Книги по МРТ КТ на английском языке / Functional Neuroimaging in Child Psychiatry Ernst 1 ed 2000

pushing our imaging techniques to progressively younger age groups. It could also involve studying unaVected but genetically vulnerable family members to identify trait or risk factors rather than state markers of the CNS functioning that are speciWc to TS.

Age-related changes

As yet, it has not been possible to control adequately for the eVects of simple age-related changes in measures followed by imaging studies. It is likely, for instance, that metabolism, blood Xow, and dopaminergic transmission in the basal ganglia all demonstrate their own speciWc developmental proWles (Shaywitz et al., 1980; Riddle et al., 1986; Seeman et al., 1987). Inadequately controlling for these age-related changes will produce variable and conXicting Wndings. Although individual matching of patient and subject groups helps to address this problem, this is still not a satisfactory substitute for knowing the developmental characteristics of the experimental measure. The same is true for sex-speciWc eVects, which assumes greater importance in TS because of the much higher prevalence of TS in males than in females. The exclusive study of adults with these disorders also introduces important ascertainment biases that may skew Wndings and severely limit how the results generalize to the larger population of patients who have TS. Follow-up studies of TS, for instance, suggest that approximately one third of children with TS will be symptom-free as adults, and another one third will have only mild symptoms that do not require clinical attention (Erenberg et al., 1987; Bruun, 1988; Leckman et al., 1998). Therefore, adults who have symptoms severe enough to be ascertained clinically are unusual representatives of all subjects who have a lifetime diagnosis of TS. Either they have a relatively unusual neurobiologic substrate that determines tic persistence (dopaminergic hyperinnervation of their striatum, for instance), or they may have a second disease process (an early dementing illness, for example) that releases the tic symptoms from the inhibitory inXuences that typically come into play in late adolescence and early childhood. These adult subjects, moreover, are very likely to have received psychotropic medications in their lifetime, either for the treatment of tics or for comorbid OCD, ADHD, depression, or anxiety (Goetz et al., 1992; Park et al., 1993), which are likely to aVect brain structure and function and confound study results. Therefore, although studies of adults with TS are important for hypothesis generation and for studies of more severe outcome, they are nevertheless fraught with methodologic limitations that severely limit our ability to interpret and generalize the

Wndings. Longitudinal imaging studies of children who either have TS or who are at risk for developing TS are, therefore, imperative to improve understanding of this disorder.

Comorbid illnesses

Another confound in functional imaging studies of TS is the high rate of comorbid illness associated with this disorder: OCD aVects as many as 60% and ADHD aVects approximately 50% of TS clinical populations, while aVective illness and anxiety disorders aVect even more patients. Despite these extraordinarily high rates of comorbid illness, TS functional imaging studies have rarely assessed the potential presence of these disorders in the patients studied, let alone examined their Wndings to determine whether these comorbidities were responsible for the Wndings.

Structure_function relationships

Another potential confound that no functional imaging study in TS has yet addressed is the possibility that group diVerences in functional measures are largely driven by underlying diVerences between groups in brain structure. The obvious example in TS is basal ganglia hypometabolism. It is possible, for instance, that the frequently reported reduction in metabolism and blood Xow to the basal ganglia is simply a consequence of the previously reported volume reductions in those same areas. Similarly, the normal relationship between morphologic characteristics of the basal ganglia and their neurotransmitter and neuroreceptor proWles is unknown. It is possible that volume reductions in the basal ganglia may account for the greater density and, therefore, elevated levels of presynaptic transporter, postsynaptic dopamine D2 receptor, and DOPA decarboxylase enzyme that have been reported in TS. Aristotle posited that structure determines function (Kirwan, 1971). Although the direction of causality may not be this clear in the brain, it seems likely that measures of structure and function are at the very least correlated. Future studies should assess these relationships in normal subjects to improve our ability to interpret deviations from those relationships in disease states. Longitudinal studies will also be able to assess how changes in structure and function are related and thereby will allow us to understand better those relationships in temporal cross-section. At the very least, structural indices should be used as covariates in the testing of statistical signiWcance of functional indices. This argues strongly for multimodal imaging studies in TS.

Normalization of metabolism and blood Xow

One last potential confound for many imaging studies of TS is the need for normalization of regional metabolic rates. Most metabolism and blood Xow studies so far have used measures of blood Xow or metabolism in either the visual or the cerebellar cortex as the basis for normalization. The investigators argue that these regions are unlikely to be involved in TS pathophysiology and should, therefore, serve as an adequate reference by which to quantify measures in regions of interest (ROIs) of more central interest to the study. However, activations seen during tic suppression are also likely to be present in metabolism and blood Xow studies of TS and this would call into question the use of visual cortex for normalization purposes. We attribute the activations in visual association cortex during tic suppression to the strategy that many subjects use to help to suppress their tics during the scanning session, that of reallocating attentional resources to visual stimuli. This attentional reallocation is thought to ªtuneº neurons in the visual cortices to visual stimuli as a general mechanism for enhancing task performance. Tuning of the neurons in visual cortex increases their activity and blood Xow (Moran and Desimone, 1985; Spitzer, 1988; Petersen et al., 1990; Heinze et al., 1994; Gratton, 1997; Shulman et al., 1997a). Normalizing regional metabolism and blood Xow by the values in a region that is itself hypermetabolic relative to normal controls will produce false reports of reduced metabolism and blood Xow in other ROIs in the TS patient group. The cerebellum also seems to be a dubious choice by which to normalize, since it is increasingly identiWed as active in attentional tasks. It also has well-known connections with other motor systems that may be abnormal in TS. These anatomic connections and functional characteristics of the cerebellum may produce diVerences in its metabolism and blood Xow in TS subjects compared with normal controls.

Candidate neural systems in the etiology of tics

Contemporary models of parallel distributed information processing recognize a general modularity in CNS organization. Rather than requiring a temporally serial processing of stimuli, these models suggest that processing in the diVerent modules occurs simultaneously in an iterative, reverberating manner. EVective information processing requires the integrity and interaction of all regional modules, without a strict one-to-one correspondence between regional location, neural computation, and overt behavior. The neural substrate of behavior, therefore,

is neither holistic nor phrenologic. Despite the general recognition that brain function is fundamentally based on parallel distributed processing, investigators still tend to localize pathologic brain functioning to relatively small, discrete brain regions. So, although mental functioning is, in part, regionally speciWed within the modules of the CNS, in our investigations of the neural substrate of tic symptoms we must remember that regional functioning also depends on brain function at spatially distributed sites. In fact, it seems probable at face value that the product of the putative TS vulnerability genes, rather than producing a discrete, localized lesion in the brain is more likely to be expressed and, therefore, more likely to produce disordered functioning within a distributed neural circuit.

The leading candidate system in the search for the neural substrate of TS (along with the substrates of OCD and ADHD comorbidity) is the circuit that loops between cortical and subcortical brain regions (Leckman et al., 1991). Named by the successive structural components within the loops, it is referred to as the cortico±striato± thalamo±cortical (CSTC) circuitry. The hypothesized function of this circuitry is consistent with the frequent cooccurrence of TS, OCD, and ADHD. Involvement of motor portions of the circuit may subserve tics and hyperkinesia. Dysfunction in portions of the circuitry that subserve higher cognitive and ideational processes may produce the premonitory urges of TS and the obsessions of OCD. The involvement of inhibitory portions of the circuit may produce the diYculties with inhibiting inappropriate, impulsive behaviors that is seen in ADHD. In short, the CSTC circuit comprises motor, associational, and inhibitory neural systems that are likely to subserve the symptoms of TS-related illnesses.

CSTC pathways

Multiple, partially overlapping, but largely parallel circuits compose the CSTC circuitry. By directing information from the cerebral cortex to the subcortex, and then back again to speciWc cortical regions, these circuits form multiple cortical±subcortical loops (Fig. 14.1, p. 242). The loops are considered parallel by virtue of their microscopic segregation from other circuits that course through the same macroscopic structures, the basal ganglia and thalamus. Multiple anatomically and functionally connected cortical regions provide input to particular subcortical portions of the circuit; these subcortical portions then project back to a limited subset of the cortical regions initially contributing to the circuit's input.

The number of anatomically discrete subdivisions of these circuits is controversial. Nevertheless, CSTC circuitry

Sources: Alexander et al., 1990; Parent and Hazrati, 1995a,b.

seems to include four major subcomponents: those loops originating from and projecting back to the sensorimotor, orbitofrontal, association, or temporolimbic cortices (Table 14.1) (Alexander et al., 1990; Parent and Hazrati, 1995a,b). Each of these expanses of cortex projects either to the caudate or putamen, which together comprise the striatum. The projections to the striatum are organized in parasagittally elongated, somatotopic domains. The information leaves these basal ganglia regions primarily through the internal segment of the globus pallidus and its brainstem counterpart, the substantia nigra pars reticulata. The loops then ascend to the thalamus and Wnally back again to the cortex. This organization is referred to as the direct pathway, which stands in contrast to the indirect pathway. In this latter circuit, the striatum projects Wrst to the external segment of the globus pallidus and then to the reticular portion of the thalamus, as well as to the subthalamic nucleus and the internal segment of the globus pallidus (see Fig. 14.3a, below). This indirect pathway is considered an intrinsic modulator of activity in the direct pathway, since the reticular thalamic nucleus exerts powerful gamma-aminobutyric acid (GABA)-mediated inhibitory inXuences on circuit elements of the direct pathway in other nuclei of the thalamus (Parent and Hazrati, 1995a). The projection from the external to the internal segment of the globus pallidus, moreover, may similarly modulate direct pathway activity in the internal segment of the globus pallidus.

Subcortical nuclei

Single-cell recordings of the functioning of normal basal ganglia neurons provide circumstantial evidence that these structures may be involved in TS pathophysiology. The recordings indicate that activity in individual neurons in the putamen is correlated with speciWc aspects of limb movement, including velocity and direction. Motor portions of the CSTC circuitry, therefore, appear to be implicated in controlling the direction of movement as well as scaling its force and speed. Tics show features of normal behavioral repertoires but are executed more frequently, rapidly, and forcefully than their normal behavioral counterparts. It is possible, therefore, that locally disinhibited functioning of CSTC circuits within the basal ganglia could produce tic-like phenomena. Further evidence for this hypothesis comes from chemical or electrical stimulation of the basal ganglia, which produce tic-like stereotypies in animal and human subjects (McLean and Delgado, 1953; Baldwin et al., 1954; Alexander and Delong, 1985; Kelley et al., 1988). Conversely, the clinical eYcacy of neuroleptic medications is thought to derive from blocking dopaminergic inXuences on the basal ganglia from nigrostriatal neurons projecting from the brainstem. Similarly, dopamine-depleting agents, such as '-methyl-p-tyrosine and tetrabenazine, have been reported to suppress tic symptoms in some patients (Sweet et al., 1976; Jankovic et

al., 1984), while L-DOPA and stimulant medications, which are dopaminergic agonists, can with varying reliability exacerbate tic symptoms (Golden, 1974; Lowe et al., 1982; Gadow et al., 1995). Finally, human studies and autopsy structural imaging studies suggest the importance of basal ganglia in TS pathophysiology. These studies have been reasonably consistent in demonstrating reduced volumes and abnormal asymmetries in the putamen and globus pallidus nuclei of children and adults with TS (Balthazar, 1956; Richardson, 1982; Peterson et al., 1993; Singer et al., 1993; Castellanos et al., 1994).

Thalamic portions of the CSTC circuit have been implicated in TS largely in the context of tic symptom changes resulting from either space-occupying or neurosurgical lesions (Leckman et al., 1993; Rauch et al., 1995; Peterson et al., 1996). While irritative space-occupying lesions in ventral thalamic nuclei may increase tic symptoms (Peterson 1996), surgical lesions to ventral, medial, and intralaminar thalamic nuceli may attenuate symptoms in some patients (Cooper, 1969; Hassler and Dieckmann, 1970; de Divitiis et al., 1977; Korzen et al., 1991; Rauch et al., 1995). Further evidence for thalamic involvement comes from intraoperative microelectrode stimulation of the ventral intermediate and ventral oralis posterior thalamic nuclei, which produces sensations similar to the premonitory urges that occur in patients with TS prior to the performance of their tics (Tasker and Dostrovsky, 1993).

Sensorimotor pathways

The sensorimotor pathways originate in part from and project back to the supplementary motor area (Table 14.1). Electrical stimulation of the supplementary motor area produces, in some patients, complex movements, vocalizations, and speech arrest, in addition to sensations that in some patients are described as an ªurgeº to move the somatotopically stimulated contralateral body region (Fried et al., 1991; Lim et al., 1994). These urges to move are reminiscent of the premonitory ªurgesº that adolescents and adults with TS describe prior to the performance of their tics (Leckman et al., 1993). Electroencephalographic (EEG) recordings have demonstrated potential changes preceding normal voluntary movements that are localized to supplementary motor areas bilaterally. The Wrst EEG study of this kind in six patients with TS was unable to detect premotor movement potentials associated with tics, although they were detected when patients mimicked their own tics (Obeso et al., 1981). Similar results were seen in a second study of Wve TS patients using a similar experimental design, although premotor movement potentials were discerned before the tics in two of the subjects (Karp

et al., 1996). These results suggest that the neural circuits that produce tics may involve the supplementary motor areas in some patients.

Orbitofrontal pathways

The orbitofrontal cortex is interconnected with the anterior cingulate and other limbic structures. The orbitofrontal cortex may contribute to the capacity for such tasks as successive discrimination, go-no-go and response reversal tasks (Rosvold and Mishkin, 1961; Drewe, 1975; Diamond and Goldman-Rakic, 1989). Lesions of this region appear to interfere with the ability to generate the internal cues that are needed to guide goal-directed behaviors (Goldman-Rakic, 1987). Lesions of the orbitofrontal cortex interfere with an animal's capacity to make appropriate changes in its behavioral set (Divac et al., 1967; Iverson and Mishkin, 1970; Mishkin and Manning, 1978); they typically disrupt the regulation of aVect, and they can produce impulsive, socially inappropriate behaviors (Luria, 1980).

Temporolimbic pathways

The limbic system consists of the amygdala and hippocampus in the temporal lobe, the cingulum, the caudate nucleus and other basal ganglia structures in the subcortex, the hypothalamus and periaqueductal gray matter in the brainstem, and connections with the associated frontal cortex. Despite the hypothesized involvement of the temporal lobe in TS pathophysiology, neurobiologic investigations of these regions are remarkably few (Jadresic, 1992; Peterson et al., 1992). The sexual and aggressive content of many complex motor and vocal tics, and of many obsessions and compulsions, also suggest the involvement of the amygdala and related circuitry. Moreover, steroid hormone receptors densely populate the human amygdala and related portions of the limbic circuitry and may mediate sex-speciWc diVerences in the prevalence of tic symptoms in the general population, as well as the sexual and aggressive content of tics in some individuals. A neuroimaging study of T2 relaxation times found evidence of abnormal tissue characteristics in the left and right amygdalae of adults with TS compared with those in normal controls (Peterson et al., 1994).

Cingulate cortex

The cingulate cortex is a heterogeneous structure that probably is a component in most of the major CSTC pathways: the supplementary motor area, orbitofrontal cortex

248B. S. Peterson and P. Thomas

and temporolimbic circuits. The cingulate receives input from the thalamus, amygdala, and motor cortex to its anterior region. It sends projections primarily to motor cortex, striatum, periaqueductal gray, and brainstem motor nuclei (Vogt et al., 1992; Bates and Goldman-Rakic, 1993). The anterior cingulate cortex is thought to subserve, among other things, attentional and executive functioning (Pardo et al., 1990; Peterson et al., 1999), and impairment of these functions probably account for some of the inattention, impulsivity, and motoric dyscontrol that commonly characterizes the TS phenotype. Electrical stimulation of the anterior cingulate in humans can produce semi-voluntary movements resembling complex motor tics (Talairach et al., 1973), and there are reports of anterior cingulotomies alleviating tic symptoms in some patients, although these were uncontrolled studies (Kurlan et al., 1990; Robertson et al., 1990).

Brainstem neuromodulators and motor nuclei

Finally, the brainstem has been implicated as one possible component of the neural substrate of TS. The various brainstem centers that generate catecholamine and indoleamine neurotransmitters are thought to play important neuromodulatory roles in the disorder. These centers project to the basal ganglia and may thereby inXuence tic symptoms. One suspected modulator of tic symptoms, for instance, includes the dopaminergic aVerent systems ascending to the basal ganglia from the midbrain substantia nigra. In addition, noradrenergic projections from the locus ceruleus modulate midbrain dopamine and frontal inhibitory centers; these eVects indirectly inXuence basal ganglia function and may thereby help to suppress tic-related behaviors. Noradrenergic systems may also mediate the exquisite stress responsivity seen in the disorder (Chappell et al., 1994). Aside from these neurotransmitter systems, limbic regions presumably have a modulating eVect via direct aVerent projections to the ventral striatum, and indirectly through projections to midbrain dopaminergic neurons. The limbic regions are hypothesized to be important in modulating the sexual and aggressive content of tic symptoms (Jadresic, 1992; Peterson et al., 1992).

Descending projections from the motor cortex and basal ganglia nuclei regulate the motor discharge of brainstem interneurons and motor nuclei. Abnormalities associated with this regulation could produce motor and vocal tics having the somatotopy characteristics of TS. A study supporting this hypothesis found an increased amplitude of one component of the blink reXex in patients with TS compared with normal controls, suggesting the presence of

increased excitability of TS brainstem regulatory interneurons (Smith and Lees, 1989). Abnormalities in these regulatory interneurons could aVect functioning of brainstem motor nuclei, which innervate the musculature most commonly aVected by tics. These nuclei innervate the muscles of the face (motor nucleus of cranial nerve VII in the midpons), neck and shoulders (spinal accessory cranial nerve XI in the medulla), larynx and pharynx (nucleus ambiguus, a portion of the vagus cranial nerve X in the medulla), tongue (hypoglossal nucleus of cranial nerve XII in the medulla), and diaphragm (descending brainstem control of high cervical spinal cord). All of these brainstem nuclei are situated closely to one another and could easily be involved in a relatively discrete pathologic process in the brainstem. Consistent with this hypothesis, the neural basis for conscious control of musculature in the extremities resides largely in the motor cortex, and the extremities are much less commonly aVected by tics than are the facial, vocal, and diaphragmatic musculature.

The basal ganglia and descending dopaminergic systems from the midbrain are also known to inXuence the activity of brainstem sensory pathways, another possible contributor to tic symptoms. For instance, disturbing the relay of proprioceptive information from the face to the mesencephalic nucleus of cranial nerveV in the pons could produce the clonus-like activity of muscle groups aVected by tics (Lawrence and Redmond, 1985; Larumbe et al., 1993).

Functional neuroimaging studies

The neural systems described above have been identiWed as candidates for the substrate of tic-related behaviors primarily through consideration of a vast preclinical literature on the normal structural and functional organization of the CNS, TS-related phenomenology, and human lesion studies. One goal of our review will be to assess which of these candidate systems are most strongly implicated in TS pathophysiology.

Resting metabolism and blood Xow studies

In vivo positron emission tomography (PET) and single photon emission computed tomography (SPECT) studies generally report a hypometabolism in cortical and subcortical brain regions of patients who have TS. These studies, however, typically do not adequately relate the implications of these results to the current knowledge of the neural circuitry and neurophysiology of TS. More importantly, in failing to make explicit the demands of

their scanning protocol, these studies usually overlook the eVects that either spontaneous tics or the suppression of tics may have on brain metabolism (Tables 14.2 and 14.3). We will review functional imaging studies of TS in chronological order of their publication.

The Wrst preliminary Xuorodeoxyglucose (FDG) PET study of 12 adults with TS (10 males, two females; mean age 33"2 years) and an equal number of normal control subjects (eight males, four females; mean age 31"2 years) reported a 15% average reduction of absolute glucose utilization rates in the frontal, cingulate, and insular cortices and in the inferior corpus striatum (p#0.01) (Chase et al., 1986). An inverse correlation (p#0.01) was seen between the severity of tics (both vocal and motor) and glucose rates in these same areas. The authors did not specify whether instructions were given to the patients to suppress their tics during scanner uptake. Additionally, the normal controls and subjects with TS were unevenly sexmatched, which may have aVected group comparisons.

SigniWcant diVerences between patients and controls were found only at horizontal levels 8.4±8.8 cm caudal to the brain vertex. This orientation bears an approximate correspondence to the level of the anterior commissure± posterior commissure (AC±PC) line in Talairach coordinates. The absence of a standard anatomic reference scheme, however, renders impossible any accurate alignment with Talairach coordinates and hence also any accurate anatomic comparison with other studies. The investigators did not specify whether the medication-free status of the 12 untreated patients was for lifetime or only recent status, and subject handedness was not reported. Finally and most importantly, the methods employed for statistical analysis were not described.

In a report published only in abstract form, brain perfusion deWcits were also found in a hexamethyl-propyle- neamine oxime (HMPAO) SPECT study of 25 patients with TS (19 males, six females; 7±48 years old) compared with 10 normal subjects whose scans were selected from a preexisting library of normal scans (Hall et al., 1990). Hypoperfusion of the basal ganglia and thalamus was seen in the patients with TS relative to the controls. Perfusion deWcits were also seen in the frontal and temporal cortices. The poorly matched controls, whose scans were selected only by convenience, and the absence of methodologic details makes critical appraisal of the study impossible.

Another HMPAO SPECT study of 20 patients with TS (17 males, three females; mean age 34.7"12 years; 18/20 right-handed; 10 with comorbid OCD) and Wve normal controls (Wve males, three females; mean age 34.7"12.5 years; Wve right-handed) failed to detect group diVerences in basal ganglia blood Xow (George et al., 1992). Instead,

the investigators noted an increased right frontal activity compared with control subjects when this region was normalized to regional cerebral blood Xow (rCBF) in the visual cortex. Yet the small number of controls and limitations in scanner resolution (7±9mm full width half maximum (FWHM) limit the generalizability of these negative Wndings. Moreover, as our discussion of subsequent studies will make clear, the choice of the visual cortex for normalization may have been misguided in that blood Xow to the region may have been aVected by the demands of the scanning protocol. Tic behaviors during tracer uptake were not described. Other investigators conducted an HMPAO SPECT study of nine patients with TS (six males, three females; mean age 29.6"7 years) and nine individually sexand age-matched controls (Riddle et al., 1992). HMPAO uptake was reduced by 4% in the left putamen and globus pallidus of the patients with TS compared with control values, although the p-value for this group comparison was not provided. In addition, a paired t-test was used to perform between-group comparisons when an unpaired test may have been more appropriate, since matching on age and sex does not control suYciently for the biological determinants of HMPAO-assessed blood Xow. The unpaired analysis probably would have eliminated the statistical signiWcance of the 4% reduction in the rCBF to the basal ganglia.

Several case reports of HMPAO SPECT blood Xow abnormalities have been reported in TS. One, published as an abstract, is that of an 11-year-old girl who had TS and who demonstrated hypoperfusion only in the region of her left basal ganglia (Sieg et al., 1992). Despite its obvious limitations, this study is consistent with others demonstrating regional hypoperfusion in patients with TS. A second case report documents the metabolic eVects of a limbic leucotomy in a 45-year-old patient with TS and OCD (Sawle et al., 1993). This patient underwent a PET scan 15 months before and 21 months after his leucotomy. His data were compared with data from six age-matched, male normal controls. Preoperatively, the caudate appeared to be the most hypermetabolic, as was the thalamus. Postoperatively, the greatest reductions in metabolic rate occurred in the caudate and anterior cingulate. Reduction in cingulate metabolism would be expected as a consequence of the surgical lesion, which is typically placed in the anteriorand inferior-most portion of the anterior cingulate gyrus beneath the genu of the corpus callosum (Rauch et al., 1995). This lesion also typically interrupts projections from orbitofrontal cortex to caudate and thalamic nuclei, which probably accounts for the reduction in caudate metabolism.

The largest and methodologically Wnest FDG PET imaging study compared normalized metabolic rates

between 16 patients with TS (14 males, two females; mean age 33"7 years) and 16 normal controls who were matched well for age but not for sex (11 males, Wve females; mean age 34"10 years) (Braun et al., 1993). Global cerebral metabolism showed no diVerences between groups. Increased relative metabolic rates were seen in superior sensorimotor cortices of the subjects with TS. Reduced metabolism was seen in the inferior limbic-associated areas, especially in paralimbic, ventral prefrontal, striatal, and brainstem regions. The largest reductions were located generally in the left hemisphere (Fig. 14.2). The basal ganglia subregions in which metabolic activity was most reduced appeared to be the limbic-associated regions of the ventral striatum. Although typically the ventral striatum is considered to comprise the nucleus accumbens and caudate, the large size of the ROI, as well as the limited scanner resolution, make it likely that activity in this ROI derived from all basal ganglia nuclei, including the putamen and globus pallidus.

Despite the many strengths of this study, it has several important limitations. Four of the patients with TS in this study were on neuroleptics and three were on other medications within 8 weeks of their scans. The remaining nine patients had a history of medication of some type. It is possible, therefore, that prior medication exposure may have contributed to the Wndings. In addition, it is possible that the patterns of metabolic diVerences in those with TS relative to controls may be explained by the implicit and explicit need to remain still or to suppress tic symptoms during radiotracer uptake. The patients, in fact, were noted to tic rarely during the period of FDG uptake, which is not uncommon for TS patients during novel and attentiondemanding experiences. Finally, the image analytic methods used in this study can be questioned. Eight slices were analyzed for signiWcance. The investigators used two methodologies to identify the ROIs. One employed 164 pieshaped ROIs in the entire axial cross-sections, while the other used 112 circular ROIs placed only in the cortex. Of all statistical tests that were performed using either method of region deWnition, 22 regions were signiWcant using both methods. Another 32 ROIs were signiWcant in one or the other analysis but not in both. The authors note that the signiWcance of regions in both analyses support claims for the reliability of their image analytic methods. Although this is true, the reliability of the ®ndings (in terms of their independent test±retest reproducibility) would be more rigorously demonstrated if the same regional group differences were identi®ed using the two analytic methods on diVerent data sets ± either between two diVerent subject groups or the between scans obtained on a single subject group at two diVerent times. Moreover, the authors did not

correct their p values for multiple comparisons; consequently the risk of type I error (false-positive group diVerences) is considerable.

Additional analyses were performed on this same data set after adding two subjects to the group of patients with TS, bringing it to a total of 18 (16 males, two females; mean age 33"7 years) (Stoetter et al., 1992). The new analysis examined the correlations between all combinations of signiWcantly activated regions. SigniWcantly diVerent interregional correlations were seen in the patients compared with controls. In the correlations involving the frontal lobe, group diVerences were seen most frequently in those with the orbitofrontal cortex. In those involving the limbic regions, group diVerences were seen most frequently in regions around the insula. The most frequent group diVerences, however, involved the ventral striatum (seen in 35% of all possible correlations). Of all the correlation analyses, those that the authors singled out for discussion were those between the ventral striatum and sensorimotor cortices. These regions were inversely correlated in the controls but positively correlated in the TS group, suggesting a disturbed functional relationship between the ventral striatum and sensorimotor cortices in the TS group. Although these altered functional relationships may represent inherent disturbances in the connectivity of the brain in TS subjects, it is also possible that the altered relationships may be caused by implicitly diVerent group behavioral demands, such as requiring subjects to remain still or to suppress tics during the scan. These demands will be greater in the TS group and may speciWcally alter relationships between ventral striatum and sensorimotor cortices.

The largest HMPAO SPECT study to date compared 50 patients with TS (38 males, 12 females; mean age 24 years; age range 7±65 years; 27 had signiWcant OCD symptoms) with 20 normal controls (nine males, 11 females; mean age 23 years) (Moriarty et al., 1995). Subjects were scanned for 30min, although patients with TS who could not suppress their disruptive tics were scanned for a shorter period of time. Consistent with previous TS neuroimaging studies, the results indicated hypoperfusion in the left caudate and anterior cingulate, as well as hypoperfusion in the left dorsolateral prefrontal cortex. Because the resolution of the scans did not allow the investigators to diVerentiate reliably between perfusion of the caudate and lenticular nuclei, the blood Xow reductions that are reported as located in the left caudate should instead be more accurately described as being located in the left basal ganglia. Additional limitations of the study include the failure to take into consideration the eVects of age on regional blood Xow values. The age range of the patients with TS