What this means in practice
Anyone on valproate should not add cannabidiol without their prescriber knowing, because that specific pair produced the highest rate of liver enzyme elevation in the trials and needs blood tests to catch. For everyone else nothing changes, and no liver benefit is claimed here in either direction.
What people say
CBD is good for your liver. It is anti-inflammatory, and detox and liver-support products are sold with cannabidiol in them.
What the evidence supports
In the epilepsy trials behind the approved label, liver enzyme readings above three times the normal ceiling appeared in 3 percent of patients taking neither valproate nor clobazam, 4 percent taking clobazam alone, 21 percent taking valproate alone, and 30 percent taking both. Two separate trials show the same fingerprint: in the Dravet syndrome dose-ranging trial, every patient whose liver enzymes rose was also on valproate, and in the tuberous sclerosis trial 18.9 percent of treated patients had raised transaminases against none on placebo. The NIH liver injury database grades cannabidiol a suspected but unproven rare cause of clinically apparent liver injury, and specifically flags the valproate combination. All of this comes from prescription doses of 10 to 25 mg/kg/day taken with routine blood monitoring, so it establishes the combination as risky at those doses and does not establish anything at consumer amounts.
How it is supposed to work
Both compounds can irritate the liver on their own, and taken together the liver test abnormalities appear far more often than with either separately. Why is not established; the label and the NIH liver injury database both describe the pattern without naming a mechanism for it.
Which way it runs
The risk is a combined one rather than one drug acting on the other. Cannabidiol does not meaningfully change valproate blood levels, and valproate does not change cannabidiol levels, yet taking both raises the rate of liver enzyme elevation far above either alone. The harm shows up in the liver readings, not in the drug concentrations.
What is still missing
The gap is not that a liver claim is overstated. It is that the direction is reversed: the only place cannabidiol has been watched closely enough for the liver to be measured, the liver is the thing being watched for.
Sources
- EPIDIOLEX (cannabidiol) oral solution — US prescribing information, SPL version 35Jazz Pharmaceuticals, Inc., label revision 2026
- Dose-Ranging Effect of Adjunctive Oral Cannabidiol vs Placebo on Convulsive Seizure Frequency in Dravet Syndrome: A Randomized Clinical TrialMiller I, Scheffer IE, Gunning B, et al., 2020
- Add-on Cannabidiol Treatment for Drug-Resistant Seizures in Tuberous Sclerosis Complex: A Placebo-Controlled Randomized Clinical TrialThiele EA, Bebin EM, Bhathal H, et al., 2021
- Cannabidiol — LiverTox: Clinical and Research Information on Drug-Induced Liver InjuryNational Institute of Diabetes and Digestive and Kidney Diseases, LiverTox, chapter NBK548890