Cannabidiol Probably Reduces Seizures in Drug-Resistant Epilepsy—But Serious Side Effects Also Rise
Cannabidiol Probably Reduces Seizures in Drug-Resistant Epilepsy, But Serious Side Effects Also Rise
A 2025 systematic review and meta-analysis of seven randomized controlled trials finds moderate-certainty evidence that cannabidiol at standard doses approximately doubles the proportion of patients achieving meaningful seizure reduction, while the higher dose also approximately doubles the risk of serious adverse events, underscoring an efficacy-safety trade-off that must guide clinical decisions.
Why This Matters
Treatment-resistant epilepsy affects roughly one-third of all people living with epilepsy and carries substantially elevated risks of morbidity, mortality, and diminished quality of life. Cannabidiol has emerged as one of the few genuinely novel therapeutic options for this population, but clinicians and patients need more than anecdotal enthusiasm; they need rigorously synthesized trial data with transparent certainty ratings. This systematic review arrives at a pivotal moment, as regulatory pathways and clinical guidelines for cannabidiol in epilepsy continue to evolve across multiple jurisdictions, making its balanced accounting of both benefit and harm directly relevant to prescribing decisions being made right now.
Clinical Summary
Drug-resistant epilepsy, defined by failure of two or more appropriately chosen and tolerated antiseizure medications, remains one of the most challenging problems in clinical neurology. Cannabidiol exerts its anticonvulsant effects through mechanisms distinct from conventional antiseizure drugs, including modulation of intracellular calcium via GPR55 and TRPV1 receptors, desensitization of voltage-gated sodium channels, and enhancement of GABAergic inhibition. This systematic review and meta-analysis, conducted according to Cochrane Handbook and PRISMA 2020 standards and registered prospectively on PROSPERO, synthesized data from seven randomized controlled trials comparing pharmaceutical-grade cannabidiol to placebo in populations with treatment-resistant epilepsy, including Dravet syndrome and Lennox-Gastaut syndrome.
The pooled results demonstrate that cannabidiol at 20 mg/kg/day probably increases the proportion of patients achieving 50% or greater monthly seizure reduction compared with placebo (RR 1.92; 95% CI 1.49 to 2.46; four RCTs, n=575, moderate certainty), with a nearly identical effect size at 10 mg/kg/day (RR 1.94; 95% CI 1.32 to 2.86; two RCTs, n=280, moderate certainty). However, cannabidiol at 20 mg/kg/day also probably increases serious adverse events (RR 2.30; 95% CI 1.36 to 3.89; four RCTs, n=583, moderate certainty), while the 10 mg/kg/day dose showed a non-significant trend toward harm (RR 1.62; 95% CI 0.92 to 2.84, low certainty). The review is limited by the small number of included trials, heterogeneous epilepsy subtypes across studies, and the absence of long-term effectiveness or quality-of-life data. The authors emphasize that additional high-quality trials are needed before definitive dose-optimization guidance or population-specific recommendations can be made.
Dr. Caplan’s Take
This review does something genuinely useful: it forces the conversation about cannabidiol and epilepsy to be honest. The efficacy signal is real and consistent across both doses, which is encouraging. But the safety signal at the higher dose is equally real. Patients and families who come to me having read that “CBD treats seizures” often expect a clean, benign intervention. What the evidence actually supports is a pharmacologically active compound with a meaningful benefit-risk trade-off that demands the same careful deliberation we apply to any antiseizure medication.
In practice, when I see patients with treatment-resistant epilepsy who are interested in cannabidiol, I discuss pharmaceutical-grade CBD specifically, not over-the-counter supplements. I review the moderate-certainty evidence for seizure reduction, and I am equally explicit about the elevated risk of serious adverse events at 20 mg/kg/day and the need for liver function monitoring given known hepatotoxicity risks, particularly with concomitant valproate. If a trial of cannabidiol is appropriate, I start at the lower dose and titrate with close clinical and laboratory follow-up.
Clinical Perspective
This review consolidates what the individual pivotal trials suggested but places those findings within a formal certainty framework. The approximately twofold increase in 50% seizure responders at both doses confirms a robust directional effect, but the moderate GRADE certainty rating reminds clinicians that this estimate could shift with future trials, particularly in underrepresented epilepsy subtypes or age groups. Critically, the evidence base for all other cannabis derivatives and synthetic cannabinoid analogs remains very low to low certainty, meaning clinicians should not extrapolate these CBD-specific findings to other cannabinoid products. Absolute risk differences, which are essential for shared decision-making conversations, are not prominently reported in the available data, representing a practical gap for clinicians attempting to translate these numbers at the bedside.
From a pharmacological safety standpoint, the hepatotoxicity signal with cannabidiol is well established and is amplified by concomitant valproate use, necessitating baseline and periodic liver function testing in all patients. Drug-drug interactions mediated through CYP3A4 and CYP2C19 pathways can alter levels of clobazam (increasing its active metabolite norclobazam) and other antiseizure medications, complicating polypharmacy regimens. One actionable recommendation clinicians can implement now is to preferentially initiate cannabidiol at 10 mg/kg/day rather than 20 mg/kg/day, given comparable efficacy point estimates but a more favorable safety profile at the lower dose, and to escalate only if clinical response is inadequate and tolerability permits.
Study at a Glance
- Study Type
- Systematic review and meta-analysis of randomized controlled trials
- Population
- Children and adults with treatment-resistant epilepsy, including Dravet syndrome and Lennox-Gastaut syndrome
- Intervention
- Pharmaceutical-grade cannabidiol at 10 mg/kg/day and 20 mg/kg/day
- Comparator
- Placebo
- Primary Outcomes
- Proportion achieving 50% or greater reduction in monthly seizure frequency; serious adverse events
- Sample Size
- 7 RCTs; up to 583 participants in pooled 20 mg/kg/day analyses, up to 280 in pooled 10 mg/kg/day analyses
- Journal
- Not specified in extracted data; Cochrane-standard systematic review published 2025
- Year
- 2025
- DOI or PMID
- PROSPERO registration CRD42024508610
- Funding Source
- Not specified in extracted data
What Kind of Evidence Is This
This is a prospectively registered systematic review and meta-analysis of randomized controlled trials, conducted following Cochrane Handbook methodology with PRISMA 2020 reporting standards, risk of bias assessment using the Cochrane tool, and certainty grading via the GRADE framework. It sits near the top of the evidence hierarchy for intervention questions. However, its conclusions are bounded entirely by the quality, number, and representativeness of the seven included RCTs, and seven trials is a small base for generalizing across a heterogeneous condition like treatment-resistant epilepsy.
How This Fits With the Broader Literature
This review consolidates and formally grades what the landmark Dravet syndrome trial by Devinsky and colleagues (2017) and the Lennox-Gastaut syndrome trials by Devinsky et al. (2018) and Thiele et al. (2018) individually demonstrated: cannabidiol has a statistically and clinically meaningful anticonvulsant effect in specific drug-resistant epilepsy syndromes. The current synthesis extends that work by stratifying outcomes by dose and applying GRADE certainty ratings, providing a more structured basis for clinical recommendations than any single trial could.
The safety findings align with post-marketing surveillance data and the FDA prescribing information for Epidiolex, which highlight hepatotoxicity and somnolence as dose-dependent concerns. Notably, the evidence gap for non-CBD cannabinoids identified here mirrors the conclusions of prior Cochrane reviews, confirming that the broader cannabinoid literature for epilepsy remains too sparse and heterogeneous to support clinical use outside of pharmaceutical-grade CBD.
Common Misreadings
The most likely overinterpretation is treating the approximately twofold seizure response rate as evidence that cannabidiol is uniformly effective and well tolerated in all treatment-resistant epilepsy. This exceeds what the data support on several fronts. First, the confidence intervals at both doses, while favoring CBD, still carry meaningful imprecision given the small trial base. Second, the concurrent doubling of serious adverse events at 20 mg/kg/day is not a minor footnote but a central finding of equal weight. Third, these results derive primarily from Dravet and Lennox-Gastaut populations and cannot be generalized to all forms of drug-resistant epilepsy without additional evidence.
Bottom Line
This systematic review provides moderate-certainty evidence that pharmaceutical-grade cannabidiol at 10 and 20 mg/kg/day probably reduces seizure frequency in treatment-resistant epilepsy, but the higher dose also probably increases serious adverse events. It does not establish long-term safety, optimal dosing, or effects on quality of life. For now, this evidence supports cautious, closely monitored use of CBD as adjunctive therapy in specific epilepsy syndromes, with a preference for initiating at the lower dose.
References
- Devinsky O, Cross JH, Laux L, et al. Trial of cannabidiol for drug-resistant seizures in the Dravet syndrome. N Engl J Med. 2017;376(21):2011-2020. doi:10.1056/NEJMoa1611618
- Devinsky O, Patel AD, Cross JH, et al. Effect of cannabidiol on drop seizures in the Lennox-Gastaut syndrome. N Engl J Med. 2018;378(20):1888-1897. doi:10.1056/NEJMoa1714631
- Thiele EA, Marsh ED, French JA, et al. Cannabidiol in patients with seizures associated with Lennox-Gastaut syndrome (GWPCARE4): a randomised, double-blind, placebo-controlled phase 3 trial. Lancet. 2018;391(10125):1085-1096. doi:10.1016/S0140-6736(18)30136-3
- PROSPERO International Prospective Register of Systematic Reviews. Registration CRD42024508610. Available at: https://www.crd.york.ac.uk/prospero/
- Higgins JPT, Thomas J, Chandler J, et al., eds. Cochrane Handbook for Systematic Reviews of Interventions. Version 6.4. Cochrane, 2023. Available at: https://training.cochrane.org/handbook
- Guyatt GH, Oxman AD, Vist GE, et al. GRADE: an emerging consensus on rating quality of evidence and strength of recommendations. BMJ. 2008;336(7650):924-926. doi:10.1136/bmj.39489.470347.AD
Have thoughts on this? Share it:
