Genomic analyses of all known putative familial hypercholesterolemia mutations showcase the value and limitations of public familial hypercholesterolemia mutation databases. (25th November 2020)

• Record Type: Journal Article
• Authors: Dobretz, K; Karunajeewa, A.S; Judja-Sato, B; Von Kaenel, T; Mach, F; Ehret, G
• Is Part Of: European heart journal. Volume 41:(2020)Supplement 2
• Publication Date: 2020-11-25
• Subjects: Lipids; Cardiology -- Periodicals; Heart -- Diseases -- Periodicals; 616.12005
• DOI: 10.1093/ehjci/ehaa946.2994
• Languages: English
• ISSNs: 0195-668X

Abstract

Abstract: Background/Introduction: Familial hypercholesterolemia (FH) is genetically very heterogeneous and genomic and locus-specific public databases describing putative FH mutations are assumed to be of limited clinical utility because of classification errors. Purpose: A description of all currently known publicly available putative FH mutations in order to determine the reliability of the classification of FH mutations described. Methods: The LOVD and ClinVar databases were interrogated for the phenotype and genes of interest. Additional information on each variant was obtained using the Bioconductor toolset, gnomAD, and GETex. Results: Currently know putative FH variants included 4, 529 variants (97.2%) in the classical FH-genes LDLR (61%), PCSK9 (10%), and APOB (29%). Single nucleotide variants constituted 83% and 17% were copy number variants. Exonic variants contributed 78%, 14% of the variants were intronic, 7% large CNV, and 1% in upstream or downstream regions. Of the 4, 529 variants, 45% were classified as pathogenic or likely-pathogenic (Fig. 1a). The ratio of exonic/intronic variants was 10.1 for pathogenic variants, 6.9 for likely pathogenic variants, 2.8 for likely benign and 1.4 for benign variants (p-value for class difference <2x10–6). For 502 frameshift mutations in exons that are particularly damaging, only 93.2% were classified as pathogenic or likely-pathogenic and 1.7% as benign or likely benign (Fig. 1b). Across 222 exon-covering deletions of >100 nucleotides, also particularly damaging, only 90.5% were classified as pathogenic or likely-pathogenic. Of the 4, 529 variants, 1, 561 (34.5%) were polymorphic in gnomAD. For these variants, the gnomAD sample size was on average 227420 (26102–282902). Of all 1, 561 polymorphic variants in gnomAD 182 (11.6%) were classified as pathogenic or likely-pathogenic (Fig. 1c) and the variant frequency ranged from 10–4 to 10–6 (average 1.8x10–5), 43 with a frequency >1.8x10–5. Among variants found in gnomAD and classified as non-pathogenic, we observed ∼2000x higher frequencies (average 0.04). Of the 4, 529 variants, 100 matched eQTLs or sQTLs in GTEx, none was annotated as pathogenic (Fig. 1d). Conclusion: We review all currently known putative FH mutations with pathogenic or likely-pathogenic variants being mostly exonic. Highly damaging mutations are largely classified as pathogenic or likely-pathogenic, but up to 9% are not classified as pathogenic in the two public FH mutation databases. Assuming an FH population prevalence of 1/250 and 2000 pathogenic variants with the most frequent variant 10x more frequent than the average, we expect most pathogenic FH mutations at frequencies <2x10–5. We found 46 pathogenic or likely pathogenic variants to have a frequencies of >2x10–5 in the general population, evidence for misclassification. No pathogenic FH variant was found among GETEx eQTLs or sQTLs. Our data showcase the utility and weaknesses of the current public FH mutation databases. Funding Acknowledgement: Type of funding source: Public hospital(s). Main funding source(s): Hôpitaux Universitaire de Genève - Fonds privés