Univariate analysis of MRI endophenotypes

Multiple regression models were specified to estimate the associationbetween genetic liability and brain structural variation at each intracerebralvoxel with gray or white matter volume density as dependent variables, geneticliability score as the key predictor variable, and age, sex, and affectionstatus as covariates. Analyses were performed separately for the familieswith schizophrenia and bipolar disorder. A map of the standardized regressionmodel coefficient of interest (β) coding the association between anatomicalvariation and genetic risk at each voxel was thresholded such that if β>1.96(probability of β<0.05), the voxel value was set to β−1.96;otherwise the voxel value was set to 0. This procedure generated a set ofsuprathreshold voxel clusters in 3 dimensions, each with a mass, or sum ofsuprathreshold voxel statistics. We tested the null hypotheses of no associationbetween brain structure and genetic risk by permutation at cluster level,as described in detail elsewhere.^14,35,36 Stringentthresholds for statistical significance were derived from the permutationdistribution so that the expected number of false-positive test results ineach map was less than 1. Significant clusters were anatomically localized,and Brodmann areas were ascribed when relevant from the coordinates of thecentroid voxel and the 2- dimensional spatial extent of each cluster in eachaxial slice in accordance with the standard atlas of Talairach and Tournoux.³⁷

Дата добавления: 2015-08-17