Missing heritability: is the gap closing? An analysis of 32 complex traits in the Lifelines Cohort Study

• Published in: European journal of human genetics : EJHG Vol. 25; no. 7; pp. 877 - 885
• Main Authors: Nolte, Ilja M, van der Most, Peter J, Alizadeh, Behrooz Z, de Bakker, Paul IW, Boezen, H Marike, Bruinenberg, Marcel, Franke, Lude, van der Harst, Pim, Navis, Gerjan, Postma, Dirkje S, Rots, Marianne G, Stolk, Ronald P, Swertz, Morris A, Wolffenbuttel, Bruce HR, Wijmenga, Cisca, Snieder, Harold
• Format: Journal Article
• Language: English
• Published: England Nature Publishing Group 01.06.2017
• Subjects: Bioinformatics; Blood Cell Count; Blood Pressure - genetics; Clinical medicine; Diabetes; Disease; Epidemiology; Gene loci; Genes, Dominant; Genetic markers; Genetic Predisposition to Disease; Genetics; Genome-Wide Association Study - methods; Genome-Wide Association Study - standards; Genomes; Health risk assessment; Heart Rate - genetics; Heritability; Humans; Morbidity; Phenotype; Polymorphism, Single Nucleotide; Quantitative Trait, Heritable; Respiration - genetics; Single-nucleotide polymorphism; Studies; Netherlands
• ISSN: 1018-4813; 1476-5438; 1476-5438
• DOI: 10.1038/ejhg.2017.50

Despite the recent explosive rise in number of genetic markers for complex disease traits identified in genome-wide association studies, there is still a large gap between the known heritability of these traits and the part explained by these markers. To gauge whether this 'heritability gap' is closing, we first identified genome-wide significant SNPs from the literature and performed replication analyses for 32 highly relevant traits from five broad disease areas in 13 436 subjects of the Lifelines Cohort. Next, we calculated the variance explained by multi-SNP genetic risk scores (GRSs) for each trait, and compared it to their broad- and narrow-sense heritabilities captured by all common SNPs. The majority of all previously-associated SNPs (median=75%) were significantly associated with their respective traits. All GRSs were significant, with unweighted GRSs generally explaining less phenotypic variance than weighted GRSs, for which the explained variance was highest for height (15.5%) and varied between 0.02 and 6.7% for the other traits. Broad-sense common-SNP heritability estimates were significant for all traits, with the additive effect of common SNPs explaining 48.9% of the variance for height and between 5.6 and 39.2% for the other traits. Dominance effects were uniformly small (0-1.5%) and not significant. On average, the variance explained by the weighted GRSs accounted for only 10.7% of the common-SNP heritability of the 32 traits. These results indicate that GRSs may not yet be ready for accurate personalized prediction of complex disease traits limiting widespread adoption in clinical practice.