Turner syndrome (TS) is a neurogenetic syndrome seen in approximately 1/2000 live female births (Stochholm et al., 2006
). Turner syndrome arises due to partial or complete absence of one X chromosome, with the most common karyotype being X-monosomy [45,XO]. Since, in karyotypically normal [46,XX] females, almost all of one of the X chromosomes in each cell is randomly “silenced” through an epigenetic process known as Lyonization (Lyon, 1961
) the TS phenotype may be due to haplo-insufficiency of those X chromosome genes that escape inactivation (Zinn and Ross, 1998
). Effective X-linked gene haploinsufficiency in TS is also influenced by the “parent-of-origin” of the intact X chromosome, as the expression of some X chromosome genes differs systematically depending on which parent the chromosome is inherited from through a process known as imprinting (Barlow, 1995
). X-chromosome haploinsufficiency in TS is also associated with ovarian failure and related estrogen and androgen deficiency, which may in themselves contribute to the TS phenotype.
Compared to the TS physical phenotype (Ogata and Matsuo, 1995
), the behavioral-cognitive and neuroanatomical phenotypes remain less well described. A better understanding of how brain and behavior are altered in TS could shed light on how X-linked genetic and epigenetic influence human neurodevelopment.
Females with TS typically show an uneven cognitive profile with pronounced deficits in visuo-spatial skills alongside verbal abilities within the normal range. Deficits are best documented for tasks involving mental rotation, visual construction, and number manipulation. There are also reports of deficits in other domains such as memory and face processing (Murphy et al., 1994
; Ross et al., 2000
). Multiple functional magnetic resonance imaging (fMRI) studies have compared brain activity in females with TS to that in controls during the execution of mental rotation and visual construction tasks, and consistently report differences within parieto-frontal systems (especially within the intra-parietal sulcus) (Hart et al., 2006
; Kesler et al., 2004
; Kesler et al., 2006
; Molko et al., 2003
). The parietal lobe has also been repeatedly implicated in TS by structural magnetic resonance imaging (sMRI), with the most consistent findings being reduced parietal lobe volumes in TS compared to controls (Brown et al., 2002
; Cutter et al., 2006
; Murphy et al., 1993
; Reiss et al., 1995
). It remains unclear however if this reduction relates to white matter volume, grey matter volume, or both. summarizes all TS sMRI studies published to date.
Summary of all structural magnetic resonance imaging studies comparing cortical anatomy between turner syndrome and typically developing controls, that were available at the time of publication.
Compared to the findings from sMRI studies of lobar volumes in TS, there is much less consistency in the findings of sMRI studies that examined brain anatomy at finer spatial resolution using hand-tracing of regions of interest (Rae et al., 2004
), or automated and spatially non-biased voxel-based morphometry (VBM) (Cutter et al., 2006
; Good et al., 2003
; Molko et al., 2004
). This is especially true for studies of the cerebral cortex. For example, CV in the left parieto-occipital junction, right primary motor and superior temporal cortex of TS individuals have been reported as both increased (Molko et al., 2003
; Rae et al., 2004
) and decreased (Cutter et al., 2006
) as compared to controls. Such contrasting findings may be due to true neurostructural heterogeneity in TS, or differences in the clinical and controls samples used across studies with respect to demographic, genetic or endocrine features. Alternatively, the fact that most conflicting findings occur in those cortical regions where people with TS also have significant differences from controls in sulcal morphology (Molko et al., 2003
; Molko et al., 2004
), suggests that group differences in cortical shape may be confounding group differences in regional CV. However, cortical shape has not been directly modeled in any of the existent studies of regional CV in TS. To date, such studies have used volume-based registration to align sMRI scans across individuals and then measured CV in a voxel-by-voxel manner (VBM). This approach has three important limitations.
Firstly, compared to surface-based methods, volume-based alignment of scans is not able to take into account the complex and highly variable folding patterns of the cortical sheet. Because these folding patterns are known to vary between females with TS and controls (Molko et al., 2003
), prior studies may not have been comparing like with like. That is, unless an attempt is made to “line-up” surface features of the cortex across scans before comparing CV between groups, then systematic differences in sulcal position between groups could result in one group’s sulcus being compared to another group’s gyrus.
Secondly, VBM studies only compare regional CV. However, CV within a given region of the cortical sheet is itself a product of two lower-order spatial properties -– cortical thickness (CT) and surface area (SA). Unless CV is broken down into CT and SA there is a risk of falsely concluding that a given cortical region is structurally unaltered in TS when in fact CT and SA for that regions may show marked alterations in “opposing” directions (such that their product - CV - does not show group differences). Conversely, group differences in CV may reflect abnormalities of CT only, SA only, or both these properties. It is important to distinguish these possibilities because measures of CT and SA capture very different sets of biological processes as evidenced by their differing evolutionary histories (Rakic, 1995
), developmental trajectories (Armstrong et al., 1995
; Sowell et al., 2007
), and genetic determinants (Panizzoni et al., 2007
). Furthermore, SA is determined by overall brain size and the degreee of cortical folding (gyrification), which are shaped by distinct influences. Decomposing cortical alterations in TS into these distinct components would help to focus investigation of the molecular and developmental pathways that lie between genetic and neurobiological alterations in TS.
Thirdly, there is emerging evidence that compared to a VBM assessment of CV across the cortex, a vertex-based assessment of CT after surface-based registration provides a more sensitive and informative marker of cortical differences between clinical groups (Bermundez P, 2008
; Park HJ, 2009
Furthermore, by inter-relating measures of CT at different vertices, it is possible to describe patterns of structural co-variance across the cortex. The extent to which cortical regions are structurally similar to each other (as indexed by CT co-variance) is neurobiologically meaningful because in typical development (i) cortical regions that show high CT co-variance also show strong shared genetic influences on CT (Schmitt et al., 2008
), and (ii) maps of CT co-variance between cortical regions have been found to strongly resemble maps of cortico-cortical connectivity derived from other techniques such as diffusion-tensor imaging (DTI) of white matter tracts (Lerch et al., 2006
) and fMRI studies of “functional connectivity” (Chen et al., 2008
; Toro et al., 2008
). Maps of CT co-variance also vary as a function of disease processes (He et al., 2008
), developmental stage, cognitive ability and skill proficiency (Bermundez P, 2008
; Lerch et al., 2006
). Therefore, if a comparison of CT-covariance amongst females with TS to CT-covariance amongst typically developing controls identifies regional differences - then such regions would represent strong candidate networks for further investigation of structural and functional dysconnectivity in TS with DTI and fMRI.
In summary, surface-based methods for measuring cortical morphometry have multiple potential advantages. However they have not yet been exploited in the study of the TS neuroanatomical phenotype. Therefore we applied a fully automated and well-validated (Kabani et al., 2001
; Shaw et al., 2008
) method for measuring cortical morphometry (MacDonald et al., 2000
) to process brain sMRI data acquired from 24 females with TS and 19 healthy female controls matched for age and verbal intelligence quotient. We have previously characterized this sample using Voxel-Based and Region of Interest methods (Cutter et al., 2006
). This allows us to qualitatively compare the findings of volume-based vs surface-based approaches to cortical morphometry. In order to better contextualize our findings, we conducted a systematic review of all sMRI studies conducted in TS to date, and summarize the main findings of these studies in .
In this paper, we apply surface-based morphometry to address the following questions. Firstly, is the well-replicated reduction of parietal lobe volume in TS contributed to by reduced parietal lobe CV? Secondly, if lobar CV alterations are present in TS - are these due to alterations in CT, SA or both? Thirdly, is there any evidence for alterations in SA and cortical folding in TS? Fourthly, can a spatially non-biased assessment of CT across the cortical sheet in TS using surface-based registration of scans clarify some of the inconsistencies in VBM findings? Fifthly, is there any evidence of selective alterations in co-coordinated CT development in TS compared to controls as indexed by the proxy measure of disrupted inter-regional correlations in CT?