Harms et al 1 suggest that volume deficits in frontal regions of interest (ROI) represent a potential endophenotype worth investigating in schizophrenia. Cortical volume is a product of thickness and surface area. Harms et al's finding that volume but not thickness or surface area show some degree of familial sharing merits a critical analysis of the study.
Their conclusion is based on examining manually parcellated frontal subregions that were compared across patients with schizophrenia, siblings and healthy controls, using global measures that exclude the ROI as covariate for volume and surface area. Whole brain average thickness has been included as a covariate for thickness calculations. Although methods similar to this have been reported elsewhere, 2 this approach seriously affects the conclusions one can draw from the results.
First, the hypothesis behind the study is based on the idea that region-specific grey matter deficits are present in schizophrenia. Let us assume that schizophrenia has a pathological mechanism that selectively affects certain brain regions but does not affect the remaining cortex to similar extent. In this case, using an ROI-subtracted measure of global volume as a covariate will incorrectly inflate the estimates. Total intracranial volume would have been a more appropriate variable.
Second, for thickness measures, the appropriateness of using global thickness as a covariate is questionable. It is difficult to construe the anatomical meaning of regional thickness covaried with total cortical or hemispheric thickness, given the wide variability across the cortex. For analysing an a priori hypothesis involving thickness of frontal regions, a global covariate of average thickness appears redundant.
Choosing global values for adjusting regional measures is influenced by various factors, including actual ROI, disease process investigated, developmental age 3 and the cortical measure collected. 4 Familial trends in cortical thickness measurements in schizophrenia shown elsewhere 5 have not been replicated in this study. In healthy individuals, it has been shown that both total cortical surface area and average cortical thickness are highly heritable but not collinear. 6 Consequently, volume needs be treated as an ambiguous measure when exploring the cortical genetic variance.
Edited by Kiriakos Xenitidis and Colin Campbell