Genomics and personalised medicine : diagnostics, deep phenotyping and pharmacogenomics in cohorts of rare disease patients
Abstract (summary)
Introduction: Clinical genetics is a rapidly evolving specialty. Diagnostic testing is moving to more agnostic screening of whole exomes and genomes. This leads to challenges in variant interpretation, variants of unknown clinical significance and secondary findings. This study looked at the utilisation of phenotypic and whole genome sequence (WGS) data to increase understanding of rare disease and pharmacogenomics. Methods: WGS was used to attempt to clarify the molecular basis of two cases of Bardet Biedl syndrome (BBS) by filtering variants in ingenuity, using bioinformatics methods to identify copy number losses or gains and examining known ciliopathy and other genes. The feasibility and accuracy of extracting pharmacogenomics data from WGS was assessed and results validated in a cohort of 84 people. Diplotypes or genotypes were determined for 18 actionable pharmacogenes and prescribing guidance was prepared. Diplotypes were also determined for 43 additional pharmacogenes. Results were validated using a commercial pharmacogenomics assay. Results: Candidate variants were identified in a number of BBS, ciliopathy and non-ciliopathy genes in each case. However no definitively pathogenic biallelic variants were identified. All patients had at least one actionable pharmacogenomic variant that could result in a change in drug dose or monitoring. The mean number of variants per patient was 3.8. Haplotype frequency data for the actionable pharmacogene haplotypes was not significantly different to population data. The majority of disagreements between WGS and SNP data were caused by poor clustering of samples during SNP genotyping resulting in ambiguous calls. Overall the discordance between WGS and SNP data for all tested pharmacogenes was less than 0.01%. Conclusions: Singleton WGS was not sufficient to identify the cause of BBS and further work is required to identify this. Possible reasons include missed variants, variants in novel genes, deep intronic variants or non-Mendelian modes of inheritance. Pharmacogenomic variants can be identified using WGS with a similar success rate to a current commercial method, but has additional advantages including the ability to review data as pharmacogenomic prescribing guidelines change. There are many challenges to introducing population-level pharmacogenomic prescribing and many potential benefits.