-drafting of the review.
