Obesity is an increasingly prevalent condition, especially in developed countries, causing 3.4 million deaths per year worldwide [1]. According to a global national survey in 2014, there has been a 10% increase in overweight populations (body mass index [BMI]>25 kg/m²) between 1980 and 2013. Furthermore, the number of children with overweight has been rapidly increasing up to 23% in developed countries, indicating a significant health concern [1,2]. Given the numerous comorbidities of obesity such as osteoarthritis, stroke, certain types of cancer, financial burdens affecting individuals and whole societies are also to be expected. The medical costs associated with obesity are estimated to have risen to $147 billion from $78.5 over the span of 10 years [2-4]. In the United States, if the current increase in the obese population continues, all adult Americans will be overweight by the year 2048 [5].

There are only limited treatment options available for obese patients. The current available anti-obesity drugs can be divided into two classes; central acting and peripheral acting. However, owing to a number of factors including misuse of the agents, abusive commercialization of pharmaceutical galenic preparations and less emphasis on classic treatment counselling, the long-term effect of these anti-treatment drugs is controversial yet. Furthermore, the vast majority of these focus on body weight alone, leading to a high rate of treatment failure [6].

Bariatric surgery is usually reserved for patients with a BMI>40 kg/m² or a BMI>35 kg/m² in the presence of significant comorbidities and most commonly involves laparoscopic gastric banding or laparoscopic Roux-en-Y gastric bypass [6]. Although bariatric surgeries have been regarded as a primary weight loss strategy for medically refractory obese patients, it has various limitations, including a high incidence of both recurrence and complications [6,7]. Even if bariatric surgery leads to weight loss and improvement in morbidity reduction, considerable weight gain often recurs approximately 2 years after surgery with a failure rate of up to 46% [2]. Moreover, as Roux-en-Y gastric bypass surgery enables the bypass of 95% of the stomach, the entire duodenum, and 150 cm of the jejunum, the risk of micronutrient deficiency, such as in calcium and various minerals, is high and patients need to be treated with life-long supplementation [2]. Considering that the fundamental cause of obesity or binge eating behavior is related to psychiatric disease linked to the reward center of the brain, the limitation of gastric bypass surgery is apparent. The primary focus should be put on a more fundamental treatment that targets the cerebral networks of the reward system [2,6,7].

The present study has two purposes: 1) to review findings from the recent literature on deep brain stimulation (DBS) for obesity or binge eating behavior and 2) to introduce a persuasive theoretical hypothesis by which DBS works for obesity. It is not meant to be an extensive review of obesity, but rather a summary of contemporary theories and practical information, which stem from animal experiments, human studies, and ongoing human trials.

Many cerebral regions are known to be involved in the mechanism of eating disorders [8]. It is widely accepted that there is “bottom-up” emotion generation emerging from subcortical, limbic neural structures and “top-down” emotion regulation by dorsal prefrontal cortical regions. An imbalance between these can result in abnormal behavior. Increased activity of the “bottom-up” stream, which means increased activity in emotion generation to external stimuli, may contribute to altered reward processing such as binge eating behavior [8,9]. Key neurotransmitter in this reward system is dopamine. The regulation of eating behavior or addiction is modulated by multiple peripheral and central systems that deliver information to the dopamine center in the reward system [10-15]. The brain dopamine reward circuitry increases the probability that behaviors that activate it (food consumption) will be repeated when encountering the same reinforcer [15]. Hence, with repeated access to highly palatable food, some individuals may exceed the inhibitory process that signals satiety and begin to compulsively consume large amounts of food, frequently leading to repulsion in this behavior [2,11-15]. Three cerebral structures have been identified through lesioning studies to have a critical role in the reward circuit, leading to excessive food consumption: the lateral hypothalamus (LH), nucleus accumbens (NA), and ventral tegmental area (VTA) [2,7]. We will discuss the individual structures in the next section.

The LH has been shown to mediate appetitive and feeding-related behaviors [12]. GABAergic neurons from the LH disinhibit dopamine neurons in the VTA and result in increased dopamine levels in the NA, leading to motivated behaviors (Fig. 1A) [14]. As a response to this increase in dopamine levels in the NA, the lateral NA predominantly disinhibits dopamine neurons in the lateral VTA, increasing dopamine levels in the NA (indirect feedback). However, the medial GABAergic inhibitory pathway from the NA to dopamine neurons in the medial VTA is much stronger (direct feedback), thus resulting in the net inhibition of VTA dopaminergic neurons (Fig. 1B) and consequently leading to a decreased dopamine concentration in the NA [16]. This feedback loop between these three structures (LH-VTA-NA) has been widely accepted as a fundamental background in reward systems, especially for eating behaviors [16].

Previous findings from animal studies on DBS for obesity or binge eating behavior are summarized in Table 1, [11,17-24]. Many studies have confirmed that the NA can be divided into two substructures, core and shell, which both mediate reward cue-driven consumptive behaviors [25-27]. However, the effect of DBS in the NA core and shell is dissociable based on the aspects of motivated behavior, “wanting” vs. “liking [28].” In a study to identify the differences in the effect on eating behavior between the two, stimulation of the NA core before binge eating (not during binge eating) induced a decrease in high-fat food intake, whereas stimulation of the NA shell during binge eating (not before) affected behavior [20]. Given that “wanting” is responsible for cravings (i.e., in anticipation before binge eating) and “liking” is the pleasurable feeling of using the potentially incentive substance (during binge eating), it is plausible that NA core stimulation might modulate the process of “wanting” and that the NA shell might be involved in the process of “liking.” The NA shell can further be divided into two sub-nuclei, based on their distinct functional role in food intake (medial NA shell vs. lateral NA shell) [21]. However, even if the same target of the NA shell is stimulated, conflicting results have also been reported: increased food intake in normal-weight rats following DBS of the medial NA shell but a decrease in food intake in obese rats [21,24,29]. Despite this inconsistency, these results indicate that the NA shell is associated with food intake and body weight, further highlighting the NA shell as an interesting target of DBS to modulate eating behavior [21]. As a potential mechanism by which DBS on the NA shell may modulate such behavior, Halpern et al. [11] suggested the hypothesis that DBS in the NA shell may lead to local release of dopamine, which in turn binds to dopamine-2-receptor (D2R), blocking the hedonic valence of the high-fat diet [24,30]. However, the efficacy was found to diminish with continuous stimulation. In addition, in some cases of behavioral context changes (e.g., other behaviors of palatable food consumption such as diet-induced obesity or a binge relapse model), DBS of the NA shell appears to be less effective [11,18]. With regard to this failure in a relapse model, Doucette et al. postulated that the decreased efficacy of NA DBS may stem from an increased motivation to consume the palatable food, which had been proven in many studies showing that “incubation of craving” during prolonged abstinence can drive a more enhanced response to specific material when animals are re-exposed to previously associated cues [18,31,32]. In addition, a recent study by Casquero-Veiga et al. [17] supported a new hypothesis that NA DBS can also decrease metabolism in the striatum and thalamus, suggesting that this induced hypometabolism may alleviate the hyperactivated status of the striatum and thalamus, which is a characteristic pathology in obesity.

Given that the LH is an anatomical hub linking the arcuate nucleus in the ventromedial hypothalamus (VMH) (in which various peripheral signals are delivered throughout fenestrated capillaries) and multiple limbic structures, its effect on obesity or binge eating disorder has received significant attention [22]. Sani et al. [22] reported that stimulation of the LH leads to weight loss without any change in food or water intake, suggesting that LH stimulation may induce metabolic changes rather than modulate eating behavior. However, in other studies, LH stimulation showed different efficacies between diet-induced obese rats and normal rats [24]. Previous studies have shown that impaired dopamine neurotransmission in diet-induced obese leads more obsession with high-energy food to compensate for the weak dopaminergic input and this altered system of dopamine neurotransmission in obesity is believed to result in this difference [24,33,34]. Consistent with this, Zhang et al. [24] reported that DBS of the LH can induce an anorexic effect caused by the activation of D2Rs only in diet-induced obese rats (not in normal rats), implying that DBS-induced higher upregulation of dopamine can account for this selective efficacy.

In addition to these targets directly related to the reward circuit, it is worth noting that neuromodulation of the VMH can also lead to weight reduction [2,10,13]. The VMH, which contains the ventromedial nuclei and arcuate nuclei, is a critical structure in regulating glucose levels and energy homeostasis [35]. There is a long-lasting consensus suggesting that the VMH can reduce weight gain by altering the metabolic rate, rather than affecting eating behavior [36,37]. Being modulated by positive signals of nutritional resources such as glucose (insulin) and leptin, the VMH not only reduces hepatic glucose output and increases peripheral glucose metabolism but also regulates appetite regulation as the satiety center of the brain [2,13]. The interaction between the VMH and structures of the reward system is schematically depicted in Fig. 2. Receiving projections from brain regions that maintain homeostasis (VMH), nuclei in the LH send projections to the reward center (VTA and NA) where they promote motivated behaviors [38]. Accordingly, in contrast to the mechanism related to dopamine levels in the reward system when the LH or NA is stimulated, the effect of VMH stimulation is likely associated with hypermetabolism, which increases basal energy consumption [23,39]. In a study by Torres et al. [23], a significant reduction in body weight and body fat was observed without any changes in hormone levels after chronic DBS stimulation of the VMH. In addition, an increase in animals’ movement velocity facilitating a higher metabolic rate during acute stimulation was noted. The authors also suggested that chronic DBS was effective only at 80 Hz stimulation, which might indicate selective inhibition of the orexigenic LH [23]. In addition, low-frequency DBS of the VMH can induce a significant reduction in weight gain without any behavioral change in minipigs, providing preclinical evidence in support of low-frequency VMH DBS as a treatment for obesity [19].

Although the exact mechanism to explain the effects of DBS on obesity or binge eating behavior has not been elucidated, an increasing number of studies have consistently revealed promising results on the clinical utility of DBS for medically refractory obesity.

The currently ongoing human trials are expected to further elucidate the role of neuromodulation in the treatment of obesity. Furthermore, other novel less-invasive methods such as optogenetics will hopefully emerge as an effective treatment in the near future.