INTRODUCTION

Electroconvulsive therapy (ECT) is well known for its strong antidepressant effect, but there is little information about the neurobiological mechanisms underlying these effects (Fink, 2001). Recent developments in cell and molecular biology have led to the hypothesis that reductions in neurotrophic signaling may be important in precipitating clinical depression (Duman and Vaidya, 1998; Jacobs et al, 2000). There is considerable evidence that antidepressant therapy exerts positive neurotrophic effects (Manji et al, 2000). Since it was shown in animal experiments that ECT induces mossy fiber sprouting in the rat hippocampus (Chen et al, 2000; Lamont et al, 2001; Vaidya et al, 1999) and stimulates neurogenesis (Madsen et al, 1998), a neurotrophic mode of antidepressive action was also proposed for ECT (Duman and Vaidya, 1998). However, neurotrophic effects are difficult to detect in the human brain in vivo. Proton magnetic resonance spectroscopy (1H MRS) poses an unprecedented opportunity for noninvasive assessment of brain metabolism in well-defined brain areas. N-acetylaspartate (NAA), an amino acid synthesized primarily in neurons (Simmons et al, 1991) and detectable by MRS, can be used as an indirect measure of neuron functionality (Sager et al, 2001; Tsai et al, 1995). Moreover, mature oligodentrocytes can express NAA in vitro (Bhakoo and Pearce, 2000) and it was suggested that glia cells are also involved in NAA/NAAG metabolism. NAA/NAAG may play an important role in cell-specific glia signaling and in the establishment and maintenance of the nervous system (Baslow, 2000). Recently, NAA increase was observed after long-term lithium therapy in cortical brain areas of bipolar patients (BIPs) (Moore et al, 2000), indicating a neurotrophic/neuroprotective effect. A recent 1H MRS study investigating bilateral hippocampi of ECT-treated patients found NAA levels unaltered (Ende et al, 2000).

There is some evidence to be derived from morphological (Ali et al, 2001; Altshuler et al, 1998), neuropathological (Sheline et al, 1998), and functional imaging studies (Drevets, 2000) that the amygdalar region is involved in the pathophysiology of depression. Here, we investigated 28 severely depressed patients by 1H MRS before and after ECT to assess probable neurotrophic and other metabolic effects related to ECT. We chose a voxel comprising the left amygdala, including the anterior hippocampus, a region centrally involved in the regulation of emotion and vigilance (Davis and Whalen, 2001).

MATERIALS AND METHODS

Patients

A total of 28 patients fulfilling the diagnostic criteria for major depressive episode (DSM-IV) and referred to ECT because of treatment resistance were enrolled in this study. Patients were classified as being severely depressed due to their high Montgomery-Asperg-rating scale (MADRS) values and length of index episode (Table 1). The diagnosis was made independently by two experienced psychiatrists employing the Munich checklist for DSM-IV diagnoses (Hiller et al, 2000). Patients with a medical history of alcohol and drug abuse or dementia were excluded. Treatment resistance was defined as failure to respond to at least two different antidepressants given longer than 4 weeks at the maximum recommended dose. All patients were off psychotropic medication between 3 and 8 days (Table 1) (only lorazepam allowed at maximum 3 mg/day) and were investigated before and after ECT. The reason for discontinuation of psychotropic medication was to provide a standard ECT monotherapy protocol; however a longer washout period was not feasible, since our patients were severely ill and a longer postponing of indicated ECT was not acceptable. After a detailed description of the study to the subjects, written informed consent was obtained. The protocol was approved by the university ethics committee and the studies were carried out in accordance with the Declaration of Helsinki. The sample consisted of 13 unipolar (UPs) and 15 BIPs. Clinical characteristics are summarized in Table 1; there was no statistically significant intergroup difference regarding relevant clinical parameters. Response was defined as a MADRS reduction >60% of baseline scores (Prudic et al, 1996). Patients with an insufficient clinical response were put on combined ECT/pharmacotherapy. After cessation of ECT monotherapy, 14 patients had not responded; five of these were remeasured after finally responding. Pharmacotherapy was different for each patient (patient 1: trijod-thyro-nine, venlafaxine; patient 2: lithium; patient 3: lithium, valproic acid; patient 4: tranylcypromine, valproic acid; patient 5: tranylcypromine, lithium). The number of additional ECTs was 8 (n=1) for the UPs and 7.3±7.2 (n=4) for the BIPs.

Table 1 Clinical Characteristics of Patient and Control Samples
Full size table

Age- and gender-matched healthy controls (unipolar controls: mean age 58.5±8.1 years; 9 female; bipolar controls: mean age 52.2±14.4 years; 10 female) were enrolled. Written informed consent was obtained and the same standardized interview as for patients was performed to rule out any underlying psychiatric or neurologic condition (Hiller et al, 2000).

ECT Treatment

ECT was applied using a customized brief-pulse device (Thymatron IV, Somatics, Lake Bluff, IL, USA) beginning with stimulus titration in the first session and continuing with the 2.5-fold stimulus dose. Motor and EEG seizure duration were monitored, and conservative criteria for adequacy were used, with the stimulus intensity being adjusted accordingly. Methohexital (0.75–1.5 mg/kg) and succinylcholine (0.5–1.0 mg/kg) were applied for anesthesia and relaxation. No atropine was applied prior to the treatment. Patients required a mean of 11.6±5 ECT treatments. In 27 of the 28 patients stimulation was right unilateral, and in one bilateral.

MRS Measurements

The MRI protocol included T1-weighted 3D spoiled gradient echo acquisition of the whole brain and T2- and proton density-weighted fast spin echo sequences in transaxial and coronal orientation. For the MRS measurements, single voxel STEAM spectroscopy (TE=20 ms, TR=2.5 s, voxel size (VOI)=3.375 cm3, number of scans=128) at 1.5 T (Magnetom SP, Siemens, Erlangen, Germany) was used (voxel position: left amygdalar region—comprising the maximum of amygdalar gray matter (GM) and the anterior one-third of the hippocampus). Voxel positions between scans were referenced to readily identifiable anatomical landmarks (inferior temporal horn).

Postprocessing and Data Analysis

To account for variations in voxel composition (cerebral spinal fluid (CSF), GM, and white matter), segmentation of the voxels was performed with a semiautomatic interactive algorithm (M Fiebich, Department of Biomedical Engineering, University of Applied Sciences, Giessen, Germany). The metabolite concentrations were corrected for partial volume effects (CSF) and normalized to the GM fraction in the voxels (Wang and Li, 1998) and coil loading. After postprocessing of our spectra, metabolite concentrations were calculated using a time domain fitting program (LC model (Provencher, 1993); Figure 1). Owing to overlapping resonances of glutamate, glutamine, and GABA at a field strength of 1.5 T, only a combined fitting of Glx (glutamine, glutamate) was performed. GABA was excluded from the processing and data analysis, since the fitting error (percent standard deviation SD) was too high to be meaningful. For robustness of results, only metabolite information with a fitting error of <20% SD was included in the final analysis. Contribution of the altered GABA levels may still influence our data. However, the GABA concentration in general is much lower (about 1 mmol/kg brain tissue (Sanacora et al, 1999)) than combined Glu/Gln (about 12–14 mmol/kg brain tissue). Also, the Glx signal is mostly dominated by glutamate (Glu) (Auer et al, 2000).

Figure 1
figure 1

Original and fitted (LC model, black line) 1H MRS spectra of the left amygdalar region of a 54-year-old female patient pre-ECT. Baseline correction, but no line broadening was applied. NAA=N-acetylaspartate, Glx=glutamate, glutamine, Cr=creatine and Cho=choline-containing compounds.

Full size image

Statistical analysis was performed with SPSS software (SPSS 10.0 for Windows, SPSS Inc., Chicago, Illinois, USA). The Mann–Whitney test was employed as a nonparametric test between groups, the Wilcoxon test as a paired test for metabolite changes after treatment, and the Pearson test for correlation analysis.

RESULTS

Before ECT, the concentrations of NAA, choline (Cho), and creatine (Cr) in patients were not statistically different from those of controls. Only Glx showed a trend towards a reduction (p=0.056). Glx reduction was significant only in patients with unipolar depression (p=0.049). Segmentation of the MRS voxels revealed that the GM fraction was significantly higher in patients (p<0.0001) compared to controls. ECT did not affect voxel composition, segmentation before and after ECT did not reveal differences. A positive correlation was found between GM fraction and duration of illness (p=0.024). Successful ECT was accompanied by increased NAA (p=0.019, Figure 2) and Glx levels (p=0.049) compared to baseline before ECT for all patients. The number of ECTs did not differ between responders and nonresponders (Table 1). Cho and Cr levels increased in responders without reaching statistical significance.

Figure 2
figure 2

NAA levels (institutional units) before and after ECT monotherapy. (a) In treatment nonresponders (n=14) NAA did not increase, while (b) in treatment responders (n=14) NAA increased significantly (nonparametric paired test, Wilcoxon, p=0.019).

Full size image

When analyzing BIPs and UPs separately, the diagnosis had no major influence on our results, with the exception of reduced Glx levels pre-ECT. Only unipolar depressed patients had significantly reduced Glx levels compared to controls (p=0.04).

Five (n=4 BIPs I, n=1 UP) of the nonresponding patients were remeasured after they had finally responded to continued ECT combined with pharmacotherapy. Hereafter, they also showed a similar increase in NAA (p=0.043; Figure 3).

Figure 3
figure 3

NAA levels (institutional units) during ECT course (n=5). Low NAA levels of ECT nonresponders increased significantly after successful treatment (nonparametric paired test, Wilcoxon, p=0.043).

Full size image

DISCUSSION

In this study we investigated a clinically highly important subgroup of severely ill, treatment-resistant patients with major depressive disorder to see whether probable neurotrophic effects may play a role in the strong antidepressant efficacy of ECT. There is increasing evidence that negatively balanced neurotrophic factors may contribute to the precipitation of severe mood disorders (Duman et al, 1997; Jacobs et al, 2000). On the other hand, as known from neurobiological studies, antidepressant strategies, for example, tricyclics, lithium, or ECT, may increase neuronal functionality/viability (Duman et al, 1999).

The major finding of this study was that ECT therapy was accompanied by a significant increase in NAA levels in responders (Figure 2). NAA is located exclusively in neurons (Simmons et al, 1991) and is suggested to be a marker of neuronal viability and functionality. It has been reported that chronic lithium therapy in therapeutic doses leads to increased NAA levels because of increased neuronal viability, probably mediated by a neuroprotective protein of B-cell lymphoma/leukemia-2 called bcl-2 (Moore et al, 2000); bcl-2 has been shown to prevent neuronal cell death in vivo and lithium therapy is able to induce its robust increase. The parallel findings of increased NAA levels after ECT as well as after lithium therapy suggest a similar neurotrophic effect.

Additionally, ECT-induced structural changes have been described in animal experiments, that is, mossy fiber/synaptic sprouting (Lamont et al, 2001; Vaidya et al, 1999) or neurogenesis (Madsen et al, 2000; Parent et al, 1997). Overall, these findings may indicate improved neuronal functioning (Moore et al, 2000).

The missing NAA increase in nonresponders probably indicates that a neurotrophic effect may be important for antidepressant efficacy. Differences between responders and nonresponders to ECT were observed by Bonne et al (1996) by single-photon emission computerized tomography (SPECT), who described reduced cerebral blood flow pre-ECT and an increase only in ECT responders. The most robust changes were observed in the cingulate gyrus, which is part of the paralimbic cortex and also but less pronounced in the parietal lobes.

Five of our 14 nonresponding patients remeasured after successful ECT combined with antidepressants (three of them additionally treated with lithium) presented a similar increase in NAA levels (Figure 3). Since three of five nonresponding patients received lithium and lithium can also increase NAA levels (Moore et al, 2000), we cannot rule out that this increase may be because of the effect of lithium and less to the positive ECT response.

Although the number of patients in the follow-on group was small, these results may indicate that patients originally not responding to ECT monotherapy can benefit from a more rigorous antidepressant strategy.

In addition, significantly reduced Glx levels were observed for UPs pre-ECT, which increased significantly after successful ECT. Similar results for unipolar depressives were observed for the left anterior cingulate (Pfleiderer et al, 2003), a region intensively interconnected with the amygdala (LeDoux, 2003). This finding indicates a pertubated Glx metabolism, which is dependent on intact neuron and/or glial cell function. Consequently, ECT may also contribute to increased NAA levels because of improved glial cell function and glutamate metabolism.

Interestingly, the analysis of voxel composition, comprising amygdala and about one-third of the anterior hippocampus, revealed an increased GM fraction in patients compared to controls before and also after ECT. Although reports on functional and structural changes of the amygdala related to depressive disorders are conflicting with respect to alterations in size (decrease vs increase) or metabolism (Altshuler et al, 1998; Drevets et al, 1999; Sheline et al, 1998), they nevertheless indicate a possible involvement of the left amygdala in the pathogenesis of major mood disorder. The observed positive correlation of GM fraction in our MR voxels with duration of illness may reflect long-term structural alterations occurring with mood disorders. This was also reported in a recent preliminary MRI study of larger left temporal lobes with increasing duration of illness (Ali et al, 2001). Consequently, the amygdalar region may be a suitable region for detecting neurotrophic effects by MRS, since it is a neuroplastic active brain area (LeDoux, 2000) and fills the voxel mainly with GM (Davis and Whalen, 2001). These may well be the neuroanatomic reasons for Ende et al (2000), not having been able to demonstrate an NAA increase after ECT in the bilateral hippocampus.

Our observations are limited by the relatively short off-medication period in this severely ill population. Consequently, pre-ECT pharmacotherapy and/or withdrawal of medication may influence metabolite levels. However, these medications had proved to be ineffective. As expected from an open label and naturalistic design, ECT responders and nonresponders may differ in various uncontrolled aspects (eg medication pre-ECT), which may influence our data. Whether neurotrophic effects may be relevant to antidepressant efficacy should be confirmed in a greater number of patients. Nevertheless our data support previously described neurotrophic effects of ECT in animals (Lamont et al, 2001; Vaidya et al, 1999; Madsen et al, 2000; Parent et al, 1997) and offer initial evidence that this may be important for effective antidepressant therapy in humans.