Effect of glucomannan on plasma lipid and glucose concentrations,
body weight, and blood pressure: systematic review and
meta-analysis1,2
Nitesh Sood, William L Baker, and Craig I Coleman
ABSTRACT
Background: Several clinical trials have investigated the impact of
glucomannan on plasma lipids, body weight, fasting blood glucose
(FBG), and blood pressure (BP), but have yielded conflicting results
and had only modest sample sizes.
Objective: The objective was to perform a meta-analysis of ran-
domized controlled trials of glucomannan to better characterize its
impact on plasma lipids, FBG, body weight, and BP.
Design: A systematic literature search of MEDLINE, EMBASE,
CINAHL, Web of Science, the Cochrane Library, and the Natural
Medicines Comprehensive Database was conducted from the earli-
est possible date through November 2007. A random-effects model
was used to calculate the weighted mean difference (WMD) and 95%
CIs as the difference between the mean for the glucomannan and
control groups. Standard methods for assessing statistical heteroge-
neity and publication bias were used.
Results: Fourteen studies (n ҃ 531) met the inclusion criteria. The
use of glucomannan significantly lowered total cholesterol
[weighted mean difference (WMD): Ҁ19.28 mg/dL; 95% CI:
Ҁ24.30, Ҁ14.26], LDL cholesterol (WMD: Ҁ15.99 mg/dL; 95%
CI: Ҁ21.31, Ҁ10.67), triglycerides (WMD: Ҁ11.08 mg/dL; 95% CI:
Ҁ22.07, Ҁ0.09), body weight (WMD: Ҁ0.79 kg; 95% CI: Ҁ1.53,
Ҁ0.05), and FBG (WMD: Ҁ7.44 mg/dL; 95% CI: Ҁ14.16, Ҁ0.72).
The use of glucomannan did not appear to significantly alter any
other study endpoints. Pediatric patients, patients receiving dietary
modification, and patients with impaired glucose metabolism did not
benefit from glucomannan to the same degree.
Conclusions: Glucomannan appears to beneficially affect total cho-
lesterol, LDL cholesterol, triglycerides, body weight, and FBG, but
not HDL cholesterol or BP. Am J Clin Nutr 2008;88:1167–75.
INTRODUCTION
More than 50 million Americans are thought to suffer from the
metabolic syndrome, which is characterized by a group of met-
abolic risk factors occurring in a single individual, including but
not limited to abdominal obesity, atherogenic dyslipidemia, el-
evated blood pressure, and insulin resistance or glucose intoler-
ance (1). Patients with the metabolic syndrome are at increased
risk of coronary heart disease, stroke, and peripheral vascular
disease as well as type 2 diabetes mellitus. According to the
American Heart Association, the primary goal for the manage-
ment of patients with the metabolic syndrome is to reduce their
Am J Clin Nutr 2008;88:1167–75. Printed in USA. © 2008 American Society for Nutrition
risk of cardiovascular disease and type 2 diabetes through smok-
ing cessation and by reducing LDL cholesterol, blood pressure,
Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017
body mass index, and glucose to recommended levels (1).
Glucomannan is a soluble fiber derived from Amorphophallus
konjac and is available in numerous over-the-counter products
such as Lipozene. Like other soluble fiber (oats, guar gum, pec-
tin, and psyllium), glucomannan has been touted for its potential
beneficial effects on the risk of coronary heart disease (2). Glu-
comannan is thought to prolong gastric emptying time, which
increases satiety, reduces body weight, decreases the ingestion of
foods that increase cholesterol and glucose concentrations, re-
duces the postprandial rise in plasma glucose, suppresses hepatic
cholesterol synthesis, and increases the fecal elimination of cho-
lesterol containing bile acids (2).
Several clinical trials (3–19) have investigated the impact of
glucomannan on total cholesterol, LDL cholesterol, HDL cho-
lesterol, triglycerides, body weight, fasting blood glucose
(FBG), systolic blood pressure (SBP), or diastolic blood pressure
(DBP), but have yielded conflicting results and had only modest
sample sizes. Although previous meta-analyses assessing the
effects of soluble fibers on these same endpoints have been pub-
lished, none have evaluated glucomannan. Therefore, we con-
ducted a meta-analysis of randomized controlled trials of gluco-
mannan to better characterize its impact on various
characteristics of the metabolic syndrome.
METHODS
Data sources
A systematic literature search of MEDLINE, EMBASE, CI-
NAHL, Web of Science, the Cochrane Library, and the Natural
Medicines Comprehensive Database was conducted from the
earliest possible date through November 2007. A search strategy
1
From the University of Connecticut Schools of Medicine (NS) and Phar-
macy (WLB and CIC), Farmington and Storrs, CT, and the Departments of
Medicine (NS) and Drug Information (WLB and CIC), Hartford Hospital,
Hartford, CT.
2
Address reprint requests and correspondence to CI Coleman, University
of Connecticut School of Pharmacy Hartford Hospital, 80 Seymour Street,
Hartford, CT 06102-5037. E-mail: ccolema@harthosp.org.
Received February 22, 2008.
Accepted for publication June 16, 2008.
1167
1168 SOOD ET AL
using the Medical Subject Headings (MeSH) and the text key-
words konjac, mannan, konjac-mannan, konjac flour, gluco-
mannan, glucomannano, konjaku, konnyaku, conjac, devil’s
tongue, voodoo lily, snake palm, Amorphophallus konjac, Amor-
phophallus rivieri, and Araceae was used. This search was then
limited to clinical trials. No language restrictions were imposed.
In addition, a manual search of references from reports of clinical
trials or review articles was performed to identify relevant trials.
When applicable, efforts were made to contact investigators for
clarification or additional data.
Study selection
Only randomized controlled trials of glucomannan that re-
ported efficacy data on at least one of the following components
of the metabolic syndrome were included in the analysis: total
cholesterol, LDL cholesterol, HDL cholesterol, triglycerides,
body weight, FBG, SBP, or DBP. Both parallel and crossover
trials were eligible for inclusion; however, crossover trials had to
have a washout period of 욷2 wk to be included in the meta-
analysis. Studies using an active control were excluded.
Validity assessment
Because they are inherent controls of bias, randomization and
double-blinding were used to assess the methodologic quality of
included trials.
Data abstraction
Through use of a standardized data abstraction tool, 2 review-
ers independently collected data; disagreements were resolved
through discussion. The following information was obtained
from each trial: author identification, year of publication, study
design, the abovementioned methodologic quality criteria,
source of study funding, study population (including study in-
clusion and exclusion criteria), sample size, duration of patient
follow-up, glucomannan dose and formulation used, use of con-
current dietary modification, effects on lipid variables (total cho-
lesterol, LDL cholesterol, HDL cholesterol, and triglycerides),
body weight, FBG, SBP, and DBP.
Statistical analysis
The mean change in lipid, glucose, body weight, and blood
pressure variables from baseline was treated as a continuous
variable and the weighted mean difference (WMD) was calcu-
lated as the difference between the mean in the glucomannan and
control groups. A DerSimonian and Laird random-effects model
(a variation on the inverse variance method, which incorporates
an assumption that the different studies are estimating different,
yet related, treatment effects) was used in calculating the WMD
and its 95% CI. For parallel trials, net changes in each of these
study variables were calculated as the difference (glucomannan
minus control) in the changes (baseline minus follow-up) in these
mean values (also referred to as the change score). For crossover
trials, net changes were calculated as the mean difference in
values at the end of the glucomannan and control periods. Be-
cause variances for net changes were not reported directly for
most studies, they were calculated from CIs, P values, or indi-
vidual variances for intervention and control groups or periods.
For parallel trials in which variance for paired differences was
reported separately for each group, we calculated a pooled vari-
ance for net change using standard methods. When the variance
for paired differences was not reported, we calculated it from
variances at baseline and at the end of follow-up. As suggested by
Follmann et al (20), we assumed a correlation coefficient of 0.5
between initial and final values. We assumed equal variances
during the trial and between intervention and control groups.
The statistical analysis was performed by using STATSDI-
RECT statistical software (version 2.4.6; StatsDirect Ltd,
Cheshire, United Kingdom), and MIX statistical software (freely
accessible at www.mix-for-meta-analysis.info). A P value
쏝0.05 was considered statistically significant for all analyses,
except where otherwise specified.
Statistical heterogeneity was addressed using the Q Statistic,
where a P value 쏝 0.10 was considered representative of signif-
icant statistical heterogeneity. Visual inspection of funnel plots
and Egger’s weighted regression statistics were used to assess for
the presence of publication bias. To assess the potential effect of
any publication bias on the meta-analysis results, the Trim and
Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017
Fill method was used, which uses funnel plot symmetry to esti-
mate the number of “missing” studies and the magnitudes of their
effects. It then re-estimates the overall effect size after imputing
any potentially “missing” studies into the meta-analysis to de-
termine whether the results of the original analysis were mark-
edly affected by publication bias.
Studies of poorer methodologic quality, such as open-label or
crossover trials, may exhibit inaccurate treatment effects. Ex-
cluding them may result in increased internal validity but could
reduce the external validity of the analysis. In addition, the se-
lection of a random-effects rather than a fixed-effects model in a
meta-analysis is controversial. The use of a random-effects
model in the calculation of CIs results in wider intervals and thus
a more conservative estimate of treatment effects when com-
pared with a fixed-effects model. To reconcile these issues, sen-
sitivity analysis was conducted whereby the meta-analysis was
reanalyzed excluding studies that were not double-blinded, ex-
cluding crossover studies, and finally using a fixed-effects model
(Mantel-Haenszel methodology). We also conducted a sensitiv-
ity analysis excluding the study by Venter et al (15), which uses
a glucomannan product that may have also contained another
pharmacologically active soluble fiber (pectin) and was limited
to studies using total daily doses of 울3 (ie, the minimum recom-
mended dose of soluble fiber by the Food and Drug Administra-
tion) (21) and 10 g glucomannan/d (ie, the maximum practical
dose of soluble fiber) (22). Additionally, subgroup analyses were
performed whereby the effects of glucomannan on study end-
points were assessed separately in subjects with impaired glu-
cose metabolism [type 2 diabetes mellitus or impaired glucose
tolerance (IGT)], in obese subjects, in adults, and in children and
in studies using or not using concurrent dietary modifications.
RESULTS
The initial search yielded 3109 potential literature citations.
Of these, only 24 were clinical trials in humans. On review of
references from identified studies, an additional 5 studies poten-
tially meeting our inclusion criteria were identified, bringing the
total number of studies for full-text review to 29. Fifteen of the 29
studies were excluded for the reasons given in Figure 1. Of note,
the study by Vita et al (17) met all the inclusion criteria but could
not be included because measures of variation around effect were
not provided and could not be estimated from the data provided.
 GLUCOMANNAN AND LIPIDS, FBG, WEIGHT, AND BP
FIGURE 1. Flow diagram of trial identification, inclusion, and exclusion. *Data from the study by Zhang et al (14) was excluded from this analysis because
of concerns regarding errors in data reporting. DBP, diastolic blood pressure; FBG, fasting blood glucose; SBP, systolic blood pressure.
Two other studies (18, 19) did not meet our criteria for inclusion
because of the lack of a washout period of adequate duration.
Thus, a total of 14 randomized controlled trials (3–16) that
evaluated 531 subjects were included in this meta-analysis (Ta-
ble 1). Two of these studies (5, 15) reported the results of 2
heterogeneous and mutually distinct populations separately in
their articles; therefore, we opted to include each of these anal-
yses in our meta-analysis separately. Eight of the studies were
conducted using a parallel study design (4, 6, 7, 11–14, 16),
whereas the other 6 studies used a crossover design. All the
crossover studies used a 2-wk washout period, except for the
study by Yoshida et al (5), which used a 4-wk washout period.
Each of the studies enrolled a relatively small number of patients
(median sample size: 20 subjects; range: 11–110 subjects) and
had a short length of follow-up (median duration: 5– 8 wk; range:
3–16 wk). All of the studies used placebo as the control, except
for 3 studies (6, 13, 14) that used diet only as the control. Each of
the included studies evaluated patients having at least one, if not,
multiple, constituents of the metabolic syndrome, including type
2 diabetes mellitus or impaired glucose tolerance (3, 5, 8, 9),
hyperlipidemia (5, 6, 9 –11, 14, 15), hypertension (9), or obesity
(4, 7, 11–13, 15). The dosage range of glucomannan used in the
included studies ranged from 1.2 to 15.1 g/d and were adminis-
tered in various forms, such as capsules, tablets, bars, biscuits,
and refined konjac meal. Nine of the studies (3, 4, 6 –9, 11–13)
administered glucomannan along with some type of dietary
modification.(Table 1) Of the 14 studies, 4 were not double-
blinded (3, 6, 13, 14). More than half (57.1%) of the studies stated
they were funded through industry (4 – 6, 8, 9, 12, 15, 16), one
was funded through academia (7), and the remainder did not
report their funding source (3, 10, 11, 13, 14).
The meta-analysis showed that the use of glucomannan ap-
peared to statistically significantly lower total cholesterol, LDL
cholesterol, triglycerides, body weight, and FBG (Figure 2).
Statistical heterogeneity was observed only in the body weight
endpoint (Q statistic P ҃ 0.04). The use of glucomannan did not
appear to significantly alter any of the other study endpoints
1169
Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017
(Table 2). Of these other endpoints, only DBP displayed statis-
tical heterogeneity (P ҃ 0.03).
Visual inspection of funnel plots (data not shown) could not
rule out publication bias for many of the analyses. Review of
Egger’s weighted regression statistics suggested that publication
bias was unlikely for all analyses (P 쏜 0.19 for all). After we
recalculated effect size estimates using the Trim and Fill meth-
odology, only our conclusion regarding glucomannan’s effect on
HDL cholesterol was significantly altered. In this case, the Trim
and Fill analysis suggests that as many as 3 studies could poten-
tially exist but were masked by publication bias and, when fac-
tored in, that glucomannan may have a small but statistically
significant detrimental effect on HDL cholesterol (reduction of
2.01 mg/dL).
The results of subgroup and sensitivity analyses are presented
in Table 2. Similar benefits in pediatric patients compared with
adults (or the base-case analysis) were not seen. In addition, it
appears that glucomannan does not have as robust an effect on
triglycerides in patients with impaired glucose metabolism or
when used in conjunction with dietary modification, the benefits
of glucomannan on weight loss are enhanced by dietary modifi-
cation, and patients with impaired glucose metabolism do not
reap the hypoglycemic benefits of glucomannan. After the ex-
clusion of studies that used a double-blind methodology, the
analyses for triglycerides and FBG went from being of borderline
statistical significance to borderline nonsignificance; however,
because the effect size estimates for both analyses remained
relatively constant, these changes were likely a result of de-
creased statistical power. No other subgroup or sensitivity anal-
yses resulted in changes in overall study conclusions.
DISCUSSION
In this meta-analysis of 14 randomized controlled trials, pa-
tients receiving glucomannan had statistically significantly
lower total cholesterol, LDL cholesterol, triglycerides, body
TABLE 1
Clinical trial characteristics1
1170

Baseline weight Baseline FBG Glucomannan

dosing and Concurrent diet
Study design Patient population Baseline lipids2 Treatment Control Treatment Control Follow-up dosage form Control group modification

mg/dL kg kg mg/dL mg/dL wk

Chearskul et al (3), Randomized, single-blind, Type 2 DM TC ҃ 191, 193; LDL-C 70 69 163 159 4 3 g/d as Placebo (3 g/d Under dietary
2007 (n ҃ 20) placebo-controlled, ҃ 92, 96; HDL-C ҃ capsules white rice control
crossover 53, 53; TG ҃ 267, 243 flour)
Wood et al (4), 2007 Randomized, double- Obese men TC ҃ 179, 177; LDL-C 94 93 92 93 12 3 g/d as pills Placebo CHO restricted
(n ҃ 29) blind, placebo- ҃ 115, 112; HDL-C (contents (60% fat,
controlled, crossover ҃ 41, 42; TG ҃ 116, NA) 30% protein,
119 10% CHO)
Yoshida et al (5), Randomized, double- Hyperlipidemic TC ҃ 221, 225; LDL-C NA NA NA NA 3 10 g/d as Placebo Dietary habits
2006 (n ҃ 29) blind, placebo- with or without ҃ 149, 152; HDL-C granola (granola stable; CHO
controlled, crossover DM ҃ 40, 41; TG ҃ 147, bars3 bars)4 replaced by
165 granola bar
Martino et al (6), Randomized, open-label, Hyperlipidemic TC ҃ 243, 254; LDL-C 30 30 NA NA 8 2–3 g/d as Diet only AAPC Step 1
2005 (n ҃ 40) diet-controlled, parallel children ҃ 172, 176; HDL-C capsules Diet (쏝10%
҃ 54, 59; TG ҃ 96, saturated fat,
90 쏝30% fat,
쏝300 mg
cholesterol)
Birketvedt et al (7), Randomized, double- Overweight TC ҃ NA; LDL-C ҃ 79 77 NA NA 5 1.24 g/d as Placebo Low calorie
2005 (n ҃ 52) blind, placebo- women NA; HDL-C ҃ NA; tablets (contents (1200 kcal,
controlled, parallel TG ҃ NA NA) 35% fat,
SOOD ET AL

15% protein,
55% CHO)
Vuksan et al (8), Randomized, double- IGT TC ҃ 240, 232; LDL-C 81 81 122 119 3 8-13 g/d as Placebo biscuits NCEP Step 2
2000 (n ҃ 11) blind, placebo- ҃ 151, 147; HDL-C biscuits (15% diet (쏝7%
controlled, crossover ҃ 39, 39; TG ҃ 248, polysaccharides, saturated fat,
257 16% 쏝30% fat,
excipients) ѿ 쏝300 mg
11 g/d of cholesterol)
hard red
wheat bran
Vuksan et al (9), Randomized, double- Hypertensive, TC ҃ 236, 225; LDL-C 86 86 173 167 3 15.1 g Placebo NCEP Step-2
1999 (n ҃ 11) blind, placebo- hyperlipidemic, ҃ 150, 138; HDL-C (mean)/d as biscuits5 ѿ diet (쏝7%
controlled, crossover type 2 DM ҃ 41, 40; TG ҃ 224, biscuits 14 g/d of saturated fat,
238 hard red 쏝30% fat,
wheat bran 쏝300 mg
cholesterol)
Arvill and Bodin Randomized, double- Hyperlipidemic TC ҃ 268, 260; LDL-C 90 90 NA NA 4 3.9 g/d as Placebo Usual diet
(10), 1995 blind, placebo- men ҃ 177, 176; HDL-C capsules (containing
(n ҃ 63) controlled, crossover ҃ 46, 48; TG ҃ 210, cornstarch,
255 lactose,
magnesium
stearate)
(Continued)
Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017
TABLE 1 (Continued)

Baseline weight Baseline FBG Glucomannan

dosing and Concurrent diet
Study design Patient population Baseline lipids2 Treatment Control Treatment Control Follow-up dosage form Control group modification

Cairella and Marchini Randomized, double- Hyperlipidemic TC ҃ 249, 246; LDL-C NA NA 117 123 8 3.87 g/d as Placebo Low calorie
(11), 1995 (n ҃ blind, placebo- obese women ҃ NA; HDL-C ҃ capsules (inactive (1200 kcal,
15) controlled, parallel NA; TG ҃ NA contents) 29% fat,
25% protein,
46% CHO)
Vido et al (12), 1993 Randomized, double- Obese children TC ҃ 194, 175; LDL-C NA NA NA NA 8 2 g/d as Placebo Well balanced
(n ҃ 60) blind, placebo- ҃ NA; HDL-C ҃ capsules (contents normocaloric
controlled, parallel NA; TG ҃ 73, 107 NA) diet
Livieri et al (13), Randomized, open-label, Obese children TC ҃ 187, 184; LDL-C NA NA NA NA 16 2-3 g/d as Diet only Balanced diet
1992 (n ҃ 53) diet-controlled, parallel ҃ NA; HDL-C ҃ capsules (30% fat,
NA; TG ҃ 82, 83 15% protein,
55% CHO)
Zhang, et al (14), Randomized, open-label, Hyperlipidemic TC ҃ 205, 199; LDL-C NA NA NA NA 6 5-10 g/d as Diet only Ordinary diet
1990 (n ҃ 110) diet-controlled, parallel adults ҃ 56, 109; HDL-C refined
҃ 47, 118; TG ҃ foods
222, 253
Venter et al (15), Randomized, double- Hyperlipidemic TC ҃ 274; LDL-C ҃ 74 74 80 80 4 4.5 g/d as Placebo Normal diet
1987 (n ҃ 18) blind, placebo adults NA; HDL-C ҃ NA; capsules (cornstarch,
controlled, crossover TG ҃ NA quantity NA)
Walsh et al (16), Randomized, double- Obese women TC ҃ 198; LDL-C ҃ 84 83 NA NA 8 3 g/d as Placebo (500 Previously
1984 (n ҃ 20) blind, placebo- 125; HDL-C ҃ NA; capsules mg starch) established
controlled, parallel TG ҃ NA eating
patterns
1
GLUCOMANNAN AND LIPIDS, FBG, WEIGHT, AND BP

CHO, carbohydrate; NA, not available; TC, total cholesterol; TG, triglycerides; C, cholesterol; DM, diabetes mellitus; IGT, impaired glucose tolerance; NCEP, National Cholesterol Education Program; FBG,
fasting blood glucose. To convert values for cholesterol from mg/dL to mmol/L, multiply by 0.0286; to convert values for TG from mg/dL to mmol/L, multiply by 0.01129.
2
First value given is for the treatment group at baseline; second value is for the control group at baseline. Venter et al (15) and Walsh et al (16) reported only means for entire study cohort.
3
Glucomannan bar composition: 360 kcal total energy, 58.4 g CHO, 6.3 g protein, 11.2 g fat, 2.3 g saturated fat, 274.9 mg Na, 14.9 g fiber, 768.6 IU vitamin A, 4 mg vitamin C, 2.3 mg Fe, and 42.6 mg Ca.
4
Placebo bar composition: 399 kcal total energy, 66.7 g CHO, 8.3 g protein, 11.2 g fat, 2.0 g saturated fat, 308.6 mg Na, 5 g fiber, 852.7 IU vitamin A, 5.3 mg vitamin C, 2.6 mg Fe, and 48.2 mg Ca.
5
Glucomannan biscuit composition (g/100 g): 6.2 g protein, 13.9 g fat, 61.2 g CHO, 1.3 g ash, 2.3 g dietary fiber, 944 kJ/100 g energy; placebo biscuit composition (g/100 g): 6.8 g protein, 14.4 g fat, 66.5 g
CHO, 1.4 g ash, 2.8 g dietary fiber, and 1011 kJ/100 g energy.
1171

Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017

1172 SOOD ET AL

Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017

FIGURE 2. Effect of glucomannan on characteristics of the metabolic syndrome. A: Total cholesterol (n ҃ 286 glucomannan, n ҃ 273 control); B: LDL
cholesterol (n ҃ 146 glucomannan, n ҃ 148 control); C: HDL cholesterol (n ҃ 138 glucomannan, n ҃ 140 control); D: Triglycerides (n ҃ 301 glucomannan,
n ҃ 287 control); E: Body weight (n ҃ 186 glucomannan, n ҃ 193 control); F: Fasting blood glucose (n ҃ 61 glucomannan, n ҃ 62 control). A DerSimonian
and Laird random-effects model was used in calculating the weighted men difference and its 95% CI. To convert values for cholesterol from mg/dL to mmol/g,
multiply by 0.0286; to convert values for triglycerides from mg/dL to mmol/g, multiply by 0.01129. *Study reported separate analyses of 2 heterogeneous and
mutually exclusive patient populations; thus, each analysis was treated as a separate study in this meta-analysis.

weight, and FBG after treatment than did control patients; how-
ever, the use of glucomannan did not appear to alter HDL cho-
lesterol or either systolic or diastolic blood pressure. Our analysis
could only show trends toward benefits with glucomannan in
children. We also found that glucomannan does not have as
robust an effect on triglycerides in patients with impaired glucose
metabolism or when used in conjunction with dietary modifica-
tion, that the benefits of glucomannan on weight loss are en-
hanced by dietary modification, and that patients with impaired
glucose metabolism do not reap the hypoglycemic benefits of
glucomannan. Because of the smaller number of studies included
in the abovementioned subgroup analyses, they should be inter-
preted with caution and considered hypothesis-generating only.
It would appear that the greatest potential cardiovascular ben-
efits from glucomannan are due to its effect on lipids. Studies
have shown that for each 1-mg/dL reduction in a patient’s LDL-
cholesterol concentration, their relative risk of having a coronary
heart disease event is decreased by 1%. Thus, the 16-mg/dL
reduction in LDL cholesterol seen in our meta-analysis with
glucomannan is not only statistically significant, but likely is also
clinically significant (23).
Glucomannan is not the only soluble fiber believed to reduce
the risk of cardiovascular disease risk by altering plasma lipids.
Two previous meta-analyses (22, 24) evaluated the hypocholes-
terolemic effects of pectin, psyllium, oats, and guar gum with
reductions in total and LDL cholesterol ranging from 7% to 15%
and 7% to 10%, respectively, along with a detrimental but small
effect on HDL cholesterol. These effects on total cholesterol,
LDL cholesterol, and HDL cholesterol are similar to those ob-
served in our meta-analysis, which suggests that this may be a
class effect of soluble fibers. Interestingly, whereas previous
meta-analyses did not demonstrate reductions in triglycerides
with soluble fibers other than glucomannan, our meta-analysis
showed a statistically significant 11-mg/dL reduction. The rea-
son behind glucomannan’s ability to preferentially lower trig-
lycerides compared with other soluble fibers is not known, but
may be related to its higher viscosity and thus its greater ability
to alter the metabolic pathways of hepatic cholesterol and li-
poprotein metabolism (2).
Glucomannan is commonly touted in the United States as an
effective over-the-counter weight-loss supplement (25, 26). In
studies lasting a mean of 5.2 wk, our meta-analysis found that
there was a statistically significant but small reduction in weight
of 0.79 kg (앒1%) with glucomannan. While studied over a
slightly longer period of time, orlistat (Alli) the only over-the-
counter weight-loss treatment approved by the Food and Drug
TABLE 2
Results of the meta-analysis of randomized controlled trials that evaluated glucomannan1
TC LDL-C HDL-C TG Body weight FBG SBP DBP

mg/dL mg/dL mg/dL mg/dL kg mg/dL mm Hg mm Hg

All studies (all doses) Ҁ19.28 Ҁ15.99 Ҁ1.36 Ҁ11.08 Ҁ0.79 Ҁ7.44 0.29 1.81
(Ҁ24.30, Ҁ14.26) (Ҁ21.31, Ҁ10.67) (Ҁ3.37, 0.66) (Ҁ22.07, Ҁ 0.09) (Ҁ1.53, Ҁ0.05) (Ҁ14.16, Ҁ0.72) (Ҁ4.54, 5.13) (Ҁ2.46, 6.09)
(n ҃ 13 studies) (n ҃ 9 studies) (n ҃ 8 studies) (n ҃ 12 studies) (n ҃ 9 studies) (n ҃ 5 studies) (n ҃ 4 studies) (n ҃ 4 studies)
Fixed-effects model Ҁ19.28 Ҁ15.99 Ҁ1.36 Ҁ11.08 Ҁ1.01 Ҁ9.37 0.21 1.66
(Ҁ24.30, Ҁ14.26) (Ҁ21.31, Ҁ10.67) (Ҁ3.37, 0.66) (Ҁ22.07, Ҁ0.09) (Ҁ1.32, Ҁ0.70) (Ҁ13.46, Ҁ5.29) (Ҁ3.72, 4.14) (Ҁ0.77, 4.10)
(n ҃ 13 studies) (n ҃ 9 studies) (n ҃ 8 studies) (n ҃ 12 studies) (n ҃ 9 studies) (n ҃ 5 studies) (n ҃ 4 studies) (n ҃ 4 studies)
Trim and fill Ҁ19.28 Ҁ15.27 Ҁ2.01 Ҁ11.32 Ҁ0.79 Ҁ12.90 0.29 1.81
(Ҁ24.30, Ҁ14.26) (Ҁ20.43, Ҁ10.10) (Ҁ3.84, Ҁ0.18) (Ҁ22.29, Ҁ0.34) (Ҁ1.53, Ҁ0.05) (Ҁ20.39, Ҁ5.40) (Ҁ4.54, 5.13) (Ҁ2.46, 6.09)
(n ҃ 13 ѿ 0 studies) (n ҃ 9 ѿ 1 studies) (n ҃ 8 ѿ 3 studies) (n ҃ 12 ѿ 1 studies) (n ҃ 9 ѿ 0 studies) (n ҃ 5 ѿ 3 studies) (n ҃ 4 ѿ 0 studies) (n ҃ 4 ѿ 0 studies)
Excluding crossover Ҁ17.74 Ҁ20.72 Ҁ1.62 Ҁ9.94 Ҁ1.30 Ҁ8.36 5.90 7.80
studies (Ҁ26.33, Ҁ9.15) (Ҁ31.39, Ҁ10.05) (Ҁ5.63, Ҁ2.39) (Ҁ26.23, 6.35) (Ҁ1.69, Ҁ0.91) (Ҁ18.55, 1.83) (Ҁ1.06, 12.86) (2.77, 12.83)
(n ҃ 7 studies)) (n ҃ 3 studies) (n ҃ 2 studies) (n ҃ 6 studies) (n ҃ 4 studies) (n ҃ 2 studies) (n ҃ 1 study) (n ҃ 1 study)
Excluding studies not Ҁ22.22 Ҁ16.09 Ҁ1.28 Ҁ7.06 Ҁ0.79 Ҁ7.25 0.29 1.81
double-blinded (Ҁ27.18, Ҁ15.26) (Ҁ21.68, Ҁ10.49) (Ҁ3.42, 0.85) (Ҁ21.75, Ҁ6.54) (Ҁ1.56, Ҁ0.02) (Ҁ15.01, 0.51) (Ҁ4.54, 5.13) (Ҁ2.46, 6.09)
(n ҃ 9 studies) (n ҃ 7 studies) (n ҃ 6 studies) (n ҃ 8 studies) (n ҃ 8 studies) (n ҃ 4 studies) (n ҃ 4 studies) (n ҃ 4 studies)
Excluding Venter et al Ҁ19.32 Ҁ16.13 Ҁ1.59 Ҁ11.34 Ҁ0.79 Ҁ7.44 0.29 1.81
(15), 1987 (Ҁ24.50, Ҁ14.15) (Ҁ21.60, Ҁ10.66) (Ҁ3.71, 0.52) (Ҁ22.67, Ҁ0.002) (Ҁ1.53, Ҁ0.05) (Ҁ14.16, Ҁ0.72) (Ҁ4.54, 5.13) (Ҁ2.46, 6.09)
(n ҃ 12 studies) (n ҃ 8 studies) (n ҃ 7 studies) (n ҃ 11 studies) (n ҃ 9 studies) (n ҃ 5 studies) (n ҃ 4 studies) (n ҃ 4 studies)
Studies of Ҁ16.65 Ҁ18.38 Ҁ1.32 Ҁ5.46 Ҁ1.28 Ҁ7.44 2.68 1.75
glucomannan with (Ҁ25.12, Ҁ8.17) (Ҁ27.84, Ҁ8.91) (Ҁ4.43, 1.78) (Ҁ19.88, 8.96) (Ҁ1.63, Ҁ0.94) (Ҁ14.16, Ҁ0.72) (Ҁ2.74, 8.10) (Ҁ4.47, 7.97)
diet modification (n ҃ 8 studies) (n ҃ 5 studies) (n ҃ 5 studies) (n ҃ 7 studies) (n ҃ 6 studies) (n ҃ 5 studies) (n ҃ 3 studies) (n ҃ 3 studies)
Studies of Ҁ21.43 Ҁ14.89 Ҁ1.38 Ҁ21.76 0.11 — Ҁ3.00 2.00
glucomannan (Ҁ28.43, Ҁ14.42) (Ҁ21.32, Ҁ8.46) (Ҁ4.02, 1.26) (Ҁ40.75, Ҁ2.77) (Ҁ0.60, 0.81) — (Ҁ8.91, 2.91) (Ҁ2.65, 6.65)
without diet (n ҃ 5 studies) (n ҃ 4 studies) (n ҃ 3 studies) (n ҃ 5 studies) (n ҃ 3 studies) (n ҃ 1 study) (n ҃ 1 study)
modification
Studies of diabetic or Ҁ16.30 Ҁ15.51 Ҁ0.21 Ҁ1.59 Ҁ0.08 Ҁ1.58 Ҁ1.46 Ҁ1.46
GLUCOMANNAN AND LIPIDS, FBG, WEIGHT, AND BP

IGT subjects only (Ҁ27.91, Ҁ4.69) (Ҁ25.17, Ҁ5.86) (Ҁ-3.98, 3.57) (Ҁ39.15, 35.97) (Ҁ3.70, Ҁ3.53) (Ҁ16.05, 12.88) (Ҁ9.52, 6.59) (Ҁ4.94, 2.02)
(n ҃ 4 studies) (n ҃ 4 studies) (n ҃ 4 studies) (n ҃ 4 studies) (n ҃ 3 studies) (n ҃ 3 studies) (n ҃ 2 studies) (n ҃ 2 studies)
Studies of obese Ҁ17.88 Ҁ20.89 Ҁ1.55 Ҁ11.30 Ҁ1.30 Ҁ8.36 5.90 7.80
subjects only (Ҁ30.56, Ҁ5.20) (Ҁ32.44, Ҁ9.34) (Ҁ6.00, 2.90) (Ҁ32.21, 9.60) (Ҁ1.69, Ҁ0.91) (Ҁ18.55, 1.83) (Ҁ1.06, 12.86) (Ҁ2.77, 12.83)
(n ҃ 5 studies) (n ҃ 2 studies) (n ҃ 1 study) (n ҃ 4 studies) (n ҃ 4 studies) (n ҃ 2 studies) (n ҃ 1 study) (n ҃ 4 studies)
Studies of adults only Ҁ21.29 Ҁ15.85 Ҁ1.33 Ҁ10.77 Ҁ0.79 Ҁ7.44 0.29 1.81
(Ҁ26.87, Ҁ15.70) (Ҁ21.27, Ҁ10.43) (Ҁ3.39, 0.73) (Ҁ25.87, 4.33) (Ҁ1.53, Ҁ0.05) (Ҁ14.16, Ҁ0.72) (Ҁ4.54, 5.13) (Ҁ2.46, 6.09)
(n ҃ 10 studies) (n ҃ 8 studies) (n ҃ 7 studies) (n ҃ 9 studies) (n ҃ 9 studies) (n ҃ 5 studies) (n ҃ 4 studies) (n ҃ 4 studies)
(Continued)
1173

Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017

1174 SOOD ET AL

Administration (FDA), has been shown to decrease body weight

A DerSimonian and Laird random-effects model was used to calculate weighted mean differences and their 95% CIs (in parentheses). To convert values for cholesterol from mg/dL to mmol/L, multiply by
0.0286; to convert values for TG from mg/dL to mmol/L, multiply by 0.01129. DBP, diastolic blood pressure; FBG, fasting blood glucose; C, cholesterol; IGT, impaired glucose tolerance; SBP, systolic blood
by 앒5% in the first 16 wk of treatment (27). Therefore, based on

10.50) (n ҃ 2
(n ҃ 1 study)

4.82 (Ҁ0.86,
7.80 (Ҁ2.77,

studies)
12.82)
current data, glucommanan’s effect on weight could be described

—
—
as mild, as could its effects on FBG.
Only a limited number of studies have evaluated the short-term
safety and tolerability of glucomannan. On the basis of the avail-
able data, glucomannan appears to be well tolerated; adverse

1.25 (Ҁ7.46, 9.97)

gastrointestinal effects such as loose stools, flatulence, diarrhea,

(n ҃ 2 studies)
(n ҃ 1 study)
5.90 (Ҁ1.06, and abdominal discomfort are the most commonly reported (15).
12.86)
SBP

Of potential greater concern are case reports of esophageal ob-

—
—

struction resulting from swelling of glucomannan tablets that

have been reported in the literature (28). In response to these and
similar cases related to glucomannan containing “jelly cup type
candies,” the FDA issued warnings about consuming certain
Ҁ8.17 (Ҁ16.84,
Ҁ2.77 (Ҁ10.37,

(n ҃ 3 studies)
(n ҃ 2 studies)

products containing glucomannan stating they felt it posed a

serious choking risk, particularly to infants, children, and the
0.50)
4.83)
FBG

—
—

elderly (29). A paucity of data exist on the long-term safety of

Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017

glucomannan.
Some limitations of this meta-analysis should be noted. First,
we included both crossover and parallel studies. Crossover stud-
ies are often subject to additional biases and are thus thought to
(n ҃ 7 studies)
(n ҃ 4 studies)

Ҁ0.80 (Ҁ1.61,
Ҁ1.25 (Ҁ1.62,
Body weight

have a lower internal validity than an equivalent parallel study,

Ҁ0.09)
Ҁ0.87)

particularly when studies have insufficient washout periods. Al-

—
—

though we believed that it was advantageous to include crossover

trials to provide additional power to our meta-analysis, we did not
include trials that did not explicitly state the presence and dura-
tion of the washout period or studies that had a washout period
쏝2 wk in duration. In addition, we conducted sensitivity analysis
Ҁ11.52 (Ҁ22.79,

(n ҃ 10 studies)
(Ҁ34.30, 12.90)

Ҁ6.97 (Ҁ23.39,
(n ҃ 3 studies)

(n ҃ 6 studies)

whereby we reanalyzed our results including only those studies

Ҁ10.80

Ҁ0.24)
9.45)

that used the more rigorous parallel design. No noteworthy

TG

changes in any of the study endpoints were noted after conduct-

ing this sensitivity analysis, which strengthened our confidence
in the conclusions of the meta-analysis. Second, as with any
meta-analysis, the potential for publication bias is a concern.
Publication bias is defined as “the tendency on the parts of in-
(n ҃ 6 studies)
(n ҃ 3 studies)

Ҁ1.49 (Ҁ3.61,
(Ҁ11.19, 7.33)

Ҁ1.68 (Ҁ5.27,
(n ҃ 1 study)

vestigators, reviewers, and editors to submit or accept manu-

HDL-C

Ҁ1.93

0.64)
1.91)

scripts for publication based on the direction or strength of the

study findings.” Although visual inspection of our meta-
analysis’ funnel plot could not rule out the presence of publica-
tion bias, a review of Egger’s weighted regression statistics and
Trim and Fill analyses showed that it was unlikely that publica-
Ҁ15.61 (Ҁ21.27,
Ҁ19.08 (Ҁ26.85,

(n ҃ 7 studies)
(n ҃ 4 studies)
(Ҁ47.50, 8.06)

tion bias significantly affected our study results.

(n ҃ 1 study)
Ҁ19.72
LDL-C

Ҁ9.95)
Ҁ9.52)

CONCLUSIONS
Glucomannan appears to beneficially affect total cholesterol,
LDL cholesterol, triglycerides, body weight, and FBG, but not
pressure; TC, total cholesterol; TG, triglycerides.

HDL cholesterol or blood pressure. Larger individual studies

Ҁ19.25 (Ҁ24.52,
Ҁ12.13 (Ҁ20.62,

(n ҃ 11 studies)
(n ҃ 3 studies)

(n ҃ 6 studies)
(Ҁ22.28, 0.60)

following patients for longer periods of time and evaluating both

Ҁ13.98)
Ҁ10.84

Ҁ3.85)

safety and efficacy are warranted and needed.

TC

The authors’ responsibilities were as follows—NS, WLB, and CIC: re-

sponsible for analyzing the data, interpreting the data and results, and writing
the manuscript; and WLB and CIC: responsible for formulating the research
question, conducting the literature search, interpreting the data and results,
TABLE 2 (Continued)

쏝10 g glucomannan/d

and writing the manuscript. We certify that none of the material in this
쏝3 g glucomannan/d

moderate-to-high

manuscript was previously published, and we have no conflicts to declare

Studies of children

(excluding high

germane to this manuscript.

(excluding

doses)
doses)

REFERENCES
only

1. Grundy SM, Cleeman JI, Daniels SR, et al. Diagnosis and management
of the metabolic syndrome: an American Heart Association/National
 GLUCOMANNAN AND LIPIDS, FBG, WEIGHT, AND BP
Heart, Lung, and Blood Institute Scientific Statement. Internet: http://
circ.ahajournals.org/cgi/content/full/112/17/e285 (accessed 20 Decem-
ber 2007).
2. Doi K. Effect of konjac fibre (glucomannan) on glucose and lipids. Eur
J Clin Nutr 1995;49(suppl 3):S190 –7.
3. Chearskul S, Sangurai S, Nitiyanant W, Kriengsinyos W, Kooptiwut S,
Harindhanavudhi T. Glycemic and lipid responses to glucomannan in
Thais with type 2 diabetes mellitus. J Med Assoc Thai 2007;90:2150 –7.
4. Wood RJ, Fernandez ML, Sharman MJ, et al. Effects of a carbohydrate-
restricted diet with and without supplemental soluble fiber on plasma
low-density lipoprotein cholesterol and other clinical markers of cardio-
vascular risk. Metab Clin Exp 2007;56:58 – 67.
5. Yoshida M, Vanstone CA, Parsons WD, Zawistowski J, Jones PJH.
Effect of plant sterols and glucomannan on lipids in individuals with and
without type II diabetes. Eur J Clin Nutr 2006;60:529 –37.
6. Martino F, Martino E, Morrone F, Carnevali E, Forcone R, Niglio T.
Effect of dietary supplementation with glucomannan on plasma total
cholesterol and low density lipoprotein cholesterol in hypercholester-
olemic children. Nutr Metab Cardiovasc Dis 2005;15:174 – 80.
7. Birketvedt GS, Shimshi M, Thom E, Florholmen J. Experiences with
three fiber supplements in weight reduction. Med Sci Monit 2005;11:
5– 8.
8. Vuskan V, Sievenpiper JL, Owen R, et al. Beneficial effects of viscous
dietary fiber from konjac-mannan in subjects with the insulin resistance
syndrome. Diabetes Care 2000;23:9 –14.
9. Vuskan V, Jenkins DJA, Spadafora P, et al. Konjac-mannan (glucoman-
nan) improves glycemia and other associated risk factors for coronary
heart disease in type 2 diabetes. A randomized controlled metabolic trial.
Diabetes Care 1999;22(6):913–9.
10. Arvill A, Bodin L. Effect of short-term ingestion of konjac glucomannan
on serum cholesterol in healthy men. Am J Clin Nutr 1995;61:585–9.
11. Cairella M, Marchini G. [Evaluation of the action of glucomannan on
metabolic parameters and on the sensation of satiation in overweight and
obese patients.] Clin Ter 1995;146:269 –74 (in Italian).
12. Vido L, Faccin P, Antonello I, Gobber D, Rigon F. Childhood obesity
treatment: double blinded trial on dietary fibres (glucomannan) versus
placebo. Paediatr Paedol 1993;28:133– 6.
13. Livieri C, Novazi F, Lorini R. [The use of highly purified glucomannan-
based fibers in childhood obesity.] Ped Med Chir 1992;14:195– 8.
14. Zhang MO, Huang CY, Wang X, Hong JR, Peng SS. The effect of foods
containing refined konjac meal on human lipid metabolism. Biomed
Environ Sci 1990;3:99 –105.
15. Venter CS, Kruger HS, Vorster HH, Serfontein WJ, Ubbink JB, De
1175
Villiers LS. The effects of the dietary fibre component konjac-
glucomannan on serum cholesterol levels of hypercholesterolemic sub-
jects. Hum Nutr:Food Serv Nutr 1987;41F:55– 61.
16. Walsh DE, Yaghoubian V, Behforooz A. Effect of glucomannan on
obese patients: a clinical study. Int J Obes 1984;8:289 –93.
17. Vita PM, Restelli A, Caspani P, Klinger R. [Chronic use of glucomannan
in the dietary treatment of severe obesity.] Minerva Med 1992;83(3):
135–9 (in Italian).
18. Chen HL, Huey-Herng Sheu W, Tai TS, Liaw YP, Chen YC. Konjac
supplement alleviated hypercholesterolemia and hyperglycemia in type
2 diabetic subjects—a randomized double-blind trial. J Am Coll Nutr
2003;22:36 – 42.
19. Biancardi G, Palmiero L, Ghirardi PE. Glucomannan in the treatment of
overweight patients with osteoarthrosis. Curr Ther Res 1989;46:908 –
12.
20. Follman D, Elliott P, Suh I, Cutler J. Variance imputation for overviews
of clinical trials with continuous response. J Clin Epidemiol 1992;45:
769 –31.
21. FDA. FDA allows whole oat foods to make health claim on reducing the
risk of heart disease. Internet: http://www.cfsan.fda.gov/앑lrd/tpoats.
Downloaded from ajcn.nutrition.org at UNIVERSITY OF GRONINGEN on June 8, 2017
html. (accessed 11 April 2008).
22. Brown L, Rosner B, Willett WW, Sacks FM. Cholesterol-lowering ef-
fects of dietary fiber: a meta-analysis. Am J Clin Nutr 1999;69:30 – 42.
23. Grundy SC, Cleeman JI, Merz CNB, et al. Implications of recent clinical
trials for the National Cholesterol Education Program Adult Treatment
Panel III guidelines. Circulation 2004;110:227–39.
24. Anderson JM, Allgood LD, Lawrence A, et al. Cholesterol-lowering
effects of psyllium intake adjunctive to diet therapy in men and women
with hypercholesterolemia: meta-analysis of 8 controlled trials. Am J
Clin Nutr 2000;71:472–9.
25. Federal Trade Commission. FTC settles claims with marketers of Fib-
erthin and Propolene. Internet: http://www.ftc.gov/opa/2005/06/fib-
erthin.shtml (accessed 31 December 2007).
26. Lipozene™ home page. Internet: http://www.lipozene.com/inner.
php?page҃63 (accessed 31 December 2007).
27. Anderson JW. Orlistat for the management of overweight individuals
and obesity: a review of potential for the 60-mg, over-the-counter dos-
age. Expert Opin Pharmacother 2007;8:1173– 42.
28. Esophageal obstruction from hygroscopic pharmacobezoar containing
glucomannan. Clin Tox 2007;45:80 –2.
29. FDA. FDA warns consumers about imported jelly cup type candy that
poses a potential choking hazard. Internet: http://www.fda.gov/oc/ po/
firmrecalls/topics/konjac.html (accessed 31 December 2007).