Molecular Psychiatry

https://doi.org/10.1038/s41380-021-01100-6

EXPERT REVIEW

Deep brain stimulation of the “medial forebrain bundle”: a strategy

to modulate the reward system and manage treatment-resistant
depression
1,2 2 2
Albert J. Fenoy ●
Joao Quevedo ●
Jair C. Soares

Received: 2 December 2020 / Revised: 21 March 2021 / Accepted: 1 April 2021

© The Author(s), under exclusive licence to Springer Nature Limited 2021

Abstract
The medial forebrain bundle—a white matter pathway projecting from the ventral tegmental area—is a structure that has
been under a lot of scrutinies recently due to its implications in the modulation of certain affective disorders such as major
depression. In the following, we will discuss major depression in the context of being a disorder dependent on multiple
relevant networks, the pathological performance of which is responsible for the manifestation of various symptoms of the
1234567890();,:
1234567890();,:

disease which extend into emotional, motivational, physiological, and also cognitive domains of daily living. We will focus
on the reward system, an evolutionarily conserved pathway whose underperformance leads to anhedonia and lack of
motivation, which are key traits in depression. In the field of deep brain stimulation (DBS), different “hypothesis-driven”
targets have been chosen as the subject of clinical trials on efficacy in the treatment-resistant depressed patient. The “medial
forebrain bundle” is one such target for DBS, and has had remarkably rapid success in alleviating depressive symptoms,
improving anhedonia and motivation. We will review what we have learned from pre-clinical animal studies on defining this
white matter tract, its connectivity, and the complex molecular (i.e., neurotransmitter) mechanisms by which its modulation
exerts its effects. Imaging studies in the form of tractographic depictions have elucidated its presence in the human brain.
Such has led to ongoing clinical trials of DBS targeting this pathway to assess efficacy, which is promising yet still lack in
sufficient numbers. Ultimately, one must confirm the mechanism of action and validate proof of antidepressant effect in
order to have such treatment become mainstream, to promote widespread improvement in the quality of life of suffering
patients.

Introduction rodent-defined pathway, as an attempt to define it from a top-

down or corticofugal (gold-standard) approach in the human
The medial forebrain bundle (MFB)—a white matter pathway cannot be performed. As it is a white matter bundle that
projecting from the ventral tegmental area (VTA)—is a projects from the VTA and extends to the nucleus accumbens
structure that has been under a lot of scrutinies recently due to (NAc) and to the prefrontal cortex (PFC), the nomenclature
its implications (by name and by physiologic presence) in the chosen was in reference to animal-defined histology and
modulation of certain affective disorders such as major studies on its role in motivation and reward [1].
depression. There is no clear correlate in the human of this In order to better understand this tract and the implications
of its modulation in treatment-resistant depression (TRD), we
need to place this white fiber bundle in the context of the
* Albert J. Fenoy
circuits considered responsible for depression.
albert.j.fenoy@uth.tmc.edu In this paper, we will discuss major depression in the
context of being a disorder dependent on multiple relevant
1
Department of Neurosurgery, McGovern Medical School, The networks, the pathological performance of which is respon-
University of Texas Health Science Center at Houston (UT
sible for the manifestation of various symptoms of the disease
Health), Houston, TX, USA
2
which extend into emotional, motivational, physiological, and
Faillace Department of Psychiatry and Behavioral Sciences,
also cognitive domains of daily living. We will focus on the
Center of Excellence on Mood Disorders, McGovern Medical
School, The University of Texas Health Science Center at reward system, an evolutionarily conserved pathway whose
Houston, (UT Health), Houston, TX, USA underperformance leads to anhedonia and lack of motivation,
 A. J. Fenoy et al.

which are key traits in depression. In the field of stereotactic MFB has also been termed the VTApp by Coenen et al. [1])
psychosurgery, specifically now deep brain stimulation as it courses antero-superiorly through the NAc to the
(DBS), different “hypothesis-driven” targets have been cho- PFC. As the VTA is considered a center of dopaminergic
sen as the subject of clinical trials on efficacy in the treatment- cellular activity, it is hypothesized that stimulation of
resistant depressed patient. The so-called MFB is one such the VTA outflow pathway may induce its clinical anti-
target for DBS and has had remarkably rapid success in depressant effect via modulation of the mesolimbic and
alleviating depressive symptoms, improving anhedonia and mesocortical dopaminergic pathways implicated in reward
motivation [2, 3]. We will review what we have learned and motivation and which are suboptimal or dysfunctional
from pre-clinical animal studies on defining this fiber tract, in MDD [15–18].
its connectivity, and the complex molecular (i.e., neuro- Understanding that MDD is a network disease implies
transmitter) mechanisms by which its modulation exerts its that MDD has significant biological components involving
effects. Imaging studies in the form of tractographic depic- a multitude of neurotransmitters that should be thought of in
tions have elucidated its presence in the human brain. Such the context of cognitive models, where genetics and
has led to ongoing clinical trials of MFB DBS to assess the inflammation also play a role [19, 20].
efficacy of anti-depressant response, which are promising yet
still lack in sufficient numbers. We close with future direc-
tions to confirm the mechanism of action and proof of anti- Network basis of disease
depressant effect.
Li et al. [21] have recently identified four relevant networks
which contribute to our understanding of affective disorders
Treatment-resistant depression—definition such as major depression: reward, affect, control, and default
mode networks [21]. These four networks play distinct roles
Major depressive disorder (MDD) is a global disease afflict- in depression, where complex, delicate interactions have
ing over 300 million people, with a lifetime prevalence of become perturbed by pathological dysregulation; a differential
15%, impacting the quality of life, public health, and multiple (network) involvement can be implicated based on the spe-
other socio-economic domains [4–6]. The majority of patients cific phenotype presented [1].
with MDD will respond to the conventional treatments of The reward network is a major driver of motivation,
medication, psychotherapy, and electroconvulsive therapy. behavior, and learning. Due to such intense motivational
However, up to approximately one-third of the patients can drive for the reward, this network has been also called the
end up being diagnosed with TRD, where they stay refractory “SEEKING system” by the neuroscientist Panskepp [22].
to all currently approved treatments [7, 8]. Patients with Dysregulation of this reward or SEEKING network has
MDD may become treatment-resistant as they may never get important implications in affective disorders, specifically
properly treated due to the heterogeneity of depression and depression [23–25]. In depression, the key symptoms of
because of the absence of clear biomarkers to help stratify anhedonia and hopelessness have been attributed to a dys-
their disease into a correct subtype; after each treatment functional reward system [1, 26–30]. Overactivation of the
failure, the chances grow toward more treatment resistance reward system can be observed during mania [31, 32] and is
[9, 10]. a hallmark in remitted depression [33].
The need to understand the biology of MDD and why The affect network is responsible for processing and
people might become treatment-resistant is growing [11–13]; regulating emotions [21]; it is the main system that deals
a better understanding of implicated neural circuits is man- with conscious and unconscious human fear [34–36]. In the
datory as it can lead to more informed ways to deliver novel world of affective neuroscience, the affect network/system
treatment, such as with neuromodulation or DBS. is intimately tied to separation distress, sadness, anxiety,
The supero-lateral branch of the MFB (slMFB, hereafter mourning, grief [37]. Feelings of rejection and isolation
referred to as MFB) has arisen as an efficacious target for have been well illustrated in imaging studies to involve
DBS as it is a point through which numerous projections activation/overactivation of this system [38], which also fall
pass linking other nodes in the neurocircuitry implicated along the lines of depressive symptoms.
with depression [1, 14]. This white fiber tract is potentially a The control network is a cognitive system responsible for
bi-directional pathway containing on the one hand myeli- motor program regulation and its limbic correlate, the top-
nated descending glutamatergic fibers from the PFC which down regulation of emotions [21, 39]. Patients suffering from
then enter the midbrain (and possibly the VTA). Under depression often suffer from a cognitive impairment which is
DBS, these fibers would be modulated and could then closely associated with the degree of emotional suffering;
activate dopaminergic neurons in the VTA, whose efferent similarly, patients suffering from OCD show cognitive
fibers are also contained in the MFB (which is why the inflexibility. Such importance of this system is seen when
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

understanding that complete treatment for depression by a of all participants post-implant demonstrated the inclusion
subcortical intervention (stimulation or lesioning) directed at a of four fiber bundles (forceps minor, uncinate fasciculus,
different network needs to be accompanied by psychotherapy cingulum, and frontostriatal fibers) as intended [46].
to access this top-down regulation [1, 40]. Such a fiber-tract-targeted approach has led to significant
The default mode network is only functional if the subject improvement in responder rates as well as long-term
is not performing any kind of action [1]. This network has responder rates of >50% for up to 8 years [46].
been associated with increased ruminations in depression.
Targeting the reward network

Modulation of networks in depression A parallel train of thought driven by hypotheses has led to a
desire to modulate the reward network, another critical
The idea that these networks are intimately involved with network in depression, dysfunction of which leads to
certain aspects of depression has spurred hypotheses that if reduced motivation (depressed mood or depression), reduced
they could be modulated, then perhaps there would be interest in previously pleasurable activities (anhedonia), and
symptom relief. As stereotactic neuromodulation of the hopelessness [29]. Research has shown that the expression
cortical-subcortical motor network is a fundamental concept of these functions is regulated in part by dopamine, pro-
in the treatment of movement disorders, targets for mod- jections of which originate in the VTA of the midbrain and
ulation in depression have been similarly developed over course superiorly through to the NAc and to the PFC.
the past 20 years, driven by the need to test hypotheses. One Early DBS studies focusing on anhedonia have targeted
such important network is the affective network. Helen several nodes/limbs of the reward network. The ventral
Mayberg pioneered the first clinical trial in TRD when she caudate/ventral striatum (VC/VS) target yielded a 53%
and her team targeted the subgenual cingulate gyrus (SCG, response rate (response defined as ≥50% improvement in
Brodman area 25, Cg25) and published their first case series depressive symptoms) in 17 patients [47] and a 21%
on DBS in TRD in 2005 [41]. response rate at 16 weeks in a randomized sham-controlled
Cg25 is believed to play an integral role in modulating multi-site trial of 30 patients [48]. A clinical trial of the
multiple networks (affect, control, possibly reward) as it is ventral anterior limb of the internal capsule (ALIC) target
involved in appetitive and aversive information processing, found a 40% response rate in 25 patients [49]. DBS to the
by regulating cortico-limbic circuits and structures asso- NAc yielded a 45% response rate at 1 and 2 years in 13
ciated with affective disorders, such as the amygdala, patients [50]. Less often used target regions have been the
anterior cingulate cortex, the caudate nucleus and the tha- bed nucleus of the stria terminals (BNST, ca. 13 patients)
lamus [42]. The drive for using this target was based on (at [51, 52], inferior thalamic peduncle region (ITP, ca. six
the time) revolutionary imaging work by Drevets et al. [43], patients) [53, 54], and the lateral habenula (LHb, ca. one
where it was found that Cg25 was hyperactive in severe patient) [55]; clinical trial ongoing (Goodman, personal
depression patients. Lozano and his team placed DBS communication). Refer to Fig. 1.
electrodes bilaterally into Cg25 with the goal of reducing its Instead of focusing on one particular node along this
activity and modulation of associated circuitry involved in pathway, it was hypothesized by Coenen et al. [1] that
depression; a remarkable anti-depressant improvement was elucidation of the axonal projection pathway as it emanates
seen in these TRD patients concomitant with normalization from the VTA, a key component of the reward system—
of Cg25 hypermetabolism [44]. Following such, over the could enable it to be useful as a target for neuromodulation.
years Cg25 has been the most commonly used target in Most of the evidence for this tract’s utility has come from
clinical trials of DBS for TRD, due to promising results in pre-clinical experimentation in the rodent, where the name
many patients. Unfortunately, a recent large multi-center of this VTA outflow tract—MFB—is used and has since
randomized sham-controlled clinical trial showed no sig- become prominent. As we will next describe this white fiber
nificant improvement leading to its discontinuation. The pathway in detail, it is important to differentiate its context
heterogeneity of targeting due to methodology among here as Coenen et al. [1, 56] have done to avoid confusion:
centers may have been one such reason, as patients at going forward, we shall delineate the MFB (mfb, in
greater than 1-year show benefit [45]; certainly, significant experimentation) in pre-clinical models vs. that in the
anti-depressant improvement has been seen when con- human (MFB, subject of DBS clinical trials).
sistently targeting elucidated fiber tracts adjacent to Cg25
[46]. Using a prospective connectomic approach to plan Description of the MFB
surgical targeting, Riva-Posse and colleagues found that
nine of eleven patients were responders at 1-year post-DBS, The mfb (in the rodent) is a very complex pathway containing
with six of them in remission [46]. A probabilistic tract map bi-directional, myelinated, and unmyelinated projections of
 A. J. Fenoy et al.

Fig. 2 VTA-NAc reward circuit. A simplified graphical representa-

tion of the major dopaminergic, glutamatergic and GABAergic con-
nections to and from the ventral tegmental area (VTA) and nucleus
accumbens (NAc) (dopaminergic = green; glutamatergic = red; and
GABAergic = blue) in the rodent brain. See text. The primary reward
circuit includes dopaminergic projections from the VTA to the NAc.
Fig. 1 Targets used in DBS for TRD. Historical targets are depicted Dopaminergic efferents also exist from the VTA to the medial pre-
that have been used in trials of DBS in TRD. Most are linked by a frontal cortex (mPFC), hippocampus (Hipp), and amygdala (Amy).
common white fiber bundle (green), the slMFB/VTApp as demonstrated The NAc receives innervation from glutamatergic monosynaptic cir-
by deterministic tractography. cuits from the mPFC, hippocampus, and amygdala. GABAergic pro-
jections also exist from the NAc to the VTA and to the lateral
hypothalamus (LH). The VTA receives glutamatergic inputs from the
diverse neurotransmitter systems, with over 50 different amygdala, lateral dorsal tegmentum (LDTg), lateral habenula (LHb),
and lateral hypothalamus, among others.
contributions in the lateral hypothalamic area including
monoaminergic, glutamatergic, and GABAergic neurons as
well as neurons containing active peptides such as neuro- hypothalamus provides direct excitatory inputs onto VTA
physin, oxytocin, vasopressin, serotonin, histamine, substance dopaminergic neurons (via orexin or glutamate) which
P, enkaphalin, and orexin, to name just a few [57, 58]. It is not promote reward. (Refer to Fig. 2).
a structure that can be visualized in its entirety using specific Of course, this inter-connected circuitry is highly com-
staining, although subcomponents of it, for example, the plex, as is the manner in which an organism can absorb and
dopaminergic pathways, can be identified histologically. The process aversive/rewarding stimuli that may influence
descending and ascending fibers between the mesencephalon future behavior [29, 60–63]. Although certain brain regions
and the forebrain connect different regions associated with may have been assigned specific behavioral functions (e.g.,
multiple functions such as motor, cognition, and emotion/ the VTA-NAc reward circuit, the amygdala is central to
mood processing [56]. The fibers of the mfb interconnect processing fear, the hippocampus in declarative memory,
nodes which include components of the reward network such the PFC in working memory, and executive control [29]), in
as the cingulate cortex, the NAc, and the VTA [29]; indeed reality, they have broader functions that are too difficult to
the VTA-NAc reward circuit in the rodent utilizes as its define—making them complexly, intimately entwined in
machinery fibers of the mfb as they course through the lateral regulating behavior. What we do know is that the mfb is the
hypothalamus and amygdala [57]. highway that connects these regions and is integral in pro-
All these structures are implicated in depression. The ducing an organism’s response (maladaptive in depression)
VTA is an important relay station in the reward circuitry to rewarding experiences.
that serves a central role in motivation and reward proces- The ascending A10 midbrain mesocorticolimbic dopa-
sing [59]. minergic fibers originate in the VTA and project via the mfb
The VTA region projects via the mfb to the NAc and via to the NAc and dorsolateral PFC [64, 65]. There is ample
a separate pathway to the PFC (see Russo and Nestler, 2013 evidence to suggest that this dopaminergic projection
[29]); dopaminergic neurons in the VTA project to the NAc, pathway is central to mediating “SEEKING” behaviors such
which is principally comprised of GABAergic medium as appetitive motivation and exploration, where, in the field
spiny neurons, and which also has GABAergic projections of affective neuroscience, these relate to emotions described
back to the VTA. The NAc also receives dense glutama- as “wanting”, “desiring”, or “anticipating” [66, 67]. Thus, it
tergic monosynaptic innervation from the medial PFC, is hypothesized that dysfunction of the dopaminergic pro-
hippocampus, amygdala. The VTA receives glutamatergic jections running via the mfb is responsible for the anhe-
inputs directly onto dopaminergic neurons from the amyg- donia and lack of motivation observed in depression [68].
dala, mPFC, and lateral dorsal tegmentum, whereas gluta- Recent technological advances such as in optogenetics
matergic projections from the lateral habenula synapse have made it possible to determine the role of dopaminergic
directly onto inhibitory GABAergic neurons, which inhibit and other specific neuronal cell types in reward-related
dopaminergic firing and promotes aversion. The lateral behavior that had not previously been possible. Optogenetic
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

stimulation of VTA dopamine neurons was found to be suf- In an effort to corroborate the behavioral effects of mfb
ficient to drive intracranial self-stimulation (ICSS) in rats and DBS, our lab has also investigated mfb DBS in a rodent
to promote a conditioned place preference [69], whereas, model of depression. Using a chronic unpredictable mild
conversely, stimulation of VTA GABAergic interneurons stress model (CUS) in the rat, 7 days of 8-h/day mfb DBS
disrupted reward and promoted a conditioned place aversion had anti-anhedonic effects (as assessed using the Froot
[29, 70–72]. Such confirmation of dopamine’s role in reward Loops® consumption assay, a variation of the sucrose
is further supported by observations that photoinhibition of consumption test [79]). As in other studies, gross locomo-
VTA dopamine neurons acutely reduces escape-related tion and body weight remained unchanged by mfb DBS.
behavior and sucrose preference, inducing a depression-like However, we also found that antidepressant effects can also
phenotype, whereas photoactivation of VTA dopaminergic be observed following shorter stimulation sessions. Rats
neurons rescues stress-induced depression-like behavior [73]. receiving 4 h of mfb DBS prior to testing showed increased
Furthermore, it has been shown that optogenetic inhibition of mobility in the forced swim test (FST) compared to con-
the VTA-NAc projection rapidly induced resilience, whereas trols, a marker of improvement of the depression-like phe-
inhibition of the VTA–mPFC projection promoted suscept- notype of helplessness. Here again, stimulation had no
ibility in mice [74]. impact on their general locomotor activity [80].
Dopamine is obviously a key contributor to reward and Taking a step deeper, in order to explore the role of
depressive-like symptoms, but it is not the only responsible dopaminergic mechanisms in mediating rewarding/anti-
mediator. As we discussed, the mfb is a pathway conducting depressive behavior, Furlanetti et al. employed a lesion
several varieties of neurotransmitters, of which dopamine is model of depression where 6-OHDA was injected into the
one. ICSS experiments in the rat have been able to help bilateral VTA targeting dopamine neurons of rats [81].
implicate certain structures necessary for rewarding effect Three weeks continuous bilateral mfb DBS was performed
but have been elusive in elucidating the key neuro- on lesioned and unlesioned control animals, where it was
transmitter mediator. found that animals with more severe loss of dopamine
neurons (i.e., more severely lesioned) had less of a response
to stimulation as compared to less-lesioned animals, which
Experimental evidence in animals— caused varied responses on their performances of stressful
modulation of the reward system vs. non-stressful behavioral tests. The data suggested that
mfb DBS can act via both dopamine-dependent and inde-
Electrical brain stimulation in the form of ICSS, pioneered pendent pathways [81]. Next, this group explored the role of
by Olds and Milner [75], has been an early useful animal D2 receptors in mediating behavior changes after mfb DBS
model to investigate reward and motivational behavior. By using an FSL rat model of depression following chronic
delivering a brief electrical pulse through an electrode D2 selective antagonist administration using [18F]DMFP
implanted in the lateral hypothalamic area, ICSS experi- PET to monitor changes in dopamine receptor availability.
ments showed rats were able to self-administer electrical When the D2 receptors were pharmacologically blocked
stimulation via lever pressing in a Skinner box [75]. The with raclopride, the depression-like phenotype was enhanced,
mfb fibers course along the walls of this area; thus, it can be whereas stimulation partially rescued social interaction and
considered that indeed ICSS here was inadvertently stimu- exploratory behaviors. The raclopride treated mfb DBS ani-
lating mfb fibers. mals had increased levels of D1 and D2 receptor mRNA,
More recently, electrical stimulation has more specifi- suggestive of stimulation mediated upregulation of dopamine
cally targeted mfb fibers in the rat to observe the behavioral receptors. This demonstrated that one potential mechanism of
response. Furlanetti et al. [76] implanted electrodes bilat- chronic mfb DBS is via the modulation of the midbrain
erally into the rat mfb slightly upstream from the VTA and dopaminergic system, including its post-synaptic receptor
delivered continuous, chronic (up to 6 weeks) DBS to show profile [82].
that chronic mfb stimulation is safe, also observing that c- Our lab has also shown that mfb DBS can produce an
fos, a marker of neural activity, was expressed in distant increase in protein levels of D2 receptors and dopamine
infralimbic and prelimbic cortices, and NAc. As these are transporter in the PFC, but in a non-depression control
all neural structures linked by the mfb, the conclusion was animal [80]. In another study using a CUS depression
that mfb DBS can modulate activity distantly [76]. Further rodent model, we recently reported that mfb DBS rescued
experiments by the same group showed that chronic mfb chronic stress-induced reduction in BDNF levels in plasma,
DBS resulted in improvement in stress-coping, exploration, cerebrospinal fluid, and the hippocampus. Furthermore,
and learning and memory performance [77]. Shorter and altered neuronal levels of inflammatory cytokines found
less chronic mfb DBS sessions were also capable of indu- after stress-driven chronic depression were differentially
cing antidepressant effects in the FSL model [78]. modulated in the presence of mfb DBS, showing that mfb
 A. J. Fenoy et al.

DBS has the potential to undo depression-related neu- natural reward-induced means, alluding to the congruency
roimmune aberrations [79]. of neuronal response [91]. Phasic activation of VTA
dopamine neurons was found to reverse a CUS-induced
depression-like phenotype [73]. Looking at influences along
Exploration into dopaminergic mechanisms pathways, as we had discussed Chaudhury et al. reported
of function that phasic activation of the VTA-NAc projection induced
susceptibility whereas inhibition induced resilience; inhibi-
If modulation of the VTA outflow tract is the mediator of tion of the VTA–mPFC projection promoted susceptibility
behavioral effect following mfb DBS in the rodent, then a to social-defeat stress [74]. In addition, phasic stimulation
further question arises as to the exact role of dopamine and of VTA dopamine neurons was also reported to influence
dopaminergic firing—how is it modulated by mfb DBS? mPFC activity and concomitantly behavior (e.g., grooming,
Microdialysis techniques have been used for decades to locomotion) [92].
evaluate dopamine release due to mfb stimulation in the These studies underscore the complexity and regulation
rodent, starting first with ICSS experiments. ICSS typically of dopamine neuron firing, which suggests that it is highly
employed a small tip electrode stimulating at frequencies context-dependent (i.e., level of stress) [93, 94]. The lessons
ranging from 50 to 100 Hz but at long pulse widths (2 ms) that we have learned from optogenetic studies are that there
[75, 83–85], which were found to increase the levels of exist at least two distinct dopamine neuronal populations
dopamine and dopamine metabolites in NAc [86]. involved in the mechanisms of the VTA: [1] a population of
Bregman et al. reported that high-frequency stimulation tonic dopaminergic neurons, firing at 1–8 Hz, and [2] a
of the mfb (130 Hz) that simulated clinical parameters population of phasic dopaminergic neurons, firing above
(100 μA amplitude, 90 μs pulse width) evoked an 10 Hz, with a longer pause between firing or inter-spike
antidepressant-like behavior in the FST of rats whereas interval [29, 95, 96].
lower 2 Hz stimulation did not; importantly, they were Such variations in firing patterns result in differential
unable to observe any significant dopamine release in the postsynaptic dopamine release, which therefore affects the
NAc [87]. This is probably due to the limitations in meth- physiological role of that specific neuron. Tonic firing is
odology, as microdialysis collections have a low temporal important but does not capture the whole physiological
resolution and as such cannot discriminate tonic vs. tran- picture. Thus, understanding how mfb DBS affects phasic
sient or phasic dopamine release [87]. Also, increased dif- dopamine firing requires the use of more technologically
ficulty arises in in vivo monitoring of dopamine because of sophisticated techniques with the improved spatiotemporal
naturally low concentrations with fast release and clearing resolution for detecting dopamine levels in vivo, currently
times from the extracellular space [88]. today in the form of fast-scan cyclic voltammetry (FSCV)
These contradicting results clearly show the importance of [97]. FSCV offers a temporal resolution of milliseconds,
choosing [1] adequate parameters of stimulation and [2] thereby elucidating the monitoring of both bursts as well as
adequate resolution of assessment technique. Clearly, the tonic stimulation-evoked dopamine release in a more
pulse width of stimulation is essential: longer pulse widths detailed fashion.
are considered to recruit more dopamine fibers when com- Initially, Ewing and colleagues reported an increased
pared to shorter pulse widths due to the chronaxie of the fiber extracellular dopamine release in the caudate nucleus/dorsal
bundle, including smaller axonal diameter and longer striatum after mfb DBS at 60 Hz [97]. Another important
refractory period [89, 90]. Moreover, if one looks at the types early finding using FSCV and single-unit recordings was
of firing patterns of dopamine neurons then one can better that 60 Hz mfb DBS induced increased spiking in substantia
understand why microdialysis is not sensitive enough to nigra dopaminergic neurons resulting in an increased
assess its release during stimulation of dopamine-containing dopamine release in the neostriatum, confirming the invol-
fiber bundles. vement of dopaminergic neurons in the neuromodulation
As we opened the discussion earlier to elucidate how attained by mfb DBS [98].
dopamine plays a role in the reward VTA-NAc circuit, the Over the years, different pre-clinical studies have ana-
field of optogenetics has helped scientists understand the lyzed optimal stimulation parameters, including stimulation
machinery of dopamine pathway-specific conduction [29]. patterns, of mfb DBS to evoke maximum dopamine release.
Work by Tsai and colleagues showed that selective stimu- It was found that the optimal frequency was between 50 and
lation of VTA dopamine neurons with a high-frequency 60 Hz; however, this frequency response was largely
phasic stimulation induced higher dopamine transients in dependent on the time-course of stimulation. Stimulation
NAc as compared to tonic stimulation [71]. Such dopamine paradigms of shorter trains with higher frequency (high
release produced by phasic stimulation was earlier found to frequency (e.g., >130 Hz), low pulse width (e.g., <60 μs)) or
be similar to dopamine released through non-stimulated longer trains with lower frequency (low frequency (e.g.,
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

<100 Hz), high pulse width (e.g., >100 μs)) were found to be evoked in the NAc in both groups, where variations in pulse
most effective, whereas longer trains of stimulation with width caused different evoked dopamine responses across
higher frequency were not effective as neurons became groups. However, the induced dopamine response was lar-
potentiated [84, 99–101]. The current intensity or amplitude ger and longer-lasting in the depression model as compared
also plays a role, where it was concluded that higher to the healthy controls [112]. This finding is suggestive of a
intensities (i.e., 200–800 μA) produced longer-lasting and divergence between the pathological and healthy state, due
larger electrochemical signals, whereas higher frequencies of to multiple possibilities [1]: physiological and anatomical
stimulation only increase their magnitude [102]. Regarding differences in the mfb tract, such as in fiber composition and
the pattern of stimulation, it was found that high-frequency density, which affect chronaxie [2]; differences in dopa-
phasic (burst-like) mfb stimulation was twice as effective as minergic neuron excitability [3]; differences in extracellular
low-frequency tonic stimulation for evoking dopamine dopamine availability [56].
release in the olfactory region of rodents [103]. The literature, thus, somewhat clarifies that the firing of
More recently, Klanker et al. [104] reported that mfb DBS dopamine neurons with resultant variation in dopamine
(120 Hz, 80 μs, 200 μA) induced an increase in dopamine release is affected by mfb DBS. Such a mechanism of action
release in the ventromedial striatum, where interestingly they is known to be fundamental to induce behavioral (reward-
were able to assess a differential response regarding neuronal ing, motivational) change. It is unlikely that mfb DBS
activation. Concomitant with mfb DBS there was a pre- causes solely phasic dopaminergic neuronal firing, and
sentation of reward, where it was observed that mfb DBS more likely that there is prolonged dopamine ramping to
increased the rate of firing of tonic DA release sustained for increase motivation over a protracted period of time [105].
about 40 s (with and without reward), whereas the phasic Hamid et al. [113] reported that the level of dopamine
release induced by reward delivery was unaffected by mfb assessed on different temporal scales, corresponding to
DBS [104]. extracellular availability and neuronal release, conveys a
This is an interesting finding: mfb DBS changed the rapidly evolving decision variable—the available reward for
baseline level of tonic firing neurons, somehow not affect- investment of effort—which is employed for both learning
ing phasic neuronal firing, which is induced by rewarding and motivational functions. Similarly, Hamid et al. [113]
stimuli. It is well known that the phasic spike activity of report that dopamine levels (and therefore, neuronal firing)
dopamine-containing neurons, and corresponding dopamine convey a single, rapidly evolving decision variable, the
transients in the striatum, are thought to underlie predictions available reward for investment of effort, which is
about future rewarding events, encoding positive and employed for both learning and motivational functions.
negative reward prediction errors [105–110]. Although it has been shown there is a sustained increase
However, although it is recognized that such transient in dopaminergic firing induced by mfb DBS [111, 112], the
dopaminergic responses to reward-predictive cues may exact route by which DBS causes such dopaminergic
initiate motivated behaviors, it is proposed that this sig- release remains unclear. As direct dopaminergic neuronal
naling alone cannot solely account for maintaining and activation seems unlikely using mfb DBS at clinical para-
directing sustained motivational drive during prolonged meters with short pulse widths, it is more probable there is a
periods of time [105]. Using FSCV, Howe et al. describe a complex circuit activated involving descending glutama-
“ramping” of dopamine signals in the ventral striatum, tergic or GABAergic neuronal synapsing [2, 114].
distinct from phasic or tonic profiles, which scaled with
both the distance and size of rewards. Such “ramping” is
strongly suggested to be ideally suited to maintain and Mechanism of action of MFB DBS—pre-
direct extended energy and motivation [105]. clinical data
In the study by Klanker et al. [104] in a non-depression
model, more insight is gained into how mfb DBS affects the As discussed earlier, the mfb is a complex fiber pathway that
dopaminergic system. Another study in the healthy pig in the rodent contains over 50 different substrates [57].
showed increased and sustained dopamine release in the Modulation of such via a DBS electrode that induces a
NAc following acute electrical stimulation (130 Hz, 0.25 ms, volume of tissue activated at a short pulse width and high-
3 V) of the VTA [111], which may correlate to a “ramping”. frequency influences all fibers within its volume, albeit those
Dobrossy and colleagues recently explored the effects of with the proper chronaxie, making the summation of
varying mfb stimulation parameters on dopamine release in stimulation-induced events to be multiple and simultaneous.
the NAc, using FSCV to compare the evoked dopamine This seems to exclude direct dopaminergic neuronal acti-
responses in both a depression model (FSL rat) vs. healthy vation. However, there is a strong focus on the reward
control [112]. Using clinically relevant mfb DBS stimula- system and its activation with an increase of dopamine, as
tion (130 Hz and 100 μs pulse width), dopamine release was confirmed by studies listed previously [78, 80, 82, 104, 112].
 A. J. Fenoy et al.

It is thus likely that there is a complex, indirect activation general. The rationale arose from the previously discussed
of dopaminergic neurons of the VTA via mfb DBS acti- pre-clinical animal studies, which also have tried to provide
vation of descending myelinated glutamatergic fibers from insight into the mechanism.
the PFC to the VTA, which causes an increase in dopami- Although a human is evolutionarily related to lesser
nergic transmission, including that in the NAc [114, 115]. species, they are not the same, and so parallelizations can
Conversely, as DBS has been shown to activate fibers both only be made up to a certain point. As expressed before,
orthodromically and antidromically [116], possibly there targets for modern stereotactic surgical procedures for
can be modulation of the ascending mesocortical pathway psychiatric diseases have arisen from analyses of historical
with antidromic modulation of the PFC, which would lesion surgery approaches, to specifically identify which
enhance the activity of the descending glutamatergic pro- nodes connected by specifically affected fiber bundles are
jections to the VTA and also produce increased dopamine most critical to induce desired effect/outcome. Imaging has
release from the NAc. Both could occur concurrently. For been a very useful tool to investigate the involved networks.
example, Taber et al. showed that direct electrical stimula- The use of diffusion tensor imaging (DTI) to perform
tion of the PFC resulted in increased dopamine in the NAc tractography has been essential to identify involved fibers
which dose-dependently increased with local NAc appli- and networks that were involved in such early lesion sur-
cation of metabotropic glutamate receptor agonists [117] as geries [120], as well as better, understanding the fiber
well as with direct intra-VTA application of the ionotropic pathways that were affected by older DBS targets (e.g.,
NMDA and AMPA receptor agonists [117, 118]. Cg25, NAc, VC/VS). Over time it has become clear that
Previous experiments using dopamine receptor and these targets are all linked by a single fiber pathway.
NMDA receptor antagonists in the setting of VTA DBS and Originally elucidated by tractography as a means to
functional magnetic resonance imaging (fMRI) showed that describe hypomanic side effects induced by singular contact
evoked blood oxygen level-dependent (BOLD) changes stimulation of the subthalamic nucleus (STN) (and its
appear to be mainly glutamatergic dependent [119]. Inter- resolution moving 1.5 mm rostrally) [32], Coenen and col-
estingly, in the study by Helbing et al. [119], dopamine leagues went on to characterize the anatomy of a newly
played a synergistic role with glutamate in eliciting BOLD described superolateral branch of the MFB (slMFB)
signal changes and was especially relevant in the setting of [121, 122]. Such nomenclature was chosen based on
continuous stimulation, suggesting that the effects of VTA reference animal (e.g., rodent) known anatomy and with
DBS are mediated by a combination of dopaminergic and similarities to its effect, namely modulation of positive
non-dopaminergic networks. In a healthy swine model, affect and the reward network. This slMFB was so named
VTA DBS has been shown to significantly increase BOLD as it is in distinction to the inferomedial branch (imMFB)
signal in multiple regions including the ipsilateral dorso- which is a branch of the MFB that traces the wall of the
lateral PFC, anterior/posterior cingulate, premotor cortex, third ventricle anteriorly until reaching the lateral hypo-
insula, primary somatosensory cortex, and the striatum thalamus [121, 122].
[111]. In this study, FSCV during VTA DBS also induced There has been much debate on this topic of nomen-
time-locked dopamine changes in the NAc, reinforcing the clature. The imMFB represents the traditional rodent mfb as
earlier suggestion that dopamine and other substrates (e.g., it follows a trans-hypothalamic route. As such, it does not
glutamate) mediate the effects of VTA DBS [111]. fully characterize the complete pathway in the human,
Thus, there are very likely multiple simultaneous circuits characterized by both the imMFB and the much larger
and transmitter pathways through which DBS to the mfb can slMFB branch. For this reason, Coenen et al. also state that
influence VTA neuronal dopaminergic release. Such impli- the superolateral branch in the human contains the VTA
cations on circuits and mediators have been explored in the projection pathway (VTApp) to describe these efferent
pre-clinical model. MFB DBS, in the human, can primarily be fibers [1]. As mentioned earlier, the slMFB is very likely, a
explored using imaging techniques—these can help identify bidirectional fiber tract system. What is next described here
and confirm implicated pathways, but can only intimate on is the use of diffusion tractography to elucidate this VTA
the mediator(s), within a modulated reward network. outflow tract and the reward network, as a means to better
understand the effects of DBS here in the human.
Use of Diffusion Imaging in humans—MFB pathway
elucidation
slMFB/VTApp description: discrepancies
Why perform DBS to the MFB? What does it modulate? with methodology
What is the mechanism? These are all spectacular questions
with answers that are inherent to the philosophy of network Evaluation of the fiber tracts of these networks described
modulation and the great mystery of how DBS works in above has been performed largely through imaging the
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

human brain. A more gold-standard approach to dissect and Although basic emotions are conserved across species
understand subcortical fiber pathways in the context of [37, 134] the human neocortex is far more complicated than
brain networks involved in psychiatric disorders like in other models (i.e., animals) where descriptions have been
depression employ a top-down or corticofugal approach [1] performed; diffusion tractography has helped to further
which can lead to in part confusing results [123–127]. describe complex human networks such as the reward
As Coenen et al. [1] explain, the reasons are probably network.
related to both brain evolution and methodology: neocor- Cortical (dlPFC, OFC, vmPFC) and subcortical regions
tical functional regions are network hubs (with task-specific (caudate, putamen, NAc, ventral striatum, anterior cingulum)
changing function) and as such receive fiber connections have been assigned as components of the reward network
from many subcortical networks, which are phylogeneti- [21, 135]. As Coenen and colleagues have described, the
cally much older (see Fig. 1 Coenen et al. [1]). Typically, VTApp is a massive fiber structure that connects the VTA
corticofugal dissection identified fiber pathways that con- with subcortical parts of the reward system (e.g., NAc, septal
nect the orbitofrontal cortex (OFC, or Brodmann’s area 11) region) as well as connecting to regions of the PFC that are
with the thalamus and the brainstem [126, 128–130]. involved with the experience and learning behavior associated
The same fibers are found when seeding (top-down) in the with reward (Brodmann’s areas BA8, 9, 10, 11, 47
ALIC [131]. [1, 122, 135]). However, the MFB cannot solely be repre-
Functionally, however, fibers from the brainstem and sented by the VTApp and contain only unidirectional,
thalamus most likely belong to different functional units unmyelinated dopaminergic axons, as it is clear that DBS is
(networks) and should not be lumped together, which able to incur an effect [136]. Again, this is likely due to
underscores the need to perform such dissection analysis modulation of descending glutamatergic fibers from the PFC
with apriori functional knowledge of involved fiber path- which activate dopaminergic neurons in the VTA, and over
ways and networks [1]. consecutive activation of the classical midline, pathway exerts
Realizing these inherent shortcomings associated with its effects.
tract-tracing, Coenen et al. proposed a corticopetal Despite the massive connection between VTA and PFC
approach to evaluate fiber tracts through imaging based on as described in previous contributions [1], in primates some
the rationale that [1] networks regulating distinctive emo- dissociative results have been reported. Injection studies in
tional behavior are largely conserved across evolutionary a standard text of descriptive macaque anatomy show
levels, where possibly evolutionary development steps and connection pathways between the precentral cortex, motor
functions are retained in the human subcortical anatomy cortex, and the midbrain [137] which also include cortico-
[22, 132] subcortical networks might be more easily iden- petal dopaminergic projection pathways [138]. On the
tifiable due to their simplicity relative to cortical networks contrary, this group reported a rich connection of the PFC to
[133], which would also lead to a more straightforward the STN [125] and defined in the macaque a “limbic hyper
identification of their associated cortical connections [1]. direct pathway” (lHDP) which realizes a connection to the
Points or nodes within such subcortical networks, when medial and anterior portion of the STN in the macaque but
interconnected to specific cortical networks subserving falls short of reaching the VTA. There is debate whether
specific purposes in behaviors conserved across evolution, this is entirely accurate, possibly influenced by viewpoint,
need to be well identified as they resemble intervention as some describe that these fibers are congruent with the
points that have been addressed in subcortical stereotactic VTApp which is the human analog of that pathway [122].
procedures like DBS or lesioning [1]. As described elsewhere [1], such an IHDP to the STN is
So, instead of looking top-down, Coenen et al. propose subjectively defined in an atlas of comparative anatomy
that one should investigate the subcortical nodes or “hubs” [139] of only reaching the medial STN and not the VTA,
to identify specific subcortical-cortical networks. Identifi- despite the fact that fibers medial and outside the STN
cation of projection pathways that bidirectionally connect (including the VTA) were found present in the referenced
cortical with subcortical structures including basal ganglia, macaque fiber injection studies [125]. What can be objec-
thalamus, midbrain, brainstem, and cerebellum [133] has tively stated is that there is a certain lack of “gold standard”
been useful to depict functional connections and network evidence (i.e., histological tract-tracing in the human) which
descriptions, such as the cortico-striato-thalamo-cortical serves as a comparison to correctly identify and name this
loop for motor function [39]. fiber pathway arising from the VTA.
However, there is a much less clear assignment of pro- Adding to the debate, Hurwitz and co-workers have
jection pathways to specific cortical networks such as reported a ponto-frontal pathway which they suspected to
reward, affect, control, and the default mode, leading to be antidromic to Arnold’s bundle; in three of five patients
poorly understood identification of relevant subcortical who had undergone anterior capsulotomies, there was
anatomy that could guide improved therapeutic options. observed here a novel T1 MRI signal, indicative of
 A. J. Fenoy et al.

Fig. 3 Illustration of the slMFB/VTApp. Illustration of the slMFB/ the target region (yellow) in several views (oblique sagittal (D), cor-
VTApp by deterministic tractography, (A) mid-sagittal, (B) coronal, onal (E), and axial (F)), with clear fiber projections to the prefrontal
and (C) axial views depicting the fibers targeted by electrode after and orbitofrontal cortices.
placement. Larger depiction of targeted fiber pathways after seeding

neurotransmitter deposits [140]. This is in line with a depression with HDRS-29 [144] score > 21, MADRS > 21,
description of dopaminergic projections reaching the PFC GAF [145] < 45, refractory to multiple medication regimens
and motor cortices from the ventral midbrain [138]. and a trial of ECT; no personality disorder as assessed by
Regardless, the discrepancies in anatomical descriptions SCID-II [146] and MCMI-III [147] and no neurological
should be evaluated in the context of the clinical effect in comorbidity. Patients were required to maintain their same
the human: modulation of this fiber tract, which can be medication for 6 weeks before and 6 months after surgery.
defined corticopetally as a projection from the VTA, indu- Psychiatric assessments were performed on a weekly basis
ces an anti-depressant effect likely through reward network by a psychiatrist independent of the programmer starting 1-
activation. week following implantation for 52 weeks.
Our target was defined using deterministic tractography
after obtaining diffusion MR imaging, as the target fibers
Clinical trials of MFB DBS cannot be seen on conventional imaging: using a midbrain
seed region posterior to the mamillary bodies, anterior to the
The first clinical trial of slMFB DBS in TRD patients arose red nucleus, and lateral to the VTA, thereby approximating
from the same group that first serendipitously observed the outflow tract of the VTA (just superolateral) [2, 32].
reversibly-induced hypomania in a ventral DBS electrode Each target point is independently calculated for each
contact targeting the STN in Parkinson’s disease [32]. In their patient’s individual anatomy; an indirect target coordinate is
initial report, Schlaepfer and colleagues found 6 out of 7 not used. Refer to Fig. 3.
patients responders described as >50% decrease in severity as In our initial publications, we reported on our first six
defined using the primary outcome measure of the patients and observed a similarly quick onset of ant-
Montgomery–Å sberg depression rating scale (MADRS) [141] depressant effects and sustained >70% response rate at
within one week of stimulation. 50% of patients continued to 1-year follow-up [3, 148]. As the initial German study turned
have sustained remission at 1-year which continued until the on the device immediately after implantation, a delineation
4-year follow-up [2, 142, 143]. between an insertional, placebo, and stimulation-induced anti-
When initially published in 2013, the results of such depressant effect could not be made in the immediate post-
rapid and sustained anti-depressant slMFB DBS were not operative period. Thus, we instituted a single-blinded sham-
seen before in any DBS clinical trial for TRD patients. In an stimulation phase for the first 4 weeks post-implantation in an
effort to independently replicate these results, our group effort to control for non-treatment effects. An “insertion
began our own clinical trial on MFB DBS in TRD using effect” was seen to be significant in other DBS studies with a
very similar inclusion/exclusion criteria as used for other similar 4-week sham stimulation period [45], possibly owing
DBS trials in depression [2, 44, 45]: severe unipolar to acute inflammatory mediators [149], or glial released
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

neurotransmitters [150, 151]. It is possible that whatever including BA 10, and involvement of certain white matter
neurochemical change that occurs from insertion and which tracts is a commonality universally required for anti-
induces mood improvement may take longer than 4 weeks to depressant effect, as one non-responder demonstrated much
subside. Interestingly, although such “insertion effects” may less extensive connectivity from the VTA to the OFC [3].
provide a confound, as there was a significant mean decrease Such an observation has prompted us to augment our
in MADRS scores between baseline pre-op an end-sham inclusion criteria (in recent patients) to show the demon-
stimulation, after only 1-week of active stimulation there was stration of adequate connectivity from the target region to
again a significant decrease in MADRS score from the end of the OFC.
sham with a more marked difference from baseline (mean To date, we have enrolled 12 patients into our ongoing
change = 15 points, [3]). It was thought, however, that a clinical trial to study the effect of MFB DBS. Currently,
longer sham stimulation phase would better separate any in this next phase of the study, we changed our protocol to a
insertion effect from treatment effect, as it is possible that 3-month single-blinded sham-stimulation phase, based on pre-
such an effect could last for up to 3 months [149]. However, it clinical data. It is unclear if this sham stimulation duration is
is highly unlikely that any response rate seen that is >50% at adequate to observe an attenuation of the initial insertional
1-year is attributed to insertion or placebo. Placebo effects are anti-depressant effect before stimulation onset; future studies
more probable at the beginning of the intervention and larger employing longer lead-in durations or even a discontinuation
in more invasive interventions [152], although patients with phase are under consideration to further prove treatment
severe TRD are less likely to develop such effects due to their efficacy [143]. Please refer to Table 1 for studies on MFB
history of non-response to many treatments [143, 153]. DBS across centers.
Nonetheless, there is a certain degree of expectation in these
patients that must be careful separated so as to not confound
true treatment effect. Lessons learned from stimulation
As safety had been established, to continue the exploration discontinuation
into the efficacy of slMFB DBS and to better identify true
treatment effects, the German group began a “Gateway trial” Inadvertent discontinuation and resultant return to baseline
of 12 months duration [143]. A double-blind randomized- pre-DBS conditions is proof in-of-itself of anti-depressant
control (DBS active vs. sham) condition was implemented in efficacy. Through the years, it has been observed that when
16 patients for the initial 8 weeks after surgery, at which point patients had inadvertent or voluntary discontinuation of
blinding was lifted and all patients received active stimulation. bilateral MFB DBS, there was a quick return within days of
Similar to the first study, all 16 patients reached the response depressive and/or anxiety symptoms [154]. This situation
criterion, and most patients (n = 10) responded within a week; occurs years after surgery and prompts return to the psy-
50% of patients were classified as remitters after 1-year chiatrist or neurosurgeon so that replacement of the depleted
of stimulation (MADRS < 10). Both groups (active/sham) implanted pulse generator (IPG) can be performed. Such
demonstrated an antidepressant effect; as in our study, observations have occurred in both the German study trial
patients had an additional antidepressant effect after initiation as well as in ours. Replacement of the IPG led to a return of
of stimulation [143]. Thus, it appears that the sham- symptom improvement, usually within days but sometimes
stimulation phase of 2 months is not long enough to sepa- within weeks [154].
rate treatment from other effects. DBS-induced changes in brain functioning seem to be
At our institution, we have completed our initial transient, indicating a functional and temporal phenomenon
assessment on ten consecutive patients to prove safety in [41, 44]. It had been observed that cerebral metabolism
MFB DBS. Although it is far too early to tell, results on the decreases occurred 48 h after DBS discontinuation in 7
anti-depressant effect are promising and awaiting publica- TRD patients [155]; as these were not associated with the
tion. We are continuing our longitudinal analysis for a development of clinical changes, it is suggested that such
period of 5 years from implantation. Our initial patient metabolic changes precede clinical manifestations. Indeed,
implanted in 2014 remains in remission, suggesting long- as we have not known the exact time when stimulation
term efficacy as seen by Coenen and Schlaepfer’s group accidentally ceased, a return of depressive symptoms likely
[142, 143]. As discussed [3], responder patients demon- did not happen immediately but took a few days to become
strated extensive fiber tract connectivity to the OFC and manifest [154].
Brodmann Area (BA) 10, with common involvement of the Interestingly, sometimes, as we have observed, if there
forceps minor among other white matter tracts connecting was a delay between IPG end-of-service and replacement,
the target region to the OFC, which is seen in other DBS re-initiation of the latest optimally employed DBS para-
studies in TRD involving other targeted locations [46]. It is meters causes side effects of diplopia that were not present
possible that such modulation of specific cortical structures, previously.
Table 1 Clinical trials of MFB DBS––based on location (as multiple papers have arisen from two key centers).
Center (surgeon/ Number of Design MADRS at time of Outcome Stimulation parameters Comments
psychiatrist) References patients inclusion
(mean ± SD)

Freiburg, Germany 24 total

(Coenen/Schlaepfer)
Schlaepfer et al. [2] 7 Case series; no 29.9 ± 8 6/7 (85%) responders, 4/7 2.86 mA mean amplitude First descriptive case series on
sham phase (75%) remitters at (130 Hz, 60 μs), bipolar slMFB DBS
12–33 weeks
Bewernick et al. [142] 1 new, f/u of Same 30 ± 7.4 6/8 (75%) responders, 4/8 Using area under the curve analysis, 7/
original 7 = 8 (50%) remitters at 52 weeks 8 had significant reduction in most
months at 200 weeks
Coenen et al. [143] 16 Case series; DB phase 29.6 ± 4 No difference in sham vs. 2.1 mA mean amplitude Sham phase not long enough to
(sham vs. DBS ON) for DBS ON for initial 8 weeks; (130 Hz, 60 μs) initially, separate treatment from
8 weeks post-op 100% responders, 50% 3.0 mA overall at insertional effect
remitters at 52 weeks 52 weeks
Houston, USA (Fenoy/ 12 total
Soares)
Fenoy et al. [148] 4 Case series; SB sham 34 ± 3 2/3 (67%) responders and 3.2 V mean amplitude Preliminary analysis of ongoing
phase for 4 weeks remitters at 26 weeks; 1 (130 Hz, 60 μs), bipolar, clinical trial
dropout 4/5 (80%) responders overall
and remitters with
Fenoy et al. [3] 2 new, f/u of Same 35 ± 2.8 >70% MADRS reduction at 3.8 ± 1.2 V mean 52 weeks summary of results of initial
original 4 = 6 52 weeks amplitude (130 Hz, 64 μs 6 pts; non-responder lacked
(SD 8.4)), bipolar, tractographic evidence of adequate
overall frontal connectivity to target
Unpublished data 4 Same Conclusion of initial enrollment of 10
pts safety trial, results pending
publication.
Unpublished data 2 Case series; SB sham New series begun with 12 weeks SB
phase for 12 weeks sham phase, enrollment ongoing
OTHER
Umea, Sweden 1 Case report 43 Non-responder at > 3 V (130 Hz, 60 μs) Comorbid OCD and anorexia; unclear
Blomsted et al. [164] 10 months; received BNST tractographic definition of target
DBS at 2 years post op
Toronto, Canada 2 Case series; at 6 months – 0/2 responder (0%), 0%, 5% 1.5 V (130 Hz, 60 μs); Tractographic targeting and
Davidson et al. [165] crossover (2 weeks ON, improvement in HAMD-17 weekly increase by 0.5 V stimulation strategies unclear
2 weeks OFF) scores at 6m until diplopia
Chicago, USA 1 Case report; 4 weeks 28 on HAM-D; Decrease in HAM-D (10 pts) 1.5–3.5 V (140 Hz, Novel use of CAT-DI, allowing 93
Sani et al. [160] post-op OFF, DB ON/ 76.4 on CAT-DI at 24 weeks; non-responder at 60–150 μs) variable over acquisitions of data over 6 months
OFF intervals 26 weeks 6 months
Response defined as ≥ 50% decrease in baseline MADRS; Remission defined as MADRS ≤ 10
SB single blind, DB double blind, SD standard deviation
A. J. Fenoy et al.
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

Induction of diplopia is one of the key intraoperative Assessment strategies in MFB DBS for TRD
findings that are used to guide the proper placement of the
electrode; the oculomotor nerve (CNIII) traverses the lateral Although the MADRS and/or Hamilton depression rating
pigmented nucleus of the midbrain as part of the VTA, scale (HDRS) have been the principal primary outcome
indicating the entry into the lateral part of the VTA [143]. measures in most clinical trials of DBS for TRD, this is
Too low a threshold of such side effect causes more inherently a subjective set of questions answered by
superficial positioning and/or different use of more rostral the patient. Such tests lack objectivity, as most patients
DBS electrode contact pairings. (including non-responders) are observed to have a renewed
Ensuring that the threshold of oculomotor side effects motivation to engage in at least some previously enjoyable
is high enough (>3 mA) usually is interpreted that there activities. Also, the employment of these outcome measure
can be utilized a large enough current (and resultant questionnaires sometimes is not often enough to capture
volume of tissue activated) to engage the MFB fibers for daily changes or trends; it has been suggested to augment
the continuous antidepressant effect. The use of bipolar this data collection by either more frequent testing using
stimulation mode usually reduces the current spread virtual methods or using new methods such as frequent
inferiorly to avoid oculomotor nerve activation. During computerized adaptive testing (CAT) of depression severity
optimal parameter selection that occurs within the first (CAT-depression inventory) that a patient implements from
few weeks of continuous stimulation, sometimes slight home [160].
diplopia can be observed (regarded by the patient as As suggested by others, an area under the curve analysis
“seeing double”, or vertiginous feelings when watching could also be helpful [142, 161, 162] as it conveys infor-
moving objects); usually, accommodation to this phe- mation about the antidepressant effect throughout the entire
nomenon occurs and use of the higher amplitude of cur- time of treatment (daily, weekly, monthly) rather than at a
rent can be used to incur the anti-depressant effect. It is given endpoint, but each day throughout the course of the
possible that re-initiation of stimulation after a delay has study. Using this method, both our group and the German
caused these oculomotor nerves to de-accommodate to study found a significant antidepressant effect in responder
stimulation, which requires a lower starting amplitude and patients for most months in the study analysis as well as a
more protracted recovery, as current needs to be ramped significant antidepressant effect comparing mean baseline
slowly to pre-discontinuation levels. with mean stimulation areas under the curve on the group
Also, conversely, in other patients experiencing a sti- level for each month of the study [3, 142]. As in other DBS
mulation OFF period years after implantation and where a depression studies, we also feel that the use of partial (25%),
return of stimulation did not cause diplopia, previously used clear (50%), and strong response (75%) differentiations
optimal parameters failed to return the patient to their would more adequately describe the treatment response
“normal” anti-depressed level; a higher current amplitude other than “non-responder/responder”. A 50% cut-off level
was necessary to re-establish a satisfactory level of anti- is an arbitrary assignment that does not reflect the benefit
depressant effect. Along these lines, it is possible that from therapy for a treatment-resistant patient [142, 163]
gliosis which slowly forms surrounding the electrodes after where a smaller effect might be associated with a substantial
chronic stimulation changes the impedance of the target gain in quality of life.
neural tissue [151, 156, 157], thereby requiring successfully Small case studies of MFB DBS with 1–2 patients have
higher current amplitudes parameters to incur effect. also been conducted in some locations, as explorations,
Such an observation begs the question of there is some looking at new methods of evaluation [160] or different
“network rheostat” that gets altered when a continuous comorbidities [164]. Recently, the Toronto group published
input of stimulation for years is suddenly removed, absent an article on 2 TRD patients who underwent MFB DBS and
for a period, and then re-initiated. Apparently, what we reported no clinical response as measured using the
have learned is that continuous stimulation is mandatory HAMD-17 scale [165]. The reported results and nature of
for effect. Plasticity of neural circuits does indeed occur, as this study are skeptically regarded, however, due to the
in dystonia, where stimulation takes sometimes months to omission of descriptive detail on planning, surgical, and
cause an effect and where withdrawal of stimulation stimulation technique: namely the lack of tractographic
slowly diminishes effect [158, 159]. Although inferences representation of the target, intraoperative verification of the
can be drawn, future functional imaging (e.g., fMRI, PET) target region with microelectrode recording, and observed
analysis will hopefully let us arrive at more concrete test stimulation responses.
conclusions. Such studies are underway, however, the It is thus clear that the surgical team must take extreme
small sample size in which we can investigate circuit care when planning a stereotactic intervention like DBS, and
plasticity in such patients makes such analysis difficult to in experimental clinical trials such as for TRD, with a very
perform quickly. small denominator, strict attention to detail is paramount.
 A. J. Fenoy et al.

As stated elsewhere [3, 143, 148], when targeting the Going forward, it is critical that each patient be viewed
MFB (slMFB), the key points of intraoperative identifica- very closely as many needs to be learned from each one’s
tion and implantation are: [1] adequate diffusion tensor response to guide future treatment and/or method of analysis
(deterministic) tractographic depiction of the target fibers to describe effect so as to perform the correct pre-and post-
(MFB); [2] multiple simultaneous microelectrode recording implant imaging to use for proper elucidation of structural,
to exclude nuclear structures (STN, substantia nigra, red functional, and electrophysiologic connectivity. Clinical trials
nucleus) from stimulation [3]; intraoperative test macro- need to be designed appropriately to separate treatment effect
stimulation through the microelectrode (cannula) showing from placebo/insertional effect and use the best possible cri-
appetitive motivation response, autonomic response (heart teria to categorize treatment response. In conjunction with this
rate increase), and a high threshold for oculomotor effect is the proper design and utility of pre-clinical studies to run in
(diplopia). a parallel fashion to explore the proposed mechanism
Final intraoperative verification of electrode location at (dopaminergic and/or glutamatergic) of modulatory effect.
the targeted fiber tracts is performed by intraoperative 3D Thus, there is still much left to discover regarding network
imaging (e.g., O-Arm, Medtronic, Minnesota, USA) and modulation before treatment of TRD through DBS is proven
then reconfirmed with postoperative helical computed to be efficacious. TRD numbers continue to climb, so it is
tomography (CT). mandatory that at some point in the future this treatment is
universally adopted worldwide. Likely, the choice of the
therapeutic target will be directed based on the disease sub-
Conclusions and future directions type and even individual phenotype, where surgical planning,
stimulation paradigms, and concomitant cognitive/behavioral
There is much need to articulate that the VTApp/MFB fiber therapy will be performed in a standardized, algorithmic
tract is an integral thruway that interconnects key regions fashion tailored to the individual. In terms of future patient
necessary for motivation and reward, not only in lesser selection, anhedonic patients will likely benefit the most from
animals but in the human. Demonstration of such integrity/ MFB DBS through increased access to the reward system.
essential nature can currently be seen clinically in the Observational methods to record meaningful results also need
human, where its modulation leads to increased motivation to be standardized across centers. It is in this way that DBS
and reversal of anhedonia, which are fundamental pillars of for TRD can be a powerful therapeutic option that can sig-
an anti-depressant response. Yet, there is no clear proof of nificantly improve quality of life.
what we are modulating, the mechanism of action, how
dopaminergic fibers/system becomes modulated given the Acknowledgements The Center of Excellence on Mood Disorders is
DBS parameters used for clinical effect. The answer to these funded by the Pat Rutherford Jr. Chair in Psychiatry, John S. Dunn
Foundation, and Anne and Don Fizer Foundation Endowment for
questions will come in future studies using functional Depression Research.
imaging as a way to observe subcortical-cortical network
reactivity and response to modulation. We have already
Compliance with ethical standards
used diffusion imaging to illustrate the structural con-
nectivity of the modulated fiber pathway of the VTApp/
Conflict of interest JQ reported no biomedical financial interests or
MFB from the target region through to the NAc and onto potential conflicts of interest. AJF serves as a consultant for Med-
the PFC/OFC. fMRI, especially in the resting state, can be tronic, Inc. JCS receives grant/research support from Bristol-Meyers
an essential way to differentiate a network activity between Squibb, Forest Laboratories, Merck and Elan Pharmaceuticals, and
DBS states (ON/OFF) which can be viewed concomitantly serves as a consultant for Pfizer, Abbot, and Astellas Pharma, Inc.
with clinical observations of behavior (level of depression/
Publisher’s note Springer Nature remains neutral with regard to
anti-depressant response), to see which regions are critically
jurisdictional claims in published maps and institutional affiliations.
involved with motivation/reward. The temporality of func-
tional changes can be perhaps better elucidated with suc-
cessive imaging techniques over short intervals of time
References
relative to the DBS state to see which cortical regions are
most implicated in behavior. Perhaps the best evidence with 1. Coenen VA, Schlaepfer TE, Sajonz B, Döbrössy M, Kaller CP,
the greatest temporal resolution is electrophysiologic evi- Urbach H, et al. Tractographic description of major subcortical
dence of subcortical-cortical connectivity [166] which can projection pathways passing the anterior limb of the internal
capsule. Corticopetal Organ Netw Relevant Psychiatr Disord
only be gleaned in the human in vivo through neurosurgical
Neuroimage Clin. 2020;25:102165.
intervention in the operating room. Time will dictate when 2. Schlaepfer TE, Bewernick BH, Kayser S, Mädler B, Coenen VA.
this can be achieved, especially given the low numbers of Rapid effects of deep brain stimulation for treatment-resistant
patients that have been enrolled in these ongoing studies. major depression. Biol Psychol. 2013;73:1204–12.
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

3. Fenoy AJ, Schulz PE, Selvaraj S, Burrows CL, Zunta-Soares G, 25. Keren H, O’Callaghan G, Vidal-Ribas P, Buzzell GA, Brotman
Durkin K, et al. A longitudinal study on deep brain stimulation of MA, Leibenluft E, et al. Reward processing in depression: a
the medial forebrain bundle for treatment-resistant depression. conceptual and meta-analytic review across FMRI and EEG
Transl Psychiatry. 2018;8:111. studies. Am. J. Psychiatry. 2018;175:1111–20.
4. DiLuca M, Olesen J. The cost of brain diseases: a burden or a 26. Anisman H, Matheson K. Stress, depression, and anhedonia:
challenge? Neuron. 2014;82:1205–8. caveats concerning animal models. Neurosci Biobehav Rev.
5. Sussman M, O’sullivan AK, Sha A, Olfson M, Menzin J. Eco- 2005;29:525–46.
nomic burden of treatment-resistant depression on the U.S. 27. Blood AJ, Iosifescu DV, Makris N, Perlis RH, Kennedy DN,
health care system. J Manag Care Spec Pharm. 2019;25:823–35. Dougherty DD, et al. Microstructural abnormalities in subcortical
6. WHO. Depression: fact sheet. Geneva: World Health Organisa- reward circuitry of subjects with major depressive disorder.
tion; 2017. http://www.who.int/mediacentre/factsheets/fs369/en/. PLoS ONE. 2010;5:e13945.
7. Rush AJ, Trivedi MH, Stewart JW, Nierenberg AA, Fava M, 28. Nestler EJ, Barrot M, DiLeone RJ, Eisch AJ, Gold SJ,
Kurian BT, et al. Combining medications to enhance depression Monteggia LM. Neurobiology of depression. Neuron. 2002;34:
outcomes (CO-MED): acute and long-term outcomes of a single- 13–25.
blind randomized study. Am J Psychiatry. 2011;168:689–701. 29. Russo SJ, Nestler EJ. The brain reward circuitry in mood dis-
8. Ruhe HG, van Rooijen G, Spijker J, Peeters FPML, Shene AH. orders. Nat Rev Neurosci. 2013;14:609–25.
Staging methods for treatment resistant depression. A systematic 30. Schlaepfer TE, Bewernick BH, Kayser S, Hurlemann R,
review. J Affect Disord. 2012;137:35–45. Coenen VA. Deep brain stimulation of the human reward
9. Rush AJ. Star-D: what have we learned? Am J Psychiatry. 2007; system for major depression–rationale, outcomes and outlook.
164:201–4. Neuropsychopharmacology. 2014;39:1303–14.
10. Rush AJ. Star-D: lessons learned and future implications. 31. Abler B, Greenhouse I, Ongur D, Walter H, Heckers S.
Depress Anxiety. 2011;28:521–4. Abnormal reward system activation in mania. Neuropsycho-
11. Williams LM. Precision psychiatry: a neural circuit taxonomy for pharmacology. 2008;33:2217–27.
depression and anxiety. Lancet Psychiatry. 2016;3:472–80. 32. Coenen VA, Honey CR, Hurwitz T, Rahman AA, Mcmaster J,
12. Drysdale AT, Grosenick L, Downar J, Dunlop K, Mansouri F, Bürgel U, et al. Medial forebrain bundle stimulation as a
Meng Y, et al. Resting-state connectivity biomarkers define pathophysiological mechanism for hypomania in subthalamic
neurophysiological subtypes of depression. Nat Med. 2017;23: nucleus deep brain stimulation for Parkinson’s disease. Neuro-
28–38. surgery. 2009;64:1106–14.
13. Akil H, Gordon J, Hen R, Javitch J, Mayberg H, Ewen B, et al. 33. Dichter GS, Kozink RV, McClernon FJ, Smoski MJ. Remitted
Treatment resistant depression: a multi-scale, systems biology major depression is characterized by reward network hyper-
approach. Neurosci Biobehav Rev. 2018;84:272–88. activation during reward anticipation and hypoactivation during
14. Coenen VA, Schlaepfer TE, Maedler B, Panksepp J. Cross-species reward outcomes. J Affect Disord. 2012;136:1126–34.
affective functions of the medial forebrain bundle-implications for 34. Ö hman A, Carlsson K, Lundqvist D. On the unconscious sub-
the treatment of affective pain and depression in humans. Neurosci cortical origin of human fear. J Phys Behav. 2007;92:180–5.
Biobehav Rev. 2011;35:1971–81. 35. Gross CT. The many paths to fear. Nat Rev Neurosci. 2012;13:
15. Coenen VA, Madler B, Schlaepfer TE. Reply to: medial fore- 651–8. Sep.
brain bundle stimulation—speed access to an old or entry into a 36. Motta SC, Carobrez AP, Canteras NS. The periaqueductal gray
new depression neurocircuit? Biol Psychol. 2013;74:e45–6. and primal emotional processing critical to influence complex
16. Castro DC, Berridge KC. Advances in the neurobiological bases defensive responses, fear learning and reward seeking. Neurosci
for food “liking” versus “wanting”. Physiol Behav. 2014;136: Biobehav Rev. 2017;76:39–47. May(Pt A).
22–30. Sep. 37. Panksepp J. Feeling the pain of social loss. Science. 2003;
17. Berridge KC, Kringelbach ML. Pleasure systems in the brain. 302:237–9.
Neuron. 2015;86:646–64. 38. Eisenberger NI, Lieberman MD, Williams KD. Does rejection
18. Döbrössy MD, Furlanetti LL, Coenen VA. Electrical stimulation hurt? An FMRI study of social exclusion. Science. 2003;302:
of the medial forebrain bundle in pre-clinical studies of psy- 290–2.
chiatric disorders. Neurosci Biobehav Rev. 2015;49:32–42. 39. Alexander GE, DeLong MR, Strick PL. Parallel organization of
19. Dean J, Keshavan M. The neurobiology of depression: an inte- functionally segregated circuits linking basal ganglia and cortex.
grated view. Asian J Psychiatry. 2017;27:101–11. Annu Rev Neurosci. 1986;9:357–81.
20. Disner SG, Beever CG, Haig EA, Beck A. Neural mechanisms of 40. Greenberg BD, Malone DA, Friehs GM, Rezai AR, Kubu CS,
the cognitive model of depression. Nat Rev Neurosci. 2011; Malloy PF, et al. Three-year outcomes in deep brain stimulation
12:467–77. for highly resistant obsessive–compulsive disorder. Neu-
21. Li BJ, Friston K, Mody M, Wang HN. A brain network model for ropsychopharmacology. 2006;31:2384–93.
depression: from symptom understanding to disease intervention. 41. Mayberg HS, Lozano AM, Voon V, McNeely HE, Seminowicz
CNS Neurosci Ther. 2018;24:1004–19. D, Hamani C, et al. Deep brain stimulation for treatment-resistant
22. Panksepp J, Wright JS. An evolutionary framework to under- depression. Neuron. 2005;45:651–60.
stand foraging, wanting, and desire: the neuropsychology of the 42. Hamani C, Mayberg H, Stone S, Laxton A, Haber S, Lozano
seeking system. Neuropsychoanalysis. 2012;14:59–75. AM. The subcallosal cingulate gyrus in the context of major
23. Alves-Pinto A, Rus OG, Reess TJ, Wohlschlager A, Wagner G, depression. Biol Psychiatry. 2011;69:301–8.
Berberich G, et al. Altered reward-related effective connectivity 43. Drevets WC, Price JL, Simpson JR Jr, Todd RD, Reich T,
in obsessive-compulsive disorder: an fMRI study. J Psychiatry Vannier M, et al. Subgenual prefrontal cortex abnormalities in
Neurosci. 2019;44:1–12. mood disorders. Nature. 1997;386:824–7.
24. Coenen VA, Schlaepfer TE, Goll P, Reinacher PC, Voderholzer 44. Lozano AM, Mayberg HS, Giacobbe P, Hamani C, Craddock
U, Terbartz van Elst L, et al. The medial forebrain bundle as a RC, Kennedy SH. Subcallosal cingulate gyrus deep brain sti-
target for deep brain stimulation for obsessive-compulsive dis- mulation for treatment-resistant depression. Biol Psychiatry.
order. CNS Spectr. 2016;493:1–8. 2008;64:461–7.
 A. J. Fenoy et al.

45. Holtzheimer PE, Husain MM, Lisanby SH, Taylor SF, Whitworth 62. Sharp C, Monterosso J, Montague PR. Neuroeconomics: a bridge
LA, McClintock S, et al. Subcallosal cingulate deep brain stimu- for translational research. Biol Psychiatry. 2012;72:87–92.
lation for treatment-resistant depression: a multisite, randomised, 63. Sesack SR, Grace AA. Cortico-Basal Ganglia reward network:
sham-controlled trial. Lancet Psychiatry. 2017;4:839–49. microcircuitry. Neuropsychopharmacology. 2010;35:27–47.
46. Riva-Posse P, Choi KS, Holtzheimer PE, Crowell AL, Garlow 64. Ikemoto S, Panksepp J. The role of nucleus accumbens
SJ, Rajendra JK, et al. A connectomic approach for sub- callosal dopamine in motivated behavior: a unifying interpretation with
cingulate deep brain stimulation surgery: prospective targeting in special reference to reward-seeking. Brain Res Brain Res Rev.
treatment-resistant depression. Mol Psychol. 2017;62:10. 1999;31:6–41.
47. Malone DA Jr, Dougherty DD, Rezai AR, Carpenter LL, Friehs 65. Alcaro A, Panksepp J. The SEEKING mind: primal neuro-
GM, Eskandar EN, et al. Deep brain stimulation of the ventral affective substrates for appetitive incentive states and their
capsule/ventral striatum for treatment-resistant depression. Biol pathological dynamics in addictions and depression. Neurosci
Psychiatry. 2009;65:267–75. Feb 15. Biobehav Rev. 2011;35:1805–20.
48. Dougherty DD, Rezai AR, Carpenter LL, Howland RH, Bhati 66. Panksepp, J. Affective neuroscience: the foundations of human
MT, O’Reardon JP, et al. A randomized sham-controlled trial of and animal emotions. New York: Oxford University Press; 1998.
deep brain stimulation of the ventral capsule/ventral striatum for 67. Wise A, McDevitt RA. Drive and reinforcement circuitry in the
chronic treatment-resistant depression. Biol Psychiatry. 2015; brain: origins, neurotransmitters, and projection fields. Neu-
78:240–8. Aug 15. ropsychopharmacol Publ Am Coll Neuropsychopharmacol.
49. Bergfeld IO, Mantione M, Hoogendoorn ML, Ruhé HG, Notten 2018;43:680–9.
P, van Laarhoven J, et al. Deep brain stimulation of the ventral 68. Heshmati M, Russo SJ. Anhedonia and the brain reward circuitry
anterior limb of the internal capsule for treatment-resistant in depression. Curr Behav Neurosci Rep. 2015;2:146–53.
depression: a randomized clinical trial. JAMA Psychiatry. 69. Witten IB, Steinberg EE, Lee SY, Davidson TJ, Zalocusky KA,
2016;73:456–64. May 1. Brodsky M. et al. Recombinase-driver rat lines: tools, techni-
50. Bewernick BH, Kayser S, Sturm V, Schlaepfer TE. Long-term ques, and optogenetic application to dopamine-mediated rein-
effects of nucleus accumbens deep brain stimulation in forcement. Neuron. 2011;72:721–33.
treatment-resistant depression: evidence for sustained efficacy. 70. Tan KR, et al. GABA neurons of the VTA drive conditioned
Neuropsychopharmacology. 2012;37:1975–85. Aug. place aversion. Neuron. 2012;73:1173–83.
51. Fitzgerald PB, Segrave R, Richardson KE, Knox LA, Herring S, 71. Tsai HC, Zhang F, Adamantidis A, Stuber GD, Bonci A, de Leca
Daskalakis ZJ, et al. A pilot study of bed nucleus of the stria L, et al. Phasic firing in dopaminergic neurons is sufficient for
terminalis deep brain stimulation in treatment resistant depres- behavioral conditioning. Science. 2009;324:1080–4.
sion. Brain Stimul. 2018;11:921–8. Jul-Aug. 72. van Zessen R, Phillips JL, Budygin EA, Stuber GD. Activation
52. Raymaekers S, Luyten L, Bervoets C, Gabriëls L, Nuttin B. of VTA GABA neurons disrupts reward consumption. Neuron.
Deep brain stimulation for treatment-resistant major depressive 2012;73:1184–94.
disorder: a comparison of two targets and long-term follow-up. 73. Tye KM, Mirzabekov JJ, Warden MR, Ferenczi EA, Tsai HC,
Transl Psychiatry. 2017;7:e1251. Oct 31. Finkelstein J, et al. Dopamine neurons modulate neural encoding
53. Lee DJ, Dallapiazza RF, De Vloo P, Elias GJB, Fomenko A, and expression of depression-related behaviour. Nature. 2013;
Boutet A, et al. Inferior thalamic peduncle deep brain stimulation 493:537–41.
for treatment-refractory obsessive-compulsive disorder: a phase 74. Chaudhury D, Walsh JJ, Friedman AK, Juarez B, Ku SM, Koo
1 pilot trial. Brain Stimul. 2019;12:344–52. Mar-Apr. JW, et al. Rapid regulation of depression-related behaviours by
54. Jiménez F, Nicolini H, Lozano AM, Piedimonte F, Salín R, control of midbrain dopamine neurons. Nature. 2013;493:532–6.
Velasco F. Electrical stimulation of the inferior thalamic 75. Olds J, Milner, P. Positive reinforcement produced by electrical
peduncle in the treatment of major depression and obsessive stimulation of the septal area and other regions of rat brain. J
compulsive disorders. World Neurosurg. 2013;80:e17–25. Comp Physiol Psychol. 1954;47:419–27.
Sep-OctS30. 76. Furlanetti LL, Coenen VA, Aranda IA, Döbrössy MD. Chronic
55. Sartorius A, Kiening KL, Kirsch P, von Gall CC, Haberkorn U, deep brain stimulation of the medial forebrain bundle reverses
Unterberg AW, et al. Remission of major depression under deep depressive-like behavior in a hemiparkinsonian rodent model.
brain stimulation of the lateral habenula in a therapy-refractory Exp Brain Res. 2015;233:3073–85.
patient. Biol Psychiatry 2010;67:e9–e11. 77. Thiele S, Furlanetti L, Pfeiffer L-M, Coenen VA, Döbrössy MD.
56. Döbrössy MD, Ramanathan C, Ashouri Vajari D, Tong Y, The effects of bilateral, continuous, and chronic deep brain sti-
Schlaepfer T, Coenen VA. Neuromodulation in psychiatric dis- mulation of the medial forebrain bundle in a rodent model of
orders: experimental and clinical evidence for reward and depression. Exp Neurol. 2018;303:153–61.
motivation network deep brain stimulation: focus on the medial 78. Edemann-Callesen H, Voget M, Empl L, Vogel M, Wieske F,
forebrain bundle. Eur J Neurosci. 2020; 15. https://doi.org/10. Rummel J, et al. Medial forebrain bundle deep brain stimulation
1111/ejn.14975. has symptom-specific anti-depressant effects in rats and as
57. Nieuwenhuys R, Geeraedts LMG, Veening JG. The medial opposed to ventromedial prefrontal cortex stimulation interacts
forebrain bundle of the rat. I. General Introduction. J Comp with the reward system. Brain Stimul. 2015;8:714–23. Jul-Aug.
Neurol. 1982;206:49–81. 79. Dandekar MP, Saxena A, Scaini G, Shin JH, Migut A, Giridharan
58. Nieuwenhuys R. The greater limbic system, the emotional motor VV, et al. Medial forebrain bundle deep brain stimulation reverses
system and the brain. Prog Brain Res. 1996;107:551–80. anhedonic-like behavior in a chronic model of depression: the
59. Lammel S, Lim BK, Malenka RC. Reward and aversion in a importance of BDNF and inflammatory cytokines. Mol Neurobiol.
heterogeneous midbrain dopamine system. Neuropharmacology. 2019;56:4364–80.
2014;76:351–9. JanPt B(0 0). 80. Dandekar MP, Luse D, Hoffmann C, Cotton P, Peery T, Ruiz C,
60. Koob GF, Le Moal M. Addiction and the brain antireward sys- et al. Increased dopamine receptor expression and anti-
tem. Annu Rev Psychol. 2008;59:29–53. depressant response following deep brain stimulation of the
61. Ikemoto S, Wise RA. Rewarding effects of the cholinergic agents medial forebrain bundle. J Affect Disord. 2017;217:80–88.
carbachol and neostigmine in the posterior ventral tegmental 81. Furlanetti LL, Coenen VA, Döbrössy MD. Ventral tegmental
area. J Neurosci. 2002;22:9895–904. area dopaminergic lesion-induced depressive phenotype in the
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

rat is reversed by deep brain stimulation of the medial forebrain characterization of the response observed in the caudate nucleus
bundle. Behav Brain Res. 2016;299:132–40. of the rat. J Neurochem. 1984;43:560–9.
82. Thiele S, Sörensen A, Weis J, Braun F, Meyer PT, Coenen VA, 100. Stamford JA, Kruk ZL, Millar J. Measurement of stimulated
et al. Deep brain stimulation of the medial forebrain bundle dopamine release in the rat by in vivo voltammetry: the influence of
in a rodent model of depression: exploring dopaminergic stimulus duration on drug responses. Neurosci Lett. 1986;69:70–73.
mechanisms with raclopride and Micro-PET. Stereotact. Funct 101. Stamford JA, Kruk ZL, Millar J. Accommodation of rat
Neurosurg. 2020;98:1–13. nigrostriatal dopamine neurones to high frequency electrical
83. Corbett D, Wise RA. Intracranial self-stimulation in relation to stimulation of the median forebrain bundle: in vivo voltammetric
the ascending dopaminergic systems of the midbrain: a moveable data. Neurosci Lett. 1987;82:172–6.
electrode mapping study. Brain Res. 1980;185:1–15. Mar 3. 102. Gratton A, Hoffer BJ, Gerhardt GA. Effects of electrical stimu-
84. Fibiger HC, LePiane FG, Jakubovic A, Phillips AG. The role of lation of brain reward sites on release of dopamine in rat: an
dopamine in intracranial self-stimulation of the ventral tegmental in vivo electrochemical study. Brain Res Bull. 1988;21:319–24.
area. J Neurosci. 1987;7:3888–96. Dec. 103. Gonon FG. Nonlinear relationship between impulse flow and
85. German DC, Bowden DM. Catecholamine systems as the neural dopamine released by rat midbrain dopaminergic neurons as stu-
substrate for intracranial self-stimulation: a hypothesis. Brain died by in vivo electrochemistry. Neuroscience. 1988;24:19–28.
Res. 1974;73:381–419. Jun 28. 104. Klanker M, Feenstra M, Willuhn I, Denys D. Deep brain sti-
86. Nakahara D, Ozaki N, Miura Y, Miura H, Nagatsu T. Increased mulation of the medial forebrain bundle elevates striatal dopa-
dopamine and serotonin metabolism in rat nucleus accumbens mine concentration without affecting spontaneous or reward-
produced by intracranial self-stimulation of medial forebrain induced phasic release. Neuroscience. 2017;364:82–92.
bundle as measured by in vivo microdialysis. Brain Res. 1989; 105. Howe MW, Tierney PL, Sandberg SG, Phillips PEM, Graybiel
495:178–81. Aug 21. AM. Prolonged dopamine signalling in striatum signals proxi-
87. Bregman T, Reznikov R, Diwan M, Raymond R, Butson CR, mity and value of distant rewards. Nature. 2013;500:575–9.
Nobrega JN, et al. Antidepressant-like effects of medial forebrain 106. Cohen JY, Haesler S, Vong L, Lowell BB, Uchida N. Neuron-
bundle deep brain stimulation in rats are not associated with type-specific signals for reward and punishment in the ventral
accumbens dopamine release. Brain Stimul. 2015;8:708–13. tegmental area. Nature. 2012;482:85–8. Jan 18.
88. Ou Y, Buchanan AM, Witt CE, Hashemi P. Frontiers in elec- 107. Matsumoto M, Hikosaka O. Two types of dopamine neuron
trochemical sensors for neurotransmitter detection: towards distinctly convey positive and negative motivational signals.
measuring neurotransmitters as chemical diagnostics for brain Nature. 2009;459:837–41. Jun 11.
disorders. Anal Methods. 2019;11:2738–55. 108. Schultz W, Dayan P, Montague PR. A neural substrate of pre-
89. Ranck JB Jr. Which elements are excited in electrical stimulation diction and reward. Science. 1997;275:1593–9. Mar 14.
of mammalian central nervous system: a review. Brain Res. 109. Waelti P, Dickinson A, Schultz W. Dopamine responses comply
1975;98:417–40. with basic assumptions of formal learning theory. Nature.
90. Kringelbach ML, Jenkinson N, Owen SLF, Aziz TZ. Transla- 2001;412:43–8. Jul 5.
tional principles of deep brain stimulation. Nat Rev Neurosci. 110. Bayer HM, Glimcher PW. Midbrain dopamine neurons encode a
2007;8:623–35. quantitative reward prediction error signal. Neuron. 2005;47:
91. Phillips PEM, Robinson DL, Stuber GD, Carelli RM, Wightman 129–41. Jul 7.
RM. Real-time measurements of phasic changes in extracellular 111. Settell ML, Testini P, Cho S, Lee JH, Blaha CD, Jo HJ, et al.
dopamine concentration in freely moving rats by fast-scan cyclic Functional circuitry effect of ventral tegmental area deep brain
voltammetry. Methods Mol Med. 2003;79:443–64. stimulation: imaging and neurochemical evidence of mesocor-
92. Lohani S, Martig AK, Deisseroth K, Witten IB, Moghaddam B. tical and mesolimbic pathway modulation. Front Neurosci.
Dopamine modulation of prefrontal cortex activity is manifold 2017;11:104.
and operates at multiple temporal and spatial scales. Cell Rep. 112. Ashouri Vajari D, Ramanathan C, Tong Y, Stieglitz T, Coenen
2019;27:99–114. e6. VA, Döbrössy MD. Medial forebrain bundle DBS differentially
93. Ungless MA, Magill PJ, Bolam JP. Uniform inhibition of modulates dopamine release in the nucleus accumbens in a
dopamine neurons in the ventral tegmental area by aversive sti- rodent model of depression. Exp Neurol. 2020;327:113224.
muli. Science. 2004;303:2040–2. 113. Hamid AA, Pettibone JR, Mabrouk OS, Hetrick VL, Schmidt R,
94. Valenti O, Gill KM, Grace AA. Different stressors produce Vander Weele CM, et al. Mesolimbic dopamine signals the value
excitation or inhibition of mesolimbic dopamine neuron activity: of work. Nat Neurosci. 2016;19:117–26.
response alteration by stress pre-exposure. Eur J Neurosci. 2012; 114. Ikemoto S. Brain reward circuitry beyond the mesolimbic dopamine
35:1312–21. system: a neurobiological theory. Neurosci Biobehav Rev. 2010;35:
95. Bunney BS, Walters JR, Roth RH, Aghajanian GK. Dopami- 129–50.
nergic neurons: effect of antipsychotic drugs and amphetamine 115. Ikemoto S. Dopamine reward circuitry: two projection systems
on single cell activity. J Pharmacol Exp Ther. 1973;185:560–71. from the ventral midbrain to the nucleus accumbens–olfactory
96. Grace AA, Bunney BS. Intracellular and extracellular electro- tubercle complex. Brain Res Rev. 2007;56:27–78.
physiology of nigral dopaminergic neurons-1. Identif Charact 116. Gradinaru V, Mogri M, Thompson KR, Henderson JM, Deis-
Neurosci. 1983;10:301–15. seroth K. Optical deconstruction of parkinsonian neural circuitry.
97. Ewing AG, Bigelow JC, Wightman RM. Direct in vivo mon- Science. 2009;324:354–9. PMID:19299587.
itoring of dopamine released from two striatal compartments in 117. Taber MT, Das S, Fibiger HC. Cortical regulation of subcortical
the rat. Science. 1983;221:169–71. dopamine release: mediation via the ventral tegmental area. J
98. Kuhr WG, Wightman RM, Rebec GV. Dopaminergic neurons: Neurochem. 1995;65:1407–10. PMID: 7643120.
simultaneous measurements of dopamine release and single-unit 118. Karreman M, Westerink BH, Moghaddam B. Excitatory amino
activity during stimulation of the medial forebrain bundle. Brain acid receptors in the ventral tegmental area regulate dopamine
Res. 1987;418:122–8. release in the ventral striatum. J Neurochem. 1996;67:601–7.
99. Kuhr WG, Ewing AG, Caudill WL, Wightman RM. Monitoring 119. Helbing C, Brocka M, Scherf T, Lippert MT, Angenstein F. The
the stimulated release of dopamine with in vivo voltammetry. I: role of the mesolimbic dopamine system in the formation of
 A. J. Fenoy et al.

blood-oxygen-level dependent responses in the medial pre- decision making in human and monkey frontal cortex. Proc Natl
frontal/anterior cingulate cortex during high-frequency stimula- Acad Sci USA 2015;112:E2695–E2704.
tion of the rat perforant pathway. J Cereb Blood Flow Metab. 136. Haber SN, Yendiki A, Jbabdi S. Four deep brain stimulation
2016;36:2177–93. Dec. targets for obsessive-compulsive disorder: Are they different?
120. Schoene-Bake JC, Parpaley Y, Weber B, Panksepp J, Hurwitz TA, Biol Psychiatry. 2020;S0006-3223:31773-X. https://doi.org/10.
Coenen VA. Tractographic analysis of historical lesion surgery for 1016/j.biopsych.2020.06.031. Online ahead of print.
depression. Neuropsychopharmacology. 2010;35:2553–63. 137. Schmahmann J, Pandya D. Fiber pathways of the brain. 1st
121. Coenen VA, Panksepp J, Hurwitz TA, Urbach H, Madler B. edition. New York: Oxford University Press;2006.
Human medial forebrain bundle (MFB) and anterior thalamic 138. Gaspar P, Neurology ISOC. Topography and collateralization of
radiation (ATR): imaging of two major subcortical pathways and the dopaminergic projections to motor and lateral prefrontal
the dynamic balance of opposite affects in understanding cortex in owl monkeys. Wiley Online Library;1992.
depression. J Neuropsychiatry Clin Neurosci. 2012;24:223–36. 139. Petersen MV, Mlakar J, Haber SN, Parent M, Smith Y, Strick
122. Coenen VA, Schumacher LV, Kaller C, Schlaepfer TE, PL, et al. Holographic reconstruction of axonal pathways in the
Reinacher PC, Egger K, et al. The anatomy of the human medial human brain. Neuron. 2019; 1–16.
forebrain bundle: ventral tegmental area connections to reward- 140. Hurwitz TA, Mandat T, Forster B, Honey C. Tract identification
associated subcortical and frontal lobe regions. NeuroImage by novel MRI signal changes following stereotactic anterior
Clin. 2018c;18:770–83. capsulotomy. Stereotact Funct Neurosurg. 2006b;84:228–35.
123. Frankle WG, Laruelle M, Haber SN. Prefrontal cortical projec- 141. Montgomery SA, Asberg M. A new depression scale designed to
tions to the midbrain in primates: evidence for a sparse con- be sensitive to change. Br J Psychiatry. 1979;134:382–9.
nection. Neuropsychopharmacology. 2006;31:1627–36. 142. Bewernick BH, Kayser S, Gippert SM, Switala C, Coenen VA,
124. Greenberg BD, Rauch SL, Haber SN. Invasive circuitry-based Schlaepfer TE. Deep brain stimulation to the medial forebrain
neurotherapeutics: stereotactic ablation and deep brain stimula- bundle for depression-long-term outcomes and a novel data
tion for OCD. Neuropsychopharmacology. 2010;35:317–36. analysis strategy. Brain Stimul. 2017;10:664–71.
125. Haynes WIA, Haber SN. The organization of prefrontal- 143. Coenen VA, Bewernick BH, Kayser S, Kilian H, Boström J,
subthalamic inputs in primates provides an anatomical sub- Greschus S, et al. Superolateral medial forebrain bundle deep
strate for both functional specificity and integration: implications brain stimulation in major depression: a gateway trial. Neu-
for Basal Ganglia models and deep brain stimulation. J Neurosci. ropsychopharmacology. 2019a;26:587.
2013;33:4804–14. 144. Hamilton M. Development of a rating scale for primary
126. Nanda P, Banks GP, Pathak YJ, Sheth SA. Connectivity-based depressive illness. Br J Soc Clin Psychol. 1967;6:278–96.
parcellation of the anterior limb of the internal capsule. Hum 145. Jones SH, Thornicroft G, Coffey M, Dunn G. A brief mental health
Brain Mapp. 2017;38:6107–17. outcome scale-reliability and validity of the Global Assessment of
127. Safadi Z, Grisot G, Jbabdi S, Behrens TE, Heilbronner SR, Functioning (GAF). Br J Psychiatry. 1995;166:654–9.
McLaughlin NCR, et al. Functional segmentation of the anterior 146. First MB, Spitzer RL, Gibbon M, Williams JBW, Benjamin LS.
limb of the internal capsule: linking white matter abnormalities to Structured clinical interview for DSM-IV Axis II personality
specific connections. J Neurosci. 2018a;38:2106–17. disorders (SCID II). Washington, DC: American Psychiatric
128. Safadi Z, Grisot G, Jbabdi S, Behrens TE, Heilbronner SR, Press;1996.
McLaughlin NCR, et al. Functional segmentation of the anterior 147. Millon T, Millon C, Davis R. Millon clinical multiaxial
limb of the internal capsule: linking white matter abnormalities to inventory-III (MCMI-III) manual. Minneapolis, MN: National
specific connections. J Neurosci. 2018b;38:2106–17. Computer Systems;1994.
129. Lehman JF, Greenberg BD, McIntyre CC, Rasmussen SA, Haber 148. Fenoy AJ, Schulz P, Selvaraj S, Burrows C, Spiker D, Cao B,
SN. Rules ventral prefrontal cortical axons use to reach their et al. Deep brain stimulation of the medial forebrain bundle:
targets: implications for diffusion tensor imaging tractography distinctive responses in resistant depression. J Affect Disord.
and deep brain stimulation for psychiatric illness. J Neurosci. 2016;203:143–51. Oct 3. PMID: 27288959.
2011;31:10392–402. 149. Perez-Caballero L, Perez-Egea R, Romero-Grimaldi C, Puigde-
130. Makris N, Rathi Y, Mouradian P, Bonmassar G, Papadimitriou mont D, Molet J, Caso JR, et al. Early responses to deep brain
G, Ing WI, et al. Variability and anatomical specificity of the stimulation in depression are modulated by anti-inflammatory
orbitofrontothalamic fibers of passage in the ventral capsule/ drugs. Mol Psychiatry. 2014;19:607–14.
ventral striatum (VC/VS): precision care for patient-specific 150. Chang SY, Shon YM, Agnesi F, Lee KH. Microthalamotomy
tractography-guided targeting of deep brain stimulation (DBS) in effect during deep brain stimulation: potential involvement of
obsessive compulsive disorder (OCD). Brain Imaging Behav. adenosine and glutamate efflux. Conf Proc IEEE Eng Med Biol
2016;10:1054–67. Soc. 2009, 3294–7.
131. Baldermann JC, Melzer C, Zapf A, Kohl S, Timmermann L, 151. Fenoy AJ, Goetz L, Chabardès S, Xia Y. Deep brain stimulation:
Tittgemeyer M, et al. Connectivity profile predictive of effective are astrocytes a key driver behind the scene? CNS Neurosci Ther
deep brain stimulation in obsessive compulsive disorder. Biol 2014;20:191–201.
Psychiatry 2019;85:735–43. 152. Naudet F, Millet B, Reymann JM, Falissard B. Improving study
132. Panksepp J. The basic emotional circuits of mammalian brains: design for anti- depressant effectiveness assessment. Int J
do animals have affective lives? Neurosci Biobehav Rev. Methods Psychiatr Res. 2013;22:217–31.
2011;35:1791–804. 153. Schatzberg AF, Kraemer HC. Use of placebo control groups in
133. Nieuwenhuys R, Voogd J, van Huijzen C. The human central evaluating efficacy of treatment of unipolar major depression.
nervous system. A Synopsis and Atlas, Fourth Edition. Heidel- Biol Psychiatry. 2000;47:736–44.
berg: Springer; 2008. 154. Kilian HM, Meyer DM, Bewernick BH, Spanier S, Coenen VA,
134. Panksepp J. Affective consciousness: core emotional feelings in Schlaepfer TE. Discontinuation of superolateral medial forebrain
animals and humans. Conscious Cogn. 2005;14:30–80. bundle deep brain stimulation for treatment-resistant depression
135. Neubert F-X, Mars RB, Sallet J, Rushworth MFS. Connectivity leads to critical relapse. Biol Psychiatry. 2018;S0006-3223:31748–7.
reveals relationship of brain areas for reward-guided learning and Sep 22. pii.
Deep brain stimulation of the “medial forebrain bundle”: a strategy to modulate the reward system and. . .

155. Martín-Blanco A, Serra-Blasco M, Pérez-Egea R, de for severe treatment-resistant depression with deep-brain stimu-
Diego-Adeliño J, Carceller-Sindreu M, Puigdemont D, et al. lation. Transl Psychiatry. 2017;7:e1207. Aug 15.
Immediate cerebral metabolic changes induced by dis- 161. Kelley ME, Franco AR, Mayberg HS, Holtzheimer PE. The
continuation of deep brain stimulation of subcallosal cingulate illness density index (IDI): a longitudinal measure of treatment
gyrus in treatment-resistant depression. J Affect Disord. 2015; efficacy. Clin Trials. 2012;9:596e604.
173:159–62. 162. Puigdemont D, Portella M, Perez-Egea R, Molet J, Gironell A,
156. Evers J, Lowery M. The active electrode in the living brain: the de Diego- Adelino J, et al. A randomized double-blind crossover
response of the brain parenchyma to chronically implanted deep trial of deep brain stimulation of the subcallosal cingulate gyrus
brain stimulation electrodes. Oper Neurosurg. 2020 Oct 19: in patients with treatment-resistant depression: a pilot study of
opaa326. https://doi.org/10.1093/ons/opaa326. relapse prevention. J psychiatry Neurosci JPN. 2015;40:130295.
157. Vedam-Mai V, Rodgers C, Gureck A, Vincent M, Ippolito G, 163. Lozano AM, Giacobbe P, Hamani C, Rizvi SJ, Kennedy SH,
Elkouzi A, et al. Deep brain stimulation associated gliosis: Kolivakis TT, et al. A multicenter pilot study of subcallosal
a post-mortem study. Parkinsonism Relat Disord. 2018;54: cingulate area deep brain stimulation for treatment-resistant
51–55. Sep. depression. J Neurosurg. 2012;116:315–22.
158. Ruge D, Cif L, Limousin P, Gonzalez V, Vasques X, Hariz MI, 164. Blomstedt P, Naesström M, Bodlund O. Deep brain stimulation
et al. Shaping reversibility? Long-term deep brain stimulation in in the bed nucleus of the stria terminalis and medial forebrain
dystonia: the relationship between effects on electrophysiology bundle in a patient with major depressive disorder and anorexia
and clinical symptoms. Brain. 2011;134:2106–15. nervosa. Clin Case Rep. 2017;5:679–84.
159. Ruge D, Tisch S, Hariz MI, Zrinzo L, Bhatia KP, Quinn NP, 165. Davidson B, Giacobbe P, Mithani K, Levitt A, Rabin JS,
et al. Deep brain stimulation effects in dystonia: time course of Lipsman N, et al. Lack of clinical response to deep brain sti-
electrophysiological changes in early treatment. Mov Disord. mulation of the medial forebrain bundle in depression. Brain
2011;26:1913–21. Stimul. 2020;13:1268–70.
160. Sani S, Busnello J, Kochanski R, Cohen Y, Gibbons RD. High- 166. Fenoy A. Challenges in deep brain stimulation for depression.
frequency measurement of depressive severity in a patient treated Braz J Psychiatry. 2020;42:347–8. Aug.