UNC13B and focal epilepsy Free

We read with interest the recent article by Wang et al.¹ regarding the potential role of variants in the gene UNC13B encoding the unc-13 homolog B protein in focal epilepsies, including one patient with benign occipital epilepsy and two with benign childhood epilepsy with centrotemporal spikes (BECTS). We are concerned that the published work¹ contains methodological issues relating to restricted selection of controls from public databases and interpretation of splicing variants.

Essentially all variants of major effect described in epilepsies are ultra-rare (<0.0005 minor allele frequency), as defined by the Epi4K consortium, and typically absent from large databases (for gnomAD v2.1.1 currently >140 000 subjects).^2–4 Inspection of the nine allele variants that the authors’ state in their abstract are ‘potentially associated with epilepsy’ reveals seven are not novel and are in fact present in the gnomAD (https://gnomad.broadinstitute.org/)³ and dbSNP (https://ncbi.nlm.nih.gov/snp/)⁵ databases (Table 1). We were surprised that the authors detected only one of these variants (c.2644G > T; p.Gly882Trp) in the sample of 10 640 Han Chinese individuals given the frequency of the respective variants within East Asian populations as reported in gnomAD.^1,3 Furthermore, the only nonsense variant p.Trp45X, and the p.Gly882Trp missense variant, are both present in gnomAD (see Table 1) despite the authors reporting in their results that they are absent.¹ These discrepancies arose because the authors restricted their UNC13B variant analysis to those identified in their patients with focal epilepsy to ‘Control-only’ samples in gnomAD. Since these ‘Control-only’ samples represent only 60 146 of 141 456 individuals on gnomAD (v2.1.1), we suggest that the additional gnomAD subsets of ‘Non-TOPMed’, ‘Non-cancer’ and ‘Non-neuro’, should also have been included in their analyses as these are also appropriate controls for a neurological disorder, as is done in conventional analyses. In general, whilst we acknowledge the gene burden analysis undertaken by the authors, we suggest greater rigor in design. The use of publicly available control databases in gene-set burden analyses, as opposed to utilizing matched control sets from other studies, without controlling for variability in sequencing platforms, analysis pipelines, or population ancestry and substructure, can lead to erroneous findings and interpretations, as has been described for previous studies.⁶

There are also concerns with the interpretation of some variants as pathogenic by the authors without supporting functional analysis.¹ The c.4008+1G > T variant does affect an invariant splice site position but is observed four times in East Asian subjects on gnomAD (Table 1), and is inherited from an unaffected mother as noted by the authors.¹ The other splice site variant (c.4330+7G > A) does not affect a residue within the consensus donor splice site sequence, and occurs at a site not typically associated with disease-causing variation.⁷

While we cannot exclude that UNC13B variants may be population-specific risk factors for focal epilepsies, the presence of the reported variants in East Asian individuals at frequencies higher than would be expected for rare dominant alleles known to cause epilepsies^2,4 draws the authors conclusions into question. Further analyses of larger cohorts with careful genotype-phenotype correlation are required to confirm any potential disease associations of UNC13B with seizure disorders.

Data availability

Data sharing is not applicable to this article as no new data were created or analysed in this study.

Competing interests

The authors report no competing interests.

References