Endocannabinoids and Stress-Related Neurospsychiatric Disorders: A Systematic Review and Meta-Analysis of Basal Concentrations and Response to Acute Psychosocial Stress

Abstract

Background:

Dysregulation of the endocannabinoid (eCB) system is implicated in various stress-related neuropsychiatric disorders (SRDs), including anxiety, depression, and post-traumatic stress disorder (PTSD). In this systematic review and meta-analysis, our objectives were to characterize circulating anandamide (AEA) and 2-arachidonoylglycerol (2-AG) concentrations at rest and in response to acute laboratory-based psychosocial stress in individuals with SRDs and without (controls). Our primary aims were to assess the effects of acute psychosocial stress on eCB concentrations in controls (Aim 1), compare baseline (prestress) eCB concentrations between individuals with SRDs and controls (Aim 2), and explore differential eCB responses to acute psychosocial stress in individuals with SRDs compared with controls (Aim 3).

Methods:

On June 8, 2023, a comprehensive review of the MEDLINE (PubMed) database was conducted to identify original articles meeting inclusion criteria. A total of 1072, 1341, and 400 articles were screened for inclusion in Aims 1, 2, and 3, respectively.

Results:

Aim 1, comprised of seven studies in controls, revealed that most studies reported stress-related increases in AEA (86%, with 43% reporting statistical significance) and 2-AG (83%, though none were statistically significant except for one study in saliva). However, meta-analyses did not support these patterns (p's>0.05). Aim 2, with 20 studies, revealed that most studies reported higher baseline concentrations of both AEA (63%, with 16% reporting statistical significance) and 2-AG (60%, with 10% reporting statistical significance) in individuals with SRDs compared with controls. Meta-analyses confirmed these findings (p's<0.05). Aim 3, which included three studies, had only one study that reported statistically different stress-related changes in 2-AG (but not AEA) between individuals with PTSD (decrease) and controls (increase), which was supported by the meta-analysis (p<0.001). Meta-analyses showed heterogeneity across studies and aims (I²=14–97%).

Conclusion:

Despite substantial heterogeneity in study characteristics, samples, and methodologies, consistent patterns emerged, including elevated baseline AEA and 2-AG in individuals with SRDs compared with controls, as well as smaller stress-related increases in 2-AG in individuals with SRDs compared with controls. To consider eCBs as reliable biomarkers and potential intervention targets for SRDs, standardized research approaches are needed to clarify the complex relationships between eCBs, SRDs, and psychosocial stress.

Introduction

Stress-related neurospsychiatric disorders (SRDs), such as post-traumatic stress disorder (PTSD), anxiety, and depression, encompass neuropsychiatric disorders where stress plays a significant role in their onset, development, and/or persistence.¹ These disorders affect roughly a quarter of the global population, a number that has surged after the onset of the COVID-19 pandemic.²

SRDs impair individuals' ability to manage everyday stress by amplifying psychobiological responses, such as heightened activation of the stress-response system to acute stressors.³ Although current first-line treatments offer relief for some, many experience only modest improvements, and rates of relapse remain high.⁴ Despite the widespread impact and debilitating nature of SRDs, identifying a reliable biomarker for risk assessment and developing effective interventions remains a significant challenge in neuropsychiatry.

Endocannabinoids (eCBs), specifically N-arachidonoylethanolamine (anandamide [AEA]) and 2-arachidonoylglycerol (2-AG), are emerging as promising biomarkers and therapeutic targets for SRDs. The discovery of AEA⁵ and 2-AG⁶ in 1992 and 1995, respectively, has sparked extensive research into their role in SRDs over the past three decades (Fig. 1). The eCB system is a complex, endogenous neuromodulatory system consisting of cannabinoid receptors and ligands distributed throughout the peripheral and central nervous systems. AEA and 2-AG are bioactive lipids with a high affinity for cannabinoid type 1 (CB1) receptors, which are primarily located in the central nervous system.⁷ ECBs play a crucial role in regulating synaptic transmission and neurotransmitter release, including gamma-aminobutyric acid and glutamate.⁸

FIG. 1.

Publications after the discovery of AEA (1992–2022) involving eCBs and SRDs. If year is not shown, there were no articles published in the MEDLINE/PubMed database for that timeframe. Search terms: (endocannabinoid OR eCB OR AEA OR anandamide OR 2-AG OR 2-arachidonylglycerol) NOT (“oxidative stress”) NOT (Review[Publication Type]) AND (PTSD OR “post-raumatic stress disorder” OR “post-traumatic stress disorder” OR depression OR “major depressive” OR anxiety OR phobia OR panic OR GAD OR SAD) NOT (Review[Publication Type]). Search performed June 8, 2023. AEA, anandamide; eCB, endocannabinoid; SRDs, stress-related neurospsychiatric disorders.

Mobilized in response to acute and chronic stress, the eCB system helps maintain homeostasis, and regulates stress- and emotion-related behaviors such as anxiety, stress coping, and fear learning.^9–13 The absence or depletion of eCB signaling, particularly AEA, is linked to increased activation of the hypothalamic–pituitary–adrenal (HPA) axis and stress-related behaviors.^10,14,15

Recent research indicates that AEA and 2-AG have complementary and interconnected roles in regulating stress. For instance, studies in preclinical models demonstrate that exposure to acute and repeated stress reduces AEA levels in specific brain regions like the amygdala, while repeated stress exposure gradually increases the release of 2-AG in the same frontolimbic regions.¹⁶

In our previous review of both clinical and preclinical studies, we found that acute physical stress—that is, acute exercise—increases circulating concentrations of both AEA and 2-AG in controls, individuals with a variety of health conditions (including SRDs), and in animal models.¹⁷ However, it is important to note that psychosocial stress, especially in the context of conditions, such as anxiety disorders (e.g., fear of social situations), PTSD resulting from interpersonal trauma (e.g., violence), and social dysfunction (e.g., difficulty maintaining relationships) in depressive disorders, may have greater relevance for the understanding and treatment of SRDs.^18–20

Psychosocial stress includes a social threat component, such as negative social evaluation or social exclusion, and can be modeled in laboratory settings using the Trier Social Stress Test (TSST) and the Maastricht Acute Stress Test (MAST), among others.^3,21 Collectively, AEA and 2-AG may exhibit diverse responses to psychosocial stress among individuals with and without SRDs, which may highlight the role of eCBs in the development and maintenance of these conditions.

Conceptual models have proposed that blunted circulating eCB concentrations and/or an attenuated response to stress contribute to risk of SRDs.^10,22 This notion is supported by preclinical models that indicate heightened eCB activity can have antidepressant and anxiolytic effects.²³ Likewise, in clinical studies, interventions that target the eCB system, such as Δ9-tetrahydrocannabinol (THC) and inhibitors of fatty acid amide hydrolase and monoacylglycerol lipase (MAGL), show promise in the treatment of SRDs.^24–27 However, findings from laboratory-based research are inconsistent regarding basal concentrations of eCBs and their response to acute psychosocial stress in individuals with SRDs as compared with those without (hereafter “controls”).^28–33 Therefore, it remains unclear whether the eCB system is dysregulated during periods of rest and/or in response to acute psychosocial stress in individuals with SRDs.

To address these inconsistencies, we conducted a systematic review and meta-analysis of the literature focusing on laboratory-based studies of eCB concentrations, specifically AEA and 2-AG, during periods of rest and in response to acute psychosocial stress. This is an emerging area of research, with no prior reviews examining the interplay of circulating eCBs and psychosocial stress in individuals with SRDs or controls.

As such, our primary aims were to assess the effects of acute psychosocial stress on eCB concentrations in controls (Aim 1), compare baseline (prestress) eCB concentrations between individuals with SRDs and controls (Aim 2), and explore differential eCB responses to acute psychosocial stress in individuals with SRDs compared with controls (Aim 3). Within each of these aims, we explored various study characteristics that may contribute to heterogeneity or influence findings, including diagnosis, biological sex, sample size, type of stressor, time of day for sample collection, feeding state, cannabis use, and the source of eCB measurement (e.g., plasma, serum, saliva, hair).¹⁷

Methods

Study identification

Three literature searches were performed in the MEDLINE (PubMed) database on June 8, 2023. This review was preregistered in OSF registries (https://doi.org/10.17605/OSF.IO/72Z9H) and was conducted in accordance with the 2020 PRISMA guidelines for systematic reviews.³⁴

Search terms

• Aim 1: (endocannabinoid OR eCB OR AEA OR anandamide OR 2-AG OR 2-arachidonylglycerol) AND (stress OR stressor OR Trier OR TSST OR MAST OR “Maastricht Acute Stress Test”) NOT (“oxidative stress”) NOT (Review[Publication Type])

• Aim 2: (endocannabinoid OR eCB OR AEA OR anandamide OR 2-AG OR 2-arachidonylglycerol) NOT (“oxidative stress”) NOT (Review[Publication Type]) AND (PTSD OR “posttraumatic stress disorder” OR “post-traumatic stress disorder” OR depression OR “major depressive” OR anxiety OR phobia OR panic OR GAD OR SAD) NOT (Review[Publication Type])

• Aim 3: (endocannabinoid OR eCB OR AEA OR anandamide OR 2-AG OR 2-arachidonylglycerol) NOT (“oxidative stress”) NOT (Review[Publication Type]) AND (PTSD OR “posttraumatic stress disorder” OR “post-traumatic stress disorder” OR depression OR “major depressive” OR anxiety OR phobia OR panic OR GAD OR SAD) AND (stress OR stressor OR Trier OR TSST OR MAST OR “Maastricht Acute Stress Test”) NOT (Review[Publication Type])

Selection process

Screening process

Studies were screened by two independent reviewers (L.C.G. and J.M.E.). The study search, screening, and selection process for each aim are shown in Figure 2. First, titles and abstracts were screened to exclude articles that did not fit the scope of this systematic review, such as nonhuman animal studies. Subsequently, articles that passed the initial screening underwent a full-text review. Articles that met the inclusion criteria for each aim were then characterized and summarized.

FIG. 2.

Flowchart for study identification and screening for Aims 1, 2, and 3. (A) Aim 1: Evaluate eCB response to acute psychosocial stress in controls; (B) Aim 2: Compare eCB concentrations at rest in controls versus in individuals SRDs; (C) Aim 3: Examine eCB response to psychological stress in individuals with SRDs compared with controls.

Study eligibility

This review included full-length original research articles published in the English language that reported on laboratory-based studies examining AEA and/or 2-AG concentrations in human subjects at rest and/or in response to acute psychosocial stress. Studies that did not utilize an acute laboratory-based psychosocial stressor were excluded for Aims 1 and 3. Studies were also excluded if the sample did not include a control group (Aims 1, 2, and 3) or an SRD group (Aims 2 and 3).

To isolate SRDs from other co-occurring disorders, studies were excluded from all aims if the SRD or control group included individuals with psychotic disorders, substance use disorders, chronic medical conditions (e.g., obesity), or if the study examined SRDs through continuous symptoms (e.g., depressive symptoms) rather than diagnostic categories. Nonclinical subgroups not otherwise excluded and different from the control group (i.e., trauma exposed without PTSD) were classified as controls and examined later in exploratory analyses. Full texts were reviewed by both L.C.G. and J.M.E., and uncertainty was discussed by L.C.G., J.M.E., and H.A.M.

Study classification

Articles were classified by (1) sample characteristics (e.g., sex, age, diagnosis); (2) sample size; (3) eCB source (e.g., plasma, serum, hair, saliva); and (4) psychosocial stressor (e.g., TSST, MAST; Aims 1 and 3 only). In addition, we noted whether studies controlled for or measured time-of-day, feeding state, or cannabis use in participants. Studies reported mixed information on these data (Supplementary Table S1). Secondary results classified SRD diagnoses separately to examine patterns across PTSD and trauma-related disorders, anxiety disorders, and depressive disorders.

Data collection process

Extracted data included sample size, group-level mean, standard deviation, and/or standard error of eCB concentrations at baseline (Aim 2), and at the pre- and post-timepoints in proximity to the psychosocial stressor for controls and SRDs (Aims 1 and 3). Authors were contacted to supplement eCB values if not reported in the published manuscript, supplementary text, figures, or tables. If exact values were not available or supplemented, PlotDigitizer (plotdigitizer.com) was used to estimate eCB values from published or Supplementary Figures.³⁵ These values were crosschecked by L.C.G., J.M.E., and H.A.M.

Extracted data also included reported statistical significance, that is, change in eCB concentrations pre- versus poststress in controls (Aim 1), group difference in basal eCBs in individuals with SRDs versus controls (Aim 2), and the group (SRD, control)×time (pre-, poststress) interaction (Aim 3). Notably, not all studies assessed or disclosed findings reporting on group differences (i.e., SRD vs. control).

Meta-analysis

A total of 7, 18, and 3 studies (AEA) and 6, 18, and 3 studies (2-AG) possessed sufficient data for inclusion in the meta-analyses for Aims 1, 2, and 3, respectively. RevMan Web (Cochrane, v.1.22.0) was utilized to conduct the meta-analysis using a fixed-effects inverse variance model on continuous between-group mean differences.³⁶ A fixed-effects model was selected given the relatively limited number of studies and given that studies were mixed in sample sizes, with the larger studies weighted more heavily than smaller studies in fixed-effects models.³⁷ However, in recognition of the heterogeneity across aims (I²=14–97%), we also report results of random effects for completeness (see Supplementary Data).³⁷ For Aim 2, multiple control groups (e.g., healthy control and trauma-exposed control) were considered separate entries.

For Aim 3, we first computed within-group change scores for change in AEA or 2-AG response from pre- to poststress, then entered those values into the model to test for mean differences between SRD and control groups. Overall effects and significance values are reported in Z and p-values, respectively. Heterogeneity was investigated using forest plots and the I² statistic. Here, I² values between 0% and 39% were classified as low heterogeneity, 40–59% as moderate, and ≥60% as high. Secondary analyses examined subgroups, for example, by diagnosis or eCB source.

Results

Aim 1: eCB response to acute psychosocial stress in controls

Study characterization

After screening 1072 search results, 7 studies reporting on 303 controls met eligibility criteria for Aim 1 (Table 1). Each study included adult samples (∼79% female) with a mean age of ∼25 years. Four studies utilized the TSST,^28–30,38 and three used the MAST.^24,39,40 All studies measured eCBs in blood (five plasma,^{24,28,29,39,40} two serum^30,38), and one study also measured eCBs in saliva.³¹ Control group sample sizes ranged from 10 to 77 participants.

Table 1.

Aim 1: endocannabinoid response to acute psychosocial stress in controls

First author	Year	DOI	Control, N	eCB source	Stressor	ΔAEA poststressor	Δ2-AG poststressor
Ney	2021	10.1016/j.biopsycho.2021.108022	77	Plasma and saliva	MAST	↓ ↓	↑ ↑a
Petrowski	2020	10.1016/j.psyneuen.2020.104905	26	Plasma	TSST	↑	↑
Mayo	2020	10.1016/j.biopsych.2019.07.034	29	Plasma	MAST	↑a	↓
Mayo	2020	10.1038/s41380-018-0215-1	75	Plasma	MAST	↑	—
Crombie	2019	10.1016/j.biopsycho.2019.04.002	10	Plasma	TSST	↑a	↑
Dlugos	2012	10.1038/npp.2012.100	71	Serum	TSST	↑a	↑
Hill	2009	10.1016/j.psyneuen.2009.03.013	15	Serum	TSST	↑	↑

N's reflect total number of participants per relevant subgroup reported in the article, N of subjects included in eCB analysis may differ. Results based on group means.

^a Indicates significance.

— Indicates not reported.

2-AG, 2-arachidonoylglycerol; AEA, anandamide; DOI, digital object identifier; eCB, endocannabinoid; MAST, Maastricht Acute Stress Test; TSST, Trier Social Stress Test.

AEA results

Six of the seven studies reported increased AEA concentrations in plasma or serum following acute psychosocial stress relative to baseline,^{24,28–30,38,40} and three of these studies reported statistical significance.^24,29,38 In contrast, the remaining study reported a decrease in AEA in plasma.³⁹ One study also reported a decrease in salivary AEA (as well as plasma) following acute psychosocial stress³⁹; however, this was not statistically significant (Table 1). Notably, all four studies utilizing the TSST reported an increase in AEA^28–30,38 (two statistically significant^29,38). Of these, two measured eCBs in plasma^28,29 (one statistically significant²⁹) and two examined eCBs in serum^30,38 (one statistically significant³⁸).

AEA meta-analysis results

There was no significant effect of acute psychosocial stress on AEA concentrations in controls (p=0.37; Fig. 3A) and a moderate level of heterogeneity across studies (p=0.008, I²=58%). A subanalysis of plasma-only studies (k=7) also did not yield significant results (p=0.27, Supplementary Fig. S1). Subanalyses examining only studies using the MAST (k=3) and TSST (k=4) revealed no significant changes in AEA (Supplementary Figs. S25 and S26). Using a random-effects model, this effect was still not significant (Supplementary Figs. S11 and S13).

FIG. 3.

Meta-analysis results: Aim 1. Forest plot indicates no significant difference in AEA (A) or 2-AG (B) concentrations following acute psychosocial stress in individuals without SRDs (controls). Square data markers indicate mean difference in eCB concentrations from baseline to poststress, with sizes reflecting the weight of the studies with fixed-effects models. Of note, some values are outside the range, which are indicated by arrows, and units differ across studies (see Supplementary Data). Horizontal lines indicate the 95% confidence intervals. The diamond marker represents the overall effect and 95% confidence interval, and the vertical line indicates the line of no effect (mean difference=0). 2-AG, 2-arachidonoylglycerol.

2-AG results

One study did not report 2-AG concentrations.⁴⁰ Of the six studies that reported 2-AG results, five reported increased 2-AG concentrations in plasma or serum following acute psychosocial stress (i.e., four TSST, one MAST).^{28–30,38,39} Only one study reported decreased 2-AG concentrations in plasma following the MAST.²⁴ Despite these apparent trends, none of these studies (in plasma or serum) reported a statistically significant change in 2-AG; however, one study that also measured saliva reported a significant increase in 2-AG immediately following stress from the MAST.³⁹ Interestingly, all four studies using the TSST reported (nonsignificant) increases in 2-AG.^28–30,38

2-AG meta-analysis results

There was no significant effect of acute psychosocial stress on 2-AG levels (p=0.38, Fig. 3B) and low heterogeneity across studies (p=0.32, I²=14%). Results were similar in a subanalysis of plasma-only studies (k=4, p=0.4, I²=0%, Supplementary Fig. S2). Subanalyses examining only studies using the MAST (k=2) and TSST (k=4) revealed no significant changes in 2-AG (Supplementary Figs. S27 and S28). Using a random-effects model, this effect remained nonsignificant (Supplementary Figs. S12 and 14).

Aim 2: basal eCB concentrations in individuals with SRDs versus controls

Study characterization

After screening 1341 search results, 20 studies reporting on 718 controls and 616 individuals with SRDs met eligibility criteria (Table 2). Each study included adult participants (mean age ∼33 years, ∼63% female). All studies but one measured eCBs in plasma (k=15) or serum (k=4); only one study examined eCBs in hair.⁴¹ The most common SRD group was PTSD^{29,31,33,42–48} (k = 10), followed by depressive disorders^{30,32,41,49–52} (seven studies; five exclusively major depressive disorder [MDD]^{30,41,49–51}), anxiety disorders (two studies; one panic disorder,²⁸ one social anxiety disorder⁵³), and one study with an SRD group comprised of multiple disorders⁵⁴ (i.e., anxiety, depression, and stress/trauma subgroups). Included studies had samples that ranged in size from 20 to 220 participants.

Table 2.

Aim 2: basal endocannabinoid concentrations in stress-related disorders versus controls

First author	Year	DOI	SRD, N	Control, N	eCB source	AEA baseline	2-AG baseline
Botsford	2023	10.1016/j.janxdis.2022.102656	42	56	Plasma	PTSD<CON	PTSD<CON
Behnke	2023	10.1080/15622975.2022.2070666	20	24	Plasma	MDD>CONa	MDD>CONb
Leen	2022	10.1177/24705470221107290	54	26	Plasma	PTSD>CON	PTSD>CON
Dos Santos	2022	10.1002/hup.2834	17	20	Plasma	SAD<CON	SAD>CON
Ney	2021	10.1002/da.23170	43	177	Plasma	PTSD<CON	PTSD>CON
Crombie	2021	10.1016/j.mhpa.2020.100366	14	26	Plasma	PTSD>CON	PTSD<CON
Bersani	2021	10.1002/hup.2779	12	12	Plasma	MDD<CON	MDD>CON
Behnke	2021	10.1016/j.cpnec.2021.100068	21	27	Hair	—	MDD>CONb
Petrowski	2020	10.1016/j.psyneuen.2020.104905	26	26	Plasma	PD>CONa	PD>CONa
Romero-Sanchiz	2019	10.1016/j.neuropharm.2019.02.026	69	47	Plasma	MDD>CON	MDD>CON
Crombie	2019	10.1016/j.biopsycho.2019.04.002	10	10	Plasma	PTSD>CON	PTSD>CON
Crombie	2018	10.1002/jts.22253	12	12	Plasma	PTSD>CON	PTSD>CON
Coccaro	2018	10.1016/j.psyneuen.2018.03.009	115	60	Plasma	DEPR<CON ANX<CON TRAUMA<CON	DEPR<CON ANX>CON TRAUMA<CON
Schaefer	2014	10.1007/s00406-013-0470-8	21	30	Serum	PTSD>CON	PTSD<CON
Neumeister	2013	10.1038/mp.2013.61	25	35	Plasma	PTSD<CONa	PTSD>CON
Hill	2013	10.1016/j.psyneuen.2013.08.004	24	22	Plasma	PTSD>CON	PTSD<CONa
Hauer	2013	10.1371/journal.pone.0062741	10	38	Plasma	PTSD>CONa	PTSD>CONa
Ho	2012	10.1186/1476-511X-11-32	28	27	Serum	DEPR>CON	DEPR<CON
Hill	2009	10.1016/j.psyneuen.2009.03.013	15	15	Serum	MDD<CONa	MDD<CONa
Hill	2008	10.1055/s-2007-993211	28	28	Serum	MDD>CON MiD>CONa	MDD<CONa MiD>CON

N's reflect total number of participants per relevant subgroup reported in the article, N of subjects included in eCB analysis may differ. Results are based on group means.

^a Indicates significance.

^b Based on group median values.

— Indicates not reported.

ANX, anxiety disorder; CON, control; DEPR, depressive disorder; MDD, major depressive disorder; MiD, minor depression; PD, panic disorder; PTSD, post-traumatic stress disorder; SAD, social anxiety disorder; SRD, stress-related disorder; TRAUMA, trauma/stress-related disorder.

AEA results

One study analyzing hair did not report AEA results.⁴¹ Of the 19 included studies, 12 studies (3 statistically significant^28,48,49) reported higher basal AEA concentrations in plasma or serum in individuals with SRDs relative to controls (Table 2). One study found mixed significance.³² Seven studies (two statistically significant^30,47) reported lower basal AEA in individuals with SRDs versus controls. Three of the four studies that measured eCBs in serum reported higher basal AEA concentrations in individuals with SRDs versus controls, although none of these reported statistical significance.^32,46,52

AEA meta-analysis results

Across studies, baseline AEA concentrations were significantly higher in individuals with SRDs compared with controls (p=0.04, Fig. 4A). There was a high level of heterogeneity across studies (p<0.001, I²=66%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S15).

FIG. 4.

Meta-analysis results: Aim 2. Forest plot indicates significantly higher baseline AEA (A) and 2-AG (B) concentrations in individuals with SRDs compared with controls. Square data markers indicate mean group differences, with sizes reflecting the weight of the studies with fixed-effects models. Of note, some values are outside the range, which are indicated by arrows, and units differ across studies (see Supplementary Data). Horizontal lines indicate the 95% confidence intervals. The diamond marker represents the overall effect and 95% confidence interval, and the vertical line indicates the line of no effect (mean difference=0).

AEA results: PTSD and trauma-related disorders

Eleven studies compared AEA concentrations in PTSD (10 studies) and trauma-related disorders (one study) versus controls, with 10 measuring AEA in plasma and 1 in serum.⁴⁶ Of the 11 studies examining PTSD and trauma-related disorders, 7 reported higher basal AEA in the PTSD group relative to controls,^{29,33,43–46,48} however only 1 study reported statistical significance.⁴⁸ Three studies in PTSD and one in stress and trauma disorders reported lower basal AEA relative to controls,^31,42,47,54 with one study reporting significance.⁴⁷

AEA meta-analysis results: PTSD and trauma-related disorders

A subanalysis of studies of PTSD and trauma-related disorders revealed significantly higher baseline AEA concentrations in individuals with PTSD compared with controls (p=0.005, Supplementary Fig. S3). There was high heterogeneity across studies (I²=83%). Using a random-effects model, this effect was nonsignificant (Supplementary Fig. S17).

Given that four studies included both a healthy control and trauma-exposed control group, we performed an exploratory follow-up analysis to test for differences in baseline AEA concentrations between healthy control and trauma-exposed control groups. There was no difference in AEA concentrations between groups (p=0.79, Supplementary Fig. S9) and low heterogeneity across studies (I²=0%).

AEA results: anxiety disorders

Three studies compared AEA concentrations in anxiety disorders versus controls.^28,53,54 One study reported significantly higher plasma AEA concentrations in panic disorder as compared with controls.²⁸ The remaining two studies reported lower AEA concentrations in plasma in individuals with social anxiety disorder⁵³ and anxiety disorders⁵⁴ in comparison with controls; however, neither study reported significance.

AEA meta-analysis results: anxiety disorders

A subanalysis of anxiety disorder studies demonstrated significantly higher baseline AEA concentrations in individuals with anxiety disorders compared with controls (p=0.004, Supplementary Fig. S4). There was low heterogeneity across studies (I²=0%). Using a random-effects model, this effect remained statistically significant (p=0.004, Supplementary Fig. S18).

AEA results: depressive disorders

Seven studies compared AEA concentrations in depressive disorders versus controls. Of those, four studies (one significant)⁴⁹ reported higher AEA concentrations in the depressed group versus control group.^32,49,51,52 One of these studies found mixed significance, wherein the minor depression subgroup demonstrated significantly higher AEA concentrations compared with controls, and the MDD subgroup displayed similar but nonsignificant elevations.³² Three studies reported the opposite pattern of lower AEA in the depressed group compared with controls,^30,50,54 with only one study reporting statistical significance in a MDD group.³⁰

AEA meta-analysis results: depressive disorders

A subanalysis of depressive disorder studies revealed significantly higher baseline AEA concentrations in individuals with depressive disorders compared with controls (p=0.007, Supplementary Fig. S5), with moderate heterogeneity across studies (I²=49%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S19).

2-AG results

Of the 20 included studies, 12 studies reported higher basal 2-AG in individuals with SRDs relative to controls. Of those 12 studies, 11 examined 2-AG in plasma (2 statistically significant^28,48) and 1 in hair.⁴¹ Six studies reported lower basal 2-AG (three statistically significant)^30,32,33 in individuals with SRDs relative to controls. Two studies reported mixed results.^32,54 One study found that 2-AG concentrations were higher in the anxiety disorder subgroup compared with controls, but lower in depressive and stress/trauma disorder subgroups relative to controls; however, neither of these group comparisons reported statistical significance.⁵⁴

In another study, 2-AG was significantly lower in the major depression subgroup compared with controls, and nonsignificantly higher in the minor depression subgroup compared with controls.³² Three of the four studies that measured 2-AG in serum reported lower basal concentrations in the SRD group (PTSD, depressive disorders, MDD),^30,46,52 and one found lower concentrations in the MDD subgroup only³² (two of these three studies examining depressive disorders found significant differences).^30,32

2-AG meta-analysis results

The meta-analysis revealed significantly higher baseline 2-AG concentrations in individuals with SRDs compared with controls (p<0.001, Fig. 4B). There was a high level of heterogeneity across studies (p<0.001, I²=64%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S16).

2-AG results: PTSD and trauma-related disorders

Overall, 11 studies compared basal 2-AG concentrations in PTSD (10 studies) and trauma-related disorders (one study) versus controls. Findings were mixed. Six studies (one significant⁴⁸) reported higher basal 2-AG in PTSD compared with controls.^{29,31,43,45,47,48} Five studies (one significant³³) reported the opposite pattern of lower basal 2-AG concentrations in PTSD^33,42,44,46 and trauma-related disorders broadly⁵⁴ compared with controls.

2-AG meta-analysis results: PTSD and trauma-related disorders

A subanalysis of PTSD and trauma-related disorders revealed no significant difference in baseline 2-AG concentrations in individuals with PTSD compared with controls (p=0.95, Supplementary Fig. S6), with moderate heterogeneity across studies (I²=53%). Given that four studies included both a healthy control and trauma-exposed control group, we performed an exploratory follow-up analysis to test for differences in baseline 2-AG concentrations in the healthy control group compared with trauma-exposed control group.

There was no difference in 2-AG concentrations between groups (p=0.3, Supplementary Fig. S10) and low heterogeneity across studies (I²=0%). Using a random-effects model, this effect was not significant (Supplementary Fig. S20).

2-AG results: anxiety disorders

Three studies compared 2-AG concentrations in anxiety disorders (social anxiety disorder,⁵³ panic disorder,²⁸ and anxiety disorders broadly⁵⁴) and controls. All three studies measured 2-AG in plasma and reported higher basal 2-AG in the anxiety disorder group relative to controls; however, this difference was only significant in one study comparing concentrations in panic disorder and controls.²⁸

2-AG meta-analysis results: anxiety disorders

A subanalysis of anxiety disorder studies revealed significantly higher 2-AG concentrations in individuals with anxiety disorders compared with controls (p<0.001, Supplementary Fig. S7), with high heterogeneity across studies (I²=72%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S21).

2-AG results: depressive disorders

Eight studies compared 2-AG concentrations in depressive disorders versus controls. Four studies (three in plasma^49–51; one in hair⁴¹; none significant) reported higher concentrations of 2-AG in the depressed group relative to controls. Three studies (one significant³⁰) reported lower 2-AG in the depressed group compared with controls.^30,52,54

One study had mixed findings for 2-AG in depressive disorders, such that the major depression subgroup showed significantly lower AEA concentrations, whereas the minor depression subgroup had nonsignificantly higher concentrations.³² Each of these three studies reporting lower 2-AG in depressive disorders analyzed serum,^30,32,52 and two of these studies reported significance.^30,32

2-AG meta-analysis results: depressive disorders

A subanalysis of depressive disorder studies demonstrated significantly higher 2-AG concentrations in individuals with depressive disorders compared with controls (p=0.003, Supplementary Fig. S8), with high heterogeneity across studies (I²=71%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S22).

Aim 3: differential eCB response to acute psychosocial stress in individuals with SRDs versus controls

Study characterization

After screening 400 search results, 3 studies met eligibility criteria (Table 3). Each included adult participants (mean age ∼27 years, ∼62% female) and measured eCBs exclusively in blood samples (two plasma,^28,29 one serum).³⁰ One study included individuals with PTSD,²⁹ one panic disorder,²⁸ and one MDD.³⁰ All three studies used the TSST.^28–30 Sample sizes ranged from 20 to 62 participants, with a total of 51 controls and 61 individuals with SRDs across all 3 studies.

Table 3.

Aim 3: endocannabinoid response to acute psychosocial stress in stress-related disorders versus controls

First author	Year	DOI	SRD	SRD, N	Control, N	eCB source	Stressor	AEA, time×group interaction?	2-AG, time×group interaction?
Petrowski	2020	10.1016/j.psyneuen.2020.104905	PD	26	26	Plasma	TSST	No	Yes
Crombie	2019	10.1016/j.biopsycho.2019.04.002	PTSD	10	10	Plasma	TSST	—	Yesa
Hill	2009	10.1016/j.psyneuen.2009.03.013	MDD	15	15	Serum	TSST	No	No

N's reflect total number of participants per relevant subgroup reported in the article, N of subjects included in eCB analysis may differ. Results are based on group means.

^a Indicates significance.

— Indicates not reported.

The study in individuals with MDD was entirely female (total n=30),³⁰ the study in PTSD was largely female (total n=20, control=70% female, PTSD=80% female),²⁹ and the study in individuals with panic disorder was about half female (total n=52; 46% female in both control and panic disorder groups).²⁸

AEA results

No significant group×time interactions were reported for AEA (Table 3). In all three studies, SRDs had AEA responses in the same direction as the control group following acute psychosocial stress; that is, SRDs (PTSD, panic disorder, MDD) and controls all exhibited a stress-related increase in AEA concentrations.

AEA meta-analysis results

The meta-analysis did not demonstrate a significant difference in stress-related changes in AEA concentrations between individuals with SRDs and controls (p=0.35, Fig. 5A), with low heterogeneity across studies (I²=36%). Using a random-effects model, this effect remained nonsignificant (Supplementary Fig. S23).

FIG. 5.

Meta-analysis results: Aim 3. (A) Forest plot indicates no significant difference in AEA response to acute psychosocial stress in individuals with SRDs compared with controls. (B) Forest plot indicating significantly lower 2-AG response to acute psychosocial stress in individuals with SRDs compared with controls. Square data markers indicate mean group differences in eCB concentrations from baseline to poststress, with sizes reflecting the weight of the studies with fixed-effects models. Of note, some values are outside the range, which are indicated by arrows, and units differ across studies (see Supplementary Data). Horizontal lines indicate the 95% confidence intervals. The diamond marker represents the overall effect and 95% confidence interval, and the vertical line indicates the line of no effect (mean difference=0).

AEA results: PTSD and trauma-related disorders

One study examined differential effects of stress following the TSST on plasma AEA concentrations in PTSD versus controls. The authors found a significant main effect of time across both PTSD and controls, such that AEA concentrations significantly increased across both groups. The authors did not report a group×time interaction in AEA results.²⁹

AEA results: anxiety disorders

One study examined the effects of the TSST on plasma AEA concentrations in panic disorder versus controls. No significant group×time effects in AEA response were observed between the panic disorder group and the control group following acute psychosocial stress. However, the panic disorder group consistently and significantly displayed higher plasma AEA concentrations pre- and poststress, and both groups showed increases in AEA concentrations from pre- to poststress.²⁸

AEA results: depressive disorders

One study in MDD did not find a group×time interaction following acute psychosocial stress, as both groups demonstrated increases in serum AEA concentrations following the TSST; moreover, although numerically higher in MDD group, the magnitude of these responses did not differ between the MDD and control groups.³⁰

2-AG results

Of the three studies examining differential effects of stress on 2-AG in individuals with SRDs versus controls, one reported a significant group×time interaction in 2-AG response to TSST (i.e., increase in controls versus decrease in PTSD).²⁹ A second study displayed a similar group×time interaction (i.e., increase in controls versus decrease in individuals with panic disorder), but this was not statistically significant.²⁸ The third study found that the 2-AG response to acute psychosocial stress in individuals with SRDs resembled controls (i.e., increase).³⁰

2-AG meta-analysis results

There was a significant group difference in stress-related change in 2-AG concentrations between individuals with SRDs and controls, wherein individuals with SRDs displayed a smaller stress-related increase in 2-AG concentrations compared with controls (p<0.001, Fig. 5B) with high heterogeneity across studies (p<0.001, I²=97%). Using a random-effects model, this effect was no longer significant (Supplementary Fig. S24).

2-AG results: PTSD and trauma-related disorders

One study in PTSD reported a significant group (SRD, control)×time (pre-, poststress) interaction in 2-AG concentrations, such that controls exhibited a stress-related increase after the TSST whereas the PTSD group demonstrated a decrease.²⁹

2-AG results: anxiety disorders

The only study in panic disorder reported a stress-related increase in plasma 2-AG in controls and a stress-related decrease in panic disorder following the TSST; however, this group×time interaction was not statistically significant.²⁸

2-AG results: depressive disorders

No significant time×group differences were reported in one study examining serum 2-AG concentrations in MDD and controls following TSST. Significant stress-related increases in 2-AG were found in both the MDD and control groups immediately following the stressor.³⁰

Discussion

This systematic review and meta-analysis is the first, to our knowledge, to characterize circulating AEA and 2-AG concentrations at rest and in response to acute psychosocial stress in individuals with SRDs and without (controls). Several findings emerged. First, while most studies reported stress-related increases in AEA (86%) and 2-AG (83%) in controls, the meta-analysis did not support these patterns. Second, the majority of studies observed higher basal (i.e., prestress) concentrations of both AEA (63%) and 2-AG (60%) in individuals with SRDs compared with controls, a pattern supported by the meta-analysis.

Third, only three studies examined differential effects of psychosocial stress on eCBs in individuals with SRDs compared with controls, with only one study reporting a significant group×time interaction indicating distinct stress-related changes in 2-AG between individuals with PTSD (decrease) and controls (increase), supported by the meta-analysis. Despite these patterns, substantial heterogeneity existed across studies, which may be related to study characteristics, sample sizes and compositions, and methodologies.

Aim 1: eCB response to acute psychosocial stress in controls

While the meta-analysis did not yield statistical confirmation of these effects, it is notable that most studies observed increased AEA and 2-AG concentrations in controls following acute psychosocial stress. This aligns with our previous findings on elevations in AEA and 2-AG concentrations in both human and nonhuman subjects after acute exercise—a form of physical stress.¹⁷ These observations suggest the eCB system's role as a stress buffer, mobilized by diverse stress types to restore physiological homeostasis and dampen stress responses tied to HPA axis overactivation.^10,14,55,56

Preclinical investigations support acute stress-induced 2-AG increases in brain regions such as the prefrontal cortex, hippocampus, and hypothalamus (for an in-depth review, see Morena et al.).⁵⁷ However, AEA concentrations typically decrease in regions such as the amygdala and hippocampus postacute stress.^58–61 Discrepancies between preclinical and clinical studies of AEA may stem from factors such as eCB measurement sources (e.g., blood samples vs. brain tissue),⁶² stressor types (e.g., TSST vs. immobilization), or behavioral paradigms used.

In fact, preclinical research suggests that stress-related increases in AEA occur in frontolimbic regions after a footshock but decrease after immobilization.^57,63–65 Moreover, the element of social threat in psychosocial stress may be more difficult to emulate and implement in preclinical models. Furthermore, chronic homotypic stress may produce differential effects on eCBs versus heterotypic stressors.⁶⁶ This review focused on acute psychosocial stress laboratory studies, contrasting with animal model stressors (i.e., footshock) that may induce pain.

Future research should investigate the relationship between peripheral and central eCB measurements, and explore the effects of various psychosocial stress paradigms on the eCB system.

Aim 2: basal eCB concentrations in individuals with SRDs versus controls

In contrast to previous conceptual models,^10,22 we did not find evidence of blunted basal eCB concentrations in individuals with SRDs when compared with controls. Rather, most studies examined here reported higher concentrations of AEA and 2-AG during resting periods or before the induction of acute stress in individuals with SRDs compared with controls, which was confirmed by the meta-analysis.

This aligns with preclinical studies that generally demonstrate heightened 2-AG concentrations in certain brain regions, such as the hippocampus, after repeated stress exposures (e.g., restraint paradigm).⁵⁹ However, the effects of repeated stress on AEA tend to exhibit the opposite pattern, with reductions in AEA concentrations observed in frontolimbic regions.⁶⁷ Higher peripheral concentrations of AEA and 2-AG in individuals with SRDs may be indicative of the impact of chronic or repeated stress on the eCB system, which may lead to abnormally elevated eCB concentrations during periods of rest.

The influence of chronic stress on the eCB system is likely intertwined with its intricate relationship with the HPA axis. For instance, prolonged exposure to corticosterone leads to a downregulation of MAGL expression in the amygdala, which may account for chronic stress-induced increases in 2-AG concentrations observed at rest.⁶⁶

In contrast, the eCB system serves to inhibit the activation of the HPA axis and plays a pivotal role in terminating the physiological stress response.^14,56 It is plausible that the eCB system undergoes adaptations in response to the habituation of the HPA axis to chronic stress, potentially resulting in the overproduction of peripheral AEA and 2-AG.^10,57 This could explain the positive correlation between circulating concentrations of AEA and cortisol observed in individuals with PTSD and control subjects.^33,57 Moreover, increased AEA concentrations have been linked to greater resilience in adults exposed to trauma (without PTSD), suggesting that the eCB system may play an adaptive role after trauma exposure.⁶⁸ However, our exploratory meta-analysis found no significant baseline eCB differences between trauma-exposed controls and healthy controls, suggesting that observed effects in individuals with SRDs reflect the expression of pathology rather than trauma exposure itself.

Moreover, a subanalysis of PTSD studies showed no significant baseline 2-AG differences, suggesting variations in 2-AG in anxiety and depressive disorders rather than PTSD. Further research is warranted to elucidate these findings.

Aim 3: differential eCB response to acute psychosocial stress in individuals with SRDs versus controls

The aberrantly elevated basal eCB tone observed in individuals with SRDs may contribute to the onset and/or persistence of these conditions by impeding the eCB system's ability to adequately respond to subsequent acute stressors. In support of this hypothesis, the study conducted by Crombie et al²⁹ was the sole study to report significant differential effects of acute stress on 2-AG concentrations in individuals with PTSD, primarily females, as compared with control subjects.

In this study,²⁹ the control group demonstrated an increase in 2-AG concentrations following the TSST, which aligns with the overall trend observed in most reviewed studies regarding the effects of acute stress on 2-AG in control subjects. Conversely, the PTSD group exhibited a decrease in 2-AG concentrations, which may indicate the possibility of a ceiling effect for 2-AG in individuals with PTSD, or an inability to appropriately respond to acute stress.

Another study by Petrowski et al. reported a similar pattern in 2-AG, with a stress-related increase in 2-AG concentrations observed in controls and a decrease in individuals with panic disorder following the TSST, although the interaction between group and time did not reach statistical significance.²⁸ However, the meta-analysis supported differential 2-AG response to acute stress in individuals with SRDs compared with controls, but no group differences in AEA. These preliminary findings should be further explored in future research.

Heterogeneity across studies and directions for future research

This systematic review and meta-analysis identified several patterns, with studies with larger sample sizes weighted more heavily in the fixed-effects meta-analysis than those with smaller sample sizes. However, there was a high level of heterogeneity across studies and many studies lacked statistical significance. Variability may stem from measurement error, diverse study characteristics, samples, and research methodologies.³

A limitation of the current literature is the that most studies (71%) recruited small sample sizes of each group (n<30), which may have contributed to low statistical power. Moreover, inconsistent inclusion/exclusion criteria and variable collection or reporting of factors influencing eCBs (e.g., substance use, medication, timing of eCB sample, time-of-day, fed state, menstrual cycle) may contribute to this heterogeneity (Supplementary Table S1).^69,70

While studies involving individuals with substance use disorders, including cannabis use disorder, were frequently excluded, some studies did not assess or report substance use, potentially affecting results. THC, the primary psychoactive compound in cannabis, binds with a high affinity to the CB1 receptor. Chronic cannabis use is associated with altered circulating eCB concentrations^71,72 and neuroendocrine stress responses, such as HPA-axis activity.^73,74 Future research should address these confounders and consider expert consensus on guidelines for collecting eCB data, akin to cortisol studies.⁷⁵

All studies except one⁴⁹ utilized blood samples to quantify eCB concentrations, with plasma being the most common (73% overall). While subanalyses supported no differences between plasma and serum studies, questions arise regarding optimal methodology for detecting eCB response to acute psychosocial stress and distinctions between SRDs and controls. These findings emphasize the need to consider the benefits (e.g., long-term retrospective measurement of stress using hair samples)⁷⁶ and drawbacks (e.g., sex differences in saliva)⁷⁷ of each sampling method when measuring eCBs. Further investigation into the strengths and weaknesses of each sampling method is needed to inform future research.

The current literature lacks studies in pediatric or adolescent populations, despite the early age of onset of many SRDs.⁷⁸ This gap is critical given the abundance of CB1 receptors in frontolimbic regions crucial for stress and emotion regulation,⁷⁹ which develop throughout childhood and adolescence.⁸⁰ Moreover, early-life stress contributes to alterations in frontolimbic circuitry and increases risk of SRDs.^81–83 Thus, studies in pediatric and adolescent populations could offer insights into eCB system's relationship with SRDs.

Furthermore, while preclinical evidence suggests stress-related sex differences in eCBs, support in human studies is limited.^39,84,85 Only one study included examined sex differences, and found significant elevations in salivary AEA compared to females.³⁹ Future research should consider the role of menstrual cycle, oral contraceptive use, and potential sex differences when exploring stress and the eCB system. In addition, the laboratory visit may induce anticipatory anxiety and mobilize the HPA axis, eliciting an eCB response.^86–88

Our review primarily focused on laboratory studies of acute psychosocial stress, rather than repeated stressors or naturalistic chronic studies.⁸⁹ Previous studies indicate that AEA and 2-AG may serve distinct functions as “tonic” (sustained/continuous) and “phasic” (short-term) regulators of stress response, respectively.⁵⁷ Thus, future research utilizing repeated stressor models could shed light on the involvement of eCBs in chronic stress.

The over-representation of PTSD in SRDs studies emphasizes the need for more research into anxiety and depressive disorders. Only three studies examined the differential effects of acute stress on eCBs in individuals with SRDs compared with controls. Moreover, there is some evidence to suggest that eCB concentrations may correlate with SRD symptoms rather than categorical diagnoses³³ as examined here, which warrants further exploration.

Conclusion

This systematic review and meta-analysis examined eCB concentrations in individuals with SRDs compared with controls, both at rest and in response to acute psychosocial stress. While most studies reported stress-related increases in AEA and 2-AG in controls, the meta-analysis did not support these findings. Higher baseline concentrations of both AEA and 2-AG were observed in individuals with SRDs compared with controls, supported by meta-analysis.

Limited research explored the differential effects of psychosocial stress on eCBs in individuals with SRDs compared with controls, but the meta-analysis supported group differences in 2-AG response to acute stress. Most studies focused on female participants, individuals with PTSD, and measured eCBs in plasma, with substantial heterogeneity across studies. Further research is needed to elucidate the complex relationships between eCBs, SRDs, and acute psychosocial stress.

Abbreviations Used

2-AG

2-arachidonoylglycerol

AEA

anandamide

ANX

anxiety disorder

CB1

cannabinoid type 1

CON

control

DEPR

depressive disorder

DOI

digital object identifier

eCB

endocannabinoid

HPA

hypothalamic–pituitary–adrenal

MAGL

monoacylglycerol lipase

MAST

Maastricht Acute Stress Test

MDD

major depressive disorder

MiD

minor depression

PD

panic disorder

PTSD

post-traumatic stress disorder

SAD

social anxiety disorder

SRD

stress-related neuropsychiatric disorder

THC

Δ9-tetrahydrocannabinol

TRAUMA

trauma/stress-related disorder

TSST

Trier Social Stress Test

Authors' Contributions

L.C.G., J.M.E., and H.A.M. conceptualized and designed the project. They analyzed and interpreted the data. All authors have critically revised the final version of the article.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

Dr. Hilary A. Marusak is supported by National Institute of Mental Health grants K01MH119241 and R01MH132830 and Eunice Kennedy Shriver National Institute of Child Health and Human Development grant R21HD105882. Dr. Clara G. Zundel is supported by National Institute of Mental Health grant F32MH133274. Ms. Samantha L. Ely is supported by a National Institute of General Medical Sciences grant T32 GM139807. Dr. Leah M. Mayo is supported by Canadian Institutes of Health Research grant 186583.

Cite this article as: Gowatch LC, Evanski JM, Ely SL, Zundel CG, Bhogal A, Carpenter C, Shampine MM, O'Mara E, Mazurka R, Barcelona J, Mayo LM, Marusak HA (2024) Endocannabinoids and stress-related neurospsychiatric disorders: a systematic review and meta-analysis of basal concentrations and response to acute psychosocial stress, Cannabis and Cannabinoid Research 9:5, 1217–1234, DOI: 10.1089/can.2023.0246.