Recent Advances in Optimizing Electroconvulsive Therapy

Abstract

This review aims to summarize and critically evaluate the evidence for recent advances and alternative approaches in electroconvulsive therapy (ECT) technique. Novel developments in ECT research are also mentioned. An EMBASE literature search was undertaken of clinical trials, case reports and research updates on novel and alternative approaches in ECT practice and research, including alternative electrode placements, variations in stimulus configuration, and novel developments (magnetic seizure therapy, focal electrical stimulation). The evidence for these approaches is reviewed, and implications for the optimizing of ECT in clinical practice are discussed. Evidence from studies suggests that unilateral ECT be given at substantially suprathreshold doses (at least 6 times seizure threshold) for maximizing efficacy, while bilateral ECT is likely to be effective at doses of 1.5–2.5 times seizure threshold. There is some evidence to support the use of bifrontal ECT, although further research is required to establish its efficacy and side-effects relative to standard unilateral and bilateral electrode placements. Other alternative electrode placements have been minimally studied. More advantageous efficacy side-effect outcomes may be achieved by reducing the pulse width and/or frequency of the ECT stimulus. Lastly, novel developments using alternative means of seizure induction (magnetic fields, focal electrical stimulation) may hold promise for the future. Clinical practice should be guided by a careful appraisal of the available evidence for alternative approaches in ECT technique.

Since its introduction in 1938, the technique of electroconvulsive therapy (ECT) has been progressively refined to improve efficacy while minimizing side-effects [1]. Some alternative approaches (e.g. bifrontal ECT) have been adopted by clinicians striving to optimize the efficacy/side-effect profile of ECT. This review aims to summarize and critically evaluate the evidence for alternative approaches and recent advances in ECT technique. Recent clinical trials, case reports and research reports on electrode placement, stimulus dose and stimulus con-figuration were obtained from a literature search using EMBASE. The findings and their implications for clinical practice are discussed. Lastly, novel developments in ECT-related research are mentioned.

Stimulus dose in unilateral and bilateral ECT

Since research in the early 1990s there has been increasing recognition of the importance of suprathreshold dosing in optimizing efficacy, particularly for right unilateral (RUL) ECT [2]. This refers to the use of electrical doses in excess of that required for inducing a generalized seizure. There are several recognized methods for achieving this [3]: empirical titration of each patient's seizure threshold (ST) at the first session, then treatment at doses corresponding to a multiple of this (detailed in [4]); algorithms based on age and gender; or a fixed high dose. Although debate continues over the best method of approaching suprathreshold dosing [5–9], the requirement that doses be substantially in excess of threshold for optimizing efficacy has been clearly established for RUL ECT [2, 4,10–14].

Where the ST titration method has been adopted, practice to date has often been to give RUL ECT at a dose equivalent to 2.5–3 times the ST and bilateral (bifrontotemporal) ECT at approximately 1.5 times ST. (As alternative approaches to bilateral ECT are discussed below, the bilateral form most commonly in use will be referred to by the more accurate description, bifrontotemporal ECT.) These stimulus doses led researchers and practitioners initially to conclude that RUL ECT was not that effective, particularly in comparison with bifrontotemporal ECT. For example, Ng et al. [4] reported that treatment with RUL ECT at 2.5 times ST led to treatment response (typically defined as 60% reduction from baseline depression scores) in only 31% of 32 patients, although another 31% achieved a 50% reduction in depression scores. Although the sample size was small, the findings are consistent across a number of other studies, with others reporting response rates of 30% [10] and 43% [11] with RUL ECT at 2.5 times ST and 39% with RUL at 2.25 times ST [12].

These suboptimal results led to the investigation of higher doses. In a double-blind randomized trial, Sackeim et al. [10] examined RUL ECT at 1.5, 2.5 and 6 times ST and bifrontotemporal ECT at 2.5 times ST. They found that increasing the dose relative to ST markedly improved response rates for RUL ECT (65% for 6 times ST, compared with 30–35% for 1.5–2.5 times ST) and that efficacy was associated with dose relative to ST rather than absolute dose. Moreover, compared with bifrontotemporal ECT at 2.5 times ST, RUL ECT at 6 times ST was found to have comparable efficacy but lesser cognitive side-effects (Mini-mental Status Examination [MMSE] [15], verbal and visual anterograde memory, retrograde memory).

The above study has been criticized for the choice of bifrontotemporal ECT at 2.5 times ST as a comparator and other groups have examined the benefits of high-dose RUL ECT relative to bifrontotemporal ECT at 1.5 times ST. McCall et al. [13] randomized 77 patients to receive RUL ECT at 8 times ST or bifrontotemporal ECT at 1.5 times ST and found no difference in efficacy or cognitive side-effects (visual and verbal anterograde memory, retrograde memory). A randomized controlled trial in Melbourne found that RUL ECT at 5 times ST had a superior response rate compared with RUL ECT at 2.5 times ST and lesser cognitive side-effects (verbal anterograde memory) compared with bifrontotemporal ECT at 1.5 times ST at 1 month follow-up, although results were not significantly different at the end of the ECT course (n = 40) [14, 16]. In a small study, Heikman et al. [11] randomized 24 patients to RUL ECT at 5 and 2.5 times ST and bifrontal ECT at ST. Although a greater proportion responded to RUL at 5 times ST (7 out of 8), than RUL at 2.5 times ST (3 out of 7) or bifrontal (3 out of 7), this difference was not significant. However, the authors report that high-dose RUL ECT led to a statistically significant faster antidepressant effect, requiring fewer treatments than the other two groups, and resulting in lesser impairment in MMSE scores than the bifrontal group.

Across the studies high-dose RUL ECT (at 5–8 times ST) appears to offer a better efficacy/side-effect compromise than bifrontotemporal ECT at 2.5 times ST, and possibly bifrontotemporal ECT at 1.5 times ST. In their review on the topic, Greenberg and Kellner [17] note that for bifrontotemporal ECT doses of 1.5 times ST may be effective and that maximal doses should not exceed 2.5 times ST as higher doses do not enhance efficacy but may increase cognitive side-effects. In contrast, they state that doses for RUL ECT should be at least 2.5 times ST, with greater efficacy at doses equivalent to 6–8 times ST. Similarly, in the report of the meta-analysis conducted by the UK ECT Review Group [18], an overall comparison of bilateral and unipolar ECT revealed that bilateral ECT was more effective. However, the authors noted that this included studies with treatment at all doses and that work by Sackeim et al. [2, 10] found RUL ECT at higher doses to be as effective as bilateral ECT with fewer adverse cognitive effects. In recognition of the greatly differing levels of effectiveness for RUL ECT at different dose levels, the Royal Australian and New Zealand College of Psychiatrists Clinical Memorandum on ECT was updated in 2005 to mention superior efficacy for RUL ECT with dosing at 6 times ST in comparison with lower doses [3].

The above findings appear to be independent of the anaesthetic agent used. The anaesthetic agents used in the studies above were methohexital [2,10–13], thiopentone [4], and thiopentone or propofol [14, 16]. Two prospective studies that randomized subjects to receive methohexital or propofol while treated with RUL ECT at doses determined by the titration method did not find significant differences in therapeutic outcomes, as measured by depression rating scales [19, 20], or cognitive outcomes (MMSE, visual and verbal memory, word generation) [20]. Overall, studies examining the effect of using propofol or methohexital [19–21], or propofol or thiopentone [22] for ECT anaesthesia, found that seizures were shorter with propofol, but that there were no significant differences in efficacy outcomes or the number of ECT treatments required for improvement.

There is scant information on the benefits of further dose increases for RUL ECT, beyond 8 times ST. Preliminary insights into this were obtained indirectly in a study by McCall et al. [12] which randomized 72 patients to receive RUL ECT at 2.25 times ST or a fixed high dose (403 mC). The main finding of this study was that the fixed high-dose group had superior efficacy (67% response vs 39% response) although greater cognitive impairment (30% vs 7% subjects had a ≥5 points decline in MMSE score post ECT compared with baseline; the fixed dose group also had more impairment in retrograde, although not anterograde memory). Seizure threshold was empirically determined for all subjects in this study, allowing a secondary analysis of subjects in the fixed dose group who had in fact received ECT at 8–12 times ST compared with those in the titration group who received ECT at 2.25 times ST. This showed that both antidepressant response and the global cognitive sideeffects (MMSE) were related to the dose relative to ST (rather than absolute dose) and that the dose–response curve appeared steeper for cognitive side-effects than for antidepressant response. Thus, dosing for RUL ECT is practically limited by the risk of cognitive side-effects, as it is for bifrontotemporal ECT. It is in this context that investigators have turned to alternative methods for enhancing the efficacy of ECT while avoiding unacceptably excessive cognitive side-effects.

Bifrontal and asymmetric bilateral ECT

While the above practitioners have investigated the use of high-dose RUL ECT, Fink [23] has argued against ST estimation by titration and unilateral electrode placement per se. He contends that RUL ECT is less effective than bilateral ECT and other placements should be used: ‘we have fiddled with this placement for more than 40 years, and no credible way has been found to make its application effective’ (p.11). Instead, he has advocated the use of bifrontotemporal and bifrontal ECT, based on the argument that the neuroendocrine effects of ECT and the effects on the hypothalamus with RUL ECT are indirect, whereas those from bifrontotemporal and ‘most likely’ bifrontal ECT are direct [23].

A more anterior bilateral placement (bifrontal ECT) was proposed by Inglis [24] to minimize effects on verbal and non-verbal cognitive functions while maintaining therapeutic effect, by avoiding stimulation of the temporal lobes and underlying hippocampal gyri. Early ECT used bifrontal placement, but there were a number of difficulties including poor induction of seizures and scalp burns (owing to the electrodes being too closely spaced) [25]. Nevertheless, a small study of anterior-bifrontal ECT was reported in 17 depressed patients by Abrams and Taylor [25]. Although there were no comparison groups, the rate of clinical improvement appeared to be intermediate between that of bifrontotemporal and RUL ECT, with memory impairment comparable to RUL ECT and less than bifrontotemporal ECT.

A Canadian group compared the cognitive and therapeutic effects of bifrontotemporal, bifrontal or RUL electrode placement using brief pulse threshold level stimulation in a double-blind randomized controlled trial in depressed patients [26, 27]. Their hypothesis was that RUL placement would impair visuospatial but spare verbal functions, bifrontotemporal ECT would affect both visuospatial and verbal functions, and bifrontal ECT would spare both. Using change in depression scores, speed of recovery and global clinical status to assess antidepressant effects, bifrontal placement was shown to be superior to RUL and bifrontotemporal ECT, with bifrontotemporal ECT intermediate between bifrontal and RUL ECT in 59 depressed patients [26]. However, RUL in this study was given at threshold doses, an approach since shown to be ineffective [2]. There were no differences in efficacy between bifrontal and bifrontotemporal ECT at 3 and 6 month follow-up. Verbal and non-verbal function and planning and sequencing ability were assessed in 40 patients from this study. Bifrontal placement produced less cognitive effects than RUL or bifrontotemporal at the end of treatment, sparing both verbal and non-verbal functions [27]. There were no differences 3 months later.

In a randomized controlled trial, bifrontal and bifrontotemporal electrode placement, using brief-pulse 1.5 times ST stimuli, were then found to be equally efficacious in a group of 48 unipolar and bipolar depressed patients, as measured by Hamilton Depression Rating Scale scores [28]. However, cognitive side-effects were only assessed using MMSE scores, which were significantly lower in the bifrontotemporal group after treatment. In the study by Heikman et al. referred to above, no statistically significant differences in response rate or MMSE changes were found between high-dose RUL (5 times ST), moderate-dose RUL (2.5 times ST) and lowdose bifrontal (1.0 times ST) ECT [11]. This study was probably insufficiently powered to find real differences between the treatment groups. Moreover, the investigators suggested the poorer outcome of bifrontal ECT in this study compared with other studies may be due to the lower stimulus level. They suggest a bifrontal stimulus dose of at least 1.5 times ST, as well as more sensitive neuropsychological testing than the MMSE.

In a retrospective study of 76 patients, Bakewell et al. [29] compared bifrontotemporal and bifrontal electrodeplacement, given at doses sufficient to achieve seizures of ≥30 seconds. Bifrontotemporal placement was significantly more efficacious, although only modestly. There was a greater risk of relapse requiring hospitalization with the bifrontal group, although this placement also resulted in a lesser rate of clinically noted cognitive side-effects. Likewise, Little et al. [30] retrospectively reviewed 30 elderly patients who had received RUL ECT at 6 times ST, or bifrontal or bifrontotemporal ECT at 1.5 times ST. From the clinical observations recorded no statistically significant differences were found between the groups in terms of response or cognitive side-effects. The limitation with these studies is the retrospective design and lack of neuropsychological testing. As naturalistic studies, however, they may be relevant to similar populations.

Thus there is some evidence for the use of bifrontal ECT, although larger randomized controlled trials comparing bifrontal, bifrontotemporal and RUL ECT at adequate doses, with detailed neuropsychological testing, are needed to clarify the relative efficacy and safety of this form of ECT, in order to establish its clinical role with respect to existing electrode placements.

Another alternative form of bilateral ECT, with ‘asymmetric bilateral right frontotemporal left frontal’ electrode placement has been advocated by Swartz [31]. This differs from standard bifrontotemporal ECT in that the left electrode is moved about 6 cm anterior, with its lateral edge ‘medial to the bony intersection ridge between the temple and the forehead’ [31]. The rationale for this arrangement is to reduce direct stimulation of the left temporal lobe and hence, cognitive side-effects. Swartz [31] reported an open trial of 10 subjects who weresuccessfully treated with this electrode placement (i.e. all subjects attained clinical ‘remission’ and average MMSE score at end of treatment was 28.4 (with no baseline score for comparison). Manly and Swartz [32] reported a further four cases who had had problematic cognitive effects with previous courses of ECT using other electrode placements, including in some cases a sine-wave stimulus. All four patients achieved good clinical improvement without decline in their MMSE scores after receiving asymmetric bilateral ECT. Unfortunately, there is only limited anecdotal evidence for the use of this form of ECT as there have been no prospective, randomized comparisons with other electrode placements, examining its relative efficacy and side-effects. Thus there is little current evidence for the clinical use of asymmetric bilateral ECT until further research is undertaken.

Stimulus characteristics

From research findings to date, it is likely that for all the above electrode placements, the enhancement of efficacy by dose increments will be limited by the accompanying increases in cognitive side-effects. A possible strategy for uncoupling this relationship between efficacy and side-effects is to alter the characteristics of the electrical stimulus. Modern ECT devices deliver a constant current (set at 0.8–0.9 A) rather than a constant voltage, as this is more reliable in inducing seizures and has a lesser risk of skin burns [33]. Thus the stimulus parameters that can be adjusted are pulse width, frequency (i.e. number of pulses per second) and the duration of the stimulus train.

An example of a major advance achieved by altering stimulus parameters is the substitution of brief pulse width (0.5–1.5 milliseconds [ms]) square-wave stimuli for the long pulse width (8 ms) sine-wave stimulation used earlier in the history of ECT. With this development efficacy was preserved (see [34] for a meta-analysis of 19 studies) while cognitive side-effects (confusion, retrograde memory) were markedly reduced [35–37]. Lisanby [38] suggests that the difference between the wave forms can be explained by the inefficient properties of the sine wave, which has a slow onset to peak, increasing the threshold for depolarization, and a slow offset, resulting in stimulation during the refractory period. However, as noted by Swartz and Manly [39], there has been no research directly comparing sine-wave and square-wave forms of equal pulse width. Thus conclusions that differences in efficiency and cognitive sideeffects result from the different wave forms per se, rather than differences in pulse width, are unproven.

Further research has investigated the effects of different stimulus parameters for square wave stimuli, including pulse width and frequency. Swartz and Manly [39] found that seizures (using the asymmetric bilateral electrode placement) were more reliably elicited and peak heart rate (postulated by Swartz as a marker for efficacy [40]) was higher with a 0.5 ms pulse width stimulus than a 1 ms pulse width stimulus.

Similarly, Sackeim [41] reported on preliminary results in 75 subjects who were randomized to receive RUL (6 times ST) or bifrontotemporal ECT (2.5 times ST) at pulse widths of 0.3 or 1.5 ms. On average, ST was 3–4 times higher with the 1.5 ms than 0.3 ms pulse width. For RUL ECT, efficacy was maintained while cognitive side-effects were markedly reduced with the 0.3 ms pulse width, although for bifrontotemporal ECT cognitive side-effects were reduced but with a corresponding loss of efficacy. (As the study is in progress, preliminary cognitive test results were only reported for retrograde amnesia, although the article notes that cognitive outcomes are superior with the 0.3 ms pulse width across ‘a large set of cognitive measures’.) The intriguing finding of a loss of efficacy with bifrontotemporal ECT at 0.3 ms pulse width offers insights into the mechanisms underpinning the efficacy of ECT. Sackeim [41] postulates that a narrower band of tissue is stimulated with the brief 0.3 ms pulse width stimulus (as there is less charge per phase), and arguing from several lines of evidence, suggests that this results in a less robust inhibitory response in the prefrontal cortex, the latter hypothesized as the critical factor in efficacy. He attributes the preservation of efficacy with RUL at 0.3 ms pulse width to dosing at 6 times ST, whereas dosing at 2.5 times ST (bifrontotemporal group) may not have resulted in stimulation of a large enough area of the prefrontal region.

From neurophysiological observations, pulse widths of 0.1–0.2 ms are optimal for neuronal depolarization [42]. However, reductions in pulse width are practically limited by the difficulty of achieving sufficient charge for a typical ECT stimulus – with very low pulse widths, pulse frequency, current amplitude and/or the duration of the stimulus train would need to be increased to levels that are non-ideal for eliciting seizures or risk skin burns.

Apart from pulse width, the effects of pulse frequency on ECT outcomes have been investigated. Theoretically, given the inherent refractory period of a neurone after depolarization, pulse frequencies of <70 Hz should be inefficient for an ECT stimulus [43]. Few studies have examined the effects of pulse frequency while holding pulse width constant. Devanand et al. [44] showed that a titration protocol relying on increases in pulse frequency for dose increments led to higher estimated STs than a protocol using a fixed low pulse frequency (30 Hz) and relying on increases in train duration for dose increments. Indirectly, these results support the greater efficiency of lower pulse frequencies.

Swartz and Manly [39] did not find a difference in seizure induction (i.e. rate of missed seizures) between stimuli of 30 Hz and 60 Hz pulse frequency, randomly varied over a course of ECT treatment. Girish et al. [45] found that 50 Hz stimuli resulted in lower STs than 200 Hz stimuli in the same patients. The same group [46] then studied the effects of these parameters in a clinical trial. Forty patients were randomly assigned to receive RUL at 2.5 times ST with 50 Hz or 200 Hz stimuli. The 50 Hz group had a significantly lower mean ST, and a trend towards better performance on the trail making test after 6 ECT, although there was no difference in efficacy, with both groups showing robust improvement.

Thus these studies suggest that ECT stimuli with shorter pulse widths and lower pulse frequencies result in lower STs, and thus lower absolute doses where the dosing-relative-to-threshold method is used. One would expect lesser cognitive side-effects as a result and there are empirical results to suggest this is the case for 0.3 ms pulse width stimuli. It is possible that the findings refiect the longer stimulus train duration used, as comparable doses are achieved by manipulations of train duration in studies where pulse width or frequency are altered while the other is held constant. Further research is required to examine the robustness of the findings above and explore the relative importance of different stimulus parameters.

Novel devel opments in ECT

In a radical development, Sackeim et al. have started investigating the effects of a unidirectional ECT stimulus [41]. As this involves a current fiowing in one direction only, one electrode is an anode and the other a cathode. Furthermore, by using a small anode (placed at the nasion) and large cathode (located at the vertex), much more focal electrical stimulation of the brain can be achieved. This novel technique has been labelled ‘focal electrically administered seizure therapy’ (FEAST). Preliminary trials with FEAST in non-human primates have demonstrated that localized seizures with ictal activity evident in prefrontal EEG recording can be elicited in the absence of a generalized seizure. This may be advantageous if prefrontal stimulation is indeed the relevant mechanism for the efficacy of ECT, as suggested by studies demonstrating strong associations between the antidepressant efficacy of ECT and reductions in cerebral blood fiow and metabolism and increases in slow-wave electroencephalographic activity in prefrontal regions after ECT [47, 48].

A recent and promising direction in ECT research has been the development of magnetic seizure therapy (MST). This uses the same principle as transcranial magnetic stimulation (TMS), that is, using magnetic fields to transfer energy across the scalp, and resulting in induced eddy currents in underlying cortex, but with machines capable of generating a much more intense magnetic field. Seizures are generated, but these are induced more focally than with traditional ECT. Lisanby et al. [38] reported preliminary data from a trial with a limited number of MST sessions (two of the first four treatments were MST and the others were conventional ECT). In all cases seizures resulted and subjective side-effects were significantly lower for the MST treatment. In addition, performance on tasks of attention, and measures of retrograde amnesia and category fiuency were all superior following the MST treatments conferring apparent advantages in cognitive side-effects, although its efficacy is yet to be established.

These new developments involve the generation of seizures under anaesthetic, an aspect of ECT that is considered disadvantageous, compared with non-convulsive therapies such as TMS. However, the efficacy of TMS in its current form does not yet equate that of ECT and the optimal method of giving TMS is still under investigation [49].

Conclusions

The practice of ECT continues to be optimized by ongoing research. Recent research has confirmed the principle of suprathreshold dosing for RUL ECT, with studies demonstrating increasing efficacy with increasing dose (up to 12 times ST), although clinically dose increases are limited by a commensurate increase in cognitive side-effects. Consistent efficacy has been demonstrated for bifrontotemporal ECT, with results suggesting that dosing should be between 1.5 and 2.5 times ST for optimal overall results in terms of efficacy and cognitive side-effects.

Other approaches have sought to improve efficacy without exacerbating cognitive side-effects. While there is some evidence for the safe and effective use of bifrontal ECT from three small randomized trials and several other reports, its efficacy and safety relative to traditional RUL and bifrontotemporal ECT need to be established with further large randomized controlled trials using adequate dosing and formal neuropsychological testing. At present there is only anecdotal evidence from a small number of case reports for the use of another alternative electrode placement: asymmetric bilateral ECT. While these case reports suggest reasonable efficacy and no major concerns with cognitive side-effects, there is no direct information on the relative benefit of this form of ECT to guide its use. Recent studies on stimulus characteristics (pulse width, frequency) suggest promise for the further optimization of ECT, such that efficacy is preserved while lowering the risk of cognitive impairment. Lastly, novel approaches such as FEAST and MST must be considered entirely experimental, but may hold promise for the future.