Recent advances in epilepsy genomics and genetic testing

Developmental and epileptic encephalopathies (DEEs) are a group of severe, early onset epilepsies characterized by refractory seizures, developmental delay or regression associated with ongoing epileptic activity, and generally poor prognosis. DEE is genetically and phenotypically heterogeneous, and there is a plethora of genetic testing options to investigate the rapidly growing list of epilepsy genes. However, more than 50% of patients with DEE remain without a genetic diagnosis despite state-of-the-art genetic testing. In this review, we discuss the major advances in epilepsy genomics that have surfaced in recent years. The goal of this review is to reach a larger audience and build a better understanding of pathogenesis and genetic testing options in DEE.


Background
The developmental and epileptic encephalopathies (DEEs) are a heterogeneous group of severe, early onset conditions characterized by developmental delay or regression associated with refractory seizures and generally poor prognosis 1 . The incidence of epilepsy is nearly 70 per 100,000 children younger than 2 years and genetic epilepsies account for more than 0.4% of the general population constituting 30% of all epilepsies 2 . The prevalence of epilepsy in the United States is 5-8 million subjects annually, while the incidence is 35-71/100,000 per year 3 , though epidemiological data specific for DEEs are just emerging. A study on a broader group of severe epilepsies beginning before 18 months found an incidence of one in 2,000 births 4-6 . Some of the most well-studied DEEs include infantile spasms and Dravet, Lennox-Gastaut, and West syndromes.
Over the last decade, next-generation sequencing (NGS) has advanced the field of human genetics and genomics significantly 7 , leading to an explosion of gene discovery across many human disorders. The number of disease-associated genes has grown to 4,132, and over 50 genes have been newly associated with epilepsy in the last three years alone 8 . However, the new technologies have also brought new challenges 9 . The ability to perform sequencing across large cohorts of affected individuals with variable but related phenotypes highlights "phenotype expansions" associated with some disease genes. For the epilepsies, patients can have clinical presentations that range from static to degenerative, clouding a clear distinction between isolated DEEs and secondary epilepsies associated with neurodevelopmental disorders (NDDs) 10 . A great benefit of using NGS is its ability to deliver clinical diagnosis in a short time, but the available "cafeteria choice" of cutting-edge genetic tests can leave medical professionals and patients' families confused.
In this review, we discuss the major advances in epilepsy genomics that have surfaced in recent years and summarize the pros and cons of genetic testing options in DEEs that could help clinicians and patients reach the end of their "diagnostic odyssey" faster and in a cost-effective way.

Genetic testing
DEE is genetically and phenotypically heterogeneous, and there is a plethora of genetic testing options ranging from gene panels, which may include a few or hundreds of genes, to exome sequencing (ES), which investigates all ~20,000 genes. These are NGS techniques, also known as massive parallel sequencing (MPS), which include a variety of approaches that facilitate simultaneous sequencing of a large number of DNA segments 11 . Whole ES and targeted gene panels have contributed incredibly towards novel gene discovery, particularly in the pediatric epilepsies 12 . Sequencing all three billion bases of the genome, genome sequencing (GS), is mostly done in research settings but will inevitably enter the clinical realm soon.
Copy-number variants (CNVs) contribute significantly to variation in the human genome. CNVs are estimated to cause 1.2% difference for every reference genome 13 . CNVs can be detected by several genomic methods including conventional karyotype (deletions/duplications >5 Mb) and chromosomal microarrays (CMA, ~100 kb-5 Mb). Other methods such as quantitative PCR and multiplex ligation-dependent probe amplification are targeted approaches to detect smaller variations (<1 kb).
The most common types of genetic causes of DEE are sequence changes, responsible for 30-40% of cases, and chromosomal deletions or duplications, responsible for 5-10% of cases 14,15 . Gene panels provide a higher sequencing depth and lower cost when compared to ES and GS but restrict the diagnosis to specific genes in the panel. Importantly, some large panels are based on ES, with restricted analysis of only the "panel" genes, so the benefit of higher depth of coverage is lost, but this opens up the possibility of future reanalysis to include the whole exome. ES also provides good sequencing depth at a lower cost; however, it is restricted to protein coding regions only. CNVs can be predicted by this method but require a secondary method to plot the breakpoints. Selection of the most appropriate test may depend on a variety of factors including age at seizure onset, severity of disease, other associated features, and patient insurance.

Novel genes in DEE
Several novel genes and disorders associated with DEE have been identified in the last few years [16][17][18] (Table 1). Many of the genes causing epilepsy encode components of neuronal ion channels leading to neuronal hyperexcitability or depletion of inhibitory mechanisms 19,20 . However, recently, several new genes coding for proteins other than ion channels have been identified, such as chromatin remodelers, intracellular signaling molecules, metabolic enzymes, transcription factors, and mitochondrial complex genes 5,21,22 . The search term "epilepsy" OR "seizure" OR "epileptic syndrome" OR "epileptic encephalopathy" from 2016 to 2019 led to 66 entries in Online Mendelian Inheritance in Man. Although comprehensive discussion of all the discoveries is beyond the scope of this review, selected major advances are highlighted below.
ES trios have revealed the influence of de novo mutations as a genetic cause of severe epilepsies (Table 1). A recent study compared de novo variants identified in individuals with variable NDDs with and without epilepsy 23 . In the subset of 1,942 subjects with NDDs with epilepsy, 33 genes were observed to have significant excess of de novo variants, three of which had limited or no previous evidence of disease association: CACNA1E, SNAP25, and GABRB2. Nine de novo missense and two truncating variants in CACNA1E variants were identified in this cohort 23 . In a subsequent study, de novo variants in CACNA1E were identified in 30 individuals with DEE 16 . Detailed phenotyping revealed refractory infantile-onset seizures, severe hypotonia, and profound developmental delay, often with congenital contractures, hyperkinetic movement disorders, macrocephaly, and early death 16 . Functional analysis revealed consistent gain-of-function effects in R-type calcium channels. Some patients were seizure free on treatment with the anti-epileptic drug topiramate, which blocks R-type calcium channels. The condition is now catalogued as early infantile epileptic encephalopathy type 69 (#MIM 618285).
The RORB gene, which encodes the retinoid-related nuclear receptor ROR-beta, was recently associated with photosensi- tive generalized epilepsy in a large family segregating a nonsense variant in the gene 24 . In the same study, two individuals with de novo coding variants in RORB and a third individual with a de novo intragenic deletion presented with significant developmental delays and behavioral abnormalities in addition to their generalized epilepsy, consistent with a diagnosis of DEE. Together, these results suggest that RORB haploinsufficiency causes a fairly consistent epilepsy phenotype but variable developmental outcomes.
Several additional recent discoveries highlight the overlap between DEEs and NDDs, with several new genes associated with syndromic epilepsy, including NBEA, FBXO11, and SMARCC2 25-27 . NBEA has long been a candidate gene for NDDs and autism 28 . Clear disease association and description of the phenotypic spectrum were recently reported after the identification of 24 de novo variants in patients with NDD, many of whom also had generalized epilepsy. Similarly, one-quarter to one-third of individuals with pathogenic variants in FBXO11 or SMARCC2, each associated with variable NDD, also have epilepsy.
Recessive genes are a rare but important cause of DEE. Inborn errors of metabolism and malformations of cortical development constitute most of the autosomal recessive epilepsies 29 . Glycosylphosphatidylinositol (GPI) anchored proteins play key roles in the human body, mainly in development and neurogenesis. Several genes involved in GPI biosynthesis and the remodeling pathway are causative of autosomal recessive epilepsy. One such gene that was recently identified is PIGB 30 . This group reported 16 patients from 10 unrelated families with early infantile epileptic encephalopathy, type 80 (#MIM 618580). Some other recessive epileptic encephalopathies are due to WWOX, TBC1D24, UBA5, and SLC13A5 31-34 . TBC1D24 is known to cause a continuum of features that were originally described as distinct, recognized Mendelian phenotypes ranging from autosomal dominant deafness to autosomal recessive epileptic encephalopathy 35 . Similarly, in addition to causing epileptic encephalopathy type 28 (#MIM 616211), WWOX is implicated as the molecular basis of spinocerebellar ataxia, type 12 (#MIM 614322) 36,37 . Both these genes are examples of a spectrum of disorders with increasingly blurred lines differentiating them as more individuals and pathogenic variants are identified. Recently, homozygous pathogenic variants in CSF1R, encoding a tyrosine kinase growth factor receptor for colony-stimulating factor-1, were identified in patients with brain abnormalities, neurodegeneration, and dysosteosclerosis 38 . This gene was previously implicated in a dominant adult-onset leukoencephalopathy. Proliferation and growth of macrophages, including microglia, require colony-stimulating factor-1 receptor (CSF1R). This study represents an under-recognized group of genes that are associated with well-described, dominant phenotypes but can also produce a different clinical picture when present in biallelic, recessive states. This is important for filtering and interpreting variants from NGS data, as candidate variants cannot be eliminated based on poor phenotypic fit 39 .

CNVs in DEE
Studies using CMA have shown that pathogenic CNVs account for 5-10% of childhood epilepsies including DEE 40-42 , and CMA is the recommended first-line genetic test if the clinical picture includes dysmorphism, intellectual disability, congenital anomalies, and other neuropsychiatric features 43 . However, NGS is increasingly being employed in the detection of CNVs. One good example is the detection of deletions in TANGO2. TANGO proteins play a crucial role in redistributing Golgi membranes into the endoplasmic reticulum 44 . Bi-allelic TANGO2 pathogenic variants have been identified as a cause of a pediatric condition with multi-organ involvement 45 . Recently, a study identified intragenic, multi-exon deletions in TANGO2 by reanalysis of ES data 45,46 . The most common disease-causing allele (55%) in one series was deletion of exons 3-9 of TANGO2 17 . ES is not yet a match for CMA for CNV detection, as it can provide data about only the protein coding or exonic regions, but it is an increasingly powerful diagnostic tool, and a growing number of algorithms are being developed to aid the detection of CNVs by NGS.
With the introduction of ES and GS, it is now possible to detect both single nucleotide variations and CNVs using an exome-or genome-wide approach with a single test 47 .

Future of epilepsy genomics
Despite state-of-the-art genetic testing, more than 50% of patients with DEE remain without a genetic diagnosis. Whole GS is increasingly being used to uncover the role of non-coding genetic material in the human genome 48,49 . Undoubtedly, massively parallel sequencing has greatly accelerated disease gene (and variant) discovery, but most studies and nearly all clinical testing employ gene panels or ES, limiting the genomic search space and the types of variants that can potentially be identified. For disorders like fragile X syndrome that are due to the expansion of triplet repeats, testing strategies other than gene panels or exome are required. Several studies have proposed a genetic testing strategy to achieve the highest clinical utility, cost-effectiveness, and diagnostic yield for individuals with epilepsy 50-52 , but specific testing algorithms are likely to change over time as new tests are introduced and the costs of existing tests decrease. New assays may be required to detect lesser-known but important molecular mechanisms.
Post-zygotic, somatic mosaic mutations are increasingly identified as an important cause of genetic disorders 22,53 . In epilepsies, many of the mutations in the mTOR pathway that lead to brain malformations are somatic mosaic mutations. Typically, leukocyte-derived DNA is used in individuals with DEE to search for germline variants, which are inherited or arise de novo in the zygote. Recent studies have demonstrated that post-zygotic somatic variants also underlie DEE 22,54-58 but can be easily missed by standard NGS tests.
Another field that has potential to uncover some of the underlying molecular mechanisms is epigenetics. Epimutations represent a class of mutational event where the epigenetic status of a genomic locus deviates significantly from the normal state 59 . Methylation of DNA and histone modifications are increasingly being implicated as causative or contributing factors for several conditions 60,61 . DNA methylation at CpG dinucleotides is the most widely studied epigenetic modification. Methylation represents an epigenetic change-a chemical modification of DNA that does not change the underlying DNA sequence. A recent study investigated the role of de novo methylation changes in NDDs using methylation chips 62 . In a cohort of 489 affected individuals, of which 16% had epilepsy, the authors identified rare differential methylation in 23% of cases when compared to controls. When the parents were able to be tested, ~40% of the methylation variants were de novo, suggesting that de novo methylation abnormalities may be causative in 5-10% of their cohort. When identified, the underlying causes of the methylation changes were varied and included CNVs, sequence variants in regulatory elements, or repeat expansions, each of which is easily missed by conventional (even next-generation) sequencing methods. In a second study of undiagnosed NDDs using a similar approach 63 , candidate differentially methylated regions in two individuals with epilepsy and intellectual disability of unknown etiology were identified.
Several techniques that enable longer read lengths (up to 200 kb), such as nanopore-based "fourth-generation" sequencing 64 and single molecule, real time (SMRT) sequencing 65 , have recently emerged. The advantages of long reads include shorter sequencing time, ability to sequence AT-or GC-rich regions and repeat stretches, and the detection of large structural abnormalities including insertions, deletions, inversions, translocations, and tandem/interspersed regions 66,67 .

Conclusion
NGS-based technologies are a mainstay of clinical diagnostic testing, and the applications and testing options will only increase as the technology, bioinformatics, and resources evolve. NGS successfully detects single nucleotide variations, structural rearrangements, and CNVs. Clinical phenotypes are now being defined by the underlying molecular basis. Interpretation of NGS data is an iterative process involving forward genetics along with a reverse phenotyping approach. The dynamic nature of data analysis should be explained to patients and their families. As more and more novel genetic and epigenetic etiologies are unveiled in DEE, the challenge for clinical and research laboratories is to make sure the testing is clinically relevant, is cost effective, and can be integrated into clinical care.

Open Peer Review Current Peer Review Status:
Editorial Note on the Review Process are written by members of the prestigious . They are commissioned and F1000 Faculty Reviews F1000 Faculty are peer reviewed before publication to ensure that the final, published version is comprehensive and accessible. The reviewers who approved the final version are listed with their names and affiliations.
The reviewers who approved this article are:

Version 1
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com

Weiping Liao
Institute of Neuroscience and the Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, China No competing interests were disclosed. Competing Interests: