Electroconvulsive Therapy: Mechanisms of Action, Clinical Considerations, and Future Directions

Abstract

Electroconvulsive therapy (ECT) is the most effective treatment for a variety of psychiatric illnesses, including treatment-resistant depression, bipolar depression, mania, catatonia, and clozapine-resistant schizophrenia. ECT is a medical and psychiatric procedure whereby electrical current is delivered to the brain under general anesthesia to induce a generalized seizure. ECT has evolved a great deal since the 1930’s. Though it has been optimized for safety and to reduce adverse effects on cognition, issues persist. There is a need to understand fundamental physiologic, cellular, and molecular mechanisms of ECT to devise strategies to optimize therapeutic outcomes. Clinical trials that set out to adjust parameters, electrode placement, adjunctive medications and patient selection are critical steps towards the goal of improving outcomes with ECT. This narrative review provides an overview of ECT, its efficacy in treating depression, the known effects on cognition, evidence of mechanisms, and future directions.

Introduction:

Electroconvulsive therapy (ECT) is a medical procedure whereby electrical current is administered through the skull under general anesthesia to induce a seizure for the treatment of a variety of psychiatric illnesses ^1–4. For over 80 years, ECT has been the most effective treatment for severe depression, mania, catatonia and clozapine-resistant schizophrenia. It has also been shown to help with suicidal ideation and in augmenting clozapine for psychosis ^5–9. There is an abundance of evidence supporting the use of ECT for treatment-resistant depression (TRD), and overall, ECT is more effective than pharmacotherapy for the treatment of depression and other mood disorders, such as bipolar depression and mania ⁴. Regardless, given the stigma of its use and adverse effects on memory, it is widely underutilized ^2,4,10. A contributing factor to why patients and their families may be reluctant to undergo ECT, or clinicians may be reluctant to refer for treatment, is the impression that little is known about the mechanisms of the therapeutic effect and cognitive side effects ^10,11. While many questions remain open, neuroscience research has made significant advances to understand the neurobiological effects of ECT. This narrative review aims to describe the evidence supporting current models that explain how ECT works therapeutically as well as what causes the cognitive side-effects.

ECT – an ultrabrief history:

Electroconvulsive therapy has evolved tremendously since it was first used in the 1930’s ^12,13. What began as bilateral stimulation with a sine wave voltage source at the power line frequency without anesthesia (referred to as “unmodified ECT”), has evolved to its present form of unilateral (typically right unilateral, RUL) stimulation with individually-titrated trains of ultrabrief or brief rectangular constant-current pulses under general anesthesia (termed “modified ECT”, i.e. modified with anesthesia), and bilateral brief pulse stimulation generally reserved for refractory or emergent cases ^14–17. The impetus for these optimizations has been to prevent physical and cognitive side effects. Despite the robust antidepressant effect of ECT, some patients still experience adverse effects on memory, which limits its clinical application.

ECT is a rapid-acting antidepressant:

Electroconvulsive therapy begins with a short index course of 6 treatments delivered 3/week, or 2/week as performed in the UK. Patients typically respond to RUL treatment with > 50% reduction in their symptoms within that two-week period and then continue with weekly maintenance treatment for a few weeks and gradually increasing the interval between maintenance treatments as clinically indicated. If there is little or suboptimal response in 2–3 weeks, particularly if there is concern for physical safety, practitioners often switch to brief-pulse RUL (change in pulse width) or bilateral treatment (change in montage), which is understood to be associated with more noticeable cognitive side effects, as we explain in the following sections. A recent epidemiological study of Medicare data compared 10,460 patients who underwent ECT to 31,160 patient-controls and showed that patients receiving ECT had lower all-cause mortality for up to 1 year following hospital discharge. Importantly, there was a significant reduction in the rate of suicide following ECT, but with no difference at 1 year follow-up ¹⁸.

Antidepressant efficacy and cognition:

Luccarelli, Henry, and colleagues have recently reviewed large datasets from Massachusetts General and McLean Hospitals and made some pertinent clinical observations. They have shown that ECT was associated with a rapid decrease in depression symptoms within the first 10 treatments; which continued for almost two years of maintenance treatment. Importantly, there was no change in cognitive functioning as measured by the Montreal Cognitive Assessment (MoCA). At their institution, regardless of the initial treatment, bilateral and brief pulse treatment parameters were more commonly used by treatment number 50, and most participants either stopped due to remitting or had continued in ECT, implying side-effects were tolerable enough on average at the end of the study period ¹⁹. Another analysis showed that 37.5% of patients who began with RUL were transitioned to bilateral treatments ²⁰. Specifically, younger patients and those with bipolar disorder were more likely to transition from brief pulse RUL to bilateral treatments, but the overall number of treatments did not differ based on age or primary diagnosis ²¹.

In a study where cognition was assessed, ECT was associated with a 40% decrease in depression symptoms (QIDS reduction from 17.1 ± 4.9 to 10.1 ± 5.2), accompanied by a reduction in MoCA scores that was statistically significant, but likely too small to be clinically significant (from 25.8 ± 3.1 to 25.4 ± 3.1); interestingly, older age was associated with a milder reduction in MoCA perhaps due to lower baseline at the initiation of treatment ²². Furthermore, in a study of 684 patients that received at least 10 ECT treatments, investigators tracked QIDS and assessed cognition with the MoCA at baseline and after treatment. Patients with baseline normal cognition (MoCA ≥ 26; N = 371) had a decrease in MoCA of 1.44±0.26 points over the course of treatment, while those with baseline impaired global cognitive function (MoCA < 26; N = 313) experienced an increase in MoCA of 1.72±0.25 points. Importantly, baseline global cognitive function was not associated with a change in depression after ECT, suggesting that impaired global cognitive function may not be a contraindication to ECT in geriatric patients ²³. It is important to note that the MoCA may not be the most appropriate assessment tool for the type of cognitive complaints patients experience with ECT (nuances discussed below) as it is a general multidomain screening test that measures many different cognitive functions and emphasizes executive function over memory. That said, it is reassuring that there is very little change in an assessment tool that quickly screens a broad range of cognitive domains, which is particularly useful in older patients.

Electrical stimulus parameters:

The characteristics of the electrical stimulus influence the therapeutic response as well as the emergence of side effects. A crucial component to the administration of ECT is electrode placement, which determines the path of current propagation through the tissue ^24,25. There are three commonly used electrode placements: RUL, bitemporal, and bifrontal; the latter two are types of bilateral placement. Historically, as mentioned above, ECT was administered bilaterally with electrodes placed on the side of the head over the temporal lobes (bitemporal placement). Because of adverse effects of having current pass directly through the left temporal lobe where language is generated for most people, practitioners typically favor beginning treatment with a RUL configuration (right temporal and midline vertex placement of electrodes). If a patient is left-handed, they may have a right-dominant hemisphere (as approximately 25% of left-handed individuals do), and steps can be taken to determine whether the electrodes should be placed left-unilaterally to avoid passing current through their dominant hemisphere in order to minimize cognitive side effects ^26,27. Electrodes can also be placed bifrontally across the forehead.

The electrical current parameters also influence the rate of depressive symptom improvement as well as the emergence of side effects. Conventional dosing of ECT depends on many factors, but pulse shape and current amplitude are generally fixed; this leaves changing the pulse width and the pulse train frequency and duration as means of protocol adjustment. The standard current amplitude used in clinical practice of modern ECT is fixed at either 800 mA or 900 mA, depending on the device and various practices; the pulse shape is rectangular with brief (0.5–1 ms) or ultrabrief (0.25–0.3 ms) pulse width; and the pulse train is comprised of monophasic pulses of alternating polarity. With these parameters fixed, the electrical dosage is conventionally individualized by increasing the number of pulses in the stimulus train via titration of the train duration and/or frequency. In ECT practice, the electrical dosage is often quantified by the total charge or energy of the stimulus train ²⁸.

Historically, researchers found that increasing electrical dosage from just above seizure threshold (ST) to 2.5 times the ST increased the efficacy of brief pulse RUL ECT, although not to the level of bilateral ECT ²⁹. Sackeim et al. showed that high-dose RUL ECT (6 × ST) is more effective than lower (1.5 × ST) or moderate (2.5 × ST) RUL ECT ³⁰. Importantly, high-dosage RUL ECT showed similar therapeutic effects to bilateral ECT (bilateral at 2.5 × ST) with less cognitive impairment. Sackeim and colleagues then compared high-dose RUL (6 × ST) to moderate suprathreshold bilateral ECT (2.5 × ST) for both brief and ultrabrief pulses, and concluded that high-dose ultrabrief RUL ECT preserved efficacy while reducing cognitive side effects, whereas ultrabrief bilateral ECT reduced cognitive side effects but was significantly less effective ¹⁵. The authors speculated that at higher dosage relative to ST, ultrabrief bilateral ECT may also increase its efficacy, although this approach has not been tested systematically ¹⁵. Taken together, a clinical approach may be to start with ultrabrief RUL ECT at a high dosage rather than bilateral ECT because of comparable efficacy and fewer cognitive adverse effects.

Sienaert and colleagues found both unilateral and bifrontal ultrabrief ECT to improve cognitive performance with a broad neuropsychological battery at 1 and 6 weeks post-treatment ^31,32. Testing stimuli of 4, 7, and 10 times ST, Quante et al. asserted that stimulus intensity of ultrabrief RUL ECT in this range does not correlate with symptom improvement measured by the Hamilton Depression Rating Scale, demonstrating that ultrabrief RUL ECT at 4 times ST is sufficient for symptom improvement ³³. A 2015 meta-analysis showed that brief versus ultrabrief pulse RUL ECT was significantly more efficacious in treating depression but also showed significantly more side effects in global cognition, anterograde learning and recall, as well as retrograde memory ³⁴.

In clinical practice, initial electrode placement is typically initiated as RUL or, in more urgent or severe cases, bitemporal or bifrontal ³⁰. This is because, generally, bilateral stimulation acts faster with a trade-off of having greater likelihood of cognitive side effects than unilateral ECT (with either ultrabrief or brief pulse width). The UK ECT Review Group found that bilateral ECT is generally more effective than unilateral ECT in improving depression symptoms, based on data from 22 studies comparing bilateral (this meta-analysis combined bifrontal and bitemporal) to unilateral ECT ⁴. However, two rigorous studies found non-inferiority of brief pulse RUL in comparison with brief pulse bifrontal and bitemporal ECT, respectively ^35,36. A randomized control trial by Kellner et al comparing bilateral, bitemporal and RUL was consistent with this, with bitemporal ECT having greater cognitive impairment ³⁷. In a meta-analysis of observational studies of cognitive function after ECT, Semkovska and McLoughlin showed that both unilateral and bitemporal ECT are associated with significant decreases in cognitive performance, mainly executive functioning, and verbal memory, within 3 days post-treatment. Bitemporal ECT had worse cognitive performance acutely, though long-term follow-up demonstrated improvements compared to baseline or reattainment of baseline performance (at 14–365 days post-treatment). Although there are concerns about acute ill effects on cognition with ECT, it is generally accepted that adverse effects on cognition (predominantly memory) will return to normal within six months of the index course of treatment ^4,38. Depending on illness severity, a protocol could begin with bilateral ECT and switch to RUL if cognitive side effects emerge or perhaps the converse—begin with high-dose RUL ECT for index and weekly treatments and then use bilateral for monthly maintenance treatments. If cognitive side effects are of particular concern, high-dose ultrabrief-pulse RUL should be considered over brief-pulse RUL or bilateral ECT.

More recently, a novel approach to electrode placement, focal electrically administered seizure therapy (FEAST), has been explored as a means of reducing the adverse cognitive effects from seizure therapy. FEAST utilizes unidirectional stimulation and an asymmetrical frontal electrode placement with a large posterior electrode (cathode) in front of the right motor cortex and a small anterior electrode (anode) above the center of the right eyebrow, over the right orbitofrontal cortex, with the aim of focusing the seizure induction in the right prefrontal cortex ^39–42.

Another experimental approach aiming to reduce the side effects of ECT is magnetic seizure therapy (MST), which utilizes high intensity and frequency transcranial magnetic stimulation (TMS) under general anesthesia to induce a seizure ⁴³. Compared to conventional ECT, MST induces a weaker and more superficial electric field ^24,44. Like in ECT, optimal MST stimulation targeting and parameters are being investigated. For example, prefrontal stimulation at higher frequencies may increase depression remission rates, whereas lower frequencies appear advantageous for suicidality remission ^45–47.

Computational, animal, and clinical studies have suggested that the electric field generated by standard ECT in the brain is markedly higher than the levels needed to depolarize neurons and induce a seizure ^48,49. The current amplitude is a key factor driving the electric field into deeper brain regions, such as the hippocampus and temporal lobe, which may contribute to adverse effects on cognition and memory. Reducing the current amplitude makes the electric field weaker and more focal ⁵⁰. Therefore, decreasing the stimulus current amplitude could be a potential means of reducing cognitive side effects ²⁸. A recent proof of concept non-randomized clinical trial suggested that ultrabrief RUL ECT with a lower current amplitude (as low as 500 mA to 600 mA) may improve cognitive outcomes by reducing reorientation time, a measure correlated with worse cognitive outcomes from ECT. Intriguingly, there was no difference in treatment response in depressive symptoms when compared to higher current ultrabrief pulse RUL ^51,52. There may be, however, an individual tradeoff between the efficacy of treating depression and preservation of cognition controlled by the current amplitude. Abbott and colleagues investigated ultrabrief RUL ECT delivered at three current amplitudes and found that 700 and 800 mA were more effective at reducing depression symptoms, while 600 mA had less cognitive side effects as measured by the Hopkins Verbal Learning Test and aspects of the Delis Kaplan Executive Function System, which together assess frontal-temporal cognitive functions (e.g., verbal fluency, cognitive flexibility, inhibition, and processing speed) ⁵³. Notably, current strength of 500 mA was not included in this trial based on evidence of insufficient therapeutic efficacy during pilot testing. Additionally, subjects enrolled in the trial were switched from ultrabrief to brief pulse stimulation, due to concerns about low efficacy of the 600 mA arm. Thus, reducing the current amplitude may diminish therapeutic effects, at least for some subjects and ECT parameter settings.

The relationship between ECT current amplitude and clinical outcomes may vary individually due to differences in the electric field delivered to the brain stemming from anatomical variability. The electric field in the brain can be modeled computationally based on individual MRI scans ²⁴. The few studies to date that modeled the ECT electric field in large groups of patients have arrived at intriguing though diverse relationships. The electric field in the left hippocampus and left amygdala were found to correlate with increases of the volume of these structures after RUL ECT, but neither the volume changes nor the electric field strength were associated with antidepressant response ⁵⁴. The latter finding contradicts other studies (often using data from a single or two centers, with smaller sample sizes but greater sample homogeneity), which demonstrated a significant relationship between increases in hippocampal and amygdala volume primarily (but also other limbic structures) and symptom improvement ^55–57. One line of investigation tried to combine genetic differences in hippocampal size after ECT and found that a single nucleotide polymorphism (SNP) in the promoter of vascular endothelial growth factor (VEGF) attenuated hippocampal growth whereas a SNP in the brain derived neurotrophic factor (BDNF) gene did not ⁵⁸.

The study of current amplitude including RUL and bitemporal electrode placement found improved antidepressant outcomes to correlate with volume increases of the right, but not left, hippocampus, which, in turn, was affected by the local electric field. In contrast to this indirect effect, stronger electric field in the right hippocampus worsened the antidepressant effect and cognitive performance ⁵⁹. A similar effect of worse antidepressant effect with a stronger electric field in the temporal lobes was observed for bilateral but not RUL ECT in another study ⁶⁰. Other studies have also observed lateralized effects, in which RUL ECT leads to right-sided volumetric changes while bilateral ECT leads to bilateral volumetric changes ⁶¹. While these limited data do not afford firm conclusions, they suggest adverse effects of direct electric field exposure in the temporal lobes, while also indicating potential beneficial indirect effect on hippocampus, perhaps mediated by the seizure and/or transsynaptic activation.

These results raise the question of the relative contribution to ECT outcomes of the electrical stimulation versus the seizure itself. The strong electrical stimulation delivered by ECT produces synchronized rhythmic firing over broad populations of neurons, akin to the brain activity during a seizure. Thus, it is possible that some of the effects of ECT are conferred or modulated by the electrical stimulus rather than the seizure. This may be a reason why, to be effective, RUL ECT has to be delivered with six times the number of pulses needed to induce a seizure. Moreover, a pilot study of intensive bifrontal subconvulsive stimulation with an ECT device reported significant therapeutic effects in the absence of a seizure ⁶². Finally, accelerated repetitive transcranial TMS, which delivers a large number of subconvulsive stimuli over a short period of time, has recently demonstrated remarkably high remission rates in treatment resistant depression ^63,64. These observations may point to using targeted stimulation to induce seizures while avoiding direct hippocampal stimulation, combined with more stimulation pulses and sessions for cumulative therapeutic effects ⁵⁰. Lower current amplitude produces somewhat less cognitive side effects, but there is substantial individual variability in efficacy versus current amplitude. Therefore, individualization of the current amplitude may be appropriate, by seizure or motor threshold titration in the amplitude domain or individual electric field modeling ^25,59,65. Broadly, ECT parameters could be explored further as pathways for improving outcomes. Optimized and individualized stimulus delivery may become an important option to lower the risk of cognitive side effects, especially in patients at higher risk or with a previous adverse outcome.

Cognition Considerations:

Studying cognition in response to ECT has a few pitfalls. The first is that depression, particularly TRD, has been shown to negatively impact many aspects cognition, mainly attention and aspects of memory ⁶⁶. Patients often report adverse effects of ECT on memory ⁶⁷. These subjective reports have been challenging to characterize ⁶⁸. This can be explained partly because studies of cognition are confounded by a few factors: First, multiple cognitive domains improve after ECT as depression negatively affects attention, executive function, and working memory; Secondly, studies typically compare MoCA, Mini-Mental Status Exam (MMSE) or the Cognitive Failures Questionnaire (CFQ) before and after ECT ⁶⁹. MMSE and MoCA are tools to assess mild to severe cognitive impairment and do not appear to capture the subtleties of subjective cognitive complaints. These screening tests do not examine retrograde episodic memory – the type of memory impairment primarily reported by patients. Certainly, if ECT caused major neurocognitive impairment, we would have stopped administering it decades ago.

The effects of ECT on memory is a major concern for patients. Many studies have evaluated cognition broadly and found non-memory cognition to improve substantially. Importantly, studies at long-term follow-up after ECT generally show improvements in memory using objective measures ⁷⁰. In the subacute period following ECT (0–3 days), tests of executive function and processing speed were impaired compared to baseline ⁷⁰. However, when measured between 4 and 14 days, most tests showed significantly improved outcomes compared with baseline, and none were below baseline ⁷⁰. This improvement is likely to have been a result of successful treatment as studies suggest that a high percentage of patients with severe depression have substantial cognitive impairment at baseline ⁷¹. Longer term, with a wide range of follow-up intervals (2 weeks to 2 years), there was no evidence of impairment in tests of non-memory cognition ⁷⁰.

The effects on memory have been variable and proven more challenging to assess. Studies of subjective report of cognitive side-effects show varying percentages of patients complaining of ‘memory loss’, depending on many factors, including the interviews were conducted ⁶⁷. When patients report cognitive side-effects, it is often loss of retrograde autobiographical memory concerning them, i.e., the loss of detailed memory for events that the person has experienced. However, both episodic memory (experiential memory) and semantic memory (factual memory) may be affected ⁷². Retrograde amnesia (memory loss) for non-autobiographical information has also been shown ⁷³. Furthermore, when objectively measured, this retrograde amnesia could persist for over one year. Objective testing of autobiographical memory, typically using the Columbia University Autobiographical Memory Interview (CUAMI), has shown that ECT memory loss exceeds the normal decay of memories over time in healthy matched participants and the degree of loss differs depending on the treatment ECT parameters used; though importantly, many of these comparisons were not to that of depressed. Importantly, subjective impairment often reduces throughout the course of treatment, correlating with clinical response, and perhaps representing negative bias associated with depressive states ^74,75. Brief assessments of cognitive side effects of ECT were reviewed by Martin et al, and an emerging tool, the Electroconvulsive Therapy Cognitive Assessment (ECCA) is gaining interest, though widespread adoption is lacking, as is thorough validation of it for autobiographical memory loss ^76,77.

Interestingly, ECT does not appear to adversely affect verbal memory long-term, even with bilateral brief pulse, with current passing directly through the left frontotemporal lobe. More appropriate tests would be those that test non-verbal and visuospatial memory, working memory, reasoning and executive function ⁶⁹. It is possible that the brain regions undergoing the most dramatic changes during a course of ECT, i.e., mesial temporal lobes and limbic structures, are the ones responsible for the efficacy of ECT, but also harbor the source of some cognitive changes – be it objective or perceived.

Ketamine has been studied as a potential augmentation strategy co-administered with ECT to alleviate side effects and/or increase efficacy. It is important to note that anesthetic doses of ketamine used for ECT seem to have a different effects on mood when compared to subanesthetic doses used to treat depression ⁷⁸. Importantly, ketamine for the treatment of depression is used at a subanesthetic dose in awake patients, so findings in anesthetized patients with higher doses of ketamine may not share the identical mechanism of action. A recent meta-analysis of add-on ketamine to ECT has shown that ketamine enhances ECT’s therapeutic effects but unfortunately also exacerbates side effects ^79,80. Ketamine is known to inhibit N-methyl-D-aspartate (NMDA) receptors and, interestingly, endogenous NMDA receptor agonist quinolinic acid (QUIN), is decreased following successful ECT – suggesting a role of NMDA receptor biology in ECT’s mechanisms. Ill-effects on cognition resulting from ECT seems to depend on many factors, including a patient’s baseline cognitive status and underlying cognitive reserve prior to obtaining ECT ^81,82. That said, there are parameters of ECT delivery that seem to directly increase the likelihood of adverse cognitive outcomes including bilateral electrode placement, current, and stimulus type ⁸³.

Additionally, general anesthetics have been shown to have varying effects on cognition and likely worse outcomes in older patients ⁸⁴. The cognitive effects of the interaction between ECT and anesthesia are beyond the scope of this review; however, it is important to highlight the fact that even though generalized anesthesia for ECT is limited to approximately 10 minutes for each session, a patient may undergo dozens of ECT treatments with the potential for adverse effects due to the accumulation of anesthetic events ⁸⁵.

Mechanisms of ECT:

γ-Aminobutyric acid and Glutamate:

γ-Aminobutyric acid (GABA) is an inhibitory neurotransmitter expressed in interneurons distributed across the brain to modulate excitatory neural circuits. It is intriguing to consider the observation that ECT appears to effectively treat a variety of psychiatric illnesses that are also effectively treated with GABAergic medications such as benzodiazepines. These include unipolar and bipolar depression, bipolar mania, and catatonia ^86–88. Low GABAergic function may be a feature in a subset of mood disorder patients, representing a genetic susceptibility, consequence of chronic stress, or more likely a combination of the two. There is evidence supporting the notion that ECT has significant anticonvulsant properties and results in decreased neural metabolic activity over the course of treatment ^89–91.

Typically, a repeated seizures throughout a course of ECT reduces seizure duration and an increases in intracortical inhibition, both of which have been correlated with clinical improvement ^91,92. In a very large retrospective study of over 3000 patients receiving an acute course of ECT there was a decrease in the mean ECT seizure duration, with the greatest change within the first 3 treatments. Older age, higher ECT dose and higher treatment number were all associated with shorter seizures ⁹³. Interestingly, this notion has been evaluated with TMS. Cortical excitability in neurocircuitry has been assessed with single and paired pulse TMS to investigate these anticonvulsant effects of ECT. After 10 sessions of RUL ECT, researchers observed an enhanced activity of inhibitory circuits in human motor cortex, as measured by both increased intracortical inhibition and cortical silent period duration, while intracortical facilitation and resting motor threshold did not change ⁹².

There is evidence suggesting GABAergic involvement in ECT’s mechanism of antidepressant action ⁹⁴. Postictal suppression of seizure activity, largely thought to be driven by GABAergic neurons, has been correlated with treatment outcomes and there may be ways to alter seizure suppression with pharmacological augmentation during ECT administration ^95–103. Compared to non-responders, ECT responders appear to have higher GABA levels at baseline and after a course of ECT. Interestingly, immediately after ECT, free plasma GABA was decreased for up to 1 hour and another study found that occipital cortex GABA levels doubled ^104,105. Two hours after a single session of ECT, GABA and amino acids plasma levels were decreased in depressed patients possibly due to shifts into the brain ¹⁰⁶. A neurophysiological explanation of an ECT mechanism is that repeated electrically-induced seizures increase the ST, and therefore mood stability, by increasing the activity of GABAergic neurons that regulate the function of neurocircuits, thereby relieving aspects of neurovegetative depression such as rumination, anxious distress and even mania or catatonia ¹⁰⁷. There are other potential mechanisms, recently reviewed by Leaver, et al, suggesting a role in the thalamus and cerebellum in the attenuation of seizure activity after ECT as the data supporting the GABA hypothesis has not been completely consistent ¹⁰⁸.

Hormonal changes:

In humans, serum prolactin levels increase in response to seizures, with a 10- to 50-fold increase observed after ECT ¹⁰⁹. This has been observed in severely depressed patients, with ECT resulting in increases of prolactin and estrogen stimulating neurophysin (ESN) for at least 6 minutes post-ECT ¹¹⁰. This was also observed for a shorter time course for nicotine stimulating neurophysin (NSN) but no other stress/hypothalamic hormones increased this dramatically, indicating that ECT has specific hormonal effects that cannot be generalized to the release of any or all pituitary or hypothalamic stress-induced hormones ¹¹⁰. Another important marker of stress, corticotropin-releasing factor (CRF), has been studied, though with mixed results. Nemeroff and colleagues found that cerebrospinal fluid (CSF) concentrations of CRF in depressed patients decreased after ECT ¹¹¹. Kling and colleagues also found that CRF levels decrease after a successful course of ECT ¹¹². However, other studies have found no significant change in CSF-CRF following ECT despite clinical improvement ^113,114. Interestingly, Banki and colleagues noted that in the 15 patients who did not relapse within 6 months of ECT treatment, the CSF-CRF concentration was indeed reduced post-ECT ¹¹³. Cortisol levels have also been reported to be increased following ECT treatment, potentially due to increased arginine vasopressin stimulation of CRF-induced acetylcholine release ^114,115. Lastly, β-endorphin was shown to be transiently increased following ECT administration in depressed patients ^115,116.

Some studies have shown that growth hormone is decreased in depressed patients following ECT, though other studies have found no statistical difference ^115,117. Generally, stress hormone and neuropeptide changes have been documented, though given the heterogeneity of sample populations, sizes, and parameters, there has been a failure to establish a consensus. While the immediate release of stress hormones such as prolactin and cortisol may be important in the mechanism of ECT, it may also simply be an epiphenomenon and marker of seizure activity with little reliable insights into the therapeutic effects of ECT ¹¹⁸.

A few other studies have attempted to identify molecular substrates that may be involved in the therapeutic effect of ECT in human subjects. The kynurenine pathway and its metabolites have emerged as potential mediators in the pathophysiology of depression and the response to ECT ¹¹⁹. Tryptophan, the amino acid basis for serotonin, degrades along the kynurenine pathway. Also, tryptophan consists of, among other compounds, QUIN, an NMDA-receptor agonist, and kynurenic acid (KYNA), an NMDA-receptor antagonist. QUIN upregulation and increased QUIN/KYNA plasma ratios have been observed in depressed patients. Schwieler et al. found that after ECT, QUIN plasma levels and QUIN/KYNA ratios were decreased in 80% of patients ¹¹⁹. The authors suggest that the capacity of ECT to decrease the neurotoxic branch of the kynurenine pathway may be of clinical importance. However, they mention that plasma QUIN levels correlate with ECT treatment but not with the response to ECT as measured by the Montgomery-Asberg Depression Scale (MADRS), consistent with another finding that QUIN levels correlate suicidality but not depression generally ¹²⁰. Furthermore, QUIN and KYNA, as the authors acknowledge, do not easily pass the blood-brain barrier (BBB), and may not reflect significant changes in the CSF. Considering ECT may disrupt the integrity of the BBB these biomarkers may consequently have relevance to depression phenotyping and may potentially become predictive of ECT response, though determining the appropriate time to assay them is of high importance ¹²¹.

Serotonin:

Neurotransmitters and their receptors have been studied extensively in depression though, to a lesser degree, in response to ECT. Serotonin (5-hydroxytryptophan, 5-HT), its receptors (5-HT_1–7), and its transporters have been investigated thoroughly in the pathogenesis of depression and the use of antidepressants, specifically selective serotonin reuptake inhibitors (SSRIs). The theory of low serotonin levels and depression has been popular, though not clearly substantiated and perhaps even untenable ¹²².

Serotonin receptor 5-HT_1A postsynaptic density was found to be decreased following ECT in depressed patients ¹²³. Ishihara and colleagues found that 5-HT_1A postsynaptic receptors are sensitized to 5-HT following ECT treatment ¹²⁴. Widespread reduction of brain 5-HT₂ receptors has been observed in depressed patients following ECT, similarly to what has been observed with antidepressants ¹²⁵. Interestingly, peripheral platelet 5-HT₂ receptor density has been reported to be increased after ECT in depressed patients ¹²⁶. Plein and Berk found that platelet 5-HT_2A receptors are desensitized to serotonin following ECT, and they hypothesized that this decreased sensitivity may be a mechanism of ECT’s antidepressant action ¹²⁷. Ishihara and Sasa noted that the 5-HT_1A increased sensitivity and perhaps subsequent decrease in 5-HT_2A receptors explain the effects of ECT; they also point out that 5-HT₃ sensitization may play an important role ^124,128. Additionally, they have shown that ECT increases the release of glutamate and GABA and decrease the receptor function of noradrenergic and dopaminergic neurons in the locus coeruleus and substantia nigra respectively, thereby increasing noradrenaline and dopamine. These effects may also be mediated through the 5-HT₃ receptor sensitization.

Magnetic resonance spectroscopy:

Investigators have utilized proton magnetic resonance spectroscopic (1H-MRS) to measure neurotransmitter levels and their metabolites in the brains of patients with major depression before and after ECT – and details can be found in recent thorough reviews ^129,130. The results have been heterogeneous, but trends are emerging. One study found no significant differences in GABA, Glutamate (Glu), glutamine (Gln), choline (Cho) or glutathione (GSH) between pre- and post-ECT, though there was a decrease in N-acetylaspartate (NAA) levels after ECT that correlated with the number of ECT sessions ¹³¹. A study in RUL ECT in patients with TRD showed that Gln concentrations were anti-correlated with their degree of depression; and after a successful ECT course, Glx (Glu + Gln) increased significantly to that of age-matched controls ¹³². Another study found increased levels Glx in depressed patient that responded to ECT ¹³³. One group found significantly reduced Glx levels in the patients’ left cingulum at baseline compared to healthy controls that normalized after ECT ¹³⁴.

A detailed anatomical 1H-MRS study measured changes in Glx, Creatinine (Cre), Cho and NAA in the dorsal (dACC) and subgenual anterior cingulate cortex (sgACC) as well as the bilateral hippocampus in patients receiving ECT for depression. At baseline, patients showed less Glx in the sgACC, less NAA in the left hippocampus and more Glx in the left hippocampus compared to controls. After ECT they observed significant increases in Cre in the dACC and sgACC and decreases in NAA within the dACC and right hippocampus. Both ECT and symptom improvement were associated with decreased Glx in the left hippocampus and increased Glx in the sgACC. Lower NAA levels in the dACC at baseline predicted ECT-induced reductions in depressive symptoms. Changes in Glx in the sgACC and hippocampus approached that if controls after ECT. This may reflect and imbalance between underactive cortical and overactive subcortical limbic circuitry in patients with depression ¹³⁵. Severe depression appears to be associated with decreased Glx and therefore increasing Glx may be important to the mechanism and a potential biomarker of ECT response ¹³⁶.

Neuroanatomical changes:

Magnetic resonance imaging (MRI) studies after ECT consistently show volumetric changes across cortical and subcortical brain regions ¹³⁷. Anatomical changes correlate with number of treatments, likely with the strength of the electric field ⁵⁴, and the placement of electrodes ⁶¹. Recent mega-analysis corroborated this by illustrating approximately 5% increase in the gray matter subfields of the hippocampus and the basolateral nuclei of the amygdala. Beyond the subcortical mesial temporal lobe structures, volumetric increases have been observed in the anterior cingulate, postcentral, fusiform gyri, the medial prefrontal cortex, supplementary motor cortex, insula, and striatum ¹³⁸. The Global ECT-MRI Research Collaboration (GEMRIC) performed a multisite analysis of individual-level subject data and showed that, except for the cerebellum, the volumetric expansion of all subcortical regions of interest negatively correlated with ventricle size, indicating that the increases was likely at the expense of cerebrospinal fluid spaces ¹³⁷. These mega-analysis findings support widespread brain changes regardless of known relation to neurocircuits implicated in depression (i.e., default mode network, cortico-striatal-thalamic, and cortico-hypothalamic circuits) ¹³⁹. Large-scale data consistently point to changes in mesial temporal lobe (e.g., hippocampus and amygdala), far from the ECT electrodes, undergoing the greatest volumetric changes. Smaller studies did not consistently find volume changes outside of the mesial temporal lobe where effect sizes seem to be larger, possibly due to the lower sample sizes and associated reduced statistical power.¹⁴⁰. Understanding how these changes support response to ECT (therapeutic and iatrogenic), and the neurocircuits involved, remains an area of exploration ¹⁰⁸.

Neurogenesis and Brain-Derived Neurotropic Factor:

Smaller hippocampi have been observed in patients with depression in MRI as well as post-mortem studies ^141,142. Intriguingly, hippocampal size is increased after ECT and may be dose- and parameter-dependent ¹⁴³. A recent post-mortem human study of patients that underwent ECT displayed an increase in doublecortin, a marker of neurogenesis, in the dentate gyrus ¹⁴⁴. Whether adult neurogenesis occurs in humans remains controversial ^145,146. Regardless, it occurs in rodents and is increased by electroconvulsive stimulation (ECS) ^147,148. The neuroplasticity hypothesis is based on work in animal models suggesting a dose-dependent increase of neurogenesis in the dentate gyrus of the hippocampus after ECS, the preclinical approach to studying mechanisms of ECT. Additionally, in mice, the behaviors associated with ECT, were shown to be reliant on neurogenesis ^149–151.

There is evidence supporting the role of neurotrophic factors in human affective disorders. Decreased brain-derived neurotrophic factor (BDNF) levels have been observed in humans with depression ¹⁵². BDNF has received considerable attention in ECS research as well as depression research broadly ¹⁵³. BDNF mediates long-term neural and behavioral plasticity during social stress. In addition, BDNF expression has been implicated as a potential mediator in depression as decreased levels may atrophy limbic structures, while its upregulation plays a role in antidepressant response ¹⁵⁴. Karege and colleagues found that depressed patients had significantly lower serum levels of BDNF, which is increased in individuals responsive to ketamine ^155,156. Additionally, patients with TRD have lower baseline levels of plasma BDNF which are significantly increased by ECT. Furthermore, Piccinni and colleagues found that BDNF levels increase at the same time of observed clinical improvement, suggesting that changes of BDNF levels in response to ECT may represent a biological, state-dependent marker of remission following ECT ¹⁵⁷. This is in line with the belief that BDNF levels are associated with clinical changes in depression and that neuroplastic changes may underlie these clinical changes ¹⁵⁸.

Gliogenesis and Angiogenesis:

Cell proliferation induced by ECS includes glial (astrocytes, oligodendrocytes, and microglia) and endothelial cells in addition to neurons. Wennstrom and colleagues reported glial cell proliferation in the adult rat hippocampus following ECS, and that inhibition of gliogenesis by corticosterone administration is mitigated by ECS ^159,160. Their group also notes that glial cell proliferation increases in the rat amygdala following ECS ¹⁶¹. As corticosterone inhibition of gliogenesis in the rat hippocampus is counteracted by ECS, the mechanism of ECS’s antidepressant effect may be influenced by glial cell proliferation. Interestingly, glial cell proliferation after ECS has also been reported in the rat frontal cortex ¹⁶². The increase in number of these glial cells may exert a neuroprotective effect and counteract the effects of stress, thereby implementing an antidepressant effect. Additionally, endothelial cell proliferation in response to ECS has been observed in the rat prefrontal cortex ¹⁶³, hypothalamic regions including the paraventricular nucleus, supraoptic nucleus, and ventromedial nucleus in response to ECS ¹⁶⁴. Endothelial cell proliferation and increased angiogenesis has also been observed in the adult rat hippocampus following ECS and this increased vascularization may be needed to support hippocampal neurogenesis ^165,166. These observed endothelial cell inductions support angiogenesis as a potentially important component of ECS’s mechanism.

Microglial cells are resident macrophages that enter the brain early in development, and while they have homeostatic roles in brain function, they can cause neuroinflammation ¹⁶⁷. One study showed that mice subjected to chronic social defeat stress and subsequent ECS had activated hippocampal microglial cells ¹⁶⁸, while a study comparing ECS to a model of status epilepticus showed no changes in hippocampal CA1 microglial density ¹⁶⁹. Taking these findings together, it is possible that the antidepressant effects of ECS may be collectively mediated by neuro-, glio-, and angiogenesis, mainly in the hippocampus and associated brain structures. Overall, given the brain regions involved, the rodent literature appears to support changes involving the hippocampus, amygdala, hypothalamus, and frontal cortical regions, though no direct interrogations of the circuits connecting these regions have been conducted.

Perspective and Conclusion:

ECT remains one of the most effective treatments for several neuropsychiatric conditions, particularly depression. Understanding the neurobiological mechanisms that underlie its effects could provide important opportunities to optimize ECT as well as reveal response biomarkers and treatment targets to engaged with novel, safe, and efficacious therapeutic approaches ¹⁷⁰. Animal models have revealed molecular and cellular insights into the effects of ECT despite their translational limitations. There are many caveats to modeling depression in rodents, including the fact that the evaluation of antidepressant treatment efficacy is state-dependent ^171–174 and cannot easily be reproduced in animals, especially in rodents. For example, ECT is frequently used in geriatric patients with severe depression, particularly with psychotic features ¹⁷⁵. Rodent studies are almost entirely performed around 12 weeks of age, corresponding to young adult in humans ^176,177. Therefore, the findings, while informative, are not likely translatable to older populations, which have a much higher risk and rate of cognitive deficits and delayed recovery times ^178,179. Additionally, while recent data suggests multiple brain region changes after ECT, we are just starting to understand the neurocircuitry of ECT effects and how to optimize a circuit-based approach, such as pairing electrode placement with specific and intentional stimulus parameter selection. For example, a recent study has shown an increase in frontal lobe activity and the executive control network with potential fMRI resting-state biomarkers that include the dorsomedial and anterior cingulate cortex after ECT ^180–182. Another study indicated that connectivity between the dorsal prefrontal region and the limbic and default-mode networks serves as a significant predictor of response to ECT ¹⁸³. Is the future of ECT one that uses fMRI-guided electrode placement to optimize antidepressant and cognitive outcomes? Or perhaps, current-amplitude-titrated or electric-field-informed dosing based on hippocampal changes? In this context, it may be relevant to disentangle the direct effects of the ECT electrical stimulus versus the seizure it induces, including drawing parallels with subconvulsive stimulation methods such as TMS.

Our understanding of the mechanisms of ECT efficacy and adverse effects on cognition are far from sufficient. Is it a coincidence that the hippocampus and amygdala are consistently implicated in the mechanisms of ECT and yet these are the same brain regions responsible for episodic/autobiographical memory, loss of which is the most concerning side effect of ECT? In addition to research of fundamental mechanisms, evaluation of mitigation approaches for optimization of outcomes is paramount. We can consider undertaking large retrospective studies on concomitant medication and comorbidities so that adjunctive treatments and screening can be implemented. Additional longitudinal studies looking at these data are necessary to begin to determine predictors or outcomes and side effects ⁵⁷. As these data emerge, meta- and mega-analyses of ECT responders and non-responders may help us triage treatments for those at risk of cognitive side effects. Further evaluation into electrode placement and stimulation parameters is needed to determine clinical approaches to improving outcomes. Additionally, there may be inflammatory and/or hormonal profiles that are relatively indicated or contraindicated. While there is no “unified theory” of ECT mechanism and how it relates to seemingly divergent psychiatric illness from mood disorders to catatonia, lines of evidence point to the modulation of dysregulated neurocircuit function in disease states by repeated application of electric fields and/or seizure activity. This is likely achieved by enhancing the regulation of aberrant neurocircuit activity at various levels of biological function (e.g. molecular, cellular, anatomical, physiological), which results in more adaptive circuit dynamics, in turn leading to euthymia. A better understanding of neurocircuits dynamics and molecular mechanisms associated with ECT efficacy and side effects would inform both research on treatment development and clinical treatment planning and would reduce stigma.