Exome no more. Over the last 15 months, we have repeatedly discussed how exome sequencing or genome sequencing is applied to neurodevelopmental disorders in order to discover new candidate genes and to assess the role of known candidate genes. We have also wondered sometimes whether exome sequencing is the most straightforward approach. Now – outpacing the two large international consortia using exome sequencing in epileptic encephalopathies – a recent study in Nature Genetics uses a different approach to uncover the genetic basis in 10% of patients with epileptic encephalopathies. Targeted resequencing or gene panel analysis is a hybrid technology between candidate gene sequencing and next generation sequencing and focuses only on a subset of candidate genes. While their study provides a comprehensive overview over the genetics of rare epilepsy syndromes, it raises the question whether the era of large-scale exome sequencing is coming to a natural end.
A balancing act. There is a fundamental question that arises when we aim to identify genes in seizure disorders: “Larger cohort or deeper sequencing”? Do we want to assess a few markers or genes in a large cohort or rather investigate a smaller, well-defined cohort with a more comprehensive technology including exome sequencing or genome sequencing. In a large cohort, you have a better chance of finding genes that are relatively rare, but you might miss the causative gene if it is not on your list of candidates. With more comprehensive technologies, you have a good chance of finding any gene, but you are limited to a smaller cohort. The genomic noise arising from this smaller cohort might mask your real finding. There is no rubber-stamp solution for this problem in rare epilepsies. Some of these epilepsies are surprisingly genetically homogeneous, such as Malignant Migrating Partial Seizures of Infancy with mutations in KCNT1. Other severe epilepsies such as Infantile Spasms, however, are genetically heterogeneous and large cohorts are needed to get an overview of the underlying genetic architecture.
The power of panels. While single candidate gene sequencing and exome sequencing represent the extremes of these approaches, other methods can combine features of both technologies. Gene panel analysis uses the technology of massive parallel sequencing, but is applied to a smaller subset of genes, thereby allowing for a rapid screening of a large cohort. Gene panel analysis has previously been shown to represent a promising technology in seizure disorders. Now, the study by Carvill and colleagues applies gene panel analysis using established and promising candidate genes to survey a cohort of 500 patients with various epileptic encephalopathies.
Two main findings. The cohort of 500 patients analyzed by Carvill and colleagues represents a mixture of various non-lesional epileptic encephalopathies including Infantile Spasms (n=81), Myoclonic Astatic Epilepsy (n=81) and Lennox-Gastaut Syndrome (n=40) as the most prominent, classifiable subsyndromes. A significant subset of patients had epileptic encephalopathy not otherwise specified (n=173). The 65 candidate genes comprised a mixed bag of known candidates, candidate genes derived from copy number variants and candidate genes derived from earlier exome sequencing studies. Applying gene panel analysis of these candidate genes, the authors identified pathogenic variants in 10% of their patients, most of which could be shown to be de novo. SCN1A, CDKL5, STXBP1 and SCN2A were the most prominent known genes and accounted for ~1% of cases each. Of the novel candidate genes, CHD2 and SYNGAP1 were the most prominent findings, also accounting for roughly 1% each. Both genes had previously been reported in other phenotypes, but were not considered epilepsy genes so far. In summary, the study by Carvill and colleagues suggests that genes for the broad phenotype of epileptic encephalopathy are by and large “1% genes”.
Candidates from copy number variations (CNVs). 33/65 candidate genes included in the study by Carvill and colleagues are candidate genes derived from microdeletions found in patients with epileptic encephalopathies. It is fascinating to see that this strategy worked out. If not encountered repeatedly, CNVs are notoriously hard to interpret and many of these variants might actually represent rare benign variants. Therefore, using candidate genes from this approach carries the inherent risk of chasing false positive findings. 2/33 CNV candidate genes (CHD2 and HNRNPU) had pathogenic variants in the gene panel study, providing us with a rough estimate of what can be expected from such a strategy: approximately 10% of CNV candidate genes may be relevant findings.
Expanding the SCN1A spectrum. Amongst the known candidate genes, the expanding spectrum of SCN1A phenotypes is one of the unexpected findings. SCN1A is the causative gene for Dravet Syndrome and Genetic Epilepsy with Febrile Seizures Plus (GEFS+). The study by Carvill and colleagues now finds three SCN1A mutations in patients with epilepsy aphasia syndromes. Epilepsy aphasia syndromes including Landau-Kleffner-Syndrome and the syndromes of continuous spikes and waves during slow sleep are conditions, in which epileptic activity in the rolandic region results in acquired speech arrest (aphasia). Patients may also have seizures, but they are usually less prominent. These syndromes, usually considered part of the spectrum of rolandic epilepsies, have little in common with Dravet Syndrome and there is little to no overlap between GEFS+ and the rolandic epilepsies. This finding underlines that the delineation of clinical spectrums is an ongoing process, a continuous back-and-forth between clinical and genetic findings. It will be one of the questions of the future if the GEFS+-related epilepsy aphasia syndromes differ from patients without an SCN1A mutation.
Don’t quit just yet. Should we throw away our exomes and do gene panels in the future? I would argue that we should not give up yet. Exome or genome sequencing still remains the only possibility to identify novel candidate genes in an hypothesis-free manner. I agree that many findings in autism, schizophrenia, intellectual disability and epilepsy have resulted in the identification of known epilepsy genes. Amongst those, SCN1A and CDKL5 appear to be the most prominent findings. However, the overall number of patients analyzed with exome or genome sequencing is still relatively small compared to cohorts in other neurodevelopmental disorders. Therefore, without going into too much detail regarding the underlying math, this is my plea: Don’t stop before the 1000 trio exome study.