Adjunctive cannabidiol in intractable pediatric epilepsy: A retrospective study on tolerability, efficacy, and safety across genetic and nongenetic etiologies

Abstract

This retrospective cohort study evaluated the tolerability, efficacy, and safety of adjunctive cannabidiol (CBD) therapy in pediatric-onset intractable epilepsy across diverse genetic and nongenetic etiologies. Twenty-nine patients aged 6 to 24 years, treated at Korea University Hospitals between April 2019 and May 2024, were included. The median follow-up duration was 14.3 months. Confirmed genetic etiologies included SCN1A-related epilepsy (6.9%); GABRB3-, SCN2A-, KCNT1-, KIF1A-, and COL4A1-related epilepsies (3.4% each); Angelman syndrome and Down syndrome (3.4% each). Presumed genetic etiologies included hemimegalencephaly (3.4%) and cortical dysplasia (6.9%). Acquired causes included hypoxic brain injury (6.9%) and CNS infection (10.3%). In 41.4% of cases, the etiology was unidentified; among them, 58.3% had a history of infantile spasms. At CBD initiation, patients were receiving a median of 5 antiseizure medications, most commonly valproic acid (93.1%), clobazam (82.8%), and levetiracetam (75.9%). The median maintenance dose of CBD was 14.2 mg/kg/d. The retention rate was above 86% at both 12 and 24 months. At 12 months, 79.3% achieved a ≥50% reduction in seizure frequency, and 34.5% achieved a ≥75% reduction without generalized motor seizures. One patient with a GABRB3 variant achieved seizure freedom. Adverse events occurred in 37.9%, most commonly somnolence and lethargy. These were mild and resolved with antiseizure medication adjustments. CBD was discontinued in 3 patients due to pneumonia, lethargy, or seizure aggravation. CBD therapy demonstrated favorable retention, efficacy, and safety profiles in pediatric-onset intractable epilepsy across a spectrum of etiologies.

1. Introduction

Despite the ongoing development of novel antiseizure medications (ASMs), a considerable proportion of patients with pediatric-onset epilepsy continues to manifest intractable seizures.^[1] Approximately 30% of pediatric epilepsy cases are refractory to ASMs, representing a significant therapeutic challenge.^[2] In recent years, cannabidiol (CBD), a nonpsychoactive compound derived from the Cannabis sativa plant, has gained increasing attention as a potential therapeutic agent for managing treatment-resistant seizures, particularly in pediatric populations.^[3] Unlike tetrahydrocannabinol, CBD lacks psychoactive effects and exerts its antiepileptic actions through several complex mechanisms.^[4] These proposed mechanisms include modulation of the endocannabinoid system via CB1 and CB2 receptors, inhibition of anandamide reuptake, reduction of neuroinflammation, and suppression of excitatory neurotransmitter release, such as glutamate. Additionally, CBD appears to exert effects through the regulation of intracellular calcium levels and adenosine signaling, mediated by GPR55, TRPV1, and ENT1 receptors.^[5,6] However, Emerging evidence suggests that the multifactorial mechanisms of CBD may underlie its efficacy across a diverse range of epileptic syndromes, highlighting its potential for broader clinical application.^[7,8]

Given the multifaceted mechanisms of CBD, further clinical evaluation of its efficacy and safety across various etiologies of drug-resistant epilepsy is warranted – particularly in patients receiving multiple ASMs.

This investigation explores treatment strategies and clinical outcomes in pediatric patients with intractable epilepsy of diverse etiologies following CBD therapy, with a particular emphasis on optimizing concomitant ASM regimens and managing adverse events.

2. Methods

2.1. Study design and participants

This retrospective cohort study reviewed medical records of pediatric-onset epilepsy patients who were treated with CBD at Korea University Hospitals between April 2019 and May 2024. To minimize selection bias, we implemented strict inclusion and exclusion criteria. Our cohort included epilepsy patients, with seizure onset before 18 years of age, who had not achieved seizure control despite treatment with at least 3 appropriate ASMs for a minimum of 12 months. Patients with incomplete or insufficient follow-up records were excluded from the study.

2.2. Data collection

Data were collected on various variables to comprehensively assess each patient’s clinical profile and treatment response. These variables included patient demographics, such as age at seizure onset, sex, and age at CBD initiation. Diagnostic information was recorded, specifying the underlying etiology. Monthly seizure frequencies were obtained from caregivers’ reports at each hospital visit. Patients lost to follow-up were accounted for in the retention rate calculation. The use of concomitant ASMs was documented, including the number and dosage of prescribed ASMs. Additional seizure control treatments, including a ketogenic diet, vagus nerve stimulation, and epilepsy surgery, were also documented. CBD treatment data were collected, including the initial prescribed dosage, subsequent adjustments, overall therapy duration, and patient response in terms of efficacy and adverse events.

2.3. Treatment protocol and response

Patients were prescribed CBD (Epidiolex^®) with a target starting dose of 10 mg/kg/d. In clinical practice, some patients required gradual titration to reach this target dose, and dosage adjustments were made according to individual clinical response and tolerability. Follow-up visits were initially scheduled at 2- to 4-week intervals. During these visits, the dosages of CBD and concomitant ASMs were evaluated and adjusted to optimize therapeutic outcomes. Seizure reduction was assessed through caregiver reports collected at each hospital visit. The primary outcome measure was the retention rate at 6, 12, and 24 months following the initiation of CBD therapy. Secondary outcomes included change in seizure burden and the occurrence of adverse effects after the initiation of CBD. Seizure frequency was recorded as the total number of seizures per month and expressed as the percentage reduction compared to baseline. Baseline seizure frequency was defined as the average monthly seizure frequency over the 6 months preceding CBD initiation. A moderate clinical response was defined as a reduction in total seizure frequency of more than 50%, while a significant response was defined as a reduction of more than 75% in the absence of generalized motor seizures. Seizure frequency reduction was assessed at 6-, 12-, and 24 months following CBD initiation. All adverse effects observed after the initiation of CBD were documented. Additionally, changes in clobazam dosage and its role as adjunct therapy during CBD treatment were monitored. A dosage changes of more or <20% was considered a clinically meaningful change. The use of other concomitant ASMs was also reviewed.

2.4. Statistical analysis

Descriptive statistics were used to summarize demographic and clinical characteristics. Continuous variables were reported as medians with minimum to maximum range, and categorical variables were expressed as frequencies and percentages. Treatment retention rates at 6, 12, and 124 months were calculated using Kaplan–Meier survival. Data from patients lost to follow-up were excluded from each period when the retention rate was analyzed. The data were analyzed using IBM SPSS Statistics for Windows (version 30.0; IBM Corp., Armonk).

3. Results

3.1. Patient demographics

This study included 29 patients aged 6 to 24 years (median: 13 years). Of the patients, 12 (41.4%) were female. The median age at seizure onset was 7.5 months (range, 0–72 months). Confirmed genetic etiologies included SCN1A-related epilepsy (n = 2, 6.9%), Down syndrome (n = 1, 3.4%), Angelman syndrome (n = 1, 3.4%), and other genetic epileptic encephalopathies (n = 5, 17.2%) caused by variants in SCN2A (n = 1, 3.4%), GABRB3 (n = 1, 3.4%), KCNT1 (n = 1, 3.4%), COL4A1 (n = 1, 3.4%), and KIF1A (n = 1, 3.4%). Other patients had presumed genetic causes due to congenital brain structural abnormalities, including hemimegalencephaly (n = 1, 3.4%) and cortical dysplasia (n = 2, 6.9%). Acquired etiologies included sequalae of hypoxic brain insult (n = 2, 6.9%) and central nervous system infection (n = 3, 10.3%). The cohort also included unidentified causes (n = 12, 41.4%), of whom 7 (7/12, 58.3%) had a prior diagnosis of infantile spasms. Of the cohort, 27 patients were clinically labeled as Lennox–Gastaut syndrome but did not fulfill the strict electroclinical diagnostic criteria; 2 patients with SCN1A variants were diagnosed with Dravet syndrome.

Patients were on a median of 5 ASMs (range: 3–6). More than 75% of patients were using valproic acid (n = 27, 93.1%), clobazam (n = 24, 82.8%), or levetiracetam (n = 22, 75.9%), with 15 (51.7%) taking all 3 as part of their ASM regimen. Additional treatments, including a ketogenic diet (n = 6, 20.7%), vagus nerve stimulation (n = 6, 20.7%), and epilepsy surgeries (lobectomy or callosotomy; n = 6, 20.7%), were administered before CBD initiation (Table 1).

Table 1.

Demographics of patients.

Variable
Total, N	29
Sex
Male, n (%)	17 (58.6)
Female, n (%)	12 (41.4)
Seizure onset age, median month (range)	7 (0–72)
Baseline total seizure frequency median (range) per month	90 (1–200)
Cannabidiol initiation age, median (range)	10.1 (3.4–21.8)
Follow-up duration, median month (range)	14.3 (3.8–56.9)
Diagnosis
Angelman syndrome, n (%)	1 (3.8)
Down syndrome, n (%)	1 (3.8)
COL4A1, n (%)	1 (3.8)
GABRB3, n (%)	1 (3.8)
KCNT1, n (%)	1 (33.3)
KIF1A, n (%)	1 (33.3)
SCN1A, n (%)	2 (7.7)
SCN2A, n (%)	1 (33.3)
History of west syndrome, n (%)	7 (26.9)
Idiopathic early onset seizures, n (%)	5 (19.2)
Congenitial brain structural abnormality, n (%)	3 (10.3)
Sequalea of hypoxic insult, n (%)	2 (7.7)
Sequalae of central nervous system infection, n (%)	3 (11.5)
Treatment
Median number of antiseizure medications (range)	5 (3–6)
Valproic acid, n (%)	27 (93.1)
Clobazam, n (%)	24 (82.8)
Levetiracetam, n (%)	22 (75.9)
Topiramate, n (%)	20 (69.0)
Lamotrigine, n (%)	11 (37.9)
Ketogenic diet, n (%)	6 (20.7)
Vagus nerve stimulation, n (%)	6 (20.7)
Epilepsy surgery, n (%)	6 (20.7)
Lobectomy, n (%)	2 (7.7)
Callosotomy, n (%)	4 (15.4)

3.2. Efficacy and adverse events

The median initial CBD dose was 4.0 mg/kg/d (range, 1.9–10.0 mg/kg/d), indicating that while 10 mg/kg/d was the recommended target starting dose, many patients began at lower doses and required individualized titration. The median time required to reach the maintenance dose was 1.2 months (range, 0.7–26 months), depending on clinical response and tolerability. The median CBD dose was 14.2 mg/kg/d (range, 7.8–21.7 mg/kg/d), and the median clobazam dose was 0.3 mg/kg/d (range, 0.1–0.6 mg/kg/d). The retention rate showed 92.3% at 6 months, 86.9% at 12 months, and 86.3% at 24 months (Fig. 1). At the 12-month follow-up, 23 patients (79.3%) demonstrated at least a moderate response. Notably, 10 patients (34.5%) showed a significant response, including 1 patient (3.45%) with a GABRB3 pathogenic variant who achieved seizure freedom at 12-month.

Figure 1.

Retention rate of cannabidiol (N = 29). Retention rates are illustrated using a Kaplan–Meier curve, depicting the proportion of patients remaining on cannabidiol treatment over time, with follow-up assessments at 6, 12, and 24 months.

Adverse events were observed in a subset of patients, including somnolence (n = 6, 20.7%), lethargy (n = 2, 6.9%), ataxia (n = 1, 3.4%), leg weakness (n = 1, 3.4%), elevated liver enzymes (n = 1, 3.4%), and elevated ammonia level (n = 1, 3.4%). These were considered mild treatment-emergent adverse events (TEAEs), and either recovered spontaneously (n = 1) or were managed through dosage adjustments of concomitant ASMs (n = 10). Other TEAEs led to CBD discontinuation in 3 (10.3%) patients. One patient with KIF1A-related epilepsy discontinued therapy due to recurrent pneumonia; another with cortical dysplasia withdrew due to severe lethargy; and a third with SCN2A-related epilepsy ceased treatment following an increase in seizure frequency.

At the initiation of CBD treatment, 24 (82.8%) patients were receiving clobazam, with a median dosage of 0.3 mg/kg/d. Clobazam was newly introduced in 2 (6.9%) patients at the onset of CBD therapy. During the follow-up period, a dosage increases of more than 20% was observed in 6 (20.7%) patients. In contrast, clobazam was dose-reduced in 2 patients due to ataxia (n = 1) and leg weakness (n = 1) and discontinued in 1 patient due to somnolence. The dosage of valproic acid was reduced in 4 patients due to somnolence (n = 2), elevated liver enzymes (n = 1), and hyperammonemia (n = 1). Perampanel was concurrently reduced with valproic acid in 1 patient due to ataxia. Oxcarbazepine was switched to eslicarbazepine acetate in 1 patient to address somnolence, and phenobarbital was also dose-reduced for the same reason (Table 2).

Table 2.

Detailed clinical characteristics of all patients treated with cannabidiol.

No.	Sex	Etiology	TEAEs during CBD therapy	CLB dosage change ≥ 20%	ASM adjustment during CBD therapy	Last CBD dosage (mg/kg/d)	Baseline total seizure frequency per month	Total seizure frequency per month	Follow-up (mo)	CBD response at 1 yr
1	M	Angelman syndrome	–	↑	–	17.8	90	45	23.5	Moderate
2	F	Down syndrome	–	↑	–	10.4	30	0.9	25.8	Significant
3	M	COL4A1	↑ Liver enzyme	–	VPA↓	20.3	150	9	34.2	Moderate
4	M	GABRB3	–	↑	–	13.6	4	<1	28.8	Seizure free
5	F	KCNT1	↑ Ammonia	–	VPA↓	8.3	240	<1	26.4	Significant
6	M	KIF1A	Pneumonia	↓	–	10.0	30	4.2	7.9	Quit
7	F	SCN1A c.2243G>A	–	–	–	7.8	1	<1	31.9	Significant
8	F	SCN1A c.5570dupT	–	–	–	14.2	1	<1	32.6	Significant
9	M	SCN2A	↑ Seizure	–	–	25.0	60	>60	3.8	Quit
10	F	Infantile spasm	Somnolence	–	VPA↓, LCM quit	18.2	300	3	5.9	N/A
11	M	Infantile spasm	–	–	–	15.6	300	150	7.5	N/A
12	M	Infantile spasm	Ataxia	↓	PER↓	10.2	600	4.2	27.6	Moderate
13	M	Infantile spasm	–	Add	–	14.3	90	<1	30.5	Significant
14	M	Infantile spasm	–	↑	VPA↑, VGB↓	20.0	420	<100	27.8	Moderate
15	M	Infantile spasm	Somnolence	–	VPA↓, LEV↑	13.3	90	30	32	Moderate
16	M	Infantile spasm	Leg weakness	–	–	8.2	30	12.9	9.5	N/A
17	M	Neonatal-onset seizure	Somnolence	–	OXC → ESL	14.1	360	180	33.6	Moderate
18	F	Infantile-onset seizure	Somnolence	↓	–	14.8	90	9	52.5	Significant
19	M	Childhood-onset seizure	–	–	LTG add	15.2	300	150	34.7	Moderate
20	M	Childhood-onset seizure	–	–	LTG, TPM add	14.3	300	60	34.1	Moderate
21	F	Childhood-onset seizure	Somnolence	–	PB↓	11.1	60	8.7	31.2	Significant
22	F	Cortical dysplasia	Lethargy	↑	–	14.3	60	30	4.4	Quit
23	M	Cortical dysplasia	–	–	–	11.7	300	3	27.7	Significant
24	F	Hemimegalencephaly	–	↓	–	20.0	90	30	35.6	Moderate
25	F	Hypoxic insult	–	Add	–	21.7	180	90	29.7	Moderate
26	F	Hypoxic insult	–	–	–	12.1	120	15	52.1	Moderate
27	F	ADEM	Somnolence	Quit	PER, ZNS add	14.3	150	2.1	56.9	Significant
28	M	Neonatal meningitis	Lethargy	↑	–	16.9	600	60	25.3	Moderate
29	M	Encephalitis	–	↓	–	9.4	23	4.2	34.4	Moderate

ADEM = acute disseminated encephalomyelitis, CBD = cannabidiol, CLB = clobazam, ESL = eslicarbazepine acetate, LCM = lacosamide, LEV = levetiracetam, LTG = lamotrigine, OXC = oxcarbazepine, PB = phenobarbital, PER = perampanel, TEAE = treatment-emergent adverse event, VGB = vigabatrin, VPA = valproic acid, ZNS = zonisamide.

4. Discussion

CBD has demonstrated consistent efficacy and safety profiles in randomized controlled trials and multicenter open-label investigations.^[3,7,9-11] Although the FDA and EMA have approved CBD for LGS, DS, and TSC, several studies have reported the efficacy of CBD in different forms of genetic epilepsy or epileptic syndromes, including those associated with SCN1A, SLC13A5, CDKL5, SCN8A, KCNT1, SYNGAP1, MECP2, and Dup15q.^[12-14] Furthermore, recent studies of real-world data have shown the clinical usefulness and safety of CBD beyond current indications.^[7,11]

Our study further demonstrated tolerability and safety of CBD across diverse genetic and nongenetic etiologies of epilepsy. A comparatively high retention rates of were observed at both 12 and 24 months, achieved with a lower median maintenance dose of CBD (14.7 mg/kg/d), alongside mild TEAEs. Detailed investigation of each patient’s clinical course revealed that 14 (48.3%) patients required adjustments to clobazam, while 11 (37.9%) patients underwent modifications to other ASMs during continued CBD therapy. Specifically, clobazam was either newly initiated (n = 2) or up-titrated (n = 6) to enhance seizure control, or it was down-titrated (n = 5) or discontinued (n = 1) to manage emergent side effects following the initiation of CBD treatment. Previous studies have reported that 56% of patients receiving CBD were on clobazam, suggesting that clobazam may enhance the effectiveness of CBD in reducing seizures.^[15-17] CBD inhibits the metabolism of clobazam and its active metabolite, N-desmethylclobazam, via CYP3A4 and CYP2C19, respectively. This leads to increased plasma concentrations.^[15,16] In a mouse model of DS, pharmacodynamic and pharmacokinetic interactions between CBD and clobazam were found to have a synergistic effect on seizure reduction N-desmethylclobazam.^[17] A study of 396 patients with LGS and 318 patients with DS found CBD to be most effective in combination with clobazam.^[18] Although, CBD may exert a synergistic effect when coadministered with clobazam, recent evidence also supports its intrinsic antiseizure properties independent of clobazam.^[7,17] Based on our observations, we do not believe that higher doses of clobazam are necessarily required when used in conjunction with CBD. We assumed that clobazam may enhance the efficacy of CBD; however, it should be used cautiously in cases where adverse effects occur, as these may impact the long-term retention of CBD therapy.

While the specific effect of clobazam on the efficacy of CBD in our cohort remains unclear, active modification of concomitant ASMs may have contributed to the low discontinuation rate of CBD (6.9%), which is notably lower than the dropout rates of CBD therapy reported in previous studies (28.8%–36%).^[10,11,19] In particular, we observed that adjustments to ASMs sharing TEAE profiles with CBD – such as elevated liver enzymes, dizziness, somnolence, or ataxia – including valproic acid, phenobarbital, oxcarbazepine, lacosamide, and perampanel, may have contributed to the improved tolerability and treatment retention observed in our cohort. Despite the small sample size of our study, this approach is particularly important for maintaining treatment adherence and minimizing adverse effects, especially in patients receiving polytherapy.

Among our cohort, GABRB3-related epilepsy showed a notably favorable response to CBD. The GABRB3 gene encodes the β3 subunit of the GABA_A receptor and is implicated in a spectrum of neurodevelopmental and epileptic disorders. Disease-causing variants can lead to either gain-of-function (GoF) or loss-of-function (LoF) effects, each associated with distinct phenotypes. GoF variants are typically linked to severe, early onset epilepsy with poor treatment response,^[20] whereas LoF variants are often more responsive to GABAergic ASMs.^[21] However, GABA_A receptor subunit variants are associated with highly variable phenotypes despite their molecular and physiological proximity.^[22] This heterogeneity underscores the need for individualized therapeutic approaches in GABRB3-associated epilepsies. CBD has also been shown to modulate GABAergic transmission, although its interactions with specific GABA_A receptor subunits, such as those encoded by GABRB3, remain under investigation. CBD may influence GABAergic signaling by altering receptor activity or indirectly affecting inhibitory neurotransmission. Both of these mechanisms could be beneficial in managing seizure disorders related to GABRB3 variants. Further research is needed to determine whether this observation is consistent across other cases with the same genetic variant or unique to this case. Additionally, studies should investigate whether GoF and LoF variants of GABRB3 differ in their response to CBD and whether earlier CBD treatment influences its efficacy.

Although, GABRB3-related epilepsy showed a favorable response to CBD treatment in our cohort, patients with SCN2A- and KIF1A-related epilepsy did not demonstrate improvement. These findings should not be interpreted as definitive variant-specific responses, as other studies have reported positive outcomes with CBD in patients with SCN2A-related epilepsy.^[19] Even in large cohort studies, data on specific genetic syndromes and individual genetic variants remain limited, making it challenging to evaluate variant-specific effects of CBD. There is a need to accumulate more clinical data on patients with specific genetic variants, rather than broader syndromic diagnoses, to better understand the etiology-specific responses to CBD treatment. Until then, caution is warranted when predicting treatment response to CBD based on genetic diagnosis alone.

Our study had several limitations. First, the retrospective design and small sample size may limit the generalizability of our findings. Additionally, because we did not measure serum levels of CBD and concomitant ASMs, our ability to accurately correlate dosages with clinical outcomes are limited. Another limitation was the reliance on subjective caregiver reports to estimate seizure frequency, which may have introduced reporting bias.

5. Conclusion

CBD demonstrated additive antiseizure properties across diverse etiologies. It is advisable that concomitant ASMs sharing TEAE profiles with CBD be cautiously adjusted to minimize early discontinuation of CBD treatment, thereby potentially maximizing the long-term benefits of CBD therapy. With this consideration, CBD may be safely utilized across both genetic and nongenetic causes of intractable pediatric epilepsy.

Author contributions

Conceptualization: Jung Hye Byeon, Baik-Lin Eun.

Data curation: Youngkyu Shim, Baik-Lin Eun.

Formal analysis: Youngkyu Shim.

Investigation: Jung Hye Byeon.

Methodology: Youngkyu Shim, Jung Hye Byeon, Baik-Lin Eun.

Software: Youngkyu Shim.

Supervision: Dong Hwa Yang, Baik-Lin Eun.

Validation: Youngkyu Shim, Dong Hwa Yang, Jung Hye Byeon, Baik-Lin Eun.

Visualization: Youngkyu Shim.

Writing – original draft: Youngkyu Shim.

Writing – review & editing: Youngkyu Shim, Dong Hwa Yang, Jung Hye Byeon, Baik-Lin Eun.