Effects of simulated exogenous and endogenous lipid microenvironments on cannabidiol disposition and cytochrome P450 2C19–mediated drug interactions: New mechanistic insights

Abstract

Cannabidiol (CBD) is an in vitro inhibitor of multiple cytochrome P450 (P450) enzymes, which has been confirmed in healthy adults. Whether these observations translate to different patient populations, including those with hepatic steatosis, is not known. Accordingly, the effects of simulated exogenous and endogenous lipid microenvironments on the P450 inhibition potential and disposition of CBD were evaluated using human liver microsomes (HLMs). Consistent with clinical observations, CYP2C19 was most sensitive to inhibition by CBD as both a component of a mixture containing triglycerides (CBD_m) and in purified form (CBD_p), with a K_i of 0.22 ± 0.03 and 0.16 ± 0.03 μM, respectively. CYP2C19 activity (S-mephenytoin 4′-hydroxylation) decreased by 25% ± 3.5% and 68% ± 0.4%, respectively, upon exposure to triglycerides and fatty acids under simulated hepatic steatotic conditions. Triglycerides and fatty acids enhanced the CYP2C19 inhibitory effects of CBD_p by 1.3- to 3-fold, respectively. The IC₅₀ for CBD_p against CYP2C19 activity in HLMs from donors with and without metabolic dysfunction–associated steatohepatitis (MASH) was 1.28 ± 0.25 and 1.12 ± 0.23 μM, respectively. Intrinsic clearance of CBD_p via P450 and UDP-glucuronosyltransferase pathways in MASH HLMs (0.33 ± 0.02 and 0.15 ± 0.02 mL/min per mg, respectively) was lower compared with non-MASH HLMs (0.73 ± 0.03 and 0.18 ± 0.04 mL/min per mg, respectively). Results from this collection of in vitro experiments provide foundational knowledge that informs future research regarding the effects of a high-fat diet and steatotic liver disease on CBD disposition and CYP2C19-mediated drug interactions.

Significance Statement

Exposure of human liver microsomes to simulated exogenous and endogenous lipid microenvironments provided new mechanistic insight into the direct effects of triglycerides and fatty acids on cannabidiol disposition and CYP2C19-mediated cannabidiol-drug interactions. Modified in vitro hepatic models, including human liver microsomes and human hepatocytes, under hepatic steatotic and metabolic dysfunction–associated steatohepatitis conditions could be applied to future studies aimed at evaluating the disposition and drug interaction potential of other drugs in these specific populations.

1. Introduction

The prevalence of cannabis use continues to rise worldwide, particularly in countries where legalization policies for recreational and/or medical purposes have been implemented.¹ The 2 main types of cannabis products are marijuana and hemp, both of which are derived from different varieties of the Cannabis sativa plant. Marijuana use in the United States remains federally illegal, but most states have legalized such use at least for medical purposes, leading to increased marijuana use.² Passage of the federal 2018 Farm Bill,³ which authorized the production of hemp and removal of hemp from the Drug Enforcement Administration’s list of controlled substances, led to an exponential increase in hemp and other cannabidiol (CBD)–containing products available over the counter.

C. sativa contains >100 cannabinoids,⁴ with CBD and Δ-9-tetrahydrocannabinol (THC) being the most extensively studied. Marijuana and hemp products differ in the amount of CBD present, with those qualifying as hemp containing ≤0.3% THC by dry weight. Unlike THC, CBD does not produce euphoric effects. Consequently, consumers often turn to hemp products to achieve perceived health benefits while avoiding the side effects associated with THC. Such products are commonly used to self-treat various conditions, including anxiety, depression, insomnia, and pain.⁵

The rising use of hemp products raises concerns for pharmacokinetic CBD-drug interactions, particularly for individuals taking multiple medications, including older adults and patients with chronic illnesses. Both CBD and THC have been reported to be potent inhibitors (K_i or IC₅₀ <1 μM) of multiple cytochrome P450 enzymes using human liver microsomes (HLMs).^6,7 Our recent clinical pharmacokinetic study involving 18 healthy adults showed that a CBD-dominant extract baked into a brownie (640 mg CBD, 20 mg THC), but not a THC-dominant extract (20 mg THC, no CBD), increased the area under the plasma concentration versus time curve (AUC) of 4 P450 probe drugs—caffeine (CYP1A2), losartan (CYP2C9), omeprazole (CYP2C19), and midazolam (CYP3A)—by at least 1.4-fold relative to baseline, with CYP2C19 being the most sensitive (3-fold increase).⁸ Whether these observations with CBD translate to other populations, including patients with hepatic steatosis or metabolic dysfunction–associated steatohepatitis (MASH), is not known.

Hepatic steatosis, a chronic liver disease marked by excessive fat accumulation in liver cells, can progress, reversibly, to the advanced stage, MASH, characterized by cellular inflammation. The global prevalence of hepatic steatosis in the adult population ranges from 23% to 32%, with MASH affecting 5% of the population, and is projected to increase to 50% by 2050.^9,10 Contributing factors include poor dietary control and a sedentary lifestyle, genetic predispositions, and associated comorbidities, including diabetes, obesity, and cardiovascular disease. Lipids in the liver consist of a mixture of triglycerides (TGs) and fatty acids (FAs),^11,12 with the latter serving as both building blocks and breakdown products of the former. FAs, with palmitic, oleic, and stearic acid being the most abundant,¹³ have been shown to alter P450 activity both in vitro and in mice.^14,15 Because CBD is highly lipophilic (LogP, 6.3), accumulation of TGs and FAs may impact the P450 inhibitory effects of CBD under steatotic conditions. Specifically, high concentrations of TGs and FAs indicative of high-fat meals and steatotic liver disease may increase the lipid partitioning of CBD, leading to reduced P450 inhibitory effects.

High TG content is indicative of not only high-fat meals but also of carrier oils in oil-based botanical products, including hemp and other CBD-containing products. Such oils are used to enhance CBD solubility in these products, which contain other constituents (eg, terpenes and cannabinoids) that may contribute to the drug interaction potential of the complex mixture. These observations raise the question: does purified CBD (CBD_p) behave similarly to CBD in the complex mixture (CBD_m)?

Based on the collective observations, the first objective of the current work was to evaluate whether CBD_p is equivalent to CBD_m as a P450 inhibitor. The second objective was to simulate the effects of various exogenous and endogenous lipid microenvironments on the hepatic P450 inhibitory effects and intestinal permeability of CBD. The aims were to (1) compare the effects of CBD_m and CBD_p as inhibitors of CYP2C19 activity (based on screening results) in HLMs, (2) evaluate the effects of CBD_p on CYP2C19 activity in HLMs exposed to select TGs and FAs, and (3) elucidate the membrane permeability of CBD upon exposure of Caco-2 cell monolayers to dietary fats (Fig. 1). Results provide foundational knowledge for future mechanistic studies to elucidate the effects of the microenvironment of hepatic steatosis on P450 inhibition by CBD, as well as inform clinical studies involving patients with steatotic liver disease.

Fig. 1.

Workflow for evaluating effects of simulated exogenous (lipids, matrix oil) and endogenous (FAs, TGs) lipid microenvironments on the CYP2C19 inhibitory effects and disposition of CBD.

2. Materials and methods

2.1. Materials and chemicals

HLMs pooled from 15 donors (male and female) without MASH (non-MASH HLMs) (cat #H0604) and from 5 donors (male and female) with MASH (MASH HLMs) (cat #HHNSH) were purchased from Xenotech, LLC (now a BioIVT company). Tributyrin (cat #HY-W011404) and palmitodiolein (cat #HY-112132) were purchased from MedChemExpress, LLC. Medium-chain TG (MCT) softgels were purchased from Panacea Life Sciences. Bovine serum albumin (BSA; cat #A4737), magnesium chloride hexahydrate (cat #M2670), uridine 5′-diphosphoglucuronic acid trisodium salt (cat #U6751), acetaminophen (cat #A7085), phenacetin (cat #77440), sodium oleate (cat #O7501), sodium palmitate (cat #P9767), and tolbutamide (cat #T0891) were purchased from Sigma Aldrich. Alprazolam (cat #14255), atenolol (cat #26284), CBD (cat #90080), CBD-d₉ (cat #16203), dextromethorphan (cat #11157), dextrorphan (cat #15886), midazolam (cat #16193), S-mephenytoin (cat #11913), 4′-hydroxymephenytoin (cat #16124), ticlopidine (cat #20770), 4-hydroxytolbutamide (cat #10008520), oleic acid (cat #90260), palmitic acid (cat #10006627), and triolein (cat #26871) were purchased from Cayman Chemical. 1′-Hydroxymidazolam (cat #H948420) and 4′-hydroxymephenytoin-d₃ (cat # H944877) were purchased from Toronto Research Chemicals. β-Nicotinamide-adenine dinucleotide phosphate tetrasodium salt (NADPH; cat #481973) was purchased from EMD Millipore. Acetonitrile Optima liquid chromatography/mass spectrometry (LC/MS)-grade (cat #A955), formic acid Optima LC/MS-grade (cat #A1171), and penicillin-streptomycin (10,000 U/mL; cat #15140-122) were purchased from Fisher Scientific. Minimum essential medium (cat #10-009-CV) was purchased from Corning. FBS (cat #30-2020) and trypsin-EDTA solution 1× (cat #30-2101) were purchased from American Type Culture Collection. All other chemicals and reagents were purchased from Sigma Aldrich.

A “broad-spectrum” hemp oil product (lot number 4230400) was purchased from CBDistillery. This product was selected for a clinical pharmacokinetic study aimed to evaluate the interaction between a hemp product high in CBD with trace THC and clopidogrel (ClinicalTrials.gov identifier: NCT 06692933). The quantity of CBD and THC in the capsule was measured using ultraperformance LC coupled to tandem MS (UPLC-MS/MS). A detailed description of the methods is provided in the Supplemental Material. The product was in the form of softgel capsules containing fractionated coconut oil, which is composed of MCTs. The oil contained 0.0702 ± 0.0095 mg CBD/mg oil; THC was below the limit of quantification (0.00042 ± 0.00027 mg THC/mg oil).

2.2. Screening of CBD_m in the hemp oil product as an inhibitor of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A activity

CBD_m was screened as an inhibitor of the 5 P450s using a cocktail approach with modifications.^16,17 In brief, CBD_m (2, 20, and 200 μg/mL hemp oil containing 0.45, 4.5, and 45 μM CBD, respectively) or the comparator, CBD_p (5 μM), was equilibrated with non-MASH HLM (1 mg/mL), BSA (0.2%), and potassium phosphate buffer (KP_i) (0.1 M, pH 7.4) at 37 °C for 5.5 minutes. BSA was added to the incubation mixture to minimize nonspecific binding and increase CBD solubility.⁷ Reactions were initiated with NADPH (1 mM final concentration). At specific time points (0–30 minutes), an aliquot was diluted 10-fold into a secondary incubation mixture containing NADPH (1 mM) and a cocktail^16,17 consisting of phenacetin (CYP1A2, 100 μM), tolbutamide (CYP2C9, 100 μM), S-mephenytoin (CYP2C19, 50 μM), dextromethorphan (CYP2D6, 5 μM), and midazolam (CYP3A, 2 μM). After 10 minutes, reactions were quenched with ice-cold methanol containing the internal standard alprazolam (50 nM). The supernatants were quantified for corresponding metabolites by UPLC-MS/MS (described briefly in Quantification of CYP Probe Metabolites, CBD, Midazolam, and Atenolol and in the Supplemental Material). Percent remaining activity was calculated as the concentration of metabolite relative to vehicle control (methanol, <1% v/v). Statistical comparisons among the 3 concentrations of CBD_m at 0 minutes for concentration-dependent inhibition and from 0 to 30 minutes for time-dependent inhibition were made using two-way ANOVA with Tukey's multiple comparisons test in GraphPad Prism (version 10). Comparisons between CBD_m (20 μg/mL) and CBD_p (5 μM) were made using one-way ANOVA with Tukey's multiple comparisons test. P values of <.05 for all tests were considered statistically significant.

2.3. General incubation conditions for reversible inhibition of CYP2C19 activity

The P450 inhibition screening results showed that CYP2C19 was most sensitive to the inhibitory effect of both CBD_m and CBD_p, prompting comparison of the reversible inhibition kinetics of CBD_m and CBD_p. Briefly, S-mephenytoin and inhibitor (CBD_m, CBD_p, or matrix oil) were equilibrated with HLMs (0.1 mg/mL), BSA (0.2%), and KP_i at 37 °C for 5.5 minutes. The canonical CYP2C19 inhibitor ticlopidine (3 μM) was used as a positive control. Details of inhibitor and substrate concentrations and testing conditions are provided below. Reactions were initiated with NADPH (1 mM) and, after 20 minutes, were quenched with ice-cold methanol containing 4′-hydroxymephenytoin-d₃ (100 nM). At least 3 different experiments were conducted. The supernatants were quantified for corresponding metabolites by UPLC-MS/MS (see Quantification of CYP Probe Metabolites, CBD, Midazolam, and Atenolol).

2.3.1. CYP2C19 activity in the presence of select triglycerides

The selected TGs represented 3 different compositions: short-chain (tributyrin), MCT, and long-chain (palmitodiolein). A range of S-mephenytoin (5–60 μM) and MCT (0–45 μg/mL) concentrations, along with the 3 TGs at a single concentration (45 μg/mL), were tested. The highest concentration (45 μg/mL) of TGs corresponded to the matrix oil (MCT) concentration in the hemp oil product containing 10 μM CBD; HLMs were assumed to be exposed to TGs to a comparable extent as CBD. The percent change in 4′-hydroxymephenytoin concentration was compared with vehicle control (methanol <1% v/v). Statistical comparisons among the 3 concentrations of MCT oil, the 3 concentrations of S-mephenytoin, and the 3 selected TGs at one concentration were made using one-way ANOVA with Tukey's multiple comparisons test (GraphPad Prism); P values of <.05 for all tests were considered statistically significant.

2.3.2. Reversible inhibition kinetics of CBD_m and CBD_p against CYP2C19 activity

A 5 × 6 matrix of S-mephenytoin concentrations (5–100 μM) and inhibitor concentrations (CBD_m [0–45 μg/mL] or CBD_p [0–10 μM]) was tested. CBD_m refers to CBD in hemp oil. The effects of the 3 selected TG compositions on the CYP2C19 inhibitory effect of CBD were next tested using an abbreviated assay consisting of a 4 × 4 matrix of S-mephenytoin (5–60 μM) and CBD_p (0–1 μM) dissolved in tributyrin, MCT, or palmitodiolein; the final mixture (CBD_p + TGs) concentration mimicked the concentration of CBD_m in hemp oil. Inhibition kinetic parameters were recovered as described^18,19 via nonlinear least-squares regression analysis using Phoenix WinNonlin (version 8.6.1.6; Certara) and either of the following equations that best fit a given dataset based on standard diagnostics:

$$\text{Competitive}\text{inhibition}:v=\frac{V_{\max }\times \left[S\right]}{K_m\left(1+\frac{\left[I\right]}{K_{\mathrm{i}}}\right)+\left[S\right]}$$ (1)

$$\text{Mixed}\text{inhibition}:v=\frac{V_{\max }\times \left[S\right]}{K_{\mathrm{m}}\left(1+\frac{\left[I\right]}{K_i}\right)+\left[S\right]\left(1+\frac{\left[I\right]}{K_{\text{ii}}}\right)}$$ (2)

where v denotes the velocity of the reaction; V_max denotes the maximum velocity; K_m denotes the Michaelis-Menten constant; [I] denotes inhibitor concentration; K_i, the competitive inhibition constant, denotes the affinity of inhibitor toward the free enzyme; and K_ii, the uncompetitive inhibition constant, denotes the affinity of inhibitor toward the enzyme-substrate complex. Eadie-Hofstee plots were constructed to assess the mode of inhibition. Initial estimates of K_i for CBD_m and CBD_p were obtained from Dixon plots.

[I], K_i, and K_ii were applied to the following mechanistic static model²⁰ to predict the drug interaction risk between CBD_p or CBD_m and the clinically relevant CYP2C19 substrate clopidogrel:

$$\text{AUCR}=\frac{1}{\left[\frac{1}{1+\left(\frac{\left[I\right]}{K_{\mathrm{i}}}\right)}\right]\ast \left[\frac{1}{1+\left(\frac{\left[I\right]}{K_{\text{ii}}}\right)}\right]\ast f_{\mathrm{m}}+\left(1-f_{\mathrm{m}}\right)}$$ (3)

where AUCR is the ratio of the AUC of clopidogrel in the presence to absence of the inhibitor, f_m denotes the fraction of hepatic clearance of clopidogrel mediated by CYP2C19 (0.45²¹), and [I] denotes the estimated CBD concentration (40 and 300 nM) from each oral dose (30 and 240 mg) used in our clinical study (ClinicalTrials.gov identifier: NCT 06692933).²² Statistical comparisons of inhibition kinetic parameters between CBD_m and CBD_p were made using the unpaired Student’s t test (GraphPad Prism); P values of <.05 for all tests were considered statistically significant.

2.3.3. IC₅₀ of CBD_p against CYP2C19 activity in non-MASH and MASH HLMs

A single concentration of S-mephenytoin (60 μM) and a range of CBD_p concentrations (0–10 μM) were tested using non-MASH HLMs (0.1 mg/mL) and MASH HLMs (0.3 mg/mL). Because of the decreased CYP2C19 activity in MASH HLMs relative to non-MASH HLMs (data not shown), a higher concentration of MASH HLMs was used to ensure that the metabolite concentration was not below the limit of quantitation. IC₅₀ was recovered via nonlinear least-squares regression analysis as described²³ using Phoenix WinNonlin.

2.3.4. CYP2C19 activity in the presence of CBD_p under simulated hepatic steatotic conditions using non-MASH HLMs

Hepatic steatotic and nonsteatotic conditions were simulated using non-MASH HLMs and FA or TG concentrations representative of those measured in livers obtained from 3 and 8 donors with and without steatosis, respectively.²⁴ Specifically, average (± SD) total lipids and TG concentrations in steatotic donors were 841 ± 297 and 381 ± 110 μg/mg protein, respectively. These concentrations equate to ∼300 μM FAs (2:1 oleic acid:palmitic acid) and ∼50 μM TGs (2:1 triolein:palmitodiolein).^13,25 Corresponding concentrations under nonsteatotic conditions were 3 times lower,²⁴ equating to 100 and 17 μM of FAs and TGs, respectively. All lipids were dissolved in DMSO, diluted further in KP_i supplemented with 2% BSA, and sonicated (oleic acid and triolein) or heated (palmitic acid), if needed. An emulsion formed for oleic acid and triolein, which became clear after addition to the aqueous incubation mixture.

The inhibitory effects of a single concentration of CBD_p (0.9 μM) on CYP2C19 activity were compared under steatotic and nonsteatotic conditions using a single concentration of S-mephenytoin (30 μM). The CBD_p concentration was selected based on the expected 50% inhibition of CYP2C19 activity, which was back-calculated from the K_i determined above. The percent decrease in 4′-hydroxymephenytoin concentration was compared with vehicle control.

Percent recovery of CBD_p in incubation mixtures without and with HLMs and percent CBD_p localized in HLMs under each condition were calculated using the following equations:

$$\%\text{recovery}\text{without}\text{HLMs}=\frac{{\mathrm{C}2}_{\mathrm{A}}-{\mathrm{C}1}_{\mathrm{A}}}{{\mathrm{C}1}_{\mathrm{A}}}\times 100$$ (4)

$$\%\text{recovery}\text{with}\text{HLMs}=\frac{{\mathrm{C}2}_{\mathrm{B}}-{\mathrm{C}1}_{\mathrm{A}}}{{\mathrm{C}1}_{\mathrm{A}}}\times 100$$ (5)

$$\%\text{CBD}\text{localized}\text{in}\text{HLMs}=\frac{{\mathrm{C}2}_{\mathrm{A}}-{\mathrm{C}2}_{\mathrm{B}}}{{\mathrm{C}2}_{\mathrm{B}}}\times 100$$ (6)

$$\text{Fold}\text{increase}\text{in}\text{CBD}\text{bound}\text{to}\text{HLMs}=\frac{{\mathrm{C}2}_{\mathrm{A}}\times \%\text{CBD}\text{localized}\text{in}\text{HLMs}\left(\text{lipid}\right)}{{\mathrm{C}1}_{\mathrm{A}}\times \%\text{CBD}\text{localized}\text{in}\text{HLMs}\left(\text{control}\right)}$$ (7)

where C1 denotes CBD_p + KP_i as the control, and C2 denotes CBD_p + lipids as described above for nonhepatic steatotic and steatotic FAs or TGs. Subgroups A and B represent incubation mixtures with and without HLMs, respectively. Equation 6 assumes that the reduction in CBD_p recovery with HLMs is due to CBD_p being localized in HLMs and is 100% bound to HLMs. These notations were inspired by those described in a previous publication involving the calculation of the partition ratio for unbound cannabinoids.²⁶ Statistical comparisons among the 4 conditions of CBD_p in the presence of lipids to CBD_p without lipids were made using one-way ANOVA with Dunnett’s multiple comparisons test (GraphPad Prism); P values of <.05 for all tests were considered statistically significant.

Following the screening assay with the P450 cocktail (see Screening of CBDm in the Hemp Oil Product as an Inhibitor of CYP1A2, CYP2C9, CYP2C19, CYP2D6, and CYP3A activity), FAs (1 and 3 mM, primary incubation concentration) were equilibrated with non-MASH HLMs and CBD_p (5 μM). Given the strong inhibition observed with CYP2C19, the experiment was repeated using 2 μM CBD_p, specifically for CYP2C19. To assess protection from FA-induced damage against P450 activity, 0.2% and 0.5% BSA (final concentration in the secondary incubation mixture) were tested. Results for CYP1A2, CYP2C9, CYP2D6, and CYP3A4 are reported using 0.5 μM CBD and for CYP2C19 using 0.2 μM CBD; these CBD concentrations refer to those in the secondary incubation mixture.

2.4. Contribution of P450s and uridine diphosphate-glucuronosyltransferases to CBD metabolism in non-MASH HLMs and MASH HLMs

The contributions of P450s and uridine diphosphate-glucuronosyltransferases (UGTs) to CBD metabolism were evaluated using HLMs and the substrate depletion method with modifications.²³ In brief, CBD_p (0.3 μM) was incubated with non-MASH HLMs (0.1 mg/mL) or MASH HLMs (0.3 mg/mL), KP_i, and BSA (0.2%). Mixtures were equilibrated at 37 °C for 5.5 minutes before initiating the reactions with NADPH (1 mM) and/or uridine 5′-diphosphoglucuronic acid trisodium salt (2.5 mM). At predetermined intervals from 0 to 20 minutes, reactions were quenched with ice-cold methanol containing CBD-d₉ (internal standard). CBD was quantified using UPLC-MS/MS (described briefly in Quantification of CYP Probe Metabolites, CBD, Midazolam, and Atenolol and in the Supplemental Material). A total of 4 independent experiments were conducted on 2 different days. Statistical comparisons between non-MASH and MASH HLMs were made using the paired Student’s t test (GraphPad Prism); P values of <.05 for all tests were considered statistically significant.

2.4.1. CBD unbound intrinsic clearance and fraction metabolized

Total hepatic intrinsic clearance (CL_int) and fraction metabolized (f_m) for CBD were determined as described previously.²⁷ Briefly, the depletion rate constant (k_dep) was estimated by fitting a log-linear regression model to the remaining CBD concentration versus time data using GraphPad Prism. CL_int by P450s, UGTs, and P450s + UGTs was calculated as the ratio of k_dep under each condition to the non-MASH HLM or MASH HLM concentration. Protein binding and unbound concentrations of CBD were assumed not to differ between incubations containing non-MASH and MASH HLMs based on a previous report showing no differences between incubations containing non-MASH HLM at 0.1 and 0.5 mg/mL.²² Unbound CL_int (CL_int,u) was calculated as the ratio of CL_int to the reported unbound fraction for CBD in human plasma (f_u,p).²² f_m was calculated using the following equations:

$$f_{\mathrm{m},\text{P450s}}=\frac{{\text{CL}}_{\mathrm{int},\text{P450s}}}{{\text{CL}}_{\mathrm{int},\text{P450s}+\text{UGTs}}\text{or}{\text{CL}}_{\mathrm{int},\text{P450s}}+{\text{CL}}_{\mathrm{int},\text{UGTs}}}$$ (8)

$$f_{\mathrm{m},\text{UGTs}}=\frac{{\text{CL}}_{\mathrm{int},\text{UGTs}}}{{\text{CL}}_{\mathrm{int},\text{P450s}+\text{UGTs}}\text{or}{\text{CL}}_{\mathrm{int},\text{P450s}}+{\text{CL}}_{\mathrm{int},\text{UGTs}}}$$ (9)

2.5. Permeability of CBD in Caco-2 cell monolayers under simulated high-fat diet conditions

The apparent permeability (P_app) of CBD (10 μM) in the absorptive (apical to basolateral) direction was determined under 4 different conditions using established methods:^28,29 CBD_p, hemp oil, CBD_p + low oleic acid (0.04 mg/mL), and CBD_p + high oleic acid (4 mg/mL). Low and high concentrations of oleic acid were used to simulate the effect of low- and high-fat diets on CBD P_app. In brief, Caco-2 cells with passage counts of 9–12 were seeded onto 12-mm Transwell polyester membrane inserts (Corning 3460) at a density of 2.5 × 10⁵ cells/cm² and allowed to grow for at least 21 days in minimum essential medium supplemented with 20% FBS and penicillin/streptomycin (100 units/0.1 mg per mL). Cell monolayers with transepithelial electrical resistance >220 Ω·cm² were used for permeability assessment. BSA (0.2%) was added to the apical compartment to minimize nonspecific binding of CBD; a higher concentration (2%) was added to the basolateral compartment to more closely mimic physiological conditions while preserving UPLC column integrity (see Quantification of CYP Probe Metabolites, CBD, Midazolam, and Atenolol) and maintaining high recovery of CBD during sample processing. P_app for CBD, the low-permeability control atenolol (10 μM), and the high-permeability control midazolam (10 μM) were determined from 10 to 120 minutes. An aliquot (100 μL) was removed from the basolateral compartment at specified times, after which the volume was replaced. Samples were diluted in methanol containing alprazolam (50 nM; internal standard for atenolol and midazolam) or CBD-d₉ (100 nM) before UPLC-MS/MS analysis. Statistical comparisons of CBD in hemp oil, CBD_p + low oleic acid (0.04 mg/mL), and CBD_p + high oleic acid (4 mg/mL) to CBD_p were made using one-way ANOVA with Tukey's multiple comparisons test (GraphPad Prism); P values of <.05 for all tests were considered statistically significant.

2.6. Quantification of P450 probe metabolites, CBD, midazolam, and atenolol

All analytes were quantified using a UPLC-MS/MS system, which consisted of a Shimadzu UPLC coupled with a SCIEX 6500+ QTRAP mass spectrometer (SCIEX). Chromatography was accomplished using an ACQUITY UPLC HSS T3 column (1.8 μm, 2.1 × 100 mm) with an ACQUITY UPLC HSS T3 VanGuard (1.8 μm, 2.1 × 5 mm) and filter column (0.2 μm, 2.1 mm) (Waters Corporation). Additional details are provided in Supplemental Table 1. All standard curves were prepared fresh on each experimental day and run twice. Accuracy was within 20% of the nominal concentration for the limit of quantitation and was 15% for all other concentrations.

3. Results

3.1. CBD_m in the hemp oil product and CBD_p have similar inhibitory effects on the 5 P450s tested

CBD_m in hemp oil showed concentration-dependent inhibition of all 5 P450s (Fig. 2). CBD_m showed strong to weak reversible inhibition of all 5 P450s, as well as moderate and weak time-dependent inhibition of CYP1A2 and CYP3A, respectively. The extent of reversible inhibition by CBD_m was highest for CYP2C19, followed by CYP2C9, CYP3A, CYP1A2, and CYP2D6. The extents of inhibition by CBD_m at 20 μg/mL (containing 4.5 μM CBD) were similar to those for CBD_p (5 μM).

Fig. 2.

Percent remaining activity of the tested P450s compared with vehicle after exposure of HLMs to CBD_m and CBD_p for 0–30 minutes in the primary incubation mixtures. Bars and error bars denote means and SDs, respectively, of at least 3 independent experiments. ∗Statistically significant (P < .027, two-way ANOVA with Tukey's multiple comparisons test). TIC, ticlopidine.

3.2. Reversible inhibition

3.2.1. All 3 compositions of TGs reduced CYP2C19 activity

A range of MCT concentrations was tested on a single concentration of S-mephenytoin, and a single concentration of MCT was tested on a range of S-mephenytoin concentrations. CYP2C19 activity decreased slightly (−15% to −26%) with increasing MCT concentrations (4.5–45 μg/mL) (Fig. 3A), and the effects of MCT (−22% to −13%) decreased with increasing S-mephenytoin concentrations (10–60 μM) (Fig. 3B). MCT and palmitodiolein (long-chain TG) decreased CYP2C19 activity to a similar extent (−25% ± 4.4% and −28% ± 4.7%, respectively), whereas tributyrin (short-chain TG) decreased CYP2C19 activity to a lesser extent (−10% ± 5.4%) (Fig. 3C).

Fig. 3.

Percent remaining CYP2C19 activity (S-mephenytoin 4′-hydroxylation) in the presence of matrix oil relative to vehicle (MeOH). (A) A range of medium-chain triglyceride (MCT) concentrations was tested on a single concentration of S-mephenytoin (10 μM). MCT concentrations matched those in the hemp oil added to incubations to achieve a final CBD concentration of 1, 3, and 10 μM. (B) A single concentration of MCT (45 μg/mL) was tested on a range of S-mephenytoin concentrations. (C) A single concentration of different triglyceride compositions (45 μg/mL) was tested on a single concentration of S-mephenytoin (10 μM). Ticlopidine (TIC) was used as the canonical CYP2C19 inhibitor. Horizontal lines and error bars denote means and SDs, respectively, of 3–4 independent experiments. ∗Statistically significant (P < .028, one-way ANOVA with Tukey's multiple comparisons test).

3.2.2. CBD_p and CBD_m exhibited a mixed mode of reversible CYP2C19 inhibition

An initial estimated K_i of ∼0.2 μM for both CBD_m and CBD_p was obtained from Dixon plots (Fig. 4A). The Eadie-Hofstee plot suggested a mixed mode of inhibition for both CBD_p and CBD_m toward CYP2C19 activity (Fig. 4B). As CBD concentration increased, the K_m of S-mephenytoin decreased while V_max increased (Supplemental Table 2). Consistent with these observations, the mixed inhibition model best described the data according to standard diagnostics,¹⁸ including the Akaike information criterion, randomness of the residuals, and SEs of parameter estimates (Fig. 4C). The K_i for CBD_m was slightly higher than that for CBD_p (0.22 ± 0.03 μM vs 0.16 ± 0.03 μM; P = .03, unpaired Student’s t test), whereas K_ii was comparable (2.00 ± 0.31 μM vs 1.54 ± 0.22 μM; P = .07). Rough estimates of K_i and K_ii for CBD_m from the abbreviated assay were comparable among the 3 different TG compositions (Supplemental Table 3). Predicted AUCR for clopidogrel upon coadministration with CBD at 30 and 240 mg was comparable between CBD_p and CBD_m (1.08 and 1.40 vs 1.11 and 1.47, respectively).

Fig. 4.

Concentration-dependent reversible inhibition of CYP2C19 activity (S-mephenytoin 4′-hydroxylation) in human liver microsomes by CBD_m as depicted in the (A) Dixon, (B) Eadie-Hofstee, and (C) Michaelis-Menten plots of the data. Legends denote S-mephenytoin (A) or CBD (B, C) concentrations. Symbols and error bars denote means and SDs, respectively, of 3–5 independent experiments. Curves denote nonlinear least-squares regression of the data using the mixed inhibition model and Phoenix WinNonlin. Corresponding plots for CBD_p were essentially superimposable with those for CBD_m. See Table 1 for comparisons between CBD_p and CBD_m.

3.2.3. FAs and TGs enhanced CBD-mediated inhibition of CYP2C19 activity

Lipids alone reduced CYP2C19 activity under both nonsteatotic and steatotic conditions, with a greater reduction observed under the latter (Table 1). Additive or synergistic inhibition of CYP2C19 activity was observed with both CBD_p and lipids. Similar results were observed with other P450s (Supplemental Table 4). The calculated contribution of CBD_p to the reduction in CYP2C19 activity in the presence of lipids was comparable to that in the absence of lipids (Table 1). Adding 0.2% BSA to the incubation mixture markedly decreased the nonspecific binding of CBD_p to the low-binding tubes. The addition of 0.5% BSA protected HLMs from FAs but also decreased the inhibitory effect of CBD on the 5 P450s (Supplemental Table 4). FAs and TGs did not enhance CBD_p solubility (Table 2). CBD_p localized in HLMs increased by 12% and 26% in the presence of 100 μM and 300 μM FAs, respectively, and slightly increased by 4% and 5% in the presence of 17 μM and 50 μM TGs, respectively. After adjusting for the fold increase in CBD_p bound to HLM in the presence of lipids (Table 2), the corrected percent reductions in CYP2C19 activity in the presence of both CBD_p and lipids were comparable to the uncorrected data, assuming minimal contribution of lipid to CYP2C19 activity reduction (Table 1). These observations indicated that FAs and TGs enhanced CBD-mediated inhibition of CYP2C19.

Table 1.

Reduction in CYP2C19 activity in the presence of FAs or TGs with CBD under both nonsteatotic and steatotic-like conditions

Data reported as mean ± SD (n = 3 independent experiments)

	TGs 17 μM (Nonsteatotic)	TGs 50 μM (Steatotic)	FAs 100 μM (Nonsteatotic)	FAs 300 μM (Steatotic)
Lipid alone (control) (%)	−9 ± 1.0	−25 ± 3.5	−17 ± 2.4	−68 ± 0.4
CBDp + lipid (%)	−55 ± 1.6∗	−62 ± 1.2∗	−61 ± 2.2∗	−84 ± 1.5∗
Contribution of CBDp in lipida (%)	−51 ± 1.4	−49 ± 0.7	−52 ± 3.7	−50 ± 4.6
Corrected CBDp + lipidb (%)	−57 ± 9.8	−61 ± 6.1	−68 ± 4.8	−79 ± 0.9

∗Statistically different from CBD_p alone (P < .004, one-way ANOVA with Dunnett’s multiple comparisons test).

^aCalculated as [(CBD_p + lipid) − lipid alone]/(lipid alone).

^bBack-calculated from the IC₅₀ curve in response to the correction for fold increase in CBD bound to HLMs, assuming minimal contribution of lipid to CYP2C19 activity reduction (see Table 3). In the absence of lipids, CBD_p alone reduced CYP2C19 activity by 49% ± 1.6%.

Table 2.

Percent recovery of CBD_p in incubations containing lipids

Data are reported as mean ± SD (n = 3 independent experiments). See CYP2C19 Activity in the Presence of CBDp Under Simulated Hepatic Steatotic Conditions Using Non-MASH HLMs for details about the various calculations

	% Recovery of CBDp Without HLMs	% Recovery of CBDp with HLMs	% CBDp Localized in HLMs	Fold Increase in CBDp Bound to HLMsa
CBDp + KPi (control)	100 ± 3.6	88 ± 2.7	14 ± 3.6	1.00 ± 0.26
CBDp + FAs 100 μM	101 ± 2.5	80 ± 3.1∗	26 ± 5.0∗	1.86 ± 0.36∗
CBDp + FAs 300 μM	102 ± 2.6	73 ± 1.1∗	40 ± 2.1∗	2.91 ± 0.15∗
CBDp + TGs 17 μM	106 ± 3.1	90 ± 3.9	18 ± 5.1	1.32 ± 0.38
CBDp + TGs 50 μM	108 ± 3.2	91 ± 2.7	19 ± 3.6	1.44 ± 0.27

∗Statistically different from CBD_p + KP_i (P < .032, one-way ANOVA with Dunnett’s multiple comparisons test).

^aAssuming that the reduction in CBD_p recovery with HLMs is due to CBD_p being localized in HLMs and all is bound to HLMs.

3.3. MASH impacts CBD_p intrinsic clearance but not CYP2C19 inhibition

Relative to non-MASH HLMs, CBD_p CL_int for P450s, UGTs, and P450s + UGTs in MASH HLMs was 55%, 17%, and 47% lower, respectively (Table 3). Corresponding f_m values were calculated using 2 different approaches. One involved determining the ratio of CL_int,P450s or CL_int,UGTs to CL_int,both (f_m,relative); the other involved determining the ratio of CL_int,P450s or CL_int,UGTs to the sum of CL_int,P450s + CL_int,UGTs (f_m,sum). Relative to non-MASH HLMs, the contribution by P450s and UGTs to CBD_p clearance decreased and increased, respectively, in MASH HLMs. CYP2C19 activity in MASH HLMs was 80% lower than that in non-MASH HLMs. The IC₅₀ for CBD_p was comparable between non-MASH and MASH HLMs (1.28 ± 0.25 and 1.12 ± 0.23 μM, respectively) (Fig. 5).

Table 3.

Intrinsic clearance of CBD and fraction metabolized by P450s and UGTs

All values are means ± SDs of 4 independent experiments.

	Non-MASH HLMs	MASH HLMs	% Change
Unbound intrinsic clearance (CLint,u) (mL/min per mg)a
P450s (NADPH only)	56.1 ± 2.3	25.6 ± 1.6∗	−55
UGTs (UDPGA only)	13.6 ± 3.5	11.4 ± 1.7	−17
Both (NADPH + UDPGA)	74.0 ± 3.7	39.3 ± 1.7∗	−47
Fraction metabolized (fm)
fm,sum P450sb	0.80 ± 0.04	0.69 ± 0.02	—
fm,sum UGTsb	0.20 ± 0.04	0.31 ± 0.02	—
fm,relative P450sc	0.76 ± 0.03	0.65 ± 0.04	—
fm,relative UGTsc	0.18 ± 0.05	0.29 ± 0.04	—

UDPGA, uridine 5′-diphosphoglucuronic acid trisodium salt.

∗Statistically different from non-MASH HLMs (P < .0005, paired Student’s t test).

^aFraction unbound of CBD in human plasma (f_u,p = 0.013).²²

^bRatio of CL_int,P450s or CL_int,UGTs to the sum of CL_int,P450s + CL_int,UGT.

^cRatio of CL_int,P450s or CL_int,UGTs to CL_int,both.

Fig. 5.

Concentration-dependent reversible inhibition of CYP2C19 activity (S-mephenytoin 4′-hydroxylation) in non-MASH HLMs (●) and MASH HLMs () by CBD_p. Symbols and error bars denote means and SDs, respectively, of 3 independent experiments.

3.4. CBD exhibits low membrane permeability in Caco-2 cell monolayers

The P_app of CBD_p and CBD in the hemp oil product was 1.65 × 10⁻⁶ ± 0.45 × 10⁻⁶ and 1.04 × 10⁻⁶ ± 0.31 × 10⁻⁶ cm/s, respectively (Fig. 6A). CBD_p P_app under the low-fat diet condition was 1.88 × 10⁻⁶ ± 0.40 × 10⁻⁶ cm/s and was unexpectedly low under the high-fat diet condition (1.70 × 10⁻⁷ ± 0.92 × 10⁻⁷ cm/s) despite disruption of monolayer integrity. CBD translocated through the monolayer after an ∼30-minute delay (Fig. 6B). The P_app of midazolam (1.20 × 10⁻⁵ ± 0.1 × 10⁻⁵ cm/s) and atenolol (8.9 × 10⁻⁷ ± 2.1 × 10⁻⁷ cm/s) was consistent with a previous report.²⁹

Fig. 6.

(A) P_app of CBD in Caco-2 cell monolayers under 4 different conditions. Midazolam and atenolol were used as a high-permeability and a low-permeability control, respectively. Horizontal lines and error bars denote means and SDs, respectively, of 3 independent experiments. (B) Normalized CBD concentration under different conditions as a function of time (n = 3), where the ratio is calculated as the concentration of CBD in the basolateral compartment (nM_b) divided by the initial concentration of CBD in the apical compartment (μM_a). ∗Statistically significant (P < .046, one-way ANOVA with Tukey's multiple comparisons test).

4. Discussion

Human exposure to botanical and other natural products has increased markedly during the last several decades, raising concerns for potential interactions with pharmaceutical drugs.^30,31 Unlike drugs, natural products are complex mixtures of active and inactive phytoconstituents. Prominent among these products are hemp oils. Constituents include CBD and other cannabinoids, terpenes, and TGs, all of which can inhibit P450 activity in vitro.^22,32,33 Hepatic exposure to these constituents concurrent with CBD may in turn alter the P450 inhibitory effects of CBD, prompting the first objective of the current study: evaluate whether CBD_p is equivalent to CBD_m in a well characterized hemp oil product as a P450 inhibitor.

Hemp oils typically contain a high TG content to enhance the solubility and shelf life of cannabinoids. Studies on drug metabolism and drug interactions typically use HLMs under simplified conditions to evaluate mechanisms, but these conditions often lack factors that reflect a physiologically relevant microenvironment. To our knowledge, results from this study are the first to demonstrate that TGs reduce CYP2C19 activity. Potential mechanisms include (1) sequestration of substrates into the lipid microenvironment or (2) prevention of substrates from entering the CYP2C19 active site. Contrary to expectation, the P450 inhibitory effects of CBD were unaltered in the presence of TGs. Both CBD_m and CBD_p, at 2 relatively low doses (30 and 240 mg), were predicted to increase the AUC of the antiplatelet drug and CYP2C19 substrate clopidogrel by similar extents (∼1.1- and ∼1.5-fold, respectively). These observations suggested that the high binding affinity of CBD for CYP2C19 surpasses the effect of TGs, further supporting that CBD_p and CBD_m are equivalent when evaluating potential CYP2C19-mediated hemp-drug interactions involving hemp oil products containing CBD as the predominant cannabinoid.

Whether CBD_p is representative of other CBD-containing products, including those with an equivalent content of the precursor CBD acid (CBDA), remains unknown. A recent clinical study showed that the systemic exposure to CBDA was significantly higher than that to CBD,³⁴ potentially leading to pronounced interactions with P450s. An in vitro study using canine liver microsomes showed that CBDA inhibited CYP2B11 and CYP2D15.³⁵ Studies involving HLMs are needed to evaluate the inhibitory effects of CBDA on human P450s, particularly CYP2C19.

The observed reduction in CYP2C19 activity upon exposure of HLMs to TGs and FAs suggested new direct interaction mechanisms, prompting the second objective of this work: evaluate the effects of a high-lipid microenvironment on CYP2C19 activity by comparing the inhibitory effects of CBD_p between (1) simulated hepatic steatotic and nonsteatotic conditions using non-MASH HLMs and (2) MASH and non-MASH HLMs. Relative to FAs, TGs decreased CYP2C19 activity to a lesser extent under both conditions. Potential explanations include (1) FAs undergo metabolism by CYP4 and CYP2E1^36,37 and can inhibit multiple P450s,¹⁴ suggesting a higher binding affinity for the enzyme(s) compared with TGs, and (2) TGs are composed of 3 FA chains and a glycerol backbone, potentially leading to steric hindrance at the enzyme binding site. Results showed that both lipids can directly reduce P450 activity in HLMs. Contrary to expectation, high lipid concentrations did not “protect” CYP2C19 from the inhibitory effect of CBD by sequestering CBD away from CYP2C19; rather, they contributed to additive or synergistic effects (Table 2), suggesting an independent inhibition mechanism. After correcting for high CBD concentrations, and assuming that all CBD is bound to HLMs in the presence of lipids, the increase in CBD concentrations matched the reduction in P450 activity observed with the uncorrected data. These results support the notion that TGs and FAs sequester CBD within the lipid microenvironment and subsequently localize CBD into HLM, thereby enhancing the CBD binding affinity for inhibition.

Significant reductions in CYP2C19 activity observed with MASH HLMs confirmed the link between reduced protein expression and MASH.³⁸ The high-lipid environment was anticipated to enhance CBD-mediated P450 inhibition, thus lowering IC₅₀. However, IC₅₀ for CBD between non-MASH and MASH was similar, suggesting that lipids are removed during HLMs preparation from liver tissue. The comparable IC₅₀ further suggested that MASH did not induce conformational changes in CYP2C19. Comprehensive models incorporating both MASH HLMs and high-lipid environments can further validate the overall impact of MASH.

There are limitations to this portion of the work. First, for the relevant experiments and interpretations thereof, enzymes were assumed to be exposed to all components of CBD_m (ie, TGs and CBD) to a comparable extent, which may not necessarily be the case in vivo. Second, FA-binding protein 1, which protects hepatocytes from FA-induced damage,³⁹ was not included in the HLMs incubations. Instead, BSA, which can bind FAs, was used. However, its protective effects against FA-induced damage to HLMs may not be comparable to FA-binding protein 1; in addition, BSA can alter the inhibitory effects of CBD (Supplemental Table 4). Third, oxidative stress impairs hepatic enzyme function by overproducing reactive oxygen species during FA metabolism.^37,40 Reactive oxygen species scavengers could be added to mitigate potential oxidative stress. Finally, effects of a high-lipid environment on the inhibitory effects of CBD were not validated with other CYP2C19 inhibitors. Addressing these limitations would improve the understanding of the effects of a high-lipid environment using HLMs.

Drug interaction risk depends on both the potency of the precipitant and associated pharmacokinetics. The risk may be higher in specific populations, including patients with hepatic steatosis or MASH, due to altered enzyme activity. A change in CBD clearance by P450s or UGTs could greatly affect the CBD inhibitory effect. CBD CL_int in MASH HLMs was half that in non-MASH HLMs (Table 3), suggesting that the duration of CBD-mediated drug interactions could be prolonged. However, the extent of reduction in CBD CL_int by P450s was greater than that by UGTs, indicating that MASH has a minimal effect on UGT activity, consistent with a previous report.⁴¹ Compared with non-MASH HLMs, f_m by MASH HLMs shifted from P450 to UGT by a value of 0.11 after adjusting for the reduction in CBD clearance. These results may help capture a more appropriate CBD clearance in these populations, rather than extrapolating from healthy populations, to evaluate both metabolism and drug interaction risk.

CBD absorption is another factor that contributes to drug interaction risk. Despite being highly lipophilic, CBD slowly permeated through Caco-2 cell monolayers (Fig. 6B), suggesting that CBD may be localized within the cells before permeating into the basolateral compartment. Such localization, which may pertain to multiple cell types, could explain, in part, why CBD exhibits a high volume of distribution^42,43 and a complex absorption phase in human pharmacokinetic studies. CBD oral bioavailability is greatly enhanced by food intake, as evidenced by up to a 10-fold increase in AUC, especially with formulations that improve CBD solubility.44, 45, 46 Likewise, higher dietary lipid intake has been shown to enhance drug solubilization by increasing bile salt and pancreatic lipase secretion.^47,48 The unexpected low P_app for CBD in the presence of oleic acid might be due to the low solubility of FAs. Indeed, the low solubility of oleic acid in the emulsion led to sedimentation, sequestering CBD within the precipitate despite disrupting the Caco-2 cell monolayer. Several compositions of fed-state simulated intestinal fluid were tested in a recent study to preserve Caco-2 cell integrity.⁴⁹ These simulated fluids could be applied in future studies.

Limitations to the permeability study include high nonspecific binding of CBD and insufficient collection times. Although the intent of adding BSA to the apical compartment was to reduce nonspecific binding of CBD, BSA appeared to hinder CBD permeation (data not shown). Given the complex absorption profile of CBD, with a peak plasma concentration occurring up to 4 hours after administration,⁴³ extending the collection time may better capture the permeability profile of CBD. Gastric permeability under the fed state with a high-fat diet is mediated by more complex mechanisms involving the lymphatic system. Addressing these limitations and further optimizing the working systems will improve the understanding of the complex permeability of CBD.

In summary, this work presents a collection of studies that generated foundational knowledge regarding CBD disposition and P450-mediated interactions among simulated exogenous and endogenous lipid microenvironments (Fig. 7). Regarding exogenous factors, results suggested that (1) despite a reduction in CYP2C19 activity by MCT, CBD in the hemp oil (CBD_m) behaved comparable to CBD_p as a CYP2C19 inhibitor; (2) CBD slowly penetrated the intestinal monolayer and persisted within enterocytes; and (3) FA concentrations indicative of a high-fat diet disrupted the cell membrane, precluding accurate determination of effects on CBD P_app. Regarding endogenous factors, results suggested that (1) FAs reduced CYP2C19 activity and enhanced CBD-mediated CYP2C19 inhibition to a greater extent compared with TGs and (2) although both CYP2C19 activity and CBD clearance in MASH HLMs were reduced, the CBD inhibitory effects on CYP2C19 activity were comparable. Overall, our modified in vitro HLM systems provide insight into potential mechanisms underlying the direct effects of lipids on CYP2C19 activity and CBD disposition. Upon further optimization of these hepatic steatotic conditions, physiologically based pharmacokinetic models could be developed and used to advance the mechanistic understanding of the effects of various lipid microenvironments on drug disposition, as well as inform future clinical studies.

Fig. 7.

Summary of effects of simulated exogenous and endogenous lipid microenvironments on the CYP2C19 inhibitory effects and disposition of CBD using human liver microsomes (HLMs) and Caco-2 cell monolayers. Results could be used to inform the development of physiologically based pharmacokinetic (PBPK) models and/or clinical studies to assess the risk of CYP2C19-mediated CBD-drug interactions in relevant specific populations, including patients with hepatic steatosis and MASH.

Conflict of interest

M.F.P. is a paid consultant for Delphian Therapeutics. All other authors declare no conflicts of interest.