Effect of divalproex on brain morphometry, chemistry, and function in youth at high-risk for bipolar disorder: a pilot study

JOURNAL OF CHILD AND ADOLESCENT PSYCHOPHARMACOLOGYVolume 19, Number 1, 2009 ª Mary Ann Liebert, Inc.
Pp. 51–59DOI: 10.1089=cap.2008.060 Effect of Divalproex on Brain Morphometry, Chemistry, and Function in Youth at High-Risk for Bipolar Disorder: A Pilot Study Kiki Chang, M.D., Asya Karchemskiy, M.S., Ryan Kelley, B.A., Meghan Howe, M.S.W., Amy Garrett, Ph.D., Nancy Adleman, B.S., and Allan Reiss, M.D.
Objective: Divalproex has been found efficacious in treating adolescents with and at high risk for bipolar disorder(BD), but little is known about the effects of mood stabilizers on the brain itself. We sought to examine the effects ofdivalproex on the structure, chemistry, and function of specific brain regions in children at high-risk for BD.
Methods: A total of 24 children with mood dysregulation but not full BD, all offspring of a parent with BD, weretreated with divalproex monotherapy for 12 weeks. A subset of 11 subjects and 6 healthy controls were scannedwith magnetic resonance imaging (MRI, magnetic resonance spectroscopy [MRS], and functional MRI [fMRI]) atbaseline and after 12 weeks.
Results: There were no significant changes in amygdalar or cortical volume found over 12 weeks. Furthermore, nochanges in neurometabolite ratios were found. However, we found the degree of decrease in prefrontal brainactivation to correlate with degree of decrease in depressive symptom severity.
Conclusions: Bipolar offspring at high risk for BD did not show gross morphometric, neurometabolite, or functionalchanges after 12 weeks of treatment with divalproex. Potential reasons include small sample size, short exposure tomedications, or lack of significant neurobiological impact of divalproex in this particular population.
primary areas of involvement in BD (Chang et al. 2004; Stra-kowski et al. 2005), and abnormalities in these areas may be Divalproex has been shown to be effective for the detected in children before the onset of fully developed BD treatment of mania in adults with bipolar disorder (BD) (Chang et al. 2006). Therefore, we sought to examine the ef- (Bowden et al. 1994), and in open studies of children with BD fects of divalproex on the structure, chemistry, and function of (Kowatch et al. 2000; Wagner et al. 2002). Furthermore, di- these brain regions in children at high risk for BD.
valproex may (Chang et al. 2003b) or may not (Findling et al.
Offspring of parents with BD are at increased risk for the 2007) be effective in treating children with subsyndromal development of BD (Lapalme et al. 1997; Chang et al. 2003a).
symptoms of BD who are at high risk for development of full Such high-risk offspring with and without psychiatric symp- BD. However, little is known about the effects of divalproex toms have been found to have increased hippocampal vol- on the brain itself. Advances in neuroimaging technology, ume (Ladouceur et al. 2008) and decreased cerebellar vermis including modalities such as magnetic resonance imaging N-acetylaspartate (NAA)(Cecil et al. 2003), although some (MRI), functional MRI (fMRI), magnetic resonance spec- neuroimaging studies in bipolar offspring have been rela- troscopy (MRS), and diffusion tensor imaging (DTI), have tively negative (Gallelli et al. 2005; Ladouceur et al. 2008; allowed for in vivo study of such effects of psychotropic Singh et al. 2008). These areas are involved in mood regulation medications. Furthermore, it may be surmised that these and prefrontal limbic circuitry that has been proposed as ab- medications act upon brain structures and circuits thought to normal in BD. Thus, it might be surmised that with amelio- be involved in the pathophysiology of BD. Prefrontal amyg- ration of mood symptoms changes in the neurobiological dalar circuits that regulate mood have been proposed to be characteristics of these areas might be detected.
Pediatric Bipolar Disorders Program, Stanford University School of Medicine, Stanford, California.
This work was supported by a grant from Abbott Laboratories, National Institutes of Health grant MH64460-01 to Dr. Chang, and a gift from the Hahn Family.
Portions of this manuscript were presented at the 53rd Annual Meeting of the American Academy of Child and Adolescent Psychiatry, San Diego, CA, October 25–30, 2006.
CHANG ET AL.
Previously, we found that children with familial BD and a disorder (ADHD), major depression, dysthymia, or cyclo- history of lithium or valproate exposure tended to have larger thymia. Additionally, subjects had to have at least moderate amygdalar volume than those without such exposure (Chang current mood symptoms, as indicated by a score of >10 on the et al. 2005). Regarding neurochemistry, however, two MRS Young Mania Rating Scale (YMRS) or a score of >30 on the studies of adults with BD failed to find significant effects Children's Depressive Rating Scale–Revised (CDRS-R). All of valproate on brain NAA, myo-inositol (mI), or glutamate- subjects (patients and healthy volunteers) were evaluated by glutamine g-butyric acid (Glx) (Silverstone et al. 2003; Fried- the affective disorders module of the Washington University man et al. 2004). Regarding brain function, overall brain in St. Louis Kiddie Schedule for Affective Disorders and activation in healthy volunteer adults was increased after 14 Schizophrenia (WASH-U-KSADS) (Geller et al. 1996; Geller days of valproate administration (Bell et al. 2005). At the et al. 2001), and the Schedule for Affective Disorders and cellular level, divalproex may have direct neurotrophic effects, Schizophrenia for School-Age Children, Present and Lifetime including increasing prefrontal bcl-2, inhibiting glycogen (K-SADS-PL) (Kaufman et al. 1997). Diagnostic decisions were synthase kinase 3B (GSK-3B), and activating the extracellular ultimately made by a board-certified child psychiatrist based signal-regulated kinase (ERK) mitogen-activated protein on personal interview, discussion with the research assistant, (MAP) kinase pathway, all putative neuroprotective effects and written notes of parental and subject responses to in- (Manji et al. 2000). Thus, we sought to study the neurobio- dividual WASH-U-KSADS questions. Current and lifetime logical effects of divalproex in a high-risk offspring population.
diagnoses were established according to DSM-IV criteria.
We hypothesized that offspring with mood and=or behavioral Response to treatment was defined by a week-8 score of 1 symptoms, but not full BD, would demonstrate increases in (very much improved) or 2 (much improved) on the Clinical amygdalar volume, increases in prefrontal NAA=Creatine- Global Impressions Scale–Change subscale (CGI-C). A subset phospho-creatine (Cr) ratios, and changes in prefrontal- of 11 consecutive subjects (the last 11 enrolled in the clinical trial after funding was obtained to include MRI in the proto- monotherapy. Because this was a pilot study with small col) were scanned with MRI, both at baseline (pretreatment, sample sizes, we sought to generate data that would lead no medications) and after 12 weeks. Furthermore, 6 healthy to hypotheses for future large-scale studies.
control subjects, matched for age, gender, and IQ, were alsoscanned at baseline and at 12 weeks to serve as a comparatorgroup for fMRI.
Eleven subsyndromal subjects were scanned using mor- This protocol was approved by the Stanford University phometric MRI, 1H-MRS, and fMRI on a 3-Tesla GE Signa Panel of Medical Research in Human Subjects. Twenty four scanner (Milwaukee, WI). Patients with BD had psychosti- children with a parent with BD, who themselves had early mulants discontinued for at least 24 hours before the scan, pri- symptoms of mood dysregulation but not full BD, were marily due to a concurrent functional MRI study of attention.
enrolled in a 12-week open label trial of divalproex mono- They were allowed to continue any other current medications, therapy (Chang et al. 2003b). Inclusion criteria for subsyn- such as mood stabilizers or antidepressants, due to the risk of dromal subjects were age 9–18 years, a biological parent with mood destabilization. Medication history was obtained from BD I or II, and a diagnosis of ‘‘subsyndromal'' BD, as defined direct interview with subjects and parent and review of below. Exclusion criteria were presence of a pervasive de- medical records when available (Table 1).
velopment disorder (such as autism or Asperger disorder), aneurological condition (such as a seizure disorder), a sub- stance use disorder, intelligence quotient (IQ) less than 80, orpresence of metallic implants or orthodontic braces, which Coronal 3D volumetric spoiled gradient echo (SPGR) series would make the MRI scan not feasible.
were obtained with the following parameters: time of repeti- Six healthy controls (group matched for age, IQ, and tion (TR) ¼ 35, time to echo (TE) ¼ 6, flip angle ¼ 45, slice handedness with subjects from the fMRI subset) were also thickness ¼ 1.5 or 1.6 mm, and matrix ¼ 256"192 for 124 sli- included in the present study. For inclusion in the control ces. The volumetric analysis was performed using BrainImage group, healthy volunteers did not have a current or lifetime software v. 5.3.7 (Stanford Psychiatry Neuroimaging La- Diagnostic and Statistical Manual of Mental Disorders, 4th edition boratory; http:==cibsr.stanford.edu) for semiautomated image (DSM-IV) (American Psychiatric Association 1994) psychia- processing and quantification.
tric diagnosis, had both parents without any psychiatric di- The processing of the scans involved removal of the non- agnosis by Structured Clinical Interview for DSM-IV Axis I brain tissue, correction of nonuniformity, and positional nor- disorders (SCID), and did not have a first- or second-degree malization to anterior and posterior commissures in a relative with BD as determined by the Family History Re- stereotactic space (Talairach and Tournoux 1988). Each brain search Diagnostic Criteria (Andreasen et al. 1977).38 was divided into lobes with a semiautomated stereotactic- An oral and written consent from the parents as well as an based parcellation method (Kates et al. 1999), based on the oral and written assent from the adolescents were obtained, raters' identification of the anterior commissure, the posterior and both the parents and the offspring were interviewed. For commissure, and a midsagittal point above the axis created by the subsyndromal group, at least one parent had BD I or II the first two points. Raters who conducted morphometric diagnosed by the SCID (First et al. 1995), administered by a analyses were blind to the diagnosis of each subject. Voxels trained master's degree-level clinician and=or board-certified comprising brain tissue were then segmented into gray mat- child and adolescent psychiatrist. For inclusion in the sub- ter, white matter, and cerebrospinal fluid (CSF) using a syndromal group, in addition to parental diagnosis of BD, all semiautomated fuzzy tissue segmentation algorithm (Reiss children either met criteria for attention-deficit=hyperactivity et al. 1998). The total brain volume (TBV) was calculated by NEUROBIOLOGICAL EFFECTS OF DIVALPROEX IN BIPOLAR OFFSPRING calculating the sum of all brain regions. Total cerebral volumewas calculated by adding cerebral total tissue with corticaland ventricular CSF. Total brain tissue was calculated byadding cerebral total tissue, cerebellar tissue, and brainstemtissue.
Amygdalae were outlined manually by reliable raters (interrater reliability > 0.9 with intraclass correlation coeffi-cient) on positionally normalized brain image stacks in thecoronal orientation. Amygdalae were traced starting on theslice demonstrating the thickest extent of the anterior com-missure and following the structure toward the posterior endof the brain. The most superior white matter tract extendingfrom the temporal lobe marked the inferior border, CSFmarked the medial border, endorhinal sulcus marked the Placement of magnetic resonance spectroscopy (MRS) voxels in bilateral dorsolateral prefrontal cortex superior border, and a thick, central white matter tract of the temporal lobe was used as the lateral border of amygdala(Fig. 1).
Brain structure volume data were first examined for rebrum and visually maintaining approximately equal parts normality to conform to the assumptions of the parametric gray and white matter (Fig. 2). An investigator blind to di- statistics employed. One-way analyses of covariance (AN- agnosis inspected each voxel placement visually to ensure COVAs) were used for comparisons of brain structure proper placement fully within the brain and that spectra volumes, with age and TBV as covariates. A p value of 0.05 contain no sizable lipid peaks or rolling baselines. MRS data (two-tailed) was chosen as the significance threshold.
were acquired using a preselected region of interest for point-resolved spectroscopy (PRESS) with a TR=TE of 2000= 35 msec. MRS scans used 32 averages, 1-kHz spectral band-width, 1 k data points, and unsuppressed water collected for For 1H-MRS, a 2"2"2-cm voxel was prescribed in the right all spectra. The MRS scan was 1 minute and 44 seconds in and then left dorsolateral prefrontal cortex (DLPFC), from the length. We were able to obtain an adequate signal-to-noise first axial slice above the lateral ventricles. Because slices were ratio with this relatively short acquisition time due to the rel- 5 mm thick, the voxel was placed anywhere from 0 mm to atively large field strength of 3T. The fully automated PRO- 5 mm above the lateral ventricles, immediately anterior to a BE=SV quantification tool (General Electric Medical System, line drawn between the anterior aspects of the lateral ventri- Milwaukee, WI) was used to process MRS data. Each of the cles, and as far lateral as possible while remaining in the ce- five spectral peaks associated with NAA, creatine-phospho-creatine (Cr), choline (Cho), mI, and H2O was quantified byLevenberg–Marquardt curve fitting over that line region us-ing the standard data processing package by GE mentionedabove.
Differences in NAA=Cr ratios from baseline to end of treat- ment were considered primary outcome measures. Second-ary, exploratory analyses of additional metabolite ratios (mI,Cho) were also conducted. Paired t-tests were used to com-pare pre- and post-valproate ratios. We used Bonferroni cor-rection to account for left and right hemispheric data, and awas set at 0.025 for our main outcome variable, NAA=Cr. Wedid not correct for exploratory comparisons of mI=Cr andCho=Cr.
Negative (e.g,. a mutilated dog), positive (e.g., puppies), and neutral (e.g., a plate) pictures that were deemed accept-able to a pediatric population were selected from the Inter-national Affective Picture System (IAPS) (Lang et al. 1997).
The three types of stimuli were organized in blocks, each withsix stimuli, with each stimulus presented for 4500 msec with a500-msec interstimulus interval. Subjects were asked to indi- Outline of the left and right amygdalae on the po- cate how each picture made them feel by pressing one of three sitionally normalized brain stack in coronal orientation. The buttons corresponding to ‘‘negatively,'' ‘‘neutrally,'' and most superior white matter tract extending from the tem- ‘‘positively.'' Stimuli were projected onto a screen using a poral lobe marked the inferior border, cerebrospinal fluid custom-built magnet compatible projection system (Sanyo, (CSF) marked the medial border, endorhinal sulcus markedthe superior border, and a thick, central white matter tract of San Diego). A custom-built button box was used to record the temporal lobe was used as the lateral border of amygdala.
CHANG ET AL.
fMRI data acquisition models that computed contrast images of negative minusneutral conditions. These models also included additional Images were acquired on a 3T GE Signa scanner using a contrast images computing repeated measures activation dif- standard GE whole head coil. The following spiral pulse se- ferences between the subject's baseline scan and week 12 scan quence parameters were used: TR ¼ 2000 msec, TE ¼ 30 msec, for the negative-neutral contrasts described above. Resultant flip angle ¼ 808, field of view (FOV) ¼ 200, 28 slices, 64"64 contrast images were analyzed using a general linear model to matrix, and 1 interleave. To reduce field inhomogeneities, an determine voxel-wise t-statistics.
automated high-order shimming method based on spiral ac-quisitions was used before acquiring functional MRI scans fMRI regions of interest analysis (Kim et al. 2000). To aid in localization of the functional data,high-resolution T1 weighted spoiled gradient recalled (SPGR) Our hypotheses of the role of the DLPFC and amygdala in 3D MRI sequences with the following parameters used: BD were tested by measuring activation in these regions using TR ¼ 35 msec, TE ¼ 6 msec, flip angle ¼ 458, FOV ¼ 24 cm, 124 spherical regions of interest (ROIs) (5 mm radius). Both right slices in coronal plane, 256"192 matrix.
(22, #2, #20) and left (#22, #2, #20) amygdala ROIs werevisually placed by a trained research assistant on a group- Image preprocessing averaged SPGR scan and examined by 2 trained neuroscien-tists to verify accuracy of placement. Placement of right (48, fMRI data were preprocessed using SPM2 (http:==www 16, 22) and left (#48, 12, 28) DLPFC ROIs were based on prior .fil.ion.ucl.ac.uk=spm). Images were reconstructed, corrected loci of activation, Brodmann areas 9=45, from a previous for movement, and normalized to Montreal Neurological In- study in which pediatric subjects with BD demonstrated stitute (MNI) coordinates. Images were then resampled every greater activation compared to healthy controls when per- 2 mm and smoothed with a 4-mm Gaussian kernel. MNI co- forming the IAPS task, and negative minus neutral pictures ordinates were transformed to stereotaxic Talairach coordi- contrast (Chang et al. 2004) (Fig. 3).
nates using a nonlinear transformation (Brett et al. 2002; Activation in the ROIs was quantified as the percentage of voxels within the ROI that surpassed a specified statisticalthreshold (Z > 1.67; p < 0.05). Activation differences in each fMRI statistical analysis ROI were extracted to a spreadsheet for statistical comparisonwith clinical scores.
Statistical analysis was performed on individual and group data using the general linear model and the theory of General statistical analyses Gaussian random fields as implemented in SPM2 (WellcomeDepartment of Cognitive Neurology, London, UK). Each Statistical analyses were completed using SPSS 12.0 (http: subject's data were globally scaled, high passed filtered at ==www.spss.com=). Independent t-tests were used in com- 120 seconds, and analyzed using a balanced design with parisons between subsyndromal subjects and healthy con- Change in dorsolateral prefrontal cortex (DLPFC) activation versus change in Hamilton Rating Score for Depression (HAM-D) score in subsyndromal bipolar disease (BD) subjects.
NEUROBIOLOGICAL EFFECTS OF DIVALPROEX IN BIPOLAR OFFSPRING trols for demographic variables, total brain volume (TBV), weeks (1501.28 $ 232.18 cm3 at baseline versus 1507.85 $ and ROI activation differences. Repeated measures analysis 236.75 cm3 after 12 weeks, p ¼ 0.96).
was used to investigate time point associations within be- Total amygdala volume in subsyndromal BD subjects did havioral ratings, ROI activation differences, and clinical not change significantly over the 12 weeks of divalproex treatment (3.70 $ 0.45 cm3 at baseline versus 3.74 $ 0.48 cm3after 12 weeks, p ¼ 0.86; Cohen d ¼ 0.08). Furthermore, no dif- ference was found in amygdalar grey matter volume (3.11 $0.21 cm3 at baseline versus 3.29 $ 0.37 cm3 after 12 weeks, Morphometric data were obtained and usable for all 11 subjects. One subject did not have follow-up MRS data and The amygdala volume in the control group also remained was excluded from the MRS analysis. For the fMRI analysis, 4 similar over the course of 12 weeks (3.79 $ 0.84 cm3 at baseline subsyndromal subjects did not have both baseline and 12- versus 4.03 $ 0.57 cm3 after 12 weeks, p ¼ 0.48; Cohen week follow-up scans and were excluded. Additionally, 1 sub- d ¼ 0.33), as did the amygdala grey matter volume (3.47 $ syndromal subject and 1 healthy control were excluded due to 0.74 cm3 at baseline versus 3.60 $ 0.56 cm3 after 12 weeks, excessive (greater than 10% of the task) combined translational and rotational movement more than 3 mm compared to thefirst scan of the series. Demographic data are given in Table 1.
There were no significant differences in percent gray and Morphometric results white matter in MRS voxels from baseline compared with There were no significant changes in TBV in subjects trea- week 12. There were no significant differences in pre- or post- ted with divalproex over 12 weeks (1549.50 $ 181.61 cm3 at divalproex NAA=Cr ratios (see Table 2). The Cohen d was 0.12 baseline versus 1545.52 $ 186.92 cm3 after 12 weeks, p ¼ 0.97).
for the left and 0.94 for the right, indicating a large effect size Healthy controls also did not have changes in TBV over 12 for a decrease in right DLPFC NAA=Cr.
Table 1. Demographics of Subjects Number of subjects serum level (mg=mL) Mean decrease in YMRS score over 12 weeks of study Mean decrease in HAM-D score Past medication exposure (%) Abbreviations: MRI, Magnetic resonance imaging; MRS, magnetic resonance spectroscopy; fMRI, functional MRI; SD, standard deviation; ADHD, attention-deficit=hyperactivity disorder; ODD, opposition defiant disorder; YMRS, Young Mania Rating Scale; HAM-D, HamiltonRating Score for Depression.
CHANG ET AL.
Table 2. 1H-MRS Results, Pre- and Posttreatment with controls. At week 12, prodromal subjects had significantly less extreme valence ratings for both negative (t ¼ #2.55,p ¼ 0.031) and positive (t ¼ 2.31, p ¼ 0.046) valences relative to healthy controls. No significant valence rating differenceswere found between groups for negative and positive va- lences at baseline or with neutral valences at either time point.
fMRI brain activation results There were no significant differences between subsyndromal and control subjects when comparing activation in the DLPFC or amygdala at baseline (respectively, t ¼ #0.54, p ¼ 0.78; t ¼ 0.49, p ¼ 0.15) or at week 12 (respectively, t ¼ #0.28, p ¼ 0.60;t ¼ 1.56, p ¼ 0.14). Similarly, there were no significant changes DLPFC ¼ dorsolateral prefontal cortex; NAA ¼ N-acetylaspartate; in DLPFC or amygdala activation between baseline and week Cr ¼ creatine-phospho-creatine; Cho ¼ choline; mI ¼ myo-inositol.
12 within the subsyndromal group (respectively, F ¼ 0.064,p ¼ 0.81; F ¼ 0.066, p ¼ 0.81) or within the control group (re-spectively, F ¼ 0.032, p ¼ 0.87; F ¼ 0.67, p ¼ 0.46).
In addition, we performed exploratory analyses on mI=Cr Repeated measures analysis resulted in a significant inter- and Cho=Cr ratios and no significant change in these ratios action between change in Hamilton Rating Score for Depres- were found. A representative spectrum from one subject is sion (HAM-D) scores and DLPFC activations during baseline shown in Fig. 4.
scans compared to week 12 scans (F ¼ 8.218, r2 ¼ 0.673,p ¼ 0.046; Fig. 3). This indicates that greater differential in fMRI behavioral results DLPFC activation from baseline to week 12 was associatedwith greater improvement in HAM-D score at week 12.
Each individual's ratings were averaged across pictures of the same valence, (negative, neutral, or positive), as classified by the IAPS, to give a subject's mean rating for each valence ofthe pictures. As expected, there was a significant effect of We found no significant changes in total brain gray matter valence, indicating that all subjects rated the positive, nega- volume, amygdalar volume, prefrontal NAA=Cr ratios, or tive, and neutral pictures significantly differently (baseline, prefrontal amygdalar activation after 12 weeks of divalproex F ¼ 44.73, p < 0.001; week 12, F ¼ 99.70, p < 0.001). However, monotherapy in bipolar offspring with subsyndromal mood repeated measures analysis indicated a significant interaction and behavioral disorders. Despite increasing power by re- (Huynh–Feldt, F ¼ 7.08, p ¼ 0.011) for week-12 behavioral peated measures analyses, this study was hampered by small valence scores between subsyndromal subjects and healthy sample size and should thus be considered as preliminary Representative magnetic resonance spectroscopy (MRS) spectra from 1 subject.
NEUROBIOLOGICAL EFFECTS OF DIVALPROEX IN BIPOLAR OFFSPRING and pilot data. However, effects sizes for morphometric and However, we did, have two interesting findings from the neurochemical change were generally small, decreasing the fMRI study. First, subsyndromal BD subjects differed from possibility of Type II error. The only large effect size found controls in their ratings of emotionally valenced pictures only was for a decrease in right DLPFC NAA=Cr in subjects treated after treatment with divalproex. It appeared that their week- with divalproex.
12 ratings of negatively valenced pictures were rated less These results are slightly surprising given the preclinical negatively and positive pictures less positively compared evidence for the neuroprotective qualities of valproate. In with ratings from healthy controls. Subjects may have been animal studies, valproate has been shown to increase levels of desensitized to the pictures because they were shown the the neuroprotective protein bcl-2 in the frontal cortex (Chen same set 12 weeks prior; however, one would expect any such et al. 1999; Manji et al. 2000) and activate protein kinases that desensitization to be similarly present in healthy controls.
mediate the effects of neurotrophic factors to stimulate neural Thus, it is possible that treatment with divalproex may have dendritic growth (Manji and Lenox 1999). Both lithium and narrowed the subjects' subjective experience of both nega- valproate have been found to have neurogenic effects in rat tivity and positivity. Given the small sample size, this is a brains and neural stem cells (Hashimoto et al. 2003; Laeng highly preliminary finding.
et al. 2004). However, there is little human data in this regard.
Second, the degree of prefrontal activation decrease was To our knowledge, there have been no prospective studies of correlated with improvement in depressive symptoms. This human brain morphometric change following valproate ad- finding might indicate why we did not find differences at ministration. Because of our finding that children with BD baseline and at week 12 between subsyndromal subjects and and a history of lithium and=or valproate exposure had controls in amygdalar or DLPFC activation. There appeared amygdalar volumes more similar to healthy controls than to be a range of both activation and behavioral response, those children with BD without such exposure, who had de- leading to heterogeneity in the sample that may have ‘‘washed creased volumes (Chang et al. 2005), we had hypothesized out'' any findings. Correlations with such variables as mood that divalproex treatment would result in increased amygdala state and response, as done here, may be one solution to ad- volume in our subjects. Thus, it is possible that lithium may dressing this heterogeneity. Furthermore, this finding might have more of this effect than valproate. Regarding neuro- indicate that prefrontal structures may be less needed to reg- chemistry, two MRS studies of adults with BD failed to find ulate emotional response after successful treatment with di- significant effects of valproate on brain NAA, mI, or Glx valproex. It is possible that subjects with greater improvement (Silverstone et al. 2003; Friedman et al. 2004), although neither in depression no longer needed to recruit prefrontal areas to of these studies was prospective. Similarly, we failed to find aid in modulating signals from hyperactive subcortical limbic significant changes in NAA, mI, and Cho to Cr ratios.
areas. Thus, in this model, DLPFC activation would reflect de- There are even fewer data regarding the effects of valproate gree of subcortical limbic activity. Therefore, divalproex may on human brain function. In a study of healthy volunteer work directly not on prefrontal areas, but potentially in sub- adults, overall brain activation was increased after 14 days cortical limbic areas, such as the amygdala. Indeed, one im- of valproate administration (Bell et al. 2005). Previously, we portant action of divalproex is potentiation of g-aminobutyric found that lamotrigine, also an anticonvulsant, led to de- acid (GABA) neurotransmission (Loscher 2002), and the ba- creases in amygdalar activation in adolescents with bipolar solateral nucleus of the amygdala (BLA) is significantly in- depression (Chang et al. 2008). Thus, we expected to find hibited by GABA-ergic interneurons (Rainnie et al. 1991).
similar results in children at-risk for BD treated with another Furthermore, electrical kindling of rat amygdala results in anticonvulsant, divalproex. Again, we did not prove this pri- decreases of such inhibitory GABA-ergic neurons in the BLA mary hypothesis.
(Callahan et al. 1991; Lehmann et al. 1998). Similar models Our results suggest that behavioral improvement in our have been proposed to occur in BD, so that the amygdala may subjects may not have been due directly to measurable have an increased flow of excitatory activity due to defi- changes in gray matter, throughout the brain, and in the ciencies in GABA-ergic inhibition (Benes and Berretta 2001).
amygdala specifically. Furthermore, it is also possible that However, it is still possible that divalproex directly affects valproate itself simply does not affect these variables. It is also prefrontal regions as well, leading to decreased activation, but possible that our subjects did not achieve a high enough brain one would then expect less regulatory control over limbic level of valproate to induce measurable change. The achieved activation, leading to worsening of mood, not improvement.
mean serum level of 82 mg=mL is in the suggested therapeutic As mentioned, this study is limited by sample size and thus range for treating adults with BD (Bowden et al. 1994), but these results should be taken as preliminary. Our subjects, toward the lower end of the range suggested for children (80– although all bipolar offspring, also presented with a variety of 120 mg=mL) (Kowatch et al. 2005). It is not known if, similar to psychiatric disorders, including ADHD, depression, anxiety, lithium (Moore 2002), children have lower brain-to-serum and cyclothymia. This heterogeneity may have led to varying ratios of valproate levels than adults due to neurophysiolog- neurobiological responses to valproate and thus our negative ical differences. Children may also require longer treatment MRI findings. A few of our subjects were also previously than 12 weeks to demonstrate change that was detectable by exposed to psychotropic medications, such as stimulants and our methods. A large 4-week trial of extended-release dival- antidepressants, which have effects on brain structure and proex did fail to demonstrate efficacy over placebo in treating function. For example, increased exposure to antidepressants children with acute mania (Wagner et al. in press). However, may lead to decreased amygdalar volume in adolescents with our subjects showed positive responses in mood symptom BD (DelBello et al. 2004). We used ratios of NAA to Cr-PCr severity reduction over 12 weeks, and thus one could rea- and did not obtain absolute concentrations of NAA due to sonably expect corresponding neurobiological change by at methodological issues. Specifically, p files for spectra were least 12 weeks in our subjects.
not saved correctly, so that later analysis with programs to CHANG ET AL.
calculate absolute concentrations, such as the LC Model, was Chang K, Steiner H, Dienes K, Adleman N, Ketter T: Bipolar not possible. Thus, changes in Cr over time may have ob- offspring: A window into bipolar disorder evolution. Biol scured actual changes in NAA concentrations. Finally, Psychiatry 53:945–951, 2003a.
there may have been changes in other regions of the brain Chang K, Adleman NE, Dienes K, Simeonova DI, Menon V, that we did not study, such as hippocampus, ventrolat- Reiss A: Anomalous prefrontal-subcortical activation in fa- eral prefrontal cortex (PFC), or anterior cingulate, where milial pediatric bipolar disorder: A functional magnetic reso- others have found neurometabolite change in response to nance imaging investigation. Arch Gen Psychiatry 61:781–792, psychotropic medications (DelBello et al. 2006; Patel et al.
Chang K, Karchemskiy A, Barnea-Goraly N, Garrett A, Simeo- Nonetheless, this is the first study to investigate the neu- nova DI, Reiss A: Reduced amygdalar gray matter volume infamilial pediatric bipolar disorder. J Am Acad Child Adolesc robiological effects of valproate in a pediatric population, and Psychiatry 44:565–573, 2005.
a population at genetic risk for BD. Our results may indicate Chang K, Howe M, Gallelli K, Miklowitz D: Prevention of pe- that behavioral change may predate neurobiological change diatric bipolar disorder: Integration of neurobiological and that was detectable by our methods. Clearly, prospective psychosocial processes. Ann NY Acad Sci 1094:235–247, 2006.
neuroimaging studies with larger samples of children with Chang KD, Dienes K, Blasey C, Adleman N, Ketter T, Steiner H: mood disorders treated with psychotropic agents over longer Divalproex monotherapy in the treatment of bipolar offspring periods are needed to clarify the neurobiological effects of with mood and behavioral disorders and at least mild affec- these medications.
tive symptoms. J Clin Psychiatry 64:936–942, 2003b.
Chang KD, Wagner C, Garrett A, Howe M, Reiss A: A pre- liminary functional magnetic resonance imaging study ofprefrontal-amygdalar activation changes in adolescents with Dr. Chang receives research support from Abbott Labora- bipolar depression treated with lamotrigine. Bipolar Disord tories, AstraZeneca, GlaxoSmithKline, Lilly, Otsuka Labora- 10:426–431, 2008.
tories, and the NIMH. He is on the speakers' board and=or is a Chen G, Zeng WZ, Yuan PX, Huang LD, Jiang YM, Zhao ZH, consultant for Abbott Laboratories, AstraZeneca, Bristol Manji HK: The mood-stabilizing agents lithium and valproate Myers' Squibb, Eli Lilly & Co., GlaxoSmithKline, and Otsuka.
robustly increase the levels of the neuroprotective protein bcl- Drs. Garrett and Reiss and Ms. Karchemskiy, Mr. Kelley, Ms.
2 in the CNS. J Neurochem 72:879–882, 1999.
Howe, and Ms. Adleman have no conflicts of interest or fi- DelBello MP, Zimmerman ME, Mills NP, Getz GE, Strakowski nancial ties to disclose.
SM: Magnetic resonance imaging analysis of amygdala andother subcortical brain regions in adolescents with bipolar disorder. Bipolar Disord 6:43–52, 2004.
The authors gratefully acknowledge the assistance of DelBello MP, Cecil KM, Adler CM, Daniels JP, Strakowski SM: Melody Chang for scan acquisition, Jessica Yee for data Neurochemical effects of olanzapine in first-hospitalizationmanic adolescents: A proton magnetic resonance spectroscopy management, and Chris Wagner for data analysis.
study. Neuropsychopharmacology 31:1264–1273, 2006.
Findling RL, Frazier TW, Youngstrom EA, McNamara NK, Stansbrey RJ, Gracious BL, Reed MD, Demeter CA, Calabrese American Psychiatric Association: Diagnostic and Statistical JR: Double-blind, placebo-controlled trial of divalproex Manual of Mental Disorders, 4th edition (DSM-IV). Wa- monotherapy in the treatment of symptomatic youth at high shington (DC): American Psychiatric Association, 1994.
risk for developing bipolar disorder. J Clin Psychiatry 68:781– Bell EC, Willson MC, Wilman AH, Dave S, Silverstone PH: Differential effects of chronic lithium and valproate on brain First MB, Spitzer RL, Gibbon M, Williams JBW: Structured activation in healthy volunteers. Hum Psychopharmacol Clinical Interview for DSM-IV Axis I Disorders - Patient Edi- 20:415–24, 2005.
tion (SCID-I=P, version 2.0) New York, NY: Biometric Re- Benes FM, Berretta S: GABAergic interneurons: Implications for search, New York State Psychiatric Institute, 1995.
understanding schizophrenia and bipolar disorder. Neu- Friedman SD, Dager SR, Parow A, Hirashima F, Demopulos C, ropsychopharmacology 25:1–2, 2001.
Stoll AL, Lyoo IK, Dunner DL, Renshaw PF. Lithium and Bowden CL, Brugger AM, Swann AC, Calabrese JR, Janicak PG, valproic acid treatment effects on brain chemistry in bipolar Petty F, Dilsaver SC, Davis JM, Rush AJ, Small JG, et al.: Ef- disorder. Biol Psychiatry 56:340–348, 2004.
ficacy of divalproex vs lithium and placebo in the treatment of Gallelli KA, Wagner CM, Karchemskiy A, Howe M, Spielman D, mania. The Depakote Mania Study Group [published erratum Reiss A, Chang KD: N-acetylaspartate levels in bipolar off- appears in JAMA271:1830, 1994] [see comments]. JAMA spring with and at high-risk for bipolar disorder. Bipolar 271:918–2, 1994.
Disord 7:589–597, 2005.
Brett M, Johnsrude IS, Owen AM: The problem of functional lo- Geller B, Zimerman B, Williams M, Bolhofner K, Craney JL, calization in the human brain. Nat Rev Neurosci 3:243–24, 2002.
DelBello MP, Soutullo C: Reliability of the Washington Uni- Callahan PM, Paris JM, Cunningham KA, Shinnick-Gallagher P: versity in St. Louis Kiddie Schedule for Affective Disorders Decrease of GABA-immunoreactive neurons in the amygdala and Schizophrenia (WASH-U-KSADS) mania and rapid cy- after electrical kindling in the rat. Brain Res 555:335–339, 1991.
cling sections. J Am Acad Child Adolesc Psychiatry 40:450– Cecil KM, DelBello MP, Sellars MC, Strakowski SM: Proton magnetic resonance spectroscopy of the frontal lobe and cer- Geller BG, Williams M, Zimerman B, Frazier J: WASH-U-KSADS ebellar vermis in children with a mood disorder and a familial (Washington University in St. Louis Kiddie Schedule for Af- risk for bipolar disorders. J Child Adolesc Psychopharmacol fective Disorders and Schizophrenia) St. Louis (Missouri): 13:545–555, 2003.
Washington University, 1996.
NEUROBIOLOGICAL EFFECTS OF DIVALPROEX IN BIPOLAR OFFSPRING Hashimoto R, Senatorov V, Kanai H, Leeds P, Chuang DM: Li- Manji HK, Moore GJ, Chen G: Clinical and preclinical evidence thium stimulates progenitor proliferation in cultured brain for the neurotrophic effects of mood stabilizers: Implications neurons. Neuroscience 117:55–61, 2003.
for the pathophysiology and treatment of manic-depressive Kates WR, Warsofsky IS, Patwardhan A, Abrams MT, Liu AM, illness. Biol Psychiatry 48:740–754, 2000.
Naidu S, Kaufmann WE, Reiss AL: Automated Talairach Moore CM: Brain-to-serum lithium ratio and age: An in vivo atlas-based parcellation and measurement of cerebral lobes in lithium magnetic resonance spectroscopy study. In: 57th An- children. Psychiatry Res 91:11–30, 1999.
nual Convention of the Society for Biological Psychiatry Phi- Kaufman J, Birmaher B, Brent D, Rao U, Flynn C, Moreci P, ladelphia, PA, p 4, 2002.
Williamson D, Ryan N: Schedule for Affective Disorders Patel NC, Delbello MP, Cecil KM, Stanford KE, Adler CM, and Schizophrenia for School-Age Children-Present and Strakowski SM: Temporal change in N-acetyl-aspartate con- Lifetime Version (K-SADS-PL): Initial reliability and valid- centrations in adolescents with bipolar depression treated with ity data. J Am Acad Child Adolesc Psychiatry 36:980–988, lithium. J Child Adolesc Psychopharmacol 18:132–139, 2008.
Rainnie DG, Asprodini EK, Shinnick-Gallagher P: Inhibitory Kim D, Adalsteinsson E, Glover G, Spielman S: SVD regulari- transmission in the basolateral amygdala. J Neurophysiol zation algorithm for improved high-order shimming. In: Proceedings of the 8th Annual Meeting of ISMRM Denver, p Reiss AL, Hennessey JG, Rubin M, Beach L, Abrams MT, War- sofsky IS, Liu AM, Links JM: Reliability and validity of an Kowatch RA, Suppes T, Carmody TJ, Bucci JP, Hume JH, Kro- algorithm for fuzzy tissue segmentation of MRI. J Comput melis M, Emslie GJ, Weinberg WA, Rush AJ: Effect size of Assist Tomogr 22:471–479, 1998.
lithium, divalproex sodium, and carbamazepine in children Silverstone PH, Wu RH, O'Donnell T, Ulrich M, Asghar SJ, and adolescents with bipolar disorder. J Am Acad Child Hanstock CC: Chronic treatment with lithium, but not sodium Adolesc Psychiatry 39:713–720, 2000.
valproate, increases cortical N-acetyl-aspartate concentrations Kowatch RA, Fristad M, Birmaher B, Wagner KD, Findling RL, in euthymic bipolar patients. Int Clin Psychopharmacol 18:73– Hellander M: Treatment guidelines for children and adoles- cents with bipolar disorder. J Am Acad Child Adolesc Psy- Singh MK, Delbello MP, Adler CM, Stanford KE, Strakowski chiatry 44:213–23, 2005.
SM: Neuroanatomical characterization of child offspring of Ladouceur CD, Almeida JR, Birmaher B, Axelson DA, Nau S, bipolar parents. J Am Acad Child Adolesc Psychiatry 47:526– Kalas C, Monk K, Kupfer DJ, Phillips ML: Subcortical gray matter volume abnormalities in healthy bipolar off- Strakowski SM, Delbello MP, Adler CM: The functional neuro- spring: Potential neuroanatomical risk marker for bipolar anatomy of bipolar disorder: A review of neuroimaging disorder? J Am Acad Child Adolesc Psychiatry 47:532–539, findings. Mol Psychiatry 10:105–116, 2005.
Talairach J, Tournoux P: Co-Planar Stereotaxic Atlas of the Laeng P, Pitts RL, Lemire AL, Drabik CE, Weiner A, Tang H, Human Brain: 3-Dimensional Proportional System: An Ap- Thyagarajan R, Mallon BS, Altar CA: The mood stabilizer proach to Cerebral Imaging. New York: Thieme Medical valproic acid stimulates GABA neurogenesis from rat fore- Publishers, Inc., 1988.
brain stem cells. J Neurochem 91:238–251, 2004.
Wagner KD, Weller EB, Carlson GA, Sachs G, Biederman J, Lang PJ, Bradley MM, Cuthbert BN: International Affective Frazier JA, Wozniak P, Tracy K, Weller RA, Bowden C: An Picture System (IAPS): Technical manual and affective ratings.
open-label trial of divalproex in children and adolescents with Gainesville (Florida): NIMH Center for the Study of Emotion bipolar disorder. J Am Acad Child Adolesc Psychiatry and Attention, 1997.
Lapalme M, Hodgins S, LaRoche C: Children of parents with Wagner KD, Redden L, Kowatch R, Wilens TE, Segal S, Chang bipolar disorder: A metaanalysis of risk for mental disorders.
KD, Wozniak P, Vigna NV, Abi-Saab W, M. S: A double-blind, Can J Psychiatry 42:623–63, 1997.
randomized, placebo-controlled trial of divalproex ER in the Lehmann H, Ebert U, Loscher W: Amygdala-kindling induces a treatment of bipolar disorder in children and adolescents. (in lasting reduction of GABA-immunoreactive neurons in a dis- press, 2009).
crete area of the ipsilateral piriform cortex. Synapse 29:299–309, 1998.
Address reprint requests to: Loscher, W: Basic pharmacology of valproate: A review after 35 Kiki D. Chang, M.D.
years of use for the treatment of epilepsy. CNS Drugs 16:669– Stanford University School of Medicine Division of Child and Adolescent Psychiatry Manji HK, Lenox RH: Ziskind-Somerfeld Research Award.
Protein kinase C signaling in the brain: Molecular transduction Stanford, CA 94305-5540 of mood stabilization in the treatment of manic-depressiveillness. Biol Psychiatry 46:1328–1351, 1999.

Source: http://pediatricbipolar.stanford.edu/pdfs/Divalproex_in_high-risk_youth.pdf