Evidence in preclinical models and healthy humans has highlighted a role for CB1 receptors and AEA in fear extinction [9, 19, 23, 42, 43]. As extinction learning forms the foundation for exposure-based CBT, our goal was to determine if elevating AEA together with exposure-based CBT would be more effective than CBT alone. While PTSD symptoms did significantly improve over time in both treatment groups, there was no effect of FAAH inhibition on PTSD symptoms, despite the fact that the drug was confirmed to elevate AEA levels. We also found no effect of FAAHi on any secondary measure, including self-reported symptoms of PTSD, anxiety, depression, or sleep quality.

Previous studies exploring FAAHi as a therapeutic intervention have shown modest results. The inhibitor used here showed some evidence of an anxiolytic effect in participants with social anxiety disorder, but did not meet the primary endpoint [21]. Subsequent analyses found that the selected dose (25 mg q.d.) may not have been high enough to maintain the level of FAAHi needed to reduce clearance of AEA at trough. Indeed, drug PK and AEA level analysis suggested that participants with drug and AEA levels in the upper tertile did experience a greater symptom reduction. Based on these findings and PK/PD modeling, we doubled our dose to 25 mg b.i.d. Despite this, we found no effect of FAAHi on PTSD symptoms. Analyses covarying for AEA or drug levels failed to show a significant effect. This same dose also recently failed to meet its primary outcome in a study of adolescents and adults with autism spectrum disorder, though small-to-moderate effects were reported on some secondary outcomes [44].

A possible reason for the failure to detect a drug effect is that the efficacy of our iCBT intervention may have resulted in a ceiling effect. However, we consider this to be unlikely for multiple reasons. First, for the first four weeks of the study, drug was administered in the absence of iCBT specifically to allow an evaluation of potential drug effects in the absence of behavioral treatment. Despite this, between-group differences were not evident on outcome measures, including the secondary outcome of self-reported PTSD symptoms, which was highly correlated with clinician-assessed PTSD measurements at weeks 0 and 12. Thus, one month of ‘FAAHi only’ had no advantage over ‘placebo only’. Furthermore, a recently completed, yet unpublished multi-center, randomized, controlled clinical trial compared another FAAHi, JZP150 to placebo in the absence of any psychotherapy. It was recently reported via press release that JZP150 was ineffective in improving PTSD symptoms over 12 weeks [45]. Collectively, available data suggest that FAAHi, whether administrated alone or with psychotherapy, does not reduce PTSD symptoms.

A potential limitation is our decision to use iCBT instead of Prolonged Exposure (PE), the gold standard PTSD treatment. However, we selected iCBT as our approach for several reasons. First, while PE is the gold standard for PTSD, it is highly resource demanding, resulting in a clinical bottleneck due to lack of available trained therapists [5]. This makes it difficult to deploy on a large scale and limits accessibility. It also requires patients to regularly commute to the clinic, which was particularly challenging during the COVID-19 pandemic, when this study was completed. Moreover, meta-analytic evidence suggests that iCBT may be as effective as face-to-face CBT in the treatment of mild to moderate disorders [34]. Finally, given the easier clinical implementation, but potentially lower efficacy of iCBT vs PE, a medication that could potentiate iCBT effects would be a particularly attractive combination to reach a large patient population with minimal burden to patients or clinical resources.

Our study was initiated based on converging data from animal models and human experiments consistently supporting the notion that elevating AEA via FAAHi might be an efficacious intervention for PTSD (for review, see ref. [9]). This conclusion was scaffolded by extensive literature that identified a critical role for CB1 receptors in fear extinction and stress response buffering [14] and human behavioral and imaging results with THC, a partial CB1 agonist [46, 47]. It was further supported by our own previous work in healthy humans, showing that FAAHi enhanced fear extinction when tested 24 h after fear acquisition and attenuates subjective and autonomic stress reactivity, as well as stress-induced negative emotions [19]. Exploring the relationship between laboratory models of fear conditioning and PTSD symptoms could provide insight into the predictive validity of these measures. However, a recent review suggests that drug effects to facilitate fear extinction in animals can translate to healthy humans, and sometimes even patients with PTSD; however, even then, these compounds are not successful in improving clinical PTSD symptoms [48]. This suggests a need for considering how outcomes from lab-based models of fear conditioning relate to real-world behaviors in patients with PTSD.

Our previous study in healthy volunteers used PF-04457845, an irreversible FAAH inhbitor [19]. In contrast, JNJ-42165279 is slowly reversible [30]. It was previously shown to dampen limbic responses during an emotional faces task, but had no effect on within-session extinction learning in healthy adult males [49]. It could be speculated that the reversible nature of FAAH inhibition by JNJ-42165279 accounts for this lack of effect. However, our previous study with PF-04457845 also failed to show a significant effect on within-session extinction, and was only effective in promoting fear extinction when tested 24 h after extinction [19]. As our iCBT treatment took place over several weeks with drug on board for the duration, we still expected to see the therapeutic effects of enhanced extinction learning, as evidenced by greater efficacy of the iCBT intervention. Thus, it could be speculated that, in clinical populations, the partial agonist activity of AEA is insufficient to robustly influence symptoms of pathological fear. If so, a more effective approach could be to increase levels of 2-AG, which acts as a full CB1 agonist, via inhibition of its degrading enzyme, monoacylglycerol lipase (MAGL) [9, 50]. Recent translational evidence suggests that deficient 2-AG signaling can promote fear generalization [51]. However, chronic administration of MAGL inhibitors can lead to desensitization of CB1 receptors [52], suggesting that dosing and duration of administration would need to be carefully considered.

Two important alternative interpretations of our data are possible. First, impaired ability to extinguish fear memories is thought to be a core symptom of PTSD, leading to the expectation that drugs with an ability to promote fear extinction under laboratory conditions would be clinically beneficial [8]. It is, however, possible that the pathological fear memories in PTSD are generalized, over-consolidated, or both, in a manner not captured by laboratory assessments of fear learning and extinction in healthy people. Secondly, while FAAHi treatment was able to elevate AEA levels in our clinical population to an extent similar to that previously observed in healthy controls [19], our findings suggest that disrupted CB1 signaling may be part of PTSD pathophysiology, preventing the expected beneficial effects of elevated ligand levels.

A limitation of this study is that 85% of participants were female, and the study was not adequately powered to detect sex or gender-specific effects. Preclinical literature has found that manipulations of both the CB1 receptor and FAAH can influence anxiety and fear extinction in males, but more recent evidence suggest that these manipulations produce incongruent or null effects in females [53–56]. Thus, the possibility exists that our results could have been influenced by a sample enriched in females. Our previous studies in healthy humans assessing genetic variation in AEA levels or pharmacological elevation did not produce gender-specific effects, though again, these studies were not powered to directly assess this potential interaction. However, recent cross-sectional data suggests that AEA levels maybe decreased in men, but not women, with PTSD [57]. Future studies systematically exploring sex- and gender-based differences in eCB function and related behaviors will be critical in determining the translational validity of these findings and subsequently developing novel pharmacotherapies more broadly.

Additional features of our sample and study design could have impacted our outcomes. For instance, we have only cursory information on trauma exposure, and more specific information, such as age at trauma or time since trauma, could be relevant to include for future studies. Further, while we predicted that FAAHi could reduce the aversiveness of the therapy, it is possible that this could also attenuate the efficacy of the therapy. However, we saw no effect of drug on ratings of distress during completing the therapy. Overall, we found that FAAHi did not enhance the response to iCBT for PTSD in our sample of primarily women. Future research should consider investigating whether eCB function is indeed altered in PTSD, and how this may be related to fear memory processes.