EAN 2026 | Mortality and SUDEP in DEEs: a Bayesian meta-analysis of RCTs and open-label extensions

So developmental and epileptic encephalopathies are among the most severe forms of epilepsy we encounter in our epilepsy daily practice and in this patient when discussing about their long-term prognosis and management we know that premature mortality and especially sudden unexpected death in epilepsy are among the most important, the most important thing to take into account. So, despite the importance of this topic, in the literature available today, we have still a lack of high-quality data regarding incident rate of mortality and SUDEP, which is mainly relying on retrospective cohorts, which are based on a few hundreds of patients, actually. So what we decided to do is to conduct a meta-analysis, including several randomized trials that enrolled patients for a new trial of antiseizure medication. So what we did was to include 37 randomized controlled trials and 15 corresponding open-label extensions. We were able then to collect information on mortality and sudden unexpected death in epilepsy, which I will refer to as SUDEP, which is brief. And especially when we then analyze also specific subgroups such as Lennox-Gastaut syndrome, Dravet syndrome, and infantile epileptic spasm syndrome.

We were able to collect data from 3,700 patients. So as I was saying, this is a considerably larger population compared to the few hundreds of patients that are included in the main epidemiological studies we have available on SUDEP in these patients. And among these patients, we had around 2,000 patients with Lennox-Gastaut syndrome, around 1,000 patients with Dravet syndrome, around 400 patients with infantile spasm, and then other subgroups, which are around 200 patients that had other genetic DEEs. So, from a methodological point of view, we chose a Bayesian approach because, because actually there were some limitations, mainly SUDEP, and mortality in general, is a very rare outcome, and the frequency, the normal frequentist method that you could use in a meta-analysis would not perform well in these settings, and the Bayesian approach allowed us to use informative priors available from the literature and to then obtain incidence rates even with rare events because actually most of the trials we included had zero events reported. So what did we find? What did we find was 37 deaths among the analyzed patients and among these 37 deaths around 15, actually 15 were SUDEP, so around 50% which is a data that is quite in line with what we know from the literature. We estimated an incidence rate of mortality of 8.8 per 1,000 person-year in the overall DEE population with SUDEP incidence rate of 3.4 per 1,000 person-year. So it’s important to highlight that around 50% of the deaths in this patient were actually sudden unexpected deaths in epilepsy, defined as possible, definite or probable SUDEP as for ILAE standardized definition.

Regarding subgroups, we focused, as I was saying, on Lennox-Gastaut syndrome, Dravet syndrome and infantile spasm. In Dravet syndrome, all deaths that we encountered were actually SUDEP, so 100% of them, which is in line with what we know on the literature, so that in Dravet syndrome, the burden of SUDEP is very high, the risk of SUDEP is very high, and it is probably in line with the high burden of generalized tonic-clonic seizures, refractory generalized tonic-clonic seizures, and also the comorbidity related to the mutation of SCN1A gene. So we estimated for these patients an incidence rate of 7.6 per 1,000 person-year, which is the highest among the different subgroups. In Lennox-Gastaut syndrome, we provide the first, from what we know from our knowledge, the first systematically and prospectively collected incidence rate for SUDEP. And what we found is that a considerably high percentage of the deaths that we recorded around one third were actually SUDEP. So we estimated an incidence rate of SUDEP for LGS of 3.4 per 1,000 person-year. And in the final subgroup, the infantile spasm group, we actually encountered the highest mortality rate, although we have to say that due to the limited number of events, these results should be interpreted carefully, and with a mortality rate of around 9 per 1,000 person-year, and SUDEP contributing in a less important way compared to the other two subgroups with around 2.7 per 1,000 person-year.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.