Age differences in endocannabinoid tone are ameliorated after recent cannabis use

Abstract

An age-related decline in endocannabinoid system (ECS) activity may contribute to conditions such as chronic pain and Alzheimer’s disease. Although cannabis is increasingly used by older adults to alleviate age-related conditions, it remains unclear how cannabinoids affect ECS activity across the lifespan. The present study assayed levels of seven endocannabinoids (AEA, 2-AG, DEA, LEA, PEA, SEA, and OEA) in a sample of adults (N = 142; younger 21–24 years, n = 38; midlife 25–54, n = 73; older 55–71, n = 31) assayed before cannabis use (baseline [pre-use]) and ~ 1 h after flower or ~ 2 h after edible cannabis use. At baseline, older adults exhibited lower AEA and DEA than younger adults, and lower LEA than midlife adults. Acute cannabis use increased AEA, DEA, LEA, PEA, SEA, and OEA across all age groups (all p < .001). 2-AG showed no increase. For AEA and DEA, increases were larger in older adults (Time×Age). These findings indicate broad endocannabinoid elevations after cannabis use regardless of age, alongside age-related differences at baseline and in acute responses.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-27618-1.

Introduction

The endocannabinoid system (ECS) comprises endocannabinoids, their receptors (CB1/CB2), and the enzymes that regulate ligand synthesis and degradation. The principal endocannabinoids are anandamide (AEA) and 2-arachidonoylglycerol (2-AG). AEA belongs to the N-acylethanolamine (NAE) family, which also includes docosatetraenylethanolamide (DEA), linoleoylethanolamide (LEA), palmitoylethanolamide (PEA), stearoylethanolamide (SEA), and oleoylethanolamide (OEA). Whereas AEA and 2-AG engage CB1/CB2, several NAEs act primarily at PPAR-α^1–3. For brevity, we refer to AEA and 2-AG as endocannabinoids and to DEA, LEA, PEA, SEA, and OEA as NAEs.

ECS signaling supports neural, immune, and metabolic homeostasis^4,5. Reduced circulating AEA has been linked to elevated anxiety⁶, PTSD symptoms⁷, and major depressive disorder⁸, and reduced 2-AG correlates with greater depressive symptom burden under stress exposure⁹. Among NAEs, PEA and OEA are the most extensively studied, with analgesic¹⁰, neuroprotective^11,12, and anti-inflammatory¹³ properties. Supplementation strategies that raise systemic PEA/OEA have been evaluated for inflammation¹⁴, chronic pain¹⁵, neurodegeneration^16,17, and mood disorders^18,19. DEA, LEA, and SEA show emerging anti-inflammatory^20–22 and neuroprotective signals²³, with SEA also implicated in pain modulation²⁴, although these effects are less well characterized. Several NAEs (notably OEA, SEA, and LEA) also support metabolic homeostasis through effects on satiety^25,26, weight²², and lipid profiles^22,26. Collectively, these observations motivate evaluation of age differences in circulating endocannabinoids/NAEs and their acute modulation by cannabis.

Exogenous cannabinoids can modulate ECS activity²⁷. Δ9-tetrahydrocannabinol (THC) is a partial CB1/CB2 agonist with well-described psychoactive and cognitive effects, whereas cannabidiol (CBD) shows low CB1/CB2 affinity but may indirectly influence ECS signaling, including by limiting eCB catabolism²⁸. Beyond receptor pharmacology, cannabis use —via THC and CBD— has been associated with reduced inflammatory signaling, with some evidence that changes in endocannabinoids/NAEs contribute to these effects^29,30.

Converging evidence indicates that eCB “tone” declines with age, beginning in midlife^31,32. In the brain, 2-AG decreases in the hippocampus, while AEA remains relatively stable in the hippocampus but declines in other regions (e.g., caudate putamen, medial prefrontal cortex)³¹. Aging is also associated with ~ 50% reductions in CB1 mRNA expression and binding in cortex³³. Mechanistically, polyunsaturated fatty-acid–derived endocannabinoids (e.g., AEA, DEA) may decline with intrinsic biological aging, reflecting broader shifts in lipid metabolism and enzyme/receptor function^31–36. In contrast, several NAEs appear more sensitive to diet and metabolic health: PEA derives from palmitic acid (saturated), SEA from stearic acid (saturated), OEA from oleic acid (monounsaturated), and LEA from linoleic acid (polyunsaturated), all influenced by dietary fat intake^37–43. These distinctions suggest that age can differentially affect specific lipids within the endocannabinoid/NAE network.

Age-related changes in the ECS have clinical relevance for domains central to healthy aging, such as mood, sleep, pain, and inflammation. An age-related attenuation of ECS activity might partly explain rising cannabis use among older adults⁴⁴, who often report using cannabis for symptom management⁴⁵. Given evidence for age-related declines in endocannabinoid/NAE tone, and the potential for cannabinoids to acutely modify circulating endocannabinoids/NAEs, it is important to determine whether age moderates these circulating lipid signals at baseline and after real-world cannabis use.

To our knowledge, no human study has tested age-related differences in circulating endocannabinoids/NAEs together with age-dependent acute responses to legal-market cannabis products. We therefore pursued two aims. First, we compared baseline plasma concentrations of seven analytes (AEA, 2-AG, DEA, LEA, PEA, SEA, OEA) across three age groups: younger (21–24 years), midlife (25–54 years), and older (55–71 years). We hypothesized lower baseline endocannabinoid/NAE levels in older adults. Second, we assessed whether acute ad libitum use of flower or edible cannabis (THC-dominant, CBD-dominant, or balanced) altered circulating endocannabinoids/NAEs and whether these changes differed by age group. Understanding which lipids are most sensitive to aging — and whether cannabis acutely shifts those signals differentially with age — might inform ECS-targeted approaches relevant to late-life health.

Methods

Participants and study design

This secondary analysis used data from two prior, non-overlapping investigations of cannabis flower⁴⁶ (n = 139) and edible products⁴⁷ (n = 52) with similar eligibility criteria and study designs. All participants provided written informed consent; protocols were approved by the University of Colorado Boulder IRB, and procedures adhered to relevant guidelines and regulations.

Recruitment was conducted via social media advertising, followed by eligibility screening over the phone. Participants were 21–71 years old, reported cannabis use ≥ 4 times in the past month, denied illicit drug or tobacco use, and reported alcohol use ≤ 2 times/week (≤ 3 drinks/occasion). Eligible participants completed an in-person baseline visit (questionnaires, cognitive tasks, venipuncture). Participants were instructed to refrain from cannabis prior to the baseline blood draw; however, time since last cannabis use was not recorded, and abstinence was not biochemically verified. Participants were randomized to a product that they then purchased from a licensed dispensary and used ad libitum for five days. The study products varied in THC/CBD potency. Flower products, which could be administered by the participants preferred method (for example pipe, bong, bubbler, joint/blunt, flower vape etc.), were THC-dominant (24% THC, 1% CBD), CBD-dominant (23% CBD, 1% THC), or balanced (9% THC, 10% CBD). Edibles were THC-dominant (10 mg THC, 0 mg CBD), CBD-dominant (25 mg CBD, 1.65 mg THC), or balanced (5 mg THC, 5 mg CBD). A follow-up visit occurred on the fifth day in a mobile pharmacology laboratory. Additional details on eligibility criteria, study design, and primary results have been published elsewhere^46,47.

For the present analysis, we restricted all inferences to participants with both baseline (pre-use) and post-use plasma obtained at the mobile visit (post-use sampling 1 h after flower use; 2 h after edible use). The complete-case analytic sample comprised N = 142: younger 21–24 y (n = 38), midlife 25–54 y (n = 73), and older 55–71 y (n = 31). All results reported in the main text use this analytic sample.

Endocannabinoid assays

Endocannabinoid/NAE quantification followed prior methods described by our group^48,49. K2EDTA blood samples were collected via venipuncture, and plasma was isolated through centrifugation. Plasma samples were acidified according to established protocols to preserve endocannabinoid integrity. Quantification was performed using a Sciex API 5500 + mass spectrometer (AB Sciex LLC, Framingham, MA) coupled with an Agilent 1260 HPLC system (Agilent Technologies, Palo Alto, CA). Seven endocannabinoids were measured:]AEA, 2-AG, DEA, LEA, PEA, SEA, and OEA. For details regarding chromatographic conditions, instrument settings, and quantitation procedures, see Sempio et al.⁴⁹.

Statistical analysis

Analyses were restricted to the complete-case sample with both baseline and post-use plasma (post-use 1 h after flower; 2 h after edible). Continuous baseline characteristics were compared across age groups by one-way ANOVA and categorical variables by χ² tests.

Endocannabinoid concentrations were right-skewed so values were natural-log–transformed prior to multivariate modeling. Distributional assumptions were examined with Shapiro–Wilk tests and Q–Q plots and homogeneity of covariance was assessed with Box’s M. Pillai’s trace was prespecified for multivariate tests. Two-sided p <.050 was considered significant. When multiple univariate tests were conducted across the seven analytes, Benjamini–Hochberg FDR control was applied. Effect sizes are reported as partial η² (η²p). Analyses were conducted in SPSS v29.0.2.0 (IBM); figures were generated in Jamovi v2.6.45.0.

Baseline differences

A one-way MANOVA tested age-group effects (younger, midlife, older) on the seven ln-transformed outcomes (AEA, 2-AG, DEA, LEA, PEA, SEA, and OEA). Significant multivariate effects were followed by univariate ANOVAs for each analyte.

Acute change in endocannabinoids/NAEs after cannabis use

For each analyte, we fit a two-way mixed ANOVA with Time (baseline, post-use; within-subjects) and Age group (between-subjects) on ln-transformed concentrations. Primary inferences focused on the Time main effect and the Time×Age group interaction. Within-group changes were summarized as ratios (post/pre-use).

Sensitivity analyses

In prespecified sensitivity analyses, we added cannabis Form (flower or edible) and Composition (THC dominant, CBD dominant, or balanced) as between-subjects factors, and centered BMI, years of cannabis use, age of first use, and The Cannabis Use Disorder Identification Test - Revised (CUDIT-R) as continuous covariates. For mixed models, we also tested Time×covariate terms. Effect sizes are η²p. Full sensitivity outputs are reported in Supplementary Tables S1–S2.

Results

Sample characteristics

Group distributions by sex, ethnicity, and study condition did not differ by age (all p ≥.100); race distribution was marginal (p =.053). Younger adults had higher CUDIT-R scores; older adults reported later age of first cannabis use and more years of use. BMI did not differ significantly (Table 1).

Table 1.

Reflects the complete-case analytic sample (n = 142). CUDIT-R, the cannabis use disorder identification Test – Revised. Lifetime years of cannabis use = current age − age of first use (abstinence periods not subtracted). *Significant omnibus ANOVA p <.050; p =.001; p <.001 among age groups. Pairwise comparisons (Tukey) significant at p <.050 are denoted: a younger vs. older, b midlife vs. older, c younger vs. midlife. #Average daily use for each product type was derived from the 30-day timeline Follow-Back (TLFB) collected at baseline (self-report questionnaire).

Characteristic [% or Mean (SD)]	Younger	Midlife	Older
	(n = 38)	(n = 73)	(n = 31)
Age*** a, b,c	22.10 (1.09)	33.90 (8.41)	62.70 (4.66)
Body Mass Index (BMI)	22.09 (3.70)	24.40 (4.95)	24.50 (3.56)
Male (%)	63%	56%	58%
White (%)	76%	77%	93%
Not Hispanic or Latino (%)	87%	92%	97%
Beck Depression Inventory-II (BDI-II)	6.87 (6.45)	8.58 (9.34)	6.39 (9.66)
Beck Anxiety Inventory (BAI)	6.21 (5.33)	7.90 (9.56)	4.30 (5.07)
Age of First Cannabis Use*** a, b	18.00 (2.45)	20.20 (6.83)	29.90 (17.60)
Lifetime Years of Cannabis Use*** a, b,c	4.08 (2.44)	13.60 (8.82)	43.50 (19.40)
CUDIT-R Score** a, c	3.70 (2.85)	2.28 (2.25)	1.45 (2.42)
Average Times Per Day Using Cannabis#
Flower	3.65 (2.61)	3.46 (2.07)	2.83 (1.83)
Edible	1.03 (0.67)	1.09 (0.59)	1.27 (0.70)
Concentrate	2.83 (2.85)	3.45 (2.17)	2.50 (2.01)

Baseline endocannabinoid concentrations

The endocannabinoid concentrations were right-skewed so we performed our analysis on ln-transformed values. A one-way MANOVA indicated a multivariate age effect across the seven analytes (Pillai’s trace = 0.241; F(14,268) = 2.629; p =.001). In a prespecified MANCOVA adjusting for BMI, years of cannabis use, CUDIT-R, and age of first use, the multivariate age effect remained significant (Pillai’s trace = 0.262; F(14,242) = 2.610; p =.002). FDR-controlled univariate ANCOVAs (across seven analytes) showed age-group differences for AEA (p =.015, q < 0.05), DEA (p =.005, q < 0.05), and LEA (p =.007, q < 0.05); PEA was not significant (p =.103). 2-AG, SEA, and OEA were not significant (all q ≥ 0.073). BMI showed a modest positive association with baseline DEA (p =.025). Directionally, older adults had lower baseline AEA and DEA than younger adults, and lower LEA than midlife adults (Fig. 1).

Fig. 1.

Baseline (pre-use) plasma endocannabinoid concentrations by age group.

Boxplots of ln(concentration [ng/mL]) for AEA, DEA, and LEA by age group (younger n = 38; midlife n = 73; older n = 31). Boxes show IQR with medians; whiskers 1.5×IQR; circles are individuals; squares denote means. One-way MANOVA across seven ln-transformed analytes: Pillai’s trace = 0.241, F(14,268) = 2.629, p =.001. FDR-controlled univariate ANCOVAs (adjusted for BMI, years of cannabis use, CUDIT-R, age of first use) showed age-group differences for AEA (p =.015, q < 0.05), DEA (p =.005, q < 0.05), and LEA (p =.007, q < 0.05). Pairwise Tukey tests: older < younger for AEA and DEA; older < midlife for LEA (FDR-adjusted q < 0.05).

Endocannabinoid response following acute cannabis use

Across the analytic sample, NAEs increased from baseline to post-use. On the ln scale, mean within-participant changes (Δln) corresponded to post/pre ratios of 1.728 (AEA), 1.637 (DEA), 1.562 (LEA), 1.419 (PEA), 1.290 (SEA), and 1.537 (OEA) (all paired t-tests p <.001). 2-AG corresponded to a post/pre ratio of 0.782 (p =.013), without an age interaction.

Age-differential responsivity was observed for AEA and DEA. In unadjusted two-way mixed ANOVAs, the Time×Age group interaction was significant for AEA and for DEA (both q = 0.021 after FDR across analytes), with larger increases in older versus younger adults. Group-specific post/pre-use ratios were: AEA—younger 1.410 (95% CI 1.206–1.648), midlife 1.796 (1.610–2.003), older 2.026 (1.712–2.397); DEA—younger 1.409 (1.213–1.636), midlife 1.638 (1.501–1.786), older 1.964 (1.691–2.281). After adjustment for BMI, years of use, and CUDIT-R, the Time×Age group interaction was attenuated and not significant for DEA (p =.152); the ordered pattern of means persisted (Fig. 2).

Fig. 2.

Acute changes in plasma endocannabinoid levels at baseline (pre-use) and post-cannabis use.

Mean (± SEM) ln(concentration [ng/mL]) for AEA and DEA at baseline (pre-use) and post-use (~ 1 h after flower; ~2 h after edible) in the analytic sample (younger n = 38; midlife n = 73; older n = 31). Two-way mixed ANOVAs (within-subjects Time; between-subjects Age group) showed robust Time effects for both analytes (p <.001). Time×Age group interactions were significant in unadjusted models for AEA and DEA (FDR q = 0.021 for both); after adjustment for BMI, years of use, and CUDIT-R, the interaction was nominal for AEA (p =.008, q = 0.058) and not significant for DEA (p =.152).

Product characteristics and additional sensitivity analyses

Adding Form (flower/edible), Composition (THC/CBD/balanced), and their interactions with Time, and including centered BMI, years of use, age of first use, and CUDIT-R as covariates did not alter inferences for the Time or Time×Age group effects. Nominal Time×Composition and Time×Composition×Form effects were observed for DEA (p =.022 and p =.009, respectively), but these did not survive multiplicity control across analytes and interaction terms (Tables S1–S2).

Discussion

In the present study, older adults (55–71 y) showed lower baseline plasma AEA, DEA, and LEA than younger adults (21–24 y), whereas 2-AG, PEA, SEA, and OEA did not differ by age. These findings align with preclinical evidence for age-related reductions in ECS signaling (e.g., CB1 expression and eCB tone)^31,32,35,37. Following ad libitum cannabis use, NAEs (AEA, DEA, LEA, PEA, SEA, OEA) increased across all age groups, while 2-AG did not. The Time×Age group interaction for AEA and DEA was significant in unadjusted models, indicating larger increases in older adults; after adjustment for BMI, years of use, and CUDIT-R, the interaction remained nominal for AEA and was not significant for DEA, but the ordered pattern persisted. Together, these results indicate age differences in endocannabinoid tone at baseline and broadly similar acute NAE increases after cannabis across the lifespan, with some evidence for greater AEA/DEA responsivity in older adults.

The selective baseline differences are biologically plausible. AEA and DEA (PUFA-derived) may track intrinsic aging processes (e.g., shifts in lipid metabolism and ECS enzymatic activity), whereas some saturated/monounsaturated NAEs (e.g., PEA, SEA, OEA) may be more diet-sensitive and thus less age-dependent^{31–35,38–43}. By contrast, 2-AG — although arachidonic acid-derived — can be tightly regulated by acute physiological state and therefore might not show parallel age effects at rest (Fig. 3)⁵⁰. These mechanistic considerations are inferential and require targeted studies.

Fig. 3.

Aging and its associated changes—such as diet, metabolism, cellular function, inflammation, and oxidative stress—affect the bioavailability of fatty acid precursors and lipid metabolism involved in endocannabinoid (EC) synthesis and degradation³¹. Polyunsaturated fatty acids (PUFAs) like arachidonic acid (AA) and docosahexaenoic acid (DHA), and saturated fatty acids (FAs) like palmitic acid (PA), linoleic acid (LA), and stearic acid (SA), are converted into N-acyl phosphatidylethanolamines (NAPE) by N-acyltransferase (NAT). NAPE is hydrolyzed by N-acyl phosphatidylethanolamine phospholipase D (NAPE-PLD) to produce N-acylethanolamines (NAEs) such as anandamide (AEA) and docosatetraenylethanolamide (DEA), targeting cannabinoid receptors CB1 and CB2. This process also produces palmitoylethanolamide (PEA), linoleoylethanolamide (LEA), and stearoylethanolamide (SEA), which primarily target PPAR-α receptors². PUFA precursors (AA and DHA) decline with age, with DHA reduction linked to cognitive decline and other age-related diseases³¹. Conversely, saturated FA precursors (PA, LA, and SA) remain relatively stable and are more influenced by diet and metabolic health than aging^37,41. Cannabinoids such as THC and CBD may alter lipid profiles through interactions with enzymes (e.g., FAAH, NAPE-PLD) and transport proteins (e.g., FABPs)^51–54 of the endocannabinoid system that regulate N-acylethanolamine (NAE) levels, as well as through CB1-mediated homeostatic regulation^53,54.

Product characteristics did not account for the findings. Neither form (flower or edible) nor composition (THC, CBD or balanced), nor their interaction, altered primary age effects or acute changes after multiplicity control. Exploratory effects involving Sex and product characteristics (e.g., Time×Composition×Sex) were small (η²p ≤.005), inconsistent across analytes, and outside the scope of the present aims. Full results are provided in Supplementary Tables S1–S2 and warrant targeted follow-up.

Strengths include standardized blood collection at two time points, a complete-case analytic cohort, and multivariate control of Type I error. Important limitations warrant emphasis. First, the cross-sectional design limits causal inference; age bins (21–24, 25–54, 55–71) are coarse and should be complemented by continuous-age analyses in future work. Second, time since last cannabis use was not ascertained and abstinence at baseline was not biochemically verified; thus, “baseline (pre-use)” may include residual effects of recent use. Third, baseline (laboratory) versus post-use (field) settings may introduce minor environmental variability. Fourth, dispensary-sourced products and ad libitum dosing improve ecological validity but reduce experimental control. Finally, missing post-use 2-AG data modestly reduced power for that analyte.

In conclusion, circulating AEA, DEA, and LEA are lower in older vs. younger adults at baseline, and acute cannabis use elevates multiple NAEs irrespective of age, with suggestive age-related differences in AEA/DEA responsivity. These observations highlight heterogeneity within the ECS across aging and support further controlled, randomized, placebo-matched studies with standardized cannabinoid dosing, verified abstinence, non-user/comparator groups, and longitudinal follow-up to determine clinical significance.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Alan W. J. Morris performed writing - original draft, writing - review & editing, methodology, and formal analysis. Raeghan L. Mueller performed writing - review & editing, investigation, project administration, and methodology. Cristina Sempio performed investigation, validation and methodology. Jost Klawitter performed investigation, validation, methodology, writing - review & editing, and provided resources. Angela D. Bryan performed writing - review & editing. L. Cinnamon Bidwell performed writing - review & editing. Kent E. Hutchison supervised the project, provided funding for the project and performed conceptualization, project administration, and writing - review & editing.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.