Sources of variability in neuroimaging findings of frontal lobe dysfunction in schizophrenia: Single subject contributions to group effects

Prefrontal hypofunction is the most frequently reported finding in functional neuroimaging studies of schizophrenia. The meaning of this so-called ‘hypofrontality’ has remained unclear, being typically interpreted as impaired recruitment of prefrontal cortical neurons. Alternatively, however, between-group differences in variability along several parameters could produce such findings. A major impediment to more precise understanding of prefrontal hypofunction is that, until recently, functional activation maps could only be assessed by using group average statistics. In such group-average data the variance may be derived from a number of neurobiologically meaningful sources. These include variability across individuals in intensity of activation, average activity within a contiguous cluster of voxels, and variability in the neuroanatomical locale of the evoked neural activation. For example, inter-individual “scatter” of functional locales could result in fewer commonly activated voxels, with resultant underestimation of cortical response. These sources of variability may have a major impact on group signals. We recently developed methods for comparing patients to controls which, by assessing the activation pattern for each individual subject, separately evaluate multiple sources of variance, including peak activation intensity, spatial extent of an activated region, and variability in location of functional neuroanatomical signal. We piloted these methods in a study of the motor cortex, an area previously reported to be hypofunctional in schizophrenia and found that although group-average level analyses suggested patients to be hypofunctional, this finding was not supported by single-subject analyses. The group level finding was produced primarily by increased variability of activation locale in the patients.; These methods were applied to a study of patients and control subjects who performed a working memory test during PET scanning. Group-level analysis revealed dorsolateral prefrontal cortex (DLPFC) hypofunction in the patients. In contrast to the pilot study, single subject analyses in these subjects revealed that DLPFC activations of patients were significantly smaller in volume and with significantly lower magnitude of activation. However, the locales of patients activations were more variable than controls activations. This analysis demonstrates that patients have deficits in multiple parameters of the neuroimaging signal and confirms that prefrontal hypofunction in patients with schizophrenia represents true pathophysiology.