As food transits through the digestive system, stomach and intestinal wall stretch activates intraganglionic laminar ending (IGLE) mechanoreceptors, which contributes to a sensation of satiety and a reduction in food consumption [50]. This has been demonstrated through experiments using gavage [51], ligation [19], and optogenetic activation [39] as well as several studies using electrical nerve stimulation. However, studies suggest that animals exposed to a high-fat diet have less vagal tone during stomach distension, which may promote overeating [15, 18,19,20,21, 52, 53]. Thus, the decision to use vagal nerve stimulation to increase vagal tone and convey a sense of fullness has been studied by several groups. Looking across the results from this study, the data shows that the sVNS group trended toward lower body weight and adiposity due to lower caloric intake but not likely due to an increase in activity that would result in increased energy expenditure nor due to a cessation of activity that would suggest discomfort.

During the baseline phase, all rats on a high-fat diet weighed significantly more than those on the standard diet. However, rats receiving sVNS attained a body weight that was greater than but not significantly different from rats kept on a standard diet. By the end of the study, rats receiving sVNS weighed approximately 595 g, compared to shams that weighed approximately 670 g. While the adiposity of rats on the high-fat diet was significantly greater than those on the standard diet, it is worth noting that the adiposity of rats receiving sVNS was significantly less than all other rats on the high-fat diet and the lean body mass was not significantly different between the group receiving sVNS and the control group on the standard diet. The reduction of adiposity found in this study was similar to the 20% reduction in visceral fat demonstrated in obese Wistar rats after 12 weeks of cervical VNS [34]. Results further suggest that the activity of rats receiving sVNS becomes more similar to rats on a standard diet than sham or non-implanted rats on a high-fat diet.

In terms of rate of weight gain between the start and end of the study, the sVNS group exhibited a 38% increase in body weight, whereas rats on the standard diet exhibited a 47% increase in body weight and rats with and without an implant on a high-fat diet exhibited a 65% and 64% increase in body weight, respectively. Therefore, the sVNS group gained weight at a lower rate than any of the other groups. This is similar to the approximately 41% and 59% weight gains observed in pigs that did and did not receive bilateral sVNS for 12 weeks using a similar stimulus paradigm, respectively [54]. When comparing the sVNS group to either of the control groups on the high-fat diet, the sVNS group exhibited a 42% decrease in weight gain, which is similar to the 48% reported during bilateral thoracic VNS in pigs [55]. While surgery and/or the implant in the sham population did reduce body weight compared to the control cohort on the high-fat diet that did not receive an implant, the reduction in body weight was transient and the two populations ended the study with similar weights and weight gains.

All rats on the high-fat diet consumed significantly less food than those on the standard diet. Although the weight of food consumed by the high-fat cohorts was less than that consumed by the cohort on a standard diet, due to the caloric density of the food, the rats on the high-fat diet consumed significantly more calories per day than those on the standard diet. This was true both during baseline and stimulus phases. Results also reveal that rats with an implant consumed significantly less food and, hence, significantly fewer calories than the cohort on a high-fat diet that did not receive an implant. Because there was a reduced caloric consumption by the sham group, we can reasonably assume that some of the reduced caloric consumption by the sVNS group was due to surgery or the implant and not due to sVNS. However, the sVNS group exhibited a further reduction in caloric intake, indicating that sVNS was effective at curtailing food consumption.

To our knowledge, this is the first study to exhaustively quantify movement and attempt to determine if there was a statistical difference in day-to-day patterns of movement between the different cohorts. Results confirm that rats receiving sVNS do not appear to behave abnormally and that the reduction in food consumption observed during sVNS is unlikely linked to lethargy or sequestration. Given the complexity of the statistical analyses, only the control group on a standard diet was compared to the other three groups. Pairwise comparisons were not conducted because corrections for multiple comparisons and the interpretation of the results were exceptionally esoteric. The results suggested that the sham group behaved differently from the control group on a standard diet during the day. Specifically, the sham group tended to move more during the day as the study progressed. It remains uncertain if this difference was meaningful, but, in general, all four cohorts moved similar amounts and exhibited similar behavior during the day. The same was not true at night.

During the night, all rats tended to be more active, though this was most pronounced in rats on the high-fat diet, resulting in a clearer difference in the different cohorts’ patterns of behavior. There were some interesting observations that came out of the data. Non-implanted rats on the high-fat diet tended to move more than any other cohort. We do not know why this occurred. It may be that this cohort was driven to forage for more food, but we did not distinguish between foraging behavior and other types of motion during the video analyses. Near the beginning of the stimulus phase, rats on the high-fat diet that were implanted moved similar amounts, but over time, these cohorts diverged and sham rats slowly increased their distance moved while sVNS rats decreased their distance moved. By the end of the study, non-implanted rats on the high-fat diet and those in the sham group were travelling similar distances. Likewise, rats on the standard diet and rats receiving sVNS were travelling similar distances, which were less than the distances traveled by the other two cohorts.

While the use of VNS to treat obesity is not new, the waveform used in this study is novel within the field of VNS. The decision to use this waveform stemmed from results in a prior study [49]. In that study, which focused on the tactile system, altering the pulse width on a pulse-by-pulse basis using a sinusoidal shaping envelope at a naturally occurring frequency produced a sensation of pressure in the phantom hand of upper extremity amputees. In [49], a frequency of 1 Hz was chosen, aligning with an average human heart rate. In this study, 0.1 Hz was chosen to align with the typical rate of antral peristalsis in rats. Because the pulse width varied between 10 and 800 μs and given the higher threshold associated with unmyelinated fibers that comprise the majority of axons in the subdiaphragmatic vagus nerves [46], the portions of the stimulus at lower pulse widths were below activation threshold while those at the larger pulse widths were above activation threshold. Computer simulations suggest that the nerves were “paced” at 0.1 Hz but with a dynamically activated (i.e., not static) population of axons on a pulse-by-pulse basis [56]. While the specific dynamic stimulus waveform used in this study is novel, another study has used dynamic waveforms. Using 30 Hz, 30-s-on, 5-min-off bilateral sVNS in pigs, Malbert et al. demonstrated that pulsons decreased daily energy consumption, particularly of high-fat diet, whereas both pulsons and more traditional sVNS reduced consumption of a high-glucose diet [32]. In a recent study, Debelle et al. demonstrated that using a dynamic and adaptive stimulation reduced food intake and hence caloric intake, ultimately reducing body weight in dogs [38].

One of the challenges within the field is that there are several VNS studies that have reported a range of results on body weight, weight gain, food consumption, adiposity, and various metabolic metrics, but these studies have used a wide range of stimulus parameters (current-controlled, voltage-controlled, varying duty cycles, varying pulse widths), target stimulus locations (cervical, thoracic, subdiaphragmatic, unilateral or bilateral), duration of stimulation (ranging from a few days to nearly a year), animal models (rats, rabbits, dogs, pigs), and animal ages. There is little consistency between these studies, making it very difficult to compare results between studies. Our study used Sprague Dawley rats that underwent surgery around day 110 and started receiving bilateral sVNS around day 140 for 90 days until 230 days of age. The sVNS was on for 30 s, off for 300 s, but only applied at nighttime. Stimulus was 3 mA at 30 Hz with a time-varying pulse width controlled by a sinusoid that had a frequency of 0.1 Hz. Other studies have used bilateral sVNS in rats [37, 57, 58], but Johannessen et al. is the most comparable. Johannessen et al. applied sVNS at 30 Hz, on for 30 s and off for 300 s. Whereas our study used a 3-mA stimulus with a variable pulse width ranging from 10 to 800 μs, Johannessen et al. used a pulse amplitude that gradually increased from 0.5 to 2.0 mA with a fixed 500-μs pulse width. Another difference was that Johannessen et al. provided up to 56 days of continuous stimulation whereas this study provided 90 days of stimulation that was only on during the night. Johannessen et al. reported that rats receiving sVNS weighed 10% less than shams at the end of the study. By the end of our study, rats receiving sVNS weighed 11% less than the sham group, putting our results in line with those presented with Johannessen et al. Our results were also similar to the 10% difference in body weights observed in pigs receiving bilateral thoracic VNS using stimulation parameters that were nearly identical to Johannessen’s but over a 98-day period [35]. Other studies using similar stimulus paradigms in different animal models, in different target locations, and/or for different durations of time have reported similar amounts of weight loss between their VNS groups and the sham groups [32, 35, 54, 55]. What begins to emerge is the possibility that this stimulus paradigm, while effective at reducing weight, is limited. Given the range of variation in the studies, comprehensive research that maps stimulus input to behavioral output is needed.

Another observation from this study is that the duration of post-surgical recovery provided by many studies may be too short. We found that weight gain in implanted animals did not return until approximately 4 weeks after surgery. This longer recovery time is likely influenced by animal model, age, target location, surgeon, surgical technique, and surgical implant. Studies that initiate VNS before full recovery may be confounded by the effects of surgery if they are not carried out long enough. Therefore, it is important to consider recovery time in the study design.

Finally, it is worth noting that rats in the control group on the standard diet do not appear to have been aged as long as rats in the other three groups (Fig. 1). In fact, 4 rats in the control group were aged significantly longer to 320 days. However, body weights collected between 190 and 270 days of age were lost due to a corrupted file. The 5 remaining rats in this control group that came later in the study were not aged beyond 190 days because there was not sufficient time to age rats to 270 days (when the corrupted data file was fixed) due to COVID disruptions that required adjusting the study timeline.

This study has limitations, the most significant of which was that stimulation of the vagus nerves was not verified by secondary means and, as such, we cannot say with certainty that the sVNS altered nerve activity. Although lead impedance was monitored daily, this only confirmed the health of the implant. Future studies should monitor the animal using secondary measures to confirm that the electrical stimulation activated the vagus nerve. Further, while the outcomes were significant and strongly suggest that electrical stimulation activated the nerves, it is unclear if the population of activated axons was exclusively afferent, exclusively efferent, or a combination of both, the latter of which is most likely. As such, techniques to monitor parasympathetic tone, such as heart rate variability and gastrointestinal transit time, should be considered. Alternatively, periodic monitoring with fMRI or PET could confirm if sVNS affected activity in the brain, particularly food-based and satiety centers such as the ventromedial hypothalamus, dorsal medial nucleus of the hypothalamus, lateral hypothalamus, and ventral tegmental area.

Another limitation of this study comes from using a rodent model. The use of rodents in preclinical studies, including obesity studies, is well-established. However, there are inherent limitations to using animal models. The body weight of Sprague Dawley rats tends to plateau around 18–24 months [59]. Waiting this long before implanting and delivering sVNS carries significant housing expenses, increases the risk of surgical mortality, and provides little time to conduct a longitudinal experiment before the obese rat would likely succumb to old age and comorbidities. As such, studies using younger rats that are still growing are limited to analyzing outcomes in terms of a reduction in weight gain rather than weight loss. This makes it difficult to compare outcomes between preclinical and clinical studies.

Another limitation associated with the use of rodents comes in the form of husbandry. Because there were percutaneous leads from the implant, rats had to be housed individually to protect the leads from damage that would occur during rough play or grooming. Individual housing also allowed practical quantification of daily food consumption and motion tracking. However, rats are social animals and isolation is known to change their behavior. As such, all rats in this study could have been affected by an unaccounted variable: isolation. To try to minimize the effects of isolation, we used large cages that allowed plenty of space to play. The cages were clear acrylic, allowing rats to see each other. Cages were open on the top, allowing rats to smell their neighbors. We also dedicated time each week to handling, petting, and interacting with each rat to provide additional enrichment.

A further limitation is associated with the specific strain of rat used in this study. As noted, these were obesity-prone Sprague Dawley rat from Charles River Labs for consistency with prior studies. Unfortunately, Charles River Labs discontinued this animal model during the study. As a result, we established breeding colonies. However, because there were a limited number of rats to start the colony, the genetic diversity within the colonies diminished over time. It is uncertain if there was any effect on outcomes, but effects should have been minimized by randomizing the cohort to which a rat was assigned.

If the ultimate goal is to translate sVNS to the clinic to treat obesity, then the technology must compete with well-established bariatric surgical treatments as well as the newer semaglutide GLP-1 and tirzepatide GLP-1/GIP agonists. Currently, the weight loss associated with these other methods exceeds weight loss with sVNS, although, as noted above, rodents were still in the growth phase during this study so the comparisons are difficult. This highlights a final limitation of this and other preclinical studies: the sVNS parameters have not been optimized. In general, optimization is a significant challenge because there are three independent stimulus parameters (pulse width, pulse amplitude/voltage, stimulus frequency), all of which can be a function of time, which itself can be applied on a pulse-by-pulse basis (short time scale) as well as used to define an on–off duty cycle (long time scale). There are myriad combinations of stimulus parameters within this multidimensional space. As such, it is likely that preclinical studies will be limited in their success and they may never achieve the weight loss seen with bariatric surgery or drugs. Indeed, it may only be possible for sVNS to achieve better outcomes when descriptive verbal feedback is available to guide stimulus parameter decisions.

The ultimate goal is to successfully translate optimized sVNS to the clinic. There are several future steps that must occur. Secondary verification is essential for translation to clinical trials. Indeed, our current studies include longitudinal fMRI and connectomics to both confirm that sVNS is activating axons and to better understand the effect of varied stimulus waveforms and chronic diet on activated pathways. We are investigating methods to optimize the stimulus parameters with the goal of establishing a standardized protocol that will allow us to set the best stimulus parameters on an individualized basis. Ideally, future studies will allow the animals to be housed together and interact socially. To this end and once there is a smaller range of near-optimal stimulus parameters, the implants should be fully internalized. This includes the cuff and the stimulator. Development of small, hermetically sealed, multichannel implantable pulse generators that support wireless programming, have either an extended battery life or support wireless recharging, and are affordable should be actively pursued. Because rats consume food periodically throughout the day, such a system does not have to be closed-loop. However, when such a system is translated to humans that eat less frequently, the implantable system should be closed-loop and triggered by stomach stretch or food consumption rather than open-loop and required to operate continuously. Finally, finding the best stimulus parameter combinations may not be possible until verbal descriptions can be used to guide parameter choices. To this end, short-term implants that piggyback off of scheduled bariatric surgeries may provide the best pathway to optimization.