A strong result in a narrow group
Patients in a selected European real-world cohort had substantial seizure reductions while their treatment was simplified around cenobamate. At the last visit, 25 of the 75 patients, or 33.3%, were seizure-free, and 51, or 68%, had at least a 50% reduction in seizure frequency. One patient, or 1.3%, had worsening. The figures describe what was observed alongside treatment simplification, not proof that cenobamate caused the outcomes.
This was not a snapshot of everyone who received the drug. The analysis included 75 of 694 patients with available Early Access Program data, or 10.7%. At the last visit, seven were taking cenobamate alone and 68 were taking it with one other antiseizure medicine. The selection favored patients who had achieved a major reduction in their medication burden, so the findings may not represent the wider cenobamate-treated population.
What the analysis measured
DECREASE pooled individual-patient data from real-world studies and registries linked to European cenobamate Early Access Programs, involving 41 centers. The main checks were seizure freedom, a response defined as at least a 50% fall in seizure frequency from baseline, and cumulative adverse events at the last visit. Cenobamate retention was a secondary outcome. Results were descriptive, based on available records, and missing values were not filled in.
Before treatment in this cohort, patients had used a median of 10 prior antiseizure medicines; the middle half had used between six and 13. The mean age was 39.5 years, with patients ranging from 19 to 65. At baseline, 46 patients were taking two other antiseizure medicines, 23 were taking three and six were taking four.
The monotherapy question remains open
The number of other antiseizure medicines fell significantly from treatment start to the last visit. By the final assessment, seven patients, or 9.3%, were on cenobamate alone, while 68, or 90.7%, were taking cenobamate with one other antiseizure medicine.
Those final regimens produced a numerical contrast, but not a settled answer. Seizure freedom was recorded in 57.1% of the monotherapy group and 30.9% of the group taking one other medicine. The rates with at least a 50% seizure reduction were 85.7% and 66.2%, respectively. Neither difference was statistically significant, with reported p-values of 0.16 for seizure freedom and 0.29 for response. Only seven patients were in the monotherapy group, so the figures cannot establish that monotherapy is superior.
An exploratory comparison by follow-up duration also showed higher rates in the longer-follow-up group. At 12 months, 20.0% of 30 patients were seizure-free and 56.7% had at least a 50% seizure reduction. Beyond 12 months, the corresponding figures among 43 patients were 41.9% and 76.7%. Because this was a post hoc, nonrandomized comparison, it does not show that longer treatment itself caused the higher rates.
Adverse events were common, but did not lead to stopping treatment
Adverse events were reported by 55 patients, or 73.3% of the cohort. The most frequently reported were somnolence at 30.7%, dizziness or vertigo at 28%, and fatigue at 26.7%. No patient discontinued cenobamate because of an adverse event. Events were reported in all seven monotherapy patients and in 48 of the 68 patients taking one other medicine, or 70.6%; that difference was not statistically significant, with a reported p-value of 0.179.
The median cenobamate dose at the last visit was 300 milligrams per day. The monotherapy group received a higher mean dose than the group taking one other medicine, 361 versus 277 milligrams per day, with a reported p-value of 0.026. That imbalance is one reason the regimen figures cannot be read as a clean test of the two strategies, since dose, tolerability and clinical decisions may also have differed.
Why the findings need a broader test
Follow-up covered a median of 22 months. Seventy-three patients, or 97%, had at least one year of follow-up, and one discontinued cenobamate at one year. The pooled studies differed in how they reported information, follow-up was not uniform, some data were missing, and the timing and reasons for reducing other medicines were not standardized. Because the study selected patients who had simplified treatment and had available follow-up, retention may be overestimated.
A separate exploratory analysis compared combinations used in at least five patients. Differences between combinations were statistically significant for seizure freedom and for the at-least-50% response measure. The highest seizure-free rates were seen with clobazam or lacosamide, while the highest response rates were seen with clobazam or carbamazepine. These subgroup findings are potentially confounded and need confirmation in larger, standardized studies; they do not establish that any one combination is superior.
The study was investigator-initiated and funded by Angelini Pharma. The authors report that the company had no role in the design, data collection, analysis, interpretation, manuscript preparation or data access. The Spanish Epilepsy Society and Italian Ministry of Health also provided support. The conflict statement lists author-specific honoraria and research-funding relationships with Angelini and other companies, while six named authors reported no conflicts.
The clearest reading is therefore modest: in a selected group already able to reduce other antiseizure medicines while taking cenobamate, many patients had fewer seizures at the last assessment, and adverse events did not lead anyone to stop the drug. The analysis cannot show that cenobamate caused those outcomes, that monotherapy is better, or that the same rates would apply outside this selected group. It also leaves the reasons and timing of medication reduction unresolved.
Paper data and sources
Original title: DECREASE study: A European pool-analysis of patients with significant reductions in concomitant antiseizure medications in cenobamate Early Access Programs.
Authors: Vicente Villanueva, Simona Lattanzi, Javier Peña-Ceballos et al.
Journal/Repository: Epilepsia open
Status: Peer-reviewed
First online: 2026-08-21
DOI: 10.1002/epi4.70344
Original paper