Highly purified cannabidiol (CBD) (approved as Epidiolex in the US) is efficacious in treating seizures associated with Dravet syndrome (DS), Lennox-Gastaut syndrome (LGS), and Tuberous Sclerosis Complex (TSC).  In Epidiolex clinical trials, frequent and dose-dependent elevations in serum alanine aminotransferase (ALT) were observed. In Epidiolex-treated patients with LGS or DS (10 or 20 mg/kg/day dosages), the incidence of ALT elevations >3x the upper limit of normal (ULN) was 21% in participants taking CBD with concomitant valproate (VPA) compared with 3% in participants not taking VPA; this interaction was not pharmacokinetic.

Here, we aimed to identify the mechanism(s) accounting for the higher incidence of ALT elevation observed in individuals treated with VPA and CBD by using a Quantitative Systems Toxicology (QST) model of hepatotoxicity (DILIsym®) to test the hypothesis that increased incidence of ALT elevation was due to VPA and CBD (or metabolites of each) inhibiting mitochondrial respiration.

Thus, we assessed any effect of CBD, and its two main plasma metabolites (7-COOH-CBD, 7-OH-CBD) using in vitro assays that can provide output variables associated with three hepatotoxic mechanisms: production of reactive oxygen stress (ROS), mitochondrial dysfunction, and inhibition of bile acid (BA) transporters. We also verified the ability of VPA to inhibit mitochondrial respiration.  These in vitro data were used to quantify drug and metabolite effects which, combined with predicted liver exposure values, parameterized the DILIsym mathematical model to predict the likelihood of liver toxicity for the following scenarios: administration of VPA only, CBD only, and VPA plus CBD.

Simulation results with VPA alone with mitogenesis in the model were consistent with clinical observations (5-10% with ALT >3x ULN): ALT levels were transiently elevated (12% simulated ALT >3x ULN) and resolved upon continued dosing. Simulations of CBD alone predicted delayed dose-dependent ALT elevations consistent with clinical experience.  Mechanistic investigation simulations revealed that 1) oxidative stress was most likely the dominant hepatoxicity mechanism underlying the predicted ALT elevations for CBD, and 2) the parent molecule, CBD, provided the largest contribution to simulated toxicity, while the two metabolites 7-COOH-CBD and 7-OH-CBD had less significant contributions.  For concomitant therapy simulations, VPA was dosed alone for two weeks prior to the addition of CBD.  The predicted frequency of ALT elevations (>3x ULN) for CBD added to VPA treatment (2/285 simulated patients) did not differ from the frequency predicted in simulations when only CBD was dosed, largely because the VPA-induced ALT elevations had resolved before the CBD-induced elevations.

We, therefore, conclude that interference with mitochondrial respiration does not account for the large increase in the frequency of ALT elevations observed in VPA-treated patients who start treatment with CBD.

To be presented at the FDA/CDER and AASLD 2021 DILI Conference XVIII, April 20-22nd.