Dr Kamilla Miskowiak 1

Title: Effects of erythropoietin on hippocampal volume and memory in mood disorders

Short title: Hippocampal effects of erythropoietin in mood disorders

Authors: Kamilla W. Miskowiak, Maj Vinberg, Julian Macoveanu, Hannelore Ehrenreich, Nicolai

Køster, Becky Inkster, Olaf B. Paulson, Lars V. Kessing, Arnold Skimminge, Hartwig R. Siebner

Affiliations: KWM, MV, NK, LVK: Psychiatric Centre Copenhagen, Copenhagen University

Hospital, Rigshospitalet, Emails: Kamilla@miskowiak.dk, maj.vinberg@regionh.dk,

ksh187@alumni.ku.dk, lars.vedel.kessing@regionh.dk

HE: Clinical Neuroscience, Max Planck Institute of Experimental Medicine, Göttingen, Email:

ehrenreich@em.mpg.de

JM, AS, HRS: Danish Research Centre for Magnetic Resonance, Copenhagen University Hospital

Hvidovre, Emails: julianm@drcmr.dk, arnolds@drcmr.dk, hartwig.siebner@drcmr.dk,

HRS: Department of Neurology, Copenhagen University Hospital Bispebjerg

BI: Department of Psychiatry, School of Clinical Medicine, University of Cambridge, Email:

bi212@medschl.cam.ac.uk

OBP: Neurobiological Research Unit, Copenhagen University Hospital, Rigshospitalet, Email:

olaf.paulson@nru.dk

Corresponding author: Kamilla Woznica Miskowiak, Psychiatric Centre Copenhagen,

Copenhagen University Hospital, Rigshospitalet, Blegdamsvej 9, DK-2100 Copenhagen, Email:

Kamilla@miskowiak.dk

Key words: Bipolar disorder, treatment-resistant depression, erythropoietin, hippocampus,

cognition, MRI

Dr Kamilla Miskowiak 2

Word count: Abstract: 248/ Article body (excl. abstract, acknowledgments, financial disclosures,

legends and references): 3,995

Figures/ tables: 2/ 1

Supplementary information: 2

Dr Kamilla Miskowiak 3

Abstract

Objective

Persistent cognitive dysfunction in depression and bipolar disorder (BD) impedes patients’

functional recovery. Erythropoietin (EPO) increases neuroplasticity and reduces cognitive

difficulties in treatment-resistant depression (TRD) and remitted BD. This magnetic resonance

imaging (MRI) study assessed the neuroanatomical basis for these effects.

Methods

Patients with TRD who were moderately depressed or BD in partial remission were randomized to

8 weekly EPO (40,000 IU) or saline infusions in a double-blind, parallel–group design. Patients

underwent MRI, memory assessment with the Rey Auditory Verbal Learning Test, and mood

ratings with the Beck Depression Inventory, Hamilton Depression Rating Scale, and Young Mania

Rating Scale at baseline and week 14. Hippocampus segmentation and analysis of hippocampal

volume, shape and gray matter density were conducted FMRIB Software Library (FSL) tools.

Memory change was analyzed with repeated-measures analysis of covariance adjusted for

depression symptoms, diagnosis, age and gender.

Results

84 patients were randomized; 1 patient withdrew and data collection was incomplete for 14; data

was thus analyzed for 69 patients (EPO: N = 35, saline: N = 34). Compared with saline, EPO was

associated mood-independent memory improvement and reversal of brain matter loss in the left

hippocampal CA1-3 and subiculum. Using the entire sample, memory improvement was associated

with subfield hippocampal volume increase independent of mood change.

Dr Kamilla Miskowiak 4

Conclusions

EPO-associated memory improvement in TRD and BD may be mediated by reversal of brain matter

loss in a subfield of the left hippocampus. EPO may provide therapeutic option for patients with

mood disorders who have impaired neuroplasticity and cognition.

Clinical trial registration: clinicaltrials.gov: NCT00916552.

Dr Kamilla Miskowiak 5

Introduction

Unipolar depression (UD) and bipolar disorder (BD) are chronic, recurrent and severe mental

disorders that cost an estimated €113 billion in Europe and $128 billion in the United States, of

which reduced workforce performance accounts for 60-80% of the expenditures (1;2). Cognitive

dysfunction is a core feature of mood disorders that often persists in remission and impedes

workforce performance (3;4). Although pharmacological treatments indirectly alleviate patients’

cognitive deficits by decreasing mood symptoms, no treatment so far has shown direct pro-

cognitive effects (5). Novel treatments that directly enhance cognitive dysfunction in mood

disorders should improve patients’ work capacity and reduce societal costs of these disorders.

Converging evidence from preclinical studies, neuroimaging studies and post-mortem studies of

patients with mood disorders suggest that cognitive dysfunction arises from disruption of

neuroplasticity and structural changes in brain regions including the hippocampus. Magnetic

resonance imaging (MRI) of UD and BD in depressed and remitted phases and post-mortem

immunohistochemical studies have revealed reduction in overall hippocampal volume (6;7) and

volumetric shrinkage in hippocampal subregions including the dentate gyrus (DG), cornu ammonis

(CA1–3) and subiculum (8-11). Preclinical studies suggest that the hippocampal volume reduction

is caused by dendritic retraction in the CA1-3 and DG, CA3 pyramidal cell death and suppression

of DG neurogenesis due to glucocorticoid overexposure (12-14). In keeping with this, post-mortem

examination of hippocampi of depressed individuals has revealed loss of dendritic branching,

dendritic spine complexity and glia (15;16), although no substantial neural atrophy or suppressed

DG neurogenesis have been observed (11;13). Together these findings support the idea that direct

upregulation of neuroplasticity might be a fruitful approach for the treatment of cognitive

dysfunction in mood disorders.

Dr Kamilla Miskowiak 6

Erythropoietin (EPO) has neurotrophic properties and is a promising treatment for cognitive

dysfunction in mood disorders. Endogenous EPO in the brain mediates neuroprotection and

neurodevelopment and plays a key role in cognitive functioning (17). Systemically administered

EPO penetrates the blood-brain barrier, has neuroprotective and neurotrophic actions, and produces

cognitive enhancement in animal models of neural injury, neurodegenerative conditions and

depression (17;18). These effects are mediated through activation of anti-apoptotic, anti-oxidant and

anti-inflammatory signaling in neurons, glial and cerebrovascular endothelial cells, and promotion

of dendritic sprouting, neurogenesis, hippocampal brain-derived neurotrophic factor (BDNF) and

long-term potentiation (17-19). Translational studies in schizophrenia and multiple sclerosis have

demonstrated that 12 weekly EPO treatments produces consistent and long-lasting enhancement of

cognitive functioning (20;21).

We have shown that single EPO versus saline administration improves neurocognitive measures of

memory and executive function and produces antidepressant-like effects in healthy and depressed

individuals independent of change in red blood cells or mood symptoms (for review see (22)).

These findings motivated us to conduct a randomized, placebo-controlled clinical trial investigating

the effects of 8 weekly doses of EPO on cognitive function and mood in TRD (23) and partially

remitted BD (24). EPO produced robust, mood-independent enhancement of verbal memory in

TRD (23) and of complex cognitive processing across attention, memory and executive function in

BD (24) as compared to saline. These cognitive effects of EPO were independent of changes in

mood symptoms and were sustained after red blood cell normalization, suggesting that they

reflected direct neurobiological actions.

The present prospective MRI investigation of 69 patients with pre- and post-treatment scans from

the above trial (23;24) aimed to assess the neuroanatomical underpinnings of the EPO-associated

improvement in cognitive function. MRI assessments in the EPO-schizophrenia trial (20) revealed

Dr Kamilla Miskowiak 7

that EPO prevented gray matter loss in regions typically affected in schizophrenia, including the

hippocampus and fronto-parietal regions (25). Given the evidence for hippocampal volume

reduction in both UD and BD, we hypothesized that EPO treatment would increase total or sub-

regional hippocampal volume, and that this increase would be associated with improved verbal

memory.

Dr Kamilla Miskowiak 8

Methods and materials

Study design and participants

This report is based on a previously reported double-blind, placebo-controlled, parallel-group study

(23;24). In brief, patients, 18-65 years of age, were recruited through Clinic for Affective Disorders,

Psychiatric Centre Copenhagen, and by advertisement on relevant websites, and screened with

Schedules for Clinical Assessment in Neuropsychiatry (SCAN) to confirm ICD-10 diagnosis.

Eligible patients had a diagnosis of major depression and fulfilled the criteria for TRD based on

assessment of their medical treatment history with the Treatment Response to Antidepressants

Questionnaire (TRAQ) (for details description see (23)) with current moderate depression

(Hamilton Depression Rating Scale 17-items (HDRS-17) score >17) or BD in partial remission

(HDRS-17 and Young Mania Rating Scale (YMRS) scores <14) with moderate to severe cognitive

difficulties according to the Cognitive and Physical Functioning Questionnaire (CPFQ) (score >4

on >2 domains). For extensive description of screening procedure, exclusion criteria and safety

precautions see (23;24). The study was approved by the local ethics committee, Danish Medicines

Agency, and Danish Data Agency, and was registered at clinicaltrials.gov (no. NCT00916552).

After complete description of the study to the participants, written informed consent was obtained.

Randomization and masking

Block randomization was performed with stratification for age (< or >35 years) and gender.

Outcome assessors were blinded to patients’ group assignment (EPO/saline) throughout the study

period and data analysis (for description of blinding procedures see (23;24)). The Good Clinical

Practice unit at Copenhagen University Hospital (www.gcp-enhed.dk/kbh) monitored that blinding

was maintained.

Dr Kamilla Miskowiak 9

Procedures

Patients were randomized to receive 8 weekly intravenous infusions of EPO (Eprex; 40,000 IU;

Janssen-Cilag) or saline (NaCl 0.9%) at the Clinic for Affective Disorders, Psychiatric Centre

Copenhagen. Cognitive function was investigated at weeks 1 (baseline), 9 (1 week post-treatment),

and 14 (6 weeks follow-up) with a neuropsychological test battery including verbal memory

measured with Rey Auditory Verbal Learning Test (RAVLT). To minimize learning effects on this

test we administered two alternate versions of the RAVLT at baseline and week 14 in a counter-

balanced fashion (24). MRI was performed at baseline and week 14 when hematological parameters

were expected to have normalized in the EPO group. Blood tests were taken at a weekly basis from

baseline to week 10 (2 weeks after treatment completion) and again in week 14. Mood symptoms

were assessed at weeks 1, 5, 9, and 14 with the Beck Depression Inventory 21-items (BDI-21),

HDRS-17, and for patient with BD also with the YMRS.

Data analysis

Behavioral analyses

The effect of EPO versus saline on verbal memory, reflected by RAVLT total recall, was examined

with ANCOVA adjusted for age, gender, diagnosis and change in depression symptoms. To

elucidate the functional implications of structural changes in the hippocampus, multiple regression

analysis was performed with verbal memory as the dependent variable and structural hippocampal

change, treatment group, age, gender, diagnosis and depression symptoms as predictive variables.

These statistical analyses were performed in Statistical Package for Social Sciences (SPSS) (version

19 by IBM).

Structural brain imaging

Dr Kamilla Miskowiak 10

We acquired a high-resolution 3-dimensional (3-D) T1-weighted MRI data set of the brain using a

magnetization-prepared rapid-gradient echo (MPRAGE) sequence with whole-brain coverage (echo

time 3.04 ms, repetition time 1550 ms, inversion time 800 ms, flip angle 9°, field of view 256 mm,

matrix 256 × 256, 1 × 1 × 1 mm3 voxels, 192 slices). We used a Magnetom Trio 3-T scanner with

an 8-channel head coil (Siemens, Erlangen, Germany). The FSL software package (version 5.0.5;

http://fsl.fmrib.ox.ac.uk/) was used for image analyses.

Segmentation of the hippocampus

The FMRIB's Integrated Registration and Segmentation Tool (FIRST) (http://fsl.fmrib.ox.ac.uk/)

was used for image segmentation and registration of the hippocampus. FIRST employs a Bayesian

probabilistic approach to segment subcortical structures using shape and appearance. In brief,

surface shape models were obtained from a database of manually segmented T1 weighted MR

images and shape and intensity were modeled jointly along the surface normals using a multivariate

Gaussian point distribution model. The model was fitted by maximizing the probability of shape

given intensity, thus incorporating prior information regarding intensity, shape and the relationship

between the two. The native output of FIRST is a surface mesh and in the conversion to voxel data

there is uncertainty as to whether the voxels that the surface passes through belong to the

structure. To correct for this, the partial volume (boundary) voxels were independently thresholded

based on the mean and variance calculated from the interior voxels (26). The total volume of the left

and right hippocampus was extracted and calculated at each time point and treatment-related

changes in hippocampal volume were assessed with analysis of covariance (ANCOVA) adjusting

for diagnosis, age, and gender in Statistical Package for the Social Sciences (SPSS, version 19 by

IBM).

Dr Kamilla Miskowiak 11

Analysis of regional shape changes in the hippocampal formation

Regional volume changes in hippocampal subregions over time were analyzed using FSL vertex

(shape) analysis. Differences in hippocampal shape over time as well as the direction of such local

shape changes were directly assessed in the meshes on a vertex-by-vertex basis. To this end, the

vertex locations from each subject at each time point were projected onto the surface normal of the

average shape provided by FIRST. The projections from different time points were then subtracted

(follow-up minus baseline) to create maps of hippocampus surface displacement for each subject

over time (with positive values indicating expansion). At the group level, these surface

displacement maps were then subjected to a voxelwise analysis in FSL with RANDOMISE, using a

general linear model with group and diagnosis as regressors and 5000 permutations. For those

hippocampal subfields where RANDOMISE revealed significant changes in surface displacement

between groups (at p < .05, FWE-corrected at a cluster level), mean displacements values were

extracted and examined in SPSS.

Voxel-based morphometry

In addition to the shape analysis of hippocampal subregions, we performed voxel based

morphometry with the FSL-VBM tool using an optimized VBM protocol (27). Structural images

from each time point were processed independently, and the within-subject changes were examined

subsequently in the statistical analysis. Pre-processing steps included extraction of the brain, gray

matter segmentation, and registration of the segmented brain maps onto the MNI 152 standard

space using non-linear registration. The resulting gray-matter segmentations were averaged and

flipped along the x-axis to create a left-right symmetric, study-specific gray matter template. All

native gray matter images were then non-linearly registered to this template and modulated for local

expansion (or contraction) due to the non-linear component of the spatial transformation.

Dr Kamilla Miskowiak 12

Modulated gray matter images from baseline and follow-up were then subtracted to quantify

changes over time. The images of the gray matter differences were smoothed with an isotropic

Gaussian kernel with a sigma of 3 mm. For left and right hippocampi, voxelwise general linear

model analysis was applied to test for time-dependent differences in hippocampal gray matter

density between groups using permutation-based non-parametric testing, correcting for multiple

comparisons across each hippocampus. The hippocampal regions were defined using the Harvard-

Oxford subcortical structural atlas. An exploratory whole-brain comparison of gray matter

difference images over time was also performed, using permutation-based non-parametric testing,

correcting for multiple comparisons across the whole brain. Statistical threshold was set at p < .05,

FWE-corrected at a cluster level.

Dr Kamilla Miskowiak 13

Results

Patient flow and characteristics

Table 1 and the CONSORT chart in the supplementary material display patient flow and

characteristics, respectively. Of the 84 patients were randomized to EPO (N = 42) or saline (N =

42), 1 patient withdrew at baseline and data collection was incomplete for 14 (for details see the

CONSORT chart in the supplementary material). 6 EPO-treated patients (3 TRD, 3 BD) were

discontinued from their infusions between weeks 5-7 due to increased thrombocyte counts (> 4x109/

L) but continued all assessments. Baseline and follow-up data was thus available and analyzed for

69 patients (EPO: N = 35 [16 TRD, 19 BD]; Saline: N = 34 [17 TRD, 17 BD]). Groups were well

matched on baseline characteristics (p > .22) (Table 1).

EPO induces regional changes in hippocampal shape

At baseline, left and right hippocampal surface showed no shape differences between EPO and

saline groups (p < .05, FWE-corrected at a cluster level). Shape analysis of left hippocampus

yielded a subfield-specific volume increase in the CA1-3 region extending into the subiculum in

EPO versus saline-treated patients (F = 16.50, FWE-corrected p = 0.0188) (see hippocampal

subfield in Figure 1.A). Post-hoc comparisons within each group revealed that this effect was driven

by a relative expansion of CA1-3 region/subiculum in EPO-treated patients (t = 3.64, df = 34, p =

.001) and a shrinkage in those given saline (t = -2.80, df = 33, p = .008) (see Figure 1.B). The sub-

regional hippocampal volume change was unrelated to changes in mood symptoms (BDI-21 scores)

in the EPO group and across all patients (EPO: r(34)=-.27, p = .12); whole cohort: r(68) = -.04, p =

.74).

There was no effect of EPO versus saline on change in right hippocampal shape at p < .05, FWE-

corrected at a cluster level. Exploratory analysis of right hippocampal shape with a lowered

Dr Kamilla Miskowiak 14

threshold of p < .10, FWE-corrected at a cluster level, also showed no trend towards an effect of

EPO versus saline.

No effect of EPO on total hippocampal volume or regional gray matter intensity

There were no baseline differences in total volume of the left or right hippocampi between groups

(p > .79; see Table 1) or any overall hippocampal volume changes over time (p > .34). ANCOVA

adjusted for age, gender and diagnosis revealed a general effect of gender, with larger hippocampal

volume in women (left: F(1,64) = 10.77, p = .002; right: F(1,64) = 14.52, p < .001) and a trend

towards smaller volume in BD versus TRD (left: F(1,64) = 3.57, p = .064; right: F(1,64) = 2.93, p =

.092). However, there was no effect of EPO versus saline on total volume change in the left or right

hippocampus (ANCOVA adjusted for age, gender and diagnosis: p > .13), also not when analyzing

the treatment effects within each patient group (TRD/ BD) separately (p > .33).

VBM analysis revealed no between-group differences in gray matter density in left or right

hippocampus at baseline and no effects of EPO versus saline on change in total hippocampal gray

matter density (p < .05, FWE-corrected at a cluster level and adjusted for diagnosis). Exploratory

whole-brain analysis also revealed no difference between groups in cortical gray matter density

change over time (p < .05, FWE-corrected at a cluster level and adjusted for diagnosis).

Increased left hippocampal subfield volume is associated with improved verbal memory

EPO improved RAVLT total recall from baseline to week 14 compared with saline independent of

changes in mood symptoms (ANCOVA adjusted for age, gender, diagnosis, and change in BDI

scores: F(1,61) = 4.49, p = .04) (see memory improvement with EPO versus saline in Figure 2.A).

Indeed, post-hoc Pearson correlations showed that verbal memory change was unrelated to changes

in mood symptoms in the EPO group (r(34) = .03, p = .88) and across all patients (r(68) = -.13, p =

Dr Kamilla Miskowiak 15

.28) (see the absence of correlation between change in memory and change in mood symptoms in

Figure 2.B). The effect of EPO on verbal memory also occurred in the absence of any general

learning effects across the whole patient cohort (p = .45).

To investigate the functional relevance of the EPO-induced hippocampal volume increase, we

performed a linear regression analysis on the entire data set with verbal memory improvement as

the dependent variable and left CA1-3/ subiculum hippocampal volume change, group, age, gender,

diagnosis and change in BDI scores as the predictive variables. This revealed a significant model

(F(6,66) = 2.69, p = .02), in which sub-regional hippocampal volume increase was the only

significant predictor (Beta = 3.72, p = .006), while changes in mood symptoms and the additional

variables had no impact on verbal memory change (p > 0.25) (see relation between hippocampal

volume increase and memory improvement in Figure 1.C).

No influence of EPO related changes in hematocrit

EPO increased the hematocrit from baseline to week 9 (23;24), but this effect had tapered off before

week 14 with no between-group difference at week 14 (t = -.71, df = 67, p = .48). Accordingly,

there were no differences in relative hematocrit change from baseline to week 14 between EPO and

saline groups (F(1,67) = 2.29, p = .14). EPO-treated patients also showed no correlation between

hematocrit increase and sub-regional hippocampal volume increase (r(34) = 0.01, p = .94) or

memory improvement (r(34) = -.09, p = .61).

Dr Kamilla Miskowiak 16

Discussion

This study is the first to prospectively elucidate the neuroanatomical underpinnings of the cognitive

improvement observed in EPO-treated patients with BD or TRD. The study demonstrated that EPO

treatment is associated with prevention of brain matter loss in a subfield of the left hippocampus

encompassing the CA1-3 and subiculum and improves verbal memory independent of changes in

mood. Across the whole cohort, verbal memory improvement was mediated by structural increase

in this left hippocampal subfield rather than by changes in mood symptoms. These effects of EPO

occurred in the absence of changes in total hippocampal volume or in hippocampal or whole-brain

gray matter volume.

In contrast with the EPO-treated patients, those given saline displayed discrete but significant left-

side volume reduction within the CA1-3 region and subiculum over a period of 3 months. This

observation points to disease-related processes in the CA1-3 and subiculum consistent with

evidence for hippocampal CA1-3 volume reduction in UD (10) and BD (9) as well as abnormal left

hippocampal subiculum shape in TRD (28). Conversely, chronic antidepressant treatment appears

to reverse the DG and CA1-3 volume reduction in UD (10) and lithium increases overall

hippocampal volume in BD (29). The ability of EPO to not only prevent shrinkage but also produce

direct volume increase in the left hippocampal CA1-3 and subiculum suggests that EPO can

mitigate disease-related hippocampal volume reduction similar to existing treatments of mood

disorders. In contrast, we found no effect of EPO on DG volume despite the ability of EPO to

enhance neurogenesis (17). A potential explanation is that all patients were on stabile antidepressant

or mood stabilizing treatment which could have obscured an EPO effect on DG volume reduction

(10). Future studies examining the effects of EPO versus saline effects in drug-free patients could

help resolve this question, although such investigation may be difficult to conduct.

Dr Kamilla Miskowiak 17

Several neurobiological mechanisms may underlie the CA1-3 and subiculum volume increase in

EPO versus saline-treated patients. Although EPO increased hematocrit during the active treatment

phase, red blood cells were normalized at the time of the follow-up scan (6 weeks after treatment

completion). This suggests that the EPO-associated hippocampal volume increase was unrelated to

unspecific changes in cerebral blood-flow. Hippocampal volume increase after long-term

antidepressant or mood-stabilizing treatment (10;29) may be mediated by reversal of dendritic spine

and glia loss and dendritic retraction in the DG and CA (30-32) and increased hippocampal nerve

growth factor (NGF), BDNF (33;34) and neurogenesis (35), respectively. The EPO-associated

increase in CA1-3 volume could reflect similar neurotrophic mechanisms given the ability of EPO

to enhance hippocampal BDNF, neurite growth and spine density (17;36). A high density of

glucocorticoid binding sites makes the subiculum particular vulnerability to stress (37) and could

explain the abnormalities in this hippocampal subfield in UD (11;28). The reversal of subiculum

shrinkage in EPO-treated patients may thus reflect neuroprotection from glucocorticoid-induced

atrophy. Another putative mechanism of EPO underlying the hippocampal effect is inhibition of

glycogen synthase kinase 3 beta (GSK3β) (38), a key activator of cell death that is involved in

mood disorders, hippocampal volume and neuroplasticity (39). Taken together, multiple

neurobiological actions may either alone or in orchestra contribute to the effects of EPO on

hippocampus and memory performance.

Interestingly, the EPO-associated sub-regional volume increase was confined to the left

hippocampus. The neurobiological mechanism for this unilateral effect is unclear. Nevertheless, the

finding is consistent with selective left hippocampal volume increase in UD after 3 years

antidepressant drug treatment (40) and in schizophrenia after 12 weeks EPO-treatment (25). In

contrast with the EPO-schizophrenia study, we observed no effects of EPO on total gray matter

volume in the hippocampus or across cortical regions. This discrepancy may be explained by

Dr Kamilla Miskowiak 18

preferential neuroprotective effects of EPO in areas marked by neural degeneration, which is more

wide-spread in schizophrenia, or – alternatively - by the 4 weeks shorter treatment in the present

study.

The study had a large sample size (N = 69) in comparison with previous MRI investigations of EPO

treatment (25) or of subfield hippocampal volume in UD (10) and BD (9). In comparison, MRI

studies of mood disorders often involve 20-40 patients (and an equal number of healthy controls) in

cross-sectional designs (9;10), and 10-30 patients in longitudinal designs (29;40). In addition, the

longitudinal study design rendered it possible to capture intra-individual structural changes in

response to EPO versus saline treatment. Finally, our assessment of the verbal memory in parallel

with the structural hippocampal measures enabled insight into the functional implications of the

EPO-associated increase in subfield hippocampal volume. It is a potential limitation that our cohort

included both patients with TRD and BD, since these mood disorders may involve differential

although partially overlapping pathogenic processes. However, disruption of neuroplasticity,

hippocampal volume reduction and cognitive deficits are common biological targets in these

disorders and the ability of EPO to enhance cognition and neuroprotection has been observed across

a range of different brain disorders (20;21;23;24). Nevertheless, there was an equal distribution of

patients with TRD and BD in the EPO and saline groups, and analyses were adjusted for patients’

diagnosis. We used three complementary methods which enabled us to capture different aspects of

hippocampal volume changes triggered by EPO treatment: FSL vertex analysis and FIRST were

used to investigate our hypothesis that EPO would increase sub-regional or total hippocampal

volume, respectively, while VBM was used to elucidate the neurobiological mechanisms of such

changes. We did not adjust the statistical threshold in the vertex analysis for the two hypotheses

regarding EPO-associated hippocampal volume change. The rationale for this was that the vertex

analysis of subfield hippocampal volume change is itself statistically highly conservative involving

Dr Kamilla Miskowiak 19

FWE-correction, which rendered the existence of false positives unlikely. However, even if we had

corrected the statistical threshold for the two comparisons, the hippocampal effect on sub-regional

hippocampal volume would have remained significant (p = 0.0188 < 0.025). Despite the EPO-

associated volume increase in the left hippocampal CA1-3 and subiculum, we observed no

corresponding change in total hippocampal volume or overall hippocampal gray matter volume.

Nevertheless, we consider the finding significant for the following reasons: First, the EPO-

associated surface expansion in the left hippocampal CA1-3/subiculum is opposite to the effects of

stress and consistent with the neurobiological effects of EPO in animal models and the left-

lateralized hippocampal effects in schizophrenia. Second, the vertex analysis estimates changes in

sub-regional hippocampal volume by the relative displacements in hippocampal surface, which is a

more sensitive measure of structural hippocampal change than total hippocampal volume or total

gray matter volume (41). In particular, direct comparisons of these techniques revealed that

localized disease-related structural change in early Parkinson’s disease was undetectable with VBM

and overall regional volume, but could be identified with vertex shape analysis. (41). Indeed, FIRST

is limited by ambiguity in the demarcation of the hippocampus versus surrounding structures (42),

while VBM is limited by ambiguous interpretability with regards to gray matter microstructure, less

precise image registration, and susceptibility to confounding effects of large anatomical differences

in other parts of the brain (43). Notably, we cannot therefore rule out that the EPO-associated

increase in sub-regional hippocampal volume could be related to increase in sub-regional gray

matter. Finally, the three methods reflect different structural measures (subfield surface

displacements, total volume and total gray matter volume, respectively). Therefore, change in one

of these measures would not necessarily be expected to be accompanied by changes in the other

measures.

Dr Kamilla Miskowiak 20

In conclusion, the study shows that EPO treatment of patients with mood disorders prevents brain

matter loss in the left hippocampal CA1-3 and subiculum and improves verbal memory.

Hippocampal subfield volume increase was associated with improvement of verbal memory

independent of changes in mood. The findings are consistent with the ability of EPO to enhance

neuroplasticity and memory in animal studies and highlight EPO as a therapeutic option for patients

with mood disorders who have impaired neuroplasticity and cognition.

Dr Kamilla Miskowiak 21

Acknowledgements

The study was funded by the Danish Ministry of Science, Innovation and Higher Education, Novo

Nordisk Foundation, Beckett Fonden, and Savværksejer Juhl’s Mindefond. The sponsors had no

role in the planning or conduct of the study or in the interpretation of the results.

We would like to thank A. John Rush MD, Professor Emeritus, National University of Singapore,

and Martin B. Jørgensen, Professor, MD, DMsc, for their valuable feedback on the manuscript,

David Cole, PhD, for statistical advice regarding the FSL analyses, and Nikolai V. Malykhin, MD,

PhD, for help with determining the hippocampal subfields that were increased by EPO versus

saline. We also thank the study nurses, Susanne Sander and Hanne Nikolajsen for their work with

logistical planning and running the trial. Professor Siebner was supported by a Grant of Excellence

on the control of actions “ContAct” from Lundbeck Foundation (R59 A5399). The Simon Spies

Foundation is acknowledged for donation of the Siemens Trio Scanner. The Lundbeck Foundation

has provided half of Dr Miskowiak’s post-doctorate salary at the Psychiatric Center Copenhagen

(2012–2015) for her to do full-time research in this period.

The findings in this paper were presented orally at the American Society for Clinical

Psychopharmacology (ASCP) meeting, Hollywood, Florida, 16-19 June 2014.

Financial disclosures

KWM reports having received consultancy fees from Lundbeck. MV discloses consultancy fees

from Eli Lilly, Lundbeck; Servier and Astra Zeneca. HE resports has submitted / holds user patents

for EPO in stroke, schizophrenia, and MS. OBP is a member of the board of directors of the Elsass

Foundation. LVK reports having been a consultant for Lundbeck, AstraZeneca and Servier within

the last 3 years. HRS discloses honoraria as reviewing editor for Neuroimage, as speaker for Biogen

Dr Kamilla Miskowiak 22

Idec Denmark A/S, and scientific advisor for Lundbeck within the past 3 years. All other authors

report no biomedical financial interests or potential conflicts of interest.

Dr Kamilla Miskowiak 23

References

(1) Olesen J, Gustavsson A, Svensson M, Wittchen HU, Jonsson B (2012): The economic cost

of brain disorders in Europe. Eur J Neurol 19(1): 155-162.

(2) Wyatt RJ, Henter I (1995): An economic evaluation of manic-depressive illness—1991. Soc

Psychiatry Psychiatr Epidemiol 30(5): 213-219.

(3) Jaeger J, Berns S, Uzelac S, Davis-Conway S. Neurocognitive deficits and disability in

major depressive disorder. Psychiatry Res 2006 Nov 29;145(1):39-48.

(4) Mur M, Portella MJ, Martinez-Aran A, Pifarre J, Vieta E (2009): Influence of clinical and

neuropsychological variables on the psychosocial and occupational outcome of remitted

bipolar patients. Psychopathology 42(3): 148-156.

(5) Dias VV, Balanza-Martinez V, Soeiro-de-Souza MG, Moreno RA, Figueira ML, Machado-

Vieira R, et al. (2012): Pharmacological approaches in bipolar disorders and the impact on

cognition: a critical overview. Acta Psychiatr Scand 126(5): 315-331.

(6) Canales-Rodriguez EJ, Pomarol-Clotet E, Radua J, Sarro S, Alonso-Lana S, del Mar BC, et

al. (2014): Structural Abnormalities in Bipolar Euthymia: A Multicontrast Molecular

Diffusion Imaging Study. Biol Psychiatry 76(3): 239-248.

(7) McKinnon MC, Yucel K, Nazarov A, MacQueen GM (2009): A meta-analysis examining

clinical predictors of hippocampal volume in patients with major depressive disorder. J

Psychiatry Neurosci 34(1): 41-54.

(8) Cole J, Toga AW, Hojatkashani C, Thompson P, Costafreda SG, Cleare AJ, et al. (2010):

Subregional hippocampal deformations in major depressive disorder. J Affect Disord 126(1-

2): 272-277.

Dr Kamilla Miskowiak 24

(9) Elvsashagen T, Westlye LT, Boen E, Hol PK, Andersson S, Andreassen OA, et al. (2013):

Evidence for reduced dentate gyrus and fimbria volume in bipolar II disorder. Bipolar

Disord 15(2): 167-176.

(10) Huang Y, Coupland NJ, Lebel RM, Carter R, Seres P, Wilman AH, et al. (2013): Structural

changes in hippocampal subfields in major depressive disorder: a high-field magnetic

resonance imaging study. Biol Psychiatry;74(1):62-8.

(11) Lucassen PJ, Muller MB, Holsboer F, Bauer J, Holtrop A, Wouda J, et al. (2001):

Hippocampal apoptosis in major depression is a minor event and absent from subareas at

risk for glucocorticoid overexposure. Am J Pathol 158(2): 453-468.

(12) Alfarez DN, Karst H, Velzing EH, Joels M, Krugers HJ (2008): Opposite effects of

glucocorticoid receptor activation on hippocampal CA1 dendritic complexity in chronically

stressed and handled animals. Hippocampus 18(1): 20-28.

(13) Czeh B, Lucassen PJ (2007): What causes the hippocampal volume decrease in depression?

Are neurogenesis, glial changes and apoptosis implicated? Eur Arch Psychiatry Clin

Neurosci 257(5): 250-260.

(14) Hageman I, Nielsen M, Wortwein G, Diemer NH, Jorgensen MB (2008): Electroconvulsive

stimulations prevent stress-induced morphological changes in the hippocampus. Stress

11(4): 282-289.

(15) Cobb JA, Simpson J, Mahajan GJ, Overholser JC, Jurjus GJ, Dieter L, et al. (2013):

Hippocampal volume and total cell numbers in major depressive disorder. J Psychiatr Res

47(3): 299-306.

(16) Stockmeier CA, Rajkowska G (2004): Cellular abnormalities in depression: evidence from

postmortem brain tissue. Dialogues Clin Neurosci 6(2): 185-197.

Dr Kamilla Miskowiak 25

(17) Sargin D, Friedrichs H, El-Kordi A, Ehrenreich H (2010): Erythropoietin as neuroprotective

and neuroregenerative treatment strategy: comprehensive overview of 12 years of preclinical

and clinical research. Best Pract Res Clin Anaesthesiol 24(4): 573-594.

(18) Girgenti MJ, Hunsberger J, Duman CH, Sathyanesan M, Terwilliger R, Newton SS (2009):

Erythropoietin induction by electroconvulsive seizure, gene regulation, and antidepressant-

like behavioral effects. Biol Psychiatry 66(3): 267-274.

(19) Adamcio B, Sargin D, Stradomska A, Medrihan L, Gertler C, Theis F, et al. (2008):

Erythropoietin enhances hippocampal long-term potentiation and memory. BMC Biol: 37.

(20) Ehrenreich H, Hinze-Selch D, Stawicki S, Aust C, Knolle-Veentjer S, Wilms S, et al.

(2007): Improvement of cognitive functions in chronic schizophrenic patients by

recombinant human erythropoietin. Mol Psychiatry 12(2): 206-220.

(21) Ehrenreich H, Fischer B, Norra C, Schellenberger F, Stender N, Stiefel M, et al. (2007):

Exploring recombinant human erythropoietin in chronic progressive multiple sclerosis.

Brain 130(10): 2577-2588.

(22) Miskowiak KW, Vinberg M, Harmer CJ, Ehrenreich H, Kessing LV (2012): Erythropoietin:

a candidate treatment for mood symptoms and memory dysfunction in depression.

Psychopharmacology (Berl) 219(3): 687-698.

(23) Miskowiak KW, Vinberg M, Christensen EM, Bukh JD, Harmer CJ, Ehrenreich H, et al.

(2014): Recombinant Human Erythropoietin for Treating Treatment-Resistant Depression:

A Double-Blind, Randomized, Placebo-Controlled Phase 2 Trial.

Neuropsychopharmacology 39(6): 1399-1408.

(24) Miskowiak K, Ehrenreich H, Christensen EM, Kessing LV, Vinberg M. (2014):

Recombinant human erythropoietin to target cognitive dysfunction in bipolar disorder: A

double-blind, randomized, placebo-controlled phase 2 trial. J Clin Psychiatry (in press)

Dr Kamilla Miskowiak 26

(25) Wustenberg T, Begemann M, Bartels C, Gefeller O, Stawicki S, Hinze-Selch D, et al.

(2011): Recombinant human erythropoietin delays loss of gray matter in chronic

schizophrenia. Mol Psychiatry 16(1): 26-36.

(26) Patenaude B, Smith SM, Kennedy DN, Jenkinson M (2011): A Bayesian model of shape

and appearance for subcortical brain segmentation. Neuroimage 56(3): 907-922.

(27) Smith SM, Jenkinson M, Woolrich MW, Beckmann CF, Behrens TE, Johansen-Berg H, et

al. (2004): Advances in functional and structural MR image analysis and implementation as

FSL. Neuroimage 23(1): S208-S219.

(28) Tae WS, Kim SS, Lee KU, Nam EC, Choi JW, Park JI (2011): Hippocampal shape

deformation in female patients with unremitting major depressive disorder. AJNR Am J

Neuroradiol 32(4): 671-676.

(29) Yucel K, McKinnon MC, Taylor VH, Macdonald K, Alda M, Young LT, et al. (2007):

Bilateral hippocampal volume increases after long-term lithium treatment in patients with

bipolar disorder: a longitudinal MRI study. Psychopharmacology (Berl) 195(3): 357-367.

(30) Bessa JM, Ferreira D, Melo I, Marques F, Cerqueira JJ, Palha JA, et al. (2009): The mood-

improving actions of antidepressants do not depend on neurogenesis but are associated with

neuronal remodeling. Mol Psychiatry 14(8): 764-773.

(31) Hajszan T, Dow A, Warner-Schmidt JL, Szigeti-Buck K, Sallam NL, Parducz A, et al.

(2009): Remodeling of hippocampal spine synapses in the rat learned helplessness model of

depression. Biol Psychiatry 65(5): 392-400.

(32) Magarinos AM, Deslandes A, McEwen BS (1999): Effects of antidepressants and

benzodiazepine treatments on the dendritic structure of CA3 pyramidal neurons after

chronic stress. Eur J Pharmacol 371(2-3): 113-122.

Dr Kamilla Miskowiak 27

(33) Frey BN, Andreazza AC, Rosa AR, Martins MR, Valvassori SS, Reus GZ, et al. (2006):

Lithium increases nerve growth factor levels in the rat hippocampus in an animal model of

mania. Behav Pharmacol 17(4): 311-318.

(34) Fukumoto T, Morinobu S, Okamoto Y, Kagaya A, Yamawaki S. (2001): Chronic lithium

treatment increases the expression of brain-derived neurotrophic factor in the rat brain.

Psychopharmacology (Berl) 158(1): 100-106.

(35) Chen G, Rajkowska G, Du F, Seraji-Bozorgzad N, Manji HK (2000): Enhancement of

hippocampal neurogenesis by lithium. J Neurochem 75(4): 1729-1734.

(36) Oh DH, Lee IY, Choi M, Kim SH, Son H (2012): Comparison of Neurite Outgrowth

Induced by Erythropoietin (EPO) and Carbamylated Erythropoietin (CEPO) in Hippocampal

Neural Progenitor Cells. Korean J Physiol Pharmacol 16(4): 281-285.

(37) Sarrieau A, Dussaillant M, Agid F, Philibert D, Agid Y, Rostene W (1986):

Autoradiographic localization of glucocorticosteroid and progesterone binding sites in the

human post-mortem brain. J Steroid Biochem 25(5B): 717-721.

(38) Ge XH, Zhu GJ, Geng DQ, Zhang ZJ, Liu CF (2012): Erythropoietin attenuates 6-

hydroxydopamine-induced apoptosis via glycogen synthase kinase 3beta-mediated

mitochondrial translocation of Bax in PC12 cells. Neurol Sci 33(6): 1249-1256.

(39) Inkster B, Nichols TE, Saemann PG, Auer DP, Holsboer F, Muglia P, et al. (2009):

Association of GSK3beta polymorphisms with brain structural changes in major depressive

disorder. Arch Gen Psychiatry 66(7): 721-728.

(40) Frodl T, Jager M, Smajstrlova I, Born C, Bottlender R, Palladino T, et al. (2008): Effect of

hippocampal and amygdala volumes on clinical outcomes in major depression: a 3-year

prospective magnetic resonance imaging study. J Psychiatry Neurosci 33(5): 423-430.

Dr Kamilla Miskowiak 28

(41) Menke RA, Szewczyk-Krolikowski K, Jbabdi S, Jenkinson M, Talbot K, Mackay CE, et al.

(2014): Comprehensive morphometry of subcortical grey matter structures in early-stage

Parkinson's disease. Hum Brain Mapp 35(4):1681-1690.

(42) Morey RA, Petty CM, Xu Y, Hayes JP, Wagner HR 2nd, Lewis DV, et al. (2009): A

comparison of automated segmentation and manual tracing for quantifying hippocampal and

amygdala volumes. Neuroimage 45(3):855-866.

(43) Mechelli A, Price CJ, Friston KJ, Ashburner J (2005): Voxel-Based Morphometry of the Human Brain:

Methods and Applications. Current Medical Imaging Reviews 1 (1): 1-9

Dr Kamilla Miskowiak 29

Table 1. Patient characteristics.

EPO (N = 35)

Saline (N = 34)

p-value

Diagnosis (no. BD/TRD) 19/16 17/17

Age, Mean (SD) 40 (10) 43 (12) 0.23

Gender (no. women) 24 23

Education, Mean (SD) 15 (4) 15 (3) 0.55

BDI baseline, Mean (SD) 26 (12) 24 (11) 0.40

HDRS baseline, Mean (SD) 14 (7) 14 (7) 0.64

YMS baseline, Mean (SD) 3 (3) 2 (2) 0.45

No. prior depressions, Mean (SD) 6 (5) 5 (4) 0.40

No. prior (hypo)manias, Mean (SD) 5 (6) 4 (4) 0.79

Bipolar subtype (no. of type II) 11 9

BMI, Mean (SD) 24 (3) 25 (3) 0.22

Total hippocampal volume at baseline

Left 5072 (483) 5036 (622) 0.79

Right 5205 (546) 5182 (518) 0.86

Dr Kamilla Miskowiak 30

Figure legend 1.a EPO reduces brain matter loss in a subfield of the left hippocampus corresponding to

CA1-3 and subiculum, as reflected by expansion in mean vertex location within this subfield in EPO versus

saline treated patients. In particular, Y = 100 and Y = 97 (hippocampus body) correspond to the subiculum

and CA2-3; Y = 103, Y = 105, and Y = 107 (hippocampus head) correspond to the subiculum and CA1-3;

and Y=94 (hippocampus tail) may be both CA and DG (Malykhin, personal communication, March 2014).

1.b. Mean hippocampal surface displacement in millimeters (reflected by change in mean vertex location) in

the left hippocampal CA1-3 and subiculum revealed expansion in the EPO-treated patients (p = .001) and

shrinkage in those given saline patients from baseline to week 14 (p = .008). 1.c. Linear relation between

subfield hippocampal volume change and change in verbal memory; there was a highly significant positive

correlation across all patients (r(68) = .40, p = .001) as well as in the EPO group (r(34) = .46, p = .005).

Figure legend 2.a. Verbal memory improvement as reflected by change in RAVLT total scores in response

to EPO (N = 35) versus saline (N = 34). The dotted line denotes the mean RAVLT total recall score of

healthy, age-matched (40-49 years) individuals of average intelligence. Mean and s.e.m are presented. 2.b.

Linear relation between change in mood symptoms (BDI-21 scores) and verbal memory (RAVLT total

score) change in the EPO group (N = 35); there was no correlation in the EPO group (r(34) = 0.03, p = .88)

or across all patients (r(68) = -.13, p = .28).

Y=94 Y=97

Y=100 Y=103

Y=105 Y=107

TOP BOTTOM

CA1-3 Subiculum

LEFT HIPPOCAMPUS A

B C

-20

-10

0

10

20

30

-0.5 -0.3 -0.1 0.1 0.3

-0.1

-0.05

0

0.05

0.1

Saline EPO

∆ Subfield hippocampal volume

∆ Memory ∆ Subfield hippocampal volume

-20

-15

-10

-5

0

5

10

15

20

25

-30 -20 -10 0 10 20

∆ Memory

∆ Mood symptoms

B A

40

42

44

46

48

50

52

54

Week 1 Week 14

EPO

SalineMem

ory

(R

AV

LT t

ota

l rec

all)