Surgery for depression is among the oldest procedures in neurosurgery, with reports of limbic leucotomy , and prefrontal lobotomy, extending back to the early 1940s [15–17]. Early attempts, however, were crude with targeting aided only by surface landmarks, and involving broad disconnections of frontal white matter tracts [16–18]. The introduction of stereotactic techniques, whereby lesions could be generated anywhere in the brain with millimeter-scale accuracy, led to the development of cingulotomy and capsulotomy , which specifically tackled disorders of mood. Cingulotomy involves bilaterally lesioning the anterior cingulate cortex approximately 2 cm posterior to the front of the corpus callosum. The procedure is generally well-tolerated and very safe, with several prospective and retrospective studies describing few adverse effects [19, 20]. A large proportion of patients also derive a significant clinical benefit, with rates of response (defined as a significant reduction in a depression ratings scale and improvements in overall functioning) ranging from 38 to 75 % [20–22]. Capsulotomy involves a lesion in the anterior limb of the internal capsule and is designed to influence fronto-subcortical circuits involved in affective regulation. In one recent paper, 8 patients with refractory MDD underwent bilateral capsulotomy, with 4 of them classified as treatment responders at 2–3 years follow-up [23]. Similar results were reported in a prospective case series of 20 patients, wherein 40 % were in remission and 50 % treatment responders, at a mean follow-up of 7 years [24].

Fifty years of clinical experience with lesions in psychiatric disorders have shown that: (1) Lesions in limbic circuits can be performed accurately and safely; and, (2) Such lesions can effectively influence pathological mood circuits, yielding positive effects in about half of otherwise refractory patients. The disadvantage of lesions , however, remains their permanence. One cannot titrate the clinical effect, ‘escalate’ or ‘reduce’ the dose, or change the location of the lesion once performed. Repeat procedures, particularly with cingulotomy , are routinely done but expose patients to additional surgical risk.

Several structures have been implicated in circuit models of MDD, including the medial and dorsolateral prefrontal cortex (mPFC; DLPFC), anterior cingulate cortex (ACC), nucleus accumbens /ventral striatum (NAcc/VS), as well as the amygdala. Neuroimaging has been the primer driver of progress in the investigation of mood circuitry, and both structural and functional abnormalities have been found in MDD patients. For example, ACC and hippocampal volumes are both diminished in patients with acute depression, with additional studies finding diffuse gray matter volume reductions. Pre-clinical models have further established a direct link between activity in the nucleus accumbens and both the enjoyment of reward (“liking”) as well as it’s pursuit (“wanting”) [25–27]. Imaging studies have linked activity in reward pathways to both the mPFC and the ventral tegmental area (VTA), a key brainstem dopaminergic center. The ability of VTA to influence both ‘top-down’ cortical centers, via mesocortical pathways, and ‘bottom-up’ regulatory centers, via mesolimbic pathways, has been proposed as a key maintenance system for depressed mood, and anhedonia specifically. For example, dysfunction in both or either mesolimbic or mesocortical systems can result in failure to anticipate or expect a rewarding outcome. Hypoactivation of NAcc in response to otherwise rewarding stimuli has been shown in conditions where reward deficits are well established, such as anorexia nervosa, further linking activity of this structure to affect-laden decisions [28, 29].

Studies performed in healthy subjects and unmedicated MDD patients have shown that activity in the ventral PFC, and subcallosal cingulate (SCC) specifically, is increased in response to sad stimuli in the former, and in the resting state in the latter [30]. This activity is attenuated with medical treatment of depression as well as with DBS in otherwise refractory patients [31, 32]. This has been found in both patients with unipolar depression as well as those with anorexia nervosa with comorbid MDD [33]. Such results have suggested that depression may be linked to a functional ‘decoupling’ of cortical-amygdalar projections, whereby increased activity in both regions leads to a failure of brain homeostatic control over affect [6]. Neurophysiological studies involving recordings directly from neurons are also informing the mechanisms of mood disturbance and the function of key limbic regions. For example, our group has shown using microelectrode recordings from single neurons in the SCC, that neurons in this region fire preferentially to negative pictures compared to neutral or positive ones [34]. Additional work in bipolar depression patients found that SCC neuronal populations undergo synchronization of firing immediately prior to making an emotional decision [35]. These results suggest that this region may be ‘programmed’ to respond to sad and depression-maintaining stimuli.

The target with the most experience to date is the subcallosal cingulate cortex (SCC). The SCC is a key node in the affective circuit, receiving inputs from a diverse range of structures including the medial prefrontal cortex, orbitofrontal cortex, anterior cingulate, nucleus accumbens and insula [36, 37]. Additional projections between the SCC and amygdala underscore the relationship, described above, between mood and it’s subcortical regulation by autonomic circuits [6, 36]. Functional imaging studies have shown that the SCC is closely involved in regulating emotions, and in particular negative emotions in both healthy subjects and patients. SCC activity has been linked to the degree of depression, and neurophysiologic studies have confirmed the preferential response of SCC neurons to negative stimuli and decision-making. As a result, the SCC has been proposed as an important node in mood circuitry and the first study of SCC DBS for refractory MDD was performed in 2005. This study included 6 patients and found that at 6-months follow-up, 4 were in remission, and that SCC perfusion, measured using PET , was significantly lower compared to baseline [32]. A larger study, published in 2008 in 20 patients followed to 1-year, found a 50 % response rate [defined as a >50 % reduction in Hamilton Depression Rating (HAMD) Scale scores] [38]. A multicenter study utilizing the same target found a more modest response rate of 29 % at 1-year, which increased to 62 % if treatment response was defined as an improvement in the HAMD by at least 40 % [39]. These results indicated that the majority of patients were either full or partial responders. Similar results were obtained by another group who reported a 50 % response rate with SCC DBS in otherwise refractory patients [40].