Serum lipid analysis and isotopic enrichment is suggestive of greater lipogenesis in young long-term cannabis users: a secondary analysis of a case-control study

Abstract

Cannabis is now legal in many countries and while numerous studies have reported on its impact on cognition and appetite regulation, none have examined fatty acid metabolism in young cannabis users. We conducted an exploratory analysis to evaluate cannabis impact on fatty acid metabolism in cannabis users (n=21) and non-cannabis users (n=16). Serum levels of some saturated and monounsaturated fatty acids, including palmitic, palmitoleic and oleic acids were higher in cannabis users compared to non-users. As palmitic acid can derive from diet or lipogenesis from sugars, we evaluated lipogenesis using a de novo lipogenesis index (palmitate/linoleic acid) and carbon specific isotope analysis, which allows for the determination of fatty acids ¹³C signature. The significantly higher de novo lipogenesis index in the cannabis users group along with a more enriched ¹³C signature of PA suggested an increase in lipogenesis. Additionally, while serum glucose concentration did not differ between groups, pyruvate and lactate were lower in the cannabis user group, with pyruvate negatively correlating with palmitic acid. Furthermore, the endocannabinoid 2-arachidonoylglycerol was elevated in cannabis users and could contribute to lipogenesis by activating the Cannabinoid receptor 1. Because palmitic acid has been suggested to increase inflammation, we measured peripheral cytokines and observed no changes in inflammatory cytokines. Finally, an anti-inflammatory metabolite of palmitic acid, palmitoylethanolamide was elevated in cannabis users. Our results suggest that lipogenic activity is increased in cannabis users, however future studies including prospective studies that control dietary intake are required.

Introduction

With its legalization in several countries, including Canada, the use of cannabis has steadily decreased stigmatization (Budney et al., 2007; CRIME, 2011; Jacobson et al., 2021). Cannabis is a plant of the Cannabaceae family and contains several active phytocannabinoids, among which delta9- tetrahydrocannabinol (D⁹-THC) and cannabidiol (CBD) are the most studied. Cannabinoids are agonists of the cannabis receptors, CB1 and CB2, which make up part of the endogenous endocannabinoid system (ECS) (Lu and Mackie, 2016). These receptors are found both in the central nervous system as well as in the periphery on both immune and adipose cells (Pertwee, 1997). Along with endogenous cannabinoids, a class of bioactive lipid mediators, regulate food intake, peripheral energy metabolism and lipid metabolism in adipose tissue (Cota et al., 2003; Di Marzo et al., 2001).

Phytocannabinoids and endocannabinoids activate the CB1 receptor, and its repeated activation is thought to lead to tolerance and dependence (Gonzalez et al., 2005; Jones et al., 1981). Chronic cannabis use can lead to cannabis use disorder (CUD), which is defined as the inability to stop using cannabis even when causing physical or psychological harm. One risk factor for the development of CUD is use at a young age. Typically, cannabis use starts during the teenage years, with a mean age at first exposure being 16–17 years old. However, relatively little is still known about THC effects, especially during adolescence and young adulthood, a phase of greater vulnerability for brain development (Dhein, 2020). Importantly, THC exposure during the adolescent period delays the maturation of the prefrontal cortex in rats (Cass et al., 2014). In Canada, 45% of people aged 15 years and over reported to have tried cannabis (Hango, 2018).

One of the effects of cannabis use and the activation of the ECS is the increase in appetite and caloric intake (Foltin et al., 1986; Foltin et al., 1988), suggesting that cannabis use can impact feeding and body weight (Hirsch and Tam, 2019). Importantly, chronic use can lead to cravings and a poor diet (Filbey et al., 2009). However, an inverse relationship between BMI and cannabis use was reported in the literature (Smit and Crespo, 2001). Despite this, while a substantial number of studies have reported on the impact of cannabis use on cognitive functions, the impact of cannabis use on metabolic processes, and more specifically on fatty acids and DNL, in humans, is still poorly understood.

Thus, we conducted a secondary analysis of a clinical study (Da Silva et al., 2019; Hafizi et al., 2017; Watts et al., 2020) to evaluate whether cannabis use could alter fatty acid metabolism, their bioactive metabolites and consequently the immune response in CU compared to non-cannabis users (NCU) following overnight abstinence.

Material and Methods

Patient recruitment and demographic:

NCU and cannabis users (CU) were first screened through an introductory phone call, and then invited for a baseline visit at the Centre for Addiction and Mental Health (CAMH). All study procedures were thoroughly explained to participants before informed consent was acquired for all participants, and the study was approved by the Research Ethics Board at CAMH (approval number 029/2010). CU were classified based on self-report, which accounts for a minimum usage of four times per week for the 12 months prior to baseline and/or met criteria for CUD including a positive cannabis urine drug screen. A past history of or current use of any psychoactive drug was an exclusion criterion for control subjects (Da Silva et al., 2019).

Assessment of inclusion and exclusion criteria for the study were performed using the Diagnostics Statistics Manual (DSM-IV) Structured Clinical Interview for Axis I disorders. Exclusion criteria included: current or past substance use disorder (except for cannabis use disorder); unstable medical or neurological illness; past or current Axis I disorder, including but not limited to major depressive disorder and/or anxiety disorders; history of severe head trauma; pregnancy or current breastfeeding; or presence of metal implants precluding a magnetic resonance imaging scan. For the assessment of cannabis use patterns, an in-house questionnaire called “Drug History Questionnaire” (DHQ) was used as previously reported and validated in (Da Silva et al., 2019). CUD categorization was affirmed using assessments from the DSM-IV and DHQ. Abstinence from cannabis for 12 hours prior to PET scan was requested and was self-reported (Da Silva et al., 2019).

Serum sample collection:

Blood samples were collected from NCU and CU prior the injection of PET radiotracers scan for previous studies reported elsewhere (Da Silva et al., 2019; Hafizi et al., 2017; Jacobson et al., 2021; Watts et al., 2020). Whole-blood samples were centrifuged at 3000 x g for 5 minutes at 4C to isolate serum. Serum samples were then stored at −80C until use (Da Silva et al., 2019).

Blood serum levels of THC, OH-THC, COOH-THC, CBD metabolites:

Briefly, 1 mL serum specimens were spiked with deuterated (D3) internal standards (Cerilliant) and then solid phase extracted using the Bond Elut Certify II (200 mg) cartridges (Agilent Technologies). The extracts were evaporated under nitrogen and BSTFA +1% TMCS-derivatized prior to the GC-MS analysis using a ThermoFisher platform: ISQ-LT single quadrupole mass spectrometer with Trace 1310 gas chromatograph fitted with a 20 m x 0.18 mm x 0.18 μm TG-5ms GC Column. The analyte quantification was in SIM mode using the target ions as follows (retention times also indicated): THC 386 Da (6.32 min); OH-THC 371 Da (7.30 min); COOH-THC 371 Da (7.82 min); CBD-390 Da (3.78 min). Calibration with internal standardization was performed with linear regression curve fits with 1/X weighting. Unknowns were quantified against a standard curve ranging from 0.2 to 200 ng/mL for THC (LOD 0.5 ng/mL), OH-THC (LOD 1 ng/mL) and COOH-THC (LOD 1 ng/mL) and from 0.1 to 200 ng/mL for CBD (LOD 0.2 ng/mL). For more details see (Da Silva et al., 2019). The levels of these metabolites were available only for a subgroup of CU subjects and maybe not be representative of the whole population. Results are reported in Table 1.

Table 1.

Demographic

	NCU	CU	P value
Number of participants (F)	16 (10)	21 (9)
Age years (Mean ± SD)	21.4 ± 1.9	23.2 ± 4.9	n.s.
BMI (Mean ± SD)	23.3 ± 4.7	24.5 ± 5.4	n.s.
Weight kg (Mean ± SD)	64.6 ± 14.0	74.2 ± 19.0	n.s.
Number of people overweight (BMI 25–30) (%of group)	3 (18.75%)	4 (19.0%)	X 2 = 9.561; p= 0.144
Number of people obese (BMI >30) (%of group)	2 (12.5%)	5 (23.8%)	X2= 8.376; p= 0.212
Tobacco	0	7	X 2 = 6.578; p= 0.010
Reported alcohol use	0	9	X 2 = 9.6; p= 0.002
Reported MDMA use	0	13	X 2 = 15.27 p= 0.000
RBANS score – Total scale (Mean ± SD)	84.9 ± 13.7	87.8 ± 12.4	n.s.
Immediate Memory (Mean ± SD)	92.7 ± 13.2	83.2 ± 11.3	P = 0.027
Visual/Spatial (Mean ± SD)	80.9 ± 19.9	88.1 ± 17.8	n.s.
Language (Mean ± SD)	79.9 ± 20.1	96.2 ± 17.0	P = 0.012
Attention (Mean ± SD)	101 ± 14.8	99.2 ± 13.2	n.s.
Delayed Memory (Mean ± SD)	88.6 ± 13.3	87.4 ± 14.2	n.s.
Cannabis
Age at first use (y) (Mean ± SD)	NA	16.3 ± 3.6 (n=21)	na
Age at regular use (y) (Mean ± SD)	NA	18.85 ± 4.4 (n=20)	na
Dose (g) (Mean ± SD)	NA	1.4 ± 0.88 (n=21)	na
Cumulative Exposure (g) (Mean ± SD)	NA	1970 ± 1645 (n=20)	na
Dose past 12 months (g) (Mean ± SD)	NA	436.6 ± 276.1 (n=20)	na
Time since last THC (h) (Mean ± SD)	NA	21.12 ± 12.3 (n=20)	na
Craving	NA	9	na
THC ng/ml (Mean ± SD)	NA	9.4 ± 10 (n=13)	na
OH-THC ng/ml (Mean ± SD)	NA	4.6 ± 4.6 (n=13)	na
COOH-THC ng/ml (Mean ± SD)	NA	84.5 ± 88.1 (n=13)	na
CBD ng/ml (Mean ± SD)	NA	0.07 ± 0.23 (n=12)	na
CUD	NA	11	na
MCQ (Mean ± SD)	NA	42.8 ± 11.8	na

BMI, body mass index; CBD, cannabidiol; CU, cannabis use; COOH-THC, 11-nor-9-carboxy-THC; CUD, cannabis use disorder; F, female; g, grams; h, hours; MCQ, Marijuana Craving Questionnaire; MDMA, 3,4-Methylenedioxymethamphetamine; na, not available; NCU, non-cannabis users; n.s.; not significative; OH-THC; 11-Hydroxy-Δ⁹-tetrahydrocannabinol; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; SD, standard deviation; THC, Tetrahydrocannabinol; y, years

In case the information was available only for the CU group, t-test was not performed, and the p-value was not available.

Fatty acid analyses:

Total lipids were extracted from serum sample following the protocol described by (Cisbani et al., 2020; Metherel et al., 2016) and modified from (Folch et al., 1957). Briefly, 100µl of serum was extracted in a 2:1 chloroform:methanol:0.88% potassium chloride solution. A known amount of the internal standard docosatrienoic acid (22:3n-3) ethyl ester was included for quantitation of individual fatty acids. Fatty acid methyl esters (FAMEs) were obtained by trans-esterification with 14% boron trifluoride in methanol. FAMEs were collected and dissolved in 0.1 mL hexane for further analysis.

FAME Quantification by Gas Chromatography (GC)–Flame Ionization Detection (FID):

FAMEs were quantified with a SP-2560 biscyanopropyl siloxane, capillary column (100 m length × 0.25 mm diameter × 0.20 μm film thickness (Supelco, Bellefonte, PA, USA) in a Varian 430 GC (Bruker; Billerica, MA, USA) as previously described (Lacombe et al., 2018; Lacombe et al., 2017). The column over program was initially set to 60°C, increased at a rate of 8.5°C/minute to 170°C and held for 6.12 minutes, increased at a rate of 4.3°C/minute to 175°C, 1.7°C/minute to 185°C, 0.8°C/minute to 190°C, and finally increased to 240°C at 8.5°C/minute and held for 27.76 minutes; totaling 60 minutes. Peaks were identified using an external reference standard (GLC 569) (Nu Chek Prep; Elysian, MN, USA) and quantified in Compass CDS (Version 3.0.0.68) by dividing the peak area under the curve (AUC) of the FA with the AUC of the internal standard (22:3n-3). GC vials were recapped and stored at −80 °C until further analysis by GC–combustion-isotope ratio mass spectrometry (GC-C-IRMS).

Here we report the relative percentages (weight %) of seven saturated FAs (SFA) (myristic acid, 14:0; palmitic acid, 16:0 (PA); margaric acid, 17:0; stearic acid, 18:0 (STA); arachidic acid, 20:0; behenic acid, 22:0; lignoceric acid, 24:0), six monounsaturated FA (MUFA) (omega-7: palmitoleic acid, 16:1n-7; vaccenic acid, 18:1n-7; omega-9: oleic acid, 18:1n-9; eicosenoic acid, 20:1n-9; erucic acid, 22:1n-9; nervonic acid, 24:1n-9) and thirteen polyunsaturated FA (PUFA) (omega-6 (n-6): linoleic acid (LA), 18:2n-6; gamma linolenic acid (GLA), 18:3n-6; eicosadienoic acid, 20:2n-6; dihomo-gamma linolenic (DGLA), 20:3n-6; arachidonic acid (ARA), 20:4n-6; docosadienoic acid, 22:2n-6; adrenic acid, 22:4n-6; docosapentaenoic acid-omega6, 22:5n-6; omega-3 (n-3): alpha linolenic, (ALA) 18:3n-3; eicosatrienoic acid, 20:3n-3; eicosapentaenoic acid (EPA), 20:5n-3; docosapentaenoic acid-omega3, 22:5n-3; docosahexaenoic acid (DHA), 22:6n-3).

Stable Carbon Isotope Analysis:

Compound specific isotope analysis (CSIA) was then performed on FAMEs using GC-C-IRMS. FAMEs were separated on a SP-2560 capillary column (100m x 0.25 mm x 0.2 µm film) (Supelco, Bellefonte, PA, USA) within a Delta V Plus Isotope Ratio Mass Spectrometer coupled to a Thermo Trace 1310 GC (Thermo Scientific). Column flow rate was 1.2mL/min and the oven temperature was initially set to 60°C, increased to 180°C at 15°C/minute, increased at 1.5°C/minute to 240°C and held for 18 minutes as previously described in (Metherel et al., 2017). IRMS chromatograms were analyzed with Isodat Workspace (version 3.0, Thermo Scientific). Certified calibrated 20-carbon FAME reference materials USGS70, USGS71, and USGS72 (Reston Stable Isotope Laboratory—United States Geological Survey, Reston, VA) were injected in triplicate along with the samples to account for carbon fractionation throughout the run and later produce linear trendlines for analysis. Isotopic values (¹³C) are reported as the difference in the ¹³C:¹²C ratio of a given compound in relation to the ¹³C:¹²C ratio of a more abundant ¹³C universal reference (δ¹³C) and it is reported in milliUrey (mUr) (equivalent to ‰ (per mil)). Of note, when δ¹³C values are compared, a value that is less negative indicates a compound that is more enriched in ¹³C. Of special relevance here, dietary added sugars from corn and sugar cane are highly enriched in ¹³C and yield higher (less negative) δ¹³C, and an emerging literature suggests they can be used as marker of DNL from added sugars (Klingel et al., 2019; Lacombe and Bazinet, 2020; Lacombe et al., 2018; Metherel et al., 2019). For details on the method see Lacombe et al., 2018 and Klingel et al., 2019 (Klingel et al., 2019; Lacombe et al., 2018). CSIA was performed by the Analytical Facility for Bioactive Molecules (AFBM), at The Hospital for Sick Children, Toronto, Canada.

Fatty acid ethanolamides:

Fatty acid ethanolamides were extracted and quantified by LC/MS/MS at the Analytical Facility for Bioactive Molecules (The Hospital for Sick Children, Toronto, Canada). Samples (200 μL), standards, and deuterated internal standards were added to Eppendorf tubes and brought to a total volume of 1 mL with water. Tubes were vortexed and then loaded onto preconditioned (1 mL methanol then 1 mL water) Oasis HLB SPE tubes. The SPE tubes were washed twice with 1 mL of 40/60 methanol/water and then the fatty acid ethanolamines were eluted into conical tubes using 2 mL of acetonitrile. The acetonitrile was evaporated under a gentle flow of nitrogen gas and the residue reconstituted in 150 μL of acetonitrile. Samples were injected onto a Kinetex XB-C18 50 × 3.0 mm column on an Agilent 1290 LC system coupled to a Sciex Q-Trap 5500 mass spectrometer. Samples were eluted using a gradient of A) 0.1% formic acid and B) 0.1% formic acid in acetonitrile. Data was collected and analyzed using SCIEX Analyst v1.7.

Glucose:

Serum glucose concentration were measured using the glucose strips with the Keto-Mojo meter (Keto-Mojo, 952 School Street 212, Napa, CA 94559). Use of glucose strips were previously validated and had variation similar to the glucose oxidate-based colorimetric assay (Anderson et al., 2010).

Lactate and Pyruvate:

Peripheral lactate and pyruvate concentrations were measured in 12 NCU and 12 CU subjects using a colorimetric L-Lactate Assay Kit and a colorimetric Pyruvate Assay Kit, respectively as described previously (Da Silva et al., 2018). Fifty µL of sample was added to 96-well plates in duplicates and incubated with 50 µL of lactate reaction mix or pyruvate reaction mix for 30 minutes at room temperature. The samples were then measured by a microplate reader at λ = 450 nm for lactate and at λ = 570nm for pyruvate. Lactate and pyruvate concentrations are reported in mmoles/L. The levels of lactate and pyruvate in the NCU group were previously reported in (Da Silva et al., 2018).

Insulin analyses:

Serum insulin concentration was measured using the Human Insulin Elisa Kit (EZHI-14K, Merck Millipore, Billerica, MA, USA) according to manufacture instructions. Absorbance was read using SkanIt Software 6.1 RE for Microplate Readers RE (ver. 6.1.0.51). Values were expressed as µU/ml. Serum samples were not available for one sample in each group. Two samples in NCU group and 7 samples in the CU group were under the detection limit, likely because of limited serum volume, and 1 sample in the CU group was over the detection limit.

Cytokine analyses:

Serum samples were thawed and eight cytokines (IFNγ, TNFα, IL-1β, IL-2, IL-6, IL-8, IL-10 and IL-12) were analyzed using Human High Sensitivity T-Cell panel (HST CYTOMAG60SK, Merck Millipore, Billerica, MA, USA) according to manufacture instruction. HsCRP was also measured in serum using high-sensitivity ELISA according to the manufacturer’s instructions (IBL-international, Hamburg, Germany). For more details consult (Da Silva et al., 2019).

Statistical analyses:

Statistical analyses were carried out following consultation with a statistician at CAMH. Chi-square test was used to evaluate group differences in the categorical variables (e.g. tobacco, MDMA and Alcohol use) as presented in Table 1. Being this a secondary exploratory analysis of a case control study, power analyses was not performed for this study. Differences in continuous variables (e.g. age, BMI) were assessed using analysis of variance (ANOVA). Differences between the two groups were assessed by unpaired t-test or Mann Whitney test, following the assessment of normality test with the Shapiro-Wilk test. Pearson’s correlation was performed to assess the relationship between variables. As an exploratory analysis, we did not conduct multiple comparison corrections to control for false discovery rates. For the purpose of this report, the significance for all statistical analysis remained at p ≤ 0.05, and conclusions should be interpreted with caution. Statistical analyses were performed using SPSS software (Version 25) and GraphPad (Version 9.0.2) and no values were deemed to be outliers nor were any excluded from the analysis. Figures were prepared with Adobe Illustrator.

Results

Demographics.

Selected subjects demographics are reported in Table 1 and previously described in (Da Silva et al., 2019). Subjects included in this study are NCU (n = 16) and long-term young CU (n = 21). Subjects did not differ for age, body weight, height, or BMI. THC is known to influence food intake and body weight and previous studies reported differences in BMI, either lower (Smit and Crespo, 2001) or higher (Mendelson, 1976) in CU compared to non-users. The reported average age at first use of cannabis for the CU group was 16.3 ± 3.6 years. CU were abstinent for about 21 hours (± 12.3h) prior to blood being withdrawn (Table 1).

Changes in SFA and MUFA in CU.

To assess whether cannabis use could affect the serum FA profile, we analyzed individual serum fatty acids and total SFA, MUFA and PUFA. The results are summarized in Table 2. We observed statistically significant differences in the levels of some FA belonging to the SFA and MUFA biosynthesis pathways, summarized in Fig. 1A. More specifically, PA was elevated in CU group compared to the NCU group (Fig. 1B; p = 0.001), along with palmitoleic acid (Fig. 1C; p = 0.003) and vaccenic acid (Fig. 1C; p = 0.02). While stearic acid (Fig. 1B; p = 0.81) did not differ between group, oleic acid (Fig. 1C; p = 0.02) was also elevated in CU. Overall, relative percentages of SFA and MUFA were elevated in CU users compared to NCU (Fig. 1B,C; p = 0.03 and p = 0.01 respectively). As subjects in the CU group reported alcohol consumption or tobacco use (Table 1) which could alter serum fatty acids (Simon et al., 1996), we excluded these subjects and statistical differences between groups for PA levels were maintained (Table 3 and 4). As 23.8% and 12.5% of CU and NCU respectively had a BMI over 30 and 19% of subjects in each group had a BMI between 25 and 30, we excluded these subjects to verify whether the BMI impacted the results observed. Indeed, statistically significant differences were maintained even when excluding subjects with a BMI over 30 or over 25 (data not shown). Furthermore, we did not detect any correlation between PA and BMI (data not shown).

Table 2.

Relative percentage (weight %) of fatty acids in serum samples of non-cannabis users and cannabis users.

	NCU (n=16)		CU (n=21)		% of difference	T-test
FAs	Mean	SEM	Mean	SEM	CU vs NCU	F	P	Lower	Upper
14:0	0.69	0.08	0.86	0.12	25.28	2.40	0.27	−0.49	0.14
16:0	20.04*	0.53	22.84	0.58	13.95	0.27	0.00	−4.43	−1.16
17:0	0.32	0.02	0.34	0.01	6.84	1.01	0.27	−0.06	0.02
18:0	7.97	0.46	7.80	0.47	−2.03	0.11	0.81	−1.21	1.53
20:0	0.26	0.13	0.11	0.00	−57.69	5.27	0.20	−0.08	0.38
22:0	0.40	0.09	0.28	0.02	−32.04	4.88	0.10	−0.02	0.28
24:0	0.20*	0.02	0.16	0.01	−22.19	2.59	0.02	0.01	0.08
SFAs	29.88*	0.88	32.39	0.70	8.38	0.06	0.03	−4.76	−0.25
16:1	0.98*	0.08	1.45	0.11	47.77	2.10	0.00	−0.77	−0.17
18:1n-7	1.54*	0.04	1.73	0.06	12.28	2.76	0.02	−0.34	−0.04
18:1n-9	17.67*	0.73	19.80	0.53	12.10	0.69	0.02	−3.92	−0.35
20:1n-9	0.21	0.02	0.23	0.01	9.78	0.01	0.35	−0.06	0.02
22:1n-9	0.03	0.01	0.03	0.01	6.84	1.96	0.82	−0.02	0.02
24:1n-9	0.38*	0.03	0.28	0.01	−26.35	6.31	0.00	0.05	0.16
MUFAs	20.86*	0.76	23.59	0.63	13.11	0.19	0.01	−4.72	−0.75
18:2n-6	31.77	1.00	31.03	0.97	−2.34	0.02	0.60	−2.12	3.61
18:3n-6	0.32	0.04	0.36	0.05	12.30	2.22	0.57	−0.18	0.10
20:2n-6	0.25	0.03	0.25	0.01	1.32	2.32	0.91	−0.06	0.06
20:3n-6	1.46	0.12	1.59	0.10	9.28	0.04	0.39	−0.45	0.18
20:4n-6	7.47	0.30	7.81	0.49	4.47	5.34	0.59	−1.59	0.93
22:2n-6	0.07	0.01	0.08	0.01	14.68	1.08	0.43	−0.04	0.02
22:4n-6	0.26	0.02	0.25	0.01	−4.81	5.67	0.49	−0.02	0.05
22:5n-6	3.90*	0.36	1.30	0.21	−66.73	1.77	0.00	1.79	3.41
N-6	45.51**	0.77	42.67	1.11	−6.23	1.70	0.06	−0.09	5.77
18:3n-3	0.85	0.10	0.81	0.06	−4.70	0.68	0.72	−0.19	0.27
20:3n-3	0.05	0.01	0.05	0.00	5.64	1.47	0.71	−0.02	0.01
20:5n-3	0.72	0.18	0.56	0.07	−22.29	1.32	0.36	−0.19	0.51
22:5n-3	0.51	0.03	0.47	0.03	−8.65	0.07	0.35	−0.05	0.14
22:6n-3	1.62	0.21	1.48	0.13	−9.04	0.75	0.55	−0.34	0.64
N-3	3.75	0.37	3.36	0.19	−10.35	1.47	0.32	−0.39	1.17
PUFAs	49.26*	0.79	46.04	1.21	−6.55	2.62	0.04	0.09	6.36
Total FA (mol/ml)	8926	606	10058	560	12.7	0.076	0.182	−2821	546
N-6/N-3	5.58	0.50	5.02	0.33	−10.08	0.13	0.34	−0.61	1.73
MUFA/SFA	0.71	0.04	0.73	0.02	3.01	0.81	0.63	−0.11	0.07
16:1n-7/16:0	0.05*	0.00	0.06	0.04	28.31	0.06	0.02	−0.25	−0.002
18:1n-9/18:0	2.33	0.16	2.44	0.11	4.69	0.32	0.59	−0.48	0.28
(18:0+18:1n-9)/16:0	1.29	0.05	1.22	0.31	−5.43	1.60	0.21	−0.042	0.19
16:0/18:2n-6	0.64*	0.03	0.75	0.03	16.65	0.08	0.02	−0.20	−0.02

Values reported as mean ± SEM. Data was analyzed with either Unpaired t-test.

^*P ≤ 0.05. CU, cannabis users; NCU, non-cannabis users C181n-9/18:0 ratio: delta 9 desaturation index; C16:0/C18:2n-6: lipogenic index; (C18:0+C18:1n-9)/C16:0: elongation index for 16:0; MUFA/SFA: total desaturation index

Figure 1. Changes in the plasma levels of saturated and monounsaturated fatty acids in cannabis users.

(A) Graphical summary of the saturated and monounsaturated fatty acid biosynthesis pathways reporting the changes in cannabis users compared to NCU (red arrow). (B, C) Graphs representing the levels of saturated (B) and monounsaturated (C) fatty acids reported as relative percentage in non-cannabis users and cannabis users. CU, n = 21; NCU, n = 16. Values reported as means ± SEM. Data was analyzed with either Unpaired t-test (16:0, 18:1n-9, SFA, MUFA, PUFA) or Mann Whitney Test (18:0; 24:0, 16:1n-7, 24:1n-9). * P ≤ 0.05; ** P ≤ 0.005. CU, cannabis users; NCU, non. See Table 2 for the complete fatty acid profile.

Table 3.

Relative percentage (weight %) of fatty acids in serum samples of non-cannabis users and cannabis users excluding alcohol users.

	NCU (n=16)		CU (n=11)		% of difference	T test
	Mean	SEM	Mean	SEM	CU vs NCU	F	P	Lower	Upper
C 14:0	0.69	0.08	0.85	0.09	24.27	0.09	0.19	−0.42	0.09
C 16:0	20.04*	0.53	23.09	0.61	15.19	0.29	0.00	−4.72	−1.37
C 17:0	0.32	0.02	0.35	0.01	11.44	5.52	0.10	−0.08	0.01
C 18:0	7.97	0.46	8.40	0.30	5.47	0.56	0.48	−1.70	0.82
C 20:0	0.26	0.13	0.11	0.01	−57.77	2.65	0.36	−0.18	0.48
C 22:0	0.40	0.09	0.31	0.02	−24.45	3.04	0.35	−0.12	0.31
C 24:0	0.20	0.02	0.17	0.01	−15.49	1.60	0.18	−0.02	0.08
SFAs	29.88*	0.88	33.28	0.77	11.39	0.66	0.01	−5.95	−0.85
C 16:1	0.98*	0.08	1.48	0.15	50.60	2.91	0.00	−0.82	−0.18
C 18:1n-7	1.54	0.04	1.72	0.09	11.08	2.38	0.07	−0.36	0.02
C 18:1n-9	17.67	0.73	19.46	0.79	10.16	0.06	0.12	−4.06	0.47
C 20:1n-9	0.21	0.02	0.22	0.02	2.56	0.13	0.84	−0.06	0.05
C 22:1n-9	0.03	0.01	0.03	0.01	−17.18	0.02	0.48	−0.01	0.02
C 24:1n-9	0.38*	0.03	0.29	0.01	−23.32	5.09	0.01	0.02	0.16
MUFAs	20.86	0.76	23.25	0.93	11.46	0.00	0.06	−4.86	0.08
C 18:2n-6	31.77	1.00	31.49	1.69	−0.90	0.66	0.88	−3.51	4.08
C 18:3n-6	0.32	0.04	0.47	0.08	47.34	1.05	0.07	−0.32	0.01
C 20:2n-6	0.25	0.03	0.26	0.02	2.25	1.85	0.88	−0.08	0.07
C 20:3n-6	1.46	0.12	1.76	0.12	20.92	0.00	0.11	−0.68	0.07
C 20:4n-6	7.47*	0.30	8.77	0.65	17.39	5.38	0.05	−2.63	0.03
C 22:2n-6	0.07	0.01	0.08	0.01	7.66	0.61	0.65	−0.03	0.02
C 22:4n-6	0.26	0.02	0.27	0.01	1.11	8.89	0.89	−0.05	0.04
C 22:5n-6	3.90*	0.36	0.67	0.19	−82.91	4.69	0.00	2.27	4.20
N-6	45.51	0.77	43.76	1.81	−3.84	2.54	0.33	−1.87	5.36
C 18:3n-3	0.85	0.10	0.74	0.07	−12.91	0.85	0.44	−0.18	0.39
C 20:3n-3	0.05	0.01	0.05	0.00	−6.07	1.23	0.74	−0.01	0.02
C 20:5n-3	0.72	0.18	0.68	0.10	−4.62	0.66	0.89	−0.44	0.51
C 22:5n-3	0.51	0.03	0.54	0.03	4.67	0.19	0.64	−0.13	0.08
C 22:6n-3	1.62	0.21	1.55	0.24	−4.62	0.01	0.82	−0.59	0.74
N-3	3.75	0.37	3.55	0.31	−5.24	0.15	0.71	−0.86	1.26
PUFAs	49.26	0.79	47.32	1.96	−3.95	4.01	0.31	−1.91	5.80
N-6/N-3	5.58	0.50	4.98	0.46	−10.79	0.20	0.41	−0.87	2.07
MUTFA‎/SFA	0.71	0.04	0.70	0.02	−2.05	1.81	0.78	−0.09	0.12
16:0/18:2n-6 ratio	0.64*	0.03	0.75	0.03	16.24	0.35	0.03	−0.20	−0.01
16:1/16:0	0.05*	0.00	0.06	0.01	29.68	0.74	0.05	−0.03	0.00
18:1n-9/18:0	2.34	0.16	2.42	0.08	3.32	1.95	0.71	−0.51	0.36
Elongation index	1.29	0.05	1.26	0.04	−2.86	2.31	0.59	−0.10	0.18
16:0 (mUr)	−25.27*	0.39	−24.02	0.22	−4.97	6.01	0.02	−2.29	−0.22
18:1n-9 (mUr)	−27.37	0.35	−27.64	0.60	1.01	2.26	0.68	−1.06	1.62
18:2n-6 (mUr)	−29.98*	0.35	−31.14	0.36	3.88	0.53	0.03	0.09	2.24
18:0 (mUr)	−25.60	0.42	−24.90	0.26	−2.73	1.33	0.21	−1.83	0.43

Table 4.

Relative percentage (weight %) of fatty acids in serum samples of non-cannabis users, cannabis users excluding alcohol users and cannabis users that reported alcohol use.

	NCU (n=16)		CU (n=12)		CU and alchool use (n=9)		% of difference
	Mean	SEM	Mean	SEM	Mean	SEM	CU vs NCU	CU (alcohol use) vs NCU
14:0	0.69	0.08	0.93	0.16	0.77	0.19	35.64	11.46
16:0	20.04*	0.53	23.44	0.70	22.04	0.95	16.95	9.94
17:0	0.32	0.02	0.34	0.02	0.33	0.02	9.07	3.90
18:0	7.97	0.46	7.59	0.77	8.10	0.44	−4.78	1.63
20:0	0.26	0.13	0.11	0.01	0.11	0.01	−59.05	−55.84
22:0	0.40	0.09	0.28	0.02	0.26	0.02	−30.09	−34.64
24:0	0.20	0.02	0.16	0.02	0.16	0.01	−22.76	−21.43
SFA	29.88	0.88	32.85	0.79	31.77	1.28	9.94	6.31
16:1	0.98*	0.08	1.58	0.18	1.28	0.10	61.07	30.04
18:1n-7	1.54	0.04	1.74	0.09	1.73	0.07	12.51	11.99
18:1n-9	17.67	0.73	19.93	0.90	19.64	0.37	12.79	11.18
20:1n-9	0.21	0.02	0.23	0.02	0.24	0.02	7.55	12.71
22:1n-9	0.03	0.01	0.03	0.00	0.04	0.02	−13.46	33.69
24:1n-9	0.38*	0.03	0.27	0.02	0.30	0.01	−29.51	−22.11
MUFAs	20.86	0.76	23.84	1.07	23.27	0.41	14.27	11.57
18:2n-6	31.77	1.00	31.29	1.63	30.67	0.74	−1.51	−3.46
18:3n-6	0.32	0.04	0.44	0.08	0.25	0.05	37.07	−20.71
20:2n-6	0.25	0.03	0.27	0.02	0.23	0.02	7.71	−7.17
20:3n-6	1.46	0.12	1.74	0.11	1.40	0.16	19.09	−3.80
20:4n-6	7.47	0.30	8.31	0.72	7.14	0.60	11.19	−4.51
22:2n-6	0.07	0.01	0.09	0.02	0.07	0.00	26.74	−1.55
22:4n-6	0.26	0.02	0.26	0.01	0.24	0.02	−0.48	−10.58
22:5n-6	3.90*#	0.36	0.97	0.24	1.74	0.33	−75.18	−55.46
N-6	45.51	0.77	43.37	1.80	41.74	1.05	−4.70	−8.27
18:3n-3	0.85	0.10	0.80	0.08	0.82	0.10	−5.38	−3.79
20:3n-3	0.05	0.01	0.05	0.00	0.06	0.01	−3.48	17.59
20:5n-3	0.72	0.18	0.62	0.10	0.47	0.07	−13.24	−34.36
22:5n-3	0.51	0.03	0.50	0.04	0.43	0.05	−2.88	−16.34
22:6n-3	1.62	0.21	1.50	0.22	1.45	0.12	−7.59	−10.99
N-3	3.75	0.37	3.47	0.29	3.22	0.20	−7.47	−14.20
PUFAs	49.26	0.79	46.84	1.93	44.96	1.17	−4.91	−8.73
N-6/N-3	5.58	0.50	4.80	0.47	5.31	0.45	−14.01	−4.83
MUFA‎/SFA	0.71	0.04	0.73	0.03	0.74	0.03	2.07	4.25
16:0/18:2n6 ratio	0.64	0.03	0.77	0.04	0.73	0.04	19.51	12.83
16:1/16:0	0.05	0.00	0.07	0.01	0.06	0.00	36.13	20.87
18:1n9/18:0	2.34	0.16	2.51	0.13	2.47	0.16	7.25	5.83
elongation index	1.29	0.05	1.22	0.04	1.23	0.05	−5.96	−5.33
16:0 (mUr)	−25.27*	0.39	−24.12	0.21	−24.41	0.37	−4.57	−3.44
18:1n-9 (mUr)	−27.37	0.35	−27.79	0.55	−27.85	0.23	1.54	1.75
18:2n-6 (mUr)	−29.98#	0.35	−31.26	0.38	−31.82	0.53	4.26	6.15
18:0 (mUr)	−25.60	0.42	−24.93	0.23	−24.94	0.39	−2.59	−2.55
Lactate	2.89	0.27	1.85	0.30	2.33	0.78	−36.02	−19.45
Pyruvate	0.07*	0.01	0.04	0.01	0.05	0.02	−46.98	−27.02

Values reported as mean ± SEM. Data was analyzed with either Unpaired t-test. * P ≤ 0.05. CU, cannabis users; NCU, non-cannabis users C181n-9/18:0 ratio: delta 9 desaturation index; C16:0/C18:2n-6: lipogenic index; (C18:0+C18:1n-9)/C16:0: elongation index for 16:0; MUFA/SFA: total desaturation index

One-way anova followed by Bonferroni post-hoc test was performed to assess differences between the three groups.

^*significantly different compared to CU;

^#significantly different compared to CU reporting alcohol use.

Increased de novo lipogenesis in CU.

Because PA was higher in CU compared to NCU, we then assessed DNL. First, we analyzed the DNL index which is the ratio between PA (16:0) and linoleic acid (18:2 n-6). Interestingly, the DNL index was elevated in the CU group (p = 0.022) (Fig. 2B) while the δ¹³C ratio of linoleic acid to PA was decreased (Fig. 2C). We also measured desaturation indices, which are ratios between 16:0/16:1n-7 or 18:0/18:1n-9 (Table 2). This desaturation index gives an indirect measure of Stearoyl-CoA desaturase-1 (SCD1) (Emken, 1994; Lee et al., 2015), but it also indirectly informs on DNL. The ratio 16:0/16:1n-7 was significantly higher in CU group compared to NCU (p = 0.02), while 18:0/18:1n-9 ratio was similar between groups (p = 0.59). Because the DNL index is only suggestive of lipogenesis and not specific (Rosqvist et al., 2019), we used CSIA to measure the contribution of DNL to PA elevation. CSIA allows to measure the natural abundance carbon isotopes of FA in the body (Lacombe and Bazinet, 2020). This measure has been applied to estimate the contribution of DNL from sugars to nonessential fatty acids (Klingel et al., 2019; Lacombe et al., 2018; Metherel et al., 2019).

Figure 2. Increased lipogenesis in cannabis users.

(A) Graph reporting the de novo lipogenesis index calculated as ratio of 16:0 and 18:2n-6 (concentration nmol/ml) in non-cannabis users and cannabis users. (B) Graph representing the carbon isotopic signature of 16:0, 18:0 and 18:2n-6. (C) Graph representing the ratio between δ¹³C of 16:0 and 18:2n-6. CU, n = 21; NCU, n = 16. Values reported as means ± SEM. Data was analyzed with either Unpaired t-test (16:0, 18:2n-6) or Mann Whitney Test (18:0). * P ≤ 0.05; ** P ≤ 0.005. CU, cannabis users; NCU, non-cannabis users

Thus, CSIA was carried out on FAMEs to determine the natural ¹³C abundance of PA (δ¹³C -PA) but also in other FAs, including STA and LA. Of note, δ¹³C -PA in CU group was significantly higher than in the NCU group (p = 0.0147) and effect that remained statistically significant upon exclusion of those who reported alcohol or tobacco use (data not shown). Importantly, this result was specific for PA, as δ¹³C-STA did not vary between groups, and the essential fatty acid linoleic acid was significantly lower in δ¹³C than in the NCU group (p = 0.0029) (Fig. 2B). Serum levels of PA as well as DNL index positively correlated with it δ¹³C -PA (r = 0.346; p = 0.039 and r = 0.392; p = 0.018, respectively). These results suggested that the increased serum levels of PA in CU are at least partly due to due to DNL. Thus, we explored the potential mechanisms that could lead to increased DNL (see summary Fig.3A).

Figure 3. Changes in sugar levels and in endocannabinoids levels.

(A) Simplified graphical summary of the lipogenic pathway reporting the changes in cannabis users compared to NCU (red arrow). (B) Graphs representing the levels of glucose, pyruvate and lactate in non-cannabis users and cannabis users. (C) Negative correlation between 16:0 and pyruvate. (D) Levels of AEA and 2-AG in the serum of non-cannabis users and cannabis users. (E) Positive correlation between the levels of 16:0 and 2-AG. CU, n = 12–21; NCU, n = 12–16. Values reported as means ± SEM. Data was analyzed with either Unpaired t-test. * P ≤ 0.05; ** P ≤ 0.005. CU, cannabis users; NCU, non-cannabis users.

Initially, we measured serum concentration of glucose, but did not detect any differences between groups (Fig. 3B) nor were glucose and PA levels correlated (data not shown). We also measured serum insulin concentrations (NCU = 24.6µU/ml ± 2.5 vs CU = 35.2µU/ml ± 7) and did not observe any difference between groups nor was it correlated with PA (data not shown). We also analyzed lactate and pyruvate concentrations which are metabolic products of glycolysis and, interestingly, they were significantly lower in the CU compared to the NCU group (Fig. 3B). Additionally, we observed that the levels of lactate and PA were negatively correlated (r = −0.480; p = 0.018) (Fig. 3C).

Stimulation of the CB1 receptor with the two endocannabinoids, AEA and 2-AG can augment DNL (Lichtman and Cravatt, 2005). While AEA did not differ between groups, 2-AG was significantly higher in the CU group (Fig. 3D). Additionally, 2-AG positively correlated with PA levels (r = 0.408; p = 0.012) (Fig. 3E) and δ¹³C-PA (r = 0.289; p = 0.087) although the latter was not statistically significant.

Impact of increased palmitic acid levels on peripheral inflammation.

FA are also implicated in the immune response with the release of mediators of immune response such as cytokines and bioactive lipid mediators. As PA has been associated with an increased immune response (Korbecki and Bajdak-Rusinek, 2019), we evaluated whether the CU group had increased circulating cytokines, but no differences were reported in the levels of IFNγ, TNFα, IL-1β, IL-2, IL-6, IL-8, IL-10 between the two groups (Table 5). Only the level of IL-12 was significantly lower in the CU group compared to the control group (p < 0.05) and also negatively correlated with levels of PA (r = −0.372; p = 0.036).

Table 5.

Cytokines levels in the plasma of non-cannabis users and cannabis users.

Peripheral inflammation
	NCU (n=16)		CU (n=16)		% of difference	t-test
	Mean	SEM	Mean	SEM	CU vs NCU	F	P	Lower	Upper
IFN-g	16.41	2.67	11.30	1.31	−31.16	2.38	0.10	−0.96	11.18
IL-10	7.95	1.31	8.43	1.81	5.99	1.28	0.83	−5.04	4.09
IL-12	4.01	0.46	2.92*	0.27	−27.03	0.94	0.05	0.00	2.16
IL-1b	1.81	0.22	1.40	0.19	−22.56	0.19	0.17	−0.19	1.00
IL-2	3.59	0.58	2.44	0.33	−32.09	4.39	0.10	−0.22	2.53
IL-6	4.02	0.62	4.12	0.95	2.30	0.68	0.94	−2.42	2.24
IL-8	6.07	0.43	6.40	0.67	5.42	2.10	0.68	−1.94	1.29
TNF-a	9.18	0.45	9.76	0.91	6.36	4.71	0.57	−2.65	1.48
CRP	1.32	0.69	1.11	0.61	−16.00	0.13	0.82	−1.67	2.09

Values reported as mean ± SEM. Data was analyzed with either Unpaired t-test or Mann Whitney Test.

^*P ≤ 0.05. CI, confidence interval; CU, cannabis users; NCU, non-cannabis users

Bioactive metabolites of PA, such as palmitoylethanolamide (PEA), an endogenous member of the ECS, have been shown to mitigate the inflammatory response (Bisogno et al., 1997). PEA was significantly increased in the CU group compared to the NCU group (Fig. 4A). Finally, to confirm the selectivity of this result, we also analyzed another ethanolamide, oleoylethanolamide (OEA) and we did not detect any statistical difference between groups (Fig. 4B).

Figure 4. Concentrations of PEA and OEA.

(A) Levels of PEA and (B) in the serum of non-cannabis users and cannabis users. CU, n = 21; NCU, n = 16. Values reported as means ± SEM. Data was analyzed with either Mann Whitney Test. * P ≤ 0.05. CU, cannabis users; NCU, non-cannabis users.

Discussion

With cannabis becoming more available and regulated, research on cannabis is slowly gaining a momentum. While the impact of cannabis on cognitive function and appetite regulation has been the focus of several studies (McDonald et al., 2019; Orr et al., 2020; van Ours and Williams, 2011), for the first time we assessed serum FA composition in a cohort of young long-term CU compared to a group of NCU (Da Silva et al., 2019). Notably, we did not observe differences in total peripheral FA concentrations. A previous study also reported that the total non-esterified FA pool and total cholesterol did not differ between CU and controls (Muniyappa et al., 2013). Similarly, we did not observe any significant differences in total FA concentrations between CU and NCU. We further examined the FA profile in this cohort of volunteers, and we observed an increase in SFA levels, and more specifically in PA, in the CU group when compared to the NCU group. Additionally, levels of MUFA, including palmitoleic acid, another marker of DNL (Johnston et al., 2018), were also significantly higher in the CU group compared to the NCU (see summary Figure 1A). Serum FA reflect dietary fat intake, adipose tissue composition as well as the activity of the enzymes that mediates the synthesis of FA, their elongation and desaturation. More specifically, PA can be acquired from diet or synthesized endogenously. Therefore, we assessed whether markers of DNL were impacted in the CU. Measuring DNL in humans can be challenging as it mostly relies on expensive tracers or liver biopsies (Paglialunga and Dehn, 2016). A less invasive approach is to measure the DNL index (Hudgins et al., 1996), which is the ratio between PA and LA. Changes in newly synthesized FA can indicate the contribution of DNL to post-prandial lipemia (Harding et al., 2015). This index suggested increased DNL in the CU group. While PA is the end product of DNL, it is also the substrate of monounsaturated FA (Moon et al., 2014). Thus, an increase in DNL could alter the desaturation index of plasma FA. We then measured the desaturation index, which is a ratio between either 16:0/16:1n-7 or 18:0/18:1n-9. This index gives an indirect measure of SCD1 (Emken, 1994; Lee et al., 2015), but it also indirectly informs on DNL. As DNL index is only suggestive of DNL (Rosqvist et al., 2019), we decided to further explore this hypothesis by measuring the carbon isotopic signature (δ¹³C) of PA by CSIA. CSIA is based on the principle that the carbon pools from added dietary sugars, namely sugar cane and corn, are more enriched in ¹³C compared to dietary FA, due to differences in the assimilation of carbon in C4 plants (Lacombe and Bazinet, 2020). When PA is enriched in ¹³C, this is indicative of an increased DNL from dietary added sugars, especially those derived from the C4 plants sugar cane and corn (Lacombe and Bazinet, 2020; Metherel et al., 2019). When we performed CSIA we observed that δ¹³C-PA was higher in CU than the NCU, suggesting that DNL at least partly contributes to the increase in PA levels observed. PA can be derived from different sources and, as commercial agriculture uses extensive corn-based feeding, carbon signature from animal products could confound the signal derived either from C4 plant-derived -sweetened beverages (Lacombe and Bazinet, 2020). Thus, we sought to see whether the increase in δ¹³C was selective for PA. To confirm the specificity of this observation, we measured the carbon signature of other FA, including STA and LA. STA, which comes from both diet and elongation of PA did not show any differences in carbon signature. Furthermore, we also assessed selectivity by measuring the signature of LA, which is only derived from diet and showed that LA was less enriched in the CU compared to the NCU group. While our results are still exploratory, and the clinical significance of increased PA in the CU group is not known, it is still quite interesting that the source of PA differs between the two groups. Indeed, our results suggests that the increase in PA is most likely explained by lipogenesis from sugars enriched in ¹³C, nevertheless, our results require confirmation in controlled studies.

Glucose and/or fructose flux through the glycolytic and lipogenic pathways in the liver contribute to DNL, and it has been shown that dietary sugars can stimulate FA synthesis (Parks et al., 2008). Increased consumption of sugar was shown to increase lipogenesis in over-feeding studies (Faeh et al., 2005; McDevitt et al., 2001). Overconsumption of fructose even for a relatively short period of time can elevated lipogenesis (Faeh et al., 2005). Also because of the direct relationship between dietary sugars and post-prandial DNL (Harding et al., 2015), we consequently hypothesized that changes in serum sugar glucose could have an impact on lipogenesis and, more specifically, on PA synthesis. However, serum glucose concentration did not differ between groups, while the two products of glycolysis, pyruvate and lactate that are upstream of PA production were downregulated suggesting that these molecules might be converted into PA more rapidly (see summary Figure 3A). Interestingly, overfeeding of sucrose did not change post-prandial FA concentration (Harding et al., 2015). However, further analyses would be necessary to confirm this hypothesis. Notably, animal studies suggested that ECS also play a role in regulating DNL (Matias et al., 2016). When primary murine adipocytes or hepatocytes were incubated with an agonist of CB1R, DNL was stimulated in a CB1-dependent manner (Cota et al., 2003; Matias et al., 2006; Osei-Hyiaman et al., 2005). Although these observations were made in rodents and not in human, we then assessed the concentration of two peripheral endocannabinoids, namely AEA and 2-AG, which could be implicated in the metabolic changes in the CU group. While AEA did not differ between the groups, 2-AG was elevated in the CU group. Additionally, growing evidence suggests that glucose metabolism is regulated by the ECS in the liver, through the CB1R (Mallat et al., 2013) and antagonists of the CB1R were reported to improve glucose tolerance and insulin resistance. Importantly, current and past cannabis use has been associated with lower levels of fasting insulin and glucose and diabetes mellitus prevalence (Le Strat and Le Foll, 2011; Penner et al., 2013). Increased levels of lactate could be a marker of insulin resistance (Berhane et al., 2015). In our study, CU subjects had a decreased lactate concentration, that could suggest a faster turnover of the metabolite, while insulin concentrations did not differ between the groups.

Maintaining energy homeostasis requires a tight regulation of appetite and peripheral metabolism (Seeley and Woods, 2003; Sharkey, 2006) and cannabis consumption might lead to metabolic dysregulation. Thus, it will be paramount to further understand the impact of increased DNL in this population. For instance, increased hepatic DNL has also been suggested to be one of the underlying causes of metabolic diseases, such as nonalcoholic fatty liver disease (NAFLD) (Lee et al., 2015; Ma et al., 2015; Zammit, 2013), however only few studies assessed NAFLD in cannabis users reporting contradictory results (Adejumo et al., 2017; Hezode et al., 2008; Muniyappa et al., 2013; Vazquez-Bourgon et al., 2019). Furthermore, while the CU were abstinent for about 21 hours, further studies should address the kinetics of increased DNL in CU upon recent exposure or prolonged abstinence. Of note, residual blood THC concentration, sustained even after a period of over four hours of abstinence (Peng et al., 2020), could maintain the activation of the CB1R in the liver and consequently the lipogenic activity. It will be imperative to design further studies to assess both DNL, NAFLD and markers of liver function in CU subjects.

The CU subjects reported in this study were included in a previous report showing increased neuroinflammation as measured by PET (Da Silva et al., 2019). Others also reported increased peripheral pro-inflammatory cytokines concentrations in cannabis users (Bayazit et al., 2017), and there is evidence that some FA can modulate the release of both pro and anti-inflammatory cytokines in vitro (Lee et al., 2003; Simon et al., 2013), thus we evaluated whether the cytokines were different between groups. Unlike a previous study (Bayazit et al., 2017), we did not detect any differences between cytokine levels except for IL-12, a critical cytokine for the development of T helper cells and the cell mediated immune response (Hsieh et al., 1993; Macatonia et al., 1993). Interestingly, most of the cytokines we reported in this study are pro-inflammatory mediators, but they were not significantly different between the groups. Additionally, peripheral cytokines were not significantly associated with central inflammation (Da Silva et al., 2019).

Limitations:

Firstly, this is a secondary analysis of a previous published case-control study (Da Silva et al., 2019; Hafizi et al., 2017; Jacobson et al., 2021; Watts et al., 2020) and thus we cannot infer direct causation. While cannabis use is associated with higher calorie intake (Rodondi et al., 2006; Smit and Crespo, 2001) and could also be associated with a poorer diet quality and increase alcohol consumption (Smit and Crespo, 2001), we did not collect detailed dietary or physical activity habits of the participants in the study and blood was not withdrawn fasting. Thus, the study design does not allow to correct for any of these measures. Being this cohort relatively small, it did not allow to evaluate for covariates, such as sex differences. Additionally, CSIA results suggested that PA is enriched in ¹³C, indicative of an increased DNL from dietary added sugars, especially those derived from the C4 plants sugar cane and corn (Lacombe and Bazinet, 2020; Metherel et al., 2019). The only C4 foods in the North American food supply are corn, sugar cane, millet and sorghum. Millet and sorghum are not popular, while corn is a major source of sugar, and it is also a part of animal feed as grain. Thus, dietary sugars, such as those derived from sugar cane and corn, have higher δ¹³C compared to dietary fat while PA from animal products is not selectively enriched with ¹³C. Thus, we also examined the ¹³C content of other FA. Our results suggests that the increase in PA is derived from DNL, however we do not know the δ¹³C content of all foods in the participants’ diet. Furthermore, we provided some complimentary analyses of lipogenesis as supporting evidence. However, while the finding of a lower δ¹³C in the essential FA, linoleic acid, in the CU group as compared to controls (Fig. 2B) suggests that the dietary δ¹³C content of the diet in the CU group was not universally higher, it does suggest that differences in food intake are likely to exist between the two groups which might confound our interpretation that lipogenesis was increased in the CU group. Thus, future studies should assess or control for the subject diets and compare δ¹³C levels in PA with more direct measures of DNL ideally in prospective or randomized studies. Additionally, as this was a secondary analysis, we could not carry out more established DNL assessments, such as the use of heavy water (Pinnick et al., 2019). Moreover, the measurement of FA composition in the total lipid fraction, as we did, could be skewed by differences in the relative proportions of different lipid fractions. Therefore, it would have been preferable to analyse a single lipid fraction, such as phospholipids, as each lipid fraction has different FA profile, and are differently influenced by diet and metabolism (Brenna et al., 2018; Hodson et al., 2008). Thus, future studies should be designed to isolate the various lipid fractions and perform lipidomic analyses to determine changes in metabolite composition.

Importantly subjects were not fasted, and THC and CBD levels were not available for all the participants, thus correlations between levels of cytokines and fatty acids with these markers might have been underpowered. Finally, correlations between the endocannabinoid system and lipogenesis were only reported in rodent models fed a controlled chow diet rich in carbohydrates. Thus, our study is the first reporting such a correlation in human samples and our observations will need to be validated in future studies. Thus, it is clear that more research needs to be done before conclusions about lipogenesis in CU can be made. Nevertheless, our study is the first to observe increased serum PA in CU using well established methods and the CU groups displayed increased in a variety of markers of lipogenesis.

Conclusion

For the first time we assessed the FA profile in young long-term CU and report that PA is elevated in CU when compared to NCU. The DNL index and CSIA analyses suggested that increased lipogenic activity is present in this population and may be fueled by dietary sugars which may be mediated by CB1R activation or both. While it is not possible to evaluate the metabolic impact of the increased PA and DNL in this cohort, we evaluated the peripheral immune response by measuring cytokines concentrations. Our results suggest that PA elevation did not contribute to an increased inflammatory response in the periphery. Further studies in larger cohorts of fasting CU are required to evaluate whether diet and physical activity can influence the PA levels as well as lipogenesis. Additionally, it will be pivotal to assess whether the activity or the levels of enzymes involved in the DNL pathway are modulated by the use of cannabis and whether similar results are mirrored in other tissues especially the brain. For the first time to our knowledge, our study suggests a new putative role of cannabis use on lipogenesis. Given the rise in legalization and use of cannabis, our study highlights a need for future studies examining the effect of cannabis use on metabolism, especially prospective studies that control for dietary intake.

Abbreviations:

2-AG

2-arachidonoylglycerol

AEA

Anandamide

CBD

cannabidiol

CB1R

cannabinoid receptor 1

CB2R

cannabinoid receptor 2

CSIA

Compound specific isotope analysis

CU

cannabis users

CUD

cannabis use disorder

THC

delta9-tetrahydrocannabinol

DNL

de novo lipogenesis

ECS

endogenous endocannabinoid system

FA

Fatty acids

MUFA

monounsaturated fatty acids

NCU

non-cannabis users

OEA

oleoylethanolamide

PA

palmitic acid

PEA

palmitoylethanolamide

PUFA

polyunsaturated fatty acids

SCD1

Stearoyl-CoA desaturase-1

SFA

saturated fatty acids