Cannabis extracts are playing an increasingly important role in clinical discussions, whether they are CBD-rich formulations, products containing THC, or combination extracts. For prescribers, the issue goes beyond potential efficacy: the question of drug interactions and the clinical safety of cannabis extracts becomes central as soon as a patient is already receiving chronic treatment, especially if that treatment relies on medications with a narrow therapeutic index.
Recent data call for a nuanced interpretation. On the one hand, human evidence remains limited compared to the volume of mechanistic hypotheses. On the other hand, documented interactions are very real, particularly with certain antiepileptics, anticoagulants, and immunosuppressants. In practice, caution does not mean systematically rejecting these products; it primarily involves better anticipating, monitoring, and standardizing the clinical approach.
Why interactions with cannabis extracts are a concern in practice
The starting point is simple: cannabinoids can alter the metabolism of other medications. Cannabis extracts containing primarily THC and/or CBD are likely to interact via several major liver enzymes, including CYP3A4, CYP2C9, and CYP2C19. This mechanism explains why the topic is of such interest to prescribers, pharmacists, and monitoring teams.
Recent reviews show that the risk is not uniform for all patients. It becomes particularly concerning when concomitant treatment has a narrow therapeutic index, meaning that small variations in concentration can lead to either a loss of efficacy or toxicity. In this context, even a moderate interaction can have disproportionate clinical consequences.
It is also important to bear in mind that the number of published cases remains relatively small compared to the number of proposed theoretical mechanisms. A recent systematic review identified 31 reports involving 889 people and 603 users of cannabis or cannabinoids. This figure does not justify alarmism, but it does confirm that the risk is neither purely speculative nor negligible.
CBD and THC: same families, different interaction profiles
CBD is currently the most closely monitored cannabinoid in terms of interactions. Recent data indicate that it inhibits CYP2C19, CYP3A4, and CYP3A5 , as well as other metabolic pathways. In short, it can increase the exposure to concurrently administered medications, with a potentially significant impact on patients taking multiple medications.
THC can also interact with several enzymes, particularly CYP3A4, CYP3A5, CYP2C9 , and CYP2C19 . However, the clinical evidence is generally less robust than for CBD, and some data suggest that the concentrations required for a marked inhibitory effect may be higher than those observed with typical use. This does not mean there is no risk, but rather a higher level of uncertainty.
For clinicians, this distinction is crucial. An extract described as “mild” or “well-tolerated” is not necessarily harmless from a pharmacokinetic standpoint, especially if it is high in CBD. Hence the importance of knowing the precise composition of the product used, ideally confirmed by laboratory analysis, as the actual dose of cannabinoids largely determines the level of vigilance required.
High-risk medications to identify before prescribing
Before introducing a cannabis extract, best practice includes conducting a complete medication review. The goal is to identify treatments most sensitive to hepatic metabolism, particularly those highly dependent on CYP3A4, CYP2C9, or CYP2C19. This step should be performed routinely at the start of treatment, as well as when increasing, reducing, or discontinuing the cannabinoid product dose.
The classes of drugs to monitor most closely are well-known from recent reviews: anticoagulants, antiepileptics, and immunosuppressants. These medications account for a large proportion of published clinical signals. Antiepileptics, in particular, represent the class most frequently encountered in recent analyses of interactions related to medical cannabis.
In practice, the greatest vigilance is required for molecules with a narrow therapeutic index. For these treatments, even small changes in exposure can lead to excessive sedation, bleeding, graft rejection, loss of disease control, or other major adverse events. The risk therefore depends less on the general category of cannabis than on the precise combination of the extract used and the associated medication.
Best-documented clinical interactions: clobazam, warfarin, tacrolimus
The combination of clobazam and CBD is among the most robustly described interactions. Significant increases in clobazam concentrations, and especially in its active metabolite, N-desmethylclobazam, have been observed with high doses of CBD. The likely mechanism involves the inhibition of CYP2C19 and CYP3A4, with a risk of increased sedation and other neurological adverse effects.
Warfarin is another major point of concern. Recent literature reports cases where cannabis or cannabinoids altered the anticoagulant effect, warranting close monitoring of the INR. In these patients, the introduction or discontinuation of a cannabis extract may necessitate dose adjustment and closer clinical monitoring to limit the risk of bleeding or thrombosis.
Tacrolimus animmunosuppressant with a narrow therapeutic index, is also among the high-risk combinations. Any variation in exposure can have significant consequences, whether toxicity or sub-immunosuppression. In this type of situation, the safest approach relies on close coordination between the prescriber, pharmacist, and referring specialist, with concentration monitoring if available.
Other sensitive situations: clopidogrel, DOACs and valproate
Clopidogrel raises a more theoretical but credible risk. Since its activation depends on CYP2C19 , inhibition of this enzyme by CBD could reduce its antiplatelet effect. Even in the absence of a large number of conclusive clinical studies, this mechanism warrants increased caution in patients at high cardiovascular risk.
Direct oral anticoagulants also deserve special attention, particularly those that are partly dependent on CYP3A4 . Recent clinical guidelines cite them among the treatments to monitor when cannabinoids are introduced. Here again, the aim is not to conclude that there is an absolute contraindication, but to avoid trivializing the risk.
Regarding valproate, a specific signal has been reported with CBD: in a phase 2 trial, a decrease in exposure to valproate and its metabolite was observed, accompanied in some subjects by an elevation of liver enzymes. This association serves as a reminder that an interaction is not always simply a matter of "more medication in the blood"; it can also manifest as altered tolerance, particularly hepatic tolerance.
Liver, transaminases and biosafety: what to monitor
Liver toxicity is an important aspect of the clinical safety of CBD-rich extracts. Several recent publications recommend considering transaminase monitoring when CBD is used in combination with other hepatotoxic drugs or treatments extensively metabolized by the liver. This precaution is particularly relevant at the start of treatment and during dose increases.
The liver signal does not mean that all CBD products are harmful to the liver. It primarily serves as a reminder that context is crucial: the dose used, the duration of exposure, the patient's individual characteristics, history of liver problems, alcohol consumption, and other medications being prescribed. In a frail or polymedicated patient, initial blood tests followed by targeted monitoring can significantly improve safety.
In practical terms, clinical symptoms alone are insufficient for the early detection of liver damage. Fatigue, nausea, or loss of appetite can be nonspecific. Hence the importance of integrating laboratory tests into the monitoring strategy when the patient's profile warrants it, rather than waiting for the appearance of more pronounced symptoms.
Route of administration, bioavailability and product quality
The risk of interaction also depends on the route of administration. Bioavailability varies considerably depending on whether it is an oil, a capsule, a sublingual extract, or another form. Recent reviews emphasize this point: the actual potency of the product and the route used greatly complicate risk assessment, as systemic exposure to cannabinoids can change from one presentation to another.
Added to this is a major problem of standardization. In practice, not all cannabis extracts are created equal: the THC and CBD content can vary significantly depending on the manufacturer, the batch, and the quality of the analytical testing. For a prescriber, this variability makes predicting interactions more difficult than with a well-standardized conventional medication.
This is where a laboratory-tested product , with a clearly displayed composition and legal compliance, becomes a real safety issue, and not just a marketing ploy. When the actual concentration is known, the dialogue between patient and healthcare professional becomes more precise: exposure can be better estimated, monitoring adjusted, and pharmacological uncertainties reduced.
Concrete guidelines for prescribers when initiating, modifying or stopping
The first rule is to conduct a complete review of treatments before prescribing or recommending any cannabis extract. This review must include prescribed medications, self-medication, supplements, and any current use of cannabinoid products. Many interactions go unnoticed simply because the patient doesn't think to mention CBD oil or other extracts purchased outside of medical channels.
The second rule is to increase monitoring whenever is initiated, increased, reduced, or discontinued . Recent sources emphasize that these changes can rapidly alter the clinical balance of concomitant treatment. In other words, a patient stabilized on an anticoagulant, antiepileptic, or immunosuppressant can become “at risk” as soon as a cannabinoid is added or removed.
Finally, it must be accepted that risk assessment still largely relies on a combination of in vitro, clinical cases, and pharmacological caution. Human evidence from 2024 and 2025 remains incomplete, but it is sufficient to justify a structured approach: start low, document the product composition, monitor relevant parameters, and regularly reassess the benefit-risk ratio.
In summary, the drug interactions and clinical safety of cannabis extracts should neither be dramatized nor minimized. The core of the problem lies primarily with CBD-rich extracts, treatments with a narrow therapeutic index, and situations where product quality is poorly understood. While the documented cases are not numerous, they are sufficient to necessitate strict prescribing practices.
For both prescribers and patients, the best strategy remains transparency and methodology: understanding the THC/CBD profile of the product, prioritizing analyzed extracts, reviewing the complete medication list, and organizing targeted follow-up. This approach makes it possible to reconcile therapeutic openness, clinical safety, and more responsible use of cannabinoids in daily practice.
