Neuromodulation and Eating Disorders

Modern non-invasive brain stimulation techniques prominently include rTMS. In rTMS, a current is passed through an electromagnetic coil to induce an increase (high frequency; HF-rTMS) or decrease (low frequency; LF-rTMS) in cortical excitability in target brain regions [9]. Overall, rTMS is well-tolerated and safe [10]. Therapeutic applications of rTMS are being investigated across psychiatric disorders; however, its use is best established in depression where protocols have been incorporated into clinical guidelines [11].

In an open 18-month follow-up of this trial, mood improvements remained broadly stable in the real rTMS group and there was a “catch-up” in the sham group, as 10/12 of these participants subsequently took the opportunity to receive real rTMS treatment [13•]. In regard to BMI change scores, by 18 months, there was a medium between-group effect, along with a higher rate of weight recovery in the real rTMS group compared to sham (BMI above 18.5 kg/m²: 46% vs. 9%; χ² = 3.67, p = .056), suggesting the mood effects of rTMS on AN precede those on BMI.

A notable finding from the participants’ qualitative feedback at the end of TIARA [13•] was a growing flexibility and relaxation around eating and food choices following rTMS. This finding was supported by data from the Food Choice Task (see [14] for task methodology). In the real rTMS group, there was a decrease in self-controlled food choices post-treatment (vs. baseline), reflected in increased selection of tasty-unhealthy foods [15]. This could have arisen because rTMS induced functional changes in the DLPFC and associated neurocircuits: This is partially supported by the arterial spin labelling data obtained in the TIARA study [16]. Unlike previous rTMS studies (e.g. [17]), participants that received real rTMS showed no differences in cerebral blood flow (CBF) at the stimulation site (DLPFC); they did, however, show greater reductions in amygdala CBF, and this was associated with longer-term weight gain [16]. It is possible that lower CBF in the amygdala was associated with reduced fear around food, and this enabled the reported increased flexibility around food post-rTMS (thus resulting in longer-term weight gain). Studies of this potential mechanism underlying weight gain following rTMS are warranted.

The DLPFC is the most widely used target for rTMS in EDs, which may not be optimal for all recipients, e.g. a recent report of rTMS to the left-DLPFC in a case of AN and comorbid depression described no improvement in AN or mood symptoms [18]. As data emerge, new protocols are being developed and alternative stimulatory targets are being examined. For example, a Canadian group targeted the dorsomedial prefrontal cortex (DMPFC) in a case series (n = 19; [19]) and reported significant improvements in core symptoms of AN (EDE global score; p = .010), anxiety (BAI; p < .001) and mood (BDI; p = .041) symptoms, from pre- to post-treatment. Their data showed that lower pre-treatment resting state functional connectivity from the DMPFC to the right frontal pole and left angular gyrus correlated with ED symptom improvement. The frontal pole is functionally coactive with the salience network regions (e.g. DMPFC), which have a role in cognitive and impulsive control [20]. Therefore, patients with weaker pre-treatment connectivity between these regions, who also showed greatest improvement in ED symptoms, may be those with reduced capacity for cognitive/impulse control and for whom DMPFC-rTMS may be particularly well-suited. To target different elements of AN (e.g. rigidity), Woodside et al. [19] highlight the potential value of stimulating the right orbitofrontal cortex, which has demonstrated potential for treating obsessive compulsive disorder (OCD) [21].

Evidence for the use of rTMS in BED is lacking, although we anticipate the findings of a double-blind, sham-controlled RCT from Maranhão et al. [31].

tDCS involves the application of a constant weak direct current via electrodes placed on the scalp to increase (anodal tDCS) or decrease (cathodal tDCS) cortical excitability [32]. Compared to other neuromodulation techniques, tDCS is easy to use, inexpensive and associated with fewer adverse effects [33, 34].

A further case report applied 2 mA anodal tDCS to the left-DLPFC in a case of AN with comorbid PTSD [37]. Improvements were seen in self-reported body dissatisfaction and weight gain, but the onset of type 1 diabetes coincided with the use of tDCS. The effect of tDCS on glucose metabolism remains unclear; however, these findings are not consistent with reports that show reduced blood-glucose in healthy controls following repeated application of tDCS (e.g. [38]).

Finally, an ongoing pilot sham-controlled RCT in Australia [39] is using high-definition tDCS (i.e. more focal stimulation) applied to the left inferior parietal lobule (IPL) in participants with AN. The IPL is a novel neuromodulation target in AN and was selected due to evidence showing reduced connectivity in this region [40] and its role in eye movement, multi-sensory integration and body image. The findings will inform future trials targeting the IPL with tDCS in EDs.

Our group has recently completed an RCT investigating the use of anodal tDCS to the right-DLPFC combined with approach bias modification (ABM) training for the treatment of adults with BED (see [43] for protocol). ABM training aims to retrain approach bias to appetitive cues into avoidance. Concurrently, administering tDCS is hypothesised to enhance neuroplasticity in pathways activated by the ABM and, in this way, is thought to amplify its efficacy. An integrated qualitative process evaluation [44] suggested that participants experienced ABM training and real/sham tDCS as an acceptable treatment combination, with few side effects; however, clinical outcomes are yet to be published.

An expanding area involves the use of at-home tDCS, which offers several advantages, including ease of access and increased compliance [45]. Several ongoing RCTs are assessing at-home tDCS as a treatment for BED. Our group is administering 10 sessions of concurrent at-home real/sham tDCS and attention bias modification training [46], and Elkfury et al. [47] are combining at-home tDCS with a nutritional counselling therapy, delivering these separately or concurrently in a multi-arm RCT.

ECT has long been used to treat mental illness, particularly when a rapid amelioration of severe and/or life-threatening symptoms is needed [50]. However, no randomised controlled trials (RCTs) have examined the efficacy of ECT in EDs. A recent systematic review and single case report by Pacilio et al. [51•] identified 14 cases with EDs treated with ECT from 11 publications. Case reports were mainly limited to AN, apart from one case of BED with comorbid obesity and bipolar disorder. Pre- and post-treatment BMI values were only reported in six cases. Of these, 50% of cases achieved a “normal” BMI over various follow-up periods, including the patient with BED who had a pre-treatment BMI of 97 kg/m² [51•]. The number of ECT sessions was highly variable across case reports (5–45), many patients had significant psychiatric comorbidities, two cases were older adults (> 75 years), and several cases were treated with ECT at a time when specialist ED treatments were not widely available. In their own report of a case with AN and recurrent depression and suicidal ideation, Pacilio et al. [51•] described improvements in mood following ECT, but no improvement in ED symptoms or BMI.

Since then, two further case reports have described the use of ECT in AN in the absence of comorbidities. One case showed gradual weight gain over 12 weeks following ECT [52]; the other case showed no improvement in ED, mood, or anxiety symptoms [53]. It is of note that in both reports where improvements in weight [52] or mood symptoms [51•] were reported, little to no change in attitudes towards weight, shape, or eating was observed. Thus, while the review and case studies available suggest potential value in using ECT to expedite weight restoration in AN [51•], they also highlight the need for adjunctive interventions that address weight and shape concerns. As such, given the evidence in favour of ECT is anecdotal and its safety and tolerability have been questioned, particularly due to the associated memory impairments, we would not presently recommend its use in EDs.

DBS is being investigated for its therapeutic potential in EDs, but, as it is an invasive procedure, its use has been limited to those with highly refractory SE-AN. DBS involves surgical implantation of electrodes into key structures (e.g. subcallosal cingulate, SCC [54]) implicated in ED pathology. In a recent randomised, sham-controlled, crossover trial [55•], DBS was applied to two different targets based on comorbidities: the SCC for affective disorders (n = 4) and nucleus accumbens (Nacc) for anxiety disorders (n = 4). This is relevant given the reported functional connectivity differences between “pure” EDs and EDs with comorbidities [56], as well as AN subtypes [57]. At 6-month follow-up (phase II), there were no significant changes in BMI from baseline to each month post-operatively. Villalba Martínez et al. [55•] surmised that as four participants underwent inpatient treatment to achieve the minimum BMI required for surgery (13 kg/m²), the preoperative BMI values did not reflect their “usual” weight. Thus, a BMI reference value (BMI-RV) was calculated from mean BMI in the preoperative period. These findings showed a significant increase (p = 0.02) from BMI-RV to BMI at 6-month follow-up, with 5/8 treatment responders (≥ 10% increase in BMI-RV). Phase III of this trial is ongoing, with five responders randomised to one of two arms (DBS switched ON/OFF or OFF/ON) and three non-responders not randomised and continuing with monthly assessments until 12-month follow-up.

The largest case series (n = 28) using DBS (to the Nacc) in AN [58] reported significant increases in BMI at 6-month (p < .001) and 2-year (p < .001) follow-up, with (a) 43% of patients achieving a BMI > 18.5 kg/m² 2 years post-operatively. In addition, obsessive–compulsive, mood and anxiety symptoms significantly decreased from baseline to 6-month and 2-year follow-up. Post hoc analysis revealed that mean BMI post-treatment was significantly lower in the binge-purge subgroup (AN-BP; 16.2 kg/m²) than the restrictive subgroup (AN-R; 19.6 kg/m²), which might mean that Nacc-DBS is less effective for weight restoration in AN-BP. Of note, the patients were described as “treatment-refractory” simply because many had BMIs < 15 kg/m² and all experienced amenorrhea [58]. However, their illness duration (mean 5.1 years; range 3–10) was markedly lower than other DBS studies in AN (e.g. [59•]) and it seems likely that some of these patients could have recovered with less invasive treatments.

Another group have recently published their ≥ 3-year follow-up data from a case series (n = 22) using DBS to the SCC in SE-AN [59•]. Mean BMI increased from baseline (14.0 kg/m²) to 1-year (p < 0.001; 17.5 kg/m²) and 3-year (p < 0.003; 16.3 kg/m²) follow-up. At the last follow-up (≥ 6 years), 3/15 patients reached BMI ≥ 18.5 kg/m² for ≥ 1 year, and an additional 3/15 reached BMIs between 17 and 18.5 kg/m². After 1-year of SCC-DBS, there was also a significant improvement in obsessive–compulsive, depression and anxiety symptoms that was not maintained at 3-year follow-up. Of note, 20% of the participants experienced hardware infections and 27% had prolonged surgical site pain, which De Vloo et al. [59•] considered may be overrepresented in SE-AN cases compared to other DBS patients. Indeed, a further report [60] described hardware infection, as well as the re-emergence of symptoms, in a case of AN-BP who requested that the stimulation be turned off 19-months postoperatively. The key risk factors for infections following DBS implantation remain largely unclear [61]. Without this information, preventative measures cannot be implemented and the burden of infection (e.g. prolonged antibiotic therapy) may compromise the risk/benefit ratio of this treatment.

Finally, DBS has been applied to the bed nucleus of the stria terminalis in a case of AN-R with significant improvements in weight and binge-purge symptoms [62]. The BNST is thought to be part of the extended amygdala and studies have increasingly used it as the stimulation target in OCD, with encouraging findings (e.g. [63]). Given that AN and OCD share some common neurobiological patterns (e.g. [64]), further examination of the BNST as a stimulation target for DBS in SE-AN seems appropriate.

Another invasive neuromodulation technique, VNS is an established treatment for several disorders (e.g. epilepsy), but it has yet to be used in EDs. However, transcutaneous auricular vagus nerve stimulation (taVNS) allows for non-invasive (i.e. without surgery) stimulation of the vagus nerve and is increasingly being studied in psychiatric disorders (see [65] for review). A recent study [66] applied 4 hours of taVNS per day over 9 weeks in a case series of mixed EDs (AN (n = 9), BN (n = 5), BED (n = 1)). Results showed a significant reduction in depression and anxiety symptoms [66]. However, ED symptoms and BMI values were not reported. Therefore, further examination of this procedure, with appropriate and comprehensive clinical outcomes, is warranted.