Quantum leap. At the Annual Meeting of the American Epilepsy Society, the Epi4K consortium presented the first data on exome sequencing in epileptic encephalopathies. This data is the most exciting finding in the field of epilepsy genetics in 2012 so far, as it provides a deep insight into the genetic architecture of Infantile Spasms (IS) and Lennox-Gastaut Syndrome (LGS). With the findings presented by the Epi4K collaborators, the epileptic encephalopathies are joining a group of neurodevelopmental disorders with a significant burden of de novo mutations. However, there are important differences that set both IS and LGS apart from diseases like autism, intellectual disability and schizophrenia.
The big picture. We have written several posts on trio exome sequencing in various neurodevelopmental disorders this year including autism, schizophrenia and intellectual disability. All studies find a significant burden of de novo mutations in trios, i.e. mutations that arise in the proband but that are not present in either parent. Many of these studies have difficulties interpreting their findings as mutations also occur at approximately the same rate in unaffected individuals. Some studies find a slight increase in patients with neurodevelopmental disorders, but these differences are not as clear-cut as one might expect. This basically means that an identified de novo mutation needs additional information to be considered pathogenic or not. This additional information is either derived from prior knowledge or through pure statistics. Prior knowledge means that this gene is known to cause the particular disorder. Statistics comes into play when a given gene occurs more than once in given a disease, a finding that is unlikely to occur by chance. In all trio exome sequencing studies on autism, for example, interpretation of findings is largely limited to a handful of recurrent genes, as most genes are so-called “single hits” that are difficult to interpret. This scenario is completely different in IS and LGS.
The genetic architecture. Let’s briefly review what we know regarding the genetic architecture of IS and LGS. Both disorders are epileptic encephalopathies and can be caused by a variety of different causes ranging from perinatal brain damage, brain malformations or neurometabolic disorders. Also, in patients with Infantile Spasms, Tuberous Sclerosis appears to be one of the more common causes. However, none of the potential causative factors is present in a significant subset of patients. These patients have IS or LGS for no apparent cause. Mutations in ARX, CDKL5, STXBP1, SPTAN1 and many other genes have been described, but their frequency in patients with epileptic encephalopathies is not known. There appears to be an important role for copy number variations, which are strong risk factors in 5-10% of cases. This leaves us with > 90% of cases with presumably genetic epileptic encephalopathies to be explained.
The Epi4K study. Epi4K, funded by the National Institutes of Health, is a large consortium on epilepsy genetics and the presentation at the meeting of the American Epilepsy Society was the first presentation of data generated by the Epi4K researchers. They analysed 165 patients with IS and LGS collected by the Epilepsy Phenome/Genome Project (EPGP) for de novo mutations that were not present in the parents using exome sequencing. Using this paradigm, the Epi4K researchers identified possibly pathogenic de novo mutations in at least 15% of patients. The list of genes includes many known genes for epileptic encephalopathies, several “single hits” that require further studies and a few genes, which appear twice in this patient cohort. In order to put the data of the mutations identified in their cohort in context, the researchers went on to assess whether these genes are “mutation-intolerant”. This method basically queries large databases such as the Exome Variant Server regarding the known mutations in large cohorts. For example, a gene with many disruptive mutations would be considered relatively mutation-tolerant, whereas a gene without any disruptive mutation would be mutation-intolerant. The mutation spectrum and frequency of a group of genes can then be compared to the data in controls. Applying this method, the Epi4K researchers find that the genes mutated in IS/LGS are usually resistant to mutations, which is another hint at a pathogenic role of these genes.
Paradigm shifts. Even though de novo mutations in IS/LGS have been described in individual cases, the Epi4K data is the first report on a large cohort of patients that has been systematically screened for de novo mutations on a genome-wide level. There are least three paradigm shifts initiated by this data.
First, de novo matters. De novo mutations contribute significantly to the genetic architecture of IS/LGS. While this statement might be trivial for many scientists, I still consider this a surprise. Only two years ago, I would not have thought that de novo mutations would have such a large impact. To me, Dravet Syndrome with the high frequency of de novo mutations in SCN1A was a singularity, an anomaly amongst a highly complex and probably impenetrable genetic architecture. I find myself surprised and relieved that some of the genetic puzzle that underlies human epilepsy appears to be solvable.
Secondly, the known knowns. The Epi4K researchers find a high frequency of known epilepsy genes. Looking back at the last decade of large-scale genomic studies, finding known genes using hypothesis-free methods is the exception, not the rule. It is reassuring that the genes that we have worked on in the past are in fact relevant. Again, this is not trivial as there is no guarantee that candidate genes detected in a small group of patients through candidate or family approaches are relevant on a population level.
Third, the negative findings. It is important to note what the Epi4K researchers did not find. There is no single gene for IS or LGS, but the genetic architecture is heterogeneous. In parallel with the findings in trio studies in autism, it can be estimated that 400-500 genes will contribute to IS/LGS. This will have profound implications for the way we will approach these disorders in clinical practice. Looking for single candidate genes will likely be futile; gene panels or even clinical exome sequencing will be more efficient. Also, the Epi4K researchers did not identify a significant burden of recessive mutations in their patient sample. Even though most patients with IS/LGS are screened for metabolic disorders, I am surprised that there is not a sizeable subgroup of patients with “hidden metabolic disorders”.
Looking forward. The Epi4K data has established de novo mutations as a common feature in IS and LGS. Taken together with the role of CNVs, we can expect that ~25% of the genetic burden of IS/LGS is explained. This leaves us with a large proportion of cases without positive genetic findings. While the rate of pathogenic de novo mutations might currently be underestimated given the unknown role of various single hits, it is unlikely that the genetic etiology is explained by this mechanism in the majority of patients. Even though our previous post on phantom heritability has suggested that we should never aim for the 100% mark in genetic studies, this data begs the question what else is out there as causative genetic variants. Either way, the Epi4K data, explaining 15% of the genetic morbidity at once, is a quantum leap in epilepsy genetics.