Circulating endocannabinoids and genetic polymorphisms as predictors of posttraumatic stress disorder symptom severity: heterogeneity in a community-based cohort

Abstract

Introduction

Each year, approximately 2.5 million people in the United States are involved in a traumatic event that requires emergent care at a trauma center [1, 2]. Epidemiological studies indicate that over 10% of the general population has a diagnosis of post-traumatic stress disorder (PTSD) [3], and those who have experienced a traumatic injury are at greater risk for the disorder than the general population [4]. PTSD following injury is negatively related to the physical and mental quality of life after trauma and occurs in 25–40% of those traumatically injured [5]. Given these data, understanding factors that predict risk for the ultimate development of PTSD will inform better clinical approaches to the treatment of trauma that prevent PTSD.

Considerable evidence indicates that the endocannabinoid signaling system (ECSS) is stress-responsive and contributes to the regulation of anxiety and fear [6]. The ECSS includes the CB1 cannabinoid receptor (CB1R) and two endogenous ligands, N-arachidonoylethanolamine (AEA) and 2-arachidonoylglycerol (2-AG) [7]. CB1Rs are highly expressed throughout the brain, including regions involved in the processing of fear and formation of memories [8]. Brain ECSS subserves activity-dependent, retrograde synaptic signaling, thus plays a vital role in moment-to-moment synaptic activity [9]. Importantly, glucocorticoids [10, 11] and corticotropin-releasing factor [12] regulate brain concentrations of 2-AG and AEA, respectively; and many preclinical studies demonstrate that the ECSS is an important link between stress exposure and changes in synaptic activity, particularly in limbic brain regions [13].

Preclinical studies demonstrate that hypoactivity of the ECSS produces effects that are similar to symptoms of PTSD, including increased anxiety-related behaviors [14], deficient extinction of aversive memories [15], and sleep dysregulation [16]. On the other hand, the elevation of AEA-mediated CB1R activation through inhibition of fatty acid amide hydrolase (FAAH) is anxiolytic under aversive conditions [14]; promotes fear extinction, a component of coping and adaptation following traumatic stress [17]; and normalizes PTSD-like symptoms in a rat model [18]. In a clinical sample exposed to the 9/11 World Trade Center attacks 4–6 years prior, diagnosis of non-remitting PTSD was significantly related to reduced concentrations of circulating 2-AG compared to trauma-exposed non-PTSD individuals [19]. Taken together, these studies support the hypothesis that hypoactive ECSS could exacerbate while increased ECSS activity could alleviate PTSD symptoms.

Polymorphisms in human genes for proteins of the ECSS have been identified that support a role for the ECSS in emotional regulation [20]. In particular, polymorphisms in the CB1R gene (CNR1) are associated with anhedonia in individuals exposed to childhood abuse [21] and could interact with other genes to contribute to anxiety phenotypes [22]. A genetic polymorphism in FAAH (rs324420) influences amygdalar reactivity in response to threatening, fearful, and angry faces in healthy subjects [17, 23]. Carriers of the A allele at this locus exhibit quicker habituation of amygdala reactivity to threat and low scores on a personality trait of stress-reactivity [17] but, in individuals with anxiety disorders, A allele carriers are more prone to anxiety [24, 25].

Thus, there is emerging evidence that the state of activation of the ECSS could contribute to the symptoms of established PTSD. However, what is not known is the relationship of the ECSS to the risk for the development of PTSD. To address this gap, we assessed serum 2-AG and AEA concentrations soon after trauma in traumatically injured individuals and examined their correlations with PTSD diagnosis and symptom burden 6–8 months later, particularly for ethnic/racial minority patients and women, who have been shown to be at greatest risk for PTSD following trauma [26]. Our central hypothesis, based upon the preclinical literature, proposed that elevated serum concentrations of 2-AG and AEA at the time of trauma would be associated with reduced PTSD symptoms 6–8 months later. Similarly, we predicted that individuals with the minor allele at rs324420 would exhibit less PTSD symptom severity. Given that several polymorphisms in the gene for the CB1R have been associated with altered risk for the development of psychopathology [13], we carried out exploratory studies of CNR1 polymorphisms at rs1049353, rs806371, and rs2180619 in association with risk for PTSD without a priori hypotheses.

Methods and materials

Results

Discussion

The overall purpose of this study was to explore and describe the associations between measures of endocannabinoid system function and post-traumatic stress disorder symptoms following injury in a community population. Our community study sample was relatively large (170 subjects) and diverse, as just over 50% were a racial or ethnic minority and 70% of the sample were male. The types of injury sustained were also diverse, just over 25% experienced an assaultive type injury and about 40% were injured in a motor vehicle accident. There was a striking difference in susceptibility for an increase in PTSD symptoms in the 6–8 months following the injury between the CNH and minority groups, with the former showing no increase in symptoms and the latter a significant increase. Thirty percent of the sample met the criteria for PTSD at 6–8 months, with minority women having the greatest PTSD symptom severity at this time point.

Our first objective was to explore the relationships between endocannabinoid concentrations at the time of hospitalization and PTSD symptoms 6–8 months later. Individuals with PTSD symptoms that persist beyond 6 months after injury are considered to have non-remitting PTSD, a condition which significantly reduces the quality of life [33]. When the data from all participants are considered together, individuals diagnosed with PTSD six-eight months after traumatic injury had significantly higher circulating AEA and a trend toward an increase in circulating 2-AG concentrations at hospitalization than those without a PTSD diagnosis. Importantly, circulating AEA concentrations at hospitalization were significantly, positively correlated with an increase in PTSD symptoms between hospitalization and 6–8 months follow-up, suggesting that elevated AEA at the time of trauma is associated with increased risk for the development of PTSD over time. When the data from men and women were considered separately, the associations between AEA concentrations at hospitalization and PCL5 scores at follow-up were more robust in women. Cortisol concentrations measured in the same blood samples trended to be lower in those who were diagnosed with PTSD at 6–8 months, which is consistent with some earlier studies particularly when the individuals had been subjected to early-life adversity [34]. Importantly, these findings suggest that the positive correlations between peritraumatic endocannabinoids and PTSD diagnosis at follow-up are not explained by higher cortisol concentrations evoked by the trauma.

The relationship between PTSD severity at follow-up and 2-AG concentrations at hospitalization was stronger and reached statistical significance when the data from ethnic and racial minorities were examined separately. In racial minority subjects only, greater 2-AG at hospitalization was significantly related to PTSD symptom severity at follow-up, with these findings particularly driven by the reexpriencing and avoidance subgroups of PTSD symptoms.

While the clear rationale for these differences based on sex and race is beyond the scope of this study, it is clear that the endocannabinoid stress response is not homogenous in humans, particularly in the acute aftermath of trauma. Further work will need to include an investigation of biologically based sex and racial differences and contextual factors that affect stress systems in the acute aftermath of a trauma. For example, recent work by our group has found that experiences of discrimination predict PTSD after trauma beyond acute stress [35]. Thus, it is possible that chronic stress prior to trauma, from such experiences as discrimination, exposure to community violence, and resource deprivation may alter the endocannabinoid system such that the endocannabinoid response to trauma is different compared to someone living in less chronically stressful conditions. Other findings suggest that prior trauma exposure impacts biologically based stress responding in the acute aftermath of a subsequent trauma, for instance, related to HPA axis functioning and the release of cortisol [34]. Importantly, manipulation of endocannabinoid signaling has been proposed as a potential target for intervention in PTSD [6] but further research on sex and race differences is warranted to inform such interventions.

Circulating AEA concentrations at hospitalization were associated most strongly with follow-up symptoms of avoidance and also trended to be associated with arousal. On the other hand, circulating 2-AG at hospitalization was most strongly associated with the reexperiencing symptom cluster at follow-up. The nonoverlapping associations of each endocannabinoid with PTSD symptoms suggest that they contribute differently to the plasticity that occurs between the time of injury and the development of non-remitting PTSD. This observation is in accord with the preclinical literature demonstrating that AEA and 2-AG have both overlapping and different roles in the regulation of amygdalar signaling [36]. We postulate that the stress of the injury mobilizes both endocannabinoids and that they contribute differently to changes in emotional circuitry in a manner that is modulated by sex and other factors.

Importantly, these findings indicate that high endocannabinoid tone in the peritraumatic period is associated with increased risk for the development of non-remitting PTSD, rather than reduced risk as originally hypothesized based upon studies showing that circulating endocannabinoids are lower in individuals with chronic PTSD [19, 37] and that an increase is associated with improved symptoms [38]. Additionally, a wealth of preclinical literature supports the notion that reduced endocannabinoid tone mirrors the symptoms of chronic PTSD [14–16]. A recent study examining proinflammatory cytokines in blood harvested within four hours of a traumatic injury found that low concentrations of interferon-gamma (IFNγ) and tissue necrosis factor-alpha (TNFα) were associated with increased chronic PTSD [39]. Like the present results, these cytokine findings are seemingly incongruent with data that proinflammatory cytokines are consistently found to be positively associated with chronic PTSD [40]. Thus, responses to trauma are dynamic and protective factors in the peritraumatic period could be associated with worsened symptoms in chronic PTSD.

While a recent study in rats found inhibition of FAAH after a trauma inhibited the development of long-term symptoms of anxiety and fear [41], other studies have found high concentrations of AEA facilitate affective, negative memories of the trauma [42]. Possibly, high concentrations of AEA contribute to PTSD through the development of negative emotional memory formation which in turn may lead to difficulty with fear extinction [43], a core component of PTSD [44]. In support of this hypothesis, Morena, Hill, and colleagues have provided evidence in rats that high FAAH activity in the amygdala is associated with low AEA concentrations and reduced HPA axis activation, anxiety, and expression of conditioned fear [45]. They suggest that, in the absence of stress, AEA tonically activates CB1 receptors on GABAergic neurons, resulting in suppression of this inhibitory input. Stress, which has been shown to increase FAAH activity and reduce AEA, would disinhibit the GABAergic neurons, resulting in suppression of amygdala activity.

Importantly, the current and previous findings suggest that role of AEA may change over time following trauma. Given that clinical studies also find that individual trajectories of PTSD symptoms emerge after trauma [5, 34], further work should elucidate the associations of trajectories in AEA concentrations with symptom change across time.

Our second objective was to examine the associations between circulating endocannabinoid concentrations and PTSD symptoms determined at the same time. Earlier studies addressing this question have provided conflicting results. In a study of individuals with long-standing PTSD following the World Trade Center attack, circulating 2-AG concentrations were lower in those with PTSD than trauma-exposed controls and AEA concentrations were negatively correlated with intrusive thoughts of the trauma [19]. On the other hand, a study of individuals with severe PTSD (mean CAPS scores of 95), found that both AEA and 2-AG were significantly elevated in individuals with PTSD compared to trauma-exposed individuals without PTSD [46]. In the current study, the strongest associations between contemporaneous measures of circulating endocannabinoids and PTSD severity were seen when the sample was stratified by sex or minority status. In women, AEA concentrations were strongly, positively correlated with overall PTSD symptom severity and with each symptom cluster individually at follow-up. These findings are contrary to some preclinical work suggesting that higher CNS AEA is related to reduced anxiety-like behavior [14], startle response, and extinction retention central to recovery following trauma [18]. However, inhibition of FAAH in female but not male rats was associated with impaired extinction of fear memory through a non-CB1 receptor mechanism [47], suggesting, as the current study does, that the relationship of FAAH and AEA to the risk for development of PTSD is sex-dependent.

Conversely, AEA concentrations and overall PTSD symptom severity were negatively correlated at hospitalization in individuals endorsing Black and American Native race and Latinx ethnicity. Among the individual symptom clusters, avoidance was most highly negatively correlated with AEA and significant, negative correlations were seen between AEA and reexperiencing and arousal, suggesting lower AEA concentrations are related to higher PTSD symptom severity in this subgroup at the time of injury. It is interesting that these associations did not emerge in the data obtained at follow-up. Our findings in ethnic and racial minorities may be an indication of ongoing sociocultural stress which has shown to increase the risk for PTSD [48], particularly in those living in urban environments. Our sample in this study was predominantly urban and further research is warranted to understand how ongoing cultural stress compounds outcomes following trauma, potentially through weathering effects as has been seen with other biomarkers such as allostatic load [49].

The lack of associations in the current study between PTSD symptoms and circulating 2-AG concentrations measured concurrently is contrary to findings that those with chronic PTSD have lower [19] or higher [46] circulating 2-AG concentrations as well as less robust recruitment of 2-AG in response to psychological or physical stress [38, 50]. It is possible that 2-AG has a more significant role in chronic stress management such that depletion of 2-AG after sustained periods of stress, such as having chronic PTSD, creates a vulnerability to ongoing stress.

Our third objective was to explore the contribution of polymorphisms in the genes for FAAH and CNR1 to the risk of developing PTSD. Genotype at a well-studied SNP in FAAH (rs324420) was significantly related to PTSD severity at follow-up but not hospitalization when the entire cohort was compared. Specifically, individuals homozygous for the minor allele A/A at this location experienced greater PTSD symptom severity 6–8 months following injury compared to individuals with at least one copy of the major allele (C). This relationship was most pronounced in men. Interestingly, women with the A/A genotype had high PCL5 scores at hospitalization (42 compared to 22 in the entire sample) although this enrichment did not reach statistical significance. There was no significant relationship between FAAH genotype and PTSD symptoms at either time point in minorities. As has been shown previously [29], individuals with A/A also had significantly greater circulating AEA concentrations. Thus, these results are consistent with our finding that circulating AEA concentrations are positively associated with PTSD severity. This finding is in contrast to some studies that have found individuals homozygous for the A allele exhibit a reduction in reactivity to stress and ultimate protection from negative emotional responses to stress exposure [29, 51]. However, other studies have found that the interaction of the minor allele and early-life adversity increases the risk for anxiety and depression later in life. Prior life trauma is a risk factor for developing subsequent PTSD [26] and the effect of prior trauma [52] coupled with the minor allele at this locus could create further vulnerability to PTSD. Despite previous work suggesting that SNPs in CB1R genes contribute to risk for pathology in adulthood following childhood trauma exposure, significant findings did not emerge related to the CNR1 polymorphisms at rs1049353 and rs2180619 and PTSD diagnosis. There was a trending association of rs806371 genotype with PTSD symptom severity at follow-up in the entire sample, an association that did not persist in the smaller sex and minority groups. The rs806371 minor variant (G) is associated with reduced expression of the CB1R [53]. Individuals with an A at rs324420 in FAAH and G at rs806371 exhibited a significantly higher PTSD severity at follow-up than those homozygous for the more common allele at each locus. This suggests the hypothesis that high AEA (resulting from low FAAH) and reduced CB1R expression can synergize to enhance the risk for PTSD following traumatic injury.

An important note related to our study findings is the validity of the utilization of circulating endocannabinoid levels to characterize a system that is largely CNS activated. The circulating concentrations of 2-AG in the samples obtained at hospitalization were very high (mean of nearly 700 nM) compared to concentrations from healthy controls (typically 60–200 nM) [20, 54] and in the samples obtained at follow-up (90 nM). Several other studies have shown that acute trauma and injury are associated with high 2-AG concentrations [55, 56]. While AEA concentrations were not abnormally high, they trended to be correlated with the 2-AG concentrations at hospitalization even though the biochemical mechanisms for their synthesis differ [7]. These data are consistent with the notion that exposure to a very strong stressor mobilizes both 2-AG and AEA in concert as the body moves further away from homeostasis. Endocannabinoids in circulation can arise from various sources, including the CNS but also peripheral tissues and blood cells, and therefore can be considered integration of endocannabinoid tone throughout the body [30]. It is difficult to assess the relationship between CNS endocannabinoid signaling and the concentrations of endocannabinoids in circulation in humans. Studies in mice treated with an inhibitor of monoacylglycerol lipase found that circulating concentrations of 2-AG were very significantly associated with brain concentrations of 2-AG [36]. Recent data in humans using PET imaging to assess the amount of FAAH enzyme while simultaneously measuring circulating AEA concentrations found a negative correlation between CNS FAAH levels and AEA concentrations in the circulation [57]. Both of these support the hypothesis that CNS endocannabinoid tone can influence circulating concentrations of 2-AG and AEA. Finally, there is a wealth of data showing significant associations and correlations between circulating endocannabinoid concentrations and measures of CNS function [13, 30], further supporting the contention that circulating endocannabinoid concentrations are a reflection of the tone of CNS endocannabinoid signaling. These findings support the hypothesized role of the endocannabinoid system as a component of the stress response system [13].

Our findings need to be considered within the context of study limitations, particularly as our results are novel and generally contradictory to preclinical work. The blood draws completed with participants were not standardized with respect to time of day or food intake, for example, which have been shown to influence circulating endocannabinoid concentrations in other studies [30]. This suggests that circadian changes, time since the last meal, and time since trauma could influence the endocannabinoid measurements. Second, while this study enrolled 170 subjects, sample size differences for allelic expression in the genetic studies limits findings, and future larger studies are needed to better understand the roles of the SNPs studied in risk for PTSD.

Our data support hypotheses that the endocannabinoid system is stress-responsive and contributes to the effects of stress on emotional processing. Our study findings suggest that endocannabinoid signaling at the time of severe stress sets in motion changes in the brain that potentiate the likelihood of non-remitting PTSD many months later. These data also indicate that the specifics of the relationship between the endocannabinoids and PTSD risk are modulated by sex, genotype, and possibly prior trauma exposure. The clinical course of PTSD is diverse [58] and these results suggest that differences in the endocannabinoid response are a contributor to this diversity. Further work is needed to continue to discover how the endocannabinoid system is related temporally to pathology risk following trauma to refine our understanding of when and how the endocannabinoid system can be targeted for intervention to prevent PTSD development.

Acknowledgements

Author contributions

TAdR-C, KJB, and CJH posed the research question, designed the study, coordinated the data and sample collection, and maintained oversight of the recruitment and follow-up of participants. All authors contributed to the manuscript. CB completed statistical analysis. TAdR-C and CJH interpreted the results and wrote the manuscript. All authors provided intellectual input and review of the manuscript.

Conflict of interest

Dr. Hillard is a member of the Scientific Advisory Boards of Phytecs, Inc and Formulate Biosciences and has equity in Formulate Biosciences.

Footnotes

References