Trio-sequencing your clinic. From the perspective of a child neurology clinic, I often wonder how much information we would gain if we performed trio exome sequencing for de novo mutations systematically in all our patients with difficult-to-treat epilepsies. Many of these patients have epilepsies that are difficult to classify and they have not been included in our existing EuroEPINOMICS working groups on defined syndromes. Now, a recent publication in Epilepsia gives us an idea what we will find if we perform family-based exome sequencing in patients with unclassified epileptic encephalopathies. Basically, you will find SCN1A and CDKL5 plus mutations in several genes that are likely pathogenic. But there is much more to this issue, which motivated me to come up with a classification scheme for epilepsy-related de novo events. 

The gold standard. We have discussed various publications relating to trio exome sequencing in intellectual disability, autism, schizophrenia and epilepsy over the last 12 months. The identification of de novo mutations by comparing exome data of patients and parents has become an established technology that has potential both as a research tool and –in the future- as a diagnostic test. However, only a fraction of the mutations found by trio exome sequencing are pathogenic, as roughly 1-2 de novo mutations are also observed in control trios. While it is tempting to think that each and every patient with an unexplained epilepsy must have a pathogenic de novo mutation, the evidence for this is only emerging at a slow rate.

SCN1A, CDKL5. Veeramah and colleagues have performed exome sequencing in 10 parent-offspring trios with therapy-resistant epileptic encephalopathies. They find de novo mutations in 9/10 probands. In 7/10 probands, these de novo mutations include known or presumable epilepsy genes. In particular, two patients with the clinical suspicion of Dravet Syndrome had mutations in SCN1A and one female patient with infantile spasms progressing to multifocal seizures had a de novo mutation in CDKL5, a clinical and genetic finding consistent with CDKL5 encephalopathy. One patient had a mutation in EEF1A2, a gene previously identified in a patient with severe intellectual disability and epilepsy. Three additional patients had mutations in plausible candidate genes. A patient with microcephaly, corpus callosum hypoplasia and epilepsy with complex partial seizures had a de novo mutation in CLCN4, coding for a CL/H+ exchanger highly expressed in the CNS. This gene is from the same family as CLCN2, which has previously been described in idiopathic generalized epilepsies. A patient with generalized tonic-clonic seizures, particularly in the setting of illnesses had a mutation in KCNH5, a gene coding for a potassium channel expressed in the CNS. Interestingly, the patient’s development was normal until the age of 6 months and the phenotypic description including hemiclonic seizures and the generalized EEG discharges are reminiscent of Dravet Syndrome. Finally, a patient with mild developmental delay and refractory complex partial seizures had a mutation in ARHGEF15, a gene coding for Ephexin 5, a protein involved in synapse development. In summary, the findings range from mutations in known genes to recurrent genes to promising genes found for the first time. This motivated me to come up with a draft classification scheme for epilepsy-related de novo mutations.

A classification scheme for epilepsy-related de novo mutations. I am going out on a limb here, but this would be one possibility to classify de novo mutations in patients with epilepsy based on the experience we have had over the last 12 months. Please note that this tentative classification does not take into account prediction tools that tell us something about the severity of the mutations. It only considers the identity of the gene and qualifies the evidence of this gene to be involved in the epilepsy phenotype. This should involve all mutations that are not synonymous, i.e. every mutation that alters the amino acid sequence or splicing. Briefly, Class 1 mutations should be de novo mutations in known and established epilepsy genes including SCN1A, CDKL5 or ARX. There is sufficient evidence that de novo mutations in these genes are involved in the epilepsy phenotype. Also mutations in novel genes where recurrent de novo events have been identified in similar phenotypes should be Class 1 mutations. Class 2 mutations are in genes that were previously described in other neurodevelopmental disorders such as intellectual disability. As it is not immediately clear why mutations in the same gene may cause different neurodevelopmental phenotypes, these genes must be approached with a certain degree of caution in the same way as de novo copy number variations with a broad phenotypic range. Also, these variants might be benign mutation hotspots, but the available control cohorts are still too small. Class 3 mutations are de novo mutations in “promising” genes including many previously not described ion channel genes expressed in the CNS and genes implicated in neuronal development. Everything else is Class 4 and of uncertain significance. Non-pathogenic variants are de novo mutations in genes that were previously found in controls. While one could still argue that these mutations confer a certain risk to disease or might be disease-related if they involve a certain region of the gene, this evidence will be very hard to establish. Please let me know what you think of this classification. I have added an additional category for de novo mutations in established epilepsy genes that appear as individual findings in existing databases (1u for “uncertain”). The significance of these variants is not immediately clear, as discussed in an earlier post.

A proposed classification scheme for epilepsy-related de novo mutations. Not every de novo mutation found in a patient with epilepsy is necessarily pathogenic. I have tried to classify the mutations based on the evidence relating to the gene involved. This classification does not take into account the severity of the mutation. Also evidence in animal models is not considered direct evidence for a role of this gene in human epilepsy. I would be happy to discuss the idea behind this classification and I am looking forward to your input.

The RES working group on unclassified EEs. Reading the paper by Veeramah and colleagues remind me of the RES working group on unclassified epilepsies, where our initial goal was to delineate novel phenotypes either through clustering of clinical features or through genetic studies. I would like to use this opportunity to remind everybody that we might discuss putting together a cohort of patients with unclassified epileptic encephalopathies for trio exome sequencing in the near future. I would refer to the next RES unclassified working group teleconference for this and please let us know if you are not included yet and would like to participate. In summary, with genetic screening technologies available at a much more rapid pace, we might be looking towards an interesting era where phenotypes will emerge based on an interesting interplay of clinical and genetic studies.

Ingo Helbig

Child Neurology Fellow and epilepsy genetics researcher at the Children’s Hospital of Philadelphia (CHOP), USA and Department of Neuropediatrics, Kiel, Germany