Eight reports from 8 studies investigated the ECS in individuals with AN [17,18,19,20,21,22,23,24]. Study designs ranged from case control, combined cross-sectional longitudinal, pre-meal/post-meal, and cross-sectional, with sample sizes ranging from 14 to 763 participants with AN and 7 to 1244 healthy controls (HC). Most participants were female or identified as women, except for 5 participants who identified as men. One study did not report participants’ gender/sex [23]. Outcomes of interest included genetic factors related to the ECS (alleles, genotypes, and haplotypes) which code for CB receptors, and circulating endocannabinoids. Further details about these reports can be found in Table 2.

Since genetic factors such as alleles and genotypes code for protein receptors in the ECS, differences in these genetic factors may lead to discrepancies in the number of available protein receptors. Three case–control studies explored alleles, genotypes, or haplotypes related to the ECS. One study found that the Val195 allele of G-protein-coupled receptor-55, which has an affinity for endocannabinoids, was significantly more frequent in participants with AN compared to HC [17]. Another study found that the 385A allele, which codes for an enzyme that degrades the endocannabinoid, arachidonoylethanolamine (AEA) was significantly less frequent in AN compared to HC [18]. The remaining study found that in participants with AN, the genotype rs806369-TT and haplotype​ rs806368/rs1049353/rs806369 of CB1 were associated with significantly lower weight and body mass index (BMI) compared to HC [19]. Additionally, the genotypes of rs806374 for CB1 and rs3003335 and rs6658703 for CB2 were associated with significantly higher occurrences of ED behaviors compared to participants with AN not carrying said genotypes.

It has been suggested that circulating endocannabinoids play roles in food intake, energy expenditure, and possibly engagement in ED behaviors [11, 21,22,23]. Both availability of receptors (the percentage of receptors that are unbound) to which circulating endocannabinoids bind and levels of circulating endocannabinoids are important to consider. One cross-sectional study found that CB1 availability was significantly increased in AN and positively associated with drive for thinness [20].

Understanding circulating endocannabinoid levels in those with AN are important to determine whether exogenous cannabinoid treatment may be able to ameliorate symptoms of the disorder. Circulating endocannabinoid levels were measured in participants with AN through plasma or hair samples in 1 combined cross-sectional longitudinal study, 2 pre-meal/post-meal studies, and 1 cross-sectional study [21,22,23,24]. Endocannabinoids measured included AEA, 2-arachidonoylglycerol (2-AG), Oleoylethanolamide (OEA), palmitoylethanolamide (PEA), and Stearoylethanolamide (SEA) which are naturally occurring and circulate within cells, adipose tissue, muscles, and the brain [25].

Because endocannabinoids are related to food intake, levels could change in relationship to consumption and whether the food is liked (hedonic) or not liked (non-hedonic) [26]. Both pre-meal/post-meal studies accounted for time since meals, 1 of which also accounted for differences in hedonic/non-hedonic eating [22, 23]. Three out of the 4 studies categorized participants with AN as underweight (UWAN) or weight restored (WRAN) [21,22,23].

All 4 studies measured AEA, 2 of which found significant differences between AN and HC. One study found significantly elevated AEA in UWAN and WRAN compared to HC [21] while the other found significantly lower AEA in UWAN and WRAN compared to HC [23]. Another study found consistently lower AEA after eating both hedonic and nonhedonic foods compared to fasting periods in both AN and HC [22]. Lower AEA in AN was associated with higher BMI and higher emotional dysregulation in a separate study [24].

All 4 studies measured 2-AG, 1 of which found significant differences between WRAN and HC [22]. In WRAN, 2-AG was significantly elevated after eating both hedonic and non-hedonic foods compared to fasting periods. HC showed significantly lower 2-AG after eating hedonic foods compared to non-hedonic foods. In a separate study, higher 2-AG was associated with poorer psychological states in those with AN [24].

Only 2 studies measured OEA, 1 of which found significantly elevated OEA in UWAN compared to HC [21]. When these participants in UWAN became weight restored, they showed significantly lower OEA compared to their baseline, indicating the possibility that OEA levels may begin to present similar to HC upon weight restoration. The other study found significantly lower OEA after eating hedonic foods compared to non-hedonic foods, which was consistent for those with AN and HC [22].

PEA was measured in 2 studies, 1 of which found significantly elevated PEA in UWAN compared to HC [21]. When these participants in UWAN became weight restored, they showed significantly lower PEA compared to results from when they were underweight, indicating the possibility that levels of PEA may begin to present similar to HC with weight restoration. The other study found significantly lower PEA after eating both hedonic and non-hedonic foods compared to fasting periods, which was consistent in AN and HC [22].

Only 1 study measured SEA and found significantly elevated SEA in UWAN compared to WRAN and HC [21]. Further, UWAN showed significantly lower SEA upon weight restoration compared to their baseline, indicating the possibility that SEA may begin to present similar to HC upon weight restoration.

Five reports from 3 studies including 1 crossover randomized controlled trial (RCT), 1 non-randomized study, and 1 case report explored cannabinoid treatment [27,28,29,30,31]. Cannabinoids were administered through capsules of THC or dronabinol, a synthetic form of THC [32]. Treatment dosages used in these reports ranged from 2 to 15 mg and length of intervention was between 4 and 6 weeks. Sample sizes ranged from 1 to 24 participants who were female or identified as women, except for 1 male from a case report. Outcomes measured included weight, ED symptoms, physical activity (PA), and adipose tissue hormones [27,28,29,30,31]. Further details about these reports can be found in Table 3.

Both the crossover RCT and case report observed weight gain in participants with AN treated with dronabinol [27, 31]. When participants from the crossover RCT received 5 mg treatment, they gained an average of 1 kg over 4 weeks, but only gained an average of 0.34 kg over 4 weeks when they received a placebo (p = 0.03 for difference) [27]. In the case report, the participant’s BMI increased from 19.5 to 21.0 kg/m2 over 6 weeks of 15 mg treatment [31]. On the other hand, the non-randomized study did not find significant weight changes after 2 mg treatment lasting 4 weeks [30].

The non-randomized study and case report observed improved ED symptoms in participants with AN during cannabinoid treatment [30, 31]. Participants from the non-randomized study reported significantly reduced ascetism and increased body care using the Eating Disorder Inventory and Eating Attitude Test, and significantly reduced depression using Beck Depression Inventory during 4 weeks of 2 mg treatment [30]. The participant from the case report reported reduced ED symptoms from items included in the Eating Disorder Examination Questionnaire over 6 weeks of 15 mg treatment [31]. Contrary to these findings, the crossover RCT found no significant changes in ED symptoms using the Eating Disorder Inventory-2 and the non-randomized study reported no significant changes in anxiety using the Spielberger State-Trait Anxiety Inventory, one of which involved 5 mg treatment lasting 4 weeks [27], while the other involved 2 mg treatment lasting 4 weeks [27, 30].

PA levels are often elevated in those with AN and contribute to maintenance of low weight status [33]. The crossover RCT found significant differences in PA during 5 mg treatment lasting 4 weeks based upon whether participants were inpatients or outpatients [28]. Dronabinol resulted in significant increases in the duration of PA in outpatients and significant increases in the intensity of PA in inpatients [28]. Conversely, in the case report, the participant’s PA urge and number of steps taken per day decreased during 15 mg treatment lasting 6 weeks [31].

Altered adipose tissue hormones, including, low leptin levels, low cortisol levels, high adiponectin levels, and low insulin-like growth factor (IGF) proteins are common in individuals with AN [34]. Both the crossover RCT (5 mg treatment lasting 4 weeks) and case report (15 mg treatment lasting 6 weeks) found no changes in leptin and an increase in cortisol during dronabinol treatment [29, 31]. Only the crossover RCT (5 mg treatment lasting 4 weeks) measured adiponectin and IGF proteins and found significantly lowered adiponectin when controlling for leptin levels and no significant changes in IGF proteins [29].

Benefits of cannabinoid treatment included weight gain, reduced ED symptoms, and reduced PA [27, 30, 31]. However, null effects/harms were also observed, including no changes in weight, ED symptoms, or adipose tissue hormones, and increased PA [27,28,29,30,31].

Four reports from 4 studies (2 retrospective cohort studies and 2 case reports) reported harms associated with cannabis use in AN [35,36,37,38]. Both participants from the case reports identified as women while the gender/sex breakdown of participants with AN could not be determined from the data included in the retrospective cohort studies. Sample sizes ranged from 1 to 16,922 participants, although the exact sample size of participants with AN from one of study could not be determined [35]. Outcomes of interest included genetic factors related to cannabis use disorder (CUD) in AN and associations between compensatory behaviors and cannabis abuse/dependence. Further details about these reports can be found in Table 3.

Two retrospective cohort studies assessed polygenic risk scores (PRS) for AN and CUD, 1 of which found that the highest PRS for AN were significantly associated with a higher risk of CUD [35], while the other did not find significant associations between PRS for AN and PRS for CUD [36].

Two case reports, each involving 1 participant with binge/purge type AN reported compensatory behaviors following cannabis use [37, 38]. One of these participants was using cannabis daily for at least 7 years [37], while the other participant was using cannabis daily for at least 3 years [38]. One participant presented to a treatment facility and was found to be suffering from cannabinoid hyperemesis syndrome (CHS), a condition associated with long term cannabis use causing recurrent vomiting [13, 37]. Although the other participant did not have CHS, appetite stimulation from cannabis appeared to encourage the participant to engage in postprandial compensatory behaviors to maintain low weight status, which was supported by a reduced BMI over the span of 3 years of daily use [38].