T37 - An Expanded Latent Variable Model Of Psychopathology With Heritability Estimates

Abstract

Background Solid evidence points towards dimensional conceptualizations of psychiatric disorders, emphasising the importance of latent factors to account for comorbidity among disorders and to provide additional clinical information to diagnostic thresholds. Therefore, continuous measures of psychopathology and factor analysis are high-potential in identifying latent sources of covariation among observed psychopathology. A consistent result from latent variable modeling studies is that two broad constructs underlie common DSM disorders: Internalizing and Externalizing. However, whether the inclusion of frequently overlooked disorders results in changes in the bi-factorial meta-structure and whether this structure is invariant across age and gender is often unclear. Finally, these phenotypic correlations indicate that disorders within the two spectra have similar etiologies. Estimating heritability and examining common genetic polymorphisms, like SNPs, provides an opportunity to quantify the shared genetic etiology of mental disorders. Methods The 7707 individuals (4084 women, 3623 men) included in the present study were selected from a sample of twins who participated in the Genetics of Sexuality and Aggression (GSA) project, launched in 2005 at the Abo Akademi University in Finland. In addition to indicators of anxiety, depression, psychopathy and alcohol use, our factorial structure included measures of eating attitudes, body image, sexual functioning, anger and aggression. We carried out a series of exploratory factor analyses to estimate measurement models, testing our hypothesis about the number of underlying factors. The results of these analyses were used to develop confirmatory factor analyses and validate the relations between the scales and the emerged latent constructs. To clarify whether the factor structures were equivalent across gender and age, measurement and structural invariance tests were performed. Genotype data are available for estimating heritability (ongoing). Results Measures of psychopathy, aggression, anger and alcohol abuse loaded on a first factor, interpreted as Externalizing; depression, anxiety and sexual distress indicators loaded on a second, Internalizing factor; eating attitudes and body image scales created a third factor we named Body. This three-factor model showed significantly better fit compared to a two-factor and a one-factor models. However, the two factor model where body-related problems loaded onto Internalizing also showed acceptable fit, consistent with the existence of two main dimensions of psychopathology. Both models were valid for women and men, although levels of specific indicators differed across genders, determining differences in factor means. Models were not equivalent across generations. Discussion We investigated latent dimensions, rather than psychiatric categories, to understand pathological patterns and to elucidate how disorders interact in shaping the structure of psychopathology. We expanded the Internalizing-Externalizing meta-structure and highlighted a novel spectra of Body-related disorders, but our data still suggests that comorbidity between mental disorders is largely accounted for by two broad latent dimensions. Based one the GSA data, we are currently carrying out an extended twin family study with Single Nucleotide Polymorphisms. Heritability estimates will be presented at the Congress.