Obesity is among the most frequent preventable causes of mortality in the developed world and is associated with several different causes of death, including heart disease, type 2 diabetes, and some forms of cancer (see Haslam and James, 2005, for an overview). As part of an effective treatment strategy for obesity, it may be useful to supplement behavioral control of food intake with the administration of drugs to modulate the processes by which the brain generates eating behavior.

The endocannabinoid system in the brain, involving cannabinoid type 1 (CB1) receptors, may regulate feeding behavior by modulating the reward signal associated with the consumption of food (Solinas et al., 2008). Rimonabant, a CB1 antagonist and inverse agonist (Pertwee, 2005), has been found to downregulate feeding and promote weight loss (Scheen et al., 2006), acting selectively on the consumption of sweet foods (Maccioni et al., 2008; Mathes et al., 2008). Furthermore, a reduced sensitivity to reward and to the effects of cannabinoid compounds has been demonstrated in animals with a lower expression of CB1 receptors in the striatum (Friemel et al., 2014).

Rimonabant was, however, withdrawn from clinical use in the EU in 2008 following mounting evidence of increased rates of depression following its use (Christensen et al., 2007). The rodent studies used in early screening of the drug largely failed to detect this potential for depressogenic side effects and in fact suggested antidepressant effects instead (Griebel et al., 2005). This highlights the importance of developing a human experimental medicine model for the development of weight-loss treatments, such that most adverse psychiatric effects of the drug can be detected before clinical use is approved.

Using a human experimental model of reward and aversion processing in the brain with functional magnetic resonance imaging (fMRI), we previously showed that 7 days’ treatment with rimonabant reduced the neural response to rewarding chocolate stimuli in key reward areas such as the ventral striatum, putamen, and orbitofrontal cortex (OFC) (Horder et al., 2010). We have also demonstrated a similar pattern of reduced response using the same task with recovered depressed patients (McCabe et al., 2009). This similarity suggests a mechanism inducing anhedonia, which could lead to the increased risk of depressive symptomatology seen in clinical use of the drug. Rimonabant also increased lateral OFC activations to the sight and taste of an aversive food, and these effects might underlie its ability to promote weight loss.

Like rimonabant, tetrahydrocannabivarin (THCv) acts on endocannabinoid CB1 receptors but does so as a neutral antagonist (Pertwee, 2005). Neutral antagonists have been suggested as a potentially safer alternative to rimonabant in the treatment of obesity (Meye et al., 2013) and appear to entail fewer side effects such as nausea, while still attenuating operant behavior in animals (Bergman et al., 2008). The increased safety and tolerability of neutral antagonists compared with rimonabant is not associated with an attenuation of the therapeutically valuable antagonistic action at CB1 receptors (Kangas et al., 2013). Most promisingly, recent animal studies have shown that neutral antagonists appear to reduce food intake and weight gain as rimonabant does, but unlike rimonabant do not suppress activity in emotion regulation areas such as the ventral tegmental area and basolateral amygdala (Meye et al., 2013). However, how THCv interacts with the reward and aversion systems in the human brain is as yet unknown.

Therefore, in this study we aim to examine the effects of THCv using our model of reward and aversion processing in the human brain. We hypothesize that as THCv is thought to be free from the depressogenic side effects seen with rimonabant, it will leave intact responses to rewarding stimuli in areas that our task (Rolls and McCabe, 2007; McCabe et al., 2009) and other studies (O’Doherty et al., 2001) have shown to be involved in reward processing, such as the anterior cingulate cortex (ACC), OFC, and striatum. To the extent that THCv nonetheless produces therapeutically valuable modulations of food aversion, we hypothesize increased activations to aversive stimuli in areas that have been shown to be involved in the processing of food aversion, such as the amygdala, insula, and lateral OFC (Zald et al., 2002; McCabe et al., 2010, 2011, 2012).

The aim of the current study was to examine the effects of a single dose (10mg) of the CB1 receptor neutral antagonist THCv on the neural processing of rewarding and aversive food stimuli in the healthy human brain. We found that there were no subjective effects of a single dose of THCv on ratings of pleasantness, wanting, and intensity of the stimuli made inside the scanner; however, we did detect significant neural differences. This study is thus further evidence that neuroimaging may be a more sensitive measure than a subjective report of the early effects of drug treatments and can provide an understanding of the neural mechanisms through which behavioral changes eventually emerge with repeated treatment (Harmer and Cowen, 2013).

In this study, we found enhancement of activation under the positive chocolate stimuli conditions in the ACC and striatum and in the striatum (caudate) mid cingulate and midbrain. Consistent with these findings, the striatal region has been shown to have a dense concentration of cannabinoid receptors located on striatal projection neurons (Herkenham et al., 1991a), and the cingulate cortex also has high levels of cannabinoid receptors (Sim-Selley et al., 2002). Both these regions have been previously highlighted in relation to reward processing and the endocannabinoid system (van Hell et al., 2012). This outcome is in direct contrast to our previous study, which found that the CB1 antagonist/inverse agonist rimonabant reduced activations to reward in key reward areas such as the OFC, ventral striatum, and putamen. We expected that, unlike rimonabant (with its depressogenic side effects), THCv would not reduce activations to the rewarding chocolate conditions in our study. In fact, we found that THCv increased activation to the sight of chocolate in a very similar region to that which was reduced under rimonabant, the ventral striatum/putamen region. These results are important in that reduced reward processing has been previously shown in patients with depression (Keedwell et al., 2005) and also in our own studies in recovered depressed patients (McCabe et al., 2009). Thus, the current results support the idea that THCv does not impair reward function, and this may be related to a potentially safer side effect profile.

Interestingly studies in obesity show a hypofunctioning striatal response to food taste, which can be further reduced by overeating (Stice et al., 2010). In our study, we examined healthy people and found effects on reward response to sight but less to taste. Given that obese individuals have a greatly reduced response to reward to begin with, THCv may produce a more pronounced enhancement of reward response to taste than observed in the present study for healthy individuals. This enhancement could act to rebalance the reward system in response to food and help obese people reduce the overeating that may occur as an overcompensation for hypofunctioning striatal regions. Our results with healthy people are a first step towards understanding the action of THCv. The present study would need to be repeated in obese individuals to clarify whether THCv modulates reward activity to food stimuli in the same way and whether over time this leads to a reduction in eating.

We also found that THCv increased activations to the aversive conditions in a ventral striatum/putamen region similar to that in which aversion-related response was reduced under rimonabant. We discussed in our previous paper how this might suggest that rimonabant reduced activity in motivational areas to both pleasant and less pleasant stimuli (Horder et al., 2010). However, in the present study, regions such as the caudate and the putamen showed increased activity under THCv for both the positive and negative visual stimuli. This might indicate that THCv is not dampening down motivational processes, as these areas are involved in motivation and action in relation to both positive and negative events. Taken together, our results support the notion that THCv is less likely to cause depression-like side effects in humans, because it does not attenuate activations to either positively or negatively valenced stimuli in key motivation areas of the brain.

We also found enhancement in the mid brain to the chocolate condition and not the aversive condition under THCv compared with placebo. Studies have revealed that CB1 receptors are on axon terminals that impinge upon midbrain neurons (Melis et al., 2004; Riegel and Lupica, 2004; Matyas et al., 2008) and that this might be a mechanism by which cannabinoids can modify reward behavior through actions in the ventral tegmental area of the midbrain (Wang and Lupica, 2014). Interestingly, we only found modulation of this region in relation to the positive reward condition, which is consistent with the report that dopamine neurons preferentially report stimuli with appetitive rather than aversive motivational value (Mirenowicz and Schultz, 1996). This result again suggests that THCv does not inhibit the neural response to reward and may even enhance it. However, it is worth mentioning that we did not find any subjective effects of the drug, which most likely was due to the acute single dose. Thus, future studies would benefit from the examination of longer term treatments and higher doses to truly map the mechanism of action and how potential subjective effects might be related to the neural responses.

In the current study, we also found that for the combination of the taste and sight of the less pleasant stimuli there was enhanced activation under THCv in the amygdala and insula. Aversive taste processing has been shown to involve the amygdala (Nitschke et al., 2006), and studies have identified distinct amygdala neuronal populations responsible for processing aversive information (Paton et al., 2006). The anterior insula is known to be part of the gustatory system (Faurion et al., 1998) but is also involved in the processing of aversive stimuli and disgust (Liu et al., 2011). Although no studies to date have directly examined the effect of THCv on the neural basis of aversive food processing in humans, it is interesting to note that one common feature between rimonabant and THCv is that both treatments increased activation to the aversive taste and sight condition in areas thought to be involved in negative information processing, such as the lateral OFC under rimonabant and the amygdala, insula, thalamus, and caudate under THCv. Although the amygdala and insula do indeed respond to positive information, in the present study we found that THCv increased amygdala and insula activity only to the aversive stimuli and in the case of the amygdala only when a taste was presented. This is likely to be a specific valence effect acting on the consummatory aspect of the unpleasant food and thus might be a mechanism by which THCv can make less pleasant food more salient and more unpleasant. By increasing the tendency on the part of appetitive regions to assess food as unpleasant, this effect may decrease time to satiety as food becomes unpleasant on repeated consumption, in turn reducing overall consumption. This is plausible in light of a study by Tallett et al. (2008), which found that the behavioral satiety sequence (time to stop feeding) was accelerated under a rimonabant and naloxone combination in rats. Because one aspect of satiety is the gradual change in the subjective experience of food from pleasant to unpleasant (Saker et al., 2014), it is possible that our finding of increased brain activity to less pleasant food stimuli under THCv might be a mechanism by which drugs that aid food reduction decrease time to satiety.

Analysis also revealed increased activity to the less pleasant food stimuli under THCv in the mid OFC. The OFC is a key player in the reward system and has been shown to be specifically related to the reward value of a stimulus and therefore to the decisions made based upon choices and preferences (Levy and Glimcher, 2012). It is also highly populated with cannabinoid receptors (Herkenham et al., 1991b). Our results indicate that THCv can modulate activity in OFC during the presentation of both the less pleasant taste and sight. This could be a mechanism by which choices and decisions made in relation to food preferences are modulated under cannabinoid antagonist treatment.

We also found that there was increased activity to the aversive stimuli under THCv in the superior temporal gyrus, an auditory and linguistic region of the human brain. A recent study examining the effects of THC on the human brain found that activity in this same region was decreased under THC in healthy volunteers and that the decrease was associated with an increase in psychotic symptoms (Winton-Brown et al., 2011). Right hemisphere activations play a role in resolving ambiguity in language (Harpaz et al., 2009), and as our results are only on the right hemisphere, perhaps THCv is having some effects on the clarity of decision making in the less pleasant conditions. This is, of course, speculative, as we did not have a task in this study that directly addressed the processing of auditory or linguistic cues.

In our recent study examining the effects of naltrexone (an opioid drug used to treat alcoholism) using the same task as described here, we found naltrexone also enhanced responses to aversive stimuli but had no effects on the striatum response to reward (Murray et al., 2014). Naltrexone is by itself a weak antiobesity agent (Atkinson et al., 1985), and this may be because it has limited effects on the ventral striatal response to reward. THCv, on the other hand, enhanced striatal responses to the sight of chocolate in the present study. As it has been shown that people “at risk” of obesity have an increased neural response to reward (Stice et al., 2008) as do binge eaters (Filbey et al., 2012; Schienle et al., 2009), we suggested that drugs such as naltrexone that reduce reward are effective as preventative treatments in those “at risk” of obesity or binge eating but not for people who are currently overweight and already have reduced reward processing.

A recent study by Wang et al. (2014) tested the effects of a combined bupropion (dopamine reuptake inhibitor) and naltrexone treatment on brain responses to food cues in overweight females. Interestingly, they found no effects of the food pictures in the striatum, a finding that the authors attributed to the task’s lack of sensitivity, but which could also indicate the hypofunctioning reward systems of the overweight volunteers. Similar to our results, they found enhancement in the anterior cingulate to food cues in drug vs placebo, which they describe as being a mechanism by which enhanced inhibitory control over food is induced due to the role of the ACC in cognitive control. Taken together, our results in healthy people show some enhancement in striatal and cognitive control areas that might make THCv a beneficial treatment in obese individuals with hypofunctioning reward systems and lack of control over food intake.

It should be noted that THCv also has a high affinity as a partial agonist for CB2 receptors, and studies have shown that CB2 receptors, although thought to be primarily involved in immune function (McPartland et al., 2014), can also be antagonized by THCv (Pertwee, 2005; Thomas et al., 2005). As CB2 receptor agonists act on the brain to induce food intake (Emadi et al., 2011), it is possible that CB2 antagonism by THCv could also reduce food intake. However, to date, how effects on CB2 receptors are related to neural responses to food in humans is as yet unknown, so effects of THCv on CB2 receptors in relation to food cannot be ruled out (Mechoulam et al., 2014).

In conclusion, our results are the first to examine the effects of THCv on the neural response to reward and aversion in the human brain. We found that THCv did not reduce reward processing but rather enhanced activations in key reward and aversion processing areas. There are contrasts between these effects and those we previously identified of rimonabant, indicating that there may be meaningful differences between the clinical effects of rimonabant and THCV. These data support functional neuroimaging as an important tool in drug development, one that allows the neural mechanisms underlying behavior to be studied in humans. It thereby may aid the detection of clinically relevant effects of treatments over shorter periods of time and speed the development of effective drugs and the rejection of unsuitable ones.

Dr. McCabe has acted as a consultant to P1Vital, Givaudan, GWpharma, the British Broadcasting Company, and Channel 4. Professor Cowen is a member of an advisory board for Lundbeck. Luke Tudge and Clare Williams report no biomedical financial interests or potential conflicts of interest.