Ablation Surgeries for Treatment-Resistant Depression: A Meta-Analysis and Systematic Review of Reported Case Series

With high morbidity and mortality, depression is a critical public health issue affecting over 264 million people (https://www.who.int/news-room/fact-sheets/detail/depression) [1]. Depression is distressing and costly to the affected individual, care providers, and the society at large [2]. A diagnosis of Major Depressive Disorder (MDD) is made by applying the criteria set out in the Diagnostic and Statistical Manual of Mental Disorders (5th) [3] or the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) [4].

Multiple treatment modalities are available for MDD and include evidence-based psychotherapies, different classes of antidepressants, noninvasive neurostimulation (e.g., transcranial magnetic stimulation and electroconvulsive therapy [ECT]), and invasive neurostimulation (deep brain stimulation – DBS). For most patients these interventions used individually, in combination or successively are effective, but a group of individuals remain unresponsive. Treatment-resistant depression (TRD) is the inclusive term but without common definition used to describe this refractory form of depression affecting between 10 and 15% of patients [5‒7]. For this group, ablative lesion procedures [8‒10] remain as a last option.

Contemporary ablative surgical targets are mostly bilateral and involve lesions in the anterior limb of the internal capsule (bilateral anterior capsulotomy – BAC), the supragenual anterior cingulate gyrus and cingulum (bilateral anterior cingulotomy – BACING), and the subgenual anterior cingulate gyrus and subcortical orbitofrontal white matter (bilateral subcaudate tractotomy – BST) [11, 12]. The combination of BACING and BST is known as limbic leucotomy (bilateral limbic leukotomy-BLL) [13]; see online supplementary A (for all online suppl. material, see www.karger.com/doi/10.1159/000526000). Lesions are produced via thermal coagulation, radioactive yttrium rods [14], gamma radiation, and more recently, MR-guided focussed ultrasound (MRgFUS) [15]. All lesion targets claim some success, but cohorts are small, depression assessment instruments differ, and inclusion and outcome criteria and follow-up duration vary. Additionally, prior reviews of evidence, including narrative reviews, examining ablative treatment options, and providing future directions, report approximative values for treatment response rate, and no aggregate value has been reported to date [8, 16, 17]. To address these concerns and to identify the best single ablative procedure for TRD, we have conducted a first meta-analysis and updated systematic review of the published case series/cohorts of ablative surgery for MDD taken from multiple databases up until January 2022, following a post hoc analysis.

We have run a full-scale meta-analysis using the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement [18] standard as well as the Flow diagram generator (http://prisma.thetacollaborative.ca/generator).

The search of the electronic medical literature using key terms and Boolean operators was done via EMBASE, MEDLINE/PubMed, PsycINFO, and CINHAL and was completed in April 2020. Additionally, we have referred to published meta-analytic type review articles to substantiate and collect any literature overlooked by our initial search strategy, shown in online supplementary B [9]. A post hoc search of the literature, PubMed and EMBASE, was carried to examine for possible studies emerging since 1st of April 2020 to January 19th, 2022.

Key terms grossly included depression/depressive disorders/treatment-resistant, capsulotomy, cingulotomy, tractotomy, leukotomy/leucotomy, radiofrequency or radiofrequency lesion/ablation/ablative/thermocoagulation, focused ultrasound, stereotactic radiosurgery, undercutting, and focused ultrasound thermal ablation (MR-guided focused ultrasound). For the selection of key terms, we have reviewed peer-reviewed neurosurgical seminal works [10, 15], and referred to our team expertise [19].

We have a priori set our selection criteria to include only studies that (A) report on primary depression; (B) report on case series/cohorts [20]; (C) have used any of the ablative/ablation surgery methods such as capsulotomy, cingulotomy, tractotomy, leukotomy/leucotomy, radiofrequency, focused ultrasound, radiofrequency thermocoagulation, stereotactic radiosurgery, undercutting, or focused ultrasound thermal ablation (MRgFUS: a noninvasive method of producing highly focal intracranial thermal lesions); (D) had unique data; and (E) were only from centers that had developed a program for and hence expertise in ablative neurosurgical interventions for mental disorder (NIMD). We have excluded (A) reviews/meta-analyses, editorials, animal studies, and conference abstracts; (B) studies in other databases, or not captured by our inclusion criteria; (C) none-English language papers; and (D) single participant case reports given observer bias and lack of cumulative expertise and the strong possibility of chance finding alone.

We have set out to examine factors moderating treatment effect, including (A) neuroanatomical location, (B) outcome measures, (C) clinical variables, and (D) confounding factors. Neuroanatomical locations related to depression set for examination were the anterior limb internal capsule, cingulate gyrus, cingulum, subgenual prefrontal lobe and/or cortex, subcaudate region, Brodmann area 25, and paraterminal gyrus. Outcome measures of response rate, as well as method of assessment had to be included and were one or more of the following instruments: Beck Depression Inventory (BDI) [21], 17 item Hamilton Depression Rating Scale (HAM-D17) [22] also identified as Hamilton Depression Rating Scale (HDRS) and Montgomery-Asberg Depression Rating Scale (MADRS) [23]]. Clinical variables were duration of follow-up, presence of comorbidities (psychiatric and neurological) at the time of surgery, and presurgical use of medications of all classes and ECT. Confounding factors included (A) incomplete patient descriptions or data; (B) more than one surgery at the same neurosurgical target to expand the area of destruction and done at different time points hence re-exposing the patient to the risk/s of another open brain surgery; and (C) surgeries at more than one of the known neurosurgical targets done at different time points with the added neurosurgical risks as per factor B. BLL is not a confounding factor as this is preplanned ablative surgery at two of the known neurosurgical targets done simultaneously in a single neurosurgical intervention. Additionally, we have gathered information on demographics (age, sex, and socioeconomic status), and adverse effects.

A responder was defined as anyone who showed a ≥50% reduction in symptoms from pretreatment baseline. This is the standard metric of response in assessing the efficacy of antidepressants [24]. In the case of ablative surgeries, the time interval to determine responder status is stricter, usually set at 12 months and not the standard 8 weeks as used in medication trials [24]. This is done presumably because ablative surgeries are irreversible interventions, unlike medications that can be stopped in the event of significant adverse effects.

Two raters (AAS and an experienced research assistant) independently reviewed and classified the abstracts based on the selection criteria as including, excluding, or not certain. In the absence of concordance between raters, discussion with a 3rd party (TAH) was carried out to ascertain a decision on the study in question. Subsequently, the Kappa concordance rate was measured between the raters. Additionally, an experienced research assistant retrieved the data and double checked with AAS for consistency.

For these analyses, relevant variables were extracted. To compare between ablative surgeries, depression responder rates were tabulated or calculated if needed at 1 year (since most studies reported this value), and longer if available. For the meta-analysis, an aggregate event rate (response rate) was calculated and subsequently presented with a forest plot using the Comprehensive Meta-Analysis Software (CMA: Ver. 2.2) [25]. Analysis of heterogeneity was carried out with CMA using the I² statistics where a value of over 50% is considered as significant variation across studies irrespective of chance alone [26].

Given the nature and limited number of studies included in the meta-analysis, we have descriptively appraised the quality of studies.

Through the search of the literature, we obtained 254 possible studies. After duplicates and case studies were removed, we retained 228 studies (abstracts) for examination. Additionally, we examined in-hand available reviews of the literature on ablative surgeries for psychiatric disorders. This yielded two reviews for us to examine [8, 15]. Lastly, the senior author (TAH) identified additional studies missed by our e-search strategy. This yielded a case series [27] and 4 other studies [28‒31] that were missed by our initial search. A final total of 235 unique studies was available for this meta-analysis. After reviewing the abstracts and titles of the studies, 201 studies were excluded as they did not meet selection criteria at the screening stage.

After a second round of reviewing of the manuscripts, 21 more studies were excluded due to various reasons (e.g., grouping individuals with differing diagnoses, or no assessment scale with data was used). An additional 3 case reports were excluded. After applying our inclusion criteria, ten unique studies [31‒39] remained to be used for meta-analysis. Four studies [15, 19, 40, 41] contained duplicate data and this information was extracted and included when missing from the original studies, as shown in online supplementary C-Table 1 [50‒61].

The Montoya study reported on 21 patients (MDD n = 6, OCD n = 15). For this meta-analysis, only data pertaining to the 6 MDD patients were examined. This led us to a final total of 6 studies to meta-analytically examine. The Kappa concordance rate between raters was 0.89 (89%) (see Fig. 1).

Six case series/cohorts [31‒36] with 84 patients (mean age ranging between 30.2 and 43.2) for which data were available contributed to this descriptive meta-analysis. Percent females included in the studies ranged from 61 to 90. Two studies reported on the history of depression in years, and no study reported on socioeconomic status. The publication year of the studies ranged from 2002 to 2019. Ablation surgery published studies have emerged from Canada, the UK, the USA, and South Korea. The mean or exact follow-up duration ranged between 12 and 84 months (median = 30 months), shown in Table 1.

With exception to one study, shown in Table 1, all studies reported that patients, prior to surgery, had received both psychotropic medications and ECT. At times, although not specified, psychotropics included various antidepressants, antipsychotics, anxiolytics, mood stabilizers, and stimulants.

The collective reported comorbidities included anxiety, eating disorders, obsessive-compulsive disorder, panic disorder, pervasive developmental disorder, posttraumatic stress disorder, psychosis, schizoaffective, and substance use disorder.

Studies used HAMD-17 (50%), BDI (50%), and MADRS (33.3%) to assess depression severity. The total percentage do not add to 100 because 3 studies used multiple depression scales.

For the assessment of neuropsychological functions both computerized (i.e., Cambridge Neuropsychological Test Automated Battery: CANTAB) as well as series of traditional paper and pencil tests were used to measure verbal and nonverbal IQ and neurocognitive abilities including attention, executive functioning, learning, and memory, as shown in Table 2.

For unique single surgeries using a random effect model, the aggregate evidence from five studies (34/78) (excluding Montoya et al. [33], see reason under discussion section BLL) with a reported response rate of ≥50% reduction in the presurgical score by standard measurement of depression was 47% (95% confidence interval [CI]: 32–63%; I² −38.86%) (see Fig. 2: Forest plot).

For the various ablative procedures, BAC gives a responder rate between 25 and 60% with an aggregate response rate of 40% (95% CI: 13–75%; N = 2), BACING gives a responder rate between 25 and 60% with an aggregate responder rate of 45% (95% CI: 30–60%; N = 2), and SCT was 71% (95% CI: 32–93%; N = 1). There are insufficient data concerning BLL as an initial and only treatment.

See Forest plot in Figure 2 for distribution of scores as well as event rates per studies and aggregate value. Interpretation of this sub-analysis should be done with caution by considering the following narrative examination.

Following a comprehensive review of the published literature, only 6 studies met our inclusion criteria. The paucity of published studies indicates how infrequently ablative surgery is being utilized for the treatment of intractable psychiatric disorders.

Here we focused exclusively on the use of ablative surgery in TRD. We looked at outcomes at 12 months or greater and in studies that used a standard measurement of depression with a benefit that was ≥50% reduction in the presurgical score.

The available data suggest that, as a single neurosurgical intervention, BST provides the best outcome (71% responder rate). BACING provides a responder rate of 41–60% (aggregate: 44%). BAC gives a responder rate between 25 and 60% (aggregate: 40%). There are insufficient data concerning BLL.

Immediate postoperative side effects were common in all ablative procedures. Most of these are transient and not clinically significant such as headache, facial swelling or weakness, nausea, dizziness, psychomotor slowing somnolence, and apathy, as shown in Table 3. In contrast important clinical side effects do occur postoperatively and include grand mal seizure and urinary incontinence following BLL [33], intracranial abscess following BACING [34] and intracerebral hemorrhage following BAC [40]. More important are long-term sequelae present after a year or more. These include grand mal seizures after BLL [33] and BACING [34], urinary incontinence after BACING [34], and permanent hemiparesis following BAC [40]. The Dundee BAC group reported epilepsy in 1 case out of 20 patients (5%) at long-term follow-up (an average of 7 years postsurgery). However, 6 of these 20 patients also had subsequent BACING done on average 3 years after BAC [36]. It is not clear from the paper if the epilepsy was in one of these patients.

Some of these side effects are the known risks of any open brain surgery such as intracerebral hemorrhage and infection and are not unique to NIMD. However, epilepsy from planned cortical destruction (BACING and BLL) and urinary incontinence from targeted ablation of the anterior cingulate gyrus (BACING) [43] are very likely side effects from the specific location of these ablations.

Subjective memory impairment is described after BLL [33], BACING [34], and BAC [31]. This has not been confirmed by objective testing. Detailed neuropsychological testing has been done only by the Dundee and Vancouver groups. In the Dundee group, both for BACING [35] and BAC [36], there was no significant decline in any neuropsychological domain. The Vancouver group also showed no significant decline for any given neuropsychological domain. Neuropsychological testing, when analyzed per individual patient, showed isolated random impairments that involved inhibitory control, verbal memory, and semantic fluency [31]. The absence of neuropsychological impairment should not be surprising as neither BACING nor BAC involves injury to the neuroanatomical structures or their major connectivity that are responsible for these capacities: Papez’s circuit (hippocampus in particular) for recent episodic memory [44], the left anterior temporal pole for semantic memory [45], and the temporoparietal cortices for calculation and visuospatial ability [46, 47].

Assessment for personality changes after BAC was done in the Vancouver and Dundee groups [31, 36]. No groups who have done BACING have specifically assessed for personality changes nor reported any changes. This is an important issue as personality changes were a major problem in the early days of BAC. The Stockholm group [48, 49], using BAC for anxiety disorders and treatment refractory obsessive-compulsive disorder (TROCD), reported frontal lobe syndromes in 27% (7/26) to 43% (10/23) of their patients, which is the presumed reason that the Dundee group switched from BAC to BACING in 2000 [36, 42]. The Dundee group data were incomplete and confounded by the fact that some had undergone subsequent anterior cingulotomy. Clinical assessment did not suggest any significant personality dysfunction. By contrast, the Vancouver group was formally assessed for frontal lobe dysfunction prior to BAC and at follow-up using the Frontal Systems Behavior Scale [31, 41]. In 1 out of the 10 patients, there was clinically significant apathy. Decreased motivation and fatigue occurred after BAC [31], but both these symptoms overlap with depression itself and were difficult to ascribe to the surgery itself. The Vancouver group also described a condition known as the burden of normality; a constellation of psychological symptoms that surface as the individual adapts to being “cured”, which has not been reported by any other center with experience with NIMD [31]. With respect to BSCT, the data provided by the Korean group concerning side effects are minimal, reporting no operative mortality and 1 case of mild transient urinary incontinence.

Two additional post hoc analyses were carried out: (A) updating the literature search and results and (B) reviewing the side effects of thermal BAC when used for TROCD [48, 62‒65] to obtain ecological validity of the side-effect profile when this procedure was used for TRD. For this last post hoc analysis, we included studies in which BAC was the only ablative procedure performed, the mechanism of injury was thermal and the reported side effects ascribable to the capsulotomy. Studies that combined BAC with any of the other ablative targets (BACING or BST) and then aggregated side effect data were excluded. We also excluded BAC produced by gamma radiation which has a different mechanism of action with difficulty producing distinct lesions, has a delayed effect with the risk of radiation associated edema and delayed radionecrosis and cyst formation [49, 66, 67]. Postoperative adverse reactions were common and transient and included urinary incontinence and delirium. Among 119 (5 + 14 + 35 + 12 + 53) reported and analyzed TROCD cases (See online suppl. D), 1 patient had an intracerebral hematoma. Long-term side effects lasting longer than 12 months included weight gain and frontal lobe deficits (executive dysfunction, apathy, and disinhibition). There were no cases of seizures postoperatively or at long-term follow-up and no long-term cases of urinary incontinence. Persistent frontal deficits were the most significant long-term consequence and most likely related to the vertical height of the capsulotomy best captured in the Swedish data where this metric is provided. According to data presented in Table 1 of manuscript by Ruck et al. [49], in 75% (3/4) of cases this was 20 mm. At this length, capsulotomy would sever the dorsal associative pathways in the ALIC compromising frontal lobe functioning [68].

Additional abstracts were reviewed and screened using our selection criteria at post hoc. We have found two studies from two centers reporting on MRgFUS ablative capsulotomy for refractory TRD. One (Korean group) of the two studies reported on 4 cases [69], and the other study (Toronto group) reported on 6 MDD cases [70]. Davidson et al. [70‒72] have published several papers since 2020 with overlapping patient samples.

Two centers have reported 12-month outcomes for TRD utilizing MRgFUS. Where overlapping patient samples are reported, we have extracted the data from the most recent report and where the sample size was largest. The Korean group (n = 4) reported a 100% responder rate with no permanent neurological, neuropsychological, or psychobehavioral complications. Their capsulotomy height was 10 mm bilaterally [69]. Like the Vancouver group, they also noted the tristolytic effect. Tristolysis is a rapid reduction in depression surfacing as early as 1-week post-ablation when BAC is used to treat TRD, which may be the marker of an effective capsulotomy [19, 41].

The Toronto group’s (n = 6) [70] results have been less promising, reporting a responder rate of 20–33% at 12 months [70‒72]. They report no significant permanent neurological, neuropsychological or psychobehavioral complications. If anything, frontal lobe behaviors improved following treatment. However, the capsulotomy sizes are small. They do not report capsulotomy vertical height (z-axis as per MNI coordinates), only indicating a lesion diameter in the axial plane of approximately 7 mm. Moreover, they describe technical challenges utilizing this methodology. In 27% of cases, utilizing the trial’s sonication parameters, they failed to create a lesion on either side of the brain [70].

Treatment for depression unresponsive to psychotherapy and medication is now mostly focussed upon either noninvasive extracranial stimulation other than ECT (transcranial magnetic stimulation) or intracranial DBS. We consider here only invasive neurosurgical procedures and hence briefly review DBS, an invasive but nondestructive NIMD. Unlike ablation procedures, DBS is adjustable and reversible and does not suffer from the negative public bias and aversion that arose in the era of lobotomies [73]. DBS is currently being explored worldwide but is not without controversies, limitations, or crucial questions. A critical issue, also applicable to ablative surgeries, is which target site in the brain is effective or, if effective, superior in the treatment of depression.

The field was initially energized by DBS of Brodmann Area 25 (subgenual cingulate cortex-SCC) [74], but subsequently a multicenter, prospective, randomized trial of SCC DBS for TRD was discontinued after 75 patients because of futility [75]. To date, 5 sites in addition to SCC have been used as target for DBS in TRD – ventral capsule/ventral striatum, nucleus accumbens, ventral anterior limb of the internal capsule, superolateral branch of the medial forebrain bundle, and posterior gyrus rectus. Current evidence emerging from meta-analyses of DBS for TRD [75‒77] have concluded that none of these target areas have convincingly proved to be more effective than the other for TRD [78]. Evidence from the recent meta-analysis based on response rate shows variable benefits and adverse effects from various causes [77]. Furthermore, the long-term effectiveness of DBS has not been well-established. DBS thus remains an unproven but potentially useful and effective invasive but nondestructive neurosurgical intervention for TRD.

Ablative neurosurgical procedures for mental illness are now mostly done for TRD and TROCD, where there is, worldwide, the most cumulative experience and consensus for these destructive interventions. Uncommonly, ablative surgery has also been used to treat refractory anorexia nervosa using BAC [79] and refractory aggression using combination surgical targets that include BAC, amygdalotomy, and hypothalamotomy [80, 81].

In contemporary practice, there are only 4 currently accepted ablative neurosurgical targets for mental disorders namely BACING, BAC, BST, and BLL. Here we focused exclusively on the use of ablative surgery in TRD. The available data suggest that, as a single neurosurgical intervention, BST provides the best outcome (71% responder rate). BACING (aggregate responder rate of 44%) and BAC (aggregate responder rate of 40%) yield almost equivalent outcomes. On closer inspection, there are several difficulties with these data.

Subcaudate tractotomy for TRD was done between 1993 and 2002 by the South Korean group [32] and reported the best results with the response rate of 71% using the HAMD scores derived from the data provided in the report. First, it is unclear how many of these 7 patients also had cingulotomy, changing their surgery into a BLL. No clinical details are provided other than the diagnosis. Moreover, it appears that preoperative HAM-D17 may have been assigned retroactively. Taking this into account, it is unclear how to interpret the Korean data as to the true efficacy of BST as a standalone procedure in depression. BST results in extensive destruction of the subgenual prefrontal region and includes bilateral accumbens areas, the paraterminal gyrus (PTG), the subcallosal extension of the cingulate gyrus, and the medial and lateral orbitofrontal cortex [82]. Lastly the long-term morbidity associated with BST is not reported, only that there was no operative mortality.

BACING has been done by two centers: Boston and Dundee [34, 42]. The Boston group with the most published experience, n = 33, followed their patients prospectively and reported a 41% responder rate after a single bilateral cingulotomy in 17 patients. The other 16 received further ablative procedures. Nine patients received a second and then a third cingulotomy. The remaining 7 patients had a SCT after the initial cingulotomy, including one who had had 2 prior additional cingulotomies. With the addition of SCT, the procedure becomes a BLL. Despite these additional ablations, the overall responder rate of these 16 patients was 25%.

The Dundee group initially used BAC but, in 1997, because of side effects, switched to BACING [36, 42], using BACING exclusively as of 2000. In their most recent report (n = 8) [35], BACING as the only procedure was done in 5 patients. In these 5, they report a response rate of 60%. Their study was retrospective.

BACING as a standalone NIMD for TRD is moreover problematic as the exact target in the cingulate gyrus was and remains uncertain but is generally placed in the dorsal anterior cingulate gyrus (Brodmann area 24) [34], as shown in Figure 3a, or at Montreal Neurological Institute (MNI) average coordinates of 10, 11, 35 (right) and −8, 11, 36 (left) representing the x, y, z coordinates, respectively, as shown in Figure 3b [35]. Cingulotomy involves destruction of both the cingulate cortex and the underlying cingulum, the white matter association bundle within the cingulate gyrus [82]. The cingulum extends from the subgenual cingulate anteriorly to the parahippocampal gyrus posteriorly, terminating in the uncal region of the medial temporal lobe [83]. Fibers continuously join and leave the cingulum and few if any connections extend the entire length of the tract [84]. Interruption of the cingulum has long been felt to be essential for the effectiveness of BACING [35].

The Dundee group (n = 8) [35] found that more anteriorly placed lesions were associated with a better response. Their data also supported ablation of the cingulate cortex rather than the cingulum as the more important variable for benefit. Their data are in marked contrast to the Boston group (n = 33) [34], who in 10/33 placed 2 additional cingulotomy lesions 7 mm and 14 mm anterior to the first ablation. These additional anterior lesions provided no additional benefit in BDI scores. A single cingulotomy led to a group BDI score reduction of 49.6%, whereas 3 bilateral lesions led to a reduction in group BDI score reduction of 46.2% [34].

Taking all of the above into account, cingulotomy, as a single procedure, at best provides a 41% responder rate. The exact target in the dorsal anterior cingulate gyrus and whether it is cingulate cortex, cingulum, or both remains uncertain. As a procedure that includes cortical injury, there is a risk, (3–20%), of developing epilepsy as well as urinary incontinence because of injury to the anterior cingulate gyrus.

BAC has been done by the Dundee [36] and Vancouver groups [41]. The Dundee group reported retrospectively on 20 out of 23 patients done between 1992 and 1999. They report a responder rate of 25% at 12 months and 50% at long-term follow-up averaging 7 years. Their data are, however, confounded by the fact that for 70% of the sample, there was neither preoperative nor 12-month HRSD-17 score which had to be imputed from the MADRS scores [36]. Moreover, baseline preoperative data are missing for 10% of the sample (2/20), and at 12 months, no metric data of any kind are available for 40% of their cohort (8/20), as shown by Table 4, page 597 of the published manuscript by Christmas et al. [36] for imputed versus actual scores and number of patients for which data was available.

The Vancouver group has been doing BAC for TRD since 2000. Between 2000 and 2014, 10 patients were followed prospectively. Using the self-rated Beck Depression Inventory to assess depression, they reported a responder rate of 60% at 12 months [31]. BAC has the advantage that the target of ablation is clear and involves the anterior limb of the internal capsule (ALIC) in its midportion as seen in the axial plane as reported by Hurwitz, shown in Figure 3c [68]. There is also now good evidence that benefits come from ablating the ventral half of the ALIC as seen in the coronal plane [17, 31, 66, 68]. The BAC lesion can now be personalized to a given patient with no loss of benefit using presurgical tractography to target limbic pathways where the likely target is thalamic fibers that project to the paraterminal gyrus [19, 31, 68]. This technique avoids damaging the dorsal associative pathways in the ALIC responsible for executive abilities. Retained executive abilities preserve cognitive flexibility and may assist in the recovery and the reversal of the demoralization that accompanies chronic severe depression [31]. As an ablation that involves deep white matter fiber tracts, no patient treated with BAC has developed epilepsy, considering the uncertainty of the Dundee data.

BLL for TRD has only been done by the Boston group [33]. Although they report on 6 patients, BLL was the initial procedure in only 2 of this cohort. The mean follow-up for the entire cohort was 26 months. One of the TRD patients had no follow-up BDI score. The only patient that did have a follow-up BDI score was a nonresponder with a 20% improvement over the preoperative score. There are thus insufficient data concerning BLL to recommend this as an initial and standalone ablative procedure for TRD or to include it in the forest plot.

Based upon the available data and analysis, BAC as a standalone NIMD appears to be the most effective and safest of the ablative procedures with a well-defined target. One limitation of this conclusion is that data are analyzed on samples that are small and all are open label studies. Moreover, since BAC is the procedure done by our group, we have been mindful to ensure that the data have been analyzed as published.

Our conclusion is however supported by a recent transdiagnostic meta-analysis of all accepted ablative procedures (cingulotomy, capsulotomy, subcaudate tractotomy, and limbic leukotomy) in mitigating depression as a symptom cluster rather than as a diagnostic category. The effect size was greatest following BAC and smallest following BACING [85]. Another limitation of the current meta-analysis is omitting the examination of heterogeneity and bias on the effect size; however, it is arguable that the ratio of the sample size (number of studies, and number of patients) to number of variables would not have allowed robust analysis and results, given current state of the ablative surgery literature. By the same token, registration with PROSPERO: an international prospective register of systematic reviews, was omitted.

NIMD are interventions of last resort and performed by very few centers and on patients who are very ill. Interventional destructive open brain surgery from a patient’s perspective is rightfully considered highly invasive. The future is now with us. Focal intracranial thermal lesions are now possible noninvasively using MRgFUS. The initial results have been mixed. Technical challenges need to be overcome, but what is becoming clearer is that effective capsulotomies must ventrally sited and of a minimum length in the ALIC (6–11 mm) but not too long (>15 mm, see Table 1 in Ruck et al. [49]) to avoid impinging upon the dorsal associative pathway [49, 68, 69].

An ethics statement is not applicable because the study is based exclusively on published literature.

The authors have no conflict of interest to declare.

The author(s) (Trevor A. Hurwitz, Christopher R. Honey, & Amir Ali Sepehry) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this study received funding from the ERIN Fund, administered through the VGH and UBC Hospital Foundation, Vancouver, Canada.

All the authors have approved this paper’s contents and have agreed to Stereotactic and Functional Neurosurgery submission policies. Each named author (Trevor A. Hurwitz, Christopher R. Honey, & Amir Ali Sepehry) has substantially contributed to planning, conducting, and reviewing the underlying research and drafting this manuscript.

The study is based exclusively on published literature data for this meta-analysis. All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.