Neurocognitive Functioning in Overweight and Obese Patients With Bipolar Disorder: Data From the Systematic Treatment Optimization Program for Early Mania (STOP-EM)

Abstract

Objective

Obesity is frequent in people with bipolar I disorder (BD I) and has a major impact on the course of the illness. Although obesity negatively influences cognitive function in patients with BD, its impact in the early phase of the disorder is unknown. We investigated the impact of overweight and obesity on cognitive functioning in clinically stable patients with BD recently recovered from their first manic episode.

Method:

Sixty-five patients with BD (25 overweight or obese and 40 normal weight) recently remitted from a first episode of mania and 37 age- and sex-matched healthy control subjects (9 overweight or obese and 28 normal weight) were included in this analysis from the Systematic Treatment Optimization Program for Early Mania (commonly referred to as STOP-EM). All subjects had their cognitive function assessed using a standard neurocognitive battery. We compared cognitive function between normal weight patients, overweight–obese patients, and normal weight healthy control subjects.

Results:

There was a negative affect of BD diagnosis on the domains of attention, verbal memory, nonverbal memory, working memory, and executive function, but we were unable to find an additional effect of weight on cognitive functioning in patients. There was a trend for a negative correlation between body mass index and nonverbal memory in the patient group.

Conclusions:

These data suggest that overweight–obesity does not negatively influence cognitive function early in the course of BD. Given that there is evidence for a negative impact of obesity later in the course of illness, there may be an opportunity to address obesity early in the course of BD.

Patients with BD exhibit impairments in multiple domains of neurocognitive functioning. Patients with euthymic BD show deficits in processing speed, attention, verbal and visual memory, and executive function at different stages of illness.^1–4 These deficits are robustly associated with poor psychosocial outcomes,⁵ and identifying predictors of cognitive impairment, particularly modifiable ones, is therefore a priority. To date, such predictors include medications, namely, mood stabilizers and atypical antipsychotics,^1,6–9 substance abuse or dependence,¹⁰ duration of illness and number of episodes,^1,6,11 childhood trauma,¹² and possibly obesity.¹³ The deficits in the broad domains of attention, learning–memory, and executive function are also present even in patients with first-episode mania, where the variables associated with the progression of BD are fewer.^14–16

Overweight and (or) obesity have been reported to have a significant impact on the clinical course of BD, and potentially on cognitive functioning. BD is associated with an increased burden of obesity-related conditions, including higher risk for metabolic syndrome, diabetes, hypertension, cardiovascular disease, and dyslipidemia.^17–19 Obesity is highly prevalent in this population^20–23 and has been shown to have a negative impact on the course of the illness.^20,22,24 Moreover, patients with BD who experience significant weight gain during the course of 1 year show impairments in their functioning.²⁵ As well, overweight patients with BD have been shown to have reduced brain volumes in important brain regions involved in mood regulation.²⁶ Thus obesity appears to negatively impact brain functioning in people with BD.

Clinical Implications

• Overweight–obesity does not negatively influence cognitive function early in the course of BD.

• Adequate management of overweight–obesity early in the course of BD may help reducing the burden associated with the disorder.

Limitations

• The sample size, particularly in the group of overweight–obese patients, was relatively modest.

• The availability of information about the course of overweight–obesity in the subjects was limited.

An extensive literature has identified an important relation between obesity and cognition. Otherwise healthy obese people show poorer performance on cognitive tests when compared with normal weight–matched control subjects, in particular, on memory and executive function–related tasks.^27–29 This effect of obesity may be expected to be even more pronounced in patients with BD, given their intrinsic vulnerability to cognitive impairment. However, to date, only one study has shown that overweight patients with BD performed poorly on tasks assessing processing speed and attention when compared with normal weight patients with BD.¹³ Participants in our study had an average duration of illness of 15 years and a mean of 7 episodes of both depression and mania, which indicates that obesity may negatively impact cognition in patients with established illness. However, the impact of overweight and obesity on cognitive functioning in patients early in the course of BD is unknown.

The study of patients with first-episode mania has the potential to provide important information about the effect of elevated BMI on cognitive functioning in the early stages of the illness, without the bias introduced by the numerous confounding factors that accrue during the course of BD (for example, duration of illness, number of previous mood episodes, and use of medication). It also has the potential to provide information about the vulnerabilities imposed by the overweight or obesity in the BD population. Establishing the nature of the potentially harmful association between obesity and BD could provide important insights into development of management strategies focusing on weight loss to help prevent an additional negative influence on the burden of BD.

We have previously reported that patients with recently remitted first-episode mania have significant impairments in various domains of neurocognitive functioning (see Torres et al¹⁴). In our present analysis, we evaluated the effect of overweight–obesity on cognitive functioning in patients with recently diagnosed BD. We hypothesized that patients would show impaired cognition, compared with a healthy comparison group, and that the presence of elevated BMI would have an additional negative effect on cognitive functioning.

Methods

Subjects

Sixty-five patients meeting criteria for BD I according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision,³⁰ were recruited from the STOP-EM at Vancouver Hospital Health Sciences Centre and affiliated sites. Patients were enrolled in STOP-EM via the inpatient Mood Disorders Clinical Research Unit at the University of British Columbia Hospital, as well as through community and hospital referrals from physicians and psychiatrists. To be included in the study, patients were required to be aged between 16 to 35 years old, and to have experienced their first manic and (or) mixed episode, with or without psychosis, and with or without comorbid conditions, within 3 months of enrolment into the study and to be clinically stable to undergo cognitive testing. Patients were excluded if they could not provide informed consent or if they had a previous undetected manic episode. A description of the full longitudinal study protocol can be found elsewhere.^14,31 Thirty-seven healthy control subjects, matched for age and sex, were recruited for comparison purposes. The study protocol was approved by the University of British Columbia ethics committee. Written informed consent was obtained from all patients and control subjects before any study procedures were conducted.

Diagnosis of BD was based on a clinical interview by a trained psychiatrist and a standardized psychiatric examination using the Mini-International Neuropsychiatric Interview³² at baseline. We collected sociodemographic data and information about the course of the illness using a standardized protocol. The clinical status of the patients was assessed by clinical interview and clinical rating scales: GAF,³³ PANSS,³⁴ BPRS,³⁵ YMRS,³⁶ and HRSD-29.³⁷ Subjects presenting with a history of major medical illness underlying their manic symptoms were excluded. Healthy control subjects were assessed with a standardized interview, and they were enrolled if they had no personal history of psychiatric illness or major medical illness and no family history of psychiatric illness in their first-degree relatives.

Body Mass Index

Patients and healthy control subjects underwent a physical examination at enrolment and follow-up. Participants were weighted in a nonfasting state in light clothing, with footwear removed. BMI was calculated using the following formula: BMI = weight (kg)/height (m²). Normal weight was defined as BMI of 18.50 to 24.99 kg/m², overweight–obesity as BMI of 25.00 kg/m² or more, and underweight as BMI of less than 18.50 kg/m².³⁸ Within the patient group, 25 were overweight or obese and 40 had normal weight; within the healthy control subject group, 9 were considered overweight or obese and 28 had normal weight.

Cognitive Assessment

Subjects were tested at baseline. The neurocognitive testing was administered by research assistants or PhD-level graduate students trained and supervised by a registered senior clinical neuropsychologist. To assure consistency across testers, the training included in-vivo observation of testing by the senior neuropsychologist. Testers administered the battery when the subjects were clinically stable (see mean mood ratings in Table 1). On average, the testing took 2 to 3 hours to administer. Subjects were allowed smoking breaks during testing sessions.

Table 1.

Demographic variables in patients with BD and healthy controls at baseline

Demographic	Patients		Control subjects
	Overweight–obese n = 25 Mean (SD)	Normal weight n = 40 Mean (SD)	Overweight–obese n = 9 Mean (SD)	Normal weight n = 28 Mean (SD)	Test F, df
Age, years	23.80 (4.54)	22.12 (4.10)	25.44 (6.80)	23.14 (4.54)	1.55, 3/98
BMI, kg/m2	27.36 (2.22)	22.07 (1.48)	28.34 (2.69)	22.20 (1.58)	68.30, 3/98a
Years of education	14.00 (1.87)	13.45 (2.42)	15.00 (2.95)	14.36 (1.87)	1.679, 3/98
Premorbid IQ	105.60 (7.72)	107.50 (7.27)	106.67 (8.80)	108.11 (6.07)	0.592, 3/98
IQ (KBIT)	101.96 (9.81)	106.30 (10.07)	105.56 (8.24)	110.21 (8.83)	3.248, 3/97a
CTQ	37.25 (10.06)	36.91 (10.61)	35.66 (12.24)	37.19 (15.15)	0.031, 3/83
	% (n)	% (n)	% (n)	% (n)	X2, df
Sex (female)	56.00 (14)	52.50 (21)	55.55 (5)	57.14 (16)	0.164, 3
Ethnicity					14.578, 9
Caucasian	68.00 (17)	77.50 (31)	66.67 (6)	60.71 (17)
Asian	08.00 (2)	17.50 (7)	22.23 (2)	32.14 (9)
Clinical variables	Mean (SD)	Mean (SD)			t, df
Age of onset of mania (years)	23.64 (4.48)	22.03 (4.08)			1.494, 63
Overall age of onset of BD (years)	20.40 (5.36)	19.62 (5.02)			0.589, 63
Number of previous depressive episodes	1.16 (1.52)	0.97 (1.56)			0.477, 63
YMRSb	1.40 (2.32)	1.35 (3.33)			0.066, 63
HRSD-29b	6.72 (7.38)	6.27 (7.61)			0.232, 63
GAF	67.04 (11.64)	67.74 (14.26)			0.206,62
BPRS	21.16 (5.99)	22.60 (6.08)			0.912, 60
PANSS	7.48 (1.15)	7.69 (1.39)			0.632, 62
	% (n)	% (n)			X2, df
Medical comorbidity	32.00 (8)	18.42 (7)			3.312, 1
Current or lifetime substance abuse or dependence	58.33 (14)	36.84 (14)			2.743, 1
Lithium	44.00 (11)	45.00 (18)			0.006, 1
Valproate	56.00 (14)	40.00 (16)			1.585, 1
Atypical antipsychotics	76.00 (19)	80.00 (32)			0.146, 1
Combination of mood stabilizers and atypical antypsichotics	76.00 (19)	70.00 (28)			0.277, 1
Antidepressants	8.00 (2)	7.50 (3)			0.005, 1

^aP < 0.05

^bScore closest to the time of cognitive testing

BD = bipoloar disorder; BMI= body mass index; BPRS = Brief Psychiatric Rating Scale; CTQ = Childhood Trauma Questionnaire; GAF = Global Assessment of Functioning; HRSD-29 = Hamilton Rating Scale for Depression, 29 items; KBIT = Kaufman Brief Intelligence Test; PANSS = Positive and Negative Symptom Scale; YMRS = Young Mania Rating Scale

Premorbid IQ and current IQ were assessed using the North American Adult Reading Test full scale IQ, and Kaufman Brief Intelligence Test, respectively. The cognitive battery was selected to assess 6 domains, which have been demonstrated to be relevant to BD.³⁹ The categorization of tasks was modelled closely after the Measurement and Treatment to Improve Cognition in Schizophrenia (commonly referred to as MATRICS) Consensus Cognitive Battery, which has been validated in BD.⁴⁰ The 6 domains assessed, and the respective measures within each domain, have been used in our previous work,^41,42 and were as follows:

1. Processing Speed: Trail-Making Test time to complete part A; Stroop Test Word and Colour Naming trials number correct; Letter Fluency number correct;

2. Attention: CANTAB RVIP discriminability score, RVIP latency score;

3. Verbal Memory: CVLT-II recall trials 1–5; CVLT-II delayed free recall trial;

4. Nonverbal Memory: CANTAB Spatial Recognition Memory per cent correct; CANTAB Pattern Recognition Memory per cent correct; CANTAB Paired Associate Learning total errors adjusted score;

5. Working Memory: Wechsler Memory Scale, Third Edition, Letter and Number Sequencing; CANTAB Spatial Working Memory between errors; and

6. Executive Function: Trail-Making Test B time; Stroop Colour and Word trial number correct; CANTAB Intra Extra Dimensional set shifting task number of extra-dimensional shifting errors; CANTAB Stockings problems solved in the minimum number of moves.

The raw scores obtained for each primary cognitive measure were converted into z scores (ranging from −4 to 4) based on demographics-adjusted normative data from test manuals. Domain scores for each subject were calculated as the average of z scores of the primary measures within each cognitive domain.

Data Analysis and Statistics

Patients and healthy control subjects were divided according to their BMI into overweight–obese or normal weight groups. To assess demographic and clinical group differences, we used Student t test, ANOVA, or chi-square statistics as appropriate. In our primary analyses, to compare groups on cognitive domain scores, we used factorial MANOVA with BMI (overweight–obese, compared with normal weight) and diagnosis (patients, compared with control subjects) as between-subject factors. Because one of the groups (overweight–obese control subjects) had a small sample size of only 9 subjects, we performed further direct comparisons exclusively between the 2 groups of patients. For these analyses, we used MANOVA, with and without controlling for current IQ. As secondary analyses, we performed Pearson correlation analysis between BMI and individual cognitive domain scores in control subjects and partial correlations controlling for mood scores (HRSD-29 and YMRS) in patients. All statistical tests were 2-tailed and had a significance threshold of α = 0.05. Data are presented as means with standard deviations. Analyses were conducted using the IBM SPSS Statistics software for windows, Version 20.0 ((IBM SPSS Inc, Armonk, NY).

Results

Demographics

Overweight–obese and normal weight patients and healthy control subjects were well matched in age (P = 0.20), sex (P = 0.98), years of education (P = 0.18), premorbid IQ (P = 0.66), and ethnicity (P = 0.10) (Table 1). Overweight–obese patients had lower current IQ, compared with normal weight control subjects (P = 0.03). Patients and control subjects had the same proportion of people who were overweight–obese and normal weight (χ² = 2.121, df = 1, P = 0.14). Additionally, no differences were found among the 2 groups of patients regarding their GAF (P = 0.84), BPRS (P = 0.36), PANSS (P = 0.53), YMRS (P = 0.95), and HRSD-29 (P = 0.82) scores, medication use, and other relevant clinical measures, such as age of onset of mania (P = 0.14), number of previous depressive episodes (P = 0.92), overall age of onset of BD (P = 0.56), medical comorbidity (P = 0.19), and alcohol and (or) substance abuse or dependence history (P < 0.10) (Table 1).

Between-Group Analysis of Cognitive Functioning

Table 2 presents the mean z scores for each domain and the results of the factorial MANOVA. The multivariate tests revealed a significant diagnosis effect (Wilks λ = 0.832, F = 3.128, df = 6/93, P = 0.008). Univariate tests revealed significant diagnosis main effects for the domains of attention (P = 0.02), verbal memory (P < 0.001), nonverbal memory (P = 0.005), working memory (P = 0.001), and executive function (P = 0.007). Post hoc analyses showed that, in all these domains, patients performed worse than healthy control subjects. The multivariate tests revealed a nonsignificant BMI group effect (Wilks λ = 0.950, F = 0.812, df = 6/93, P = 0.56), and all univariate tests were also nonsignificant (P > 0.05). The multivariate tests for the interaction between the presence of the BD diagnosis and the presence of elevated BMI revealed no significant results (Wilks λ = 0.970, F = 0.473, df = 6/93, P = 0.83). The tests of between-subject effects for this interaction for all 6 domains did not show significant results: processing speed (P = 0.20), attention (P = 0.69), verbal memory (P = 0.90), nonverbal memory (P = 0.77), working memory (P = 0.68), and executive function (P = 0.50). When the groups of patients were compared with each other (excluding healthy control subjects), the multivariate test indicated a nonsignificant BMI group effect: Hotelling T = 0.039; F = 0.372 df = 6/58, P = 0.89 (not controlling for current IQ), and Hotelling T = 0.067; F = 0.621, df = 6/56, P = 0.71 (controlling for current IQ).

Table 2.

Mean z score on each cognitive domain in patients with bipolar disorder and healthy subjects

Cognitive domain	Patients		Control subjects
	Overweight–obese n = 25 Mean (SD)	Normal weight n = 40 Mean (SD)	Overweight–obese n = 9 Mean (SD)	Normal weight n = 28 Mean (SD)	Main effect of diagnosis F, df	Main effect of elevated BMI F, df	Interaction between diagnosis and elevated BMI F, df
Processing speed	−0.49 (0.61)	−0.42 (0.68)	−0.48 (0.76)	0.00 (0.78)	1.855, 1/98	2.959, 1/98	1.653, 1/98
Attention	−0.18 (0.99)	−0.25 (0.78)	0.16 (0.71)	0.24 (0.68	5.198, 1/98a	0.000, 1/98	0.161, 1/98
Verbal memory	−0.09 (1.09)	−0.26 (1.05)	0.79 (1.04)	0.56 (0.88)	13.530, 1/98a	0.704, 1/98	0.016, 1/98
Nonverbal memory	−0.13 (0.87)	−0.03 (0.70)	0.36 (0.43)	0.37 (0.49)	8.365, 1/98a	0.119, 1/98	0.085, 1/98
Working memory	−0.37 (0.87)	−0.22 (0.99)	0.41 (0.54)	0.39 (0.75)	12.472, 1/98a	0.102, 1/98	0.174, 1/98
Executive function	−0.22 (0.71)	−0.12 (0.79)	0.12 (0.67)	0.45 (0.65)	7.655, 1/98a	1.623, 1/98	0.457, 1/98

^aP < 0.05

BMI = body mass index

Correlation Between Body Mass Index and Cognitive Functioning

Table 3 presents the correlations between BMI and the cognitive domains for patients and healthy control subjects. We found a trend for a negative correlation between the BMI and the score of the cognitive domain of nonverbal memory in the patient group (r = −0.246, r² = 0.06, P = 0.05), controlling for mood scores (YMRS and HRSD-29).

Table 3.

Correlations between body mass index and the score on cognitive domains in patients with bipolar disorder and healthy controls at baseline

Cognitive domain	Patientsa n = 61		Control subjects n = 37
	r	P	r	P
Processing speed	−0.035	0.79	−0.193	0.25
Attention	0.068	0.60	−0.006	0.97
Verbal memory	0.081	0.53	0.233	0.16
Nonverbal memory	−0.246	0.05	0.098	0.56
Working memory	−0.078	0.54	0.201	0.23
Executive function	0.007	0.98	−0.050	0.77

^aControlling for mood scores (Young Mania Rating Scale and Hamilton Rating Scale for Depression Rating Scale, 29 items)

Discussion

To our knowledge, this is the first study to explore the relation between overweight–obesity and neurocognitive functioning in a sample of patients with BD who recently recovered from their first episode of mania. The results obtained from the direct comparisons between groups indicate that the cognitive problems early in the course of BD were most likely due to the illness itself and not related to obesity or overweight. The finding of a diagnosis effect is consistent with results from our prior study (see Torres et al¹⁴). Moreover, except for a small, statistically nonsignificant, negative correlation between BMI and performance on the domain of nonverbal memory in the patient group, there were no correlations between cognitive domains and BMI within the patient group. Similarly, there were no significant correlations between BMI and neurocognitive domain scores in the healthy control group. Although some of the participants in the current report were also included in Torres et al,¹⁴ our present study was different from the prior report in the following 2 ways: the sample size has increased by 20 patients in the ongoing longitudinal study; and, the current focus is on the effects of weight status and cognitive function, which was not assessed in the prior report.

Otherwise healthy overweight–obese adults reportedly exhibit impairments mostly in the cognitive domains of memory and executive function.²⁷ We were unable to show any cognitive impairment in our overweight–obese healthy control group in comparison to their normal weight counterparts. However, our statistical methods likely lacked power to detect those differences owing to the small number of overweight–obese healthy control subjects, and thus our study was unable to sufficiently address whether obesity associates with cognitive impairment in healthy control subjects. Nonetheless, we replicated some findings from previous reports^1,2,40 regarding the impact of BD itself on cognition in the comparison between patients and healthy subjects. Although we replicated the findings from previous literature, we did not detect any differences in the neurocognitive functioning between the overweight–obese and normal weight group of patients. These results were similar when the group of healthy subjects was included or excluded in the analyses. Consequently, given the modest sample size, a type II error cannot be excluded.

As this study was conducted in Vancouver, which has a large Asian population, our study subjects included some subjects of Asian ethnic background. There is a debate about the appropriate BMI cut-offs for categorizing this population into normal weight and overweight groups.⁴³ Therefore, one could argue that different BMI cut-offs might have yielded significant effects of BMI on cognition. However, the number of Asian subjects in each group in our study sample was very small, and comparable between the groups, making such a possibility unlikely. Further, correlation analysis is not impacted by different BMI cutoffs, and the assertion that categorization of patients based on BMI into low and high groups did not affect the results also supports the finding of no significant correlation between BMI and cognition in this study.

Although there were no statistically significant differences in comorbidity rates between the various groups in our sample, the rates were numerically higher in the overweight–obese group. There is some evidence that cognitive impairments are more pronounced in the population with BD and alcohol dependence comorbidity,⁴⁴ and it is possible that the impact of comorbidity might have biased our findings. However, if there were a negative effect of substance abuse comorbidity in our sample, this would have biased our overweight patient group to have even more cognitive deficits, given that the comorbidity rates were nonsignificantly higher in our overweight group. Even with this potential bias, that we failed to observe a cognitive difference between overweight–obese and normal weight patients suggests that comorbidity did not confound our results.

Study subjects were allowed to take smoking breaks during the neurocognitive testing sessions. There is some evidence that smoking may impact cognitive functioning.⁴⁵ We did not systematically collect information about smoking during the cognitive testing sessions in our study, and therefore, we cannot exclude the potential confounding of smoking on our study results.

Another potential explanation for the lack of effect of obesity on cognitive functioning could relate to the limited information available about the course of obesity in the subjects included in our sample. Our baseline assessment recorded the information about the presence of elevated BMI at the time of recruitment, but the duration of the current obesity was not registered. Therefore, our sample is likely to include subjects who had been obese for only a few weeks to few months at the most. It can be implied from the literature^46–51 that obesity, rather than being a short-term risk factor for cognitive dysfunction, may impose its deleterious effects on cognition over time. Interestingly, several reports point to impairments in cognitive functioning even among young obese subjects.²⁷ Obese children or adolescents are outperformed by their normal weight comparison group on tests of general intellectual ability,⁵² attention,^53,54 and executive function.^53–55 The cross-sectional design of most of these studies limits the conclusions about a definite causative factor. Nonetheless, it is possible that longer duration of obesity leads to a more significant impact on cognitive functioning. Indeed, longitudinal observations concluded that long-term obesity was associated with poor performance on batteries of tests assessing cognition.^49,51,56,57 Additionally, people who had gained more weight during the follow-up period experienced an additional worsening on test scores.⁵¹ Therefore, it is conceivable that the cognitive deficits associated with obesity take place over time and may not be detectable early in the course of BD, particularly if patients have not been obese for a long period of time.

Another possible explanation for our results is the potentially positive influence of the pharmacological treatment received by the subjects enrolled in our study. Obesity has been associated with alterations in circulating proteins associated with neuronal survival. More specifically, obese people have been shown to have lower levels of BDNF, which promotes neuronal cell survival.^58–60 Additionally, plasma levels of BDNF were negatively correlated to BMI.⁶⁰ Peripheral levels of BDNF have also been associated with cognitive functioning. Patients from STOP-EM received treatment as per Canadian Network for Mood and Anxiety Treatments (commonly referred to as CANMAT) guidelines for managing BD. These guidelines recommend as first-line treatment mood stabilizers or antipsychotics in monotherapy or in combination.^61,62 Mood stabilizers and antipsychotics are known to induce increases in the neurotrophic factors, and in particular BDNF, in BD^63–65 and schizophrenia.⁶⁶ In line with this, it is possible that the additional negative impact of the elevated BMI may have been overcome by optimal pharmacological treatment provided to the patients, and it could be the reason why we were unable to detect a potentially negative effect of obesity on cognitive functioning in patients in our analyses. However, as we have not collected serum samples for BDNF levels in all our subjects, we are unable to verify this hypothesis.

Atypical antipsychotics and mood stabilizers that are widely used to treat BD are associated with weight gain.^67–69 In this regard, it is possible that some patients who presented elevated BMI at baseline may have experienced a recent weight gain. Patients were recruited within 3 months of their recovery from their first manic episode and most of them were treated with different combinations of atypical antipsychotics and mood stabilizers. Hence the future longitudinal analyses of our full sample may provide important information about the long-term effects of overweight–obesity on cognitive functioning in patients with BD. From our present study and the results obtained in established patients,¹³ the implication is that even though elevated BMI does not impose a negative impact on cognitive functioning in BD early in the course of the illness, if unaddressed, it may lead to future deficits.

Our main limitations were a modest sample size, particularly in the overweight–obese group of healthy subjects, and the lack of information about the course of obesity prior to intake into the STOP-EM study. Another weakness is our limited ability to address the influence of medications regarding obesity and even cognitive functioning. Nonetheless, our study has several strengths. It is the first attempt to evaluate the impact of obesity on cognitive function in a first-episode mania sample with BD. While cognitive function in long-standing BD is affected by the confounds of age, duration of illness and treatment, multiple mood episodes, and comorbid conditions, these factors are minimized in a sample of early BD, allowing us to assess the impact of obesity on cognition more directly. Longitudinal assessments of obesity coupled with evaluation of serum markers and cognitive function in BD may be helpful in understanding this further in future studies. The observation that obesity does not seem to worsen cognitive functioning in early BD is potentially clinically relevant, and indicates that adequate management to control obesity along with optimal treatment of BD early in its course has tremendous potential in reducing the disease burden related to BD.

Abbreviations

BD

bipolar disorder

BDNF

brain-derived neurotrophic factor

BMI

body mass index

BRPS

Brief Psychiatric Rating Scale

CANTAB

Cambridge Neuropsychological Test Automated Battery

CVLT-II

California Verbal Learning Test—Second Edition

GAF

Global Assessment of Functioning Scale

HRSD-29

Hamilton Rating Scale for Depression, 29 items

PANSS

Positive and Negative Syndrome Scale

RVIP

Rapid Visual Information Processing

STOP-EM

Systematic Treatment Optimization Program for Early Mania

YMRS

Young Mania Rating Scale