Recent structural and functional imaging findings in schizophrenia Margaret A. Niznikiewicz, Marek Kubicki and Martha E. Shenton

Transcription

1 Recent structural and functional imaging findings in schizophrenia Margaret A Niznikiewicz, Marek Kubicki and Martha E Shenton Purpose of review Schizophrenia is a severe mental disorder that affects nearly 1% of the general population and has long been a challenge for both clinicians and researchers in terms of treatment and etiology More recently, evidence has amassed that suggests that schizophrenia is a brain disorder, and that some aspects of this disorder appear to be genetic It has only been in the past decade, however, that adequate neuroimaging tools have become available to study brain abnormalities in schizophrenia These technologic advances have led to a much better understanding of both structural and functional brain abnormalities in schizophrenia, and allowed construction of comprehensive theories regarding the neural underpinnings of schizophrenia The present review provides an account of research activities in the fields of structural magnetic resonance imaging (MRI), functional MRI, and diffusion tensor imaging during the past year, dating from August 2001 to August 2002 Recent findings In the review we discuss MRI findings in chronic schizophrenia and in first episode schizophrenia, and we include information on family studies We note that the two major hypotheses regarding schizophrenia (neurodevelopmental and/or neurodegenerative) provide a theoretical framework for most studies For the functional MRI studies, we note that questions focus on identifying networks of brain regions that are abnormal and contribute to functional abnormalities Summary Jointly, MRI, functional MRI, and diffusion tensor imaging research suggests that schizophrenia is a disorder that impacts on several brain regions, leading most notably to abnormalities in both frontal and temporal lobes, as well as to abnormal connections between those two major structures Keywords brain abnormalities, diffusion tensor imaging, functional magnetic resonance imaging, neuroimaging, schizophrenia Curr Opin Psychiatry 16: # 2003 Lippincott Williams & Wilkins Boston Veterans Affairs Healthcare System, Clinical Neuroscience Division, Laboratory of Neuroscience, Department of Psychiatry, Harvard Medical School, Boston, Massachusetts, USA Correspondence to Margaret A Niznikiewicz, PhD, Department of Psychiatry-116A, Boston VA Healthcare System, 940 Belmont Street, Brockton, MA 02301, USA Fax: ; margaret_niznikiewicz@hmsharvardedu Current Opinion in Psychiatry 2003, 16: Abbreviations ACC anterior cingulate cortex CSP cavum septi pellucidi DLPFC dorsolateral prefrontal cortex DTI diffusion tensor imaging (f)mri (functional) magnetic resonance imaging PTSD post-traumatic stress disorder STG superior temporal gyrus VBM voxel-based morphometry # 2003 Lippincott Williams & Wilkins Introduction Interest in elucidating the structural and functional brain abnormalities that occur in schizophrenia has a long history, dating back to Kraepelin s [1] and Bleuler s [2] early descriptions of schizophrenia For Kraepelin, the disorder, which he described as dementia praecox, or early dementia, was defined primarily by the course or outcome For Bleuler the disorder was defined more by the marked cognitive and emotional dysfunctions, and by a split between emotions and cognition, or what he described as schizophrenia However, both of those seminal thinkers believed that, ultimately, schizophrenia would be linked to a brain disorder Early work in this field was thus particularly discouraging because there were no tools available that allowed investigators to discern subtle brain abnormalities With the advent of neuroimaging techniques, beginning with computerassisted tomography findings in 1976 [3] and the first magnetic resonance imaging (MRI) scan of a schizophrenic brain in 1984 [4], we now know more about brain abnormalities in schizophrenia than at any other time in the history of schizophrenia research In the present review we highlight important recent structural and functional findings [ie MRI, functional (f)mri, and diffusion tensor imaging (DTI)] in schizophrenia Structurally, those findings suggest predominantly enlarged lateral ventricles and temporal lobe abnormalities, but functionally, they suggest more temporal frontal connectivity abnormalities We frame these findings within the context of research findings reported during the past 15 years We highlight structural MRI studies that address neurodevelopmental and/or neuroprogressive hypotheses of brain abnormalities, and functional studies that address specific hypotheses regarding several domains of cognitive dysfunction in schizophrenia We end with a brief review of DTI findings that demonstrate connectivity abnormalities in DOI: /01yco b 123

2 124 Schizophrenia schizophrenia, followed by a summary and discussion of future directions in neuroimaging studies in schizophrenia It is increasingly recognized that schizophrenia is a disease of the brain that profoundly affects most aspects of human function Although the exact nature of the dysfunction and its etiology remain unknown, substantial progress has been made in identifying brain abnormalities in schizophrenia and in mapping abnormal cognitive functions onto brain structures One reason for this progress is that the quality of in-vivo images using MRI has increased dramatically over the past 15 years We are now able to acquire and evaluate 1 mm slices through the entire brain, a spatial resolution that was previously impossible [5 ] Advances in fmri have also made it possible to go beyond examining isolated brain regions to examining interconnected neural networks that are probably implicated in schizophrenia [6 9] Additionally, the recent advent of DTI renders possible the investigation of abnormal connectivity among brain areas because it is well suited to examining white matter in the brain (for review [10 ]) In this review we present the most recent findings in schizophrenia from structural MRI, fmri, and DTI studies reported between 1 August 2001 and August 2002 We begin with a review of MRI structural findings in chronic schizophrenia and in first episode schizophrenia This is then followed by a review of fmri and DTI findings in schizophrenia We frame these recent findings within the context of MRI findings over the past 15 years Magnetic resonance imaging: structural findings in chronic schizophrenia Magnetic resonance imaging has continued to provide important information regarding the brain structure in schizophrenia, with most studies motivated by a keen interest in pursuing the possible causes of schizophrenia and in identifying regions affected by the disease process Overview and context Two reviews of MRI structural findings in schizophrenia have recently been published, one by Shenton et al [5 ] and the other by Kasai et al [11 ], that are quite comprehensive The review by Shenton et al spans the time period from 1988 to 2000, and reviews more than 190 MRI studies In patients with schizophrenia, as compared with normal control individuals, it documents the following: lateral ventricular enlargement (80% of studies); third ventricle enlargement (73% of studies); medial temporal lobe volume reductions (74% of studies), including amygdala, hippocampus, and parahippocampal gyrus; and neocortical superior temporal gyrus (STG) volume reductions (100% of studies), particularly on the left, with the percentage decreasing to 67% of studies when STG gray and white matter are combined There is also moderate evidence for frontal lobe volume reduction in schizophrenia (59% of studies), particularly prefrontal and orbitofrontal regions, and parietal lobe abnormalities (60% of studies), including both supramarginal and angular gyri abnormalities; the latter regions are important in language processing Finally, enlarged cavum septi pellucidi (CSP; 92% of studies), basal ganglia abnormalities (68% of studies), corpus callosum abnormalities (63% of studies), thalamus abnormalities (42% of studies), and cerebellar abnormalities (31% of studies) are also evident in schizophrenia Figure 1 [12] provides a lateral view of the brain that depicts the locations of the frontal, parietal, temporal, and occipital lobes, as well as major gyri in the brain, including STG Previous MRI findings thus highlight the importance of temporal lobe abnormalities in schizophrenia However, this is not to suggest that the temporal lobe is the only brain region that is abnormal in schizophrenia In fact, the temporal lobe is highly interconnected with other brain regions, including the frontal lobe [13 19], and normal brain function depends on such interconnectivity In the present review of fmri and DTI findings in schizophrenia, the role of neural connectivity in schizophrenia is further highlighted Neurodegenerative hypothesis There are two prevailing hypotheses concerning the etiology of brain abnormalities in schizophrenia: neurodevelopmental and neurodegenerative Shenton et al [5 ], as well as Weinberger and McClure [20 ], discussed the merits of these two theoretical approaches, and a review paper by Okubo et al [21] focused primarily on evidence for progressive changes in schizophrenia Although these approaches are not described in detail here, it is important to point out that neurodevelopmental and neurodegenerative approaches have a long history in schizophrenia They are often viewed as mutually exclusive, but are increasingly recognized as complementary [20 ], with the suggestion that schizophrenia (like many neurologic diseases) may be characterized by both neurodevelopmental and neurodegenerative processes Several, if not most, of the studies discussed here adopt those two perspectives as their theoretical anchor points Evidence for progressive changes in gray matter in chronic schizophrenia has been reported in separate studies conducted by Velakoulis et al [22], Hulshoff et al [23], and Convit et al [24] Velakoulis et al [22] used a voxel-based morphometry (VBM) approach to

3 Structural and functional imaging in schizophrenia Niznikiewicz et al 125 Figure 1 Lateral view of the human brain From Carpenter and Sutin [12]; reprinted with permission from Williams & Wilkins, New York, USA Inf, inferior; Sup, superior Percentral sulcus Sup frontal sulcus Central sulcus Postcentral sulcus Sup parietal lobule Intraparietal sulcus Inf parietal lobule Parieto-occipital sulcus Inf frontal sulcus Preoccipital notch Orbital gyri Lateral sulcus Horizontal fissure Medulla oblongata study gray matter changes across all brain regions, as a function of illness duration This approach offers a quick way to evaluate multiple brain regions at once, across many individuals Negative correlations (unrelated to the age of onset) with illness duration were found for right medial temporal, medial cerebellar and bilateral anterior cingulate, and white matter volume in the right posterior limb of the internal capsule Positive correlations were noted for globus pallidus (Table 1), suggesting the role for medication in the latter volumetric changes Evidence for progressive changes was also reported by Hulshoff et al [23] In their crosssectional study of gray and white matter changes in 159 schizophrenic patients and 158 normal control individuals (Table 1), a steeper regression for gray matter volume changes as a function of age was found in the schizophrenia group These findings support an earlier, important longitudinal study conducted by Mathalon et al [43 ] In patients with schizophrenia as compared with control individuals, those investigators reported progressive volume decline from baseline to rescan 4 years later both in frontal and temporal gray matter, as well as expansions in sulci and lateral ventricles Moreover, Mathalon et al observed that cortical gray matter in these brain regions showed an accelerated decline over time in those patients with the most severe symptoms These observations were interpreted as suggesting that a progressive pathologic process is evident in at least some patients with schizophrenia A different strategy was adopted by Convit et al [24], who parcellated the frontal lobes into subregions and compared the volumes of the subregions between young normal control individuals, old normal control individuals, and young schizophrenia patients (Table 1) Patients with schizophrenia and older normal control individuals both showed smaller volumes in the superior frontal gyrus and orbitofrontal regions, suggesting that both older control individuals and younger patients with schizophrenia exhibit frontal lobe volume reduction as compared with younger control individuals However, these similarities in volume reduction do not necessarily suggest similar mechanisms Moreover, differences in education between groups may have also influenced the findings In another study, Lawrie et al [44] evaluated change over time in magnetic resonance volume of the temporal lobe and amygdala hippocampal complex in control individuals and in relatives of patients with schizophrenia (defined as being in a family in which there were at least two first-degree or second-degree relatives diagnosed with schizophrenia; Table 2) Previous results showed a reduction in the amygdala hippocampal complex at baseline in relatives of patients with schizophrenia A subset of those people was then followed up over time, and results revealed no changes in temporal lobe or in the amygdala hippocampal complex in relatives at risk for schizophrenia (ie volume reductions noted at baseline were stable and did not change over time) However, when psychotic relatives

4 126 Schizophrenia Table 1 Magnetic resonance imaging structural findings in chronic schizophrenia Reference Ananth [25 ] Magnet/slice thickness 2 Tesla/15 mm Participants (n) Mean age (years) Sex [male/female (n/n)] MRI measures and major findings Chemerinski [26] Chow [27 ] Convit [24] Erbagci [28] Falkai [29] Goldstein [30 ] Hagino [31] 15 Tesla/15 mm and 3 4 mm 15 Tesla/15 mm 15 Tesla/12 mm 10 Tesla/3 mm (5-mm interslice gap) 15 Tesla/12 mm 15 Tesla/31 mm 15 Tesla/1 mm SZ 20 NC 20 SZ 45 NC 45 SZ 14 NC 14 (Note: 22q11 deletion syndrome and SZ) SZ 9 NC young 9 NC old 9 SZ 26 NC 29 SZ(MA) 14 SZ(U) 12 NC 10 SZ 40 NC 48 SZ 86 NC 79 SZ 378 NC 386 SZ 300 NC 305 SZ 275 NC 282 SZ 35 NC young 374 NC old 698 SZ 347 NC 286 SZ(MA) 336 SZ(U) 298 NC 244 (NC were younger) SZ male 466 SZ female 414 NC male 416 NC female 390 SZ 293 NC 24 SZ 10/10 NC 10/10 Males only SZ 7/7 NC 7/7 Males only SZ 11/15 NC 11/18 SZ(MA) 8/6 SZ(U) 8/4 NC 5/5 SZ 27/13 NC 27/13 SZ 46/40 NC 44/35 Voxel-based morphometry was used to measure global and regional gray matter Global gray matter differences were observed, but more importantly regional gray matter differences were observed, particularly in the left mediodorsal thalamus Other differences between groups involved the occipitoparietal cortex; premotor, medial, and orbital prefrontal cortices; and inferolateral temporal lobe White matter differences were observed between groups in the lateral optic radiation of the occipital cortex Of note, abnormalities in ventral and medial prefrontal cortices correlated with a positive family history of SZ There was no proportional reduction in gray matter with age in SZ Total cerebral gray matter and total cerebral surface area, as well as ventral frontal cortex (comprised of orbitofrontal and straight gyrus) were measured Findings showed no group differences for total cerebral gray matter, total cerebral surface area, or ventral frontal cortex or orbitofrontal cortex, but right straight gyrus showed smaller volume and shape in SZ The volume of ventral frontal cortex was negatively correlated with social dysfunction in SZ Total gray and white matter, lateral ventricles, as well as frontal, temporal, and parietal gray and white matter were measured Findings showed that total gray matter, and frontal, temporal and left parietal lobe gray matter were significantly reduced in SZ This was most prominent in the frontal and temporal lobes There was also an observed increase in the lateral ventricles in SZ Smaller volumes of superior and orbital frontal gyri were observed in both old NC and in SZ as compared with younger NC Old NC and SZ thus did not differ This suggests similarities between the SZ and aging NC in frontal lobe abnormalities (There were both age and education differences between groups) Third ventricle and adhesio interthalamica were measured Results showed that the adhesio interthalamica was more often absent in SZ than in NC Third ventricle was not correlated with presence or absence of adhesio interthalamica, as predicted Sylvian fissure and amygdala hippocampal complex were measured Results showed that amygdala hippocampus volume was reduced, and there was an abnormal asymmetry in SZ(U) Sylvian fissure did not show differences among groups The authors suggested that brain abnormalities involving medial temporal lobe structures are not familial 22 regions of interest were evaluated Relative to male NC, male SZ were characterized by reductions in frontomedial cortex, middle frontal cortex, cingulate gyri, paracingulate gyri, Heschl s gyrus, and Broca s area, but with increased volume in posterior cingulate and basal forebrain Planum temporale was also smaller on the right in male SZ, but larger on the right in female SZ Relative to female NC, female SZ were characterized by reductions in fronto-orbital gyri, basal forebrain, anterior cingulate gyri, and posterior supramarginal gyri, but exhibited increases in volume in cingulate gyri and right planum temporale No differences were noted for subcortical gray matter or for CSF Cavum septum pellucidum (the membrane separating the lateral ventricles) was measured Findings showed no group differences in the prevalence of large cavum septum pellucidi (744% in SZ; 747% in NC), and there were also no sex-mediated differences Such large cavum in NC is unusual, with most other studies reporting prevalence between 5 and 13% (continued opposite )

5 Structural and functional imaging in schizophrenia Niznikiewicz et al 127 Table 1 (continued ) Reference Hulshoff [23] Meisenzahl [32] Meisenzahl [33] Okugawa [34] Rajarethinam [35] Suzuki [36 ] Magnet/slice thickness 15 Tesla/12 and 16 mm 15 Tesla/3 mm 15 Tesla/3 mm 15 Tesla/15 mm 15 Tesla/3 mm 15 Tesla/1 mm Takahashi [37] 15 Tesla/1 mm Velakoulis [22] 15 Tesla/3 mm Velakoulis [38] 15 Tesla/15 mm Wang [39 ] 15 Tesla/125 mm Participants (n) SZ 159 NC 158 SZ 30 NC 30 SZ 44 NC 48 SZ 32 NC 32 SZ 20 NC 20 SZ 45 NC 42 SZ 40 NC 40 SZ 39 (Illness duration: 2 31 years) SZ 45 NC 139 SZ 15 NC 15 Mean age (years) SZ 356 NC 377 (Age range ) SZ 294 NC 291 SZ 302 NC 303 SZ 393 NC 386 SZ 335 NC 339 SZ 264 NC 261 SZ: male 264; female 259 NC male 255; female 248 SZ 36 (Range19 60) SZ 341 NC 3005 SZ 329 NC 309 Sex [male/female (n/n)] SZ 112/47 NC 106/52 Males only Males only Males only Males only SZ 23/22 NC 22/20 SZ 20/20 NC 20/20 MRI measures and major findings Total volume of cerebral and cereballar gray and white matter, and lateral and third ventricles were measured Total brain volume, total cerebral gray matter, and both prefrontal gray and white matter were reduced in SZ, and lateral ventricles and third ventricles were increased in SZ Results also showed steeper regression slopes for age and gray matter volume in SZ as compared with NC 3-mm slices were reformatted and resliced into 15-mm Planum temporale was measured No volumetric or asymmetry differences were observed, regardless of planum temporale definitions used 3-mm slices were reformatted and resliced into 15-mm Bifrontal temporal gray matter volume and generalized white matter volume deficits associated with allele 2 carriers Gray matter, white matter, and CSF were measured in the frontal, temporal, parietal, and occipital lobes Results showed gray and white matter reduction in the temporal region in SZ, as well as white matter reduction that was more widespread, and increased CSF in whole brain and in the frontal and temporal lobes in SZ The amygdala hippocampus complex (both amygdala and hippocampus) was evaluated on 1-mm reformatted 3-mm slices Results showed no group differences for the amygdala hippocampal complex, but negative correlations were observed between left amygdala and thought disorder, between left hippocampus and negative symptoms, and between left anterior and posterior hippocampus and both negative and positive symptoms in SZ A new method, adapted from fmri, namely statistical probability mapping (SPM), was used to measure gray matter and white matter of the brain Gray matter in males: gray matter was decreased in left superior temporal and middle frontal gyrus, and bilateral anterior cingulate gyrus Gray matter in females: gray matter was decreased in right anterior cingulate gyrus, the medial part of the right superior frontal gyrus, and right middle frontal gyrus Gray matter was increased in right precuneus, postcentral gyrus, left cuneus and precuneus, and right cerebellum White matter: both males and females exhibited decreases in the bilateral anterior limb of the internal capsule and superior occipitofrontal fasciculus Females also exhibited increased bilateral parietal lobes The volume of whole brain gray and white matter, and anterior cingulate gray and white matter was evaluated Results showed that right anterior cingulate (ACG) gray matter is reduced in female SZ, and there is a lack of right4left asymmetry in female SZ ACG white matter in female SZ also lacked normal right4left asymmetry 34/5 Voxel-based morphometry was used to measure temporal, cingulate, and cerebellar abnormalities Findings showed that right medial temporal, cerebellar, and bilateral anterior cingulate gray matter volumes correlated negatively with illness duration, and right globus pallidus volume correlated positively with illness duration 877%/59% Both two-dimensional measure of shape (volume loss in the head of the hippocampus) and volume distinguished between the two groups Males only Principal components analysis based on left right asymmetry vector fields was used; group differences were reported in the subiculum (continued overleaf )

6 128 Schizophrenia Table 1 (continued ) Reference Wible [40] Zuffante [41] Yucel [42] Magnet/slice thickness 15 Tesla/15 mm 15 Tesla/125 mm 15 Tesla/15 mm Participants (n) SZ 17 NC 17 SZ 23 NC 23 SZ 55 NC 75 Mean age (years) SZ 44 NC 40 SZ 433 NC 465 SZ 3599 NC 2908 Sex [male/female (n/n)] Males only Males only Males only MRI measures and major findings No group differences were observed for prefrontal gray matter, but right prefrontal white matter was reduced in SZ and this was also correlated with right hippocampal volume No group differences were observed between groups for volume of area 46 (middle frontal gyrus), although deficits were noted in SZ in both spatial and nonspatial working memory tasks Reduced folding of the left anterior cingulate cortex was found in SZ CSF, cerebrospinal fluid; MRI, magnetic resonance imaging; NC, normal control; SZ, chronic schizophrenia; SZ(MA), multiply affected family (in which there is more than one member with SZ); SZ(U), uni-affected family (in which only one member is diagnosed with SZ) were examined separately, there was a reduction in temporal lobe volume over time Lawrie et al suggested that there are abnormalities in temporal lobe structures in relatives at risk for schizophrenia, but that no changes occur in those structures over time; In contrast, such changes are seen in individuals who develop psychotic features, further suggesting that changes over time are observed only when psychosis is evident One negative finding among reports of progressive changes was reported by Ananth et al [25 ], who did not find proportional gray matter reduction with age in patients with schizophrenia However, they did report left mediodorsal thalamus abnormalities using VBM, as well as differences in occipitoparietal cortex, premotor, medial, orbital prefrontal cortices, and inferolateral temporal lobe in patients with schizophrenia as compared with control individuals (Table 1) VBM is a relatively new and promising technique, but it needs further refinement, including comparison with manual measurements of brain regions of interest (which are the standard for comparison), before it can widely be accepted as an accurate measure of brain regions of interest Neurodevelopmental hypothesis Support for a neurodevelopmental origin of schizophrenia is often sought by studying brain structures in which perinatal maturation coincides with increased risk for schizophrenia (ie second and third trimester) These include midline structures that are proposed to mediate attention and information processing [57], including CSP (two thin leaflets that fuse together as the corpus callosum and hippocampus develop; for more extensive review [5 ]) Other midline structures include the adhesio interthalamica (single or multiple connections between the medial surfaces of the two thalami), the corpus callosum, and the hippocampus Finally, other structures that may be affected by adverse perinatal event(s) include: the sulcal gyral pattern of the brain, as well as regions of the brain that are lateralized for specialized functions such as language, including STG and a component of that gyrus, namely the planum temporale (for review [5 ]) Midline structures Evidence emerging from studies of midline structures suggests subtle neurodevelopmental abnormalities affecting these brain areas Erbagci et al [28], for example, examined the adhesio interthalamica to determine whether they were present or absent (abnormal) in schizophrenia They found adhesio interthalamica to be absent more often in patients with schizophrenia than in normal control subject individuals, suggesting a neurodevelopmental abnormality in at least a subset of patients diagnosed with schizophrenia (Table 1) Hagino et al [31] examined CSP, another brain anomaly that is believed to be neurodevelopmental in origin They did not find group differences in the prevalence of large CSP in a group of 86 schizophrenia patients and 79 normal control individuals However, this finding is not consistent with 92% of studies showing enlarged CSP in patients with schizophrenia (for review [5 ]) Moreover, those investigators reported that 747% of control individuals showed abnormal CSP This figure is much higher than the range 5 13% that was generally reported for control individuals in previous studies Narr et al [47] examined the genetic contributions to altered shape of the corpus callosum in a study of dizygotic and monozygotic twins discordant for schizophrenia (Table 2) No group volume differences were found between affected and nonaffected individuals, but both affected and unaffected monozygotic twins shared an upward bowing of the callosum, which is suggestive of volumetric enlargement of the ventricles and is consistent with genetic rather than environmental influences on corpus callosum anatomy

7 Structural and functional imaging in schizophrenia Niznikiewicz et al 129 Table 2 Magnetic resonance imaging structural findings in childhood/adolescent onset schizophrenia or in relatives of patients diagnosed with schizophrenia Reference Relatives of SZ Magnet/slice thickness Participants (n) Mean age (years) Sex [male/female (n/n)] MRI measures and major findings Cannon [45 ] Harris [46] Lawrie [44] Narr [47] O Driscoll [48 ] Seidman [49 ] 15 Tesla/5 mm 15 Tesla/15 or 17 mm 10 Tesla/188 mm 10 Tesla/12 mm 15 Tesla/1 mm 15 Tesla/3 mm SZ 64 SZSIBS 51 NC 54 SZ 6 UP 12 NC 6 RELSZ 66 NC 20 Discordant cotwins: MZ 20; DZ 20 Control cotwins: MZ 20; DZ 20 RELSZ 20 NC 14 RELSZ 45 SZ 18 NC 48 (RELSZ = families with one or two 1st degree relatives with SZ) SZ 405 SZSIBS 403 NC 407 (SZ = SZ or SZAFF; SZSIBS = SZ or SZAFFSIB) SZ 38 UP 69 NC 38 RELSZ 231 NC 229 (RELSZ = at least two 1st or 2nd degree relatives with SZ) Discordant cotwins: MZ 483; DZ 49 Control cotwins: MZ 483; DZ 479 RELSZ 354 NC 362 RELSZ 446 SZ 432 NC 401 (Group differences in sex, education, and IQ) SZ 32/32 SZSIBS 22/29 NC 23/31 SZ 5/1 UP: positive history 4/2; negative history 2/4 NC 5/1 RELSZ 34/32 NC 13/7 Discordant cotwins: MZ 10/10; DZ 10/10 Control cotwins: MZ 10/10; DZ 10/10 RELSZ 9/11 NC 5/ 9 RELSZ 17/28 SZ 10/8 NC 27/21 FH predicted gray matter volume loss in the patient and sibling groups, most strongly in the temporal lobe An increase in CSF and sulcal enlargement was associated with hypoxia in patients only, whereas FH and sulcal enlargement in the temporal lobe were observed in siblings only The authors concluded that FH was associated with increased brain abnormalities in SZ and in their nonpsychotic siblings, but not in NC Whole brain and hippocampus were measured Results showed that positive history parents had a larger hippocampus volume than did SZ The authors speculated that positive history parents have some compensatory mechanisms or protective factors that result in larger hippocampi than in their affected children Previously these investigators reported reduced amygdala hippocampal complex volume in people at risk for SZ In this study, a subset of these patients, along with healthy NC, were rescanned on average 2 years later Results showed that there was not a reduction in temporal lobe and amygdala hippocampal complex over time for all high-risk relatives Instead, there was volume reduction in the right temporal lobe in high-risk individuals with psychotic features This finding suggests that brain structures may change over time with psychotic symptoms, but may remain stable (ie reduced but no progression) in individuals at risk for SZ who do not evince psychotic symptoms Lateral ventricle and third ventricle volume were evaluated, as well as corpus callosum area and vertical displacement (upward bowing) Results showed that both affected and unaffected MZ cotwins had significant callosal displacements There were no differences among groups in corpus callosum area measures Of note, lateral and third ventricle volume were associated with corpus callosum displacement The authors suggested that upward bowing of the corpus callosum may be a useful developmental marker for screening for SZ Amygdala anterior hippocampus and posterior hippocampus were evaluated Results showed reduced volume of amydgala anterior hippocampal volume as well as poor delayed verbal memory in RELSZ as compared with NC Across all participants there was a correlation between delayed verbal memory and amygdala anterior hippocampus The data provide an empiric link between verbal memory deficits and volumetric abnormalities in the amygdala anterior hippocampus in RELSZ Volumes of total cerebrum and of hippocampi were measured Results showed significantly smaller left hippocampus volumes in RELSZ, particularly for multiplex families (ie those with more than one person with SZ), but no differences were found in hippocampal volumes between SZ and their relatives Additionally, verbal memory and left hippocampal volume were positively correlated, particularly in the multiplex RELSZ The authors suggested that reduced hippocampal volumes and verbal declarative memory deficits reflect a vulnerability to SZ (continued overleaf )

8 130 Schizophrenia Table 2 (continued ) Reference Steel [50] Magnet/slice thickness 10 Tesla/15 mm Some sections tested on another 10 Tesla scanner/15 mm Participants (n) Mean age (years) Sex [male/female (n/n)] MRI measures and major findings Van Erp [51 ] 15 Tesla/13 mm 6 sibships (n = 18) [SIBSHIP = 1 patient with SZ, one obligate carrier (inherited genetic risk, no SZ), one noncarrier and no SZ] SZ 72 (60 SZ/ 12 SZAFF) US 58 NC 53 SZ sibships 462 Obligate 49 Noncarrier 452 SZ 402 US 407 NC 409 7/11 Whole brain, prefrontal and temporal lobes, caudate, lentiform and thalamic nuclei, as well as amygdala hippocampal complex, lateral, third and fourth ventricles were measured Results showed overall reduction in cortical gray matter in SZ siblings (5%) with pronounced loss in the amygdala hippocampus (12%) Obligate carriers were similar to SZ siblings in exhibiting amygdala hippocampal complex volume reduction, and were similar to unaffected noncarriers in exhibiting whole brain gray matter volume reduction Obligate carriers had smaller ventricles than did SZ siblings The authors concluded that reduced cortical gray matter was associated with phenotypic SZ, whereas reductions in medial temporal lobe structures such as the amygdala hippocampus were associated with a genetic risk for SZ Males only Hippocampus was measured and smaller hippocampus volume was found in SZ with FH, followed by SZ without hypoxia, their siblings, and NC There was no association between hypoxia and hippocampal volume in NC Of note, smaller hippocampus correlated with earlier age of onset in SZ Childhood/adolescent onset of SZ James [52] Levitt [53] Matsumoto [54] Thompson [55 ] 15 Tesla/5 mm 15 Tesla/14 mm 15 Tesla/15 mm 15 Tesla/ 15 mm AOSZ 16 NC 16 COSZ 13 NC 20 AOSZ 40 NC 40 COSZ 12 NC 12 AOSZ 166 NC 160 (Mean age at baseline MRI scan differed between groups) COSZ 142 NC 120 (IQ differences between COSZ and NC) AOSZ 155 NC 157 COSZ 139 NC 135 (Onset by age 12) AOSZ 9/7 NC 9/7 Children AOSZ 20/20 NC 20/20 COSZ 6/6 NC 6/6 Whole brain, lateral ventricles, third ventricle, temporal lobe, and medial temporal lobe structures (amygdala and hippocampus) were measured in adolescents with SZ and in NC SZ had follow-up scans on average 27 years later, and NC had follow-up scans on average 17 years later Results showed generalized ventricular enlargement (lateral and third ventricle), particularly in male SZ, which was most prominent on the left, as well as left amygdala volume reduction in SZ, and a trend toward left hippocampal volume reduction in SZ There was no progression of tissue loss in brain volumes or in lateral and third ventricular volumes over time Total brain, temporal lobe volume, amygdala, and hippocampus volume were measured Results showed that amygdala volume, especially on the left, was larger in the COSZ group than in NC There was also a trend finding for a reversal of the normal right4left amygdala asymmetry in COSZ Total gray and white matter of the superior temporal gyrus was measured in early onset SZ Total brain volume was also measured Results showed that total gray and white matter volume of the right superior temporal gyrus was smaller in AOSZ than in NC Bilateral volumes in AOSZ were correlated with age at onset of psychosis, whereas severity of thought disorder and hallucinations were inversely related to right superior temporal gyrus volume The authors suggested that these findings may reflect a neurodevelopmental disruption This study followed individuals over 5 years with three separate MRI scans Results showed that deficits progress from parietal areas, to temporal lobes, and sensorimotor and dorsolateral prefrontal cortices in patients with COSZ (continued overleaf )

9 Structural and functional imaging in schizophrenia Niznikiewicz et al 131 Table 2 (continued ) Reference Other Magnet/slice thickness Participants (n) Mean age (years) Sex [male/female (n/n)] MRI measures and major findings McCreadie [56] 05 Tesla/3 mm SZDYS 31 SZNDYS 31 NC 31 SZDYS 43 SZNDYS 44 NC 43 (Never medicated SZ from South India, with and without DYS) SZDYS 18/13 SZNDYS 18/13 NC 18/13 Cerebral hemisphere and lateral ventricles were measured on two contiguous slices Caudate and lentiform nucleus were measured Results showed that left lentiform gyrus was larger in SZDYS whereas right ventricle : hemisphere ratio was larger in SZNDYS than in NC When volume of lentiform nucleus was corrected for hemisphere in a smaller group of patients, volume differences between groups were not present The authors suggested that SZDYS may have striatal pathology that may represent a subset of SZ AOSZ, adolescent onset schizophrenia; COSZ, childhood onset schizophrenia; CSF, cerebrospinal fluid; DYS, dyskinesia; DZ, dyzygotic twin; FH, fetal hypoxia; IQ, intelligence quotient; MRI, magnetic resonance imaging; MZ, monozygotic twin; NC, normal control; RELSZ, relative of schizophrenia; SZ, schizophrenia; SZAFF, schizoaffective disorder; SZDYS, schizophrenia with dyskinesia; SZNDYS, schizophrenia without dyskinesia; UP, unaffected parents; US, unaffected siblings Amygdala hippocampal complex Three recent studies examined the amygdala hippocampal complex in schizophrenia Velakoulis et al [38] used both traditional volumetric measures and shape measures to evaluate the hippocampus in schizophrenia Both shape-based and volume-based analyses suggested reduced hippocampal volume in the patient group Specifically, the shape-based analysis indicated volumetric loss in the posterior portion of the hippocampus (Table 1) Those researchers suggested that abnormal receptor distribution in early development could lead to altered hippocampal development in terms both of shape and connectivity Figure 2 provides MRI images that depict the amygdala hippocampal complex, as well as the temporal lobe, frontal lobe, lateral ventricles, and STG Although amygdala and hippocampal volume reductions in schizophrenia are among the more robust findings, not all studies report volume reductions For example, in a recent study conducted by Rajarethinam et al [35], no group differences were found in hippocampus and amygdala volumes, although left amygdala volume correlated negatively with the severity of thought disorder in the patient group, and left hippocampal volume was correlated with negative symptoms in the patient group Of note, in that study 3-mm slices were used, as compared with use of 15-mm slices by Velakoulis et al [38] This difference in spatial resolution may explain the negative finding because differences in spatial resolution can adversely effect the ability to detect differences between groups Of further note, that reductions in left amygdala volume were correlated with increased formal thought disorder confirms several earlier studies that showed similar correlations (for review [5 ]) Falkai et al [29] were interested in evaluating families in which only the proband had schizophrenia versus families in which multiple family members were affected with schizophrenia They found more reduction as well as abnormal asymmetry in the amygdala hippocampal complex in families in which only the proband was diagnosed with schizophrenia, as compared with control individuals and with probands from families with multiple diagnoses of schizophrenia Those investigators interpreted this finding as suggesting that brain abnormalities in medial temporal lobe regions are not familial The question of how genetic transmission might impact on reduction in hippocampal volume was also addressed in a familial MRI study conducted by Harris et al [46] Hippocampal volumes were studied in patients with schizophrenia, in their parents (families in which only one parent had an ancestral history of schizophrenia), and in normal control individuals Unaffected parents with an ancestral history of schizophrenia had larger hippocampal volumes than did their affected offspring (Table 2) Those authors concluded that reduced hippocampal volume is not a genetically transmitted risk factor In addition, the unaffected parents with ancestral history of schizophrenia had the largest hippocampal volumes among the three groups, suggesting that such increases in volume may represent a protective or compensatory factor in a person at genetic risk for schizophrenia Of note, however, the proportion of males and females in the positive and negative history groups and in the control group was not equal, and this in itself might have affected the findings The findings reported by Harris et al [46] are not consistent with findings from a study conducted by Seidman et al [49 ] Those investigators examined

10 132 Schizophrenia Figure 2 Coronal 15 mm view of the brain Coronal view Sagittal view Right Anterior Frontal lobe Lateral ventricle Temporal lobe Superior tenporal gyrus Amygdalahippocampal complex Left Posterior found in the unaffected relatives was not significantly different from that found in patients with schizophrenia Seidman et al also reported a correlation between deficits in verbal memory performance and reduced left hippocampal volume This association is similar to that reported in previous structural MRI studies (for review [5 ]) and to that reported by O Driscoll et al [48 ], who found reduced anterior amygdala hippocampus in relatives of patients with schizophrenia as compared with control individuals, which was correlated with impairments in verbal memory These data thus provide an empiric link between verbal memory deficits and brain abnormalities in the amygdala hippocampal complex both in schizophrenia and in relatives of patients with schizophrenia Similar findings were also reported by Van Erp et al [51 ] They found the smallest hippocampal volume in patients with schizophrenia, followed by their full siblings, and normal control individuals In addition, the smallest hippocampal volumes were observed in schizophrenic patients with fetal hypoxia, a relationship not noted in any other group Sagittal view Front Back Coronal 15 mm view of the brain shows the temporal lobe, frontal lobe, superior temporal gyrus, amygdala hippocampal complex, and the lateral ventricles The location of the coronal slice is depicted in the sagittal view (Courtesy of Susan Demeo, Magdalena Hale Spencer, and Anders Brun) hippocampal volume as a marker of vulnerability for schizophrenia in both simplex (one person in the family with schizophrenia) and multiplex (two people in the family with schizophrenia) unaffected relatives of schizophrenic patients, patients with schizophrenia, and normal control individuals (Table 2) Using this genetic sampling approach, left hippocampus was found to be reduced in the unaffected relatives, especially those from the multiplex families Moreover, the volume The effect of fetal hypoxia on cortical gray matter loss and on volume of cerebrospinal fluid was also assessed in a study conducted by Cannon et al [45 ] in a group of patients with schizophrenia, full siblings of schizophrenic patients, and matched normal control individuals (Table 2) Fetal hypoxia predicted reductions in gray matter volume and increased levels of cerebrospinal fluid in patients and their siblings, but not in control individuals Of note, hypoxia did not correlate with white matter changes in any group These findings point to a scenario in which genetic vulnerability and environmental factors may interact in the development of schizophrenia, because brain regions particularly vulnerable to hypoxia, such as the hippocampus, were especially impacted in at least a subset of patients diagnosed with schizophrenia Finally, Steel et al [50] compared volumes of specific brain regions, including the amygdala hippocampal complex, in sibships with either one patient with schizophrenia, one obligate carrier without the disorder but with affected offspring, and one nonaffected noncarrier Obligate carriers shared reduced amygdala hippocampal volumes with schizophrenic patients, but not the whole brain, or frontal or temporal lobe reductions found in schizophrenic patients Those findings suggest that reduced cortical structures are associated with the phenotype of schizophrenia, whereas the reduction in the amygdala hippocampal complex is associated with genetic risk for schizophrenia in the absence of the disorder Of note in this regard, a recent twin study was conducted in post-traumatic stress

11 Structural and functional imaging in schizophrenia Niznikiewicz et al 133 disorder (PTSD) monozygotic twins discordant for PTSD [58] It found that the unaffected twin also showed reduced amygdala hippocampal volume as compared with monozygotic twins without PTSD, suggesting that such volume reduction may reflect a genetic vulnerability or predisposition to PTSD Cortical and subcortical sex differences The role of sex differences in mediating gene expression in schizophrenia was examined by Goldstein et al [30 ] They assessed a range of cortical and subcortical brain volumes, and hypothesized that sex-mediated brain differences in men and women with schizophrenia are present in cortical but not in subcortical regions Indeed, differential patterns of cortical brain abnormalities were found in schizophrenic men and women relative to their control groups (Table 1) No sex differences were found in asymmetry, with the exception of the planum temporale, in which male patients exhibited smaller planum temporale on the right and female patients showed larger planum temporale on the right, resulting in greater planum temporale symmetry in women This finding is not consistent with that of Meisenzahl et al [32], who reported no asymmetry differences in males using three different methods of planum temporale definition As expected, however, Goldstein et al found no sex-mediated group differences in subcortical regions, despite normally existing sex dimorphisms in these areas The authors interpreted their findings as supporting the role of neurodevelopmental factors in schizophrenia, especially because they relate to timing of the release of gonadal hormones and their relative distribution between cortical and subcortical brain regions Cingulate cortex and schizophrenia The anterior cingulate cortex (ACC) was examined in three studies Yucel et al [42] studied the morphology of the ACC in a group of 55 male patients and 75 normal control individuals The surface morphology of the ACC was used to classify paracingulate sulcus either as prominent, present, or absent The results suggest that patients had less leftward folding of the anterior cingulate than did control individuals, and these differences existed over and above those found in the entire left hemisphere The results were viewed as supporting evidence for the role of neurodevelopmental factors in the etiology of schizophrenia because cortical folding is established in the second and third trimesters of pregnancy, remains stable throughout life, and is not influenced by environmental factors such as alcohol abuse, neuroleptic exposure, or age In contrast to Yucel et al [42], two other studies [30,37] found group differences in ACC to be sex related (ie, observed only in women) Additionally, Takahashi et al [37] found reduced gray and white matter in the ACC, along with reduced asymmetry in women, in a study that examined ACC gray and white matter using highresolution three dimensional MRI Frontal cortex and schizophrenia Three studies evaluated the role of frontal cortex in schizophrenia In the first of those, Chemerinski et al [26] looked at the relationship between social function in male patients with schizophrenia and the morphology of the ventral frontal cortex, parcellated into orbitofrontal gyrus and straight gyrus Both volume and surface measures were completed (Table 1) Only the volume of the straight gyrus was found to be smaller in the patient group and, in agreement with the study hypothesis, was correlated with measures of social adjustment, both before and after illness onset This relationship became especially clear when negative symptoms were controlled for, suggesting that negative symptomatology and poor social function may represent two distinct features of the illness These data are consistent with previous findings suggesting that premorbid functioning, particularly poor premorbid social functioning, may be related to more frontal lobe abnormalities in schizophrenia (for review [5 ]) In the second study, Wible et al [40] analyzed the relative volumes of gray and white matter of the prefrontal cortex and its correlations with clinical variables Although no differences were found in gray matter, white matter was reduced in the right hemisphere in the patient group, and it correlated with volume of the right hippocampus in the patients only Also, patients with high negative symptom scores had lower white matter volumes than did those with low negative scores The fact that the patients in the study were characterized by more negative symptoms suggests the importance of evaluating more homogeneous patient groups, because this may make a large difference with respect to which brain regions are most affected Differences in genetic loading may also be important because there is some indication that patients with more negative symptoms may have more neurodevelopmental abnormalities and an earlier age of onset (for review [5 ]) Finally, in the third study, that conducted by Zuffante et al [41], a lack of volumetric changes in the frontal lobes (area 46) was noted, despite the presence of deficits on tasks of both spatial and nonspatial working memory Schizophrenia and gene expression More direct studies of genetic factors that might be associated with schizophrenia include those by Chow et al [27 ] and Meisenzahl et al [33] The latter investigators [33] examined the relationship between

12 134 Schizophrenia brain volumetric measures and individual genotypes Specifically, because higher plasma levels of interleukin- 1b are found in blood samples of schizophrenic patients, they examined the relationship between interleukin-1b allele 2 carriers and volumetric brain abnormalities Schizophrenic patients who were carriers had smaller volumes in the frontal and temporal gray matter and a generalized deficit in white matter volume There was no association between the interleukin-1b polymorphism at position 511 and schizophrenia Chow et al [27 ], in a preliminary study of 14 adults with schizophrenia who also suffered from 22q deletion syndrome (a disorder associated with increased risk for schizophrenia) and 14 age-matched and sex-matched control individuals, found that the patient group exhibited smaller total gray matter volumes in the temporal and frontal lobes White matter did not show volumetric differences between groups Taken together, these results suggest a scenario of how genetically mediated risk for schizophrenia might interact with other genetic and environmental vulnerability factors to result in schizophrenia Summary MRI structural findings in schizophrenia over the past year are generally in agreement with previous structural MRI findings in schizophrenia Multiple brain regions appear to be affected in schizophrenia, although rather than generalized deficits these abnormalities are predominantly in temporal lobe, including medial temporal lobe (eg amygdala hippocampal complex) and neocortical temporal lobe (ie STG, which includes planum temporale and Heschl s gyrus; see Goldstein et al [30 ]), followed by frontal, parietal [27,30 ], and occipital lobe abnormalities [34] In addition, family members of patients with schizophrenia appear also to evince brain abnormalities, including amygdala hippocampal volume anomalies [44,46,48,49 ] and corpus callosum anomalies [47] The recent and growing interest in studies of families and relatives is particularly important; such investigations provide crucial information because brain abnormalities observed in nonpsychotic relatives of patients with schizophrenia provide the best indicator of potential vulnerability markers of schizophrenia, independent of psychosis Such studies also provide important clues to possible early neurodevelopmental anomalies that reflect both liability or vulnerability to schizophrenia, and relate to the expression of schizophrenia Finally, we note new methodologic advances, such as the use of VBM [22,25,36 ], which offer a quick way to evaluate multiple brain areas across many individuals, in contrast to the usual labor intensive manual tracing of brain regions of interest Additionally, a more sophisticated method of asymmetry definition based on a vector field [39 ] offers what may represent a better assessment of asymmetry in such structures as hippocampus These methods, which need further refinement, will probably offer new possibilities for generating hypotheses that can then be followed up in more carefully delineated manual region of interest investigations Magnetic resonance imaging studies: first episode schizophrenia The extent to which brain abnormalities found in chronic schizophrenia are also present in first episode schizophrenia is the focus of a number of MRI studies Individuals with a first episode of schizophrenia provide a unique opportunity to examine the effects of a psychiatric process without the confounding factors of chronicity and exposure to chronic neuroleptic medications Additionally, studying first episode schizophrenia patients, as well as childhood and adolescent onset schizophrenia, and at-risk populations such as relatives of schizophrenic patients (see previous section) and individuals with schizotypal personality disorder is a fruitful strategy to elucidate the etiology of schizophrenia and its cognitive correlates [5,59,60] Again, as with the MRI studies of chronic schizophrenia, both neurodevelopmental and neurodegenerative models of schizophrenia provide a theoretical framework for many of the studies Initial abnormalities The majority of studies document that brain abnormalities exist at the onset of the illness Multiple brain areas approach Two studies examined multiple brain areas to look for evidence of an initial insult [61,62] Cahn et al [61] (see Table 3 for brain regions studied) found enlargement of the third ventricle only, suggesting to these investigators that other brain abnormalities reported in schizophrenia may appear later in the course of illness and also may be medication related Using the strategy of comparing volumetric changes across different clinical populations, Salokangas et al [62] examined differences in several brain areas (Table 3) in first episode schizophrenia, and in depression, with and without psychotic features First episode patients evinced reduced left frontal lobe volumes, whereas depressive patients with psychosis evinced enlarged ventricular and posterior sulcal cerebrospinal fluid volumes Patients with depression without psychotic features only evinced larger white matter volumes than other clinical groups

13 Structural and functional imaging in schizophrenia Niznikiewicz et al 135 Table 3 Magnetic resonance imaging structural findings in first episode schizophrenia Reference Cahn [61] Frumin [63] Gunduz [64] Joyal [65] Keshavan [66] McCarley [67 ] Salokangas [62] Sumich [68] Szesko [69] Lee [70 ] Wood [71] Magnet/slice thickness 15 Tesla/ 16 mm 15 Tesla/ 15 mm 15 Tesla/ 5 mm, 1 mm in plane resolution 15 Tesla/ 15 mm 15 Tesla/ 5mm (1-mm interslice gap) 15 Tesla/ 15 mm 15 Tesla/ 54 mm 15 Tesla/ 15 mm 10 Tesla/ 31 mm 15 Tesla/ 15 mm 15 Tesla/ 15 mm Participants (n) SZ 20 NC 20 SZ 14 AFF 19 NC 18 SZ = 51 (SZAFF 4, SZPHRE13) NC 28 SZ 18 NC 22 SZ 31 Non-SZP 12 NC 31 SZ 15 AFF 18 NC 18 SZ 11 PD 20 Non-PD 17 NC 19 SZ 25 NC 16 SZ 75 NC 36 SZ 22 AFF 20 NC 24 SZ 30 CHRSZ 12 NC 26 Mean age (years) SZ 2763 NC 2724 (Neuroleptic naïve) SZ 281 AFF 258 NC 245 (Minimal neuroleptic exposure) SZ 245 NC 258 (Neuroleptic naïve and minimal exposure) SZ 28 NC 30 (Neuroleptic naïve) SZ 2420 Non-SZP 2283 NC 2509 SZ 276 AFF 235 NC 249 SZ 366 PD 340 Non-PD 384 NC 305 SZ 240 NC 270 (Minimally treated with neuroleptics) SZ 247 SZ 273 NC = 253 SZ 260 AFF 226 NC 240 SZ 218 CHRSZ 336 NC 238 Sex [male/female (n/n)] SZ 16/4 NC 16/4 SZ 11/3 AFF 14/5 NC 16/2 SZ 37/14 NC 17/11 SZ 11/7 NC 14/8 SZ 20/11 Non-SZP 6/6 NC 20/11 SZ 12/3 AFF 15/3 NC 15/3 SZ 3/8 PD 8/12 Non-PD 8/9 NC 12/7 All male SZ 43/32 NC 24/12 SZ 17/5 AFF 15/5 NC 21/3 SZ 19/11 CHRSZ 11/1 NC 14/12 MRI measures and major findings ICC, total brain, frontal lobe (gray and white matter), cerebellum, hippocampus, parahippocampal gyrus, thalamus, caudate nucleus, and lateral and third ventricles were measured Only third ventricle differed in SZ as compared with NC, with enlarged third ventricle in SZ The investigators concluded that this paucity of findings in first episode SZ suggests that brain abnormalities develop over time as seen in chronic SZ, which may therefore be related to a progression of the illness or to the effects of neuroleptic medications over time (There were differences in education between groups) Corpus callosum area and shape measures were completed Group differences were noted in shape for SZ versus NC, but no differences between groups were observed for area measures of the corpus callosum Total brain volume, caudate nucleus, putamen, nucleus accumbens volume, and subcommissural limbic forebrain volume were measured Findings showed no group differences in volumes of the basal ganglia, although age was correlated with volumes of caudate and putamen in NC (There were differences in education and parental socioeconomic status between groups) Entorhinal cortex and whole brain volume were measured Findings showed smaller entorhinal cortex volume in SZ versus NC Area measures of the corpus callosum were completed Findings showed smaller area of corpus callosum, its anterior genu, anterior body, isthmus, and anterior splenium in SZ as compared with NC and non-szp Agerelated increases in corpus callosum size observed in NC were not seen in patients, suggesting neurodevelopmental abnormalities ICC, total gray and white matter, planum temporale, and Heschl s gyrus were measured Findings showed smaller gray matter volume of the left planum temporale and smaller Heschl s total gray matter volume in SZ, which differed from AFF and NC Frontal and temporal lobe gray and white matter were measured, as well as lateral ventricles and sulci Findings showed reduced gray matter volume in left frontal lobe in SZ group as well as larger ventricular and posterior sulcal cerebrospinal fluid volumes in the PD group, and larger white matter volumes in the non-pd group Whole brain, Heschl s gyrus, planum temporale, hippocampus, and amygdala were measured Findings showed smaller hippocampus volumes and smaller left planum temporale volumes in the SZ group Hippocampal volume and ICC were measured Findings showed that anterior hippocampal volume was correlated with motor and executive functions in men with SZ Whole brain gray and white matter and CSF were measured Findings showed smaller volume of left fusiform gyrus in SZ versus NC and AFF Right fusiform gyrus volume was also smaller in SZ as compared with the NC group only ICC, temporal lobe, hippocampus, and whole brain were measured in a follow-up study averaging 22 years following the first MRI scan Findings showed whole brain volume loss in both chronic and first episode SZ at follow up AFF, affective disorder with psychosis; CHRSZ, chronic schizophrenia; CSF, cerebrospinal fluid; ICC, intracranial contents (brain, ventricular system, and subarachnoidal CSF); MRI, magnetic resonance imaging; NC, normal control; non-szp, nonschizophrenic psychotic; PD, psychotic depression; SZ, first episode schizophrenia; SZAFF, schizoaffective; SZPHRE, schizophreniform

14 136 Schizophrenia Temporal lobe structures (amygdala hippocampal complex, entorhinal cortex, and superior temporal gyrus) Four studies looked directly at STG volumetric abnormalities [53,54,67,68] [The study conducted by Levitt et al [53] is discussed in detail under Change over time (see below)] A study conducted by McCarley et al [67 ] focused on gray matter in STG, amygdala hippocampal complex, and parahippocampal gyrus in first episode schizophrenia, patients with a first episode of affective psychosis (mainly manic), and in normal control subject individuals (Fig 2) Patients with first episode schizophrenia had smaller gray matter volumes of left posterior STG, left planum temporale, and total Heschl s gyrus volume relative to control individuals and to affective patients Similar conclusions were drawn in a study conducted by Sumich et al [68], in which first episode schizophrenic patients were found to have smaller volumes of hippocampus bilaterally and smaller left planum temporale relative to control individuals In a study of adolescent persons with early onset schizophrenia, Matsumoto et al [54] found reduced right rather than left superior temporal gyrus, and severity of thought disorder and hallucinations were inversely correlated with volume of the right superior temporal gyrus, suggesting that this pattern of abnormalities may be characteristic of early onset schizophrenia Of note, this asymmetry finding is the reverse of that reported in adult onset schizophrenia, in which left STG is frequently reported to be reduced [67 ] The relationship between hippocampal volume and neuropsychological function was evaluated by Szeszko et al [69] in 43 men and 32 women with first episode schizophrenia Anterior hippocampus was correlated with both executive and motor functions in men, although no significant correlations emerged for women These results further suggest that medial temporal lobe abnormalities are present at first episode and that such abnormalities are associated with what are considered frontal lobe functions, therefore further suggesting an abnormality in fronto temporal neural circuitry in schizophrenia These findings also suggest a role for sex in mediating both structural and functional abnormalities in schizophrenia Joyal et al [65] measured the entorhinal cortex in neuroleptic naïve patients with schizophrenia and control individuals Those investigators reported reduced entorhinal cortical volume in the patient group This finding is consistent with previous reports of entorhinal cortical volume reduction in schizophrenia (for review [5 ]) Fusiform gyrus Lee et al [70 ] reported reduced left fusiform gyrus in first episode patients with schizophrenia relative to first episode patients with affective psychosis (mainly manic) and normal control individuals The fusiform gyrus volume on the right was smaller in schizophrenic patients as compared with controls but not smaller than in the affective group This gyrus is important in the identification of human faces, and volume reduction here may be relevant to problems that patients with schizophrenia experience with respect to social cues Midline structures Keshavan et al [66] measured the area of the corpus callosum in neuroleptic naïve first episode patients, nonpsychotic patients, and normal control individuals Reduced area was found in the anterior genu, body, isthmus, and splenium subdivisions of the corpus callosum, which connect heteromodal association areas, but not in those areas that connect primary cortices, in first episode schizophrenia patients, followed by nonschizophrenic, and psychotic patients In addition, agerelated growth in the size of the corpus callosum was present in normal persons but not in first episode schizophrenia patients The authors viewed these results as supporting a neurodevelopmental abnormality, which impacts primarily on association cortices Frumin et al [63] also reported abnormal shape but not volume in first episode schizophrenia patients, but not in affective psychosis patients Subcortical structures Gunduz et al [64] examined the integrity of parts of the basal ganglia (caudate nucleus, nucleus accumbens, putamen, subcommissural limbic forebrain) in first episode, medication naïve schizophrenia patients No volumetric differences were found in basal ganglia structures between neuroleptic naïve first episode schizophrenic patients (n = 53) and normal controls (n = 28) Of note also is the study conducted by McCreadie et al [56], in which gray matter volume in components of the basal ganglia (putamen and caudate nucleus), as well as volume of the lateral ventricles, were measured in never medicated patients in rural India, who were estimated to have been ill for approximately 10 years One group of patients suffered from dyskinesia whereas the other did not Results showed larger left putamen in the dyskinesia patients; the patients without dyskinesia evinced a higher lateral ventricle : hemisphere ratio, especially on the right side The authors concluded that schizophrenic pathology may interfere with normal, agerelated changes in the basal ganglia, which are independent of medication effects (Table 2) Several methodologic limitations preclude clear-cut interpretation of the findings

15 Structural and functional imaging in schizophrenia Niznikiewicz et al 137 Change over time Change over time was explicitly addressed in four studies [52,53,55,71] In the study conducted by Levitt et al [53], volumetric changes in medial temporal lobe regions were evaluated in a group of early psychosis schizophrenic patients (mean age 14 years; mean time of onset to scan time 42 years) Although childhood onset schizophrenia patients are different from first episode populations, in this group no volumetric changes were observed in the hippocampus, but the amygdala was enlarged in the patient group, with the group difference more pronounced on the left side There was also a trend toward left greater than right asymmetry in the amygdala volume in the patient group (Table 2) Both the study by Thompson et al [55 ] and that by Wood et al [71] found evidence of ongoing neurodegenerative processes, whereas James et al [52] did not Using a new method for evaluating the brain, Thompson et al [55 ] tracked the rate of progression of gray matter loss in children with childhood onset schizophrenia (aged years) Results suggest an age-dependent loss of gray matter, which is progressive and most pervasive in the parietal lobe early, with progression to superior frontal and motor areas, and lateral temporal and dorsolateral prefrontal cortex Of particular note, the rate of progression in this group surpassed normal gray matter loss observed in normal control individuals Furthermore, to examine progressive changes over time, Wood et al [71] conducted a study of first episode and chronic patients with a 22-year time interval between MRI scans No time-related changes were observed in the hippocampus or in the temporal lobe in the first episode group However, whole brain volumetric loss was observed in both first episode and chronic groups, and the rates did not differ between the two patient groups In contrast, James et al [52], in their study of adolescent onset schizophrenia cases, demonstrated enlarged lateral and third ventricles and reduced left amygdala, and a trend toward left hippocampal volume reduction at baseline as compared with control individuals (Table 2) Scans taken at an average of 17 years later for control individuals and 27 years later for the adolescent onset group revealed no reduction over time The authors interpreted these findings as suggesting a nonprogressive neurodevelopmental disorder that impacts on brain structures before the onset of clinical symptoms However, the small sample, the different time intervals between scans for patient and control groups, and the relatively thick 3-mm slices all limit the generalizability of these findings It is also quite likely that early onset schizophrenia constitutes a subtype of schizophrenia that is unrelated to adult onset schizophrenia Summary First episode patients evince brain abnormalities, most notably in medial temporal lobe (amygdala hippocampal complex, entorhinal cortex), STG, corpus callosum, left planum temporale and left Heschl s gyrus, left frontal gray matter, and left fusiform gyrus They also generally have larger lateral ventricles than do control individuals These findings are similar to those reported in chronic schizophrenia In addition, there is some indication that initial brain abnormalities are present at the onset of illness, suggesting neurodevelopmental influences Furthermore, there is some indication that certain brain regions may show progressive changes over time, although far more research work following a large number of first episode patients over time is needed before conclusions can be drawn regarding brain changes over time Functional brain imaging studies fmri has been increasingly used to study abnormal cognitive processes in schizophrenia because it is less invasive and costly than positron emission tomography and because it promises better localization of function than does positron emission tomography Because the fmri signal reflects, indirectly, the relationship between neural activity in response to a stimulus and blood flow levels within brain regions, it can, at least in principle, aid in identifying brain regions that exhibit abnormal response to cognitive task requirements As such, fmri can complement other imaging techniques such as structural MRI, which offers an excellent spatial resolution, and event-related potentials, which offer excellent temporal resolution, to provide information on where in the brain cognitive operations are disturbed However, its ultimate success crucially depends upon appropriate experimental designs and data analysis techniques [72,73] Most recent fmri studies focused on deficits identified by neuropsychologic and clinical research as central to schizophrenia, including working memory, language function, and thought disorder and auditory hallucinations, as well as social cognition and correlates of abnormal saccadic eye movements More recently, fmri has also been used to ascertain the impact of medication and behavioral therapy treatments In addition, studies have examined abnormal motor function [74,75] and some have focused on face and affect processing [76,77] In most studies, the focus has been on identifying a network of brain regions that exhibit abnormal activation and that are involved in the cognitive processes under investigation Figure 3 provides an illustration of fmri activation differences between normal control indivi-

16 138 Schizophrenia duals and patients with schizophrenia, in the inferior frontal lobe as a result of semantic encoding operations Of note, the activation observed in control individuals was lacking in patients with schizophrenia Below, we review relevant fmri studies conducted over the past year; the studies are grouped according to the type of cognitive operation investigated Executive function Prefrontal cortex is involved in various aspects of executive functioning and has been studied extensively in schizophrenia Weinberger et al [78 ], for example, viewed prefrontal dysfunction and its association with impaired working memory function as central to the pathophysiology of schizophrenia Those investigators suggested that the catechol-o-methyl-transferase gene is involved in regulation of frontal lobe function, especially with respect to working memory tasks, and as such may represent a candidate genetic risk factor for schizophrenia Two central objectives regarding the role of prefrontal cortex in schizophrenia are to identify whether the conditions that lead to abnormal dorsolateral prefrontal cortex (DLPFC) response are manifested as either hypoactivation or hyperactivation, and to identify the network of functional connections between brain areas that are involved in supporting specific cognitive operations With regard to the latter, the connections between DLPFC and temporal cortex are especially relevant to our understanding of functional abnormalities in schizophrenia Barch et al [79 ] looked at the role of prefrontal cortex in working and long-term memory deficits in schizophrenia during the processing of verbal and nonverbal stimuli Those investigators found that right DLPFC showed decreased activation in both working memory and long-term memory tasks in patients with schizophrenia (Table 4), as well as decreased activation in hippocampal regions, basal ganglia, thalamus, and parietal cortex This suggests that in schizophrenia there are impairments in the neural circuits that are necessary to perform cognitive routines common to both working memory and long-term memory tasks, rather than impairments in discrete functional areas However, in a study of logical reasoning conducted by Ramsey et al [81], prefrontal activation was normal in medicated patients, after correcting for performance between patients and normal control individuals This result is consistent with some previous studies [96,97] that did not find hypofrontality in patients, when both patients and control individuals performed equally well on a task (Table 4) In the same design, nonmedicated patients showed activation levels that were higher than those in normal control individuals These results suggest that performance levels across groups may be an important factor in interpreting the results of activation studies Additionally, the result found in unmedicated patients suggests that efficiency of neural communication may be compromised in schizophrenia Similarly, no group differences in the DLPFC were found by Paulus et al [82] in a study of decision making (Table 4) However, abnormal activation was found within the fronto-parietal network in patients with schizophrenia (ie less activation in the inferior, medial prefrontal, and right superior temporal cortex, and more activation in the postcentral and inferior parietal cortex) However, interpretation of these results is difficult because, in addition to the complexity of the process studied, nonmedicated patients as well as patients medicated with typical and atypical neuroleptics were included in the study An indication of abnormal fronto-parietal connection in schizophrenia was also found by Honey et al [83] In the working memory study, in which no group differences were found in performance accuracy, a correlation was found between increased reaction time and activation in parietal cortex in the normal control group, whereas no such correlation was detected in the patient group (Table 4) The functional integrity of prefrontal brain regions in mediating inhibitory function was investigated by Rubia et al [84] using stop and go/no go inhibitory motor tasks In the stop task there was less activation in the left dorsolateral superior frontal gyrus and an increase in subcortical areas in the schizophrenia group, with equivalent behavioral performances in the patient and normal control groups, suggesting involvement of an abnormal neural network in inhibitory function The role of ACC in mediating error monitoring was examined in normal control individuals and schizophrenic patients using an event-related fmri design during a continuous performance task [85 ] When the target detection was made more challenging by degrading the quality of the visual display, more errors were committed by both groups Healthy individuals showed a responserelated increase in activation in the anterior cingulate gyrus related to error detection Such an increase in anterior cingulate gyrus activation was not observed in schizophrenic patients Moreover, anterior cingulate gyrus activation was associated with impairments in performance adjustment, suggesting that abnormal internal error monitoring in schizophrenia may be related to functional abnormalities in the cingulate cortex A

17 Structural and functional imaging in schizophrenia Niznikiewicz et al 139 caveat to these results is that all patients were medicated, which might have contributed to the limited response in the cingulate Kumari et al [86] examined functional correlates of procedural learning often found to be abnormal in schizophrenia In normal control individuals there was increased activity in brain regions associated with procedural learning, including striatum, thalamus, cerebellum, precuneus, medial frontal lobe, and cingulate gyrus In schizophrenic patients, however, the only structure activated was the anterior portion of the inferior gyrus Importantly, patients with schizophrenia did not exhibit evidence of procedural learning Again, the results indicate a possible role of activated structures in procedural learning, and abnormal procedural learning in patients, which may stem from the inadequate engagement of these structures Overall, results of many of the studies involving different aspects of memory and executive function indicate that functional abnormalities in schizophrenia are not only a matter of reduced activation in the same brain region as in normal control individuals (as demonstrated in some Figure 3 Semantic processing functional magnetic resonance imaging study z = 8 mm T value Results are shown for a semantic processing functional magnetic resonance imaging study The axial image shows difference in activation in the left inferior prefrontal cortex (area in yellow) between nine control individuals and nine schizophrenia patients, which reflects group differences in semantic encoding in a levels of processing paradigm (Courtesy of Marek Kubicki) studies) but also a matter of engaging different regions or, alternatively, using different strategies mediated by the same region, as demonstrated by other studies For example, Zorrilla et al [80], using a picture recognition paradigm, showed that in the patients there was a positive correlation between recognition memory and activation in the parahippocampal and hippocampus gyri during encoding, whereas in normal control individuals there was a negative correlation between these two variables Saccadic eye movements Two recent papers discussed neural substrates of saccadic inhibition deficits in schizophrenia, which are believed to represent a vulnerability marker for schizophrenia and have been associated with a dysfunction in the dorsolateral prefrontal cortex [98 100] However, the neural bases of the antisaccadic deficit are not well understood [101,102] The two studies also provide nonconvergent results, probably due to different methodologies (Table 4) In one of them, McDowell et al [87] analyzed blood oxygenation level dependent signal change between refixation (fixation baseline) and antisaccades (refixation saccade baseline) Under these conditions, only DLPFC was found to exhibit less activation in patients with schizophrenia during antisaccade generation (Table 4) In the other study, Raemaekers et al [88] used an eventrelated fmri design Brain activity generated during saccade and antisaccade tasks were compared across brain regions involved both in the generation of saccades and in structures involved in the transmission of an inhibitory signal to brainstem regions In this design, group difference existed in the striatum but not in the DLPFC, suggesting that striatum structures are part of a network that governs saccadic eye movement Language function in schizophrenia Abnormal language function, in its many manifestations, including thought disordered speech and verbal hallucinations, is a hallmark symptom of schizophrenia It has been proposed that abnormal lateralization of language areas may be an underlying factor Functional lateralization for language was examined by Sommer et al [89 ] Those investigators used a lateralization index that was calculated for several areas supporting language function (Table 4) Lower scores on the lateralization index were found in schizophrenic patients, which was due to increased activation in the right hemisphere rather than to reduced activation in the left hemisphere, suggesting an inefficiency in inhibiting the nondominant hemisphere The lateralization index was also correlated with the severity of hallucinations

18 140 Schizophrenia Table 4 Functional magnetic resonance imaging findings in schizophrenia Reference Participants (n) Mean age (years) Executive and memory function studies Sex [male/female (n/n)] Task design Major findings Barch [79 ] SZ 38 NC 48 Zorilla [80] SZ 8 NC 10 Ramsey [81] Exp 1: SZ 10; NC 10 Exp 2: SZ (medication naïve) 11; NC 10 Paulus [82] SZ 15 NC 15 Honey [83] SZ 20 NC 20 Rubia [84] SZ 6 NC 7 Carter [85 ] SZ 17 NC 16 Kumari [86] SZ 6 NC 6 Saccadic eye movement studies SZ 363 NC 365 SZ 542 NC 619 Exp 1: SZ 289; NC 241 Exp 2: SZ 277; NC 284 SZ 417 NC 410 SZ 346 NC 393 SZ 400 NC 400 SZ 335 NC 341 SZ 3467 NC 3138 SZ 24/14 NC 23/25 SZ 3/5 NC 8/2 Exp 1: SZ 9/1; NC 7/3 Exp 2: SZ 8/3; NC 7/3 All male All male All male SZ 12/5 NC 11/5 All male WM and LTM tasks Both verbal and nonverbal stimuli were presented within three tasks: WM-n-back task; encoding task; and recognition task Block design; whole brain coverage Picture encoding and recognition task Logical reasoning, a version of XT-task Block design; whole brain coverage Two-choice prediction task Block design; whole brain coverage Verbal n-back task Blocked periodic BA design; whole brain coverage Motor response inhibition stop and go/no-go tasks Block design; whole brain coverage Error monitoring task Subjects detected visual cues in either clear or degraded display conditions in alternating blocks Event-related design; whole brain coverage Procedural, sequence learning task Block design; whole brain coverage Impaired activation was found in SZ in the right DLPFC and the medial temporal lobe for both WM and LTM tasks Other regions found to have impaired activation during both WM and LTM included basal ganglia, thalamus, and parietal cortex In NC, a negative correlation was found between level of activation in parahippocampal gyrus and hippocampus for the recognition task In SZ, a positive correlation between the same variables was found After correcting for performance, brain activation was not different between NC and medicated SZ In unmedicated SZ, brain activity was elevated relative to NC No differences in activation were noted in the right prefrontal cortex in SZ and NC Less activation was found in inferior, medial prefrontal, and right superior temporal gyrus in SZ More activation was also found in right prefrontal and bilateral parietal cortex in medicated SZ relative to unmedicated SZ No group difference was found in activation levels in the prefrontal and parietal regions A lack of correlation between reaction time and level of activation in parietal regions was found in SZ In both the stop and go/no-go tasks, reduced activation was observed in the left anterior cingulate in SZ During the stop task, increased signal was observed in the right and left dorsomedial and ventromedial thalamus, and right putamen in SZ NC activated anterior cingulate, right medial frontal, and left posterior parietal cortex during error commission No errorrelated increase in brain activity was found in SZ Procedural learning was associated with activation in the striatum, thalamus, cerebellum, precuneus, medial frontal lobe, and cingulate gyrus in NC In SZ, only anterior inferior gyrus was activated McDowell [87] Raemaekers [88] SZ 14 NC 13 SZ 16 NC 17 SZ 370 NC 350 SZ 279 NC 259 SZ 10/4 NC 12/1 SZ 13/3 NC 10/7 Refixation saccade and antisaccade tasks Block design; whole brain coverage Occulomotor task: prosaccade, antisaccade, and active fixation Event-related design; whole brain coverage SZ patients did not show increased DLPFC activation during the antisaccade task as was observed in NC Overall, activation was reduced in SZ relative to NC in all regions The interaction between level of activation during the inhibitory task and illness was significant for striatum only SZ did not show the activity in the striatum, which was present in NC during antisaccade (inhibitory activity) (continued opposite )

19 Structural and functional imaging in schizophrenia Niznikiewicz et al 141 Table 4 (continued ) Participants Reference (n) Language studies Mean age (years) Sex [male/female (n/n)] Task design Major findings Sommer [89 ] SZ 12 NC 12 Kircher [90] SZ 6 NC 6 Kircher [91 ] SZ 6 NC 6 Lawrie [44] SZ 8 NC 10 Bentaleb [92] SZ 1 NC 1 SZ 270 NC 280 SZ 343 NC 340 SZ 343 NC 340 SZ 286 NC 264 SZ 36 NC 36 All male All male All male SZ 3/5 NC 5/5 Female case study Shegrill [93] SZ 1 SZ 36 Male single case study Verb generation and semantic decision reverse read task Block design; whole brain coverage Speech generation task Seven Rorschach cards were presented on the screen during scanning and subjects were asked to describe them Event-related design; whole brain coverage Speech generation task Seven Rorschach cards were presented on the screen during scanning and subjects were asked to describe them Event-related design; whole brain coverage Sentence completion task Block design; whole brain coverage Patient with auditory hallucinations scanned during hallucinations and while listening to external speech Block design; whole brain coverage Single patient imaged during somatic and auditory verbal hallucinations Event-related-like design; whole brain coverage Reduced language-laterality activation in the SZ group was related to increased activation in the right hemisphere rather than to decreased activity in the left hemisphere In SZ, severity of thought disorder was correlated positively with activity in the cerebellar vermis, right body of caudate, and right precentral gyrus Severity of thought disorder was negatively correlated with left superior temporal gyrus, and the posterior part of middle temporal gyrus NC activated more the left, whereas SZ activated more the right middle temporal gyrus Bilateral activation of DLPFC and left middle/superior temporal gyrus was observed in both groups Functional frontotemporal connectivity was diminished in SZ Auditory verbal hallucinations were associated with increased activity in the left primary auditory cortex and the right middle temporal gyrus Somatic hallucinations were associated with activation in the primary somatosensory and posterior parietal cortex, and auditory hallucinations were associated with middle and superior temporal cortex activation Motor function studies Kodama [74] SZ 9 NC 10 Muller [75] SZ(HA) 10 SZ(OL) 10 SZ(UN) 10 NC 10 SZ 232 NC 249 SZ(HA) 320 SZ(OL) 302 SZ(UN) 327 NC 328 SZ 6/3 NC 6/4 SZ(HA) 10 SZ(OL) 10 SZ(UN) 7/3 NC 10 Finger tapping task Block design Slices covering prefrontal cortex (axial slices above the corpus callosum) Finger tapping task Block design; whole brain coverage Less activation was found in the premotor areas in SZ; after training, the activation level increased in the left premotor area in SZ and decreased in NC Right premotor cortex, putamen, and left cerebellum had higher activation in SZ(UN) than in NC SZ(UN) exhibited higher level of activation than other treated SZ patients across all regions studied Face and emotion processing studies Kosaka [76] SZ 12 NC 12 Quintana [77] SZ 8 NC 8 SZ 260 NC 244 SZ 3522 NC 2925 SZ 6/6 NC 6/6 SZ 6/2 NC 6/2 Facial emotional intensity judgment task and the size of a geometric shape judgment task Block design; coronal slices covering amygdala only Visual-motor task of cue and stimulus matching with either a circle or a facial diagram Block design; axial slices superior to the middle temporal lobe Positive facial emotion identification activated the amygdala bilaterally in both groups, but greater activation of the right amygdala was found in SZ relative to NC Negative face emotion identification did not produce group differences, but bilateral amygdala activation was found in SZ whereas only right amygdala activation was found in NC Motor cortical areas involved in the representation and execution of facial movement were activated in SZ only during processing of facial expressions (continued overleaf )

20 142 Schizophrenia Table 4 (continued ) Reference Participants (n) Mean age (years) Drug and behavioral therapy efficacy studies Sex [male/female (n/n)] Task design Major findings Wykes [94 ] SZ(CRT) 6 SZ(CT) 6 NC 6 Stephan [95 ] SZ 6 NC 6 SZ(CRT) 350 SZ(CT) 360 NC 360 SZ 258 NC 275 All male SZ 5/1 NC 5/1 Two back, working memory task, a vigilance task Block design; 10 axial slices in the anterior commissure-posterior commissure plane Alternating finger tapping task of the right and left hand Subjects were scanned twice, 3 weeks apart, with SZ being medicated at the 2nd scan Block design; whole brain coverage As a result of cognitive remediation therapy, increased values of FPQ were found in SZ in right inferior frontal gyrus, and bilateral occipital cortex, and decreased values of FPQ were found in NC in the left inferior/middle frontal gyrus, right frontal cortex, and right inferior frontal cortex This study examined CFC as a function of olanzapine treatment with seed voxel correlational analysis Results suggest that there is a normalization of CFC after olanzapine treatment CFC, cerebellum functional connectivity; DLPFC, dorsolateral prefrontal cortex; Exp, experiment; FPQ, fundamental power quotient; LTM, long-term memory; NC, Normal control; SZ, schizophrenia; SZ(CT), SZ receiving cognitive therapy; SZ(CRT), SZ receiving cognitive remedial therapy; SZ(HA), SZ receiving haldol; SZ(OL), SZ receiving olanzapine; SZ(UN), untreated SZ; WM, working memory Two fmri studies conducted by Kirscher and coworkers [90,91 ] focused on thought disorder and its relationship to brain activation fmri was performed while thought disordered and normal control individuals talked about Rorschach cards In the control individuals the amount of words articulated was positively correlated with activation in the left STG, whereas in the patients it was correlated with activation in the right STG Those authors interpreted these results as evidence for abnormal lateralization of areas involved in word retrieval and speech planning and monitoring They proposed that relatively greater activation of the right STG in the patients may directly contribute to loosening of associations, because the right STG is presumably involved in generating broad semantic fields [103,104] whereas the left STG is involved in maintaining narrow semantic fields so that contextually inappropriate meanings are suppressed early Functional correlates of auditory hallucinations The relationship between auditory hallucinations and functional connectivity between frontal and temporal structures was examined in an fmri study conducted by Lawrie et al [105 ] Both patients and control individuals showed equivalent levels of DLPFC activation in the sentence completion task, but the functional correlation coefficient between activity in left DLPFC and left medial/superior temporal cortex was lower in the patient group, and this correlation connectivity coefficient was lower still in patients with hallucinations (Table 4) This finding of abnormal connectivity between frontal and temporal areas is in agreement with the findings of a study conducted by Ford et al [106] A direct approach of mapping auditory hallucinations onto brain regions was taken by Bentaleb et al [92] in a single case study In that study a female patient with schizophrenia was scanned while experiencing auditory verbal hallucinations, and while listening to external speech that eliminated the subjective experience of auditory hallucinations Experiencing auditory hallucinations was associated with increased activity in the primary auditory cortex and medial temporal gyrus Another single case was reported by Shegrill et al [93], who acquired fmri signal during somatosensory and auditory hallucinations Somatosensory hallucinations activated primary somatosensory and parietal cortices, whereas auditory hallucinations activated the middle and superior temporal cortex Both of these studies suggested that experiencing hallucinations in a given modality (auditory or somatosensory) involves activation of sensory areas that are normally involved in processing sensory information in that modality Functional magnetic resonance imaging: a tool to study the efficacy of clinical interventions Wykes et al [94 ] used fmri to study functional brain changes after cognitive remediation therapy aimed at improving information processing during working memory tasks All individuals were scanned twice Decreased activation was found in normal individuals on repeat scans, whereas increased activation of the right inferior frontal gyrus was observed in schizophrenic patients, especially those who underwent cognitive remediation therapy (Table 4) Because no other measures such as medication or symptom severity changed across the two scanning sessions, the authors suggested that these

21 Structural and functional imaging in schizophrenia Niznikiewicz et al 143 results were attributable to practice and remediation therapy benefits An interesting report from Stephan et al [95 ] discussed the role of olanzapine in normalizing the functional connectivity between cerebellar activity and thalamic cortical circuitry during finger tapping in patients with schizophrenia Using seed voxel correlation analysis, which was introduced by Horwitz and coworkers [107,108], the study explored cerebellar functional connectivity, and it was concluded that olanzapine improved cerebellar functional connectivity for the right but not for the left cerebellum (Table 4) Summary Continuing the tradition of more recent reports [109], most studies reported during the past year used activation paradigms that employed experimental designs traditionally used in neuropsychology such as the n-back task, verb generation or finger tapping, along with clinical tools such as Rorschach cards The experiments were designed to identify which brain regions show abnormal activation in tasks probing executive and memory processes, language processes, and motor function The overwhelming number of studies suggests that for any given task there is not a single abnormal brain region that is related to abnormal function, but instead there is a network of brain regions that are affected and that, together, contribute to functional abnormalities in schizophrenia Abnormal connectivity between temporal and frontal brain regions continues to figure prominently in functional abnormalities in schizophrenia Importantly, some studies indicated that patients with schizophrenia not only show a nonoptimal activation in brain regions used by normal control individuals in a given task, but they also might use different brain regions, either as part of compensatory mechanisms or faulty neural connections Finally, as a new development, a handful of studies used fmri as a tool to look at the efficacy of both medication and behavior therapy interventions This latter application of fmri will probably figure more prominently in future studies, particularly in studies testing the efficacy of new medications in targeting cognitive and clinical symptoms in schizophrenia Diffusion tensor imaging DTI is another promising tool that permits further examination of abnormal functional connectivity in schizophrenia, and thus brings us closer to understanding schizophrenia as a disease of disconnections, or disruptions, in important brain areas that work in concert, and not in isolation, in cognition and behavior DTI is a relatively new technique that can be used to visualize and measure the diffusion of water in brain tissue, and it is particularly useful for evaluating white matter abnormalities in the brain Thus far seven DTI studies have been conducted in schizophrenia [ ,116 ] to investigate white matter abnormalities in multiple brain regions, including prefrontal and temporal white matter, corpus callosum, and uncinate fasciculus (for review [10 ]) Figure 4 shows a single coronal slice of a tensor map (see left image; this map is primarily used for visualization purposes) and an anisotropy map (see right image; this map is used for diffusion quantification), both of which were derived from DTI data Figure 5 shows the newest application of DTI, namely in-vivo white matter fiber tractography, a method that will render possible both dissection and quantification of major white matter pathways in the human brain in vivo In a recent study, Kubicki et al [116 ] used DTI to examine the structural integrity of white matter tracks Figure 4 Diffusion tensor imaging Coronal slices (4-mm thick) are shown that were acquired on a 3 Tesla magnet using diffusion tensor imaging These images are from a normal control individual (a) A diffusion tensor map, in which the blue lines represent the in-plane component of the diffusion (ie white matter fiber tracts are traveling parallel to the acquisition plane) The out-of-plane components of diffusion are color coded (from yellow to dark orange), and represent fibers that are perpendicular to the acquisition plane (ie the darker the color, the stronger the out-of-plane diffusion) (b) A fractional anisotropy map a rotationally independent measure that is frequently used for quantifying diffusion (Courtesy of Marek Kubicki) (a) (b)

22 144 Schizophrenia of the uncinate fasciculus, the most prominent fiber track that connects the frontal and temporal lobes Those investigators found a lack of normal left4right asymmetry in a fractional anisotropy measure (a measure of water diffusion within the fibers, and thus an indirect measure of the integrity of the fibers) in patients with schizophrenia as compared with normal control individuals Findings from that study are but another piece of evidence in support of abnormal connectivity patterns between the frontal and temporal lobes in schizophrenia Conclusion The picture emerging from current MRI and fmri research, as well as from the small number of DTI studies thus far conducted, is that schizophrenia is a brain disorder in which multiple areas are affected not in a random manner but along the lines of functional connectivity, with prominent roles played by both frontal and temporal cortical structures, but also with involvement of relevant subcortical brain regions As new experimental designs are informed by advances in cognitive neuroscience, and new analytic techniques optimize information that can be gleaned from fmri designs and from new advances in structural MRI and DTI, the field of schizophrenia will be enriched with a more comprehensive understanding of this devastating disease In the domain of fmri research, advances in analyzing fmri data are important for meaningful interpretation of results Accordingly, future directions should include careful experimental designs that afford a better understanding of the neural connectivity involved in supporting cognitive tasks It is also becoming apparent that medication status of patients is an important factor, both as a potential confounding factor and as a variable of interest with respect to the efficacy of medication on cognitive tasks evaluated by fmri, as well as on possible neuroprotective effects with respect to ameliorating possible changes in brain structure over time as evaluated by structural MRI and DTI Finally, more studies are needed that evaluate relatives of patients with schizophrenia in order to elucidate further Figure 5 In-vivo three-dimensional fiber tractography of the human brain An in-vivo three-dimensional fiber tractography of the human brain is shown, which was created using diffusion tensor imaging data The data used to create this three-dimensional model were acquired on a 3 Tesla magnet (Courtesy of Hae Jong Park)