Psychopharmacology

, Volume 207, Issue 1, pp 133–142

Effects of erythropoietin on emotional processing biases in patients with major depression: an exploratory fMRI study

Authors

    • Department of PsychiatryUniversity Hospital of Copenhagen, Rigshospitalet
  • Elisa Favaron
    • Department of PsychiatryUniversity of Oxford
  • Sepehr Hafizi
    • Department of PsychiatryUniversity of Oxford
  • Becky Inkster
    • Faculty of MedicineImperial College London
  • Guy M. Goodwin
    • Department of PsychiatryUniversity of Oxford
  • Philip J. Cowen
    • Department of PsychiatryUniversity of Oxford
  • Catherine J. Harmer
    • Department of PsychiatryUniversity of Oxford
Original Investigation

DOI: 10.1007/s00213-009-1641-1

Cite this article as:
Miskowiak, K.W., Favaron, E., Hafizi, S. et al. Psychopharmacology (2009) 207: 133. doi:10.1007/s00213-009-1641-1

Abstract

Introduction

Erythropoietin (Epo) has neurotrophic effects and may be a novel therapeutic agent in the treatment of depression. We have found antidepressant-like effects of Epo on emotional processing and mood in healthy volunteers.

Objective

The current study aimed to explore the effects of Epo on the neural processing of emotional information in depressed patients.

Materials and methods

Seventeen patients with acute major depressive disorder were randomised to receive Epo (40,000 IU) or saline iv in a double-blind, parallel-group design. On day 3, we assessed neural responses to positive, negative and neutral pictures during fMRI followed by picture recall after the scan. Mood and blood parameters were assessed at baseline and on day 3.

Results

Epo reduced neural response to negative vs. positive pictures 3 days post-administration in a network of areas including the hippocampus, ventromedial prefrontal and parietal cortex. After the scan, Epo-treated patients showed improved memory compared with those that were given placebo. The effects occurred in the absence of changes in mood or haematological parameters, suggesting that they originated from direct neurobiological actions of Epo.

Conclusions

These findings are similar to the effects of conventional antidepressants and opposite to the negative biases in depression. The central effects of Epo therefore deserve further investigation as a potential antidepressant mechanism.

Keywords

AntidepressantHumanEmotionDepressionNeuroimaging

Introduction

Depression is the most common psychiatric disorder worldwide and has severe implications in terms of quality of life and economic burden associated with loss of work and health-care costs (Murray and Lopez 1997). The World Health Organization has therefore identified depression as an urgent health priority with the pressing need for more effective treatment options. Current antidepressant drug strategies, which require several weeks of treatment before efficacy can be confirmed, have only partial or no success for a significant proportion of patients and fail to target the neurocognitive impairments associated with the disorder even after full clinical recovery (for review, see Marvel and Paradiso 2004).

Compelling evidence points to a central role for neuroplasticity in the pathophysiology and long-term antidepressant drug treatment of depression (for review, see Berton and Nestler 2006). Compounds that directly increase neural plasticity and resilience may represent novel targets for treatment development. Although best known for its role in regulation of red blood cells, erythropoietin (Epo) is also produced in the brain by neurons and astrocytes and exerts vital neurobiological actions (for review, see Brines and Cerami 2005). Systemically administered Epo crosses the blood–brain barrier and has neuroprotective and neurotrophic effects in animal models of acute brain damage and chronic neurodegenerative conditions (Brines et al. 2000). In humans, long-term Epo administration improves neuropsychological function in patients with schizophrenia (Ehrenreich et al. 2007a) and multiple sclerosis (Ehrenreich et al. 2007b). Multiple mechanisms may mediate these actions, including up-regulation of brain-derived neurotrophic factor (BDNF; Viviani et al. 2005), hippocampal plasticity (Adamcio et al. 2008) and neurogenesis (Chen et al. 2007). Such actions in humans would make Epo an intriguing candidate for treatment of psychiatric disorders marked by neural dysfunction such as depression.

Negative biases in the processing and memory of emotional information are directly related to the severity and persistence of depression (Bradley et al. 1995). According to a recent hypothesis, antidepressant drugs may exert their clinical effects at a systems level through reversal of these negative biases (for review, see Harmer 2008). This reversal appears to occur prior to change in mood and symptoms. Thus, changes in automatic biases may provide a platform for cognitive restructuring and learning, thereby contributing to relatively slow improvement of mood and symptoms. These effects are seen across antidepressant drug classes early on in the treatment of healthy volunteers and patients (Harmer 2008). In contrast, compounds without reliable antidepressant properties such as benzodiazepines and NK1-receptor antagonists have limited or no effects on the same affective biases (Harmer 2008; Murphy et al. 2008). Consistent with evidence that neurotrophins such as Epo may be candidate treatments for depression, we have also previously demonstrated that single administration of Epo modulates neural and cognitive response to emotional information in healthy volunteers 3 and 7 days post-administration (Miskowiak et al. 2007a, 2008), similar to effects seen with conventional antidepressants (Harmer 2008).

Functional neuroimaging methods allow early biomarkers of illness and treatment effects to be examined. Functional magnetic resonance imaging (fMRI) studies report hyper-activation to negative vs. neutral pictures from the International Affective Picture System (IAPS; Lang et al. 1997) within the medial and dorsolateral prefrontal cortex (mPFC and dlPFC, respectively), amygdala (Wagner et al. 2004), hippocampus (Hamilton and Gotlib 2008) and occipital cortex (Davidson 2003) in depression. Over-recruitment of the hippocampus and amygdala during encoding of negatively valenced IAPS pictures was associated with greater memory for negative pictures in depressed vs. healthy individuals after one week later (Hamilton and Gotlib 2008). Depressed patients have also been found to display hypo-activation to positively valenced IAPS pictures in a broad network including prefrontal, temporal, parietal and limbic regions (Schaefer et al. 2006); these neural effects were reversed after successful antidepressant treatment.

The present study aimed to investigate whether the neurocognitive effects of Epo in healthy volunteers translate into a patient population. In particular, the study aimed to explore the effects of Epo on neural responses to affect-laden pictures in acutely depressed patients. Given the previously mentioned effects of depression and their resolution with antidepressant medication, we hypothesized that Epo would reduce neural response during encoding of negative compared with positive IAPS pictures in the m-dlPFC, occipito-parietal cortex and limbic regions. If the antidepressant-like effects of Epo in healthy volunteers translate into a patient population, this would provide support for a clinical trial with repeated administration of Epo.

Subjects and methods

Subjects

The study was approved by the Oxfordshire Research Ethics Committee. Patients with acute major depression were recruited through outpatient clinics at the Warneford Hospital, Oxford. Patients were screened through a medical examination and psychiatric interview using the Structured Clinical Interview for DSM-Clinical Version to confirm diagnosis and the Hamilton Rating Scale for Depression (HRSD; Hamilton 1960) to determine illness severity. All patients but two were on antidepressant medication (for details, see Table 1). No change in medication was made in the 2 weeks prior to or during the study. Exclusion criteria were: bipolar disorder, comorbid schizophrenia or substance misuse, significant risk of suicide, diabetes, epilepsy, hypertension, coronary disease and thrombosis, previous exposure to erythropoietin, pregnancy, or medication which could increase the risk of adverse effects of Epo, including contraceptives. In addition, fMRI required the exclusion of patients with a cardiac pacemaker, any mechanical implants, claustrophobia or who were left-handed. Patients' medical family history was assessed through a structured medical interview to exclude anyone with a first-degree family history of blood clotting or seizure disorders. A baseline blood test was taken to check that hematocrit, haemoglobin (Hb), renal function, liver function and ferritin levels were normal, and a pregnancy test was performed on female patients. After complete description of the study to the subjects, written informed consent was obtained.
Table 1

Demographic information for each group. Mean values with SD in brackets

 

Epo (n = 9)

Placebo (n = 8)

Age

34.6 (8.5)

34.3 (12.3)

Years of education

17.7 (4.7)

16.0 (3.0)

Antidepressant medication

7

5

 SSRI (citalopram, sertraline, fluoxetine)

4

3

 Dual action (mirtazapine, venlafaxine)

2

1

 MAOI (tranylcypromine, moclobemide)

1

2

 Antipsychotics (olanzapine, quetiapine)

2

1

 Lithium

0

1

 Benzodiazepines

3

1

 Thyroxin

0

1

 No medication

1

1

Mood

 HRSD day 0

17.7 (4.6)

19.1 (4.6)

 HRSD day 3

15.1 (6.0)

17.5 (5.6)

 CAS day 0

13.0 (4.7)

10.9 (4.8)

 CAS day 3

11.2 (4.9)

10.1 (5.2)

 BDI day 0

27.6 (13.0)

30.5 (9.0)

 BDI day 3

22.9 (9.5)

27.3 (9.9)

 HADS day 0

26.1 (8.0)

27.6 (8.5)

 HADS day 3

21.1 (5.0)

22.9 (10.7)

Blood parameters

 Hb day 0

14.34 (1.44)

14.26 (1.08)

 Hb day 3

14.16 (1.29)

14.08 (0.81)

 Haematocrit day 0

0.426 (0.040)

0.426 (0.031)

 Haematocrit day 3

0.423 (0.037)

0.422 (0.023)

 RCC day 0

4.84 (0.42)

4.81 (0.31)

 RCC day 3

4.77 (0.43)

4.75 (0.28)

Experimental design

Seventeen right-handed patients were randomly allocated to receive an intravenous injection of Epo (Eprex, Janssen-Cilag, High Wycombe, UK; 40,000 IU/ml) (n = 9; six males and three females) or placebo (n = 8; five males and three females) in a double-blind between-groups design. The groups were comparable in terms of age (p > 0.95), years of education (p > 0.40) and baseline mood measured with the HRSD, Clinical Anxiety Scale (CAS; Snaith et al. 1982), Hospital Anxiety and Depression Scale (HADS; Snaith and Zigmond 1986) and Beck's Depression Inventory (BDI; Beck et al. 1961) immediately before Epo/saline administration (day 0; all p values > 0.38; see Table 1). Most patients were taking an antidepressant either alone or with an accepted add-on treatment, i.e. low-dose antipsychotic, lithium or thyroxin, and this was comparable between groups (see Table 1). Following Epo/saline administration, blood pressure, wellbeing and subjective state was monitored for 2 h. Daily ratings of mood and subjective state were obtained on day −1 and for 3 days post-administration. On day 3, we assessed neuronal response to affect-laden pictures during fMRI followed by picture recall. Blood pressure was measured immediately after scanning and a follow-up blood test was taken to check for any changes in red cell mass in Epo vs. placebo-treated patients (for visual representation of the experimental design, see Fig. 1).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1641-1/MediaObjects/213_2009_1641_Fig1_HTML.gif
Fig. 1

Experimental design. Patients with acute major depression (mean HRSD score ± SD, 18.4 ± 4.6) were screened with a medical examination, psychiatric interview and a blood test. No change in medication was made in the 2 weeks prior to or during the study. Patients were randomised to receive Epo (40.000 IU) or saline in a double-blind, parallel-group design and subjected to fMRI and a blood test on day 3

Functional magnetic resonance imaging paradigm

Visual stimuli were projected from a computer using e-prime software (version 1.0; Psychology Software Tools Inc., Pittsburgh, PA) onto an opaque screen at the foot of the scanner bore. Patients viewed the screen through angled mirrors and responded by pressing the keys of a response pad with their thumbs. Thereby, accuracy and reaction time could be investigated.

Pictures of affect-laden scenes were taken from the IAPS. Sixty pictures (20 positive, 20 negative and 20 neutral) were presented on the screen in a random order interspersed with 20 fixation crosses in an event-related design. Stimulus duration was 2,000 ms, and the inter-stimulus interval varied between 5,000 and 8,000 ms with a mean of 6,500 ms, leading to a total task time of 5,200 ms (8 min and 40 s). Half of the pictures in each category and half of the fixation crosses were presented on the left side and half on the right side of the screen in a counter-balanced fashion. Subjects were instructed to indicate the side of the screen on which the stimulus appeared by pressing the left or right key on a response pad as quickly as possible. The lateralised presentations served to keep patients alert and engaged with the task.

To explore whether drug-related effects on processing of affect-laden pictures were confounded by global effects of Epo on baseline blood flow or neuronal coupling, neural activation was assessed with a visual stimulation paradigm: a flashing checkerboard (8 Hz) was presented in blocks of 15 s alternating with 15 s of a fixation cross for a total of 20 cycles (5 min). Subjects were instructed to lie with their eyes open during this time.

After the scan, patients were asked to recall as many IAPS pictures as possible. The number of correctly recalled positive, neutral and negative pictures as well as false recollections was recorded.

Functional magnetic resonance imaging data acquisition

Imaging data were collected using a Siemens Sonata scanner operating at 1.5 T, at the Oxford Centre for Clinical Magnetic Resonance Research. Functional imaging consisted of 35 T2*-weighted echo-planar image slices [repetition time (TR) = 3,000 ms, echo time (TE) = 54 ms, \( {\text{matrix}} = 128 \times 128 \)], 1.5 × 1.5 × 4.5 mm3 voxels. To facilitate later co-registration of the fMRI data into standard space, we also acquired a Turbo FLASH sequence [TR = 12 ms, TE = 5.65 ms] voxel size = 1 mm3. The first two EPI images in each session were discarded to avoid T1 equilibration effects.

Functional magnetic resonance imaging data analysis

fMRI data were pre-processed and analysed using FMRIB Expert Analysis Tool (FEAT) version 5.63, part of FMRIB Software Library (FSL version 3.3 β) (www.fmrib.ox.ac.uk/fsl). After realignment, spatial normalization and spatial smoothing (Gaussian kernel, 5 mm), the time series in each session were high-pass filtered (maximum of 0.04 Hz). FSL was used to compute individual subject analyses in which the time series were pre-whitened to remove temporal autocorrelation (Woolrich et al. 2001). Three experimental conditions were modelled: negative, positive and neutral pictures. Each condition was modelled separately by convolving trials with a canonical haemodynamic response function (Boynton et al. 1996). The following contrasts were chosen to maximize sensitivity to a drug effect on encoding of affect-laden information: negative–positive and positive–negative. All analyses at the group level employed a full mixed effects approach (Woolrich et al. 2004). Z (Gaussian T) statistic images were thresholded using clusters determined at Z = 2.0 and a corrected cluster significance of p = 0.05. For brain regions in which significant drug group by task interactions were observed in the whole-brain analyses, cluster maxima were localized using Talairach co-ordinates (Talairach and Tournoux 1988). Mean percent signal change in these regions was examined with analysis of variance to identify the profile of drug effect. To assess hippocampal and amygdala response during picture encoding regions of interest (ROIs) for the left and right hippocampus and amygdala in standard space were obtained using mri3dX (Lancaster et al. 2000). Mean percent signal change in these regions during the three experimental conditions was computed and compared between the groups. For the control stimulation paradigm, we compared mean percent signal change in patients given Epo vs. placebo within a region of the occipital (calcarine) cortex activated by photic stimuli (Maldjian et al. 2003).

Mood and subjective ratings

On the day 0, baseline mood and subjective state was assessed using the HRSD, CAS, BDI and HADS. Transient subjective state was assessed with the Befindlichkeits Scale (BFS; Von Zerssen et al. 1974), State Trait Anxiety Inventory (STAI; Spielberger 1983) and visual analogue scales (VAS) of relevant mood states at times −15 min and +120 min pre- and post-injection. Mood was recorded on the day before and every morning for 3 days after Epo/saline administration using the Positive And Negative Affect Scales (PANAS; Watson et al. 1988), BFS, STAI and VAS. On day 3, baseline mood was assessed with the BDI, HADS, BFS, STAI and VAS before the scan.

Statistical analysis of behavioural and mood data

Behavioural data and mood ratings were analysed using repeated measures analysis of variance with group as the between-subjects factor and time as the within-subjects factor (all tests were two-tailed). Significant interactions were analysed further using simple main effect analyses. Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS, version 15.0).

Results

Biological results

Analysis of blood samples before and after Epo/placebo administration revealed no effects of Epo on the levels of Hb (p > 0.84), hematocrit (p > 0.79) or RCC (p > 0.69) (see Table 2). Measurement of blood pressure after fMRI (day 3) revealed no effects of Epo on systolic or diastolic blood pressure (all p values > 0.56).
Table 2

Peak cluster activation in brain regions of significantly increased BOLD response over the threshold of Z = 2.0, P = 0.05, corrected, during encoding of affect-laden pictures in placebo-treated patients (main effect of task) and effects of Epo on day 3 post-administration

Task and Region

Z value

Coordinates

X

Y

Z

Negative pictures

 Main effect of task

 Medial occipital gyrus (BA 19)

5.23

26

−98

12

 Cuneus (BA 18)

4.68

−26

−98

10

 Fusiform gyrus (BA 19)

4.4

−24

−56

−14

 Medial frontal gyrus (BA 9)

3.63

48

14

30

 Medial frontal gyrus (BA 6)

3.41

0

10

56

 Superior temporal gyrus (BA 38)

3.36

−40

24

−22

 Medial frontal gyrus (BA 9)

3.41

−14

46

34

Positive pictures

 Main effect of task

 Medial occipital gyrus (BA 19)

4.84

26

−98

12

 Lingual gyrus (BA 19)

4.22

−20

−76

−8

 Superior frontal gyrus (BA 9)

3.73

−6

54

36

 Inferior frontal gyrus (BA 45)

3.06

−56

24

2

Neutral pictures

 Main effect of task

 Fusiform gyrus (BA 37)

4.54

−26

−54

−12

 Fusiform gyrus (BA 19)

4.85

26

−72

−12

 Medial frontal gyrus (BA 8)

3.2

−42

10

36

Negative > positive pictures

 Main effect of task

 Inferior frontal gyrus (BA 45)

2.8

60

22

8

 Placebo > Epo

 Inferior parietal gyrus (BA 40)

3.27

54

−34

52

 Medial frontal gyrus (BA 10)

3.45

18

48

−4

Positive > neutral pictures

 Epo > Placebo

 Inferior parietal gyrus (BA 40)

3.06

50

−66

42

Coordinates (X, Y and Z) refer to peak activation within each cluster identified using this threshold

BA Brodmann area

Blood-oxygenation-level-dependent (BOLD) response during encoding of affect-laden pictures

Main effect of task in placebo-treated patients

Hippocampus ROI

Encoding of picture stimuli activated the hippocampus bilaterally (paired t tests: t = 4.03, df = 16, p = 0.001) across subjects consistent with previous findings (Miskowiak et al. 2007b). Encoding of positive and negative pictures produced greater bilateral hippocampal response than neutral pictures (main effect of valence: F(2, 30) = 3.24, p = 0.05; post-hoc paired t test: t = 2.31, df = 16, p = 0.035). The left hippocampus was significantly more engaged in picture encoding than the right hippocampus (F(1, 15) = 4.47, p = 0.05).

Amygdala ROI

Presentations of negative, positive and neutral pictures produced bilateral amygdala activation across subjects (negative: t = 4.57, df = 16, p < 0.001; positive: t = 4.30, df = 16, p = 0.01; neutral: t = 3.53, df = 16, p = 0.003). The left amygdala was significantly more activated by picture stimuli than the right amygdala (main effect of lateralization: F(1, 30) = 6.69, p = 0.021).

Whole brain analysis

Processing of positive, negative and neutral pictures activated largely overlapping fronto-occipital networks including regions in the mPFC, fusiform gyrus and occipital cortex, consistent with previous reports (Britton et al. 2006; for cluster maxima, see Table 2). Negative pictures produced greater activation in the right ventrolateral (vl) PFC compared with positive pictures (for cluster maximum, see Table 2). This region was determined as the main effect of task ROI for investigation of drug effects in the preceding section.

Group x task interactions

Hippocampus ROI

Epo reduced neural response in the left hippocampus to negative compared with positive pictures (F(1, 15) = 4.78, p = 0.045; Fig. 2) in the absence of effects on right-side hippocampal response (ps > 0.14).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1641-1/MediaObjects/213_2009_1641_Fig2_HTML.gif
Fig. 2

BOLD signal change in the left hippocampus during encoding of negative and positive pictures. Epo (N = 9) reduced left hippocampal BOLD signal change during encoding of negative vs. positive pictures compared with placebo (N = 8; F(1, 15) = 4.78, p = 0.045). Values represent mean percentage signal change. Error bars show the s.e.m

Amygdala ROI

There were no effects of Epo on the left or right amygdala response to positive vs. negative pictures (p values > 0.20). However, Epo influenced lateralization of amygdala response during picture encoding (F(1, 15) = 10.79, p = 0.005); while placebo-treated patients displayed greater left vs. right amygdala activation (paired t tests: t = 7.26, df = 7, p < 0.001), this effect was abolished in the Epo-group (p > 0.55).

Main effect of task ROI

Extraction of signal change from the ventrolateral prefrontal region responding specifically to negative vs. positive pictures in placebo-treated patients revealed that Epo down-regulated response to negative pictures compared with placebo in this region (F(1, 15) = 20.78, p < 0.001; negative: t = −2.40, df = 15, p = 0.03; Fig. 3; for cluster maximum, see Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1641-1/MediaObjects/213_2009_1641_Fig3_HTML.gif
Fig. 3

Neural response to negative versus positive affective pictures in the right ventrolateral prefrontal cortex. Extraction of signal change from the region activated by negative vs. positive pictures in placebo-treated patients (main effect of task; area marked with green) revealed that Epo specifically reduced response during encoding of negative pictures (F(1, 15) = 20.78, p < 0.001; negative: t = −2.40; df = 15, p = 0.03) compared with placebo in this region involved in negative affect processing. For cluster maxima, see Table 2. Values represent mean percentage signal change. Error bars show the s.e.m

Whole-brain analyses

Epo modulated neural activation to negative vs. positive pictures in the right-side vm-vlPFC and parietal cortex (Fig. 4; for cluster maxima, see Table 2). Extraction of blood-oxygenation-level-dependent (BOLD) signal change from these regions revealed that Epo reduced neural response to negative pictures while increasing response to positive pictures in the vm-vlPFC (F(1, 15) = 57.80, p < 0.001; negative: t = −3.71, df = 15, p = 0.002 and positive: t = 3.29, df = 15, p = 0.005; Fig. 4) and reduced response to negative pictures in the parietal cluster (F(1, 15) = 46.86, p < 0.001; negative: t = −2.57, df = 15, p = 0.021; Fig. 4; for cluster maxima, see Table 2).
https://static-content.springer.com/image/art%3A10.1007%2Fs00213-009-1641-1/MediaObjects/213_2009_1641_Fig4_HTML.gif
Fig. 4

Neural response to negative versus positive affective pictures of the right ventromedial prefrontal and parietal cortex. Whole brain exploratory analysis revealed that Epo-modulated neural response to negative versus positive pictures in the right vmPFC and parietal cortex (areas marked with dark and light blue). Extraction of signal change from these regions revealed that Epo reduced neural response to negative pictures while increasing response to positive pictures in the vmPFC (F(1, 15) = 57.80, p < 0.001; negative: t = −3.71, df = 15, p = 0.002 and positive: t = 3.29, df = 15, p = 0.005) and reduced response to negative pictures in the parietal cluster (F(1, 15) = 46.86, p < 0.001; negative: t = -2.57, df = 15, p = 0.021) compared with placebo. For cluster maxima, see Table 2. Values represent mean percentage signal change. Error bars show the s.e.m

Visual stimulation control experiment

In the region of the occipital (calcarine) cortex identified as an ROI, mean percent BOLD signal change revealed no differences between groups (t = 1.05, df = 15, p = 0.31), suggesting that the observed effects of Epo did not result from global haemodynamic changes.

Picture recall

Epo had no effect on the total recall of positive, negative or neutral pictures (all p values > 0.11). However, Epo-treated patients produced fewer false recollections (self-invented pictures; t = −3.29, df = 15, p = 0.005).

Mood and subjective state

Day 0

There were no differences between groups in baseline subjective state measured with the STAI-state (p > 0.42), BFS (p > 0.19) and VAS of relevant mood states (all p values > 0.17). STAI-state ratings revealed reduced anxiety over time across all patients (F(1, 14) = 24.65, p < 0.001) with a greater reduction in patients given Epo vs. placebo (F(1, 14) = 7.52, p = 0.016; post-hoc t tests: p values > 0.34). BFS ratings showed general improvement of mood over time (F(1, 14) = 16.14, p = 0.001) but this was not influenced by Epo (p values > 0.14). VAS ratings of happiness and sadness revealed improvement for all patients (happiness: F(1, 14) = 12.17, p = 0.004; sadness: F(1, 14) = 14.99, p = 0.007) but this was not influenced by Epo (all p values > 0.30). There were no changes over time and no effects of Epo in VAS-ratings of anxiety, alertness or nausea (all p values > 0.10). There was a trend that dizziness increased over time (F(1, 14) = 4.37, p = 0.055) and a group x time interaction (F(1, 14) = 4.91, p = 0.044) driven by a trend that dizziness increased in the placebo-group (t = 1.88, df = 6, p = 0.11).

Daily ratings

Daily ratings of mood on the PANAS revealed reduced negative affect across all patients (F(3, 42) = 4.72, p = 0.006) but no effects of Epo (p values > 0.47). BFS revealed no changes in mood or energy over time (p > 0.12) or effects of Epo (all p values > 0.49). VAS ratings also showed no differential effects of Epo over time (all p values > 0.21).

Day 3

All patients experienced mood improvement from day 0 to day 3, as reflected by a reduction in scores on the HRSD (F(1, 15) = 11.80, p = 0.004), CAS (F(1, 15) = 6.44, p = 0.023), BDI (F(1, 15) = 11.39, p = 0.005) and HADS (F(1, 17) = 12.45, p = 0.003; see Table 1). However, no effects of Epo were demonstrated on these scales (all p values > 0.32). Baseline BFS and STAI-state ratings on the scan day revealed no differences between the groups in subjective mood (p values > 0.49).

Discussion

The present study is the first to demonstrate evidence of direct neurobiological effects of Epo on neuronal and cognitive function in individuals with major depressive disorder. Three days after administration, a single dose of Epo (40,000 IU) vs. saline reduced left hippocampal response during encoding of negative vs. positive pictures. This was paired with decreased activation to negative vs. positive pictures in vm-vlPFC and parietal cortex. After the scan, Epo-treated patients demonstrated fewer memory intrusions or false memories. Notably, the effects occurred in the absence of changes in mood, red cell mass or global blood flow in the brain, suggesting that they originated from direct neurobiological actions of Epo in this emotion-processing circuit. These preliminary findings are similar to the effects of Epo seen in healthy volunteers and suggest that findings from these models can be translated into a patient group.

Encoding of negative IAPS pictures produced greater right-side vlPFC response than positive pictures in our sample of acutely depressed patients, consistent with previous research in healthy volunteers (Pourtois and Vuilleumier 2006) and depression (Lang et al. 1997; Davidson 2003). Notably, Epo reduced neural response to negative pictures compared with placebo in this region specifically involved in negative affect processing. Epo also down-regulated neural response to negative vs. positive pictures in the right-side parietal cortex at the level of the whole brain analysis. A similar parietal-occipital circuitry has previously been implicated in attention to emotional material (Pourtois and Vuilleumier 2006) and replicates previous effects of Epo seen in healthy volunteers (Miskowiak et al. 2007a). As such Epo would be predicted to decrease exaggerated threat-relevant processing observed across depression and anxiety disorders (Williams et al. 1997).

Epo also reduced vmPFC responses to negative affective stimuli again opposite to effects seen during anxiety and depression (for review, see Drevets 2000) but similar to the effects of successful antidepressant drug treatment (Brody et al. 1999). Notably, over-recruitment of the mPFC during transient sadness may reflect disturbed mood regulation (Beauregard et al. 1998) while mPFC hypo-response to pleasant picture stimuli has been hypothesized to underlie abnormal positive affect processing in depression (Schaefer et al. 2006). The effect of Epo on vmPFC responses to affect-laden picture stimuli is hence consistent with a putative role of Epo in mood regulation and affective processing in acutely depressed patients in a way that is compatible with an antidepressant drug action (Harmer 2008).

This is an exploratory study in a small patient sample. It illustrates preliminary evidence for the potential for experimental paradigms to aid proof of concept in psychiatry as in other medical areas (Dawson and Goodwin 2005). The primary limitation of the study is the small sample size, which may have led to type II errors. In particular, the acute mood improvement following administration of Epo vs. placebo to healthy volunteers (Miskowiak et al. 2007a, 2008) was not replicated in these acutely depressed patients, though decreases in anxiety levels were seen. It is possible that repeated Epo administration is necessary for mood effects to occur in clinically depressed patients, as it is the case with conventional antidepressants. However, it is also possible that the absence of mood effects was due to lack of statistical power, given the relatively low number of patients and considerable variation in symptom presentation and depression severity. The majority of patients were also taking antidepressant medication, which may have had non-specific effects on outcome measures. Although the two groups were well matched for age, gender, years of education and mood, similar to our previous studies of Epo (Miskowiak et al. 2007a, 2008) and antidepressants (Harmer et al. 2003, 2006; Miskowiak et al. 2007c; Norbury et al. 2008), replication of these effects in a larger group is therefore warranted. Further, a longitudinal study of the same patients scanned twice (before and after Epo/ placebo administration) would have supported more robust inferences about the effects of Epo on emotional processing but a cross-sectional design was chosen to minimize learning and habituation effects.

From a clinical perspective, the haematopoietic activities of Epo with repeated administration may limit its clinical use. Future studies may wish to assess whether such unwanted haematopoietic activities could be avoided with infrequent Epo administration while maintaining the neurotrophic effects. This also warrants studies evaluating the effects of novel Epo derivatives such as carbamylated Epo and asialoEPO, which exert neurotrophic actions in the absence of haematological effects (for review, see Brines and Cerami 2005). If the relevant effects seen for Epo were also seen for these compounds in healthy volunteer models and in depressed patients, they could have potential value as novel candidates for long-term treatment of depression.

In conclusion, the present study demonstrated antidepressant-like effects of Epo on neural response to affective picture stimuli in depressed patients. This translation of effects of Epo in healthy volunteers to a clinically depressed population raises the exciting possibility that Epo may be a candidate agent for future antidepressant treatment strategies. These findings provide the ground for a large-scale clinical trial with repeated Epo administration and longer follow-up.

Acknowledgements

We thank Drs. Digby Quested, Susan Shaw, Mary-Jane Attenburrow and Peter Sargent for their help with patient recruitment and Drs. Mike Browning, Danilo Arnone, Sarah McTavish and Matthew Taylor for the medical assistance. The study was supported by the Lundbeck Foundation, Denmark (grant no. 94/04).

Copyright information

© Springer-Verlag 2009