Evaluating the consistency of SMARCB1 variant classification and assertions of genotype-phenotype relationships in ClinVar.

Pathogenic SMARCB1 variants are associated with multiple Mendelian syndromes: Schwannomatosis (SM), Rhabdoid Tumour Predisposition Syndrome (RTPS1), and/or Coffin-Siris Syndrome (CSS). Although some data suggests genotype/phenotype relationships based on mutation type and location, this is not used consistently potentially due to inconsistent or absent phenotypic data. We used ClinVar, the largest public platform of variant classifications, to evaluate variant-specific data from clinical laboratories and cited publications to assess reporting consistency and genotype-phenotype relationships. When available, we extracted the mutation type, HGVS nomenclature, submitters, publications, and disease conditions. We compared the disease assertions made by laboratory submissions with those made in cited publications. Across 59 SMARCB1 pathogenic variants, we identified 91 laboratory submissions (38 single and 20 multiple laboratories) and 40 cited articles. Of 91 submissions, 40 did not provide any disease assertion. Nonsense RTPS1 variants (n = 23) were mostly located in exon 2, with some in exons 1, 4 and 5, and other variant types were limited to frameshifts and large deletions/duplications. SM variants (n = 10) were found throughout SMARCB1 with diverse variant types. CSS variants (n = 6) were single amino-acid deletions, missense and frameshift variants limited to exons 8 and 9. Although ClinVar is a rich source of variant data and extensively used during variant classification, the frequent absence of disease assertions and inconsistent reporting impedes implementation of clear genotype-phenotype relationships for SMARCB1. Given the significant clinical impact of these diagnoses a more standardized way of reporting pathogenic variants for diseases associated with multiple Mendelian disorders is needed.