The most interaction-heavy substance in the whole hemp category, and the one most often assumed to be inert. It inhibits several CYPs and the phase-2 conjugating enzymes, and at the doses used in the epilepsy literature it carries a transaminase-elevation signal of its own.
Also known as: CBD, cannabidiol, Epidiolex, hemp extract. Category: supplement.
What it does, mechanism by mechanism
CYP2C19 inhibition — inhibits, moderate
Slows the enzyme handling clopidogrel activation, PPIs and some benzodiazepines. For clopidogrel the effect is loss of antiplatelet activity, not accumulation.
The best-documented clinical example in the cannabinoid field: CBD raises N-desmethylclobazam, the active metabolite of clobazam, with sedation as the visible consequence. That interaction was characterised in children with refractory epilepsy and is in the product labelling.
Slows the enzyme that metabolises roughly half of all prescription drugs. A CYP3A4 substrate taken with a CYP3A4 inhibitor reaches higher blood levels than its dose implies.
Slows the phase-2 conjugation that makes a compound water-soluble enough to excrete. Phase 2 is the step most interaction checkers skip. Cannabinoids are heavily glucuronidated, so this is not a side issue in this corpus.
A phase-2 effect on top of the phase-1 ones, which is why the interaction surface is wider than a CYP table alone suggests.
Two or more agents with documented liver injury signals taken together. Not a pharmacokinetic interaction; the additivity is at the organ.
Dose-dependent transaminase elevations are documented at the high milligram-per-kilogram doses used in the epilepsy trials, and the signal is larger when valproate is co-administered. Whether the low doses in consumer products carry the same signal is not established.
The interaction profile scales with dose, and consumer products span three orders of magnitude of dose with labelling that product surveys repeatedly find inaccurate. A stack built on an assumed CBD dose is built on a number that may not be true.
This list is what is IN the table. It is not the set of substances this interacts with — that set is
larger and partly unknown, and the mechanism is the thing to carry to a substance we have not listed.
A clean result means NO DOCUMENTED INTERACTION IN THIS DATASET. It does not mean safe, and it is not a clearance. Most substances are not in this dataset at all, and for many pairs that are, nobody has ever studied the combination.
Serotonergic drugs and the serotonin-toxicity mechanism
Dietary tyramine and L-dopa loads
The major cytochrome P450 pathways: CYP3A4, CYP2D6, CYP1A2, CYP2C9, CYP2C19 — inhibition and induction
P-glycoprotein inhibition and induction
11β-HSD2 inhibition (the licorice mechanism) and the potassium consequences that follow it
QT prolongation as an additive pharmacodynamic axis
Culinary seasonings and common foods with documented pharmacological activity
A selected set of narrow-therapeutic-index drugs where those shifts matter most
The endocannabinoid enzymes and transport: FAAH, MAGL, endocannabinoid membrane transport, CB1 and CB2
Additive CNS depression and GABA-A positive modulation — the alcohol / benzodiazepine / opioid / kava axis
The phytocannabinoids delta-9-THC, cannabidiol and the converted cannabinoids, as both substrates and inhibitors
Synthetic full CB1 agonists as a class, and why they are pharmacologically unlike cannabis
CYP2E1, and phase-2 glucuronidation and sulfation where a specific entry names them
The sedative and potentiator botanicals of the kava literature, and dietary L-dopa from Mucuna
Not in this dataset
Any substance not named in this dataset — which is most substances. There are tens of thousands of marketed drugs and this table holds fewer than a hundred entries.
Phase-2 conjugation (UGT, SULT, NAT2, COMT) except where a specific entry names it. The oilahuasca corpus turns heavily on phase 2 and this engine models it only in passing.
Pharmacogenomics. CYP2D6 and CYP2C19 are strongly polymorphic; a poor metaboliser and an ultra-rapid metaboliser can have opposite outcomes from the same pair, and this engine does not know your genotype.
Dose, timing, duration, formulation and route — all of which change whether a documented interaction is clinically real for you.
Renal and hepatic impairment, age, pregnancy, and body composition.
Bleeding and antiplatelet risk, hypoglycaemia, anticholinergic load, and most other pharmacodynamic axes beyond the ones listed above. Additive CNS depression and GABA-A modulation ARE now modelled — see the covers list — but the absence of a sedation finding still only means the agents you named are not on that axis in this dataset.
Herb–herb interactions outside the named entries, and essentially the whole botanical world: most plants have no interaction literature at all.
Allergy, intolerance, and contamination or adulteration of unregulated products.
Anything published after the last-reviewed date below.
Bleeding and antiplatelet risk, which is the mechanism that matters most for garlic, ginkgo and several other common supplements. It is not modelled at all, so a clean result says nothing about it.
Whether any of the natural FAAH, MAGL or transport inhibition reported in vitro occurs at all at a dose a person would take. For most of these compounds nobody has measured it.
The actual contents of an unregulated cannabinoid product. This engine models named compounds; an unidentified isomer or side-product in a converted-cannabinoid product is outside it by construction.
Dose. Every cannabinoid interaction here scales with dose, and consumer product labelling for this category is repeatedly found inaccurate in published surveys.
Inhalation-specific hazards — thermal degradation products, diluents chosen for rheology rather than for inhalation toxicology, and carrier and adulterant contamination.
101 substances, 33 mechanisms,
101 citations. Last reviewed .
Primary literature (every DOI resolved against the Crossref API) and FDA drug labelling. There is no free, openly-licensed, comprehensive drug-interaction dataset to draw on; NLM retired its Drug Interaction API on 2024-01-02 and DrugBank's interaction set is a commercial licence.
References
Jiang R, Yamaori S, Okamoto Y, Yamamoto I, Watanabe K (2013). Cannabidiol is a potent inhibitor of the catalytic activity of cytochrome P450 2C19. Drug Metabolism and Pharmacokinetics. doi:10.2133/dmpk.DMPK-12-RG-129
Geffrey AL, Pollack SF, Bruno PL, Thiele EA (2015). Drug-drug interaction between clobazam and cannabidiol in children with refractory epilepsy. Epilepsia. doi:10.1111/epi.13060
U.S. Food and Drug Administration (2018). EPIDIOLEX (cannabidiol) oral solution — prescribing information, including the transaminase-elevation and clobazam interaction sections. FDA Structured Product Labeling. https://www.accessdata.fda.gov/scripts/cder/daf/
Brown JD, Winterstein AG (2019). Potential Adverse Drug Events and Drug-Drug Interactions with Medical and Consumer Cannabidiol (CBD) Use. Journal of Clinical Medicine. doi:10.3390/jcm8070989
Stott C, White L, Wright S, Wilbraham D, Guy G (2013). A phase I, open-label, randomized, crossover study in three parallel groups to evaluate the effect of Rifampicin, Ketoconazole, and Omeprazole on the pharmacokinetics of THC/CBD oromucosal spray in healthy volunteers. SpringerPlus. doi:10.1186/2193-1801-2-236
Every DOI above was resolved against the Crossref API on 2026-09-09 and the returned
title checked against the one printed here. Three DOIs in the first draft resolved to real but different
papers and were corrected before publication.