Heymsfield SB, Allison DB, Vasselli JR, Pietrobelli A, Greenfield D, Nunez C. Garcinia cambogia (Hydroxycitric Acid) as a Potential Antiobesity AgentA Randomized Controlled Trial. JAMA. 1998;280(18):1596–1600. doi:10.1001/jama.280.18.1596
From the Department of Medicine, Obesity Research Center, St Luke's–Roosevelt Hospital, Columbia University College of Physicians and Surgeons, New York, NY.
Context.— Hydroxycitric acid, the active ingredient in the herbal compound Garcinia cambogia, competitively inhibits the extramitochondrial
enzyme adenosine triphosphate–citrate (pro-3S)-lyase. As a citrate cleavage
enzyme that may play an essential role in de novo lipogenesis inhibition, G cambogia is claimed to lower body weight and reduce fat
mass in humans.
Objective.— To evaluate the efficacy of G cambogia for
body weight and fat mass loss in overweight human subjects.
Design.— Twelve-week randomized, double-blind, placebo-controlled trial.
Setting.— Outpatient weight control research unit.
Participants.— Overweight men and women subjects (mean body mass index [weight in kilograms
divided by the square of height in meters], approximately 32 kg/m2).
Intervention.— Subjects were randomized to receive either active herbal compound (1500
mg of hydroxycitric acid per day) or placebo, and both groups were prescribed
a high-fiber, low-energy diet. The treatment period was 12 weeks. Body weight
was evaluated every other week and fat mass was measured at weeks 0 and 12.
Main Outcome Measures.— Body weight change and fat mass change.
Results.— A total of 135 subjects were randomized to either active hydroxycitric
acid (n=66) or placebo (n=69); 42 (64%) in the active hydroxycitric acid group
and 42 (61%) in the placebo group completed 12 weeks of treatment (P=.74). Patients in both groups lost a significant amount of weight
during the 12-week treatment period (P<.001);
however, between-group weight loss differences were not statistically significant
(mean [SD], 3.2 [3.3] kg vs 4.1 [3.9] kg; P=.14).
There were no significant differences in estimated percentage of body fat
mass loss between treatment groups, and the fraction of subject weight loss
as fat was not influenced by treatment group.
Conclusions.— Garcinia cambogia failed to produce significant
weight loss and fat mass loss beyond that observed with placebo.
EXCESSIVE ADIPOSITY and its concomitant health risks are among the most
common conditions managed by health care practitioners. The limited long-term
effectiveness of conventional weight management, including behavioral therapy,1 is the impetus of major efforts aimed at developing
alternative pharmacologic2 and surgical weight
reduction treatment strategies.3 A rapidly
growing therapeutic area, and one widely embraced by the general public, is
the use of herbal weight loss products.
An herb-derived compound, hydroxycitric acid, is now incorporated into
many commercial weight loss products. Obtained from extracts of related plants
native to India, mainly Garcinia cambogia and Garcinia indica , hydroxycitric acid was first identified
by Watson and Lowenstein4,5 in
the late 1960s as a potent competitive inhibitor of the extramitochondrial
enzyme adenosine triphosphate–citrate (pro-3
S)-lyase. These investigators and others subsequently demonstrated both in
vitro and in vivo that hydroxycitric acid in animals not only inhibited the
actions of citrate cleavage enzyme and suppressed de novo fatty acid synthesis,6 but also increased rates of hepatic glycogen synthesis,7 suppressed food intake,8
and decreased body weight gain.9
Although hydroxycitric acid appears to be a promising experimental weight
control agent, studies in humans are limited and results have been contradictory10- 14
(also R. Ramos, J. Flores Saenz, F. Alarcon, unpublished data, 1996, and G.
Kaats, D. Pullin, L. Parker, S. Keith, unpublished data, 1996). Supporting
evidence of human hydroxycitric acid efficacy for weight control is based
largely on studies with small sample sizes,11,12
studies that failed to include a placebo-treated group,10
and use of inaccurate measures of body lipid change.12
Although hydroxycitric acid effectiveness remains unclear, at least 14 separate
hydroxycitric acid–containing products are presently sold over-the-counter
to consumers.15 This investigation was designed
to overcome limitations of earlier studies and examine the effectiveness of
hydroxycitric acid for weight loss and fat mass reduction in a rigorous controlled
We tested 2 primary hypotheses in a randomized, double-blind, placebo-controlled
trial: (1) G cambogia produces a greater reduction
in body weight than placebo, and (2) G cambogia produces
a greater reduction in total body fat mass than placebo. Advertisements were
placed in local newspapers, and overweight subjects who responded and met
entry criteria during a telephone screening interview were scheduled for a
baseline visit. The evaluation included a physical examination, electrocardiogram,
and screening blood studies. Subjects meeting entry criteria were seen within
2 weeks for randomization at treatment week 0. Subjects were assigned to placebo
or active compound with equal probability through a random number generator.
The protocol with active herbal compound included G cambogia extract (50% hydroxycitric acid by chemical analysis), taken
3 times daily as two 500-mg caplets 30 minutes before meal ingestion. Total
daily dose was G cambogia extract, 3000 mg, and hydroxycitric
acid, 1500 mg. Placebo-treated subjects followed an identical protocol in
which active compound was replaced with inert ingredients. Subjects taking
active compound or placebo were provided a high-fiber, 5040-kJ/d diet plan,
with 20%, 50%, and 30% of energy as fat, carbohydrate, and protein, respectively.
The recommended daily food provision was divided into 3 meals with an evening
snack. Subjects were asked to maintain a stable physical activity level and
return for evaluation every 2 weeks for a total treatment interval of 12 weeks.
Body weight was measured at each visit, and clinical information, including
potential herb or weight loss adverse effects, was obtained. Biweekly pill
counts and diaries were used to check patient medication compliance. Diet
compliance was not quantitatively monitored during the study.
The study was approved by the institutional review board of St Luke's–Roosevelt
Hospital Center, New York, NY, and all subjects gave written consent prior
Subjects were overweight but otherwise healthy adults aged 18 to 65
years who had a body mass index (BMI, defined as weight in kilograms divided
by the square of height in meters) of more than 27 kg/m2and at
most 38 kg/m2. Subjects were excluded if they were pregnant, had
any clinically significant medical condition, were taking prescription medications
or appetite suppressants on a regular basis, had a history of alcohol or other
drug abuse, were allergic to any of the study products, or had dieted with
weight loss in the past 6 months.
Body weight and height were measured to the nearest 0.1 kg and 0.5 cm
using a digital scale (Weight Tronix, New York, NY) and stadiometer (Holtain,
Crosswell, Wales), respectively. Total body fat mass was measured at baseline
and at the 12-week visit using several different procedures.
A pencil-beam dual-energy x-ray absorptiometry (DXA) scanner (Lunar
DPX, Madison, Wis) was used to estimate total body fat mass. Subjects completed
the slow-mode whole body scan and fat mass estimates were provided by Lunar,
Version 3.6g, software.16 The technical error
of DXA percentage fat mass estimates in our laboratory is 3.1%.17
The remaining body fat mass measurement methods used in our laboratory for
this study included underwater weighing,18
skinfold thicknesses,19 and bioimpedance analysis.20
Based on previous research,1 we estimated
that a study that included at least 30 completed subjects in each of 2 groups
would have more than 80% power at the 2-tailed α level of .05 to detect
any significant differences in body weight.
The 2 study hypotheses were tested in separate sets of statistical analyses.
Statistical models were used in which the outcome variable, either loss of
body weight or percentage of fat mass, was set as dependent variable and assigned
treatment and other covariates were set as independent variables in an intent-to-treat
analysis.21 Within the intent-to-treat analysis,
missing data due to measurement failure or subject dropout were imputed by
carrying the last observation forward (LOCF).22
The baseline value of the dependent variable (ie, initial body weight or percentage
of fat mass) served as a potential independent variable in each analysis.
Patient age and sex also served as additional independent variables. All analyses
were conducted at the 2-tailed α level of .05.
For each of the 2 dependent variables, a set of secondary analyses were
conducted, including (1) evaluation of completers only; (2) imputation of
all missing data with a regression procedure rather than the LOCF; (3) imputation
of missing data using the EM23 algorithm rather
than the LOCF; (4) use of weight loss slopes as outcomes24
rather than the simple baseline to final measurement change when more than
2 time points for weight were available; (5) performance of a full repeated-measures
analysis of variance using all time points; and (6) performance of a multivariate
analysis of covariance using all time points simultaneously in the statistical
model. In no case did any of these secondary sensitivity analyses lead to
different conclusions than the primary LOCF intent-to-treat analysis. We therefore
report only the results of the primary intent-to-treat analysis.
At baseline, DXA readings were unavailable for several subjects who
had technically poor scans or who were evaluated during a brief period in
which the DXA system was undergoing repair. However, each of these subjects
had 1 or more measurements of fat mass taken with the other techniques mentioned
herein and summarized in earlier articles.16- 20
Estimates of total body fat mass for these subjects by DXA were inferred using
single imputation plus random error models based on multiple regression analysis
of all other available measurements of fat mass for that subject, as described
by Graham et al.25 Similarly, several subjects
completed the entire course of treatment and received some measurement of
body fat mass after treatment but not by DXA. For these subjects, estimates
of total body fat mass by DXA also were imputed using the same statistical
methods and the other available measurements of body fat mass.
The purported fat-mobilizing properties of hydroxycitric acid were evaluated
by computing the slope of change in fat mass vs change in body weight for
the 2 treatment groups. Assuming approximately a zero intercept for this relation,
the anticipated regression line slopes should approach 0.7 to 0.8, the generally
acknowledged fraction of weight loss as fat mass in obesity trials.26 Promotion of fat mass loss by active hydroxycitric
acid would be associated with an increased fraction of weight loss as fat
Group results are expressed as mean (SD) in text and tables. Data were
analyzed using the statistical programs SPSSWIN, Version 7.5, and SPSSMVA,
Version 7.5 (SPSS Inc, Chicago, Ill).
At baseline, 180 moderately overweight subjects were screened and, of
those, 135 were randomized to placebo and active compound (Table 1 and Figure 1).
There were 69 subjects (BMI, 31.9 [3.1] kg/m2) in the placebo-treated
group (14 men and 55 women) and 66 subjects (BMI, 31.3 [2.8] kg/m2)
in the G cambogia– treated group (5 men and
Of the 69 placebo-treated subjects, 42 (61%) completed the 12-week protocol.
The reasons for subject withdrawal (27 cases) are summarized in Figure 1. Of the 66 subjects randomized to active compound, 42
(64%) completed the 12 weeks of treatment. The reasons for subject withdrawal
from this group (24 cases) are also summarized in Figure 1. There were no significant differences in age, body weight,
or BMI between subjects who withdrew from the study and those who completed
the 12-week protocol. There was also no significant difference between the
2 groups in the proportion of subjects who completed the entire course of
treatment (χ2=0.11, P =.74). Among
subjects completing the 12 weeks of treatment, medication compliance was 88.6%
(10.9%) and 92.1% (10.0%) in the treatment and placebo groups, respectively
Primary Analysis.— The weight loss curves for placebo and treatment groups are shown in Figure 2 for subjects in the intent-to-treat
analysis with LOCF. The estimated mean (SD) [median (interquartile range)]
weight loss for the placebo group was 4.1 (3.9) [3.9 (4.7)] kg and for the
treatment group was 3.2 (3.3) [2.6 (4.1)] kg. The weight loss within each
group was significantly different from baseline (t134=11.795, P <.001), although between-group
weight loss differences were not statistically significant (t133=1.474, P =.14). Body weight
change differences remained nonsignificant after controlling for patient starting
weight, sex, and age. Assumptions of the applied parametric statistical analysis
such as homogeneity of variance and normality of residuals were tested and
no meaningful violations were detected. Given the lack of significant findings,
questions of statistical power are important. Therefore, using the observed
distributions of weight change and the within-group SD thereof, we estimated
that the power of the current study to detect differences between the treatment
and placebo groups in terms of weight change was 89% to detect a between-group
difference in weight loss as small as 2 kg at the 2-tailed α level of
Secondary Analyses.— In no case did any secondary analysis indicate any statistically significant
effect for the active compound to produce more weight loss than placebo.
Primary Analysis.— Results for body fat mass analysis were imputed for 9 baseline and 4
post–weight loss subjects. With the LOCF intent-to-treat analysis, the
estimated mean (SD) [median (interquartile range)] percentage of body fat
mass loss for the placebo group was 2.16% (2.06%) [2.20% (2.7%)] and the estimated
percentage of fat mass loss for the treatment group was 1.44% (2.15%) [1.60%
(1.9%)]. This difference was tested using the Welch test because the variances
were significantly heterogeneous by the Levene test (P
for variance heterogeneity=.03). Using the Welch test, the placebo and treatment
group mean differences were not statistically significant (t129=1.7, P =.08). This finding
was consistent with that of the ordinary t test (t132=1.78, P =.08).
Using analysis of covariance with age, sex, and pretest percentage of fat
mass as covariates, the percentage of fat mass differences also was nonsignificant
Secondary Analyses.— As for weight loss, all of the secondary analyses were consistent with
the primary analysis. That is, in no case did analysis indicate any statistically
significant effect for the active compound to produce a different percentage
of body fat mass loss than the placebo.
Examination of the change in fat mass relative to change in body weight
derived using least squares regression analysis for all subjects combined
resulted in the relation, Δfat mass (kg)=0.77 × Δbody weight
(kg) − 0.44, with r=0.89 and P <.001. The association was not changed significantly (P>.91) by adding treatment group as a second independent variable,
even after adjusting for 3 additional potential covariates: initial body weight,
sex, and age.
No patient was removed from the study protocol for a treatment-related
adverse event, and the number of reported adverse events was not significantly
different between the placebo and treatment groups (eg, headache, 12 vs 9,
respectively; upper respiratory tract symptoms, 13 vs 16, respectively; and
gastrointestinal tract symptoms, 6 vs 13, respectively).
In 1883 von Lippmann isolated hydroxycitric acid, a minor constituent
of sugar beets.27 More than half a century
later, in 1941, Martius and Maué28 discovered
that the (+) isomer of a racemic hydroxycitric acid mixture is attacked by
the enzyme isocitrate dehydrogenase. The (−) hydroxycitric acid isomer
of hydroxycitric acid was first isolated by Lewis and Neelakantan in 1964,29 and by 1969 Watson and colleagues5
reported the powerful inhibition by (−) hydroxycitric acid of citrate
cleavage enzyme. Evidently, the additional hydroxyl group's steric position,
compared with citric acid, enhances its binding affinity and competitively
inhibits catalytic action by the enzyme. Citrate, entering the cytoplasm from
mitochondria, cannot be cleaved to release acetyl coenzyme A, the substrate
for de novo fatty acid synthesis. Despite these century-old, well-grounded
observations, there has been little effort to critically test the basic assumption
underlying therapeutic use of hydroxycitric acid in overweight humans: that
hydroxycitric acid inhibition of lipid synthesis will significantly reduce
body fat mass beyond that observed with a placebo capsule.
The present study, carried out during a 12-week evaluation period and
using accepted experimental design and in vivo analytic methods, failed to
support the hypothesis that hydroxycitric acid as prescribed promotes either
additional weight or fat mass loss beyond that observed with placebo. Specifically,
body weight and fat mass change during the 12-week study period did not differ
significantly between placebo and treatment groups. These results apply to
both the primary and secondary statistical analyses. Additionally, there were
no observed selective fat-mobilizing effects specifically attributable to
the active agent, hydroxycitric acid.
Seven earlier G cambogia trials have appeared
in peer-reviewed literature,11,14
as abstracts,12,13 and in industrial
publications as an open-label study10 and randomized
controlled trials.11- 14
We chose to collectively review these studies even though G cambogia typically was used in combination with other ingredients
for the claimed purpose of enhancing weight loss.
Of the 7 studies reviewed, 5 reported significant (P<.05) effects of G cambogia alone or in
combination with other ingredients on body weight or fat mass loss in overweight
humans (Table 2). These earlier
studies all have limitations when specifically considering G cambogia as a weight loss agent, including lack of placebo control
or double-blinding in 1 study,10 coadministration
of G cambogia in combination with other potentially
active ingredients in 5 studies,10,11,13,14
use of an inaccurate body composition method (near-infrared interactance)12 in 1 study, and failure as of yet to publish study
results in peer-reviewed literature in all but 213,14
of the 7 studies. However, our present investigation, carried out using accepted
clinical trial design procedures and applying accurate body composition methods,
failed to support a specific weight loss effect of G cambogia administered as recommended. The present 12-week study period also
exceeded in duration all previous study treatment periods, which ranged from
4 to 8 weeks.
In our present investigation we failed to detect a weight loss or fat-mobilizing
effect of active herb. The question therefore arises whether there exist conditions
differing from those used in the present study that might support hydroxycitric
acid efficacy. The 5040-kJ/d low-fat diet recommended in our current study
was intended to mimic diets commonly prescribed as a component of weight control
programs. The possibility exists that the lipid synthesis–inhibiting
properties of hydroxycitric acid may be more evident in subjects relapsing
following a failed diet attempt, particularly if high-carbohydrate foods are
Another concern is related to the timing and dosage considerations of
hydroxycitric acid. Sullivan and colleagues31
showed that the effects of hydroxycitric acid in animals depend on time of
administration in relation to a meal, with hydroxycitric acid maximally effective
when administered 30 to 60 minutes prior to feeding. The approach used in
our study and the others we reviewed suggested hydroxycitric acid ingestion
about 30 minutes prior to meal intake, the lower end of the maximally effective
range. A related concern is that hydroxycitric acid provided in divided doses
also was found to be more effective than the same amount given as a single
dose.8 Although divided doses typically are
used in weight loss protocols, human doses ranging between 750 and 1500 mg/d
of hydroxycitric acid are at the extreme low end of the in vivo dose-response
range established by Sullivan and colleagues.32
Thus, in light of the many requirements for its effective use, it seems unlikely
that the maximal effects of hydroxycitric acid will be realized in human weight
loss studies unless treatment conditions are well defined and patient diet
and medication compliance are tightly monitored.
Our study explored product safety only in the form of clinical evaluations
and reported adverse events. No significant differences were observed between
placebo and treatment groups in number of reported adverse events and no subjects
were removed from the study for a treatment-related adverse event. Additional
studies, potentially with larger subject groups, are needed to gather specific
information on the long-term safety of G cambogia.
An important concern in all pharmacological trials, particularly those
in which herbal products are evaluated, is the amount and bioavailability
of the active agent. As standard procedure, we confirmed the presence and
quantity of hydroxycitric acid in the supplied capsules using an independent
testing laboratory. However, we did not measure hydroxycitric acid blood levels
or evaluate tissue or cytosolic citrate-cleavage enzyme activity. Although
the format of our experiment closely resembles current use of G cambogia as a weight loss product, our conclusions should not be
interpreted as a failure to support the validity of the biochemical effects
of hydroxycitric acid identified by earlier investigators.
In conclusion, our study evaluated the hypothesis that the active ingredient
of G cambogia, hydroxycitric acid, has beneficial
weight and fat mass loss effects. Our findings, obtained in a prospective,
randomized, double-blind study, failed to detect either weight loss or fat-mobilizing
effects of hydroxycitric acid beyond those of placebo. These observations,
the first, to our knowledge, to appear in a peer-reviewed article using currently
accepted experimental and statistical methods, do not support a role as currently
prescribed for the widely used herb G cambogia as
a facilitator of weight loss.