Fenfluramine cut seizures across a range of severe epilepsies in a registry study
A Spanish registry followed 166 patients with treatment-resistant epileptic encephalopathies for a year. Median seizure frequency fell 68.8%, though the design cannot match a placebo-controlled trial.
Fenfluramine, once withdrawn worldwide as a diet drug, has had an unlikely second life in epilepsy. Sold now as Fintepla by UCB, it is approved for two rare, severe childhood epilepsies — Dravet syndrome and Lennox-Gastaut syndrome — on the strength of placebo-controlled trials [s2]. A new study asks a different question: how the drug performs in ordinary clinical practice, across a broader mix of patients than those trials enrolled [s1].
The answer, from a Spanish registry, is broadly positive — but the way the study was done means its numbers should be read as description, not as trial-grade proof [s1].
What the study did
Researchers drew on the Spanish Epilepsy Society's registry of developmental and epileptic encephalopathies (DEEs), a group of conditions defined by drug-resistant seizures alongside developmental slowing or regression [s1]. This was a multicentre, retrospective, real-world analysis: patients had started fenfluramine at least 12 months before the database closed, and their charts were reviewed at 3, 6 and 12 months [s1].
The 166 patients spanned more than the two licensed indications. They included 84 with Lennox-Gastaut syndrome, 42 with Dravet syndrome, and 40 with other DEEs [s1]. Their mean age was 16.6 years (standard deviation 12.1), split between 111 children (median age 10.1 years) and 55 adults (median 26.5 years), and they had already tried and failed a median of three anti-seizure medications [s1]. Notably, the study was funded by the Spanish Epilepsy Society rather than the drug's manufacturer [s1].
What it found
At 12 months, the median fenfluramine dose was 0.49 mg/kg per day, and 77.1% of patients were still taking it [s1]. Median seizure frequency had fallen 68.8% from baseline (p < 0.001), with significant reductions in each diagnostic group — Lennox-Gastaut (p < 0.001), Dravet (p < 0.001) and other DEEs (p = 0.004) [s1]. At the one-year mark, 61% of patients had at least halved their seizures, and there was no statistically significant difference in response by age [s1].
The registry also tracked effects beyond seizure counts. The mean number of concomitant anti-seizure drugs fell significantly (p = 0.004), as did the proportion of patients needing rescue medication (p < 0.001) [s1]. Clinicians recorded meaningful improvement on a global-impression scale in cognition for 59.3% of patients, behaviour for 40.9%, sleep for 24.3%, and a caregiver domain for 44.8% [s1].
On tolerability, adverse events were reported in 68.1% of patients, mostly mild to moderate, and led to discontinuation in 12%, again without significant differences by age [s1]. The retention figure and the discontinuation rate pull in the same direction: most patients who started the drug were still on it a year later, which in a population that had already exhausted a median of three other medications is itself a signal that clinicians and families judged it worth continuing [s1].
How to read it
The strengths here are real: a sizeable cohort, a full year of follow-up, patients far more varied than a trial would admit, and funding from a professional society rather than the manufacturer [s1]. That combination is exactly what makes real-world evidence useful for gauging whether a drug's trial performance survives everyday use.
But the design caps how far the numbers can be pushed. This was retrospective and, crucially, uncontrolled — there was no placebo group, so a 68.8% median fall cannot be attributed to fenfluramine alone [s1]. Seizure counts fluctuate, regress toward the mean, and depend on caregiver diaries; concurrent changes to other medications, captured here as a drop in concomitant drugs, further muddy any single-cause reading [s1]. The softer outcomes — the global-impression ratings of cognition, behaviour and sleep — are clinician judgements without blinding, the kind most vulnerable to expectation [s1]. And extending an effectiveness signal to "other DEEs" beyond the two licensed indications is a use the controlled trials did not test [s1].
What to watch
Fenfluramine carries a known cardiovascular caution rooted in its diet-drug history — the reason it was pulled as an appetite suppressant was valvular heart disease and pulmonary hypertension — which is why its epilepsy prescribing includes echocardiographic monitoring; a registry reporting adverse events broadly, over a single year, does not resolve long-term cardiac safety [s2]. The open questions are whether the benefit in non-Dravet, non-Lennox-Gastaut epilepsies holds up in controlled trials, and whether a year of real-world response persists over the many years these conditions are managed [s1].
This article describes research and regulatory records and is not medical advice. Epilepsy treatment decisions are for patients, families and their clinicians.
Sources
- Effectiveness and tolerability of fenfluramine in pediatric and adult patients with developmental and epileptic encephalopathies: A multicenter, retrospective, real-world clinical-practice study — Epilepsia , June 23, 2026
- Drugs@FDA: Fintepla (fenfluramine), NDA 212102 — sponsor UCB Inc. — U.S. Food and Drug Administration (openFDA drug/drugsfda API)
More on
New potassium-channel drug cut focal seizures in a phase 3 epilepsy trial
In 374 adults with drug-resistant focal epilepsy, the 25 mg dose of azetukalner more than halved monthly seizures versus placebo over 12 weeks — one trial of a drug that is not yet approved.
Breastfeeding on epilepsy drugs didn't harm child development at age 6
In the MONEAD cohort, six-year-olds whose mothers breastfed while taking antiseizure medicines scored no worse on language or adaptive behaviour — and longer breastfeeding tracked with slightly better scores.
An experimental gene therapy cut seizures 83% in children with a severe epilepsy syndrome
ETX101 doesn't replace the faulty gene behind Dravet syndrome — it turns up the healthy copy that's still there. New data from 21 children show the effect holding for a year.
First global tally of early-life epilepsy points to developmental syndromes
A meta-analysis of 86 population-based studies estimated that about 236 infants per 100,000 live births develop epilepsy in the first two years, with developmental encephalopathies the largest share.