The dorsolateral prefrontal cortex (DLPFC) has figured prominently as abnormal in schizophrenia. Weinberger et al. (Weinberger et al., 1986) performed a now classic study implicating DLPFC in the working memory impairments observed in patients with schizophrenia. However, while the fMRI signal was attributed in this study to the site of working memory in DLPFC, the anatomical location of this activation pattern in the prefrontal cortex was not entirely clear, nor is it clear today.
An alternative and successful approach to localizing the site of working memory in DLPFC comes from work on nonhuman primates, where neurons in Brodmann area (BA) 46 and 9 exhibit persistent activity during working memory tasks (Goldman-Rakic, 1987). BA46 and BA9 also play an important role in working memory in humans (Petrides et al., 1993), and patients with schizophrenia show deficits in physiological activation especially in the BA 46 (Cannon et al., 2005; Perlstein et al., 2001). BA 46 is, however, difficult to localize with neuroimaging methods because neuroimaging methods tend to use gyrification pattern as a reference system, and BA46, defined based on cytoarchitectonic findings on post-mortem brains, does not follow the gyrification pattern. Brodmann originally mapped one human brain into various cytoarchitectonic areas (Brodmann, 1909). Later, the mapping of area 46 was done on five human brains (Rajkowska Goldman-Rakic, 1995) (Figure 1B) and was found to have major regions of overlap located between coronal Talairach coordinates +50 and +29 in the AP plane (Figure 1C). In comparison to Brodmann’s definition, area 46 is more rostral and is of smaller extension. In morphometric nomenclature, the region of BA46 is a part of the middle frontal gyrus (MFG) (Rajkowska Goldman-Rakic, 1995).
Two morphometric studies have explored BA46 in human subjects. The localization of BA46 is difficult to confirm because of the high variability of the gyral and sulcal patterns in individual subjects (Bartley et al., 1997; Noga et al., 1996) and because Brodmann areas do not always follow gyral patterns. Thus presently it is impossible to exactly determine the borders of BA46 in living subjects. Al-Hakim and co-workers nonetheless developed semi-automated rules to define an “average” BA46 in the MFG to be used in structural and functional imaging studies in living human subjects (Al-Hakim, 2006).
The second study, performed by Zuffante and coworkers (Zuffante et al., 2001), attempted to delineate the region of BA46 in schizophrenia subjects. These investigators were able to draw on coronal slices between the Talairach coordinates y=32 and y=45, but experienced technical difficulties between Talairach coordinate y=45 and y=50, and for this reason did not include this part. The problems encountered by Zuffante and colleagues are not surprising. The complexity and variation of the gyrification pattern of the cortex in the rostral part of MFG makes manual drawings in a 2-dimensional (2-D) view very challenging.
A promising approach to delineate the prefrontal cortex is based on 3D-reconstructions of the brain surface. Previously, 3D brain surface reconstruction were used as a visual guide to manually draw the ROIs of the prefrontal areas (Buchanan et al., 1998; Wible et al., 1997). In the current study, we used FreeSurfer software (http://surfer.nmr.mgh.harvard.edu/) to create a 3D reconstruction of the white matter surface. This software also automatically parcellates the brain into subregions, with the MFG delineated into rostral and caudal parts with high accuracy (Desikan et al., 2006). The caudal part of the MFG corresponds to the premotor area, while the rostral portion encapsulates a number of Brodmann areas, including BA46 (personal communication with Rahul Desikan). In addition, FreeSurfer has a tool that cuts brain regions along the coronal plane, which makes it possible to focus still further on the rostral part of the MFG by using the anterior and posterior boundaries as suggested by the cytoachitectonic studies. For the purposes of this study, we more narrowly delineated FreeSurfer based parcellations of the rostral MFG in order to depict BA46 and we refer to this region as “mid-rostral middle frontal gyrus” (MR-MFG).