Topiramate: Long-Term Maintenance of
Weight Loss Induced by a Low-Calorie Diet in
Arne Astrup,* Ian Caterson,† Pierre Zelissen,‡ Bernard Guy-Grand,§ Michele Carruba,¶ Brian Levy,**
Xiang Sun,** and Martin Fitchet** for the OBES-004 Study Group
ASTRUP, ARNE, IAN CATERSON, PIERRE ZELISSEN,
BRIAN LEVY, XIANG SUN, AND MARTIN FITCHET.
Topiramate: long-term maintenance of weight loss induced
by a low-calorie diet in obese subjects. Obes Res. 2004;12:
Objective: To examine the safety and efficacy of topiramate
(TPM) for maintaining weight following a low-calorie diet.
Research Methods and Procedures: Obese subjects (30 ?
BMI ? 50 kg/m2) 18 to 75 years old received a low-calorie
diet for 8 weeks. Those who lost ?8% of their initial weight
received TPM (96 or 192 mg/d) or placebo; all were on a
lifestyle modification plan. Sixty weeks of medication were
planned. Sponsor ended study early to develop a new con-
trolled-release formulation with the potential to enhance
tolerability and simplify dosing in this patient population.
Efficacy was analyzed in subjects who completed 44 weeks
of treatment before study termination.
Results: Of the 701 subjects enrolled, 80% lost ?8% of
their initial body weight and were randomized; 293 were
analyzed for efficacy. Most withdrawals were due to pre-
mature termination of the study. Subjects receiving TPM
lost 15.4% (96 mg/d) and 16.5% (192 mg/d) of their enroll-
ment weight by week 44, compared with 8.9% in the pla-
cebo group (p ? 0.001). Subjects on TPM continued to lose
weight after the run-in, whereas those on placebo regained
weight. Significantly more TPM subjects lost 5%, 10%, or
15% of their randomization weight than placebo. Most
adverse events were related to the central nervous system.
Discussion: During a treatment period of 44 weeks, TPM
was generally well tolerated, and subjects maintained
weight loss initially achieved by a low-calorie diet—and
produced additional clinically significant weight loss be-
yond that achieved by a low-calorie diet.
Key words: topiramate, weight loss, diet, drug therapy,
Obesity is a serious and growing global problem (1). One
large study in the United States reported a prevalence of
19.8% in 2000, a 61% increase from 1991 (2), whereas
another large study found an increase from 22.9% in 1988
to 1994 to 30.5% in 1999 to 2000 (3). Two large-scale
European surveys in the middle and late 1980s reported
obesity prevalence ranging from 9% to 18% for men and
from 12% to 24% for women (4). A 1997 World Health
Organization (WHO)1survey found rates of obesity in most
countries of Western Europe ranging from 10% to 25% (5).
Obesity in the rest of the world has been less well docu-
mented, but a prevalence of 5% has been reported in certain
urban areas of Asia, with an increasing prevalence noted
among rural populations, as well (6).
This epidemic of obesity has serious consequences for
public health, increasing the risk of type 2 diabetes (7,8),
hypertension (8–10), dyslipidemia (8,9), coronary artery
Received for review December 1, 2003.
Accepted in final form July 28, 2004.
The costs of publication of this article were defrayed, in part, by the payment of page
charges. This article must, therefore, be hereby marked “advertisement” in accordance with
18 U.S.C. Section 1734 solely to indicate this fact.
*Royal Veterinary and Agricultural University, Frederiksberg, Denmark; †School of Mo-
lecular and Microbial Biosciences, University of Sydney, Sydney, Australia; ‡Department of
Internal Medicine/Endocrinology, University Medical Center, Utrecht, The Netherlands;
§University Pierre et Marie Curie, Paris, France; ¶Department of Preclinical Sciences,
University of Milan, Milan, Italy; and **Johnson and Johnson Pharmaceutical Research and
Development, LLC, Raritan, New Jersey.
Address correspondence to Arne Astrup, The Department of Human Nutrition and LMC,
The Royal Veterinary and Agricultural University, 30 Rolighedsvej, 1958 Frederiksberg C,
Copyright © 2004 NAASO
1Nonstandard abbreviations: WHO, World Health Organization; TPM, topiramate; CNS,
central nervous system; MITT, modified intent-to-treat; LOCF, last observation carried
forward; ITT, intent-to-treat.
1658OBESITY RESEARCH Vol. 12 No. 10 October 2004
disease (8,11), ischemic stroke (12), and osteoarthritis (13).
For both sexes and at all ages, mortality from all obesity-
related causes, including cancer and cardiovascular disease,
increases with the severity of the obesity (14).
Dietary restriction, especially in combination with exer-
cise programs and behavioral modification, remains the
preferred initial treatment approach to weight reduction in
obese subjects. Such approaches are often effective in the
short term, but long-term results have generally been dis-
appointing. Those who lose weight through dietary restric-
tion and behavioral modification typically regain about two-
thirds of the lost weight within 1 year and almost all of it by
the end of 5 years (15). In the Diabetes Prevention Program,
an intensive lifestyle treatment led to a weight loss of ?6.6
kg in the 1st year; after the 1st year, weight started to
increase gradually, resulting in approximately a 4-kg weight
loss maintained at 4 years. Nonetheless, this weight loss was
sufficient to reduce the incidence of type 2 diabetes in this
high-risk population by 58% (16). A number of studies have
examined the use of pharmacological therapies to assist
weight loss and maintenance, including some in which
medication effects were followed for as long as 2 years
(17–21). To the best of our knowledge, however, only one
previous study has examined the use of pharmacological
therapy for maintenance and possible promotion of further
weight loss in subjects who lost a predefined amount of
weight by nonpharmacological means (21).
Preclinical studies of topiramate (TPM) in several animal
models have suggested that TPM is likely to be an effective
weight loss treatment (22). A subsequent 6-month dose-
ranging study in which subjects were randomly assigned to
either placebo or TPM in doses ranging from 64 to 384
mg/d (23) reported significantly greater weight loss on all
doses compared with placebo.
The objective of the current study was to investigate the
efficacy and safety of TPM for maintaining weight loss in
obese subjects who had initially lost weight through a
low-calorie diet without pharmacological assistance. The
study was initially planned as 60-week duration on study
Research Methods and Procedures
Subjects were eligible for enrollment if they were be-
tween 18 and 75 years of age and had a BMI ?30 to ?50
kg/m2. Subjects with a BMI of ?30 kg/m2to ?50 kg/m2
were eligible if they had controlled hypertension and/or
dyslipidemia with a stable medication regimen. Subjects
with diabetes were ineligible unless they were newly diag-
nosed with diabetes by means of oral glucose tolerance test
at the enrollment visit, and antidiabetic medication was not
deemed necessary by the investigator. Subjects were also
ineligible if they had significant cardiovascular, hepatic, or
renal disease, a history or family history of kidney stones,
uncontrolled thyroid disease, or significant central nervous
system (CNS)-related or psychiatric disorders. To be eligi-
ble, a subject’s weight had to have been stable for at least 3
months and smoking habits stable for at least 2 months
before enrollment. Female subjects of childbearing potential
were required to use an approved method of contraception.
Subjects were recruited from centers in Europe and Austra-
lia. The study was carried out from August 2000 to June
2002. It was conducted in accordance with the Declaration
of Helsinki and International Conference on Harmonisation
of Technical Requirements for Registration of Pharmaceu-
ticals for Human Use (ICH) Good Clinical Practice and
approved by Ethics Committees at all sites. All subjects
provided written informed consent before enrollment.
Medications that were not allowed during the study pe-
riod included antiseizure medications, antiparkinsonian
medications, antidepressants, tranquilizers, sedatives and
agents that might affect weight or food absorption, such as
glucocorticoids, anorexigenic agents (prescription or over-
the-counter), orlistat, nonfiber laxatives, and thyroid hor-
mone (except as part of a stable treatment regimen).
The study was a randomized, double-blind, placebo-con-
trolled, parallel-group multicenter trial. It was originally
designed to last a total of 74 weeks: an 8-week nonpharma-
cological low-calorie (800 to 1000 kcal/d) weight-loss
run-in phase, an 8-week titration phase, a 52-week mainte-
nance phase, and a 6-week drug taper and follow-up phase
(Figure 1). Due to the sponsor’s decision to terminate the
study prematurely, efficacy data were analyzed from a pre-
defined (i.e., before unblinding) modified intent-to-treat
(MITT) population. This MITT population consisted of
randomized individuals who had at least one dose of study
medication, at least one postbaseline efficacy assessment,
and the opportunity to complete at least 44 weeks of med-
ication (8-week titration phase and 36-week maintenance
phase) before the sponsor’s announcement of study termi-
nation. The rationale for the MITT analysis is provided in
the statistical analysis section. The safety population con-
sisted of all individuals who had at least one dose of study
medication and who provided any postbaseline safety infor-
mation while on medication.
Enrollment and Run-in Phase
At the initial enrollment visit (week ?8), subjects were
evaluated with a medical examination including history,
(weight, height, waist and hip circumferences), an oral
glucose tolerance test, and a 12-lead electrocardiogram.
Blood and urine samples were collected for laboratory anal-
TPM for Maintenance of Weight Loss, Astrup et al.
OBESITY RESEARCH Vol. 12 No. 10 October 20041659
ysis. Enrolled subjects who met inclusion/exclusion criteria
then entered the 8-week nonpharmacological weight loss
phase described below. Subjects who lost at least 8% of
their enrollment body weight during the 8-week run-in
phase were eligible for randomization to pharmacological
treatment (TPM or placebo), provided that they continued to
meet inclusion and exclusion criteria. BMI inclusion criteria
at baseline were ?30 to ?50 kg/m2or BMI ?27 to ?50
kg/m2in the presence of controlled hypertension or dyslip-
During the initial 8-week run-in phase, subjects received
a low-calorie, nutritionally balanced diet containing 800 to
1000 kcal/d. The diet make-up was at the discretion of
individual centers; many centers used a proprietary liquid
formula diet or mixed formula/solid diet.
At randomization to pharmacological therapy (after the
8-week low-calorie diet run-in), all subjects participated in
a standardized commercially available behavioral modifica-
tion program known as Pathways to Change (Johnson &
Johnson Healthcare Systems Inc., Piscataway, NJ), which
continued throughout the remainder of the study. Pathways
to Change focuses on lifestyle and self-management as
related to weight loss and obesity. The low-calorie diet was
replaced by an individualized diet with an energy content
?600 kcal/d less than the subject’s calculated total energy
requirement (24). At week 32, the total energy requirement
was recalculated and the caloric content of the diet modified
accordingly. A standardized lesson plan administered by
clinic staff on a one-to-one basis concentrated on topics
associated with diet, nutrition, physical activity, psychoso-
cial structuring, and support. Each lesson topic for each
subject visit was unique but consistent for that visit at all
centers. In addition, each lesson was accompanied by inter-
active, user-friendly subject support materials.
Randomization, Titration, and Maintenance
Randomization codes were generated by the sponsor us-
ing the RandGen program. Subjects were allocated in ran-
domly permuted blocks stratified by study center. Subjects,
study investigators and staff, and sponsor were blinded
throughout the study to the treatment allocation.
An 8-week titration schedule was used as shown in Fig-
ure 1. All subjects randomized to TPM were started on a
dose of 16 mg in the evening for the 1st week and then
titrated upward over an 8-week period. Doses of 16 mg were
administered both morning and evening for the 2nd week,
and TPM was administered daily in two divided doses
thereafter. Each week thereafter, the dose was increased by
32 mg/d (in two divided doses) until the assigned dose was
reached. This occurred at the beginning of week 4 for those
assigned to 96 mg/d TPM and the beginning of week 7 for
those assigned to 192 mg/d TPM. The subjects were then
maintained at their assigned dose for the next 52 weeks. At
completion or termination of the study, medication was
tapered over a 2-week period.
Dosage Adjustment Related to Adverse Events
If intolerable side effects occurred at any dose, the clin-
ical site could reduce the dose by one level and continue
Figure 1: Study design.
TPM for Maintenance of Weight Loss, Astrup et al.
1660 OBESITY RESEARCH Vol. 12 No. 10 October 2004
treatment at this reduced dose for the remainder of the
study. Only one dose reduction was allowed. Subjects were
withdrawn from the study if intolerable adverse events
continued despite down-titration.
The primary efficacy end point was the mean percentage
change in body weight from enrollment (including the
8-week low-calorie diet run-in) to week 44, using the last
observation carried forward (LOCF) approach. Secondary
end points included the percentage change in body weight
from baseline (randomization to study drug, after comple-
tion of the 8-week low-calorie diet run-in), absolute change
in body weight from enrollment and baseline, number and
proportion of subjects losing at least 5%, 10%, and 15% of
their body weight (5%, 10%, and 15% responders) from
enrollment and from baseline, and the number and propor-
tion of subjects who, during titration and maintenance pe-
riods, maintained 50%, 75%, and 100% of the weight loss
achieved during the 8-week low-calorie diet run-in period.
Changes in lipid profile and systolic and diastolic blood
pressure were also predefined end points.
Safety evaluations were based on adverse events either
spontaneously reported by the subject or elicited by general,
nondirect questioning and were coded according to a mod-
ified WHO Adverse Reaction Terminology dictionary. Se-
verity was assessed as mild, moderate, or marked in accor-
dance with the impact on the subject’s daily life, i.e.,
minimal, noticeable, or substantial, respectively. Safety was
also assessed by physical examination, pulse and blood
pressure measurements, 12-lead electrocardiogram, and
standard urinalysis, hematology, and blood chemistry pan-
The sample size was determined based on the aim of
achieving at least 90% power to detect a 5% difference in
the mean percentage weight loss between the placebo and
the TPM-treated arms. These calculations used an estimated
SD of 11.0%, which was obtained from previously pub-
lished work on pharmacological weight loss. Commercial
(nQuery) software was used to perform the calculations.
The primary efficacy analysis population was predefined
(before unblinding) as an MITT population. This modifica-
tion was necessary because after the announcement of the
premature termination of the study, it was not possible to
guarantee that data collected subsequently would not be
subject to many uncontrollable factors or biases. The orig-
inal intent-to-treat (ITT) population consisted of all random-
ized subjects who received at least one dose of study drug
and provided at least one postbaseline efficacy evaluation.
The MITT population decision allowed only data collected
before the closedown announcement and only from subjects
who had the opportunity to complete a predetermined num-
ber of weeks (44 weeks for this study) of treatment before
the study closedown announcement. Therefore, subjects
were included in this population if their enrollment date was
on or before a particular date before the announcement of
the program termination. All subjects meeting this criterion
were included if they received at least one postrandomiza-
tion efficacy assessment. The primary population for effi-
cacy analysis is designated an MITT population to reflect
the fact that it was the ITT population that was appropriately
modified to omit all subjects randomized after the specified
date and to omit all data collected after the announcement of
the closedown of the study.
To evaluate the effect of TPM at different dose levels
compared with placebo and to adjust for multiple compar-
isons for the two active doses, the step-down multiple
testing framework of Dunnett and Tamhane (25) for com-
paring treatments with a control was used. This method
provides a family-wise false-positive protection rate of 0.05.
For the primary efficacy measure, the two-sided signifi-
cance level was 0.05. The primary efficacy end point was
analyzed using analysis of covariance, with treatment, cen-
ter, and treatment-by-center as factors and with enrollment
values and gender as covariates. Response rates were ana-
lyzed using the Cochran-Mantel-Haenszel test stratified by
gender and center. Except as otherwise noted, missing val-
ues were imputed on the basis of LOCF.
A total of 701 subjects were enrolled, of whom 561
(80%) were randomized to treatment after losing at least 8%
of their initial body weight during the 8-week nonpharma-
cological run-in period. Of these, 557 were included in the
safety population, whereas 293 were randomized early
enough to potentially allow completion of week 44 and
were, therefore, included in the MITT population. The spon-
sor’s decision to prematurely terminate the study accounted
for the majority (308/561 or 55%) of withdrawals of ran-
domized subjects (Figure 2). The first patient visit occurred
on August 25, 2000 and the last on June 14, 2002.
Demographic and baseline characteristics (at randomiza-
tion after 8-week low-calorie diet run-in) for the safety
population are listed in Table 1. The average age of subjects
was 43.8 years. Seventy-six percent were women, and al-
most all were white. Treatment groups were well-balanced,
with no notable differences on any variable. During the
titration and maintenance phases, 75 subjects (13%) with-
drew due to an adverse event (discussed in detail below), 49
(9%) due to subject choice, 7 (1%) for lack of efficacy, and
TPM for Maintenance of Weight Loss, Astrup et al.
OBESITY RESEARCH Vol. 12 No. 10 October 20041661
20 (4%) for other reasons. Nine subjects (2%) were lost to
follow-up (Figure 2). One subject in the 192 mg/d group
withdrew due to an adverse event during follow-up phase.
During the 8-week low-calorie diet run-in phase, all
groups in the MITT population experienced a similar mean
percentage weight loss: 10.6%, 10.9%, and 10.8% for those
subsequently randomized to placebo and 96 and 192 mg/d,
respectively, corresponding to mean decreases in weight of
11.6, 11.9, and 12.3 kg.
From enrollment (including the 8-week low-calorie
diet run-in) to the end of week 44, TPM significantly
decreased weight by 15.4% and 16.5% in the 96 and
192 mg/d groups, respectively, compared with 8.9%
in the placebogroup (MITT
(p ? 0.001) (Figure 3). This corresponds to mean de-
creases in weight of 9.9, 17.0, and 18.7 kg in the placebo
and 96 and 192 mg/d TPM groups, respectively, with
corresponding mean enrollment values of 108.3, 109.5,
and 111.9 kg. A supportive analysis of the 554 ITT
subjects (i.e., all randomized with at least one on-treat-
ment efficacy evaluation) yielded similar highly signifi-
When analyzed from randomization (i.e., after comple-
tion of the low-calorie diet run-in), TPM treatment fur-
ther decreased weight by 5.2% and 6.4% in the 96 and
192 mg/d groups, respectively, to week 44 compared with
a gain of 1.8% in the placebo group (MITT population,
LOCF) (p ? 0.001). This corresponds to mean decreases
in weight of 5.0 and 6.4 kg in the TPM 96 and 192 mg/d
groups compared with a gain of 1.7 kg in the placebo
Figure 2: Disposition of subjects.
TPM for Maintenance of Weight Loss, Astrup et al.
1662 OBESITY RESEARCH Vol. 12 No. 10 October 2004