Skip Navigation LinksHome > June 2014 - Volume 30 - Issue 2 > What Does the Electroencephalogram Tell Us About the Mechani...
Journal of ECT:
doi: 10.1097/YCT.0000000000000144
Invited Reviews

What Does the Electroencephalogram Tell Us About the Mechanisms of Action of ECT in Major Depressive Disorders?

Farzan, Faranak PhD*; Boutros, Nash N. MD; Blumberger, Daniel M. MD, MSc, FRCPC*; Daskalakis, Zafiris J. MD, PhD, FRCP(C)*

Free Access
Article Outline
Collapse Box

Author Information

From the *Centre for Addiction and Mental Health, Temerty Centre for Therapeutic Brain Intervention, Department of Psychiatry, University of Toronto, Ontario, Canada and †Department of Psychiatry and Neurosciences, University of Missouri, Kansas City, MO.

Received for publication January 31, 2014; accepted March 20, 2014.

Reprints: Faranak Farzan, PhD, Centre for Addiction and Mental Health, Temerty Centre for Therapeutic Brain Intervention, University of Toronto, 1001 Queen St, Unit 4-118A, Toronto, Ontario, Canada (e-mail:

Conflict of Interest: DMB receives research support from CIHR, Brain and Behaviour Research Foundation (formerly NARSAD), the Temerty Family through the CAMH Foundation and the Campbell Research Institute; and equipment support-in-kind for an investigator-initiated study from MagVenture/Tonika and research and in-kind equipment support for an investigator-initiated study from Brainsway Ltd. In the past 5 years, ZJD received research and equipment in-kind support for an investigator-initiated study through Brainsway Inc and a travel allowance through Merck. ZJD has also received speaker funding through Sepracor Inc and AstraZeneca, served on the advisory board for Hoffmann-La Roche Limited and Merck, and received speaker support from Eli Lilly. This work was supported by the Ontario Mental Health Foundation (OMHF), CIHR, the Brain and Behaviour Research Foundation, the Temerty Family, and the Grant Family and through the Centre for Addiction and Mental Health (CAMH) Foundation and the Campbell Institute. FF and NNB have no conflict of interest to disclose.

Collapse Box


Abstract: Electroconvulsive therapy (ECT) remains to be one of the most effective treatment options in treatment-resistant major depressive disorder (MDD). From the early days, researchers have embarked on extracting information from electroencephalography (EEG) recordings before, during, and after ECT to identify neurophysiological targets of ECT and discover EEG predictors of response to ECT in patients with MDD. In this article, we provide an overview of visually detected and quantitative EEG features that could help in furthering our understanding of the mechanisms of action of ECT in MDD. We further discuss the EEG findings in the context of postulated hypotheses of ECT therapeutic pathways. We introduce an alternative and unifying hypothesis suggesting that ECT may exert its therapeutic efficacy through resetting the aberrant functional connectivity and promoting the generation of new and healthy connections in brain regions implicated in MDD pathophysiology, a mechanism that may be in part mediated by the ECT-induced activation of inhibitory and neuroplasticity mechanisms. We further discuss the added value of EEG markers in the larger context of ECT research and as complementary to neuroimaging and genetic markers. We conclude by drawing attention to the need for longitudinal studies in large cohort of patients and the need for standardization and validation of EEG algorithms of functional connectivity across studies to facilitate the translation of EEG correlates of ECT response in routine clinical practice.

Shortly after the introduction of electroconvulsive therapy (ECT) as a treatment for severe mental disorders, electroencephalography (EEG) was used to evaluate electrophysiological changes during the ictal phase of the treatment. 1 One concern that has preoccupied ECT researchers is whether any EEG changes that occur secondary to ECT can predict treatment response or are simply an epiphenomenon. In addition to the interest in predicting response to ECT, there is the desire to understand the mechanism through which ECT exerts its therapeutic effects in major depressive disorder (MDD).

The principal function of the electrical stimulus delivered with ECT is to induce a generalized tonic-clonic seizure. The progression of the ECT-induced seizure, as seen through EEG, starts with a very brief, and may at times not be detectable, electrodecremental period with very low–voltage fast activity. This is immediately followed by a high-amplitude polyspike phase, which marks the tonic and then early clonic phases of the induced seizure. This phase lasts a few seconds (usually <30 seconds) and progressively morphs into a regular pattern of polyspike and wave discharges that begin at a frequency of 4 to 6 Hz and eventually slowing to 1 to 2 Hz. This phase may constitute the bulk of the seizure activity and may not be very regular. 2 Induced seizures may abruptly end 3 or just continue to break up with fewer spike discharges with steadily diminishing amplitudes ending in a postictal state of diffuse slowing of the EEG with high-amplitude delta activity followed by a period of postictal suppression (Fig. 1). However, it should be noted that this pattern is not always observed, and sometimes, slow waves do not develop and the degree of post suppression may be minimal or even absent.

Figure 1
Figure 1
Image Tools

To date, numerous studies have extracted information from EEG recordings with 3 overarching aims. These include: (a) identifying predictive EEG markers of ECT therapeutic efficacy, (b) determining optimal parameters of stimulation, and (c) understanding the mechanism of action of ECT via monitoring of the ECT-related electrophysiological changes. A limited number of studies have also studied EEG to identify potential correlates of ECT-induced cognitive adverse effects. Depending on the aim of each specific study, EEG recordings are obtained at one or several time points relative to ECT administration including: (1) baseline (ie, before ECT treatment), (2) during ECT (ie, peri-ictally), (3) between ECT sessions (ie, interictally), and (4) after ECT treatment trial. The recording of the EEG during the ECT procedure is different from interictal EEGs in 3 major ways. First, in ictal studies, the number of recording electrodes is usually extremely limited (in the order of 1–2 electrodes) and mainly used to monitor seizure activity. The second fundamental difference is that interictally, patients are awake; whereas during ECT, patients are both anesthetized and paralyzed.

In this article, we introduce and review the evidence from both the visually inspected EEG (vEEG, without computer assistance) and computer-assisted quantitative EEG analysis (qEEG) for common EEG features across ECT studies in MDD that could help in furthering our understanding of the electrophysiological changes that may underlie the therapeutic efficacy in MDD or could predict response to treatment. We describe the previous hypotheses regarding the mechanisms of action of ECT in MDD and propose a modified hypothesis, taking into account ECT-related electrophysiological changes. We highlight the importance of EEG markers in understanding the mechanism of action of ECT in heterogeneous illnesses such as MDD that arise as a result of complex interaction between several genes and environmental factors. We conclude by discussing shortcomings of current EEG studies and briefly discuss the future direction.

Back to Top | Article Outline
The vEEG Correlates of ECT Therapy

Early literature examining the relationship between EEG variables and clinical response to ECT was inconclusive. 4 More consistent correlations between vEEG features and ECT therapeutic effects did emerge in later studies. 5 Some of the main visually inspected EEG features include ictal and peri-ictal EEG parameters such as polyspike-phase maximal amplitude, polyspike-phase duration, slow-wave phase maximum amplitude, slow-wave phase duration, regularity (global seizure strength), stereotypy (global seizure patterning), and postictal suppression (as seen below in the third panel of Fig. 1 between the 37 and 40 seconds). Burst suppression index is another method of quantifying postictal suppression. 6 This index can be defined for a fixed time window as the ratio of total suppression period during this time window divided by the duration of the time window. As discussed next, the association between these features and clinical outcomes has been explored in several previous studies.

The amplitude of ictal EEG during ECT varies significantly across the scalp but with a maximum typically over the frontocentral sagittal area. 7 Sackeim et al (1996) reported a greater increase in frontal and prefrontal low frequencies (0.5–3.5 Hz) and spectral amplitudes over the treatment course in therapeutic responders to ECT. 5 This work underscores the importance of the frontal regions in mediating the therapeutic response to ECT. In fact, evidence has been accumulating that ECT might lead to regeneration of neurons in the frontal cortex. In an animal model of ECT, it was reported that ECT, monoamine-oxidase inhibitors, and selective serotonin reuptake inhibitors increased the expression of brain-derived neurotrophic factor (BDNF) and its receptors in the frontal cortex and that this effect was not present with nonantidepressant psychotropic factors. 8 The state of postictal depression, which is observed in the immediate postictal period, has been found to be associated with a rapid and marked increase in BDNF expression and activity. 9 This effect on BDNF is also linked to the interictal slowing, which usually follows the immediate postictal suppression and provides for possible mechanism of continued action of ECT between and after the course of treatment. 10

Early work 11 provided data suggesting that the degree of EEG suppression immediately after the electrical stimulus as well as the degree of postictal suppression can be predictors of clinical efficacy. Suppes et al 12 provided additional data supporting the observation that the duration of the electrical silence (Fig. 1) after the application of the electrical stimulus (in this study assessed at 5 seconds) is a measure of the strength of the response to the electrical stimulus and could be a predictor of clinical response. To our knowledge, this work has yet to be replicated in larger sample studies.

Another potentially important factor that could be related to the therapeutic action of ECT is the effect on seizure threshold. It has been long observed that a variable increase in seizure threshold occurs over the course of treatment. 13 It was also reported that postictal suppression and BS index decreased as a function of number of ECT sessions. Furthermore, the midictal amplitude and the postictal suppression were highest in the first ECT session. 14 Krystal et al (1998) provided additional evidence that if the seizure threshold does not rise during the course of treatment the clinical outcome is likely to be unfavorable with the opposite being true. 15 This observation is particularly germane to the purpose of this review for two reasons. First, it relates to the kindling theory of mood disorders. 16 If kindling (ie, the lowering seizure threshold with repeated stimulation) plays a role in the pathogenesis of mood disorders, then one would predict that raising the seizure threshold (as with many anticonvulsant mood stabilizers) would lead to amelioration of mood symptoms. Such changes may also be related to the therapeutic effects of other forms of brain stimulation such as rTMS, as this latter technique has also been noticed to modify seizure threshold. 17

Finally, seizure duration does not seem to correlate with therapeutic response. 18 Other seizure characteristics such as the amplitudes and frequencies of the induced spikes, polyspikes, and spike wave discharges have received significant examination in relation to the therapeutic response, and while promising, further research is needed. 19

Back to Top | Article Outline


Parallel to digitization of EEG recordings and implementation of various linear and nonlinear EEG signal analysis algorithms, a growing list of EEG features are introduced that permit quantifying different electrophysiological characteristics of a healthy brain from reactivity in one area to nonlinear and long-range interaction between two or several functional units (eg, interaction between oscillatory activities or brain areas). Furthermore, the concurrent or offline combination of EEG with other neuroimaging and neuromodulatory methods such as functional magnetic resonance imaging (fMRI) or transcranial magnetic simulation (TMS) 20 has provided further insight into neurophysiological impairments in MDD and treatment-related amelioration of the underlying impairments. In this section, we first provide an overview of key EEG features that are commonly reported to differ between patients with MDD and healthy controls followed by discussion of several EEG features that are modified after ECT treatment or predict response to treatment.

Back to Top | Article Outline
Electrophysiological Impairments in MDD

There is much literature regarding the EEG abnormities across several behavioral states including but not limited to resting state, cognitive performance, and sleep (for a recent and comprehensive review, refer to Olbrich and Arns 21 ). While there are no absolutely consistent findings, which is not surprising given the diagnostic heterogeneity of mood disorders and the lack of standardized EEG recording and analysis guidelines across studies, there are a number of EEG abnormalities that have been documented.

Back to Top | Article Outline
Spectral Power

If EEG is considered a linear dynamical system, the signal recorded at electrode k and at time point t (ie, V k (t)) can be represented linearly through the Fourier series 22 as a function of a series (ie, N series) of sine waves described in terms of amplitude (A kn ), frequency (f kn ), and phase ( kn ). In this model, amplitude represents the maximum vertical peak of the sine wave (in μV), frequency is the number of complete cycles per second (in Hz), and phase describes the time point position with respect to the beginning of the sine wave.

Equation (Uncited)
Equation (Uncited)
Image Tools

Electroencephalographic signals can be converted from time domain to frequency domain using transfer functions such as fast Fourier transform. 23 A large number of EEG studies have examined the two QEEG measures of absolute power (squared of amplitude) and relative power of specific frequency bands in frequency domain. Relative power is calculated as the ratio of the absolute power in a specific band to the power across all frequency bands.

A large number of QEEG studies in MDD have examined the baseline difference in relative and absolute EEG power spectrum for cortical oscillatory activities categorized within delta (1–3 Hz), theta (4–7 Hz), alpha (8–12 Hz), beta (12–30 Hz), and gamma (>30 Hz) frequency bands. Among these frequency bands, the most commonly reported differences between patients and healthy controls are increases in the absolute alpha power in parietal and frontal regions. 24,25 Consistently, high baseline alpha power were associated with positive response to antidepressants (eg, Ulrich et al 26 ), and alpha power was used to distinguish responders and nonresponders vis-à-vis antipsychotic treatments (eg, Bruder et al 27 ). Moreover, recent studies have revealed that the BDNF Val66Met polymorphism may be indirectly associated with depression severity by mediating the EEG alpha oscillations in eyes-closed resting state. 28 Collectively, these studies support a functional role for alpha rhythms in at least a subgroup of patients with MDD.

In addition to alpha, increases of frontal midline theta oscillations have been observed in patients with MDD (eg, Jaworska et al 29 ), and it has been proposed that the degree of theta oscillation abnormalities may predict response to antidepressant treatments (eg, Arns et al 30 and Pizzagalli et al 31 ). Using source localization and implanted electrodes in primates, midline frontal theta oscillations have been associated with the activity of the anterior cingulate cortex (ACC), and consistently, ACC is considered to be a generator of theta oscillations. 32 Fewer abnormalities have been observed in fast frequency bands (beta to gamma oscillations), although some evidence exists of increased beta range activity in depressed patients (eg, 33 ).

Electroencephalographic cordance is another QEEG measurement developed in 1990s. 34 The algorithm for this feature involves a combination of relative and absolute power including the reattribution of EEG power, followed by spatial normalization of absolute and relative power, and combination of z-transformed absolute and relative power. This feature has been shown to have a strong correlation with brain perfusion 34 and is another EEG feature that has been explored as a biomarker and predictor of treatment response in MDD. One study, for example, has identified an association between baseline cordance and treatment response to antidepressants. 35

Interhemispheric asymmetry of frontal alpha power (eg, ratio of left to right alpha power) is another EEG feature that has been commonly detected in patients with MDD (recently reviewed in Olbrich and Arns 21 ). Several studies have reported that MDD is characterized by low alpha oscillations in the right prefrontal cortex and high alpha oscillations in the left prefrontal cortex. However, several factors such as EEG montage and asymmetric bone thickness can affect this feature, and thus far, this feature has not been shown to have a strong prognostic value in patients with MDD (discussed in Olbrich and Arns 21 ).

Back to Top | Article Outline
Functional Connectivity

An extended neural circuitry and network has been implicated in the pathophysiology of MDD including the medial prefrontal cortex, and limbic, striatal, thalamic, and basal forebrain structures. 36 The pathophysiology of MDD has been linked to impairment in functional connectivity, 37,38 in particular, involving structures such as the dorsolateral prefrontal cortex (DLPFC), ACC, thalamus, and hypothalamus. 39 In fact, the enhanced EEG power in theta and alpha bands is suggested to be a result of impaired connectivity between cortical and subcortical gray matter structures. Aside from EEG power analysis, which may indirectly reflect integrity of functional connectivity, several EEG features have been introduced to examine the interaction between two or more brain regions or cortical oscillations.

Electroencephalographic features of functional connectivity fall within two broad classes. The first, and more commonly used, are measures of undirected (ie, without quantification of the direction of information flow) connectivity, such as correlation, cross-frequency phase-phase or amplitude-phase coupling, coherence, and synchrony. The second set of measures capture direction of information flow. It is generally accepted that the direction of the brain signal propagation is associated with the direction of the information flow and may reflect the direction of influence of one brain area on another. 40 The measures that have been explored to estimate the causality between two or a cascade of sensors, are based on the Granger causality principle. 41 Granger causality states that if the past information in signal x i (t) can predict the signal x j (t), then x j (t) is caused by x i (t). A few measures that have been used to study the causality in EEG analysis include the directed transfer function and partial directed coherence. The directed transfer function is the multichannel estimator of the intensity of signal propagation between brain regions. The partial directed coherence 42 is similar to directed transfer function, and it provides a frequency domain measure for Granger causality (a more extensive review of these measures can be found elsewhere 43,44 ).

The aforementioned measures reflect the functional connectivity between two or more functional units such as brain areas, network nodes, and specific oscillatory activities. In MDD, decreases (eg, Lee et al 45 ) as well as increases of functional connectivity measures (eg, Fingelkurts et al 37 and Leuchter et al 38 ) have been reported in resting-state condition, albeit the latter seems to be more commonly reported. It has been argued that this discrepancy is likely related to the effect of volume conduction and sensitivity of EEG connectivity measures to factors such as choice of reference electrode (interested readers could refer to Haufe et al 46 ).

Quantification of the EEG signal complexity is another approach that may provide an index for assessing the integrity of the neuronal long-range correlation across multiple temporal and spatial scales, 47 in effect, an index of functional connectivity. A recently developed EEG feature of complexity is multiscale entropy (MSE), which quantifies the richness of interneuronal interaction across multiple temporal scales. 47 This feature was developed based on the observation that the outputs of complex systems exhibit structures with long-range correlation across multiple spatial and temporal scales and cannot be simply described by perfect regularity or complete randomness. 47 Recent preliminary evidence illustrated that MSE was increased in a small sample of patients with depression. 48

Back to Top | Article Outline
Quantitative EEG Correlates of ECT Therapy

Several ECT studies have quantified ECT-related changes in EEG power using longitudinal designs.5,49,50 The most consistent effect of ECT that outlasts the stimulation session is generalized slowing of the EEG. 51 The diffuse slowing, while it builds up during the course of ECT, begins to decrease immediately after the course of ECT ends. The diffuse slowing can take one of two major forms: regular activity in the delta and theta ranges or intermittent runs of regular delta activity that are usually bilaterally synchronous but maximally active in the frontal regions. Involvement of the frontal regions may prove crucial for mediating the therapeutic response to ECT. Although both forms of diffuse EEG slowing are correlates of a diffuse encephalopathic process, the two forms are not necessarily the same physiologically, as intermittent slowing that is maximal in frontal regions tends to correlate with delirium, whereas regular diffuse slowing may correlate more with degenerative disorders like dementia. 52 It should be noted that this potential difference in EEG changes that occur after ECT has not been explored any further in the literature. Furthermore, follow-up EEG studies have varied in their estimates of when the EEG slowing completely abates but varies from hours to months. There has been a suggestion that if the EEG was already abnormal before ECT, it takes significantly longer to return to baseline. 53

Among one of the early studies was also an observation that immediately after unilateral ECT, the treated side had more slow wave delta activity (1.2–4 Hz) and less alpha and beta oscillations, whereas more alpha and beta oscillations were observed on the untreated site. 49 Furthermore, induction of slow wave activity in the prefrontal cortex was linked with the magnitude of symptomatic improvement. 5 In another study, quantitative and visual EEG features were assessed to compare the electrophysiological changes after clinically ineffective versus effective ECT using a broad dosing range for right unilateral ECT or high-dose bilateral ECT. 50 In this study, only two frontal EEG electrodes were used, and although EEG features could not distinguish responders from nonresponders, the greater ictal power, delta coherence, and postictal suppression modestly predicted the treatment efficacy. It was suggested that these EEG features may reflect the individual differences in the strength of inhibitory processes that terminate the seizure, which may also explain the interindividual differences in the therapeutic efficacy of ECT. Finally, in a longitudinal ECT study, a group of patients were followed up to 26 years from the first course of ECT treatment. In this study, rather inconsistent with other studies, the baseline EEG assessment had revealed no EEG abnormities. However, after an immediate course of ECT, relative delta and theta power were increased prominently in the precentral regions; 7 to 41 months after ECT, a significant decrease in relative delta, theta, and alpha power were observed compared to before treatment; and finally, years after the first ECT treatment, only 2 of 10 patients showed pathological EEG. 54 Collectively, these studies suggest that modulation of slow oscillatory activities (delta to alpha oscillations) is likely one mechanism by which ECT exerts its therapeutic efficacy.

Electroconvulsive therapy is also shown to modify the frontal midline theta oscillations. Previous imaging studies using positron emission tomography have revealed region-specific changes in blood flow during and immediately after ECT such as a decrease in the subgenual ACC and medial prefrontal cortex and an increase in thalamus. 55 Interestingly, impairment in ACC is among one of the most studied topics in depression research and in understanding the therapeutic efficacy of antidepressants. As examples, pre-ECT theta cordance in the central electrodes overlaying the cingulate cortex was strongly associated with decrease in depression scores over the course of the ECT treatment. 56 A recent ECT study in patients with psychotic depression 57 revealed a frequency-specific modification of subgenual ACC theta activity 2 to 3 weeks after a course of ECT treatment. Furthermore, this study found a positive association between increased subgenual ACC theta activity and decreased psychotic symptoms. In addition, low subgenual ACC theta activity before ECT treatment was shown to predict antipsychotic response to ECT. 57 In an earlier study, higher baseline theta activity within rostal ACC was reported to predict antidepressant response to ECT. 31 Given the heterogeneity of major depression and differences across studies with regard to brain regions affected by ECT treatment, it is possible that ECT interacts with the underlying pathological processes and exerts differential but region-specific normalization of aberrant activities (such as within ACC) across patient populations.

The cellular neurobiology of depression has also been in part related to changes in neuroplasticity mechanisms. 58 In a recent study, Transcranial magnetic stimulation combined with electroencephalography (TMS-EEG) method was used to examine the modulation of the frontal cortex excitability in response to a course of ECT treatment in patients with severe MDD. 20 Transcranial magnetic stimulation EEG was applied to middle-caudal portion of the superior frontal gyrus, and it was reported that the early TMS evoked potentials that likely reflect the integrity of cortical excitability and glutamatergic activation, was enhanced posttreatment in all 8 tested patients. Furthermore, a nonsignificant but trending associating was also identified between ECT-related changes in cortical excitability and clinical improvement. It was, therefore, proposed that ECT may exert its therapeutic efficacy through promoting synaptic potentiation, neuroplasticity and growth of neurons and glial cells, which was argued may lead to amelioration of impairments in neuroplasticity and deficient BDNF expression in the frontal cortex and likely hippocampus in MDD. 20

In addition to regional changes, ECT is reported to modify the brain functional connectivity, more likely toward inducing a reduction in connectivity between specific network nodes. In one of the early studies, Roemer et al reported that a higher pre-ECT frontal interhemispheric coherence predicted better clinical response as compared to patients with less frontal coherence. 59 In addition, a recent study revealed an ECT-related decrease in the complexity of the resting-state EEG using the MSE feature, 48 which as previously discussed, has been shown to be abnormally elevated in patients with depression. Okazaki et al (2013) examined neural complexity in 3 subjects before, during, and 5 days after last treatment. They reported all 3 subjects to have shown a decrease in EEG complexity associated with immediate ECT treatments and then tended to begin to revert after the cessation of treatments. In addition to MSE, fractal dimension is another nonlinear EEG feature that provides a statistical index of complexity. It was shown that a smaller fractal dimension of postseizure EEG in the first ECT treatment session could predict the clinical response at 2 weeks of treatment. 60 The largest Lyapunov exponent is another nonlinear EEG feature that was shown to be smaller in ECT responders. 61 Moreover, abnormal glutamatergic modulation of resting-state functional connectivity (using fMRI) has been observed between pregenual ACC and anterior insula in patients with depression, 62 although it remains unclear whether or not ECT acts on this aberrant connectivity and whether this dynamic could be captured by EEG. Future longitudinal (ie, pre/post-ECT) EEG studies should further investigate the ECT-related changes in functional connectivity.

Finally, it should be noted that in addition to providing mechanistic insights or predictors of treatment response, EEG can further be used to identify markers of ECT-related cognitive adverse effects as has been explored by Nobler and Sackeim 63 and Sackeim et al. 64 In these studies, findings were generally consistent with previous reports that post-ECT disorientation was accompanied by increases in delta power in the anterior frontotemporal regions. Furthermore, a greater increase in the global delta relative to theta power was associated with post-ECT decrease in global cognitive states. Furthermore, increases in the frontotemporal theta oscillations were associated with magnitude of retrograde amnesia for autobiographical events. These findings suggest that there might be either an overlap or an effect demarcation zone between mechanisms that mediate ECT-related therapeutic effects and those that give rise to cognitive adverse effects.

Back to Top | Article Outline


In summary, the qEEG and vEEG literature have revealed important insights into the mechanism of action of ECT treatment and introduced several clues in the search for the robust predictors of ECT response. Current electrophysiological evidence further confirms and refines some of the existing hypotheses of ECT mechanisms of action such as, for example, anticonvulsant, 65 neurotrophic, 66 and hyperconnectivity 67 hypotheses. Furthermore, the evidence also suggests there can be improved linkage and unification of existing hypotheses as briefly discussed next (Table 1).

Table 1
Table 1
Image Tools

One of the earlier hypotheses of ECT therapeutic action is the anticonvulsant hypothesis. 68 In this model, it is proposed that the consecutive increase in EEG seizure threshold across the course of ECT treatment is associated with ECT therapeutic effect. Further electrophysiological and neurobiological support for this hypothesis are the reported correlation between postictal suppression and clinical response, reduced blood flow in frontal regions, the lowered metabolic state after ECT and increases in postictal slow wave activities, most of which are likely mediated by ECT-induced activation of inhibitory gamma-aminobutyric acid (GABA)ergic mechanisms. In this regard, it is suggested that the postictal inhibitory response to the seizure, rather than the seizure itself is therapeutic. 68 The same group subsequently suggested that the presence of postictal EEG suppression and prefrontal slowing in effective ECT are suggestive of relative enhancement of prefrontal inhibition being involved in the mechanism of action of ECT. 5 Finally, a TMS study illustrated the potentiation of neurophysiological markers of GABAergic-mediated inhibitory neurotransmission in the motor cortex after ECT. 74 The proposition that ECT may potentiate the prefrontal inhibitory mechanism opens a major door that allows for the finding of a common mechanism of action between ECT and rTMS where no seizures are produced. 17

An alternative ECT hypothesis is the neurotrophic hypothesis that posits that ECT promotes neurogenesis and cell proliferation. 66 In this case, ECT-related increases in the expression of BDNF, neuropeptide molecules, and transcription factors in structures such as hippocampus are considered to underlie the ECT therapeutic mechanism. 69,70 The neurotropic hypothesis may in fact be consistent with the implication of neuroplasticity mechanisms in the pathophysiology of MDD and results of a recent TMS-EEG study that showed an ECT-related potentiation of early TMS evoked potentials in the prefrontal cortex of patients with depression, 20 as discussed previously.

Another hypothesis is the hyperconnectivity hypothesis that proposes that ECT exerts its therapeutic efficacy through modification of the aberrant functional connectivity. 67 Previous studies have reported the hyperconnectivity of limbic and cognitive networks in MDD. 71 Functional MRI studies have also revealed reduction in global functional connectivity in the DLPFC (ie, a measure of connectivity of DLPFC to other brain regions) in response to ECT. 67 It has further been suggested that the ECT-induced deceleration of neural activity in the frontotemporal regions may reflect alteration of the functional connectivity and is likely strongly related to the therapeutic effects of ECT. 70

Back to Top | Article Outline
Connectivity Resetting Hypothesis

In line with the aforementioned hypothesis, we propose a connectivity resetting hypothesis. The connectivity resetting hypothesis postulates that ECT exerts its therapeutic efficacy through resetting aberrant neural connectivity, likely mediated through activating thalamocortical pathways and central inhibitory mechanisms, and increasing the possibility of formation of newer and healthier connection by promoting neurogenesis. This hypothesis is formulated by integrating several observations. First, the prime function of the electrical stimulus used with ECT is to induce a generalized major motor seizure. The occurrence of the generalized seizure is almost instantaneous despite the fact that the stimuli are delivered focally either bilaterally or unilaterally. This suggests that the centrencephalic mechanisms like thalamocortical pathways are instantly engaged.

Second, as discussed, ECT is shown to modify cortical oscillatory activities in particular delta, theta, and alpha oscillations. These findings as well as the ECT-related increase in regional blood flow in thalamus and reduction in medical prefrontal cortex and ACC may further provide evidence for the role of thalamus in ECT mechanism of action. 55 Thalamus is considered as the main brain oscillation pacemaker, and thus could be an important mediator of any effects on EEG oscillations. Cortical oscillatory activities arise from a complex combination of mechanisms such as intrinsic properties of neurons and presence and differential properties of various voltage-gated ion channels in brain tissues but importantly also as a result of short- and long-range interactions between cortical, subcortical, and cerebellar structures including the activity of the corticothalamocortical loop. Several lines of evidence suggest that thalamus and thalamocortical pathways play a major role in generation and modulation of cortical oscillations. Whereas the wake alpha activity seems to be generated mainly in the cortex, it is suggested that thalamic nuclei and thalamocortical pathways play a role in generation and modulation of alpha oscillations.72,73 Higher-frequency cortical rhythms can be generated either in the cortex or the thalamus and represent coherent activity in the corticothalamocortical networks. Delta waves can be generated in either cortex or thalamus, but very slow waves (<1 Hz) are likely generated in the thalamus. 75 Furthermore, thalamus through its connectivity to ACC likely also plays a major role in modulation of ACC theta oscillations that are repeatedly reported in MDD literature and whose modulation has been reported in response to ECT treatment. Therefore, impairment of oscillatory activities in MDD and the ECT-related restoration of oscillatory activities may in fact reflect modifications of short- or long-range pathways likely mediated by thalamic activation.

Third, the combination of greater ictal coherence, greater morphological regularity, and lower postictal coherence as correlates of favorable response suggest that it is important for the stimulus to be strong enough to take over the thalamic pacemaking function and possibly resetting it. One could invoke a tendency of aberrant systems to revert to their normal functional state when a stimulus allows this to occur.

Finally, the resetting hypothesis may be supported by the commonly reported impairment in the functional connectivity in patients with MDD and the accumulating evidence for ECT-related alternation of functional connectivity after ECT such as the observed reduction in neural complexity by Okazaki et al (2013). In future studies, this hypothesis should be assessed by examining effective versus ineffective (but seizure-producing ECT) on restoring functional connectivity in MDD. 21

Back to Top | Article Outline
Added Value of EEG Markers in ECT Research

There are several key advantages in using electrophysiological markers in understanding the pathophysiological characteristics of heterogeneous illnesses such as MDD and investigating the effect of ECT treatment. First, electrophysiological markers are intermediate to genes and behavior. Therefore, in illnesses that arise from a complex interaction between several genes and environmental factor, identification of intermediate markers will further assist in discovering the underlying genes. For example, as previously discussed, a recent study has revealed that the BDNF Val66Met polymorphism may be indirectly associated with depression severity by mediating the EEG alpha oscillations in eyes-closed resting state. 28 Future ECT studies with linkage of electrophysiological and genetic assessments are warranted. 76 Second, electrophysiological markers closely reflect the electrical activities of neural tissue and thus the function of the central nervous system. This is while genetic markers are not necessarily brain specific, for instance, if acquired through periphery using blood or saliva samples. Third, electrophysiological markers are direct recordings from neural activity at high temporal resolution, which is not possible to quantify with other neuroimaging modalities such as fMRI. Finally, the recent advances in the portability and lower cost of current EEG technology have increased the translational value of EEG markers. Recently, qEEG has been FDA approved for being used in combination with clinical assessment to assist the diagnosis of attention-deficit/hyperactivity disorder. 77 As discussed next, future research studies should further explore the use of qEEG for diagnosis of subtypes of MDD and differentiating responders and nonresponders to ECT treatment.

Back to Top | Article Outline
Limitations of EEG Markers and Recommendation for Future Direction

Several limitations should also be carefully considered related to both EEG markers and use of EEG in ECT research. An inherent shortcoming of EEG in general is its low spatial resolution compared to other modalities such as fMRI. Although various source localization methods can be used to identify subcortical sources of EEG signals, these methods are subjected to several a priori assumptions that may not be necessarily valid, and further EEG signals predominantly reflect the activity of the cortical tissues. Future multimodal studies may consider combining EEG recording with concurrent fMRI, magnetoencephalography, electrocorticogram, or recordings with deep brain-implanted electrodes to further verify the sources of EEG markers. Furthermore, TMS-EEG is another powerful multimodal tool that should be further used in future ECT research to more systematically assess ECT-related changes to cortical inhibition 78,79 and neuroplasticity 80 in prefrontal cortex, and cortical and interhemispheric connectivity 81 and integrity of thalamocortical circuitry. 82 In addition to these shortcomings, the ECT research would benefit from further studies examining the application of qEEG in differentiating responders from nonresponders or to identify the most effective stimulus properties, as the efficacy of ECT treatment is suggested to be influenced by stimulus intensity, pulse width, and electrode placement. 83

Moreover, comprehensive longitudinal studies are needed to examine closely the ECT-related modification of EEG features over time as a function of change in clinical symptoms. In years ahead, data mining of large samples of heterogeneous MDD and normative control databases are needed to accelerate further the discovery of unknown patterns and associations that could guide the classification of MDD subtypes and prediction of antidepressant response. Parallel to this, validation of existing EEG features and signal processing algorithms are needed to confirm the neurophysiological mechanisms that each feature subserves. In particular, special attention should be paid to confirm the technical validity and test-retest reliability of existing EEG features of functional connectivity and also to design novel EEG analysis algorithms that robustly capture the brain multiscale network interaction. Such attempts to enhance data mining and algorithm development should be followed by implementation of machine learning systems that would learn from the previously derived patterns to predict diagnosis, distinguish responders from nonresponders, and become progressively more accurate in their predication as more data are added to the system. This guided data-driven approach would overcome some of the shortcomings of the hypothesis-driven approaches that are mainly concerned with examination of a priori hypothesized feature (eg, alpha power), which may leave several clinically relevant EEG features undetected. Currently, a handful of studies have begun to apply machine learning algorithms to distinguish patients with MDD from healthy subjects or to predict response to antidepressants (eg, Hosseinifard 84 and Khodayari-Rostamabad 85 ).

Back to Top | Article Outline


Collectively, it seems plausible that in MDD genetic factors (eg, BDNF Val66Met polymorphism) and stress-inducing environmental factors interact in a complex manner to lead to alteration of hypothalamus-pituitary-adrenal stress axis, 86 inhibitory, and neuroplasticity mechanisms that combined likely result in induction of aberrant functional connectivity within an extended neural network. 36 The wealth of electrophysiological studies to date collectivity suggests that ECT likely ameliorates these symptoms through a combination of mechanisms including the resetting of oscillatory activities, perhaps by acting on the thalamic pacemakers, and furthermore by engaging the activation of inhibitory GABAergic mechanisms that could further facilitate dissociating the aberrant connectivity between different near and remote brain regions (eg, perhaps involving DLPFC, ACC and hippocampus). This is while after ECT, the engagement of neurotrophic mechanisms such as increases in BDNF and cell proliferation in the hippocampus and frontal regions promotes formation of new and healthy network connectivity.

However, given the heterogeneity of findings such as variability in the baseline abnormalities and implicated neural pathways, and ECT-related EEG modifications across studies and patient populations, it is unlikely that one mechanism can explain how ECT works in all subjects. However, this variability may be in part explained by the proposed connectivity resetting hypothesis, that is, ECT likely resets connectivity in an extended network and through resetting the thalamic rhythm it weakens or inhibits any aberrant functional connectivity. This may seem as if ECT has differential therapeutic mechanisms that vary as a function of the underlying pathology. In the future studies, assessment of the EEG and TMS-EEG indices of functional connectivity (and resting-state fMRI) before and after ECT compared with healthy subjects would help ascertain this issue. In addition, given that potentiation of inhibitory mechanisms seems to be one of the most dominant features of ECT and likely clinically relevant (eg, anticonvulsant hypothesis), future studies should ascertain the functional role of activation of inhibitory mechanisms in mediating or maintaining the ECT-induced changes in the brain functional connectivity. Finally, more extensive research studies and larger data sets are needed to improve the accuracy and sensitivity of data mining methods, design and validate analysis algorithms, and assess the predictive accuracy of learning algorithms in untested data sets. The identification of robust and reliable EEG correlates of ECT response will provide invaluable insight for designing the future generation of treatments that have optimal therapeutic efficacy and fewer adverse effects.

Back to Top | Article Outline


1. , , . ECT of schizophrenia. ECT of schizophrenia. Lancet . 1939;12:1353–1355.

2. . Electroencephalographic Correlates of ECT. Psychopharmacol Bull 18. 1982;2:78–81.

3. . New developments in electroconvulsive therapy. Dis Nerv Syst 1976;6:356–358.

4. . EEG related to electroconvulsive therapy. In: , , eds. EEG and Evoked Potentials in Psychiatry and Behavioral Neurology . Boston: Butterworth; 1983.

5. , , , The effects of electroconvulsive therapy on quantitative electroencephalograms. Relationship to clinical outcome. Arch Gen Psychiatry 53. 1996;9:814–824.

6. , , , Burst suppression: a more valid marker of postictal central inhibition? J ECT . 2013;1:25–28.

7. , , , The electroencephalographic pattern during electroconvulsive therapy. III. Analysis of frontotemporal and nasopharyngeal spectral energy. Clin Electroencephalogr . 1986;2:66–77.

8. , , . Regulation of BDNF and TRKB mRNA in rat brain by chronic electroconvulsive seizure and antidepressant drug treatments. J Neurosci . 1995;11:7539–7547.

9. , , , Rapid increase of BDNF MRNA levels in cortical neurons following spreading depression: regulation by glutamatergic mechanisms independent of seizure activity. Brain Res Mol Brain Res . 1993;4:277–286.

10. , . EEG monitoring of ECT seizure. In: . The Clinical Science of Electroconvulsive Therapy . Washington DC: American Psychiatric Press; 1993.

11. , , , The effects of ECT stimulus dose and electrode placement on the ictal electroencephalogram: xan intraindividual crossover study. Biol Psychiatry . 34. 1993;11:759–767.

12. , , , Is postictal electrical silence a predictor of response to electroconvulsive therapy? J Affect Disord 41. 1996;1:55–58.

13. , , . Stimulus intensity, seizure threshold, and seizure duration: impact on the efficacy and safety of electroconvulsive therapy. Psychiatr Clin North Am 14. 1991;4:803–843.

14. , , , Effect of ECT treatment number on the ictal EEG. Psychiatry Res . 1996;2:179–189.

15. , , , Changes in seizure threshold over the course of electroconvulsive therapy affect therapeutic response and are detected by ictal EEG ratings. J Neuropsychiatry Clin Neurosci . 1998;2:178–186.

16. . Kindling and sensitization as models for affective episode recurrence, cyclicity, and tolerance phenomena. Neurosci Biobehav Rev 31. 2007;6:858–873.

17. , , , EEG effects of ECT: implications for RTMS. Depress Anxiety . 2000;3:157–165.

18. . Beyond seizure duration as a measure of treatment quality. Convuls Ther 9. 1993;1:1–7.

19. , , . The ictal EEG as a marker of adequate stimulus intensity with unilateral ECT. J Neuropsychiatry Clin Neurosci . 1995;3:295–303.

20. , , , Transcranial magnetic stimulation-evoked EEG/cortical potentials in physiological and pathological aging. Neuroreport . 2011;12:592–597.

21. , . EEG biomarkers in major depressive disorder: discriminative power and prediction of treatment response. Int Rev Psychiatry 25. 2013;5:604–618.

22. . Fourier-Analyse Von Elektrencephalogrammen Des Menschen. Pflüger’s Archiv für die gesamte Physiologie des Menschen und der Tiere . 1932;1:106–112.

23. . Fourier-Analyse Von Elektrencephalogrammen Des Menschen. Pfluger Archiv Physiologie . 1932:106–112.

24. , , , EEG power spectra at early stages of depressive disorders. J Clin Neurophysiol . 2009;6:401–406.

25. , . Electrocortical features of depression and their clinical utility in assessing antidepressant treatment outcome. Can J Psychiatry . 58. 2013;9:509–514.

26. , , , Interrelation between changes in the EEG and psychopathology under pharmacotherapy for endogenous depression. A contribution to the predictor question. Pharmacopsychiatry 17. 1984;6:178–183.

27. , , , Electroencephalographic alpha measures predict therapeutic response to a selective serotonin reuptake inhibitor antidepressant: pre- and post-treatment findings. Biol Psychiatry . 2008;12:1171–1177.

28. , , , EEG alpha power as an intermediate measure between brain-derived neurotrophic factor Val66Met and depression severity in patients with major depressive disorder. J Clin Neurophysiol 30. 2013;3:261–267.

29. , , , Alpha power, alpha asymmetry and anterior cingulate cortex activity in depressed males and females. J Psychiatr Res . 2012;11:1483–1491.

30. , , , Neurophysiological predictors of non-response to rTMS in depression. Brain Stimul 2012;4:569–576.

31. , , , Anterior cingulate activity as a predictor of degree of treatment response in major depression: evidence from brain electrical tomography analysis. Am J Psychiatry 158. 2001;3:405–415.

32. , , . Direct recording of theta oscillations in primate prefrontal and anterior cingulate cortices. J Neurophysiol 95. 2006;5:2987–3000.

33. , , , EEG power, frequency, asymmetry and coherence in male depression. Psychiatry Res . 2001;2:123–140.

34. , , , Relationship between brain electrical activity and cortical perfusion in normal subjects. Psychiatry Res . 1999;2:125–140.

35. , , , Neurophysiologic predictors of treatment response to fluoxetine in major depression. Psychiatry Res . 1999;3:263–273.

36. , . Neural circuits underlying the pathophysiology of mood disorders. Trends Cogn Sci 16. 2012;1:61–71.

37. , , , Impaired functional connectivity at EEG alpha and theta frequency bands in major depression. Hum Brain Mapp . 2007;3:247–261.

38. , , , Resting-state quantitative electroencephalography reveals increased neurophysiologic connectivity in depression. PLoS One . 2012;2:e32508.

39. , , , The relationship between brain oscillatory activity and therapeutic effectiveness of transcranial magnetic stimulation in the treatment of major depressive disorder. Front Hum Neurosci . 2013;37.

40. , , , Determination of information flow direction among brain structures by a modified directed transfer function (DDTF) method. J Neurosci Methods . 2003;1–2:195–207.

41. . Investigating causal relations by econometric modelsand cross-spectral methods. Econometrica . 1969:424–438.

42. , . Partial directed coherence: a new concept in neural structure determination. Biol Cybern . 2001;6:463–474.

43. . Nonlinear dynamical analysis of EEG and MEG: review of an emerging field. Clin Neurophysiol 116. 2005;10:2266–2301.

44. , . Graph theoretical analysis of complex networks in the brain. Nonlinear Biomed Phys 1. 2007;1:3.

45. , , , Cortical mechanisms of the symptomatology in major depressive disorder: a resting EEG study. J Affect Disord . 2011;1–3:243–250.

46. , , , A critical assessment of connectivity measures for EEG data: a simulation study. Neuroimage . 2013:120–133.

47. , , . Multiscale entropy analysis of biological signals. Phys Rev E Stat Nonlin Soft Matter Phys . 2005;(2 Pt 1) 71:021906.

48. , , , Effects of electroconvulsive therapy on neural complexity in patients with depression: report of three cases. J Affect Disord 150. 2013;2:389–392.

49. , , , EEG immediately after unilateral ECT. Acta Psychiatr Scand . 1978;3:231–244.

50. , , , Seizure expression during electroconvulsive therapy: relationships with clinical outcome and cognitive side effects. Neuropsychopharmacology 29. 2004;4:813–825.

51. , , . Electrophysiology of ECT. In: , , , eds. Psychopharmacology: A Generation of Progress . New York: Raven Press; 1978.

52. , , , Standard Electroencephalography in Clinical Psychiatry: A Practical Handbook . John Wiley & Sons; 2011.

53. , . The significance of abnormal electroencephalograms prior to electroconvulsive therapy. J Nerv Ment Dis 107. 1948;3:220–227.

54. , , , Long-term follow-up in depressed patients treated with electroconvulsive therapy. J ECT . 21. 2005;4:214–220.

55. , , , Changes in regional cerebral blood flow during acute electroconvulsive therapy in patients with depression: positron emission tomographic study. Br J Psychiatry 190. 2007:63–68.

56. , , , Pretreatment neurophysiologic function and ECT response in depression. J ECT 20. 2004;3:142–144.

57. , , , Antipsychotic effect of electroconvulsive therapy is related to normalization of subgenual cingulate theta activity in psychotic depression. J Psychiatr Res . 2009;5:553–560.

58. , , . The cellular neurobiology of depression. Nat Med . 2001;5:541–547.

59. , , , Relationship between pretreatment electroencephalographic coherence measures and subsequent response to electroconvulsive therapy: a preliminary study. Neuropsychobiology 24. 1990;3:121–124.

60. , , , Post-seizure eeg fractal dimension of first ECT predicts antidepressant response at two weeks. J Affect Disord . 1999;1-3:235–238.

61. , , , The largest lyapunov exponent of the EEG during ECT seizures as a measure of ECT seizure adequacy. Electroencephalogr Clin Neurophysiol . 1997;6:599–606.

62. , , , Glutamatergic and resting-state functional connectivity correlates of severity in major depression - the role of pregenual anterior cingulate cortex and anterior insula. Front Syst Neurosci . 2010;4:33.

63. , . Neurobiological correlates of the cognitive side effects of electroconvulsive therapy. J ECT . 24. 2008;1:40–45.

64. , , , Electrophysiological correlates of the adverse cognitive effects of electroconvulsive therapy. J ECT 16. 2000;2:110–120.

65. . The anticonvulsant hypothesis of the mechanisms of action of ECT: current status. J ECT 15. 1999;1:5–26.

66. . Exploration of new molecular mechanisms for antidepressant actions of electroconvulsive seizure. Biol Pharm Bull 34. 2011;7:939–944.

67. , , , Electroconvulsive therapy reduces frontal cortical connectivity in severe depressive disorder. Proc Natl Acad Sci U S A 109. 2012;14:5464–5468.

68. . Central issues regarding the mechanisms of action of electroconvulsive therapy: directions for future research. Psychopharmacol Bull 30. 1994;3:281–308.

69. , , , A review of longitudinal electroconvulsive therapy: neuroimaging investigations. J Geriatr Psychiatry Neurol . 2013;27:33–46.

70. , . Electroconvulsive treatment: hypotheses about mechanisms of action. Front Psychiatry . 2013;4:94.

71. , , , Resting-state functional mri in depression unmasks increased connectivity between networks via the dorsal nexus. Proc Natl Acad Sci U S A 107. 2010;24:11020–11025.

72. , . Thalamo-cortical mechanisms underlying changes in amplitude and frequency of human alpha oscillations. Neuroimage . 2013:150–163.

73. , , , Prediction of electroencephalographic spectra from neurophysiology. Phys Rev E Stat Nonlin Soft Matter Phys 63. 2001;2 (Pt 1):021903.

74. , , , Effects of right unilateral electroconvulsive therapy on motor cortical excitability in depressive patients. J Psychiatr Res . 2006;4:322–327.

75. , , , The slow (< 1 Hz) oscillation in reticular thalamic and thalamocortical neurons: scenario of sleep rhythm generation in interacting thalamic and neocortical networks. J Neurosci 13. 1993;8:3284–3299.

76. , , . Major depressive disorder: new clinical, neurobiological, and treatment perspectives. Lancet . 379. 2012;9820:1045–1055.

77. , , , Characterization of the theta to beta ratio in ADHD: identifying potential sources of heterogeneity. J Atten Disord . 2013;5:384–392.

78. , , , The EEG correlates of the TMS-induced EMG silent period in humans. Neuroimage . 2013:120–134.

79. , , , Reliability of long-interval cortical inhibition in healthy human subjects: a TMS-EEG study. J Neurophysiol . 2010;3:1339–1346.

80. , , , PAS-induced potentiation of cortical-evoked activity in the dorsolateral prefrontal cortex. Neuropsychopharmacology 38. 2013;12:2545–2552.

81. , , , The role of the corpus callosum in transcranial magnetic stimulation induced interhemispheric signal propagation. Biol Psychiatry 68. 2010;9:825–831.

82. , , , Natural frequencies of human corticothalamic circuits. J Neurosci 29. 2009;24:7679–7685.

83. , , , Effects of pulse width and electrode placement on the efficacy and cognitive effects of electroconvulsive therapy. Brain Stimul 1. 2008;2:71–83.

84. , , . Classifying depression patients and normal subjects using machine learning techniques and nonlinear features from EEG signal. Comput Methods Programs Biomed . 2013;3:339–345.

85. , , , Diagnosis of psychiatric disorders using EEG data and employing a statistical decision model. Conf Proc IEEE Eng Med Biol Soc . 2010;2010:4006–4009.

86. , , , HPA axis normalization, estimated by DEX/CRH test, but less alteration on cerebral glucose metabolism in depressed patients receiving ECT after medication treatment failures. Acta Psychiatr Scand 112. 2005;4:257–265.


electroconvulsive therapy; electroencephalography; major depressive disorder; functional connectivity; neuroplasticity; inhibition

© 2014 by Lippincott Williams & Wilkins


Article Tools



Search for Similar Articles
You may search for similar articles that contain these same keywords or you may modify the keyword list to augment your search.