Joining forces. The EuroEPINOMICS-RES consortium and Epi4K/EPGP are currently joining forces for genetic studies on epileptic encephalopathies. A first collaborative study focuses on de novo mutations in Infantile Spasms and Lennox-Gastaut-Syndrome. In the last two years, after decades of disappointment, we have finally managed to accomplish a breakthrough in understanding the genetic basis of epileptic encephalopathies. The method of trio-based exome sequencing works amazingly well to identify the genetic cause, and the field currently has the crucial momentum to reach the next level of research. Let’s briefly review why we need international collaborations to disentangle the genetic architecture of the epileptic encephalopathies.
De novo mutations in epileptic encephalopathies. The tide was already about to turn on the budding technology of trio-based exome sequencing. Even though initial studies in autism, intellectual disability and schizophrenia had shown promising results of this costly, but comprehensive technology, the genetic variability and lack of recurrent findings was disturbing. The frequency of de novo mutations in unaffected siblings was found to be only marginally lower than in affected sibs, hinting at substantial genomic noise drowning out any possible signal from truly causative genes. In addition, very few trios were found to have a de novo mutation in the same gene. During this phase of increasing reluctance, the first studies on epileptic encephalopathies were published – and their results turned skepticism into new hope.
Genes in waiting. The epileptic encephalopathies probably stand out as the single group of neurodevelopmental disorders in which a significant fraction of cases is explained by de novo mutations. This is both due to the fact that we stumble upon known genes in roughly 15% of cases and because the discovery rate of novel genes seems to be higher than in autism or intellectual disability. Accordingly, even small increases in sample sizes may yield novel genes for epileptic encephalopathies. While it is premature to delineate an upper limit of how much can be explained by de novo mutations, the existing data at least suggests that we have finally identified a technology that allows us to take a quantum leap forward. One novel gene for every 20-30 trios sequenced would be a rough estimate. And “novel” does not mean that we don’t know the gene yet. It is already before our eyes, but we fail to appreciate its importance in the flood of other, non-causative mutations.
When SCN1A gets a p-value. I remember a discussion during the RES kick-off meeting in 2011 when we wondered whether SCN1A in Dravet Syndrome has a p-value. Let me explain. SCN1A mutations can be identified in more than 80% of patients with Dravet Syndrome. There is no doubt regarding the causative role of the mutation. There is no need to express this in statistical terms, and we quickly dismissed this idea. De novo mutations, especially when identified in more than one patient, were considered to be causative. In 2013, we are not so sure about this “double hit” paradigm anymore. While there is no doubt regarding SCN1A and Dravet Syndrome, we know about the amazing variability of the human genome. If we apply the “double hit” paradigm, we will stumble upon genes like Titin (TTN). In fact, there are several patients in recent studies with de novo mutations in this gene. Titin is the largest gene of the human genome, and it is relatively mutation-prone. Accordingly, finding mutations in this gene is not a big surprise. Wait, did I just use the word “surprise”?
Level of surprise. Titin is not the only gene with de novo mutations that we usually dismiss. We are looking for genes with de novo mutations that surprise us. And, in statistical terms, our level of surprise is the p-value. But how can we measure this? Based on the size of a gene, the amount of sequence covered by exome sequencing and the overall number of de novo mutations in a group of patients, it is possible to estimate whether a certain gene has more de novo mutations than expected. This would be a strong indicator that this gene is involved in epileptic encephalopathies. This “real estate analysis” helps you argue why de novo mutations in SCN1A are likely causative compared to de novo mutations in Titin. There is, however, one problem.
The 1% disease. If we know one thing about the genetic architecture of the epileptic encephalopathies, it is the remarkable genetic heterogeneity. SCN1A in Dravet Syndrome and GRIN2A in epilepsy-aphasia are notable and unusual exceptions from this rule. For the majority of epileptic encephalopathies, however, recurrent mutations in a specific gene are rare, and few genes are identified in more than 1% of patients. Accordingly, there will only be very few patients with de novo mutations in a given gene, and this gene might not pass the statistical threshold. That’s where international collaborations come into play.
The E2 working group. Increasing sample size and finding more de novo mutations that allow a given gene to rise above the level of genomic noise is the only chance to demonstrate the causative role. EuroEPINOMICS-RES and EPGP/Epi4K are currently performing a joint analysis of their Infantile Spasms/Lennox-Gastaut cohorts, culminating in the largest study of its kind so far. Independent of the number of novel genes identified, this study on almost 400 trios will allow for the most comprehensive insight into the genetic architecture of epileptic encephalopathies to date.