© 2015 John Wiley & Sons A/S Published by John Wiley & Sons Ltd.

Bipolar Disorders 2015: 17: 496–506

BIPOLAR DISORDERS

Original Article

Lithium treatment and hippocampal subfields and amygdala volumes in bipolar disorder Hartberg CB, Jørgensen KN, Haukvik UK, Westlye LT, Melle I, Andreassen OA, Agartz I. Lithium treatment and hippocampal subfields and amygdala volumes in bipolar disorder. Bipolar Disord 2015: 17: 496–506. © 2015 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd. Objectives: Results from magnetic resonance imaging (MRI) studies are heterogeneous with regard to hippocampal and amygdala volume alterations in bipolar disorder (BD). Lithium treatment may influence both structures. It is unknown if lithium treatment has distinct effects on hippocampal subfield volumes and if subfield volumes change over the course of illness in BD.

Cecilie Bhandari Hartberga,b, Kjetil Nordbø Jørgensena,b, Unn Kristin Haukvikb, Lars Tjelta Westlyeb,c, Ingrid Melleb, Ole Andreas Andreassenb and Ingrid Agartza,b a Department of Psychiatric Research, Diakonhjemmet Hospital, bNORMENT/K.G. Jebsen Centre for Psychosis Research, Institute of Clinical Medicine, University of Oslo and Oslo University Hospital, cDepartment of Psychology, University of Oslo, Oslo, Norway

Methods: MRI scans were obtained for 34 lithium-treated patients with BD (Li+), 147 patients with BD who were not treated with lithium (NonLi), and 300 healthy controls. Hippocampal total and subfield volumes and amygdala volumes were automatically estimated using Freesurfer. General linear models were used to investigate volume differences between groups and the effects of illness course and lithium treatment. Results: The Non-Li BD group displayed significantly smaller bilateral cornu ammonis (CA) 2/3 and CA4/dentate gyrus (DG) subfields, total hippocampal volumes, right CA1 and right subiculum subfields, and left amygdala volume compared to healthy controls. There were no differences between the Li+ BD and either the Non-Li BD or the healthy control groups. In patients with numerous affective episodes, Non-Li BD patients had smaller left CA1 and CA2/3 volumes compared to Li+ BD patients and healthy controls. There were positive associations between lithium treatment duration and left amygdala volume. Conclusions: Hippocampal subfield and amygdala volumes were reduced in Non-Li BD patients compared to healthy controls, whereas the Li+ BD volumes were no different from those in Non-Li BD patients or healthy controls. Over the course of BD, lithium treatment might counteract reductions specifically in the left CA1 and CA2/3 hippocampal subfields and amygdala volumes, in accordance with the suggested neuroprotective effects of lithium.

Magnetic resonance imaging (MRI) of the brain is an important method for achieving a better understanding of the underlying neurobiology in severe mental illness. However, the results from previous neuroimaging studies of patients with bipolar disorder (BD) have been contradictory, and a metaanalysis revealed that the only consistent findings across MRI studies are increased lateral ventricle volumes and white matter hyperintensities (1). Limbic structures are of particular interest in BD

496

doi: 10.1111/bdi.12295 Key words: adult – affective episodes – crosssectional – illness course – illness duration – MRI – neuroimaging methods Received 29 August 2014, revised and accepted for publication 25 November 2014 Corresponding author: Cecilie Bhandari Hartberg, M.D., Ph.D. Diakonhjemmet Hospital PB 85 Vinderen 0319 Oslo Norway Fax: +47 22 45 85 01 E-mail: [email protected]

due to their role in emotional processing and memory functions, which are often found to be impaired in BD (2, 3). Whereas single studies on BD have shown smaller, same size, and larger limbic structures, meta-analyses have demonstrated hippocampal and amygdala volumes to be of a similar size in subjects with BD and healthy controls (4, 5). Lithium is considered to be standard medication in the long-term treatment of BD (6). Kempton

Lithium and limbic volumes in BD et al. (1) showed increasing global gray matter volumes with increasing proportion of patients using lithium medication, and larger hippocampi in lithium-treated patients with BD have been demonstrated in cross-sectional (7–9) and longitudinal studies (10, 11). In a meta-analysis, patients with BD with minimal lithium exposure demonstrated smaller hippocampal volumes than lithium-treated patients and healthy controls (12). There is also evidence for increased amygdala volumes in lithium-treated patients (9, 13–15). Longitudinal MRI studies have demonstrated an increase in gray matter in patients with BD after 4–12 weeks of lithium treatment (11, 16, 17). Lithium may exert a neuroprotective effect on the gray matter but the underlying mechanisms of the lithium-associated volume increase are unknown. Results from rodent studies have suggested that lithium increases neurogenesis in specific parts [i.e., the dental gyrus of the hippocampus (18)], while other studies have shown metabolite changes (19). It has been speculated that the gray matter increases observed with lithium use may be related to the osmotic effects of lithium, with subsequent swelling of neurons (20). A study investigating regional hippocampal effects reported tissue reduction corresponding to the right hippocampal cornu ammonis (CA) 1/2 and subiculum subfields in unmedicated compared to lithium-treated patients with BD and healthy controls (21). A recent study on patients with bipolar II disorder using the same segmentation algorithm as the present study reported smaller fimbria volumes in unmedicated compared to medicated patients; however, only one of the patients used lithium (22). Previous results from our research group, using partly overlapping samples, demonstrated smaller total hippocampal volumes (23) as well as hippocampal subfield volumes (24) in patients with BD compared to healthy controls. There were no effects of current lithium dose. In addition to specific medication effects, structural brain changes in BD may be related to illness course – for example, illness duration, number of episodes, history of psychosis, or type of BD. Some studies have shown associations between reduced total hippocampal volumes and illness duration (25) and duration of affective episodes (26), while others have failed to show associations (9). Studies have shown smaller amygdala volumes in pediatric BD (27–29) and larger volumes in adult BD (30). One study reported positive associations with the number of manic episodes and prior hospitalizations (31), while one review reported an increase in amygdala volume with age (32). However, a recent

review reported stable amygdala volumes in adult BD (33), so it is unknown if amygdala volumes change over the course of illness. Taken together, the literature on brain structure changes in BD is inconsistent. Previous studies have been mainly small and the patient samples heterogeneous, mostly due to differences in medication and illness course. Methodological differences across studies, such as different segmentation procedures and difficulties in segmenting tissues (e.g., the amygdale), may account for some of the discrepancies (34). Nevertheless, structural changes are found in the limbic structures implicated in the pathophysiology of BD (23). As one of the most commonly used medications in BD (i.e., lithium) has been reported to affect the hippocampus and the amygdala, it is important to investigate within-structure effects. Various hippocampal subregions have different cellular profiles, so specific subregional effects of lithium treatment can be expected. If we find specific subfield effects in the hippocampus, we can develop more-specific hypotheses to help to clarify the underlying pathophysiology, which is mainly unknown. Interestingly, despite the localized lithium-related structural changes reported in the dentate gyrus (DG) of the hippocampus in animal models (18), only a few human MRI studies have investigated the extent to which such changes can be related to medication and illness course. Previous studies on subregional hippocampal alterations in lithium-treated patients with BD have utilized surface-based methods (21), which allow for the estimation of shape differences. Methodological advances have enabled automated and reliable estimations of hippocampal subfield volumes in large subject samples (35), which have facilitated targeted studies on anatomically specific volumetric changes associated with medical treatment. The aim of the present study was twofold: (i) to compare volumes of the total hippocampus, the hippocampal subfields, and the amygdala between patients with BD treated with lithium (Li+), patients with BD with no history of lithium treatment (Non-Li), and healthy controls. Based on the results from previous studies, we hypothesized larger global as well as specific subfield volumes in Li+ compared to non-Li BD patients and healthy controls, in particular in the DG (18); and (ii) to explore the putative statistical effects of illness course (i.e., the number of affective episodes, illness duration, and duration of lithium treatment) on the total hippocampal, hippocampal subfield, and amygdala volumes.

497

Hartberg et al. Materials and methods Participants

All participants were recruited as part of the Thematically Organized Psychosis (TOP) Research Study (Oslo, Norway), an ongoing study on schizophrenia and bipolar spectrum disorders. Inpatients and outpatients alike were referred from psychiatric units from four major hospitals in the greater Oslo area. To ensure a representative control group, the controls were randomly selected from statistical records from the same catchment area as the patient groups, and contacted by letter inviting them to participate. All participants gave informed consent to participate. The study was approved by the Regional Committee for Medical Research Ethics and the Norwegian Data Inspectorate. Participants were excluded if they met the following criteria: a history of hospitalized head injury, a neurological disorder, IQ < 70 points, or age outside the range 18–65 years. The healthy control sample was evaluated by a clinical interview about severe mental disorder symptoms and the Primary Care Evaluation of Mental Disorders (PRIME-MD) (36). Controls were excluded if they had had a diagnosis of drug abuse/dependency in the previous three months, if they or any of their first-degree relatives had a lifetime history of a severe psychiatric disorder, or if they had a history of medical problems thought to interfere with brain function (37). Clinical characterization

The subject sample in the present study consisted of patients with a DSM-IV diagnosis within the BD spectrum (n = 181); they were divided into two clinical groups according to lithium treatment (n = 34) or non-lithium treatment (n = 147) at the time of scanning, and healthy controls (n = 300) (Table 1). Trained physicians and clinical psychologists performed the clinical assessments. Diagnosis was based on the Structured Clinical Interview for DSM-IV Axis I disorders (SCID-I) (38). Current depressive symptoms were rated using the Inventory of Depressive Symptomatology (IDS)–Clinician Rating (39) and current manic symptoms were rated using the Young Mania Rating Scale (YMRS) (40). Psychosocial functioning in patients was assessed using the Global Assessment of Functioning (GAF) scale, split version. Data on the dosage and duration of medication treatment were derived from interviews and medical records at the time of scanning. Illness duration was defined as the period between illness onset (defined as the first affective episode) and the time of MRI scanning (37).

498

Lithium medication is considered to be one of the standard treatments for BD in Norway and is used according to international and national guidelines. Participants who reported previous, but no current lithium treatment were excluded from the study (n = 11) as the effects of lithium on brain structure may persist after treatment discontinuation (41). Lithium blood levels were measured at the time of clinical interviews. In order to obtain a clinically representative sample, all patients were included, regardless of drug use, but abuse/dependency disorder was diagnosed (SCID module E) if present. Urine toxicology was available for a subset of the participants and the findings have been included in Supplementary Table 1. Demographic and clinical data are displayed in Table 1. MR image acquisition

All participants underwent MRI scanning on a 1.5T Siemens Magnetom Sonata scanner (Siemens Medical Solutions, Erlangen, Germany) equipped with a standard head coil. After a conventional three-plane localizer, two sagittal T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) volumes were acquired using the Siemens tfl3d1_ns pulse sequence (echo time = 3.93 msec, repetition time = 2,730 msec, inversion time = 1,000 msec, flip angle = 7°, field of view = 24 cm, voxel size = 1.33 9 0.94 9 1 mm3, number of partitions = 160). Acquisition parameters were optimized for increased gray/ white matter contrast. There were no major scanner upgrades during the study period, and patients and controls were scanned interchangeably. MR image processing

FreeSurfer version 5.2.0 (http://surfer.nmr. mgh.harvard.edu/) was used to estimate the volumes of the hippocampal subfields (35) (Supplementary Fig. 1), total hippocampal volume, amygdala volume, and intracranial volume (ICV) (42, 43). The hippocampal subfield segmentation technique used was based on a Bayesian modeling approach and manual delineations of each hippocampal subfield. A region of interest (ROI) around the hippocampal formation (94 9 66 9 144 voxels) was automatically assigned to each image using an affine mutual information-based registration technique by first aligning the whole-brain template and then the ROI template only. The hippocampal subfield volumes obtained with this method have been compared to manual hippocampal subfield tracings, and reliability measures were good for the larger subfields CA2/3, CA4/DG, and

77 (52) 62 (42) 8 (5) 56.2 (11.4) [28–84] 53.4 (12.6) [28–82] 2 [0–28] 15 [0–53] 27.3 (10.1) [7–63] 5.8 [0.5–40] 6 [1–252] 2 [0–245] 3 [0–90] 44 (35) 73 (53)

30 (88) 3 (9) 1 (3) 57.7 (11.9) [38–82] 53.2 (10.3) [35–76] 1 [0-11] 14 [4–53] 27.2 (9.4) [16–47] 4.8 [0.3–32] 7 [1–90] 2.5 [0–50] 4 [0–40] 12 (39) 21 (66)

36 (26) 52 (37) 67 (48) 14 (10) 4 (3) 3 (2)

0.6 (0.2) [0.4–0.9] 13 (38) 14 (41) 10 (29) 2 (6) 0 (0) 0 (0)

12 [2–115]

35.3 (11.8) [18–65] 59 (40) 13.4 (2.3) [9–20] 110 (86) 1.581 (155) [1.262–2.047]

BD Non-Li (n = 147)

34.7 (10.8) [20–65] 10 (29) 13.8 (2.4) [9–18] 27 (93) 1.582 (175) [1.306–2.016]

BD Li+ (n = 34)

Bipolar disorder

35.0 (9.6) [18–59] 158 (53) 14.2 (2.3) [9–20] 243 (92) 1,622 (169) [1,199–2,210]

Healthy controls (n = 300)

0.943 0.420 0.762 0.744 0.997 0.671 0.206

F = 0.01 U = 0.65 U = 0.09 U = 0.11 U < 0.01 v2 = 0.18 v2 = 1.60

0.500 0.331 0.401

0.480 0.927 0.043 0.949

F = 0.50 F = 0.01 U = 4.08 U < 0.01

v2 = 0.45 v2 = 0.95 v2 = 0.71

< 0.001 < 0.001

v2 = 14.69 v2 = 13.35

0.145 0.664 0.052

0.933 0.004 0.015 0.386 0.032

F = 0.07 v2 = 10.86 F = 4.26 v2 = 4.15 F = 3.47

v2 = 2.12 v2 = 0.19 v2 = 3.77

p-value

Test statistic

Statistics

Significant results (p < 0.05) are presented in bold. BD = bipolar disorder; BD-I = bipolar I disorder; BD-II = bipolar II disorder; BD-NOS = bipolar disorder not otherwise specified; GAF = Global Assessment of Functioning; ICV = intracranial volume; IDS = Inventory of Depression Scale; Li+ = lithium treated; Non-Li = not treated with lithium; SD = standard deviation; YMRS = Young Mania Rating Scale. Missing data: Education: BD Li+ (n = 5), BD Non-Li (n = 17), healthy controls (n = 37); Handedness: BD Li+ (n = 5), BD Non-Li (n = 19), healthy controls (n = 36); GAF-f: BD Non-Li (n = 6); YMRS: BD Li+ (n = 1), BD Non-Li (n = 12); IDS: BD Li+ (n = 1), BD Non-Li (n = 20); Age at onset: BD Li+ (n = 2), BD Non-Li (n = 10); Elevated episodes: BD Non-Li (n = 24); Depressive episodes: BD Li+ (n = 1), BD Non-Li (n = 18); Duration Li medication: BD Li+ (n = 1); Li blood level: BD Li+ (n = 10). a Euthymic = IDS ≤ 13 and YMRS ≤ 7.

Age, years, mean (SD) [range] Gender, male, n (%) Education, years, mean (SD) [range] Handedness, right, n (%) ICV, cm3, mean (SD) [range] Clinical variables BD-I, n (%) BD-II, n (%) BD-NOS, n (%) GAF symptom, mean (SD) [range] GAF function, mean (SD) [range] YMRS score, median [range] IDS score, median [range] Illness course Age at onset, years, mean (SD) [range] Illness duration, years, median [range] No. of affective episodes, median [range] No. of elevated episodes, median [range] No. of depressive episodes, median [range] Euthymic, n (%)a History of psychosis, n (%) Medication Duration of Li medication, months, median [range] Li blood level, mean (SD) [range] Antidepressants, n (%) Antipsychotic agents, n (%) Antiepileptic agents, n (%) Abuse/dependency Alcohol, n (%) Cannabis, n (%) Stimulants, n (%)

Table 1. . Demographics and clinical data

Lithium and limbic volumes in BD

499

Hartberg et al. subiculum, with acceptable reliability for the CA1, presubiculum, and fimbria (35). In the present study, we included the CA1, CA2/3, subiculum, presubiculum, CA4/DG, and fimbria subfields. The MRI post-processing procedures were fully automated without manual editing. All segmented scans were inspected visually following standard procedures. Statistical analysis

All statistical analyses were performed using the statistical package SPSS version 18 (IBM, SPSS Inc., Chicago, IL, USA). Group differences in demographic and clinical data were analyzed using analysis of variance for normally distributed data, Mann–Whitney U-tests for non-normally distributed continuous data, and chi-square tests for categorical data. Hippocampal and amygdala volume differences between groups were tested using analysis of covariance, with structure volume as the dependent variable, group as the fixed factor, and age and ICV as covariates. The effects of gender were ruled out by including it as a covariate in this model, with no changes in significance threshold. Hence, the gender variable was left out of the final statistical model. For structures showing overall group differences, we performed post-hoc pairwise comparisons and the statistical tests were considered significant for Bonferroni-corrected p-values < 0.05. The analyses were repeated with the subgroup of bipolar I disorder (BD-I) exclusively, and with history of psychosis and antipsychotic, antiepileptic, and antidepressant medication as covariates. As urine toxicology data were not obtained from all participants, we chose to repeat the analyses with only those who had been diagnosed with drug/alcohol abuse or dependency at the time of inclusion to the study. Analysis of illness course effects on hippocampal subfield or amygdala volumes. We investigated the interaction effects between group and illness course variables – that is, illness duration and number of affective episodes. Previous cross-sectional studies have shown lithium treatment effects to be more pronounced or evident at a later stage of the illness (12). The illness course variable was highly skewed and was thus dichotomized by the median value – that is, short illness duration ≤5 years, long illness duration >5 years; and fewer affective episodes ≤6, and numerous affective episodes >6. The dichotomized variables were entered in the analyses and checked for interaction effects with group. For structure volumes for which an interaction effect

500

was found, the between-group analyses were repeated for the subgroup of patients at a later stage of illness (i.e., with long illness duration or numerous affective episodes). Hierarchical linear regression forced-entry models were used to study the strength of associations between structure volumes and lithium treatment duration in the Li+ BD group. For each structure, volume was entered as the dependent variable, and age and ICV were entered as independent variables in the first block, while duration of lithium treatment was entered in the second block. We report uncorrected and corrected results after multiple comparison control with a false discovery rate (FDR) of 5%.

Results Group differences in demographics and clinical data

There was a higher proportion of men in the healthy control group compared to the Li+ BD and Non-Li groups, and a higher proportion of BD-I diagnoses in the Li+ group compared to the Non-Li BD group. Both the Li+ and Non-Li BD groups had fewer years of education than healthy controls. The non-Li group had higher YMRS scores than the Li+ group but there was no difference in the proportion of patients with symptoms of mania between BD groups: 15% (five) in the Li+ group and 21% (28) in the Non-Li group (v2 = 0.53, p = 0.469). There was no difference in the proportion of patients with current symptoms of depression between BD groups: 55% (17) in the Li+ group and 54% (68) in the Non-Li group (v2 = 0.02, p = 0.897). There were no other group differences in demographic or clinical data (Table 1). Hippocampal total and subfield volumes and amygdala volumes

We found significant overall group differences for right CA1 and subiculum subfields, bilateral CA2/3, CA4/DG subfields and total hippocampal volumes, and left amygdala volumes across all groups. The results remained significant after corrections with an FDR (Benjamini–Hochberg) of 5%. Pairwise group comparisons showed smaller volumes in the Non-Li BD group compared to healthy controls for the right CA1 and subiculum subfields, bilateral CA2/3, CA4/DG subfields and total hippocampal volumes, and left amygdala volumes (see Table 2 for Bonferroni-corrected pvalues, and Supplementary Fig. 2). There were no significant differences between Li+ patients with BD and healthy controls or between Li+ patients

Lithium and limbic volumes in BD Table 2. . Statistical comparisons of hippocampal and amygdala volumes (mm3) across groups Bipolar disorder

Left Presubiculum CA1 CA 2/3 Fimbria Subiculum CA4/DG Hippocampal total Amygdala Right Presubiculum CA1 CA 2/3 Fimbria Subiculum CA4/DG Hippocampal total Amygdala

Healthy controls

Li+ (n = 34)

Non-Li (n = 147)

(n = 300)

Mean (SE)

Mean (SE)

Mean (SE)

Statistics ANCOVA F(2.475)

p-value

469 (8.6) 327 (6.6) 999 (19.4) 70 (3.6) 647 (11.0) 554 (10.7) 4,113 (59.2) 1,591 (32.9)

466 (4.1) 323 (3.2) 986 (9.3) 74 (1.7) 647 (5.3) 551 (5.1) 4,070 (28.5) 1,560 (15.9)

475 (2.9) 329 (2.2) 1,023 (6.5) 75 (1.2) 662 (3.7) 571 (3.6) 4,203 (20.0) 1,618 (11.1)

1.6 0.9 5.3 0.6 3.2 5.5 7.6 4.5

0.195 0.391 0.005 0.530 0.041 0.004 0.001 0.011

443 (7.9) 346 (7.2) 1,049 (19.2) 58 (3.5) 658 (11.0) 584 (11.0) 4,237 (58.6) 1,661 (31.9)

446 (3.8) 334 (3.5) 1,035 (9.3) 64 (1.7) 644 (5.3) 575 (5.3) 4,128 (28.3) 1,605 (15.4)

454 (2.7) 348 (2.4) 1,072 (6.4) 65 (1.2) 664 (3.7) 596 (3.7) 4,276 (19.8) 1,649 (10.8)

1.8 4.9 5.7 1.8 4.9 5.1 9.1 3.1

0.166 0.008 0.004 0.166 0.008 0.006 6) affective episodes showed that the Non-Li BD group (n = 57) had significantly smaller left CA1 and CA2/3 volumes compared to the Li+ BD (n = 17) group and healthy controls (Fig. 1). There were no significant differences in demographic or clinical variables between the subgroup of patients and healthy controls. There were no significant interaction effects with illness duration. Regression analyses of the Li+ BD group showed significant positive associations between lithium treatment duration and left amygdala volumes (R2 change = 0.079, b = 0.29, t = 2.15, p = 0.040). There was a trend for an association with the right amygdala (R2 change = 0.065,

b = 0.26, t = 1.82, p = 0.079), but not for hippocampal volumes. The effects of illness course are reported with uncorrected p values (significance level 0.05). These nominal significance levels did not remain significant after corrections for multiple comparisons. Discussion

To the best of our knowledge, the present study is the first to address the putative effects of lithium on hippocampal subfield volumes in BD. We found significantly smaller hippocampal subfield and amygdala volumes in the Non-Li BD group compared to healthy controls, while the volumes in the Li+ BD group were no different from either the Non-Li group or healthy controls. However, for patients with numerous (>6) affective episodes, the left CA1 and CA 2/3 volumes in the Non-Li BD group were smaller than those in the Li+ BD group. The duration of lithium treatment was associated with increased left amygdala volumes within the Li+ BD group. The findings of smaller total hippocampal volumes in the Non-Li BD group compared to healthy controls are in accordance with previous studies, summarized in a meta-analysis by Hajek et al. (12). As previously shown in the present sample (24), specific hippocampal subfields are smaller

501

Hartberg et al. A

B

Fig. 1. Results from analyses of the subgroup of patients with a history of more than six affective episodes and healthy controls (HC). The histogram shows (A) lower left hippocampal CA1 volume in Non-Li patients with bipolar disorder (BD) compared to Li+ patients with bipolar disorder (BD) (p = 0.023, Cohen’s d = 0.53) and healthy controls (p = 0.028, Cohen’s d = 0.29), and (B) lower left hippocampal CA2/3 volume in Non-Li patients with BD compared to Li+ patients with BD (p = 0.048, Cohen’s d = 0.49) and healthy controls (p = 0.003, Cohen’s d = 0.39). There were no differences between Li+ patients with BD and healthy controls in either case. Li+ = lithium-treated; Non-Li = not treated with lithium.

in BD than in healthy controls. We did not confirm our hypothesis of larger volumes in the Li+ BD group as compared to the Non-Li BD group and healthy controls. This may have been because of statistical power issues as, for the majority of structures, the Li+ BD group showed numerically larger volumes compared to the Non-Li BD group, and the effect sizes showed that the mean of most Li+ BD volumes was intermediate to the mean volumes in healthy controls and the Non-Li BD group (Supplementary Table 2). However, in line with the present results, previous studies of patients with BD that have included patients at an early stage of illness have failed to show group differences between lithium-treated patients and healthy controls (25, 44). Several studies have suggested that the association between lithium treatment and hippocampal volumes is complex and possibly nonlinear (17, 21, 25). Our findings of larger volumes specifically in the CA1 and CA2/3 in Li+ BD compared to Non-Li BD patients with numerous affective episodes (>6)

502

are in concordance with the previous MRI studies on regional surface changes of the hippocampus (21, 45). The association between left amygdala volume and the duration of lithium treatment is also substantiated by the existing literature (9, 13). As lithium exerts its effect on various molecular targets, including changes in gene transcription (46), both acute and chronic effects can be expected. Although there were no differences between the Li+ BD and Non-Li BD groups when all patients were analyzed, such differences were found between the subgroups of patients with BD with numerous episodes (>6). This could lead one to speculate that the neuroprotective effects of lithium on specific subfield volumes accumulate during the course of illness. Of clinical significance, an increased number of episodes in BD has been related to illness progression as evidenced by cognitive deterioration, increases in comorbid somatic diseases, and poor outcome (47). Hypothetically, if lithium treatment is associated with a smaller decrease in hippocampal volumes, lithium may counteract illness progression through its neuroprotective properties, as hippocampal subfield volumes have been related neurocognitive functions known to be impaired in BD (48, 49). Neuropathophysiology

The unique role of each hippocampal subregion is not fully known. The CA1 subfield, and particularly the subiculum, provides the main hippocampal outflow. The CA1 receives and integrates information from the entorhinal cortex and CA2/3 region, while the subiculum receives input from CA1. It has been shown that lithium can enhance pre- and post-synaptic excitatory signaling in CA1. Changes in the CA1 can also influence other structures through interactions with a wide range of subcortical and cortical regions, including the amygdala, which plays an important functional role in BD (50). So far, neurogenesis has been localized only to the DG of the hippocampus (18). We did not find a lithium-related volume increase in the DG. However, newborn neurons project onto the CA3, where we found larger volumes in Li+ patients compared to Non-Li patients with BD with numerous affective episodes. As such, the present findings may indirectly support lithium involvement in neurogenesis. Alternatively, the newborn cells may integrate with the hippocampal trisynaptic circuit and stimulate growth in CA1. In addition, lithium may act as a neuroprotective agent by acting through one or more of the following mechanisms: modulation of excitatory neurotransmitter-related

Lithium and limbic volumes in BD neurotoxicity [inactivating N-methyl-D-aspartate (NMDA) receptors and increasing the release of caminobutyric acid (GABA)] (51, 52), increasing neuroprotective proteins such as brain-derived neurotrophic factor and B-cell lymphoma-2, and decreasing pro-apoptotic enzymes such as glycogen synthase kinase-3 (53, 54). N-acetyl-aspartate, which is a marker of neuronal viability and function, has been found to be increased at therapeutic levels of lithium treatment (55). Additional neuroprotective effects are reduced proinflammatory status and decreased oxidative stress (56). The findings in these studies suggest that lithium helps to preserve gray matter volumes in BD, rather than cause regrowth – that is, it has neuroprotective rather than neuroregenerative effects. This would also explain the present comparable volumes in the lithium-treated group versus the healthy controls, which have also been reported in other studies. It has been proposed that estimated gray matter changes can be a result of the osmotic effects of lithium, leading to neuronal swelling and a subsequent increase in gray matter (21). Vernon and Hajek (57) pointed out that white matter does not seem to be influenced by lithium treatment, rendering pure osmotic action less likely to be the only explanation for volume increases in gray matter. Limitations

The patients were not randomized to the treatment groups, so a selection bias cannot be excluded. Even though the size of the present sample of Li+ patients with BD was equal to or larger than that in previous studies, using a larger sample size may detect additional subfield tissue expansion in the Li+ group. The Li+ BD group was small (n = 34) as compared to the other groups [Non-Li (n = 147), healthy controls (n = 300)], which may have skewed the results from the group comparisons; however, the results were in the expected direction. There were group differences in gender, BD subtype, mania rating score, and proportion of patients with BD treated with antiepileptic agents. However, we corrected for gender differences and the analyses were repeated in the BD-I subgroup, which yielded highly similar results. We did not have systematic information about previous lithium levels to check for past compliance with treatment. Duration of treatment was noted as reported by the subjects and from their medical records. Furthermore, we corrected for neuroleptic medications other than lithium. Previous studies have not shown a consistent effect of antiepileptic agents on brain structure (15). Mood states have been shown to influence limbic structure volumes (58). How-

ever, the preliminary analyses showed no correlation between mania or depression symptom scores and brain structure in the present study. Drug or alcohol abuse may have an effect on the volumes of brain structures, but the number of subjects with ongoing drug or alcohol abuse in the present study was small, so we did not perform separate analyses for these subjects. Using a magnetic field strength higher than 1.5 T, with targeted and longer MRI acquisition, and the use of manual tracings would probably have increased the reliability and accuracy of subfield measurements. However, increasing the imaging time also increases the risk of motion artifacts, which decrease the accuracy of measurements. In a recent study, large significant positive correlations were reported between segmentation results from using magnetic field strengths of 1.5 T and 3 T for total hippocampal volume and most subfields, and between images with shorter and longer scan times (59). Automated measurements of cellular subfields may not be as precise as manual delineated ones but are systematic and repeatable and enable analysis in large imaging studies. Manual labeling is time consuming and not feasible in studies such as the present one, with several hundred participants and repeated scanning sessions. The results from the present and other recent studies (60, 61) suggest that research on automated methods and their ability to reflect the exact anatomical boundaries is needed. Optimal studies would include manual segmentation of at least a set of the images, and a high resolution, preferably using both T1- and T2-weighted images with good contrast resolution. The present study was based on cross-sectional data and the conclusions regarding changes over time should therefore be interpreted with caution. Conclusions

We found smaller hippocampal subfield and amygdala volumes in the Non-Li BD group compared to healthy controls, but the volumes in the Li+ BD group were not significantly different from either the Non-Li group or healthy controls. Follow-up analyses in a subgroup of patients with numerous affective episodes may indicate that lithium treatment increases or counteracts reductions in the left CA1 and CA2/3 hippocampal subfield volumes in BD. Both the results from the subgroup of BD with numerous affective episodes and the association between amygdala volume and lithium treatment are in accordance with the suggested neuroprotective effects of lithium. However,

503

Hartberg et al. the results should be interpreted with caution and replicated in independent studies. Future studies on BD should take into account that lithium treatment effects may be localized and depend on illness stage. Long-term longitudinal studies and randomized controlled trials are needed to disentangle the unique effects of lithium from possible confounding factors. Acknowledgements The authors thank the patients and controls for participating in the study, and the TOP study group members for contributing to data collection. The study was supported by grants to the TOP study group by the Research Council of Norway (#223273, #167153/V50, #163070/V50), the South-Eastern Norway Regional Health Authority (#2004-123, #2008-039), and the KG Jebsen Foundation.

Author roles CBH and KNJ wrote the first draft of the paper. IA planned, supervised, and coordinated the work. CBH and KNJ analyzed data and interpreted the results. All authors contributed to and have approved the final manuscript.

Disclosures OAA has received speakers’ honoraria from GlaxoSmithKline, Eli Lilly & Co., Otsuka and Lundbeck; however, these were without relevance to the subject matter of the manuscript. CBH, KNJ, UKH, LTW, IM, and IA report no direct or indirect financial interests or potential conflicts of interest relevant to the subject of the manuscript.

References 1. Kempton MJ, Geddes JR, Ettinger U, Williams SC, Grasby PM. Meta-analysis, database, and meta-regression of 98 structural imaging studies in bipolar disorder. Arch Gen Psychiatry 2008; 65: 1017–1032. 2. Torres IJ, Boudreau VG, Yatham LN. Neuropsychological functioning in euthymic bipolar disorder: a meta-analysis. Acta Psychiatr Scand Suppl 2007; 434: 17–26. 3. Mercer L, Becerra R. A unique emotional processing profile of euthymic bipolar disorder? A critical review J Affect Disord 2013; 146: 295–309. 4. Hajek T, Kopecek M, Kozeny J, Gunde E, Alda M, Hoschl C. Amygdala volumes in mood disorders – metaanalysis of magnetic resonance volumetry studies. J Affect Disord 2009; 115: 395–410. 5. Ellison-Wright I, Bullmore E. Anatomy of bipolar disorder and schizophrenia: a meta-analysis. Schizophr Res 2010; 117: 1–12. 6. Guidelines NICE. Bipolar Disorder: The Management of Bipolar Disorder in Adults, Children and Adolescents, in Primary and Secondary Care. London: National Institute for Health and Care Excellence, 2006. 7. Yucel K, Taylor VH, McKinnon MC et al. Bilateral hippocampal volume increase in patients with bipolar disorder and short-term lithium treatment. Neuropsychopharmacology 2008; 33: 361–367.

504

8. Foland LC, Altshuler LL, Sugar CA et al. Increased volume of the amygdala and hippocampus in bipolar patients treated with lithium. NeuroReport 2008; 19: 221–224. 9. Hallahan B, Newell J, Soares JC et al. Structural magnetic resonance imaging in bipolar disorder: an international collaborative mega-analysis of individual adult patient data. Biol Psychiatry 2011; 69: 326–335. 10. Yucel K, McKinnon MC, Taylor VH et al. Bilateral hippocampal volume increases after long-term lithium treatment in patients with bipolar disorder: a longitudinal MRI study. Psychopharmacology 2007; 195: 357–367. 11. Lyoo IK, Dager SR, Kim JE et al. Lithium-induced gray matter volume increase as a neural correlate of treatment response in bipolar disorder: a longitudinal brain imaging study. Neuropsychopharmacology 2010; 35: 1743– 1750. 12. Hajek T, Kopecek M, Hoschl C, Alda M. Smaller hippocampal volumes in patients with bipolar disorder are masked by exposure to lithium: a meta-analysis. J Psychiatry Neurosci 2012; 37: 333–343. 13. Usher J, Menzel P, Schneider-Axmann T et al. Increased right amygdala volume in lithium-treated patients with bipolar I disorder. Acta Psychiatr Scand 2010; 121: 119– 124. 14. Savitz J, Nugent AC, Bogers W et al. Amygdala volume in depressed patients with bipolar disorder assessed using high resolution 3T MRI: the impact of medication. NeuroImage 2010; 49: 2966–2976. 15. Germana C, Kempton MJ, Sarnicola A et al. The effects of lithium and anticonvulsants on brain structure in bipolar disorder. Acta Psychiatr Scand 2010; 122: 481–487. 16. Moore GJ, Bebchuk JM, Wilds IB, Chen G, Manji HK. Lithium-induced increase in human brain grey matter. Lancet 2000; 356: 1241–1242. 17. Selek S, Nicoletti M, Zunta-Soares GB et al. A longitudinal study of fronto-limbic brain structures in patients with bipolar I disorder during lithium treatment. J Affect Disord 2013; 150: 629–633. 18. Chen G, Rajkowska G, Du F, Seraji-Bozorgzad N, Manji HK. Enhancement of hippocampal neurogenesis by lithium. J Neurochem 2000; 75: 1729–1734. 19. Moore CM, Breeze JL, Gruber SA et al. Choline, myoinositol and mood in bipolar disorder: a proton magnetic resonance spectroscopic imaging study of the anterior cingulate cortex. Bipolar Disord 2000; 2: 207–216. 20. Phatak P, Shaldivin A, King LS, Shapiro P, Regenold WT. Lithium and inositol: effects on brain water homeostasis in the rat. Psychopharmacology 2006; 186: 41–47. 21. Bearden CE, Thompson PM, Dutton RA et al. Threedimensional mapping of hippocampal anatomy in unmedicated and lithium-treated patients with bipolar disorder. Neuropsychopharmacology 2008; 33: 1229–1238. ashagen T, Westlye LT, Bøen E et al. Evidence for 22. Elvs reduced dentate gyrus and fimbria volume in bipolar II disorder. Bipolar Disord 2013; 15: 167–176. 23. Rimol LM, Hartberg CB, Nesvag R et al. Cortical thickness and subcortical volumes in schizophrenia and bipolar disorder. Biol Psychiatry 2010; 68: 41–50. 24. Haukvik UK, Westlye LT, Mørch-Johnsen L et al. In vivo hippocampal subfield volumes in schizophrenia and bipolar disorder. Biol Psychiatry 2015; 77: 581–588. 25. Hajek T, Cullis J, Novak T et al. Hippocampal volumes in bipolar disorders: opposing effects of illness burden and lithium treatment. Bipolar Disord 2012; 14: 261–270. 26. Wijeratne C, Sachdev S, Wen W et al. Hippocampal and amygdala volumes in an older bipolar disorder sample. Int Psychogeriatr 2013; 25: 54–60.

Lithium and limbic volumes in BD 27. Chen BK, Sassi R, Axelson D et al. Cross-sectional study of abnormal amygdala development in adolescents and young adults with bipolar disorder. Biol Psychiatry 2004; 56: 399–405. 28. Chang K, Karchemskiy A, Barnea-Goraly N, Garrett A, Simeonova DI, Reiss A. Reduced amygdalar gray matter volume in familial pediatric bipolar disorder. J Am Acad Child Adolesc Psychiatry 2005; 44: 565–573. 29. Dickstein DP, Milham MP, Nugent AC et al. Frontotemporal alterations in pediatric bipolar disorder: results of a voxel-based morphometry study. Arch Gen Psychiatry 2005; 62: 734–741. 30. Brambilla P, Hatch JP, Soares JC. Limbic changes identified by imaging in bipolar patients. Curr Psychiatry Rep 2008; 10: 505–509. 31. Foland-Ross LC, Thompson PM, Sugar CA et al. Threedimensional mapping of hippocampal and amygdalar structure in euthymic adults with bipolar disorder not treated with lithium. Psychiatry Res 2013; 211: 195–201. 32. Usher J, Leucht S, Falkai P, Scherk H. Correlation between amygdala volume and age in bipolar disorder – a systematic review and meta-analysis of structural MRI studies. Psychiatry Res 2010; 182: 1–8. 33. Lim CS, Baldessarini RJ, Vieta E, Yucel M, Bora E, Sim K. Longitudinal neuroimaging and neuropsychological changes in bipolar disorder patients: review of the evidence. Neurosci Biobehav Rev 2013; 37: 418–435. 34. Entis JJ, Doerga P, Barrett LF, Dickerson BC. A reliable protocol for the manual segmentation of the human amygdala and its subregions using ultra-high resolution MRI. NeuroImage 2012; 60: 1226–1235. 35. Van Leemput K, Bakkour A, Benner T et al. Automated segmentation of hippocampal subfields from ultra-high resolution in vivo MRI. Hippocampus 2009; 19: 549–557. 36. Spitzer RL, Williams JB, Kroenke K et al. Utility of a new procedure for diagnosing mental disorders in primary care. The PRIME-MD 1000 study. JAMA 1994; 272: 1749– 1756. 37. Hartberg CB, Sundet K, Rimol LM et al. Brain cortical thickness and surface area correlates of neurocognitive performance in patients with schizophrenia, bipolar disorder, and healthy adults. J Int Neuropsychol Soc 2011; 17: 1080–1093. 38. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSM-IV Axis I Disorders (SCID I). New York, NY: Biometrics Research, New York State Psychiatric Institute, 1997. 39. Rush AJ, Gullion CM, Basco MR, Jarrett RB, Trivedi MH. The Inventory of Depressive Symptomatology (IDS): psychometric properties. Psychol Med 1996; 26: 477–486. 40. Young RC, Biggs JT, Ziegler VE, Meyer DA. A rating scale for mania: reliability, validity and sensitivity. Br J Psychiatry 1978; 133: 429–435. 41. Vernon AC, Natesan S, Crum WR et al. Contrasting effects of haloperidol and lithium on rodent brain structure: a magnetic resonance imaging study with postmortem confirmation. Biol Psychiatry 2012; 71: 855–863. 42. Fischl B, Salat DH, Busa E et al. Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain. Neuron 2002; 33: 341–355. 43. Buckner RL, Head D, Parker J et al. A unified approach for morphometric and functional data analysis in young, old, and demented adults using automated atlas-based head size normalization: reliability and validation against

44.

45.

46.

47.

48. 49.

50.

51.

52.

53.

54.

55.

56.

57.

58.

59.

60.

61.

manual measurement of total intracranial volume. NeuroImage 2004; 23: 724–738. McDonald C, Marshall N, Sham PC et al. Regional brain morphometry in patients with schizophrenia or bipolar disorder and their unaffected relatives. Am J Psychiatry 2006; 163: 478–487. van Erp TG, Thompson PM, Kieseppa T et al. Hippocampal morphology in lithium and non-lithium-treated bipolar I disorder patients, non-bipolar co-twins, and control twins. Hum Brain Mapp 2012; 33: 501–510. Malhi GS, Tanious M, Das P, Coulston CM, Berk M. Potential mechanisms of action of lithium in bipolar disorder. Current understanding. CNS Drugs 2013; 27: 135–153. Post RM, Fleming J, Kapczinski F. Neurobiological correlates of illness progression in the recurrent affective disorders. J Psychiatr Res 2012; 46: 561–573. Kesner RP. Behavioral functions of the CA3 subregion of the hippocampus. Learn Mem 2007; 14: 771–781. Travis SG, Huang Y, Fujiwara E et al. High field structural MRI reveals specific episodic memory correlates in the subfields of the hippocampus. Neuropsychologia 2014; 53: 233–245. Schloesser RJ, Martinowich K, Manji HK. Mood-stabilizing drugs: mechanisms of action. Trend Neurosci 2012; 35: 36–46. Nonaka S, Hough CJ, Chuang DM. Chronic lithium treatment robustly protects neurons in the central nervous system against excitotoxicity by inhibiting N-methyl-Daspartate receptor-mediated calcium influx. Proc Natl Acad Sci USA 1998; 95: 2642–2647. Brunello N, Tascedda F. Cellular mechanisms and second messengers: relevance to the psychopharmacology of bipolar disorders. Int J Neuropsychopharmacol 2003; 6: 181– 189. Gould TD, Chen G, Manji HK. In vivo evidence in the brain for lithium inhibition of glycogen synthase kinase-3. Neuropsychopharmacology 2004; 29: 32–38. Malhi GS, Tanious M, Das P, Berk M. The science and practice of lithium therapy. Aust N Z J Psychiatry 2012; 46: 192–211. Moore GJ, Bebchuk JM, Hasanat K et al. Lithium increases N-acetyl-aspartate in the human brain: in vivo evidence in support of bcl-2’s neurotrophic effects? Biol Psychiatry 2000; 48: 1–8. Machado-Vieira R, Manji HK, Zarate CA Jr. The role of lithium in the treatment of bipolar disorder: convergent evidence for neurotrophic effects as a unifying hypothesis. Bipolar Disord 2009; 11 (Suppl. 2): 92–109. Vernon AC, Hajek T. Effects of lithium on magnetic resonance imaging signal might not preclude increases in brain volume after chronic lithium treatment. Biol Psychiatry 2013; 74: e39–e40. Foland-Ross LC, Brooks JO 3rd, Mintz J et al. Moodstate effects on amygdala volume in bipolar disorder. J Affect Disord 2012; 139: 298–301. Krogsrud SK, Tamnes CK, Fjell AM et al. Development of hippocampal subfield volumes from 4 to 22 years. Hum Brain Mapp 2014; 35: 5646–5657. de Flores R, La Joie R, Landeau B et al. Effects of age and Alzheimer’s disease on hippocampal subfields: comparison between manual and freesurfer volumetry. Hum Brain Mapp 2015; 36: 463–474. Wisse LE, Biessels GJ, Geerlings MI. A critical appraisal of the hippocampal subfield segmentation package in FreeSurfer. Front Aging Neurosci 2014; 6: 261.

505

Hartberg et al. Supporting Information Additional Supporting Information may be found in the online version of this article: Fig. S1. Hippocampal subfield segmentation. Coronal (left) and sagittal (right) view. Color code: red, CA1; blue, CA2/3; dark brown, CA4/dentate gyrus; purple, fimbria; orange, presubiculum; green, subiculum; light blue, hippocampal fissure; light yellow, ‘the remaining’ hippocampus. Fig. S2. Comparisons of hippocampal and amygdala volumes (mm3) across groups. The figure presents mean structure volumes, with a 95% confidence interval for all volumes where

506

significant differences were found across groups. Post-hoc pairwise group comparisons showed that, for all of these volumes, the Non-Li BD group displayed significantly smaller volumes than the healthy control group (p < 0.05, Bonferroni corrected). There were no significant differences between Li+ BD patients and healthy controls or between Li+ BD patients and Non-Li BD patients. Table S1. Positive drug screening (urine test on day of magnetic resonance imaging scan). Table S2. Effect sizes (Cohen’s d) for pairwise comparisons of hippocampal and amygdala volumes.