Biased pathogenic assertions of loss of function variants challenge molecular diagnosis of admixed individuals

Abstract

Diagnosis of individuals affected by monogenic disorders was significantly improved by next-generation sequencing targeting clinically relevant genes. Whole exomes yield a large number of variants that require several filtering steps, prioritization, and pathogenicity classification. Among the criteria recommended by ACMG, those that rely on population databases critically affect analyses of individuals with underrepresented ancestries. Population-specific allelic frequencies need consideration when characterizing potential deleteriousness of variants. An orthogonal input for classification is annotation of variants previously classified as pathogenic as a criterion that provide supporting evidence widely sourced at ClinVar. We used a whole-genome dataset from a census-based cohort of 1,171 elderly individuals from São Paulo, Brazil, highly admixed, and unaffected by severe monogenic disorders, to investigate if pathogenic assertions in ClinVar are enriched with higher proportions of European ancestry, indicating bias. Potential loss of function (pLOF) variants were filtered from 4,250 genes associated with Mendelian disorders and annotated with ClinVar assertions. Over 1,800 single nucleotide pLOF variants were included, 381 had non-benign assertions. Among carriers (N = 463), average European ancestry was significantly

Variant interpretation using population databases: Lessons from gnomAD

“…For population-groups not well represented in gnomAD, we would expect these numbers to be even higher. Often there is limited evidence to rule out pathogenicity for the variants not observed in a population database, resulting in an increased number of variants of uncertain significance in clinical genetic testing and highlighting the need for continued aggregation of sequencing data to improve the accuracy of genetic test interpretation (Naslavsky et al, 2021). The gain from increased sample size and improved representation is demonstrated by the decrease in number of unique variants per individual when utilizing the entire gnomAD dataset versus v2 exomes only (Figure S1 and Table S2).…”

Bartter Syndrome-Related Variants Distribution: Brazilian Data and Its Comparison with Worldwide Cohorts

“…The indication of genetic test should be driven according to the genetic spectrum variants for each specific population, which could be timing and money saving [42], which was demonstrated in our study. Moreover, the lack of sufficient information on genetic basis in non-European non-North American populations has been widely discussed as a limitation to reproduce data from genome-wide association studies (GWAS) for a number of disorders [14], and the current study expanded the genetic spectrum of BS mutations including data from an admixed population with different ethnic background. BS type 3 was the most frequent type of BS in this Brazilian cohort, similar to what has been reported in other cohorts [29,31,34,41].…”

“…The clinical and genetic profiles of BS have been poorly characterized in developing countries, in non-Caucasian populations and, particularly, in highly admixed populations. Moreover, the underrepresentation of those populations represents a limitation in the use of the current genomic databases as reference for the analysis of new variants [14]. Brazilian population is unique owing to its diverse ancestry and high rate of miscegenation [15].…”

Clinical genomics and precision medicine

“…However, because populations with European ancestry are the most studied, we know much more about the mutations that cause diseases in individuals of European origin. For instance, loss-of-function mutations, if placed in fragments of our genomes that are of African and Native American origins, are three times less likely to be present in the ClinVar dataset than loss-of-function mutations placed in fragments of European origins ( Naslavsky et al, 2021 ). While it is not clear yet how the effects of ancestry should be incorporated into clinics (including the Polygenic Risk Scores, see below), there are at least three cases in which ancestry is relevant: (i) mutations causative of rare genetic variants may be associated with some ancestries.…”