Abstract

Tuberous sclerosis complex (TSC) is a multisystem syndrome classically associated with the occurrence of focal brain dysplasias. We used structural magnetic resonance imaging to test for neuroradiological abnormalities in TSC (tubers, white matter lesions, and subependymal nodules) and to explore the relationships between these lesions and computational morphometric abnormalities of gray and white matter distribution. We tested memory function in TSC and investigated the relationship between memory function and both morphometric variation and lesion load. Patients demonstrated deficits bilaterally in volume of subcortical gray matter regions including thalamus, basal ganglia, insula, and cerebellum, as well as white matter deficits bilaterally in intrahemispheric tracts. Morphometric deficits could not be explained as local effects of lesions. Patients demonstrated deficits in executive working memory and recall memory, sparing recognition. Structure–function mapping showed long-term and working memory function was positively correlated with gray matter density (in thalamus, caudate nucleus, and frontal cortex) but not with lesion load. The neuroanatomical endophenotype of TSC is more extensive than previously recognized and comprises abnormalities in the distribution of gray and white matter in addition to classical lesions. Normal intelligence quotient patients with TSC show a profile of long-term and working memory impairment that is related to gray matter deficits in thalamus and basal ganglia components of fronto-striatal circuits.

Introduction

Tuberous sclerosis complex (TSC) is a multisystem genetic disorder with an autosomal dominant pattern of inheritance and an incidence of approximately 1:6000 births (Osborne and others 1991). Two causative genes have been identified: TSC1, located on chromosome 9q34 (van Slegtenhorst and others 1997), and TSC2, located on chromosome 16p13.3 (Nellist and others 1993). The disorder has a highly variable clinical phenotype, with a wide range of physical and neuropsychiatric disorders including epilepsy, mental retardation, autism spectrum disorders, and hyperkinetic disorder (Prather and de Vries 2004; de Vries and others 2005). The most severely affected cases can be profoundly handicapped, whereas some individuals are largely unscathed by the disease (Hunt and Dennis 1987). The neuropathology of TSC has classically been described in terms of cortical tubers or hamartomata, white matter abnormalities, and subependymal nodules (SENs) (Scheithauer and Reagan 1999). SENs are located immediately beneath the ependymal epithelium lining the ventricles and may undergo neoplastic transformation to form subependymal giant cell astrocytomas (Scheithauer and Reagan 1999). Cortical tubers are clusters of developmentally abnormal cells that have radially migrated to cortex from the periventricular zone; the track of migration is indicated by the persistence of heterotopic dysplastic neurons in adult white matter (Gomez 1988; Braffman and others 1992; Scheithauer and Reagan 1999).

Most prior neuroimaging studies of TSC have focused on the number and distribution of tubers and nodules; for example, seeking to correlate indices of neuroradiological “lesion load” with epilepsy, mental retardation, autism, or other psychopathological outcomes (Jambaque and others 1991; Shepherd and others 1995; Bolton and Griffiths 1997; Goodman and others 1997; Seri and others 1999; Jambaque and others 2000; Weber and others 2000; Bolton and others 2002; O'Callaghan and others 2004). One such finding is an inverse correlation between neuroradiological “lesion load” (number of tubers or nodules) and global cognitive function (intelligence quotient [IQ]), although findings have been inconsistent; see Ridler and others (2004) for review. Tubers are located predominantly in neocortical regions, especially frontal and parietal cortex, as well as anterior cingulate cortex. Nodules are embedded in ventricle walls and extend into subcortical nuclei, most frequently into the caudate nucleus, putamen, and thalamus (Ridler and others 2004). However, we have previously used computational morphometry to demonstrate structural abnormalities of gray and white matter over and above the classical lesions in patients with TSC (Ridler and others 2001). Specifically, we demonstrated bilaterally symmetric gray matter deficits in thalamus, basal ganglia, and medial temporal lobe, as well as extensive, bilaterally symmetric deficits in white matter constituting major intrahemispheric tracts. An issue arising from this prior study, which is directly addressed by the current work, is the possibility that gray and white matter morphometric abnormalities might simply be attributable to the classical lesions of the disorder.

The cognitive manifestations of this complex and highly variable neuroanatomical phenotype are not yet clearly defined. There are now a few studies seeking to investigate the neuropsychological profile of individuals with TSC; see Prather and de Vries (2004) for a recent review. These authors suggested that attentional-executive deficits are consistent findings in the disorder and proposed that such functional abnormalities were due to aberrations in fronto-striatal circuits (Alexander and others 1986). There have, however, been no systematic investigations of memory functions in TSC nor any attempt to examine the neuroanatomical substrates of specific cognitive deficits.

The present study was designed with three main aims in mind. First, to confirm and replicate the finding of morphometric abnormalities in the brain and to test that these were not simply attributable to tubers, white matter lesions, or SENs; second, to investigate the profile of memory function in normally intelligent adults with and without TSC; third, to explore correlations between memory performance and both brain morphometry and lesion load in search of the putative neuroanatomical substrate of memory deficits in TSC.

Materials and Methods

Subjects

Twenty-five adults with TSC (17 males, 8 females) and 25 healthy comparison subjects (16 males, 9 females) were studied. The groups were well matched for age, sex, handedness, and IQ; see Table 1 and Supplementary Data Table 1 for details. All participants had normal intelligence as assessed by both the National Adult Reading Test (Nelson and Willison 1991) and their age-adjusted scores on the Wechsler Abbreviated Scale of Intelligence (Wechsler 1999). All participants with TSC satisfied standard operationalized diagnostic criteria (Roach and others 1998). No subject suffered from other neurological or psychiatric disorder (except depression, which was present in 2 patients). Patients were interviewed using a structured questionnaire to ascertain the presence or absence of epilepsy and the characteristics of their epilepsy, that is, age at onset, duration and type of seizures, and current medication details (see Table 1 and for more details see Supplementary Data Table 1). At the time of the magnetic resonance imaging (MR), antiepileptic treatment included monotherapy or polytherapy with carbamazepine (patients N = 7), lamotrigine (N = 4), phenytoin (N = 3), sodium valproate (N = 2), levetiracetam (N = 1), olanzapine (N = 1), oxcarbazepine (N = 1), phenobarbitone (N = 1), prednisolone (N = 1), and vigabatrin (N = 1). Note that 9 of the patients had never suffered from seizures. All healthy volunteers satisfied the following inclusion criteria: no personal or family history of neurological or psychiatric disorder, no history of prolonged loss of consciousness, no contraindication to MRI. The study was approved by the Local Research Ethics Committee (Addenbrooke's NHS Trust, UK), and all participants gave informed written consent.

Table 1

Demographic and clinical characteristics of the sample

Characteristic Adults with TSC Healthy volunteers Between-group differences 
Sample size (n25 25  
Gender, male:female 17:8 16:9  
Handedness, right:left 19:6 20:5  
Average age (SD) 39.3 (10.7) 34.3 (11.0) t = −1.620 (df = 48), P = 0.111 
Average WASI IQ (SD) 116.9 (12.5) 116.7 (10.8) t = −0.049 (df = 48), P = 0.962 
Average NART IQ (SD) 114.1 (9.3) 115.4 (5.1) t = 0.606 (df = 48), P = 0.548 
Epilepsy    
    Ever 16/25   
    Mean duration (years) 14.13 (10.9)   
    Age of onset 8.67 (9.5)   
    Seizures currently uncontrolled   
    Seizure free, unmedicated   
    Seizure free, medicated   
    Infantile spasms   
Characteristic Adults with TSC Healthy volunteers Between-group differences 
Sample size (n25 25  
Gender, male:female 17:8 16:9  
Handedness, right:left 19:6 20:5  
Average age (SD) 39.3 (10.7) 34.3 (11.0) t = −1.620 (df = 48), P = 0.111 
Average WASI IQ (SD) 116.9 (12.5) 116.7 (10.8) t = −0.049 (df = 48), P = 0.962 
Average NART IQ (SD) 114.1 (9.3) 115.4 (5.1) t = 0.606 (df = 48), P = 0.548 
Epilepsy    
    Ever 16/25   
    Mean duration (years) 14.13 (10.9)   
    Age of onset 8.67 (9.5)   
    Seizures currently uncontrolled   
    Seizure free, unmedicated   
    Seizure free, medicated   
    Infantile spasms   

Note: WASI IQ, Wechsler Abbreviated Scale of Intelligence (Wechsler 1999) IQ; NART IQ, National Adult Reading Test (Nelson and Willison 1991) IQ; Seizure free, patients who had not suffered from a seizure in the year preceding testing; Seizures currently uncontrolled, all patients who had suffered from a seizure in the last 12 months despite antiepileptic medication.

Structural MRI data acquisition

We used a GE Signa system (General Electric, Milwaukee, WI) operating at 1.5 T at the Magnetic Resonance Imaging and Spectroscopy Unit, Addenbrooke's Hospital, Cambridge, UK. A preliminary localizing scan in the sagittal plane was used to identify anterior and posterior commissures and to prescribe acquisition of a dual-echo, fast spin-echo (FSE) data set in a near-axial plane parallel to the intercommissural line. Whole-brain coverage was provided by 40 contiguous, interleaved, 4-mm-thick proton density–weighted and T2-weighted images acquired with the following parameters: repetition time (TR) = 5625 ms; echo times (TEs) = 20 ms and 102 ms with an 8-echo train length; matrix size = 256 × 256; and field of view = 22 cm, giving an in-plane resolution of 0.859 mm. The total acquisition time was 12 min and 10 s. T2-weighted fluid-attenuated inversion recovery (FLAIR) data were also acquired in the same (near-axial) orientation as the dual-echo FSE data with TR = 10 002 ms, TE = 112.5 ms with 2 excitations, inversion time = 2250 ms, matrix size = 256 x 256, slice thickness = 4 mm. The total acquisition time was 14 min and 40 s.

Structural MRI Data: Segmentation, Normalization and Parcellation

MRI data were segmented and probabilistic maps of gray matter, white matter, and cerebrospinal fluid (CSF) were created for each subject. Tissue classification maps were coregistered with a customized template image in standard stereotactic space (Talairach and Tournoux 1988), using a 12-parameter affine transformation. Both segmentation and normalization were implemented using BAMM software (http://www-bmu.psychiatry.cam.ac.uk/software: Brammer and others 1997; Suckling, Brammer, and others 1999; Suckling, Sigmundsson, and others 1999).

To facilitate later regional parcellation of the gray matter maps, each of these was also coregistered with the Montreal Neurological Institute (MNI) single-subject high-resolution T1-weighted image (MNI; http://www.bic.mni.mcgill.ca) using an identical procedure. The automated anatomical labeling (AAL) regionally parcellated template image (Schmahmann and others 1999; Tzourio-Mazoyer and others 2002), also in the space of the MNI T1 image, was then used to estimate regional mean gray matter densities in each of 116 regions, which were aggregated to give 10 lobar and major subcortical structure measures in each hemisphere for each subject.

MRI Data: Statistical Analysis

Group differences in gray matter, white matter, and CSF volume were explored separately by fitting a linear regression model at each intracerebral voxel. Loci of significant between-group differences were identified by a nonparametric permutation test of suprathreshold voxel clusters, as described in detail elsewhere (Bullmore and others 1999; Suckling and Bullmore 2004). Thresholds for statistical significance were set such that the clusterwise probability of type 1 error P < 0.003 (for gray and white matter) and the expected number of false-positive tests was less than one per map.

MRI Data: Neuroradiological Assessment

To explore the relationship between neuroradiological lesion load, cognitive function, and morphometric abnormalities, we identified tubers and nodules by expert neuroradiological examination of the FLAIR and FSE data. Lesion locations were manually marked on the imaging data in native space, and maps of lesion location were then coregistered in standard space by the same procedures used to coregister the gray matter maps; see Ridler and others (2004) for detail. This procedure allowed us to estimate lesion load both in terms of whole-brain number of tubers and also in terms of lesion density (percent region occupancy) in each of the 10 lobar and major subcortical regions of interest defined by appropriately aggregating sublobar regional volumes defined by the MNI parcellated template image (Schmahmann and others 1999; Tzourio-Mazoyer and others 2002). The regional distribution of lesions in this group of patients and the relationships between the classical lesions are presented in detail elsewhere (Ridler and others 2004).

Neuropsychological tests

All subjects were examined using a battery of neuropsychological tests (for test details see Table 2). Group means and standard deviations (SDs) were estimated for all outcome measures of each test, and between-group differences were assessed by t-tests. In order to provide comprehensive yet parsimonious measurements of cognitive functioning, the dimensionality of the neuropsychological data was reduced by summarizing the neuropsychological battery into cognitive domains; see Table 2. We computed a summary score of performance in each domain for each subject by standardizing all individual test scores (subtracting healthy group mean and dividing by healthy group SD) and averaging them within domain for each subject. This procedure reduced the total number of psychological variables per subject from 37 raw scores to 6 domain-specific Z-scores.

Table 2

Neuropsychological tests summarized under six cognitive domains

1. Long-term memory 
     Rey–Osterrieth Complex Figure —delayed recall (Rey 1941; Osterrieth 1944
     CANTAB nonverbal paired associate learning (level 6)—total errors (Sahakian and others 1988; Fowler and others 1995
     WMS-III Logical Memory—immediate and delayed recall (Weschler 1997
     WMS-III Verbal Paired Associates—immediate and delayed recall (Weschler 1997
     Camden faces—total number correct (Warrington 1996
     Long-term memory encoding and recognition Task—number correct deep and shallow encoding levels (Craik and Lockhart 1972; Fletcher and others 2003
2. Verbal working memory 
     WMS-III Digit Span—forward and backward (Weschler 1997
     Verbal Sternberg (loads 1, 3, and 5)—total number correct (Sternberg 1969a, 1969b, 1975) 
3. Spatial working memory 
     CANTAB spatial span—span (Owen and others 1990
     CANTAB spatial working memory—strategy score, both between- and within-search errors (Owen and others 1990; Joyce and Robbins 1991
     Spatial Sternberg task (loads 1, 3, and 5)—total number correct (Sternberg 1969a, 1969b, 1975
4. Inhibition 
     MARS go/no-go Task—percent inhibition, omission, and prepotent errors (Rubia and others 2001
     MARS stop task—omission and prepotent errors (Rubia and others 2001
     MARS switch task—number correct for switch and repeat trials, omission, and prepotent errors (Rubia and others 2001
5. Processing speed 
     Spatial Sternberg task—learning slope and 0-intercept (Sternberg 1969a, 1969b, 1975
     Verbal Sternberg task—learning slope and 0-intercept (Sternberg 1969a, 1969b, 1975
     MARS II go/no-go—mean reaction time (Rubia and others 2001
     MARS II stop—mean reaction time and stop signal delay (Rubia and others 2001
     MARS II switch—mean reaction time (Rubia and others 2001
     Long-term memory encoding and recognition task—mean reaction times for correct and incorrect responses (Craik and Lockhart 1972; Fletcher and others 2003
6. Visuospatial skills 
     Rey–Osterrieth Complex Figure Test—copy (Rey 1941; Osterrieth 1944
1. Long-term memory 
     Rey–Osterrieth Complex Figure —delayed recall (Rey 1941; Osterrieth 1944
     CANTAB nonverbal paired associate learning (level 6)—total errors (Sahakian and others 1988; Fowler and others 1995
     WMS-III Logical Memory—immediate and delayed recall (Weschler 1997
     WMS-III Verbal Paired Associates—immediate and delayed recall (Weschler 1997
     Camden faces—total number correct (Warrington 1996
     Long-term memory encoding and recognition Task—number correct deep and shallow encoding levels (Craik and Lockhart 1972; Fletcher and others 2003
2. Verbal working memory 
     WMS-III Digit Span—forward and backward (Weschler 1997
     Verbal Sternberg (loads 1, 3, and 5)—total number correct (Sternberg 1969a, 1969b, 1975) 
3. Spatial working memory 
     CANTAB spatial span—span (Owen and others 1990
     CANTAB spatial working memory—strategy score, both between- and within-search errors (Owen and others 1990; Joyce and Robbins 1991
     Spatial Sternberg task (loads 1, 3, and 5)—total number correct (Sternberg 1969a, 1969b, 1975
4. Inhibition 
     MARS go/no-go Task—percent inhibition, omission, and prepotent errors (Rubia and others 2001
     MARS stop task—omission and prepotent errors (Rubia and others 2001
     MARS switch task—number correct for switch and repeat trials, omission, and prepotent errors (Rubia and others 2001
5. Processing speed 
     Spatial Sternberg task—learning slope and 0-intercept (Sternberg 1969a, 1969b, 1975
     Verbal Sternberg task—learning slope and 0-intercept (Sternberg 1969a, 1969b, 1975
     MARS II go/no-go—mean reaction time (Rubia and others 2001
     MARS II stop—mean reaction time and stop signal delay (Rubia and others 2001
     MARS II switch—mean reaction time (Rubia and others 2001
     Long-term memory encoding and recognition task—mean reaction times for correct and incorrect responses (Craik and Lockhart 1972; Fletcher and others 2003
6. Visuospatial skills 
     Rey–Osterrieth Complex Figure Test—copy (Rey 1941; Osterrieth 1944

Note: WMS-III is the Wechsler Memory Scale 3rd edition, CANTAB is the Cambridge Neuropsychological Test Automated Battery, and MARS II is the Maudsley Attention and Response Suppression task battery II.

Structure–function Analysis of Neuropsychological and MRI Data

Associations between cognitive domain scores of interest and gray matter volume were tested by computational morphometry separately for each domain, and corroborated at a regional level of analysis, using a simple linear regression model with cognitive domain score as the independent variable. Voxel-level statistic maps, representing local strength of association between each cognitive domain score and gray matter structure (across all subjects), were tested for statistical significance using a nonparametric permutation test of the mass or sum of suprathreshold voxel clusters (Bullmore and others 1999; Suckling and Bullmore 2004). For whole-brain maps, the size of each clusterwise test was set such that the expected number of false-positive tests in each map was less than 1: clusterwise P < 0.005.

Cortical and subcortical areas of significant association identified in this way by whole-brain mapping were corroborated by fitting the same regression model to gray matter volumes in corresponding regions defined by the AAL template image and testing the strength of regional association by a t-test on the standardized regression coefficients (allowing a direct comparison with analysis performed on identically parcellated radiological measures of classical lesions). We also fitted an analysis of covariance (ANCOVA) model to estimate the main effect of the diagnostic group (healthy volunteers vs. people with TSC), the effect of cognitive domain scores, and the group × cognitive performance interaction on gray matter volume in the regions of association identified in the morphometric analysis of the entire sample. The same process was completed independently for each cognitive domain score.

Results

Structural MRI Data: Case–Control Differences

Whole-Brain and Total Tissue Class Volumes

There were no significant differences in whole-brain volume (2-sample t-test, t = −0.236, df = 48, P < 0.814) or total gray matter (t = 0.529, df = 48, P < 0.599), white matter (t = −0.687, df = 48, P < 0.495), or CSF (t = −0.553, df = 48, P < 0.583) volumes when patients with TSC were compared with healthy volunteers. Note that these whole-brain measures were estimated prior to spatial normalization of the images in each subject's native space; the subsequent morphometric results, estimated after spatial normalization to a standard template image, will have effectively controlled for intersubject differences in global brain size.

Computational Morphometry of Gray Matter

The patients with TSC demonstrated extensive and bilaterally symmetric deficits of gray matter in the cerebellum, midbrain, thalamus, basal ganglia (caudate, putamen, and globus pallidus), and insula. There were also deficits in anterior cingulate cortex and right medial temporal lobe; see panel D of Figure 1. Over all these regions of deficit, gray matter volume was reduced by approximately 19% in patients with TSC (2-sample t-test, t = 7.132, df = 48, P < 0.001).

Figure 1

Gray matter deficits in patients with TSC and maps locating gray matter variation significantly associated with memory test scores. Panels (AC) present areas (yellow) where greater tissue volume was predicted by a high cognitive domain score over all subjects (N = 50). Panel (D) illustrates the case–control comparison in gray matter volume (where red denotes regions of relative deficit in the TSC group). Panel (E) represents the TSC group's lesion density map showing frequency of tubers (red/green color scale) and SENs (blue/purple color scale) and is presented here to aid comparisons (see also Ridler and others 2004). The left side of each panel represents the right side of the brain; the z coordinates for each axial slice in the standard space of Talairach and Tournoux are given in millimeters.

Figure 1

Gray matter deficits in patients with TSC and maps locating gray matter variation significantly associated with memory test scores. Panels (AC) present areas (yellow) where greater tissue volume was predicted by a high cognitive domain score over all subjects (N = 50). Panel (D) illustrates the case–control comparison in gray matter volume (where red denotes regions of relative deficit in the TSC group). Panel (E) represents the TSC group's lesion density map showing frequency of tubers (red/green color scale) and SENs (blue/purple color scale) and is presented here to aid comparisons (see also Ridler and others 2004). The left side of each panel represents the right side of the brain; the z coordinates for each axial slice in the standard space of Talairach and Tournoux are given in millimeters.

Figure 2

Between-group comparisons of proportional gray matter volumes in major cortical lobes and subcortical structures. Also shown on the same scale are proportional volumes of SENs (dark gray) and tubers (light gray). Subcortical gray matter deficits remain significant in the patients with TSC, even on the conservative assumption that all voxels representing radiological lesions were classified as tissues other than gray matter (white matter or CSF). Asterisks indicate regions with significant between-group difference in gray matter volume; 2-sample t-test, P < 0.01.

Figure 2

Between-group comparisons of proportional gray matter volumes in major cortical lobes and subcortical structures. Also shown on the same scale are proportional volumes of SENs (dark gray) and tubers (light gray). Subcortical gray matter deficits remain significant in the patients with TSC, even on the conservative assumption that all voxels representing radiological lesions were classified as tissues other than gray matter (white matter or CSF). Asterisks indicate regions with significant between-group difference in gray matter volume; 2-sample t-test, P < 0.01.

Regional Volumetrics of Gray Matter

The patients with TSC demonstrated significant gray matter deficits in basal ganglia, thalamus, and cerebellum regions defined by a parcellated anatomical template image; see Table 3. It is important to note that these regional deficits remained significant even after gray matter volumes in the patient group were combined with tuber and nodule volumes in the corresponding regions, implying that the gray matter deficits were not simply attributable to classification of subcortical tubers or nodules as white matter or CSF; see Figure 2.

Table 3

Regional distributions of tubers and nodules in patients with TSC and regional gray matter volumes in both patients with TSC and healthy volunteers

Anatomical region Lesion load Gray matter volume (ml) 
 SENs Tubers 
 % of subjects mean % occupancy % bilateral % of subjects mean % occupancy % bilateral TSC (SD) Healthy volunteers (SD) t P 
Frontal 24.0 0.010 66.7 92.0 0.543 91.3 213.60 (11.7) 213.77 (13.3) 0.048 0.963 
Occipital 0.0 0.000  76.0 0.344 63.2 96.18 (5.3) 98.52 (6.3) 1.419 0.162 
Parietal 0.0 0.000  88.0 0.698 63.6 71.62 (6.0) 73.62 (4.6) 1.322 0.193 
Temporal cortical 12.0 0.000 0.0 80.0 0.346 40.0 83.48 (4.0) 81.26 (5.5) −1.641 0.107 
Temporal subcortical 16.0 0.006 25.0 8.0 0.023 50.0 36.62 (3.7) 36.16 (3.7) −0.441 0.661 
Cingulate 16.0 0.004 25.0 76.0 0.437 42.1 32.86 (1.8) 32.69 (2.0) −0.309 0.758 
Insula 16.0 0.005 50.0 32.0 0.156 0.0 16.31 (1.1) 16.31 (1.3) 0.012 0.991 
Basal ganglia 56.0 0.661 71.4 8.0 0.006 0.0 19.98 (1.6) 21.33 (1.2) 3.402 0.001** 
Thalamus 48.0 0.181 50.0 0.0 0.000  8.73 (1.0) 9.84 (0.7) 4.428 0.001** 
Cerebellum 0.0 0.000  8.0 0.047 0.0 67.42 (12.7) 74.93 (7.8) 2.518 0.015* 
Total 64.0 0.023 68.75 92.0 0.476 100.0 848.72 (39.8) 860.52 (41.2) 1.030 0.308 
Anatomical region Lesion load Gray matter volume (ml) 
 SENs Tubers 
 % of subjects mean % occupancy % bilateral % of subjects mean % occupancy % bilateral TSC (SD) Healthy volunteers (SD) t P 
Frontal 24.0 0.010 66.7 92.0 0.543 91.3 213.60 (11.7) 213.77 (13.3) 0.048 0.963 
Occipital 0.0 0.000  76.0 0.344 63.2 96.18 (5.3) 98.52 (6.3) 1.419 0.162 
Parietal 0.0 0.000  88.0 0.698 63.6 71.62 (6.0) 73.62 (4.6) 1.322 0.193 
Temporal cortical 12.0 0.000 0.0 80.0 0.346 40.0 83.48 (4.0) 81.26 (5.5) −1.641 0.107 
Temporal subcortical 16.0 0.006 25.0 8.0 0.023 50.0 36.62 (3.7) 36.16 (3.7) −0.441 0.661 
Cingulate 16.0 0.004 25.0 76.0 0.437 42.1 32.86 (1.8) 32.69 (2.0) −0.309 0.758 
Insula 16.0 0.005 50.0 32.0 0.156 0.0 16.31 (1.1) 16.31 (1.3) 0.012 0.991 
Basal ganglia 56.0 0.661 71.4 8.0 0.006 0.0 19.98 (1.6) 21.33 (1.2) 3.402 0.001** 
Thalamus 48.0 0.181 50.0 0.0 0.000  8.73 (1.0) 9.84 (0.7) 4.428 0.001** 
Cerebellum 0.0 0.000  8.0 0.047 0.0 67.42 (12.7) 74.93 (7.8) 2.518 0.015* 
Total 64.0 0.023 68.75 92.0 0.476 100.0 848.72 (39.8) 860.52 (41.2) 1.030 0.308 

Note: This table summarizes the lesion load profile, presented in a form to aid the direct comparison of regional gray matter volumes with regional volumes of tubers and nodules. The frequency and distribution of lesions in this sample of patients with TSC is presented in detail elsewhere (Ridler and others 2004). % of subjects, percent of subjects with lesion affecting that structure; mean % occupancy, the mean percent of lesion occupancy (lesion volume in a region/total volume of the region × 100); % bilateral, percent of the subjects with a bilateral lesion in a given structure (measures of cerebellar lesion laterality exclude the cerebellar vermis); P-values less than 0.05 are given in bold; ** P < 0.001.

Computational Morphometry of White Matter

The patients with TSC demonstrated extensive and bilaterally symmetric deficits of white matter in the territory of major intrahemispheric tracts including the superior and inferior longitudinal fasciculi, subgyral white matter in the inferior parietal and medial temporal lobes, the tapetum, and the optic radiation. Over all these regions of deficit, white matter volume was reduced by approximately 15% in patients with TSC (t = 11.151, df = 48, P < 0.001). There were also areas of white matter excess (t = −6.041, df = 48, P < 0.001), which complemented and were located adjacent to the regions of gray matter deficit already described in cerebellum, thalamus, basal ganglia, and insula.

Computational Morphometry of CSF

The patients with TSC demonstrated significantly increased CSF volume in the lateral ventricles bilaterally and in the third ventricle. The relative increase of CSF in these regions was 28% (t = −4.5, df = 48, P < 0.001). There was also an area of relative decrease of extracerebral CSF by 24% (t = 4.9, df = 48, P < 0.001). These changes were in the absence of obstructive hydrocephalus.

Relationships between Morphometric Abnormalities and Radiological Lesions

The frequency and distribution of lesions in this sample of patients with TSC is presented in detail elsewhere (Ridler and others 2004). A summary of the lesion load profile is presented in Table 3. By inspection of Figure 1, the distribution of nodules, tubers, and white matter abnormalities (panel E) is directly comparable with the distribution of morphometric deficits in gray and white matter (panel D). It is evident that there was little exact anatomical coincidence between these two measures of neuroanatomical abnormality. More quantitatively, less than 11% of the gray or white matter deficits identified using morphometric analysis coincided anatomically with the locations of radiological lesions in one or more subjects.

There was, however, some evidence for correlation between morphometric and radiological abnormalities. Total SEN volume (r = −0.41, P = 0.043) and total tuber count (r = −0.53, P = 0.006) were both negatively correlated with white matter volume in the right frontal region of relative deficit in the TSC group. Total tuber volume was also negatively correlated with gray matter volume in bilateral thalamic and basal ganglia regions of relative deficit in the TSC group (r = −0.383, P = 0.058), although this association was not quite significant.

At a regional level of analysis, volumes of tubers and nodules in cortical and subcortical regions, respectively, were negatively correlated with gray matter volumes both locally and remotely. For example, tuber volume in the occipital lobe was negatively correlated with local (occipital) gray matter volume (r = −0.491, P < 0.013); tuber volume in frontal lobe was negatively correlated with gray matter volumes of remote regions (lateral temporal lobe [r = −0.422, P < 0.035] and insula [r = −0.406, P < 0.035]); and tuber volume in lateral temporal lobe was negatively correlated with gray matter volumes of medial temporal lobe (r = −0.480, P < 0.015) and thalamus (r = −0.523, P < 0.007). There were also negative correlations between SEN volume in basal ganglia and gray matter volumes of both basal ganglia (r = −0.393, P < 0.052) and thalamus (r = −0.389, P < 0.055), but these were not quite significant.

Neuropsychological Tests: Case–Control Differences

There was an uneven profile of cognitive impairment in the TSC group (see Table 4). Patients were significantly impaired on tests of long-term memory, especially immediate recall rather than recognition, and on tests of working memory, especially the backward digit span and the between-search errors score of the Cambridge Neuropsychological Test Automated Battery spatial working memory task. Although patients encoded less information in recall tasks, they showed no increased rate of forgetting after a delay. In contrast, there were few significant differences between groups on measures of inhibition, processing speed, or visuospatial function (Table 4).

Table 4

Results of neuropsychological testing

Cognitive domains and component neuropsychological tests Measure Healthy volunteers (N = 25) Adults with TSC (N = 25) F (df), ANOVA P-value 
Long-term memory      
    Recall memory      
     Rey–Osterrieth Complex Figure Figure recall % 59.89 (13.8) 50.63 (18.1) 4.917 (1,46) 0.032 
     CANTAB PAL Errors (level 6 only) 15.12 (10.6) 8.72 (3.6) 8.090 (1,48) 0.007 
     WMS-III Story Immediate recall 16.36 (3.3) 12.92 (3.8) 12.04 (1,48) 0.001** 
     WMS-III Story Delayed recall % 89.82 (16.4) 86.66 (19.3) 0.390 (1,48) 0.535 
     WMS-III Verbal PAL Immediate recall 20.88 (7.7) 14.80 (8.8) 6.757 (1,48) 0.012 
     WMS-III Verbal PAL delayed recall % 96.67 (18.6) 105.12 (66.6) 0.374 (1,48) 0.544 
    Recognition memory      
     CAMDEN Faces % Recognized 24.00 (1.4) 23.12(2.1) 1.733 (1,48) 0.089 
     Word Encoding/Recognition Deep Correct 55.36 (12.7) 57(9.5) 0.259 (1,48) 0.613 
     Word Encoding/Recognition Shallow Correct 31.56 (9.48) 33.79(11.4) 0.557 (1,48) 0.459 
Verbal working memory      
    WMS-III Digit Span Forward score 10.93 (2.2) 9.64(2.1) 3.411 (1,38) 0.073 
    WMS-III Digit Span Backward score 8.33 (2.7) 6(1.7) 11.190 (1,38) 0.002 
    Verbal Sternberg % correct 91.00 (8.1) 89.93(7.3) 0.231 (1,46) 0.633 
Spatial working memory      
    CANTAB Spatial Span Span 6.88 (1.3) 5.72 (1.1) 11.587 (1,48) 0.001** 
    CANTAB SWM Between-search error 17.20 (2.9) 33.92 (18.6) 13.070 (1,48) 0.001** 
    CANTAB SWM Within search errors 1.24 (2.6) 2.24 (2.9) 1.661 (1,48) 0.204 
    CANTAB SWM Strategy 31.48 (5.1) 33.48 (5.3) 1.868 (1,48) 0.178 
    Spatial Sternberg % correct 93.26 (4.1) 91.66 (7.8) 0.800 (1,47) 0.376 
Inhibition      
    MARS Go/no-go Percent inhibited 80.10 (11.1) 82.66 (11) 0.660 (1,47) 0.421 
    MARS Go/no-go Omission errors 1.42 (3.9) 1.16 (2.4) 0.516 (1, 47) 0.476 
    MARS Go/no-go Prepotent errors 5.00 (4.6) 4.43 (11.2) 0.053 (1,47) 0.819 
    MARS Stop Omission errors 1.29 (2.5) 3.75 (4.3) 6.095 (1,48) 0.017 
    MARS Stop Prepotent 1.13 (2.5) 2.49 (6.1) 1.061 (1,48) 0.308 
    MARS Switch Switch % correct 93.13 (6.1) 86.30 (21.7) 2.271 (1,48) 0.138 
    MARS Switch Repeat % correct 95.08 (4.7) 87.21(20.0) 4.452 (1, 48) 0.040 
    MARS Switch Omission errors 1.20 (2.2) 2.21 (4.3) 1.111 (1,48) 0.297 
    MARS Switch Prepotent errors 1.56 (3.1) 6.66 (18.5) 1.842 (1,48) 0.181 
Processing speed      
    Spatial Sternberg Learning slope 93.40 (39.4) 93.15 (74.9) 0.036 (1,47) 0.851 
    Spatial Sternberg 0-intercept 606.64 (172.6) 751.27 (347.3) 3.362 (1,47) 0.073 
    Verbal Sternberg Learning slope 57.48 (39.7) 58.71 (47.9) 0.009 (1,46) 0.923 
    Verbal Sternberg 0-intercept 687.4 3(192.4) 744.80 (39.7) 0.896 (1,46) 0.349 
    MARS Go/no-go MRT 301.19 (59) 313.70 (54) 0.593 (1,44) 0.460 
    MARS Stop SSD 97.12 (49) 105.28 (51) 0.328 (1,48) 0.569 
    MARS Stop MRT 551.79 (149) 586.1 (160) 0.613 (1,48) 0.437 
    MARS Switch MRT 587.3 (103) 631.17 (163) 1.282 (1,48) 0.263 
    Word encoding/recognition MRT correct responses 1014.5 (209.5) 1056.7 (184.9) 0.555 (1,48) 0.460 
    Word encoding/recognition MRT incorrect responses 1175.6 (262.6) 1221.0 (181.6) 0.491 (1,48) 0.481 
Visuospatial skills      
    Rey–Osterrieth Complex Figure Figure—copy 34.10 (2.0) 35.25 (1.0) 6.065 (1,46) 0.018 
Cognitive domains and component neuropsychological tests Measure Healthy volunteers (N = 25) Adults with TSC (N = 25) F (df), ANOVA P-value 
Long-term memory      
    Recall memory      
     Rey–Osterrieth Complex Figure Figure recall % 59.89 (13.8) 50.63 (18.1) 4.917 (1,46) 0.032 
     CANTAB PAL Errors (level 6 only) 15.12 (10.6) 8.72 (3.6) 8.090 (1,48) 0.007 
     WMS-III Story Immediate recall 16.36 (3.3) 12.92 (3.8) 12.04 (1,48) 0.001** 
     WMS-III Story Delayed recall % 89.82 (16.4) 86.66 (19.3) 0.390 (1,48) 0.535 
     WMS-III Verbal PAL Immediate recall 20.88 (7.7) 14.80 (8.8) 6.757 (1,48) 0.012 
     WMS-III Verbal PAL delayed recall % 96.67 (18.6) 105.12 (66.6) 0.374 (1,48) 0.544 
    Recognition memory      
     CAMDEN Faces % Recognized 24.00 (1.4) 23.12(2.1) 1.733 (1,48) 0.089 
     Word Encoding/Recognition Deep Correct 55.36 (12.7) 57(9.5) 0.259 (1,48) 0.613 
     Word Encoding/Recognition Shallow Correct 31.56 (9.48) 33.79(11.4) 0.557 (1,48) 0.459 
Verbal working memory      
    WMS-III Digit Span Forward score 10.93 (2.2) 9.64(2.1) 3.411 (1,38) 0.073 
    WMS-III Digit Span Backward score 8.33 (2.7) 6(1.7) 11.190 (1,38) 0.002 
    Verbal Sternberg % correct 91.00 (8.1) 89.93(7.3) 0.231 (1,46) 0.633 
Spatial working memory      
    CANTAB Spatial Span Span 6.88 (1.3) 5.72 (1.1) 11.587 (1,48) 0.001** 
    CANTAB SWM Between-search error 17.20 (2.9) 33.92 (18.6) 13.070 (1,48) 0.001** 
    CANTAB SWM Within search errors 1.24 (2.6) 2.24 (2.9) 1.661 (1,48) 0.204 
    CANTAB SWM Strategy 31.48 (5.1) 33.48 (5.3) 1.868 (1,48) 0.178 
    Spatial Sternberg % correct 93.26 (4.1) 91.66 (7.8) 0.800 (1,47) 0.376 
Inhibition      
    MARS Go/no-go Percent inhibited 80.10 (11.1) 82.66 (11) 0.660 (1,47) 0.421 
    MARS Go/no-go Omission errors 1.42 (3.9) 1.16 (2.4) 0.516 (1, 47) 0.476 
    MARS Go/no-go Prepotent errors 5.00 (4.6) 4.43 (11.2) 0.053 (1,47) 0.819 
    MARS Stop Omission errors 1.29 (2.5) 3.75 (4.3) 6.095 (1,48) 0.017 
    MARS Stop Prepotent 1.13 (2.5) 2.49 (6.1) 1.061 (1,48) 0.308 
    MARS Switch Switch % correct 93.13 (6.1) 86.30 (21.7) 2.271 (1,48) 0.138 
    MARS Switch Repeat % correct 95.08 (4.7) 87.21(20.0) 4.452 (1, 48) 0.040 
    MARS Switch Omission errors 1.20 (2.2) 2.21 (4.3) 1.111 (1,48) 0.297 
    MARS Switch Prepotent errors 1.56 (3.1) 6.66 (18.5) 1.842 (1,48) 0.181 
Processing speed      
    Spatial Sternberg Learning slope 93.40 (39.4) 93.15 (74.9) 0.036 (1,47) 0.851 
    Spatial Sternberg 0-intercept 606.64 (172.6) 751.27 (347.3) 3.362 (1,47) 0.073 
    Verbal Sternberg Learning slope 57.48 (39.7) 58.71 (47.9) 0.009 (1,46) 0.923 
    Verbal Sternberg 0-intercept 687.4 3(192.4) 744.80 (39.7) 0.896 (1,46) 0.349 
    MARS Go/no-go MRT 301.19 (59) 313.70 (54) 0.593 (1,44) 0.460 
    MARS Stop SSD 97.12 (49) 105.28 (51) 0.328 (1,48) 0.569 
    MARS Stop MRT 551.79 (149) 586.1 (160) 0.613 (1,48) 0.437 
    MARS Switch MRT 587.3 (103) 631.17 (163) 1.282 (1,48) 0.263 
    Word encoding/recognition MRT correct responses 1014.5 (209.5) 1056.7 (184.9) 0.555 (1,48) 0.460 
    Word encoding/recognition MRT incorrect responses 1175.6 (262.6) 1221.0 (181.6) 0.491 (1,48) 0.481 
Visuospatial skills      
    Rey–Osterrieth Complex Figure Figure—copy 34.10 (2.0) 35.25 (1.0) 6.065 (1,46) 0.018 

Note: Group mean scores and between-group comparisons are shown for each outcome measure of multiple tests of long-term memory, working memory, inhibition, processing speed, and visuospatial function. P-values less than 0.01 are given in bold. ANOVA, analysis of variance; CANTAB, Cambridge Neuropsychological Test Automated Battery; WMS-III, Wechsler Memory Scale 3rd Edition; MARS, Maudsley Attention and Response Suppression Task Battery; PAL, Paired Associate Learning; SWM, Spatial Working Memory; MRT; Mean Reaction Time; SSD, Stop-Signal delay. **indicates P < 0.0013 (survived Bonferroni correction for multiple comparisons).

This pattern of results in the raw scores was confirmed by the analysis of between-group differences on the summary scores for each domain (Table 5). Patients with TSC were significantly (P < 0.01) impaired on average over all tests of long-term memory, verbal working memory, and spatial working memory; but not significantly impaired on tests of inhibition, processing speed, or visuospatial skills.

Table 5

Summary scores on domain-specific cognitive function tests in patients with TSC

Domain of cognitive functioning Mean (SD) Z-score TSC Range of Z-score TSC Mean (SD) Z-score control Range of Z-score control t P-value 
Long-term memory −0.528 (0.717) −2.267 to 1.032 0 (0.432) −0.971 to 1.181 3.154 0.003 
Verbal working memory −1.056 (0.983) −3.281 to 0.796 0 (1.000) −1.533 to 2.835 3.765 0.001 
Spatial working memory −0.841 (0.931) −2.895 to 0.587 0 (0.704) −1.688 to 1.137 3.604 0.001 
Inhibition −0.727 (1.729) −5.790 to 0.607 0 (0.416) −1.278 to 0.563 2.045 0.046 
Processing speed −0.265 (0.520) −1.404 to 0.666 0 (0.489) −0.990 to 1.078 1.855 0.070 
Visuospatial Skills −1.109 (1.971) −6.06 to 0.730 0 (1.00) −2.18 to 0.730 2.463 0.018 
Domain of cognitive functioning Mean (SD) Z-score TSC Range of Z-score TSC Mean (SD) Z-score control Range of Z-score control t P-value 
Long-term memory −0.528 (0.717) −2.267 to 1.032 0 (0.432) −0.971 to 1.181 3.154 0.003 
Verbal working memory −1.056 (0.983) −3.281 to 0.796 0 (1.000) −1.533 to 2.835 3.765 0.001 
Spatial working memory −0.841 (0.931) −2.895 to 0.587 0 (0.704) −1.688 to 1.137 3.604 0.001 
Inhibition −0.727 (1.729) −5.790 to 0.607 0 (0.416) −1.278 to 0.563 2.045 0.046 
Processing speed −0.265 (0.520) −1.404 to 0.666 0 (0.489) −0.990 to 1.078 1.855 0.070 
Visuospatial Skills −1.109 (1.971) −6.06 to 0.730 0 (1.00) −2.18 to 0.730 2.463 0.018 

Note: The t-statistics and P-values refer to tests of the null hypothesis of zero between-group difference in group mean domain summary scores; for all domains, the healthy group mean score is zero by definition; P-values less than 0.05 are given in bold.

There were no significant correlations between age of seizure onset and cognitive domain scores within the subset of participants with TSC who had a lifetime history of seizures (N = 16). There was no significant difference in gray matter volume in deficit areas between the TSC patients with a history of seizures and the TSC patients without seizures (F1,23 = 1.508, P = 0.232). There was, however, still a significant difference in gray matter volume in deficit areas between healthy comparison subjects (n = 25) and subjects with TSC without a history of seizures (F1,32 = 42.9, P < 0.001).

Structure–Function Associations

Long-Term Memory

There was a significant positive correlation over all subjects between long-term memory Z-score and gray matter volume in bilateral thalamus (Talairach coordinates: 9, −10, 15), bilateral caudate nucleus, and left anterior cingulate cortex (approximate Brodmann area [BA] 32; Fig. 1A). In other words, better long-term memory performance was associated with greater gray matter volume in these regions.

Total gray matter volume estimated for each subject over all these regions of morphometrically defined structure–function association was correlated with long-term memory score (r = 0.616, df = 49, P < 0.00001; see Fig. 3A). An ANCOVA model fitted to the long-term memory scores using these anatomical summary data showed no significant effect of group (F1,46 = 2.714, P < 0.106), a significant effect of gray matter volume (F1,46 = 15.928, P < 0.001), but no significant group × gray matter interaction (F1,46 = 0.151, P < 0.700).

Figure 3

Structure–function associations between gray matter volumes and memory test scores. Panel (A) represents association between gray matter volume in morphometrically identified areas (bilateral thalamus, caudate, and anterior cingulate cortex) and long-term memory scores. Panel (B) represents association between gray matter volume in morphometrically identified areas (bilateral thalamus and caudate nucleus; right parahippocampal gyrus; and medial, middle, and superior frontal gyri) and verbal working memory scores. Panel (C) represents association between gray matter volume in morphometrically defined areas (left anterior cingulate cortex, bilateral thalamus, and caudate nucleus) and spatial working memory scores. Associations were morphometrically identified on the basis of the whole study sample and therefore included data from both the healthy comparison and TSC groups (N = 50). The solid lines and filled circles represent the association separately estimated for the healthy volunteers, and the broken lines and open circles represent the association separately estimated for subjects with TSC. For domain scores, a positive score represented an above-average performance and a negative score a below-average performance.

Figure 3

Structure–function associations between gray matter volumes and memory test scores. Panel (A) represents association between gray matter volume in morphometrically identified areas (bilateral thalamus, caudate, and anterior cingulate cortex) and long-term memory scores. Panel (B) represents association between gray matter volume in morphometrically identified areas (bilateral thalamus and caudate nucleus; right parahippocampal gyrus; and medial, middle, and superior frontal gyri) and verbal working memory scores. Panel (C) represents association between gray matter volume in morphometrically defined areas (left anterior cingulate cortex, bilateral thalamus, and caudate nucleus) and spatial working memory scores. Associations were morphometrically identified on the basis of the whole study sample and therefore included data from both the healthy comparison and TSC groups (N = 50). The solid lines and filled circles represent the association separately estimated for the healthy volunteers, and the broken lines and open circles represent the association separately estimated for subjects with TSC. For domain scores, a positive score represented an above-average performance and a negative score a below-average performance.

Verbal Working Memory

There was a significant positive correlation between verbal working memory Z-score and gray matter volume in bilateral thalamus (Talairach coordinates: 4, −13, 5), bilateral caudate nucleus, right parahippocampal gyrus (BA 30), and several regions of frontal cortex including medial, middle, and superior frontal gyri (BA 6, 8, and 9; Fig. 1B). In other words, better verbal working memory performance was associated with greater gray matter volume in these regions.

Total gray matter volume estimated for each subject over all these regions of morphometrically defined structure–function association was correlated with verbal working memory score (r = 0.616, df = 39, P < 0.0001; see Fig. 3B). An ANCOVA model fitted to the verbal working memory scores using these anatomical summary data showed a significant effect of group (F1,36 = 6.498, P < 0.015), a significant effect of gray matter volume (F1,36 = 69.392, P < 0.001), but no significant group × gray matter interaction (F1,36 = 1.733, P < 0.196).

Spatial Working Memory

There was a significant positive correlation between spatial working memory Z-score and gray matter volume in bilateral thalamus (Talairach coordinates: 2, −2, 13), bilateral caudate nucleus, and left anterior cingulate cortex (Fig. 1C). In other words, better spatial working memory performance was associated with greater gray matter volume in these regions.

Total gray matter volume estimated for each subject over all these regions of morphometrically defined structure–function association was correlated with spatial working memory score (r = 0.600, df = 49, P < 0.0001; see Fig. 3C). An ANCOVA model fitted to the spatial working memory scores using these anatomical summary data showed no significant effect of group (F1,46 = 2.268, P < 0.139), a significant effect of gray matter volume (F1,46 = 10.030, P < 0.003), and a significant group × gray matter interaction (F1,46 = 4.472, P < 0.040).

Structure–Function Associations: Volumetric Results

We sought to corroborate the results of computational morphometry by testing for association between memory performance and regional volumes estimated by automated parcellation in thalamus and basal ganglia. Gray matter volume in thalamus was positively correlated with verbal working memory (r = 0.363, df = 39, P < 0.021), spatial working memory (r = 0.330, df = 49, P < 0.019), and long-term memory (r = 0.271, df = 49, P < 0.057). Gray matter volume in basal ganglia was positively correlated with verbal working memory (r = 0.307, df = 39, P < 0.054), spatial working memory (r = 0.349, df= 49, P < 0.013), and long-term memory (r = 0.273, df = 49, P < 0.055).

Structure–Function Associations: Neuroradiological Lesion Load

The total number of tubers, the total volume of tubers, and the total volume of SENs were estimated for each patient with TSC. There was no significant correlation between any of these global measures of lesion load and cognitive performance summarized by any of the domain-specific Z-scores. There were also no significant correlations between regional measures of lesion load (see Table 3) and cognitive performance summarized by any of the domain-specific Z-scores.

Discussion

We have reported 3 main findings: evidence for a bilaterally symmetric, predominantly subcortical pattern of gray matter deficit in patients with TSC; evidence for an uneven profile of memory impairment implicating long-term and working memory in these high-functioning patients compared with IQ-matched volunteers; and evidence that memory performance was positively correlated with gray matter density in thalamic, striatal, and prefrontal brain regions. It therefore seems likely that a key neuroanatomical substrate of selective memory deficit in these patients with TSC is gray matter deficit in subcortical nuclei (thalamus and caudate). In contrast, there was no evidence for a significant association between cognitive impairment and neuroradiological lesion load in patients with TSC.

Extending the Definition of Neuroanatomical Abnormalities in TSC

We have demonstrated extensive and symmetric deficits of gray matter in subcortical structures (cerebellum, thalamus, and basal ganglia) that broadly replicate the bilaterally symmetric deficits of subcortical gray matter that we reported in a previous morphometric analysis of an independent study of adults with TSC (Ridler and others 2001). We have also replicated our previous finding of bilaterally symmetric and extensive deficits of white matter in the territories of major intrahemispheric tracts (Ridler and others 2001). Interestingly, there is significant overlap between deficit regions reported here and subsequent findings identified using a manual measurement technique (Yorns and others 2001; DiMario 2004), despite different samples and methods. Moreover, in the current study, we have coregistered whole-brain maps of gray matter and white matter in a standard space with radiologically defined maps of tubers and nodules. This has both facilitated a quantitative analysis of classical lesion load and allowed us to explore systematically the nature of association between morphometric and radiological markers of neuroanatomical abnormality in patients with TSC.

We have previously reported that the distribution of tubers and nodules is regionally heterogeneous but not significantly lateralized and that the density of tubers and nodules is positively correlated (Ridler and others 2004). There was a minor degree of exact anatomical coincidence between the distributions of tubers and nodules and the distributions of gray and white matter deficit in people with TSC. However, there was only one example (in occipital cortex) of a strong negative correlation between tuber volume and local gray matter volume and only one marginally significant example of a negative correlation between nodule volume and local gray matter volume (in basal ganglia). This suggests that not all parenchymal deficits directly reflect local lesion load. We also tested the related hypothesis that deficits could apparently arise due to local replacement of typical gray matter by atypical tubers or nodules that might be less probably classified as gray matter by our segmentation algorithm. However, the volumes of radiological lesions were generally small compared with the scale of between-group differences in gray matter, and gray matter deficits in cerebellum, thalamus, and basal ganglia remained significant even after correction of regional volumes for the presence of tubers or nodules in the group of patients with TSC.

In addition to these results indicating that gray matter deficits could not simply be explained by the local density of classical lesions, we also found a number of negative correlations between tuber density and gray matter volume in remote cortical or subcortical structures and a negative correlation between tuber density and right frontal white matter. This finding of morphometric abnormalities correlated with a higher lesion load (whether measured by lesion volume, as in the current study, or lesion number as in some previous studies) has some support from the literature. For example, there are prior reports of a relationship between lesion number and both increased ventricular size (DiMario 2004) and decreased subcortical gray matter volume (Ridler and others 2001). Prior data show that radial migration of tubers to cortex disrupts subjacent white matter and leaves a trail of dysplastic neuronal precursor cells (Braffman and others 1992). These architectural and cellular abnormalities of white matter in the path of tuber migration might be expected to interfere with normal axonal guidance mechanisms that are important in orienting axonal projection from normal cortex in the vicinity of the tuber to distant cortical and subcortical targets. In short, radial migration of tubers to, frontal cortex may, for example, disrupt later developmental processes important for securing anatomical afferentation of remote cortical and subcortical regions by frontal cortex; and relative absence of afferentation is known to have a negative effect on the growth of cortical and subcortical structures (Kolb and others 1983, 1998). This mechanistic hypothesis could be tested in future studies using diffusion tensor imaging to quantify more precisely white matter projections between cortical regions of high tuber density and distantly interconnected cortical and subcortical regions that show negatively correlated reductions in gray matter volume.

Memory Function and Its Neuroanatomical Correlates in TSC

The cognitive profile was uneven for patients with TSC, with patients performing as well as IQ-matched controls on most tasks and disproportionately badly on tasks within several domains. There were significant group differences on all measures of cued or free recall for both verbal and visual modalities. This was specific to immediate recall of information, and there was not an increased rate of forgetting with delay. There were also no differences in any of the measures of recognition memory, despite the fact that verbal recognition scores showed no ceiling effect (a problem that often complicates interpretation of recognition performance). There were significant differences in both the spatial and verbal working memory tasks, which involved executive processes. However, there were no differences on either the verbal or spatial Sternberg tasks, which involved maintenance of information in working memory but no manipulation. There were no salient group differences in any measures of either inhibition or speed of information processing. Therefore, these were not considered to be the underlying cause of TSC group deficits for long-term memory retrieval or executively demanding working memory.

At first glance, this profile seems more “frontal” than might be expected in a classical amnesic syndrome due to medial temporal lobe or diencephalic damage. For example, impairment of recall in the context of spared recognition, no exacerbation of recall as a function of increasing delay, and intact forward digit span but impaired backward digit span are all compatible with a frontal profile of memory impairment (Lezak 1983; Kopelman 2002). There were, however, also differences between the profile of TSC deficits and the classical profile of frontal lobe patients. In fact, the profile of deficit in the TSC group was consistent with neither a classical amnesic syndrome nor a classical frontal syndrome. It is interesting to note that the profiles of both cognitive and brain structural abnormalities share some similarities with the patterns found in normal aging (Rubin 1999).

Results, however, showed clearest similarities to the neuropsychological profile produced by basal ganglia and thalamic abnormalities that has sometimes been termed “frontal-like” or fronto-striatal in nature (Owen and others 1990). The TSC long-term memory profile was similar to that of patients with early Huntington's disease (a basal ganglia disorder) (Lawrence and others 2000) and patients with thalamic lesions (Van der Werf and others 2003; Zoppelt and others 2003). Both of these subcortical disorders cause specific deficits in long-term memory recall, without an increased rate of forgetting, while recognition is usually spared. In relation to working memory, there were also subtle differences between the pattern of test results observed in these subcortical disorders and in frontal lobe lesions. For example, the pattern of SWM deficit in both TSC and early subcortical dementia (Huntington's disease and Parkinson's disease) spared strategy and within-search errors while between-search errors were increased (Owen and others 1990, 1992, 1993, 1997, 1998; Lawrence and others 1998, 2000); whereas patients with frontal lobe lesions typically show an increase in both between- and within-search errors and impairment in the use of an effective strategy (Owen and others 1990; Robbins and others 1999). These long-term and working memory deficits in basal ganglia and thalamic disorders are possibly caused by interference with a critical memory circuit including diencephalic structures, cingulate cortex, temporal lobes, and frontal cortex (Kopelman 2002). One generalization emerging from these and other data is that aspects of both long-term and working memory function may depend on integrated function of similar fronto-striato-thalamic circuits (Alexander and others 1986; Robbins and others 1999; Watkins and others 2000; Green and others 2002; LaBerge 2002). In this context, it is not surprising that subcortical gray matter deficits in TSC are associated with impaired long-term recall and manipulation in working memory (although the dependence of memory in TSC on fronto-striatal circuits also damaged by neurodegenerative disorders should not, of course, be taken to mean that TSC is a progressively dementing neurodegenerative disorder).

We found that volumetric changes in radiologically normal gray matter—most consistently, thalamus, and caudate nucleus—predicted cognitive performance in these domains. The Talairach coordinates of the thalamic locus most consistently or strongly correlated with long-term memory deficit suggest that deficits in the dorsomedial nucleus, implying disruption of the prefronto-caudate-thalamic circuit, are especially prejudicial to recall from long-term memory. The strong correlation of bilateral caudate and thalamic regions and performance in both working memory domains also fits well with other evidence of subcortical involvement in executive components of working memory via fronto-striatal circuits. Furthermore, there appears to be an overlap between the fronto-striatal circuits involved in the executive or attentional control aspects of both working memory and the retrieval aspects of long-term memory tasks that was disrupted on several tests in TSC.

Memory performance was generally dependent on some of the areas identified as abnormal by morphometric case–control comparison rather than on the volume of radiological lesions. There was, in fact, no relationship between global or regional lesion volume and neuropsychological performance. However, the percentage of brain volume affected by lesions in this sample was relatively small, and lesions are known to develop early in brain development, providing ample opportunity for compensatory cortical reorganization.

Molecular Mechanisms of Memory Impairment in TSC

We know that TSC is a neuronal migration disorder with focal and widespread anatomic and histological abnormalities identifiable within the brains of patients with TSC (DiMario 2004). These migrational disturbances are a direct result of disturbed cellular control owing to a loss of the functional interaction of the hamartin–tuberin complex with its downstream nuclear signaling components (DiMario 2004). As a consequence, regional morphometric brain disturbances have been described both in conjunction with and separate from migration lesions (Ridler and others 2001; DiMario 2004).

It is interesting to note that memory deficits were specifically implicated in our study of TSC when we recall that one of the downstream targets of hamartin and tuberin is molecular target of rapamycin (mTOR) (Au and others 2004). mTOR is implicated in the cellular mechanisms of memory (Kelleher and others 2004) and, although present in the classical brain lesions, tuberin and hamartin are unable to regulate the activity of downstream targets like mTOR in patients with TSC (Han and others 2004). It will be important in future studies to integrate molecular and system-level markers relevant to memory dysfunction in TSC and, indeed, to consider a possible therapeutic role for rapamycin in prevention or remediation of cognitive deficits associated with this disorder.

Inhibition and Information-Processing Speed

An important question in the neuropsychological profiles of individuals with TSC is the extent to which findings such as memory deficits are specific or secondary to other factors such as general intellectual ability, response inhibition, or processing speed. Here we studied IQ-matched samples, thus controlling experimentally for effects of general ability. We also reported that there were no significant group differences on tasks of response inhibition or processing speed. It is important to note that the discussion so far has concentrated on group differences. However, there was an increased variability of performance rates in the TSC group on all measures of both inhibition and processing speed, and, in both domains, there were some individuals who were significantly impaired, with scores at or below the 5th percentile of performance by the healthy comparison subjects. In short, findings suggest that the profile of working memory deficits and impaired recall but intact recognition is not attributable to these general factors. However, it seems likely that, for a subset of subjects, inhibition or information processing deficits may be a factor in poor performance on memory measures.

Methodological Issues

Some methodological issues deserve comment. First, we note that several of our patients had a history of epilepsy, which may be associated with brain structural changes in its own right. However, our results indicate that gray matter deficits could not be solely attributed either to the effects of seizures or to antiepileptic medication as patients with TSC and no history of seizures still showed significant gray matter deficits in the same subcortical regions as those with a history of epilepsy and antiepileptic medication. Second, concerns regarding the validity of between-group differences defined by computational morphometry and their possible sensitivity to imperfect registration of regional boundaries (Bookstein 2001) have been addressed by the complementary analysis of gray matter differences at a regional level of analysis, using a prior anatomically parcellated template image (Schmahmann and others 1999; Tzourio-Mazoyer and others 2002). These volumetric results substantially corroborated those obtained by computational morphometry in terms of key deficits in thalamus, basal ganglia, and cerebellum. Third, we acknowledge that the sample size is modest (N = 50 for between-group comparisons) and is reduced further for some analyses, for example, the within-group correlational analyses of the relationship between regional gray matter volumes and classical lesion load in the patient group (N = 25). Under these circumstances, we can expect some degree of type 2 error; largely for this reason, we have highlighted some trend correlations (0.05 < P < 0.1) that might be significant in a larger study. Fourth, we have studied a group of individuals with TSC who have normal intellectual abilities (mean IQ = 117). About 55% of individuals with TSC are known to have normal global abilities, and our results are likely to be fairly representative of this group. Extending this approach to include the remaining 45% of individuals may help to shed some light on possible structural determinants of global abilities in TSC. Fifth, our group of patients was not genotyped and may have been heterogeneous in terms of underlying genetic mutations. It is an interesting question for future research to explore possible endophenotypic differences in brain structure between individuals with mutations of TSC1 and TSC2 genes.

Supplementary Material

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.

KR was supported by the Medical Research Council (UK) and the Tuberous Sclerosis Association; MRI data acquisition was generously supported by a grant to PFB from the Tuberous Sclerosis Association. Software development was supported by a Human Brain Project grant from the National Institute of Biomedical Imaging and Bioengineering and the National Institute of Mental Health. Thanks to PC Fletcher for the use of long-term memory tasks and input into psychological study design. Conflict of Interest: None declared.

References

Alexander
GE
DeLong
MR
Strick
PL
Parallel organization of functionally segregated circuits linking basal ganglia and cortex
Annu Rev Neurosci
 , 
1986
, vol. 
9
 (pg. 
357
-
381
)
Au
KS
Williams
AT
Gambello
MJ
Northrup
H
Molecular genetic basis of tuberous sclerosis complex: from bench to bedside
J Child Neurol
 , 
2004
, vol. 
19
 (pg. 
699
-
709
)
Bolton
PF
Griffiths
PD
Association of tuberous sclerosis of temporal lobes with autism and atypical autism
Lancet
 , 
1997
, vol. 
349
 (pg. 
392
-
395
)
Bolton
PF
Park
RJ
Higgins
JN
Griffiths
PD
Pickles
A
Neuro-epileptic determinants of autism spectrum disorders in tuberous sclerosis complex
Brain
 , 
2002
, vol. 
125
 (pg. 
1247
-
1255
)
Bookstein
FL
“Voxel-based morphometry” should not be used with imperfectly registered images
Neuroimage
 , 
2001
, vol. 
14
 (pg. 
1454
-
1462
)
Braffman
BH
Bilaniuk
LT
Naidich
TP
Altman
NR
Post
MJ
Quencer
RM
Zimmerman
RA
Brody
BA
MR imaging of tuberous sclerosis: pathogenesis of this phakomatosis, use of gadopentetate dimeglumine, and literature review
Radiology
 , 
1992
, vol. 
183
 (pg. 
227
-
238
)
Brammer
MJ
Bullmore
ET
Simmons
A
Williams
SC
Grasby
PM
Howard
RJ
Woodruff
PW
Rabe-Hesketh
S
Generic brain activation mapping in functional magnetic resonance imaging: a nonparametric approach
Magn Reson Imaging
 , 
1997
, vol. 
15
 (pg. 
763
-
770
)
Bullmore
ET
Suckling
J
Overmeyer
S
Rabe-Hesketh
S
Taylor
E
Brammer
MJ
Global, voxel, and cluster tests, by theory and permutation, for a difference between two groups of structural MR images of the brain
IEEE Trans Med Imaging
 , 
1999
, vol. 
18
 (pg. 
32
-
42
)
Craik
FIM
Lockhart
RS
Levels of processing: a framework for memory research
J Verb Learn Verb Behav
 , 
1972
, vol. 
11
 (pg. 
671
-
684
)
de Vries
P
Humphrey
A
McCartney
D
Prather
P
Bolton
P
Hunt
A
Assel
M
Chugani
D
Creighton
L
Craig Elias
C
Franz
D
Hubbeling
A
King
B
Mastbaum
C
Medcalf
J
Mulder
J
Ridler
K
Rothberg
B
Santosh
P
Whittemore
V
Consensus clinical guidelines for the assessment of cognitive and behavioural problems in tuberous sclerosis
Eur Child Adolesc Psychiatry
 , 
2005
, vol. 
14
 (pg. 
183
-
190
)
Jr
DiMarioFJ
Brain abnormalities in tuberous sclerosis complex
J Child Neurol
 , 
2004
, vol. 
19
 (pg. 
650
-
657
)
Fletcher
PC
Stephenson
CM
Carpenter
TA
Donovan
T
Bullmorel
ET
Regional brain activations predicting subsequent memory success: an event-related fMRI study of the influence of encoding tasks
Cortex
 , 
2003
, vol. 
39
 (pg. 
1009
-
1026
)
Fowler
KS
Conway
EL
Semple
JM
Louis
WJ
Computerized delayed matching to sample and paired associate performance in the early detection of dementia
Appl Neuropsychol
 , 
1995
, vol. 
2
 (pg. 
71
-
78
)
Gomez
MR
Gomez
MR
Neurologic and psychiatric features
Tuberous sclerosis
 , 
1988
New York
Raven Press
(pg. 
21
-
36
)
Goodman
M
Lamm
SH
Engel
A
Shepherd
CW
Houser
OW
Gomez
MR
Cortical tuber count: a biomarker indicating neurologic severity of tuberous sclerosis complex
J Child Neurol
 , 
1997
, vol. 
12
 (pg. 
85
-
90
)
Green
J
McDonald
WM
Vitek
JL
Haber
M
Barnhart
H
Bakay
RA
Evatt
M
Freeman
A
Wahlay
N
Triche
S
Sirockman
B
DeLong
MR
Neuropsychological and psychiatric sequelae of pallidotomy for PD: clinical trial findings
Neurology
 , 
2002
, vol. 
58
 (pg. 
858
-
865
)
Han
S
Santos
TM
Puga
A
Roy
J
Thiele
EA
McCollin
M
Stemmer-Rachamimov
A
Ramesh
V
Phosphorylation of tuberin as a novel mechanism for somatic inactivation of the tuberous sclerosis complex proteins in brain lesions
Cancer Res
 , 
2004
, vol. 
64
 (pg. 
812
-
816
)
Hunt
A
Dennis
J
Psychiatric disorder among children with tuberous sclerosis
Dev Med Child Neurol
 , 
1987
, vol. 
29
 (pg. 
190
-
198
)
Jambaque
I
Chiron
C
Dumas
C
Mumford
J
Dulac
O
Mental and behavioural outcome of infantile epilepsy treated by vigabatrin in tuberous sclerosis patients
Epilepsy Res
 , 
2000
, vol. 
38
 (pg. 
151
-
160
)
Jambaque
I
Cusmai
R
Curatolo
P
Cortesi
F
Perrot
C
Dulac
O
Neuropsychological aspects of tuberous sclerosis in relation to epilepsy and MRI findings
Dev Med Child Neurol
 , 
1991
, vol. 
33
 (pg. 
698
-
705
)
Joyce
EM
Robbins
TW
Frontal lobe function in Korsakoff and non-Korsakoff alcoholics: planning and spatial working memory
Neuropsychologia
 , 
1991
, vol. 
29
 (pg. 
709
-
723
)
Kelleher
RJ
III
Govindarajan
A
Tonegawa
S
Translational regulatory mechanisms in persistent forms of synaptic plasticity
Neuron
 , 
2004
, vol. 
44
 (pg. 
59
-
73
)
Kolb
B
Cioe
J
Muirhead
D
Cerebral morphology and functional sparing after prenatal frontal cortex lesions in rats
Behav Brain Res
 , 
1998
, vol. 
91
 (pg. 
143
-
155
)
Kolb
B
Sutherland
RJ
Whisaw
IQ
Neonatal hemidecortication or frontal cortex ablation produces similar behavioral sparing but opposite effects on morphogenesis of remaining cortex
Behav Neurosci
 , 
1983
, vol. 
97
 (pg. 
154
-
158
)
Kopelman
MD
Disorders of memory
Brain
 , 
2002
, vol. 
125
 (pg. 
2152
-
2190
)
LaBerge
D
Attentional control: brief and prolonged
Psychol Res
 , 
2002
, vol. 
66
 (pg. 
220
-
233
)
Lawrence
AD
Hodges
JR
Rosser
AE
Kershaw
A
Ffrench-Constant
C
Rubinsztein
DC
Robbins
TW
Sahakian
BJ
Evidence for specific cognitive deficits in preclinical Huntington's disease
Brain
 , 
1998
, vol. 
121
 
Pt 7
(pg. 
1329
-
1341
)
Lawrence
AD
Watkins
LH
Sahakian
BJ
Hodges
JR
Robbins
TW
Visual object and visuospatial cognition in Huntington's disease: implications for information processing in corticostriatal circuits
Brain
 , 
2000
, vol. 
123
 
Pt 7
(pg. 
1349
-
1364
)
Lezak
MD
Neuropsychological assessment
1983
2nd ed
Oxford University Press
Nellist
M
Brook-Carter
PT
Connor
JM
Kwiatkowski
DJ
Johnson
P
Sampson
JR
Identification of markers flanking the tuberous sclerosis locus on chromosome 9 (TSC1)
J Med Genet
 , 
1993
, vol. 
30
 (pg. 
224
-
227
)
Nelson
H
Willison
J
National adult reading test (NART): test manual (part II)
 , 
1991
Windsor, UK
NFER-Nelson
O'Callaghan
FJ
Harris
T
Joinson
C
Bolton
P
Noakes
M
Presdee
D
Renowden
S
Shiell
A
Martyn
CN
Osborne
JP
The relation of infantile spasms, tubers, and intelligence in tuberous sclerosis complex
Arch Dis Child
 , 
2004
, vol. 
89
 (pg. 
530
-
533
)
Osborne
JP
Fryer
A
Webb
D
Epidemiology of tuberous sclerosis
Ann N Y Acad Sci
 , 
1991
, vol. 
615
 (pg. 
125
-
127
)
Osterrieth
PA
Le test de copie d‘une figure complex: contribution A l’etude de la perception et de la memoire
Arch Psychol
 , 
1944
, vol. 
30
 (pg. 
286
-
356
)
Owen
AM
Beksinska
M
James
M
Leigh
PN
Summers
BA
Marsden
CD
Quinn
NP
Sahakian
BJ
Robbins
TW
Visuospatial memory deficits at different stages of Parkinson's disease
Neuropsychologia
 , 
1993
, vol. 
31
 (pg. 
627
-
644
)
Owen
AM
Downes
JJ
Sahakian
BJ
Polkey
CE
Robbins
TW
Planning and spatial working memory following frontal lobe lesions in man
Neuropsychologia
 , 
1990
, vol. 
28
 (pg. 
1021
-
1034
)
Owen
AM
Doyon
J
Dagher
A
Sadikot
A
Evans
AC
Abnormal basal ganglia outflow in Parkinson's disease identified with PET. Implications for higher cortical functions
Brain
 , 
1998
, vol. 
121
 
Pt 5
(pg. 
949
-
965
)
Owen
AM
Iddon
JL
Hodges
JR
Summers
BA
Robbins
TW
Spatial and non-spatial working memory at different stages of Parkinson's disease
Neuropsychologia
 , 
1997
, vol. 
35
 (pg. 
519
-
532
)
Owen
AM
James
M
Leigh
PN
Summers
BA
Marsden
CD
Quinn
NP
Lange
KW
Robbins
TW
Fronto-striatal cognitive deficits at different stages of Parkinson's disease
Brain
 , 
1992
, vol. 
115
 
Pt 6
(pg. 
1727
-
1751
)
Prather
P
de Vries
PJ
Behavioral and cognitive aspects of tuberous sclerosis complex
J Child Neurol
 , 
2004
, vol. 
19
 (pg. 
666
-
674
)
Rey
A
L‘examen psychologque dans les cas d’encephalopathic traumatique
Arch Psychol
 , 
1941
, vol. 
28
 (pg. 
286
-
340
)
Ridler
K
Bullmore
ET
De Vries
PJ
Suckling
J
Barker
GJ
Meara
SJ
Williams
SC
Bolton
PF
Widespread anatomical abnormalities of grey and white matter structure in tuberous sclerosis
Psychol Med
 , 
2001
, vol. 
31
 (pg. 
1437
-
1446
)
Ridler
K
Suckling
J
Higgins
N
Bolton
PF
Bullmore
ET
Standardised whole brain mapping of tubers and subependymal nodules in tuberous sclerosis complex
J Child Neurol
 , 
2004
, vol. 
19
 (pg. 
658
-
665
)
Roach
ES
Gomez
MR
Northrup
H
Tuberous sclerosis complex consensus conference: revised clinical diagnostic criteria
J Child Neurol
 , 
1998
, vol. 
13
 (pg. 
624
-
628
)
Robbins
TW
Owen
AM
Sahakian
BJ
Ron
MA
David
AS
The neuropsychology of basal ganglia disorders
Disorders of the brain and mind
 , 
1999
Cambridge
Cambridge University Press
(pg. 
57
-
83
)
Rubia
K
Taylor
E
Smith
AB
Oksanen
H
Overmeyer
S
Newman
S
Oksannen
H
Neuropsychological analyses of impulsiveness in childhood hyperactivity
Br J Psychiatry
 , 
2001
, vol. 
179
 (pg. 
138
-
143
)
Rubin
DC
Frontal-striatal circuits in cognitive Aging: evidence for caudate involvement
Aging Neuropsychol Cogn
 , 
1999
, vol. 
6
 (pg. 
241
-
259
)
Sahakian
BJ
Morris
RG
Evenden
JL
Heald
A
Levy
R
Philpot
M
Robbins
TW
A comparative study of visuospatial memory and learning in Alzheimer-type dementia and Parkinson's disease
Brain
 , 
1988
, vol. 
111
 (pg. 
695
-
718
)
Scheithauer
BW
Reagan
TJ
Gomez
MR
Neuropathology
Tuberous sclerosis complex
 , 
1999
3rd ed
Oxford
Oxford University Press
(pg. 
101
-
144
)
Schmahmann
JD
Doyon
J
McDonald
D
Holmes
C
Lavoie
K
Hurwitz
AS
Kabani
N
Toga
A
Evans
A
Petrides
M
Three-dimensional MRI atlas of the human cerebellum in proportional stereotaxic space
Neuroimage
 , 
1999
, vol. 
10
 (pg. 
233
-
260
)
Seri
S
Cerquiglini
A
Pisani
F
Curatolo
P
Autism in tuberous sclerosis: evoked potential evidence for a deficit in auditory sensory processing
Clin Neurophysiol
 , 
1999
, vol. 
110
 (pg. 
1825
-
1830
)
Shepherd
CW
Houser
OW
Gomez
MR
MR findings in tuberous sclerosis complex and correlation with seizure development and mental impairment
Am J Neuroradiol
 , 
1995
, vol. 
16
 (pg. 
149
-
155
)
Sternberg
S
Discovery of processing stages: extensions of Donders' method
Acta Psychol
 , 
1969
, vol. 
30
 (pg. 
276
-
315
)
Sternberg
S
Memory-scanning: mental processes revealed by reaction-time experiments
Am Sci
 , 
1969
, vol. 
57
 (pg. 
421
-
457
)
Sternberg
S
Memory scanning: new findings and current controversies
Q J Exp Psychol
 , 
1975
, vol. 
27
 (pg. 
1
-
32
)
Suckling
J
Brammer
MJ
Lingford-Hughes
A
Bullmore
ET
Removal of extracerebral tissues in dual-echo magnetic resonance images via linear scale-space features
Magn Reson Imaging
 , 
1999
, vol. 
17
 (pg. 
247
-
256
)
Suckling
J
Bullmore
E
Permutation tests for factorially designed neuroimaging experiments
Hum Brain Mapp
 , 
2004
, vol. 
22
 (pg. 
193
-
205
)
Suckling
J
Sigmundsson
T
Greenwood
K
Bullmore
ET
A modified fuzzy clustering algorithm for operator independent brain tissue classification of dual echo MR images
Magn Reson Imaging
 , 
1999
, vol. 
17
 (pg. 
1065
-
1076
)
Talairach
J
Tournoux
P
Co-planar stereotaxic atlas of the human brain
1988
New York
Thieme Medical Publishers
Tzourio-Mazoyer
N
Landeau
B
Papathanassiou
D
Crivello
F
Etard
O
Delcroix
N
Mazoyer
B
Joliot
M
Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain
Neuroimage
 , 
2002
, vol. 
15
 (pg. 
273
-
289
)
Van der Werf
YD
Scheltens
P
Lindeboom
J
Witter
MP
Uylings
HB
Jolles
J
Deficits of memory, executive functioning and attention following infarction in the thalamus; a study of 22 cases with localised lesions
Neuropsychologia
 , 
2003
, vol. 
41
 (pg. 
1330
-
1344
)
van Slegtenhorst
M
de Hoogt
R
Hermans
C
Nellist
M
Janssen
B
Verhoef
S
Lindhout
D
van den Ouweland
A
Halley
D
Young
J
Burley
M
Jeremiah
S
Woodward
K
Nahmias
J
Fox
M
Ekong
R
Osborne
J
Wolfe
J
Povey
S
Snell
RG
Cheadle
JP
Jones
AC
Tachataki
M
Ravine
D
Sampson
JR
Reeve
MP
Richardson
P
Wilmer
F
Munro
C
Hawkins
TL
Sepp
T
Ali
JBM
Ward
S
Green
AJ
Yates
JRW
Kwiatkowska
J
Henske
EP
Short
MP
Haines
JH
Jozwiak
S
Kwiatkowski
DJ
Identification of the tuberous sclerosis gene TSC1 on chromosome 9q34
Science
 , 
1997
, vol. 
277
 (pg. 
805
-
808
)
Warrington
EK
The Camden memory tests manual
 , 
1996
Hove, UK
Psychology Press
Watkins
LH
Rogers
RD
Lawrence
AD
Sahakian
BJ
Rosser
AE
Robbins
TW
Impaired planning but intact decision making in early Huntington's disease: implications for specific fronto-striatal pathology
Neuropsychologia
 , 
2000
, vol. 
38
 (pg. 
1112
-
1125
)
Weber
AM
Egelhoff
JC
McKellop
JM
Franz
DN
Autism and the cerebellum: evidence from tuberous sclerosis
J Autism Dev Disord
 , 
2000
, vol. 
30
 (pg. 
511
-
517
)
Wechsler
D
Wechsler abbreviated scale of intelligence (WASI)
 , 
1999
San Antonio, TX
Psychological Corporation
Weschler
D
The Wechsler memory scale-III UK manual
 , 
1997
San Antonio, TX
Psychological Corporation
Yorns
W
Burleson
JA
DiMario
FJ
Brain morphometry in tuberous sclerosis complex
J Investig Med
 , 
2001
, vol. 
49
 pg. 
221A
 
Zoppelt
D
Koch
B
Schwarz
M
Daum
I
Involvement of the mediodorsal thalamic nucleus in mediating recollection and familiarity
Neuropsychologia
 , 
2003
, vol. 
41
 (pg. 
1160
-
1170
)