Background
Psychiatric patients who receive long-term mood-stabilizing agents can gain excess body weight [1-3]. Weight gain is induced by the majority of antipsychotics [1,4], lithium [2,5,6], antidepressants [7] and valproate [5,8,9]. This could be a major problem with compliance and is one of the main reasons for discontinuation of treatment [1,10]. Increased body weight is also a major health hazard. Mortality rates for both men and women show a strong association with the body mass index (BMI) [11]. The risk starts to increase when the BMI is >25 kg/m2 and doubles for BMI values of >40 kg/m2, where the correlation curve becomes progressively steeper [11]. It has been estimated that obesity causes 300,000 deaths per year in the USA alone [12].
Topiramate is a relatively new antiepileptic drug. Its effects include sodium-channel-blocking activity, enhancement of cerebral GABA concentrations and antagonism of AMPA/kainate receptors, which leads to a decreased glutamate-mediated excitation [13]. It has been shown to cause weight loss in a variety of treatment populations. In an add-on uncontrolled study on 34 epileptic patients there was a 5.9 kg weight loss at 1 year, while among the obese patients in that trial (BMI> = 30 kg/m2) the weight loss reached 10.9 kg [14]. The weight reducing potential of topiramate has been used to treat obesity in subjects with no concomitant psychiatric disorder or epilepsy. Bray et al, [15] treated 385 subjects in a randomised, double-blind, placebo-controlled study and reported weight losses of 4.8%–6.3% of body weight for increasing doses of topiramate treatment. Topiramate was also effective in the treatment of binge-eating disorder. In an open-label study of 13 patients topiramate resulted in 11.8 kg mean weight loss over a period of 3–30 months [16]. McElroy et al, [17] treated 61 outpatients affected with binge-eating associated with obesity for 14 weeks in a placebo-controlled trial. Mean weight loss was 5.9 kg and the binge-eating frequency was also reduced. The beneficial effects on weight and binge-eating were sustained during a 42-weeks open continuation of that trial [18], but a number of patients dropped out from the study. Another double-blind study on 69 patients with bulimia nervosa reported improvements in self-esteem and anxiety, in addition to the behavioural dimensions of bulimia [19].
Several studies have investigated the use of topiramate in bipolar disorders. Marcotte [20] evaluated 58 psychiatric patients, mostly with bipolar affective disorder. They were treated with add-on topiramate for a mean of 16 weeks. 62% were rated improved (weight changes were not reported). McElroy et al, [21] gave topiramate to 56 outpatients, either manic, depressed, or euthymic. The manic patients were reported improved, while the depressed (N = 11) did not change. Significant decreases in weight and BMI were recorded at every follow-up point, reaching 6.0 kg after 1 year. Gupta et al [22] performed a retrospective chart review of 5 patients with bipolar or schizoaffective disorder and reported a mean weight loss of 10 kg. Vieta et al, [23] treated 25 patients with treatment-resistant bipolar spectrum disorders in an open study and reported significant improvements in rating scales for depression and mania. Over 50% of all patients were considered to be responding to topiramate. Ten patients in that study experienced weight loss. Guille and Sachs, [24] treated 14 patients with refractory bipolar illness for an average of 22.4 weeks. Nine of them (64%) improved. Four patients with BMI>28 kg/m2 experienced mean weight loss of 13.5 kg. In an open study topiramate was given together with Risperidone to 58 bipolar patients during a manic episode, of whom 41 completed the 12 months follow-up [25]. A significant improvement in rating scales for mania was recorded and relapse rates were lower compared to the preceding year. The mean weight of patients decreased by 1.1 kg by the end of the study. The same team examined the effect of the co-administration of topiramate and olanzapine over a 12 months interval [26]. Thirteen patients completed 12 months of treatment and their mania and depression ratings improved significantly. At the endpoint they showed 0.5 kg of weight loss, an impressive result given the known problems of weight gain associated with olanzapine [4]. A study using retrospective chart review on patients given topiramate showed a mean loss of 1.2 kg [27]. Topiramate was also used to treat acute manic episodes. In one study 18 patients with manic, mixed, or rapid-cycling episodes, resistant to current mood-stabilizers were given topiramate [28]. There were 12 patients who were judged responders after 5 weeks, and they had a mean weight loss of 4.3 kg. In a study conducted by Grunze et al, [29] 11 patients were given add-on topiramate, of whom 7 showed good improvement. In a study reported by Bozikas et al, [30], 14 manic patients received topiramate either as monotherapy, or in combination with antipsychotics for 4 weeks. Response rate was 61.5% and all patients tolerated topiramate well. Both weight loss and weight gain were reported in that study. Calabrese et al, [31] treated 10 patients hospitalised with mania with topiramate monotherapy for 4 weeks and 8 patients were described as responding. In a retrospective chart review of 76 patients, 13% showed a moderate to marked improvement on the Clinical Global Impression scale (CGI) [32]. There was a weight loss in 50% of patients in that sample, with a mean weight loss of 6.5 kg. A better response and higher weight loss correlated with higher doses of medication. Topiramate has been used in the treatment of the depressive phase of bipolar disorder. McIntyre et al, [33] compared bupropion and topiramate in 36 depressed BP outpatients in an open trial lasting 8 weeks. There was a similarly good improvement in both groups, with a 56% response rate on the Hamilton Rating Scale for Depression in the topiramate group (N = 18). The weight loss in this study was 5.8 kg in the topiramate group and 6 patients withdrew due to side effects. Lykouras and Hatzimanolis, [34] gave topiramate to 56 patients who had suffered relapses of bipolar disorder in the previous 12 months as an add-on treatment. There was a significant reduction of new manic and depressive episodes in the 50 patients who completed 12 months treatment.
All these open studies should be considered in the context of four large unpublished placebo-controlled trials of topiramate for the treatment of acute mania, which produced negative results (cited in Yatham [35]). They indicate clearly that monotherapy with topiramate has no acute anti-manic property. In contrast, almost all of the above open trials used topiramate as an add-on therapy.
Many patients in our Affective Disorders Clinic were concerned with excessive weight gain which they blamed at least partly on their treatments. As weight loss was a listed side effect of topiramate and several small studies had already reported on its use in bipolar affective disorders, we decided to use it as an add-on mood stabilizer in patients who had weight problems. We reasoned that topiramate could also have mood-stabilizing properties, like several other anticonvulsants, a finding already reported in open trials. We also felt that this off-licence use of topiramate was justified, as this intervention enabled patients to comply better with the rest of their medication, thus reducing potential relapses of illness, and that the weight loss could improve their risk for cardio-vascular disorders. We were not aware of the results from the controlled trials until all the patients described here were already taking topiramate.

Patients
Patients in our mood disorders clinic are seen in regular intervals for long-term maintenance of their illness (some have attended for over 7 years now). As part of the service that we provide, patients have regular blood tests, reviews of medication and checks of weight, usually at intervals of 3 or 4 months. Topiramate was given to some obese patients as part of our normal clinical practice, rather than as an experimental intervention (therefore we sought no Ethics committee approval). We decided to publish our results only after the patients had taken topiramate for a considerable time, because we felt that we should share our experience with other health professionals, as the results looked so clear. The only difference to treatment as usual was that we checked the weight of these patients on a more regular basis than we would have normally done. All 12 patients included in this report had a pre-topiramate BMI of >30 kg/m2, i.e. they were all obese according to the National Heart, Lung and Blood Institute guidelines [36]. According to the British National Formulary guidelines [37], an anti-obesity drug should be considered only for those with a BMI of 30 or greater (page 207), and we felt that this recommendation should be followed in the case of Topiramate, even if this drug is not licenced for this use. The mean BMI of patients at the start was 37.95 kg/m2, SD = 5.66 and the mean weight was 104.38 kg, SD = 21.46 kg. The patients had the following diagnoses: Bipolar Affective Disorder type I (BPI) = 5, BPII = 4 and Unipolar Depression (UD) = 3. Four of them were rapid cyclers. Two of the three unipolar patients had co-morbid epilepsy and the third one reached a BMI of nearly 45 (extreme obesity). We felt that it was justified to give her this medication on general health grounds, although benefits from topiramate on unipolar depression had not been demonstrated before. Our clinic treats chronic and treatment resistant patients, and 9 of the 12 patients could be described as treatment resistant or incomplete responders, who had either frequent relapses of mood disorder (more than 2 per year), or remained chronically depressed (i.e. had not reached criteria for remission for a period of at least several weeks over the previous two years) despite adequate treatment. One more overweight patient was prescribed topiramate. He did not like its effect and stopped treatment after one month. He is therefore not included in any analysis.

Concomitant medication
The nine treatment resistant/incomplete responder patients took combinations of medications: all of them took antidepressants, seven took antipsychotics, five took lithium and five took anticonvulsants (Table 1). One other patient was in a full remission on lithium monotherapy and one had recently been discharged from hospital after a manic relapse, having been relatively stable for several years prior to that. In four cases an anticonvulsant was exchanged with topiramate, in the remaining cases topiramate was given as an add-on treatment. No one received monotherapy of topiramate.
Table 1  Characteristics of the patients and changes of their weight during the study. Both individual and mean weight changes are presented. p-values are based on paired samples tests for the patients who have completed the corresponding follow-up interval. The dose of topiramate is the highest dose reached. Under "Concomitant medication" the drugs that are undelined were exchanged with topiramate.

Topiramate dosage
Topiramate dose was increased in steps of 25–50 mg every one to two weeks. After some of the first patients complained of paraesthesia and headaches, we routinely started making the increases only at 2-weekly intervals. The final dose of topiramate ranged between 200 and 600 mg, with a mean of 296 mg. The dose was increased until we could see a clear effect on the weight, or the patient reported some side effects. The patient whose topiramate was increased to 600 mg/day suffers with poorly controlled epilepsy and we increased the dose in order to achieve a better seizures control, rather than weight loss.

Statistical analysis
The level of statistical significance for weight loss at each time-point during the follow-up, compared to baseline, was determined with a paired-samples t-test. This test assumes that comparisons at the four time points are independent, whreas the observations within individuals are, of course correlated. Therefore the test for statistical significance is slightly anti-conservative.