Neural correlates of memory in depression measured by brain perfusion SPECT at rest

Authors


Hideki Azuma, MD, PhD, Department of Psychiatry and Cognitive-Behavioral Medicine, Nagoya City University Graduate School of Medical Sciences, Mizuho-cho, Mizuho-ku, Nagoya 467-8601, Japan. Email: azma@med.nagoya-cu.ac.jp

Abstract

Aim:  Brain metabolism activated studies have indicated associations between memory and the anterior cingulate cortex and hippocampus in patients with depression. The aim of the present study was therefore to investigate memory function, measured as performance on the Wechsler Memory Scale–Revised (WMS-R), and its relationship to brain perfusion using single-photon emission computed tomography (SPECT) at rest in patients with depression.

Methods:  The Hamilton Rating Scale for Depression (HAMD) and WMS-R were measured for 17 patients with depression by an independent clinical evaluation team. Voxel-based correlation analyses were performed with statistical parametric mapping at an extent threshold of 200 voxels. Associations were controlled for state and trait factors.

Results:  WMS-R measurements of verbal, visual, and general memory were inversely correlated with brain perfusion in the right anterior cingulate cortex, left premotor cortices, and both regions, respectively. The HAMD directly correlated with brain perfusion in the right anterior cingulate cortex.

Conclusion:  Brain perfusion SPECT measurements of the anterior cingulate cortex at rest were associated with the severity of depression and immediate memory scores measured with the WMS-R.

IN SUBJECTS WITH untreated depression, positron emission tomography (PET) of cerebral glucose metabolism has identified compromised function in the left prefrontal cortex,1–3 bilateral prefrontal cortex,4–6 medial prefrontal cortex, anterior cingulate cortex,7–9 orbitofrontal cortex,10,11 the striatum,12 and the thalamus.13

Of the various cognitive functions, verbal memory declines considerably during depression.14–19 During memory tasks, functional magnetic resonance imaging (fMRI) and activation PET have shown that the prefrontal area, anterior cingulate cortex, and/or hippocampal area exhibited decreased activity or hypometabolism in patients with depression compared to normal controls.20–22 A few reports used single-photon emission computed tomography (SPECT) to show that brain perfusion at rest in the anterior and posterior cingulate cortex and hippocampus was associated with immediate and delayed recall on the adult verbal learning test (AVLT).23,24 We were interested in investigating whether memory, measured on the Wechsler Memory Scale–Revised (WSM-R),25 was correlated with perfusion of different brain regions.

In the present study we investigated the correlations between brain perfusion and memory, as measured on the WMS-R, in patients with depression. In comparisons of SPECT analyses, we controlled for the number of depressive episodes, the duration of the index episode, the severity of depression, age at onset, and age at examination, because memory function in patients with depression varies according to these state and trait factors.

METHODS

Subjects and basic procedures

This study included 17 consecutive patients diagnosed with depression who had been admitted to the Department of Psychiatry at Nagoya City University Hospital. All subjects had experienced a recent episode of unipolar major depression, as described by the DSM-IV.26 Diagnoses were determined by psychiatrists in a clinical interview and a review of psychiatric records. Patients with the following conditions were excluded from the study: schizophrenia, schizoaffective disorder, substance abuse or substance dependence, significant neurological illness or any other significant medical illness, or grave abnormalities on brain computed tomography (CT).

Evaluation of depression severity and administration of the WMS-R and the Mini Mental State Examination (MMSE) were conducted by an independent clinical team that included a research psychiatrist and a speech therapist.27 Depression severity was measured with the Structured Interview Guide of the Hamilton Depression Rating Scale (HAMD).28 SPECT was carried out after admission, with no more than 10 days between the neuropsychological evaluations and the SPECT (mean ± SD, 4.6 ± 2.7 days; range, 1–10 days). None of the 17 patients had visual or auditory deficits, difficulties with language, or impaired orientation at baseline. Antidepressant regimens remained unchanged during the study. The study protocol was approved by the Ethics Committee of the Nagoya City University Medical School. All subjects were informed about the purposes and procedures of the study and provided written consent to participate in the study.

The SPECT was performed for eight women and nine men with a mean age (±SD) of 51.3 ± 12.5 years; all were right-handed. Standard laboratory examinations were normal for all patients. All patients were diagnosed with major depressive disorder; there were no patients with bipolar disorder.

The ratio of patients with single-episode major depressive disorder to those with recurrent major depressive disorder was 5:12. The mean duration of depressive episodes for the group was 24.5 ± 33.5 months; the mean number of episodes was 1.5 ± 1.5; the mean age at onset was 44.4 ± 15.2 years, and the mean HAMD score was 16.8 ± 7.4. On admission, each subject had received 2.0 ± 0.8 antidepressant medication trials. The antidepressant dose during the study was 30.8 ± 19.1 mg/day of fluoxetine equivalents, calculated according to the World Health Organization (WHO) defined daily dosage method (WHO Collaborative Centre for Drug Statistics Methodology, ATC/DDD System, 2006). Several of the 17 patients were taking other medications, including antipsychotics (three patients), benzodiazepines (three patients), and an anti-Parkinson's drug (one patient).

SPECT procedure

The SPECT was conducted at the Department of Radiology of Nagoya City University Hospital. All subjects were studied in the supine resting position in a quiet room with the patient's head positioned in a molded head holder. Approximately 755–825 MBq of 99mTc-ECD was injected i.v. after a 5–10-min rest period. Fifteen minutes after injection, scans were performed using a triple head gamma camera equipped with low-energy, high-resolution, parallel hole collimators (Siemens Multispect3, Berlin, Germany). The energy window was set at 140 keV with a 15% width. The images were collected within a 128 × 128 matrix with a pixel size of 2.46 mm. The acquisition time for each projection was 20 s, with a total imaging time of 24 min for all 72 projections. Images were reconstructed by filtered back-projection (Ramp filters) with Butterworth prefiltering (a cut-off frequency of 0.38 cycle/cm and an order of 4). Attenuation correction was performed using Chang's algorithm with an attenuation coefficient of 0.1 cm−1. Spatial resolution of a full width at half-maximum in the reconstructed images was 7.6 mm.

SPECT image analysis

All images were transferred from the SPECT imaging unit to a Windows workstation. Statistical analysis of all data was conducted on a voxel-by-voxel basis using Statistical Parametric Mapping 2 (SPM) software (Wellcome Department of Cognitive Neurology, Institute of Neurology, London, UK) implemented in Matlab 6.5 (Mathworks, Sherborn, MA, USA).29 The Siemens Multispect3 image files were converted into ANALYZE format. All images were spatially normalized using 12-parameter affine (linear) transformation. A further 12-parameter non-linear iteration algorithm was applied with basis functions of 7 × 8 × 7 on the stereotactic Montreal Neurological Institute (MNI) 99mTc-ECD template to remove inter-subject anatomical variability (voxel size, 2 × 2 × 2 mm). The gray matter threshold was set at 0.8.

Smoothing of spatially normalized images was performed by proportionally scaling the global cerebral blood flow to 50. Ultimately, the statistical parametric map constituted the set of values used for comparison. The SPM{t} maps were then transformed to normally distributed units (SPM{z}) and examined at a threshold voxel level of uncorrected P = 0.001, with an extent threshold of 200 voxels. This relatively large cluster size was chosen in order to cancel out possible noise due to multiple comparisons. The significance of identified regions was assessed with P corrected for multiple comparisons. A corrected P < 0.05 was considered statistically significant at the cluster level.

We investigated the correlations between brain perfusion and the WMS-R subscales by means of the ‘multi-subjects: covariates only’ option. The covariate was the score on each subset of the WMS-R. We input the age at examination, age at onset, and the presence or absence of a recurrent depressive episode as ‘nuisance variables’ to account for their confounding effects on both perfusion data and memory in depression. We assumed that the age at onset and the presence or absence of recurrence were also associated with the severity of cognitive impairment.14,15,17,19 The HAMD was input as a ‘nuisance variable’ to control for the degree of brain perfusion during depression. Alternatively, we explored correlations between brain regions and the HAMD by inputting the age at examination, age at onset, presence or absence of recurrence, and verbal memory score from the WMS-R as nuisance variables. An extent threshold of 200 voxels was used to ensure the corrected P = 0.05 (false discovery rate). The SPM contrast +1 or −1 was defined in each evaluation to search for a correlation.

RESULTS

Table 1 lists the scores observed on the WMS-R and MMSE for the 17 patients. It also provides the normative mean ± SD scores for a standardized Japanese sample.30 Impairments (≤1SD below the mean) on the WMS-R subscales were observed in nine patients (52.9%; 9/17) for verbal memory, three (17.6%; 3/17) for visual memory, six (35.3%; 6/17) for general memory, two (11.8%; 2/17) for attention/concentration, and six (35.3%; 6/17) for delayed memory. There was no statistically significant correlation between the HAMD and WMS-R scores.

Table 1.  MMSE and WMS-R scores for 17 patients with depression
 Test Score
Mean ± SD
Normative test scores in the Japanese standardized sample
Mean ± SD
  • The WMS-R consisted of five subscales: verbal memory, visual memory, general memory, attention/concentration, and delayed memory. A higher score indicates better memory performance. The definition of impairments is ≤1SD below the mean on the WMS-R subscales.

  • MMSE, Mini-Mental State Examination; WMS-R, Wechsler Memory Scale–Revised.

MMSE28 ± 1.9
WMS-R  
Verbal memory85.9 ± 18.8100.2 ± 14.8
Visual memory97.4 ± 15.6100.1 ± 13.5
General memory87.4 ± 19.1100.1 ± 14.7
Attention/concentration105.1 ± 15.5100.1 ± 14.7
Delayed memory91.1 ± 19.7100.4 ± 14.6

Each subset of the WMS-R or the HAMD was investigated separately for its potential correlation with the brain perfusion SPECT taken at rest; the threshold significances described in the previous section were used for this analysis (Table 2). The right anterior cingulate cortex and left premotor cortices emerged as significant brain regions (Fig. 1). We then used Spearman's rho to investigate the correlations between the identified areas of brain perfusion and each subset of the WMS-R or HAMD. Verbal memory was negatively correlated with brain perfusion in the right anterior cingulate cortex (Brodmann's area 24, ρ = −0.67, P = 0.003; Fig. 1). Visual memory was negatively correlated with perfusion in the left premotor cortices (Brodmann's area 6, ρ = −0.74, P = 0.001; Brodmann's area 8, ρ = −0.74, P = 0.001). General memory was negatively correlated with perfusion in the right anterior cingulate cortex (Brodmann's area 24, ρ = −0.70, P = 0.002) and left premotor cortices (Brodmann's area 6, ρ = −0.70, P = 0.002; Brodmann's area 8, ρ = −0.61, P = 0.009). The HAMD was positively correlated with perfusion in the right anterior cingulate cortex (Brodmann's 24, ρ = 0.92, P < 0.001). No correlations were detected between perfusion and attention/concentration or delayed memory.

Table 2.  Brain perfusion SPECT vs WMS-R and HAMD
 Cluster levelVoxel level
Cluster extentCorrected PCortical regionZ score of maximumTalairach coordinatesCortical regionBrodmann's area
  • P < 0.05, corrected for multiple comparisons at the cluster level, was considered statistically significant.

  • For the WMS-R subsets, we controlled for age at examination, age at onset, the presence or absence of recurrence, and HAMD. Verbal memory was negatively correlated with brain perfusion in the right anterior cingulate cortex (Brodmann's area 24). Visual memory was negatively correlated with the left premotor cortices (Brodmann's area 6, 8). General memory was negatively correlated with the right anterior cingulate cortex (Brodmann's area 24) and left premotor cortices (Brodmann's area 6, 8). No correlations were detected for attention/concentration or delayed memory. For HAMD, we controlled for the age at examination, age at onset, presence or absence of recurrence, and the WMS-R verbal memory. The HAMD was positively correlated with the right anterior cingulate cortex (Brodmann's area 24).

  • HAMD, Hamilton Rating Scale for Depression; L, left; R, right; WMS-R, Wechsler Memory Scale–Revised.

WMS-R subsets
Verbal memory3050.03R limbic lobe5.0310, −5, 26Cingulate gyrus24
Visual memory5350.002L frontal lobe4.42−31, 22, 50Superior frontal gyrus8
  L frontal lobe4.33−22, 18, 61Superior frontal gyrus6
  L frontal lobe3.97−24, 24, 50Superior frontal gyrus8
General memory2710.04R limbic lobe4.6211, −5, 27Cingulate gyrus24
  R limbic lobe4.5112, −7, 31Cingulate gyrus24
3020.03L frontal lobe4.47−22, 18, 60Superior frontal gyrus6
  L frontal lobe4.10−27, 20, 57Middle frontal gyrus6
  L frontal lobe4.08−31, 22, 54Middle frontal gyrus6
  L frontal lobe3.89−34, 16, 45Middle frontal gyrus8
  L frontal lobe3.37−15, 14, 64Superior frontal gyrus6
  L frontal lobe3.19−10, 12, 65Superior frontal gyrus6
HAMD213<0.001R limbic lobe5.2512, −9, 31Cingulate gyrus24
Figure 1.

Z maps of brain perfusion are shown for the (a) verbal, (b) visual, and (c) general memory subscales of the Wechsler Memory Scale–Revised (WMS-R) and (d) Hamilton Rating Scale for Depression (HAMD). Each z map reached a threshold of uncorrected P = 0.001. In the case of HAMD, the threshold was set at a corrected P = 0.05 (false discovery rate). The significance of the identified region was assessed using corrected P < 0.05 at the cluster level. (a) Verbal memory was negatively correlated with brain perfusion in the right anterior cingulate cortex (Brodmann's area 24). (b) Visual memory was negatively correlated with the left premotor cortices (Brodmann's area 6, 8). (c) General memory was negatively correlated with the right anterior cingulate cortex (Brodmann's area 24) and left premotor cortices (Brodmann's area 6, 8). (d) The HAMD was positively correlated with the right anterior cingulate cortex (Brodmann's area 24). No correlations with attention/concentration or delayed memory were detected. See Table 2 for details of the Talairach coordinates. L, left; R, right.

DISCUSSION

The present results demonstrate that immediate memory was inversely correlated with brain perfusion of the right anterior cingulate cortex and/or premotor cortices. Some subsets of the WMS-R had better correlations than others; verbal memory, visual memory, and general memory had good correlations, but delayed memory had no correlation. The brain regions identified in the present study were consistent with regions identified in prior activation studies of depression.20–22 In addition, the HAMD was positively correlated with the right anterior cingulate cortex after controlling for WMS-R verbal memory. These results, however, should be interpreted with caution. We found no correlation between the severity of depression and the subsets of memory. Therefore, the present results do not indicate, for example, that a depressive patient with high brain perfusion in the anterior cingulate cortex will manifest a severely depressive state with poor memory performance.

In previous activated fMRI or PET, an associative learning paradigm or verbal memory tasks were given to the patients with depression and healthy controls. The results showed that the activity or metabolism in the prefrontal cortex, anterior cingulate cortex, and right hippocampal area was higher in controls than that observed in patients with depression.21,22 These results suggested that patients with depression have deficits in hippocampal and anterior cingulate function. Furthermore, on fMRI, working memory task was associated with the increased activity of prefrontal cortex in the patients with depression.31,32

In contrast, in correlation PET or SPECT performed in the resting state with a memory test, Ebmeier et al. found that perfusion in the anterior and posterior cingulate cortex, and hippocampus were probably relevant to symptom changes.24 Nevertheless, Bench et al. and Austin et al. showed that perfusion in the medial frontal cortex, cerebellar vermis, and posterior cingulate were the significant regions.2,23 In the present study the verbal memory (Brodmann's area 24) of the WMS-R was associated with significant perfusion of the anterior cingulate cortex; this region was consistent with the regions identified in brain activated studies. The present results for visual memory (Brodmann's area, 6,8) and general memory (Brodmann's area, 6,8 and 24) may also be consistent with the results observed in the brain activated studies.21,22 The present results showed a negative correlation between memory function and brain perfusion, although the significance of this negative association was unclear, because the present study was not compared with normal controls. Further experimental study is needed to investigate the correlation between subsets of the WMS-R and brain perfusion.

We found a positive correlation between the HAMD and brain perfusion in the anterior cingulate cortex of the present patients, after controlling for age, age of onset, number of episodes, and verbal memory from the WMS-R. Austin et al. reported a correlation between the HAMD and the posterior cingulate cortex in a correlation SPECT study, after controlling for age and medication.23 Osuch et al. performed a correlation PET study and reported that the HAMD was correlated to brain metabolism in the bilateral medial frontal, right anterior cingulate, and right dorsolateral prefrontal cortices, after controlling for age, gender, and anxiety score.33 The differences between the methods and control factors may have influenced the results. The present study was a SPECT study performed at rest without normal controls, thus, we could not determine whether there was hyperperfusion in the anterior cingulate cortex of the present patients.

The function of the anterior cingulate cortex has been associated with emotional, sensory, response selection, and cognitive processes.34 Furthermore, studies of cingulate epilepsy syndrome have provided evidence of the function of the anterior cingulate cortex in movement, affect, and social behavior.35 Thus, the cingulate gyrus can be viewed as both an amplifier and filter, interconnecting the emotional and cognitive components of the mind.34

The present study had several limitations that should be considered. First, this was not an activation study and we did not compare patients with matched controls; thus, the results should be interpreted cautiously and considered as purely investigational. Second, the present patients had drug-resistant depression and continued to take antidepressants and other medications during the evaluation. It is unclear whether the degree of medication might influence brain perfusion. Bench et al. reported that there was no significant difference between brain perfusion in medicated and non-medicated patients in the regions where the maximum difference was found among patients with depression compared to normal controls.2 Third, the present patient ages ranged widely, and it would have been better to correct the partial volumes with MRI data, because the volume of the anterior cingulate cortex is associated with an aging effect.36 In the present study we did control for age and other trait and state factors.

In conclusion the present study demonstrated that in patients with depression, verbal memory, visual memory, and general memory measured on the WMS-R were inversely correlated with perfusion of the right anterior cingulate cortex, left premotor cortices, and both regions, respectively. Consistent with fMRI or PET studies, we showed that perfusion of the anterior cingulate cortex in patients with depression was associated with the severity of depression and immediate memory.

ACKNOWLEDGMENTS

We would like to thank the 17 patients who voluntarily participated in the study and the senior residents who cooperated in data collection.

Ancillary