Genome-wide association (GWA) studies have revolutionized the field of genetics and identified many genetic variants associated to complex traits. However, a large part of the phenotypic variance remains unexplained. The limited yield is often attributed to traits being genetically highly complex (i.e., many variants are involved, each of small effect), and GWA studies may be underpowered to robustly detect very small genetic effects. An alternative explanation, however, can be found in how complex traits are operationalized in genetics studies. Especially in studies on traits like depression, personality and schizophrenia trait measures often consist of composite scores (e.g., case-control status, sum scores), aggregating information from many different symptoms (e.g., in depression: irritable mood, insomnia, worrying, concentration problems, and suicidal ideation). If the aggregated symptoms are genetically heterogeneous, item-specific genetic signals are strongly diluted when composite scores are analyzed.
In a large population sample (UK biobank; N~113,000), we studied in detail the genetic heterogeneity of 12 neuroticism items, which correlate strongly with depression. GWA analyses were conducted on the 12 items separately, and their sum. To assess genetic heterogeneity, we 1) estimated the SNP-based heritability of individual items, 2) established genetic correlations among the items and with external variables, and 3) examined the overlap in associated SNPs and genes across items.
We identified 19 independent genomic loci that reached genome-wide significance (GWS) at P < 5´10-8. Eight of these loci were only significant for (one or more) individual items, and not for the sum score. Six loci reached GWS for the sum score but not in individual items. Inter-item genetic correlations ranged from .31 to .92, suggesting substantial genetic heterogeneity. In addition, genetic correlations between neuroticism items and a number of external traits (e.g., BMI, IQ, bipolar disorder) showed variation in both strength and direction.
Our findings suggest that symptom-specific analyses can increase the genetic signal and yield novel biological information. Moreover, the variants identified through the use of a composite score may not be representative of all (or even the majority) of the individual items.
Gene Finding Strategies , Psychopathology (e.g., Internalizing, Externalizing, Psychosis) , Other
