Schizophrenia is associated with a diverse array of clinical symptoms and cognitive deficits (Addington, Addington & Maticka-Tyndale, Reference Addington, Addington and Maticka-Tyndale1991; Andreasen et al., Reference Andreasen, Nopoulos, O'Leary, Miller, Wassink and Flaum1999; Johnstone, Owens, Frith, & Crow, Reference Johnstone, Owens, Frith and Crow1987). The symptoms include hallucinations, delusions, disorganized thought and speech, and negative symptoms such as avolition, alogia, and flattened affect (Andreasen & Olsen, Reference Andreasen and Olsen1982). In addition to the clinical symptoms, there are often cognitive deficits that span multiple neuropsychological domains (White, Ho, Ward, Leary, & Andreasen, Reference White, Ho, Ward, O'Leary and Andreasen2006). The diversity of these symptoms suggests perturbations that act on global scales of brain function. As the brain operates through an orchestration of intricately tuned neural networks, a disruption in the connectivity within or between these networks is one of the leading hypotheses underlying schizophrenia (Andreasen, Paradiso, & O'Leary, Reference Andreasen, Paradiso and O'Leary1998; Friston, Reference Friston1998; Konrad & Winterer, Reference Konrad and Winterer2008; Weinberger, Berman, Suddath, & Torrey, Reference Weinberger, Berman, Suddath and Torrey1992). Evidence supporting this theory has emerged through post mortem (Benes, Reference Benes1989; Colter et al., Reference Colter, Battal, Crow, Johnstone, Brown and Bruton1987; Davis et al., Reference Davis, Stewart, Friedman, Buchsbaum, Harvey and Hof2003; Heckers, Heinsen, Geiger, & Beckmann, Reference Heckers, Heinsen, Geiger and Beckmann1991), genetic (Hakak et al., Reference Hakak, Walker, Li, Wong, Davis and Buxbaum2001), and neuroimaging studies (Agartz, Andersson, & Skare, Reference Agartz, Andersson and Skare2001; Ardekani, Neirenberg, Hoptman, Javitt, & Lim, Reference Ardekani, Nierenberg, Hoptman, Javitt and Lim2003; Buchsbaum et al., Reference Buchsbaum, Tang, Peled, Gudbjartsson, Lu and Hazlett1998; Kendi et al., Reference Kendi, Kendi, Lehericy, Ducros, Lim and Ugurbil2008; Kubicki et al., Reference Kubicki, McCarley, Westin, Park, Maier and Kikinis2007; Kumra et al., Reference Kumra, Ashtari, McMeniman, Vogel, Augustin and Becker2004, Reference Kumra, Ashtari, Cervellione, Henderson, Kester and Roofeh2005; Lim et al., Reference Lim, Hedehus, Moseley, de Crespigny, Sullivan and Pfefferbaum1999; Rose et al., Reference Rose, Chalk, Janke, Strudwick, Windus and Hannah2006; Volkow et al., Reference Volkow, Wolf, Brodie, Cancro, Overall and Rhoades1988; White et al., Reference White, Kendi, Lehericy, Kendi, Karatekin and Guimaraes2007).

Theories of aberrant connectivity in distributed neural networks in schizophrenia have paralleled, and perhaps even driven, the exponential growth in techniques applied to measure brain connectivity. Measures of neural connectivity are divided into structural, functional, and effective connectivity, each which evaluates a different aspect of connectivity, although these aspects are interrelated. Structural connectivity refers to measures of the physical presence and pathways of the actual neural fibers. It is commonly established in animal models through anatomical tract tracing techniques (Köbbert et al., Reference Köbbert, Apps, Bechmann, Lanciego, Mey and Thanos2000), or less reliably in noninvasive human studies utilizing techniques such as diffusion tensor imaging (DTI; Shimony, Snyder, Conturo, & Corbetta, Reference Shimony, Snyder, Conturo and Corbetta2004). Although the structural connectivity of the human brain is not directly known, because of experimental limitations, more readily available noninvasive measures of functional connectivity, particularly for the “resting state,” have been taken as a proxy for human anatomical connections (Greicius, Supekar, Menon, & Dougherty, Reference Greicius, Supekar, Menon and Dougherty2009). Functional connectivity describes the functional association between brain regions that possess correlated temporal activation patterns (Friston, Reference Friston1994), as demonstrated by various functional imaging techniques. The previous two aspects of brain connectivity are linked through effective connectivity, which reflects the causal influence of the activity in one brain region on another. Thus, although functional connectivity is essentially model free, effective connectivity relies on a model that explains how functional connectivity is mediated by structural connectivity (Friston, Reference Friston1994).

The triad of structural (Konrad & Winterer, Reference Konrad and Winterer2008; Kubicki et al., Reference Kubicki, McCarley, Westin, Park, Maier and Kikinis2007), functional (Andreasen et al., Reference Andreasen, Nopoulos, O'Leary, Miller, Wassink and Flaum1999; Meyer-Lindenberg et al., Reference Meyer-Lindenberg, Poline, Kohn, Holt, Egan and Weinberger2001, Reference Meyer-Lindenberg, Olsen, Kohn, Brown, Egan and Weinberger2005; Micheloyannis et al., Reference Micheloyannis, Pachou, Stam, Breakspear, Bitsios and Vourkas2006), and effective (Schlosser et al., Reference Schlosser, Gesierich, Kaufmann, Vucurevic, Hunsche and Gawehn2003; Winterer, Coppola, Egan, Goldberg, & Weinberger, Reference Winterer, Coppola, Egan, Goldberg and Weinberger2003) connectivity has been shown to be aberrant in patients with schizophrenia. Early positron emission tomography studies demonstrated regional abnormalities (Volkow et al., Reference Volkow, Wolf, Brodie, Cancro, Overall and Rhoades1988), suggesting that the impairment is focused on specific neural subnetworks, such as hippocampal–prefrontal and cortical–cerebellar–thalamic–cortical connectivity (Andreasen et al., Reference Andreasen, Nopoulos, O'Leary, Miller, Wassink and Flaum1999). Aberrant connectivity in schizophrenia has been shown to be associated with the level of clinical (Lawrie et al., Reference Lawrie, Buechel, Whalley, Frith, Friston and Johnstone2002) and cognitive impairments (Meyer-Lindenberg et al., Reference Meyer-Lindenberg, Poline, Kohn, Holt, Egan and Weinberger2001), and can be altered with the administration of antipsychotic medication (Stephan et al., Reference Stephan, Magnotta, White, Arndt, Flaum and O'Leary2001).

It has recently has been argued that structural connectivity also contributes to developmental processes that influence the shape of the folded cerebral cortex through axonal tension forces (Van Essen, Reference Van Essen1997). The overall interaction of these forces produces the characteristic convolutions of the human brain and leads to a compact layout that optimizes the transit of neuronal signals between brain regions (Hilgetag & Barbas, Reference Hilgetag and Barbas2005; Rakic, Reference Rakic1988; Van Essen, Reference Van Essen1997; Zilles, Armstrong, Schleicher, & Kretschmann, Reference Zilles, Armstrong, Schleicher and Kretschmann1988). Thus, the characteristic morphology of the brain, with its many gyri and sulci, may correspond to its efficient function (see Figure 1). Conversely, if form corresponds to function, and neural function is disrupted in schizophrenia, then is form disrupted as well? Is it possible that buried within the fissures and folds that make up the convoluted landscape of the cerebral cortex exists structural evidence of the disrupted connectivity reported in schizophrenia?

Figure 1.

The trajectories and densities of projections support the axonal tension hypothesis of cortical folding. (a) Prediction of the axonal tension hypothesis: denser projections should be straighter in the adult brain because of the global balancing of tension forces; straight (st, thick line), intermediate (im, short dashed line), and curved (cu, long dashed line) projections link a tracer injection site (dark spot) with different areas. (b) An example of the densities and trajectories of straight, intermediate, and curved projections as labeled by neuronal tract tracing in the primate prefrontal cortex. Denser projections form straighter paths. (c) The average summed density of straight (st), intermediate (im), and curved (cu) projections linking cortical areas, demonstrating that straight connections dominate in the global tug of war. Adapted from “Role of Mechanical Factors in the Morphology of the Primate Cerebral Cortex,” by C. C. Hilgetag and H. Barbas, 2006, PLoS Computational Biology, 2(3), e22. Copyright 2006 International Society for Computational Biology. Adapted with permission. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

The goal of this paper is to provide a review of recent evidence for disruptions in the surface morphology or gyrification of the brain in schizophrenia. We start by reviewing the development and theories of the underlying mechanisms associated with gyrification. We then provide a review of studies of gyrification abnormalities in schizophrenia. Finally, we present evidence that abnormalities of gyrification may reflect abnormalities in the underlying neural connectivity. Because gyrification occurs during a relatively short window during prenatal development, specific changes seen in patients with schizophrenia may help elucidate the interplay between neurodevelopment and illness related changes. Recent theories of gyrification provide an important link for understanding the relationship between brain surface morphology and connectivity.

Theories of Gyrification

By the end of the second trimester (26 weeks gestational age) the developing brain is lissencephalic, with a generally smooth cortical surface (Welker, Reference Welker, Jones and Peters1990). Following the 26th week the brain surface morphology changes rapidly, and by term at 40 weeks the characteristic adult patterns of sulci and gyri are present. Thus, the major steps in the process of gyrification occur during the third trimester, which is a time of considerable brain growth (Neal, Takahashi, Silva, Tiao, Walsh, & Sheen, Reference Neal, Takahashi, Silva, Tiao, Walsh and Sheen2007). Genetic information forms the basis of the convolution patterns of the brain (Piao et al., Reference Piao, Hill, Bodell, Chang, Basel-Vanagaite and Straussberg2004). However, because even monozygotic twins show considerable differences in their folding patterns (Thompson, Cannon, et al., Reference Thompson, Cannon, Narr, van Erp, Poutanen and Huttunen2001; White, Andreasen, & Nopoulos, Reference White, Andreasen and Nopoulos2002), the characteristic convolutions are not specified directly from genetic information; they are probably specified via biological and physical self-organization, that is, through a multitude of interactions involving biological processes, such as regulatory metabolic cascades or anabolic growth, and physical forces, such as volume competition or biomechanical tissue tension (Hilgetag & Barbas, Reference Hilgetag and Barbas2005, Reference Hilgetag and Barbas2006; Richman, Stewart, Hutchinson, & Caviness, Reference Richman, Stewart, Hutchinson and Caviness1975; Toro & Burnod, Reference Toro and Burnod2005; Van Essen, Reference Van Essen1997; Welker, Reference Welker, Jones and Peters1990).

Early theories on brain gyrification hypothesized that differential growth rates in different brain regions could lead to competing volume demands and tension between these regions, resulting in local tissue deformation (His, Reference His1874). One school of morphogenetic concepts of the brain suggests that regions that are developing into gyri may grow at a faster rate than those destined to be become sulci (Kriegstein, Noctor, & Martinez-Cerdeno, Reference Kriegstein, Noctor and Martinez-Cerdeno2006). Thus, convolutions may be established by active, localized growth. This active growth could also occur in combination with friction between the different brain layers against each other, coupled with the friction with the surrounding skull (Le Gross Clark, Reference Le Gross Clark, LeGross and Medawar1945). This differential growth expansion of the individual cortical layers may lead to a buckling of the laminar cortical sheet (Hilgetag & Barbas, Reference Hilgetag and Barbas2005, Reference Hilgetag and Barbas2006; Richman et al., Reference Richman, Stewart, Hutchinson and Caviness1975; Toro & Burnod, Reference Toro and Burnod2005; Van Essen, Reference Van Essen1997; Welker, Reference Welker, Jones and Peters1990).

Theories of cortical morphogenesis should also consider the relationship between the development of brain convolutions and the pruning of neuronal and synaptic connections (Goldman-Rakic, Reference Goldman-Rakic1980; Goldman-Rakic & Rakic, Reference Goldman-Rakic, Rakic, Geschwind and Galaburda1984). It has been suggested that viscoelastic tension exerted by corticocortical fibers (Franze, Reichenbach, & Käs, Reference Franze, Reichenbach, Käs, Kamkin and Kiseleva2008) may contribute to shaping cortical convolutions (Van Essen, Reference Van Essen1997, Reference Van Essen2007). Specifically, this theory proposes that neuronal connectivity that develops with neuronal migration during the second trimester produces fiber tension that draws densely interconnected regions closer together, forming outward bulging gyri between them, whereas more sparsely connected regions drift apart and are separated by inward sulci. The tension, although very small for an individual axon (Heidemann, Lamoureux, & Buxbaum, Reference Heidemann, Lamoureux and Buxbaum1995), is summed by the very large population of neurons, thus creating differential forces that interact globally with the rapidly growing brain to form the gyri and sulci. Within this concept, the characteristic pattern of the convolutions is explained by the highly specific organization of the underlying connectivity (Sporns, Chialvo, Kaiser, & Hilgetag, Reference Sporns, Chialvo, Kaiser and Hilgetag2004).

If tension produced by neuronal connections is involved in the mechanisms of gyrification, then changes in the patterns of the gyri and sulci are an expected outcome of changes (intrinsic or experimental) that alter the connectivity between brain regions. Local and remote changes of gyrification have been observed after experimental white-matter lesions in the developing primate brain (Goldman-Rakic, Reference Goldman-Rakic1980; Goldman-Rakic & Rakic, Reference Goldman-Rakic, Rakic, Geschwind and Galaburda1984). The axonal tension concept has also been supported by recent experimental findings in the primate brain showing a strong correlation between the densities and trajectories of fiber projections, with denser fibers being straighter and most projections following straight or only mildly curved trajectories (Hilgetag & Barbas, Reference Hilgetag and Barbas2006; Toro & Burnod, Reference Toro and Burnod2005).

In summary, starting from genetically specified parameters, such as differential growth rates, various physical processes may interact to produce and organize the surface morphology of the cerebral cortex during brain development. In particular, experimental findings support the concept of a relationship between characteristic cortical convolutions and the specific layout of cortical fiber connections.

Measures of Gyrification

There have been a number of different techniques to quantify brain gyrification, and different approaches are used to quantify different aspects of gyrification. For example, sulcal depth may be considered as one characteristic feature; because sulci can branch in complex patterns, greater sulcal depth does not necessarily translate into a quantitative increase in other gyrification measures. Other features are sulcal frequency (Narr et al., Reference Narr, Thompson, Sharma, Moussai, Zoumalan, Rayman and Toga2001) or cortical complexity, which typically relates an estimate of brain surface area to volume, accounting for the two- (2-D) and three-dimensional (3-D) differences between surface area and volume measurements.

Zilles et al. (Reference Zilles, Armstrong, Schleicher and Kretschmann1988) utilized coronal sections of postmortem brains to measure the ratio of the length of the outer folded surface of the brain (including sulci) to the length of the outer surface excluding sulci. This measure, known as the gyrification index (GI), has been applied to study both the phylogeny (Armstrong et al., Reference Armstrong, Curtis, Buxhoeveden, Fregoe, Zilles and Casanova1991; Zilles et al., Reference Zilles, Armstrong, Schleicher and Kretschmann1988) and ontogeny (Armstrong, Schleicher, Omran, Curtis, & Zilles, Reference Armstrong, Schleicher, Omran, Curtis and Zilles1995) of cortical gyrification (see Figure 2). Brains that have a higher degree of cortical folding yield larger values of the GI. Anterior to posterior maps of human GI measures have shown greater gyrification in the prefrontal and temporal/parietal association regions of the brain (Zilles et al., Reference Zilles, Armstrong, Schleicher and Kretschmann1988). During development, the GI begins to increase prior to the third trimester and plateaus at birth. It is interesting that this plateau appears to remain constant across the life span, despite the rapid brain development and changes associated with aging (Armstrong et al., Reference Armstrong, Schleicher, Omran, Curtis and Zilles1995). However, because developmental studies typically require large samples, and because postmortem samples in younger individuals are difficult to obtain, it would be beneficial for the study of Armstrong et al. (Reference Armstrong, Schleicher, Omran, Curtis and Zilles1995) to be replicated with a larger sample size to confirm this finding. Age-related differences in gyrification measures using magnetic resonance imaging (MRI) techniques have been demonstrated (Magnotta et al., Reference Magnotta, Andreasen, Schultz, Harris, Cizadlo and Heckel1999; White, Su, Schmidt, Kao, & Sapiro et al., 2010), although factors such as cause of death, differences in the interval between fixation of the brains, brain shrinkage, or the generation of brain slices may complicate direct comparisons between MRI and postmortem GI measures.

Figure 2.

The gyrification index (GI) is computed by dividing the length of the inner cortical contour (purple line online only) by the length of the outer contour (blue line online only). This yields a unitless measure that reflects a quantification of gyrification of the brain. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

Alternative methods for measuring the surface morphology of the brain involve regional measurements of curvature (i.e., convexity and concavity) of the cortical surface from MR images (Koenderink & van Doorn, Reference Koenderink and van Doorn1992; Luders et al., Reference Luders, Thompson, Narr, Toga, Jancke and Gaser2006; Magnotta et al., Reference Magnotta, Andreasen, Schultz, Harris, Cizadlo and Heckel1999). These methods first define a triangular isosurface covering a layer within the cortical rim. Once the isosurface is constructed, the angles between vectors normal to the triangular isosurfaces are used to calculate the regional curvature of the cortex. These regional measures can be utilized to study regional curvature between groups (Luders et al., Reference Luders, Narr, Thompson, Rex, Jancke and Steinmetz2004, Reference Luders, Thompson, Narr, Toga, Jancke and Gaser2006) or averaged to study global changes (Magnotta et al., Reference Magnotta, Andreasen, Schultz, Harris, Cizadlo and Heckel1999; Nopoulos, Flaum, O'Leary, Andreasen, Reference Nopoulos, Flaum, O'Leary and Andreasen2000; White et al., Reference White, Andreasen and Nopoulos2002).

Gyrification Abnormalities in Schizophrenia

The ability to quantify various aspects of brain surface complexity has existed for decades (Thompson, Schwartz, Lin, Khan, & Toga, Reference Thompson, Schwartz, Lin, Khan and Toga1996; Zilles et al., Reference Zilles, Armstrong, Schleicher and Kretschmann1988), but there have been relatively few studies that applied these techniques to gyrification in schizophrenia (Table 1). The majority of these studies have employed the traditional 2-D GI measure to evaluate differences between patients and controls. These studies yielded mixed results showing GIs in patients that are higher (Harris, Yates, et al., Reference Harris, Yates, Miller, Best, Johnstone and Lawrie2004; Vogeley et al., Reference Vogeley, Schneider-Axmann, Pfeiffer, Tepest, Bayer and Bogerts2000, Reference Vogeley, Tepest, Pfeiffer, Schneider-Axmann, Maier and Honer2001), lower (Bonnici et al., Reference Bonnici, William, Moorhead, Stanfield, Harris and Owens2007; Kulynych, Luevano, Jones, & Weinberger, Reference Kulynych, Luevano, Jones and Weinberger1997; McIntosh et al., Reference McIntosh, Moorhead, McKirdy, Hall, Sussmann and Stanfield2009; Sallet et al., Reference Sallet, Elkis, Alves, Oliveira, Sassi and Campi de Castro2003; Wheeler & Harper, Reference Wheeler and Harper2007), or no different than controls (Highley et al., Reference Highley, DeLisi, Roberts, Webb, Relja and Razi2003).

Table 1.

Studies examining gyrification in schizophrenia spectrum disorders

Note: M, male; F, female; Schiz, chronic schizophrenia; GI, gyrification index; MRI, magnetic resonance imaging; NR, not reported; EOS, early-onset schizophrenia (children and adolescents with schizophrenia; FES, first-episode schizophrenia; HR-C, high risk who converted to schizophrenia; HR-NC, high risk who have not converted; HR, high risk; SIS, Structured Scale for Schizotypy (+ is above threshold/high score and – is below threshold/low score); MR, mental retardation; Dx, diagnosis; BPD, bipolar disorder.

The earliest study utilized coronal MRI slices from the left hemisphere to calculate the GI and reported an overall reduction in GI in both anterior and posterior regions in nine patients with chronic schizophrenia (Kulynych et al., Reference Kulynych, Luevano, Jones and Weinberger1997). Subsequent studies using MRI-based GI measures have replicated the finding of reduced GI in the frontal cortex (Bonnici et al., Reference Bonnici, William, Moorhead, Stanfield, Harris and Owens2007) and global decreases in the GI (Sallet et al., Reference Sallet, Elkis, Alves, Oliveira, Sassi and Campi de Castro2003). Using 50-μm coronal slices in a postmortem study of posterior cingulate cortex, Wheeler and Harper (Reference Wheeler and Harper2007) reported lower GI in the rostral cingulate, with a trend for a reduced GI in the medial and caudal regions. These studies demonstrating a decrease in the GI were all performed on patients with chronic schizophrenia.

There have been two postmortem studies of gyrification in patients with schizophrenia. The first study found an increase in frontal GI in a group of patients with chronic schizophrenia (Vogeley et al., Reference Vogeley, Schneider-Axmann, Pfeiffer, Tepest, Bayer and Bogerts2000). The second postmortem study focused only on the posterior cingulate and found a reduction in GI. Because typical whole-brain 2-D measures of the GI follow the entire contour of the brain, this finding of a reduced GI in the cingulate is difficult to compare to the more global GI measures.

The report of an increased GI in frontal regions, as shown in one of the postmortem studies, has also been reported in MRI studies of the frontal cortex in patients with chronic (Falkai et al., Reference Falkai, Honer, Kamer, Dustert, Vogeley and Schneider-Axmann2007; Vogeley et al., Reference Vogeley, Tepest, Pfeiffer, Schneider-Axmann, Maier and Honer2001) and first-episode schizophrenia (Harris, Yates, et al., Reference Harris, Yates, Miller, Best, Johnstone and Lawrie2004). Moreover, increased GI has been identified in temporal lobe structures (Harris, Yates, et al., Reference Harris, Yates, Miller, Best, Johnstone and Lawrie2004), but not occipitoparietal regions (Falkai et al., Reference Falkai, Honer, Kamer, Dustert, Vogeley and Schneider-Axmann2007; Harris, Yates, et al., Reference Harris, Yates, Miller, Best, Johnstone and Lawrie2004). However, findings of lower GI (Bonnici et al., Reference Bonnici, William, Moorhead, Stanfield, Harris and Owens2007; McIntosh et al., Reference McIntosh, Moorhead, McKirdy, Hall, Sussmann and Stanfield2009) have also been reported and include studies using the same techniques that reported an increase in GI (Harris et al., Reference Harris, Moorhead, Miller, McIntosh, Bonnici and Owens2007; Harris, Yates, et al. Reference Harris, Yates, Miller, Best, Johnstone and Lawrie2004). These apparent discrepant findings may point to the heterogeneity of gyrification or to the heterogeneity in the populations and the methodologies utilized to study GI.

Several approaches have studied whether abnormal surface folding is an endophenotypic marker (Gottesman & Gould, Reference Gottesman and Gould2003) for schizophrenia (Falkai et al., Reference Falkai, Honer, Kamer, Dustert, Vogeley and Schneider-Axmann2007; Vogeley et al., Reference Vogeley, Tepest, Pfeiffer, Schneider-Axmann, Maier and Honer2001). Evidence from studies of unaffected siblings and individuals at risk for developing schizophrenia support gyrification abnormalities a s being more severe in individuals who go on to develop the illness (Harris, Whalley, et al., Reference Harris, Whalley, Yates, Miller, Johnstone and Lawrie2004; Vogeley et al., Reference Vogeley, Tepest, Pfeiffer, Schneider-Axmann, Maier and Honer2001). Thus, it is possible that abnormalities of gyrification, which may surface early in cortical development (Armstrong et al., Reference Armstrong, Schleicher, Omran, Curtis and Zilles1995), are trait markers of the illness, which differentiate those at increased risk who eventually develop the illness (Harris et al., Reference Harris, Moorhead, Miller, McIntosh, Bonnici and Owens2007; Harris, Whalley, et al. Reference Harris, Whalley, Yates, Miller, Johnstone and Lawrie2004). An increase in the right frontal gyrification in individuals at a high risk for developing schizophrenia has been shown to be predictive for developing schizophrenia (Harris et al., Reference Harris, Moorhead, Miller, McIntosh, Bonnici and Owens2007; Harris, Whalley, et al., Reference Harris, Whalley, Yates, Miller, Johnstone and Lawrie2004; Stanfield et al., 2007).

Several studies have deviated from the traditional GI approach to measurements of gyrification and have utilized either 3-D GI measures of cortical surface complexity (Cachia et al., Reference Cachia, Paillere-Martinot, Galinowski, Januel, de Beaurepaire and Bellivier2008; White, Andreasen, Nopoulos, & Magnotta, Reference White, Andreasen, Nopoulos and Magnotta2003; Wiegand et al., Reference Wiegand, Warfield, Levitt, Hirayasu, Salisbury and Heckers2005) or 3-D curvature indices of the gryi and sulci (White et al., Reference White, Andreasen, Nopoulos and Magnotta2003). Two studies found no difference in surface complexity, one using a 3-D adaptation of the GI (Wiegand et al., Reference Wiegand, Warfield, Levitt, Hirayasu, Salisbury and Heckers2005) and the other using a comparison of surface area to volume, corrected to the two-thirds power (White et al., Reference White, Andreasen and Nopoulos2002). However, surface area measures are highly variable and patient–control differences may be more difficult to assess. Using a 3-D adaptation of the GI, Cachia et al. (Reference Cachia, Paillere-Martinot, Galinowski, Januel, de Beaurepaire and Bellivier2008) found that patients with chronic schizophrenia had lower sulcal complexity that was more pronounced in the temporal and left frontal lobes.

Direct comparisons between the methodologies that utilized surface curvature and those that implemented the GI are difficult, as the GI does not differentiate between gyral and sulcal regions. Thus, it is possible that sulcal widening may be cancelled by gyral or global brain changes (Kulynych et al., Reference Kulynych, Luevano, Jones and Weinberger1997).This is highlighted in children and adolescents with schizophrenia showing opposing changes in gyral and sulcal curvature in the frontal and temporal brain regions (White et al., Reference White, Andreasen, Nopoulos and Magnotta2003). These findings have been replicated and have been shown to be more pronounced in a younger sample of children and adolescent with schizophrenia (R. K. Lenroot, personal communication, 2006). In this sample, the sulci developed less curvature (i.e., became broader) but the gyri became more peaked, demonstrating an increased curvature (Figure 3). This alteration in morphology results in increased sulcal width, which has also been reported in adolescents and adults with schizophrenia (Allmann, Reference Allmann2000; Andreasen et al., Reference Andreasen, Flashman, Flaum, Arndt, Swayze V and O'Leary1994; Giedd et al., Reference Giedd, Jeffries, Blumenthal, Castellanos, Vaituzis and Fernandez1999; Pfefferbaum et al., Reference Pfefferbaum, Zipursky, Lim, Zatz, Stahl and Jernigan1988; Weinberger, Torrey, Neophytides, & Wyatt, Reference Weinberger, Torrey, Neophytides and Wyatt1979).

Figure 3.

Children and adolescents with schizophrenia demonstrate differences in gyrification that include (a) a widening of the sulci and a more peaked appearance of gyri compared to (b) controls. These measures can be quantified regionally following the identification of (c) whole brain sulci and gyri. Adapted from “Gyrification Abnormalities in Childhood- and Adolescent-Onset Schizophrenia,” by T. White, N. C. Andreasen, P. Nopoulos, and V. Magnotta, 2003, Biological Psychiatry, 54, 418–426. Copyright 2003 by Elsevier. Adapted with permission. [A color version of this figure can be viewed online at journals.cambridge.org/dpp]

Neural Connectivity in Schizophrenia

New approaches and insights from the field of complex network studies (Strogatz, Reference Strogatz2001) are currently entering systems neuroscience. Applied to the global organization of neural networks, such approaches have revealed high modularity, complexity, and efficiency of the different types of anatomical and functional neural connectivity (Sporns et al., Reference Sporns, Chialvo, Kaiser and Hilgetag2004). Efficient connectivity, for instance, can be assessed by the average length of all shortest paths (ASP) across a network, where the shortest paths represent the smallest number of edges required to link two nodes. The ASP can be taken as an estimate for the minimal number of processing steps in a neural system, and it appears that neural networks are organized to maintain very short ASP (Kaiser & Hilgetag, Reference Kaiser and Hilgetag2006). Conversely, network alterations have been observed in schizophrenia, which lead to increased ASP and which may be associated with a loss of global network efficiency and functional impairments (see Figure 1; Liu et al., Reference Liu, Liang, Zhou, He, Hao and Song2008; Micheloyannis et al., Reference Micheloyannis, Pachou, Stam, Breakspear, Bitsios and Vourkas2006).

Network anomalies in schizophrenia can be identified in measures of functional of as well as structural connectivity (Karlsgodt, Sun, et al., Reference Karlsgodt, Sun, Jimenez, Lutkenhoff, Willhite and van Erp2008); however, a large number of recent studies have focused on white-matter integrity and structural connectivity as indicated by the noninvasive approach of DTI. These studies demonstrated mostly reductions of indicators of structural connectivity, such as fractional anisotropy, in particular for frontal and temporal white matter (Kyriakopoulos, Bargiotas, Barker, & Frangou, Reference Kyriakopoulos, Bargiotas, Barker and Frangou2008; White, Nelson, & Lim, Reference White, Nelson and Lim2008). The diminished integrity of structural connectivity may result from weaker connections, perhaps also due the loss of oligodendroglia (Konrad & Winterer, Reference Konrad and Winterer2008), or through the disconnection of network nodes and appears to precede the onset of the disease (Begre & Koenig, Reference Begre and Koenig2008). Nonetheless, there are also occasional reports of increases in functional (resting state) connectivity in schizophrenia, for instance, in paranoid schizophrenic subjects (Broyd et al., Reference Broyd, Demanuele, Debener, Helps, James and Sonuga-Barke2009).

The approach of DTI has also indicated reductions in specific structural connection systems, such as the uncinate fasciculus, a fiber tract connecting the temporal pole near the amygdala and the orbitofrontal/ventrolateral prefrontal cortex, or connections of the cingulate gyrus (Karlsgodt, Sun, et al., Reference Karlsgodt, Sun, Jimenez, Lutkenhoff, Willhite and van Erp2008). Such weakened connections can be associated with specific functional aspects of schizophrenia. For example, the estimated strength of the left superior longitudinal fasciculus, which is decreased in schizophrenic patients, is correlated with performance in a verbal working memory task (Karlsgodt, van Erp, et al., Reference Karlsgodt, van Erp, Poldrack, Bearden, Nuechterlein and Cannon2008). However, the exact relationship between variations in projection systems and global or local aberrations of gyrification in schizophrenia is currently still unclear for two reasons. First, the results of the DTI investigations are themselves not consistent (Kyriakopoulos et al., Reference Kyriakopoulos, Bargiotas, Barker and Frangou2008; White et al., Reference White, Nelson and Lim2008). Investigated fiber systems varied across studies, and several studies investigating the same systems disagreed on whether they showed consistent changes in schizophrenic patients compared to healthy subjects (Kanaan et al., Reference Kanaan, Kim, Kaufmann, Pearlson, Barker and McGuire2005). Moreover, DTI findings, such as that of a weakened uncinate fasciculus in schizophrenic patients (Burns et al., Reference Burns, Job, Bastin, Whalley, Macgillivray and Johnstone2003), may be contradicted by corresponding studies of postmortem material (Highley, Walker, Esiri, Crow, & Harrison, Reference Highley, Walker, Esiri, Crow and Harrison2002). Second, changes of structural connectivity as well as gyrification in schizophrenia appear to be predominantly associated with the frontal lobes. However, although changes in connectivity mostly manifest as weakened projections, schizophrenia may go along with reduced as well as increased gyrification of the frontal gray matter (cf. Table 2). These issues still need to be investigated more systematically. For example, structural MR studies showed that the gray to white matter ratio in frontal and temporal lobes does not decrease over the life span in schizophrenic patients as it does in healthy subjects (Bartzokis et al., Reference Bartzokis, Nuechterlein, Lu, Gitlin, Rogers and Mintz2003).

Table 2.

Consolidated results of gyrification studies of schizophrenia and individuals at high risk

Note: GI, gyrification index. No change in the GI was noted by Highley et al. (Reference Highley, DeLisi, Roberts, Webb, Relja and Razi2003).