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

Introduction

The emergence of newer, noninvasive approaches to brain research since the 1980s, for example magnetic resonance spectroscopy (MRS) and functional magnetic resonance imaging (fMRI), has provided Wndings of critical relevance to childhood-onset anxiety disorders. These techniques are particularly well suited for studying pediatric populations where radiation exposure limits the use of positron emission tomography (PET) and single photon emission computed tomography (SPECT). This chapter will discuss the use of these techniques in childhood-onset anxiety disorders.

For many years, investigations in adult psychiatric disorders have been extended to child populations with the intent of verifying the Wnding in children in a ªtop-downº fashion. In the process, the developmental focus was often lost. Recent investigations have kept the developmental focus at the forefront, while applying a systems neuroscience approach to childhood-onset neuropsychiatric disorders, including pediatric anxiety. Understanding the underlying neutral substrate of childhood-onset anxiety disorders requires knowledge of the normal functional development of the brain. Recent investigations have begun to reWne and map systematically the developmental trajectory of brain functions in normal children and to apply this knowledge to the study of childhood-onset neuropsychiatric disorders (Casey et al., 1995, 1997b; Rosenberg et al., 1997a,b; Rosenberg and Keshavan, 1998). The techniques of fMRI and MRS provide in vivo noninvasive approaches for studying neural circuitry in childhood anxiety disorders. Moreover, these techniques permit delineation of critical developmental ªwindowsº of abnormality in brain anatomy and function in pediatric anxiety disorders.

Implementing a neuroscience approach to develop-

mental systems in the study of childhood-onset anxiety disorders oVers several critical advantages. Anxiety disorders frequently emerge during childhood even though they are not always diagnosed at their time of onset. In contrast to other neuropsychiatric disorders such as depression, the presentation of many anxiety disorders in children is quite similar to their presentation in adulthood, suggesting that the risk for certain core anxiety symptoms emerges during early childhood development. Studies of anxiety disorders during childhood near the time of onset can, therefore, minimize potential confounds of illness chronicity and treatment eVects and begin to clarify the contributions of neurodevelopmental abnormalities to the pathogenesis of anxiety disorders.

A developmental approach to examining brain circuits can best explain the frequent childhood and adolescent onset of anxiety, as well as the modiWcation of its presentation with age. Because fMRI and MRS are noninvasive and without radiation risks, repeated studies can be performed to elucidate the longitudinal course of anxiety as well as the impact of psychotropic medication, cognitive behavioral therapy, and illness duration on brain function and chemistry.

This chapter will review brain imaging studies in various childhood-onset anxiety disorders, including obsessivecompulsive disorder (OCD), generalized anxiety disorder (GAD), panic disorder, speciWc phobia, social phobia, and post-traumatic stress disorder (PTSD), as well as suggest directions for future investigation. Not all anxiety disorders have been examined by neuroimaging techniques. For example, to our knowledge there are no published neuroimaging studies in patients with separation anxiety disorder, despite its high prevalence (Benjamin et al., 1990) and early age of onset (Keller et al., 1992). OCD is the most investigated anxiety disorder and will occupy the largest place in this review. Finally, it is important to note that

many fMRI and MRS studies have examined only a limited number of brain regions. Therefore, brain regions that are not mentioned either show no abnormalities or have not been analyzed.

One major concern in the study of anxiety disorders is the potential confound of ªnormativeº anxiety that arises from participating in a medical procedure. Indeed, diVerences in brain activity have been shown as a function of anxiety state using both [18F]-Xuorodeoxyglucose (FDG) PET and the 133Xe inhalation technique (Gur et al., 1987). Consequently, we discuss initially techniques of simulation training to reduce anxiety in subjects prior to functional neuroimaging data collection.

Simulation training for anxiety reduction

As with any medical procedure, brain imaging studies elicit anxiety in participants, particularly in children. To minimize levels of anxiety during MRI scanning, Rosenberg et al. (1997d) trained pediatric subjects in a simulation scanner that mimicked the actual MRI scanning environment. The simulation MRI scanner was equipped with a genuine scanner patient tube, 55 cm in diameter (kindly supplied by General Electric), a sound system, gurney bed, mirrors, and an airXow apparatus located in the bore of the scanner. The training process involved a graded exposure behavioral program. Parents and their children Wrst toured the MRI center, where they were encouraged to ask questions and observe MRI studies in progress. Subjects then entered the simulation scanning room and were gradually introduced to the scanning procedure. Subjects were asked to lie down on the moveable patient table and were given ear plugs. Once the radio frequency coil was placed over the subject's head, rhythmic sounds of the MR procedure were played over a high-Wdelity audio system. During the entire simulation procedure, all children had their pulse, blood pressure, and respiration continuously monitored. The Subjective Units of Discomfort Scale (SUDS), which measures distress on a scale of 0 to 100 (0 being no distress and 100 being overwhelming distress), was also administered (Wolpe et al., 1973). After the MRI simulation, children under the age of 12 received stickers and treats as rewards.

Rosenberg et al. (1997d) examined the eYcacy of simulation scanning in 16 healthy pediatric subjects who were 7±17 years of age and in 16 ageand sex-matched, severely anxious children with OCD. Both healthy controls and OCD children exhibited a signiWcant decrease in both subjective and objective anxiety as a result of the preliminary simulation scanning. Moreover, when levels

of anxiety of pediatric control subjects, who were not trained in the simulation scanner, were compared with control subjects trained in the simulation scanner, the pretrained group experienced signiWcantly less subjective and objective distress during MRI data acquisition. None of the subjects who had been trained in the simulation scanner suVered a claustrophobic reaction in the actual scanner, whereas 10% of pediatric subjects (n5 10) not trained in the simulation scanner had a claustrophobic reaction necessitating discontinuation of the MRI procedure. This reduced anxiety resulted in improved comfort and cooperation during data collection, reduced head motion artifact, and increased signal-to-noise ratios (SNR). Parents and their children found the simulation experience reassuring and were grateful for the opportunity to ªpracticeº and become comfortable with the MRI scanner. More striking were the results that occurred in patients with OCD, who began with higher levels of anxiety and demonstrated a more dramatic decrease in anxiety levels.

Simulation scanning may reduce the need for sedation for completion of MRI studies in children. This is important because sedation may introduce confounds when interpreting functional neuroimaging data. The simulation scanner also reduces the costs of scanning since the children are less apprehensive and require less time to settle down. The simulation scanning process typically takes 15±30min in most children; however, it may take longer when children require extra reassurance (Rosenberg et al., 1997d). It should be noted, however, that constructing an actual simulation scanner is expensive; typical costs for scanner construction, equipment, and scanner time lie in the range of US $10000±15000. Because of these high costs, some groups have adopted alternative strategies for desensitizing children, such as presenting videotapes that demonstrate the procedures, utilizing a working scanner when it is not in use for data collection, or constructing less-expensive simulation scanners. MRI studies can also be greatly facilitated in those institutions that have an MRI center located in a children's hospital. An environment designed to make the appearance of a hospital less intimidating can further help children habituate to the actual scanning process.

Another potential confound is pharmacologically induced anxiety. For example, stimulants like caVeine may have anxiogenic eVects. In fact, Moore et al. (1999) used 1H MRS to demonstrate that intravenous caVeine induced changes in lactate concentration in the anterior cingulate and insular cortices in healthy volunteers. Therefore, it is important to control the amount of caVeine (caVeine-con- taining sodas, chocolate) consumed by children.

Obsessive-compulsive disorder

Since the late 1980s, there has been an increasing recognition of OCD as a severe, highly prevalent, and chronically disabling disorder characterized by repetitive, ritualistic thoughts, ideas, impulses, and behaviors over which individuals have little control. Characteristic obsessions and compulsions include pathologic doubting, fear of germs, need for symmetry, checking, washing, hoarding, and religious scrupulosity. The lifetime prevalence of OCD is 2±3% (Flament et al., 1988; Valleni-Basile et al., 1994; Hanna, 1995), and as many as 80% of all patients have a disease onset in children and adolescence (Pauls et al., 1995). The mean age of onset of OCD in referred children and adolescents ranges from 9 years (Riddle et al., 1990) to 10.7 years of age (Last and Strauss, 1989).

Neurobiologic studies have begun to clarify the pathophysiology of OCD. Converging lines of evidence support basal ganglia±frontal cortical pathway dysfunction as a basis for OCD (Modell et al., 1989; Wise and Rapoport, 1989; Baxter et al., 1992; Insel, 1992) (Fig. 13.1, p. 242). Brain structures related to fear (i.e., amygdala, hippocampus) act in concert with the frontal thalamic loop, which is implicated in OCD (Fig. 13.2, p. 242). Increased rates of OCD symptoms in basal ganglia disorders such as Tourette's syndrome, Sydenham's chorea, and postencephalitic parkinsonism (von Economo, 1931; Pitman et al., 1987) and the fact that psychosurgical lesions of ventral prefrontal cortex (anterior cingulum) can decrease OCD symptoms (Jenike et al., 1991) provide indirect support for this hypothesis. More direct evidence comes from functional neuroimaging studies in adult patients with childhood onset of OCD. These studies have demonstrated increased metabolic activity in ventral prefrontal cortex and the head of the caudate nucleus (Benkelfat et al., 1990; Hoehn-Saric et al., 1991; Baxter et al., 1992; Swedo et al., 1992a,b; Rauch et al., 1994; Breiter and Rauch, 1996; Schwartz et al., 1996). However, despite the high prevalence and frequent childhood onset, there have been few studies of children with this disorder. The neuroimaging Wndings in OCD will be discussed in terms of the structures aVected, the networks identiWed by activation tasks, and the changes detected in neurotransmitters.

Structures implicated in obsessive-compulsive disorder

Basal ganglia and ventricular size

In two early studies using computed tomographic (CT) scans. Behar et al, (1984) found signiWcantly larger ventric-

ular brain ratios in 16 adolescents with OCD (13 males, three females) relative to ageand sex-matched control subjects, whereas Luxenberg et al. (1988) found signiWcantly smaller caudate volumes in late adolescent and young adult male patients with OCD compared with controls. Caudate abnormalities were most pronounced in younger patients with OCD (J. Rapoport, personal communication). Using MRI, Rosenberg et al. (1997c) examined striatal volumes in 19 nondepressed, treatmentnaive pediatric patients with OCD, ages 8±17 years, and 19 ageand sex-matched controls. The patients with OCD had signiWcantly smaller striatal volumes than the controls, and their striatal volumes negatively correlated with OCD symptom severity. Striatal volumes, however, did not correlate with duration of illness, age of onset of illness, age of child, or intracranial volume. Children with OCD also had increased third ventricular volumes relative to ageand sex-matched controls. Consistent with these Wndings, Giedd et al. (1995) also found selective anatomic basal ganglia abnormalities in 24 pediatric patients with Sydenham's chorea and OCD compared with 48 matched controls. The sizes of the caudate, putamen, and globus pallidus were larger in the Sydenham's chorea group compared with controls; no size diVerences were found in total cerebral, prefrontal, midfrontal, or thalamic volumes.

Frontal cortex and corpus callosum

Consistent with a prior study in adult patients with OCD (Robinson et al., 1995), Rosenberg and colleagues (1997c) did not Wnd any signiWcant diVerences in MRI measurements of total prefrontal cortical gray or white matter volumes between 19 pediatric patients with OCD, aged 7±17 years and 19 ageand sex-matched controls. However, prior functional neuroimaging studies in adult patients with OCD had found abnormally high glucose metabolic activity in ventral prefrontal cortex (Baxter, 1992). Morphometric measurement of the total prefrontal cortex may not be sensitive enough to detect subtle case±control diVerences in structures such as ventral prefrontal cortex. Indeed, subsequent studies of the regional morphology of the corpus callosum are consistent with ventral prefrontal deviance (Rosenberg et al., 1997b). SpeciWcally, abnormalities in the genu of the corpus callosum were identiWed. Fibers that connect the ventral prefrontal cortex and the striatum cross between the cerebral hemispheres at this point (Seltzer and Pandya, 1986). Patients with OCD who were between the ages of 7.2±17.7 years were found to have larger genu sizes than ageand sex-matched controls. The increased genu sizes correlated positively with OCD symptom severity but not with illness duration. Increased genu size of the corpus callosum in

Fig. 13.3. Anterior genu signal intensity in patients with

obsessive-compulsive disorder (OCD) and controls (*F(1,37) 5 5.47; p5 0.025). (Created by Frank P. MacMaster.)

patients with OCD could be the result of increased myelinization of Wbers in that area relative to that in controls. In fact, the normal age-related increase in the total corpus callosum area (Giedd et al., 1996; Rajapakse et al., 1996) was absent in patients with OCD. Earlier myelinization could result in increased corpus callosum size in these patients. However, no signiWcant diVerence in corpus callosum area between adult patients with OCD and controls has been observed (Jenike et al., 1996), consistent with the hypothesis that myelinization of axons may be more gradual in healthy children than in those with OCD, only catching up toward the end of adolescence or in early adulthood. It is also possible that these Wndings may reXect a greater number of synaptic connections or less pruning in young patients with OCD, resulting in increased axonal density.

The following Wndings support the hypothesis of enhanced myelinization of corpus callosum Wbers in OCD. MacMaster et al. (1999) compared regional corpus callosum signal intensity from midsagittal MRI in 21 treatmentnaive patients with OCD, 7.2±17.2 years of age, and 21 case-matched healthy controls. Although controversial, the measure of MRI signal brightness, or signal intensity (SI), is believed to reXect myelination (van der Knapp and Valk, 1989; Laissy et al., 1993; Belmonte et al., 1995).

Fig. 13.4. Genu signal intensity in relation to symptom severity (r5 0.55; p5 0.01) as measured by the Children's Yale±Brown Obsessive Compulsive Scale (CY-BOCS). (Created by Frank P. MacMaster.)

MacMaster et al. (1999) found lower SI in the anterior genu region of the corpus callosum connecting the ventral prefrontal cortex and the striatum in OCD compared with controls (Fig. 13.3). The lower was the genu SI, the more severe were the OCD symptoms (Fig. 13.4). No correlation was detected between illness duration and genu SI. MacMaster and colleagues (1999) hypothesized that a deWcit in myelination occurring in patients with OCD may result in abnormal signal transmission and function of ventral prefrontal±striatal circuitry.

Anterior cingulate

More recently, Rosenberg and Keshavan (1998) reported increased ventral prefrontal cortex and anterior cingulate volumes in 21 treatment-naive pediatric patients with OCD relative to ageand sex-matched controls. Increased volumes were associated with increased OCD symptom severity, but not with illness duration. No anatomic abnormalities were observed in posterior cingulate cortex, dorsolateral prefrontal cortex, or temporal cortex. Moreover, an age-related increase in anterior cingulate volumes in healthy pediatric controls was not observed in these pediatric patients. Increased anterior cingulate volumes, particularly in younger patients with OCD, were consistent with prior observations of increased genu size of the corpus callosum (Rosenberg et al., 1997b). Larger anterior cingulate volumes tended to be associated with smaller striatal volumes in pediatric patients with OCD,

(Rosenberg et al., 1997c; Rosenberg and Keshavan, 1998). This neuroanatomic dysplasia in OCD might represent an increase in pruning and/or a decrease in neural tissue in the striatum, whereas the increased ventral prefrontal cortical volumes might represent a delay in pruning and/or an excess of neural brain elements. The increased volume observed in the anterior cingulate in OCD may be indicative of a delay in the normal pruning processes that occur during this period, further implicating a putative developmental abnormality.

Cognitive/behavioral activation tasks

Another strategy used to examine neural networks in OCD involves the use of functional neuroimaging in conjunction with activation tasks. This neurocognitive/behavioral approach has the advantage of exploiting the richness of neuropsychologic research, which develops theoretical models and provides a priori hypotheses. Such experimental paradigms require the development and testing of activation tasks appropriate for the model to be tested and for use with brain imaging technologies.

Response inhibition tasks

Based on the theory that OCD may represent the expression of abnormal inhibitory processes, tasks of inhibition have been sought for use in brain imaging studies. Along these lines, Rosenberg et al. (1997a) measured three core prefrontal functions in treatment-naive pediatric patients with OCD (aged 7 to 18 years) and ageand sex-matched

healthy control subjects: (i) inhibition of contextually inappropriate responses, (ii) delayed responses involving working memory (requiring a response to new situations on the basis of stored information without beneWt of sensory cue information), and (iii) the ability to prepare a response in advance of its initiation. The Antisaccadic Response Suppression Task was used to test these functions. In this task, subjects Wxate their eyes on a central target (such as a red dot), then look the same distance, but in the opposite direction of targets presented to the left or right of the center Wxation (8, 16, or 24 degrees). Therefore, if a target appeared to the left, subjects need to shift their eyes 8° to the right. This test requires the suppression of a powerful reXexive response to look toward novel peripheral targets and the ability to drive the eyes volitionally away from a target when no cue exists in the periphery. The inability of subjects to drive their eyes away from and in the opposite direction of the peripheral target is classiWed as a response suppression failure. A selective deWcit in neurobehavioral response inhibition was observed in patients with OCD, particularly in younger patients, compared with ageand sex-matched controls (Fig. 13.5); these patients showed no deWcits on measures of delayed response and preparatory set. Controls began to master neurobehavioral response inhibitions tasks approximately 5 years earlier than did patients with OCD; as a result, comparable performance was observed between 16-year-old patients with OCD and 11-year-old healthy control subjects. In addition, response inhibition failures correlated signiWcantly with striatal and ventral prefrontal cortical volumetric MRI

measures (Rosenberg and Keshavan, 1998), suggesting a developmental neurobiologic model for relating anatomic Wndings with clinical and behavioral OCD symptoms.

Ventral prefrontal±striatal circuits appear to play a crucial role in mediating responses suppression (Rosvold and Mishkin, 1961; Luria, 1966; Goldman and Rosvold, 1970; Iversen and Mishkin, 1970; Goodglass and Kaplan, 1972; Passingham, 1972; Rosenkilde, 1979; Stuss and Benson, 1983; Diamond, 1990; Cummings, 1993; Sweeney et al., 1996), whereas dorsal prefrontal cortex has been shown to mediate delayed response and preparatory set functions (Goldman and Rosvold, 1970; Funahashi et al., 1989). Consistent with this, functional neuroimaging studies in healthy volunteers have reported functional activation of ventral prefrontal cortico±striatal circuitry in tasks of inhibition such as the antisaccadic oculomotor response inhibition task (Sweeney et al., 1996) and the go- no-go task (Casey et al., 1997). Consequently, dysplasia in ventral prefrontal±striatal circuits might give rise to OCD by disrupting a neurocognitive function that mediates ongoing purposive behaviors. The localization of a neuroanatomic abnormality that correlates with deWcits in neurobehavioral tasks known to be subserved by this brain region provides an ideal focus for functional neuroimaging using fMRI. The fMRI technique can monitor in vivo the spatial localization and time course of functional abnormalities. Such studies are now being conducted in our laboratory.

Symptom provocation

Another type of behavioral activation paradigm used to study OCD is that designed to provoke symptoms. Using fMRI, Breiter et al. (1996) studied 10 adult patients with OCD and Wve normal control subjects during control and experimental conditions designed to induce OCD symptoms. The subjects were asked to hold a neutral stimulus object, with the knowledge that a feared stimulus object would be introduced half-way through the scan. The feared stimulus objects included blindly faked versions of

(i) tissue soaked in toilet water, (ii) plastic bags from contaminated waste barrels, and (iii) tissue with nose eZuvium. Increased activation of the ventral prefrontal±striatal circuits was found in patients with OCD but not in healthy controls. Abnormal activation was also observed in these patients in temporolimbic regions, i.e., the hippocampus. Adapting the technique designed and developed by Rauch and colleagues (1994) and Breiter and associates (1996), Rosenberg and Thulborn have initiated similar studies in pediatric patients with OCD with comparable Wndings (unpublished data).

Conclusion

Findings from activation studies appear promising. Thus far, they support neurocognitive and neurofunctional theories of OCD. The challenge of neuroimaging research is to further the understanding of mechanisms underlying the hypothesized processes in order to provide clues on how to modify these processes and develop focused, rational, and eVective therapeutic interventions. In this regard, the elucidation of the role of neurochemical substrates in the dysfunction of the neural circuits involved in OCD is critical.

Neurochemical substrates

Serotonin

Pharmacologic treatment studies have suggested a critical role for serotonin in OCD. Serotonin reuptake inhibitors (e.g., Xuxovamine, sertraline) are eVective in treating OCD, whereas dopamine and noradrenaline reuptake inhibitors (e.g., amitriptyline, desipramine) are less eVective (Pigott, 1996). Other studies have found the serotonin transporter protein (5HTPR) capacity, as indexed in platelets by 3H- paroxetine, to be decreased in pediatric patients with OCD (Sallee et al., 1996). Researchers have also found the brain regions with the greatest binding of the potent selective serotonin reuptake inhibitor (SSRI) citalopram to be the caudate nucleus and nucleus accumbens (Insel, 1992). Moreover, the ventral prefrontal cortex and the striatum are densely innervated by serotonin-containing neurons (Smith and Parent, 1986). Finally, animal studies have shown that a sustained administration of SSRI to guinea pigs increases serotonin release by desensitizing terminal serotonin autoreceptors in the orbital prefrontal cortex. Findings such as these have spawned the ªserotonin hypothesis of OCDº, which postulates that OCD is mediated by serotoninergic alterations in ventral prefrontal±striatal circuits.

Studies investigating alterations of neurotransmitters in blood and the serotonin metabolite 5-hydroxyindoleacetic acid (5-HIAA) in cerebrospinal Xuid (CSF) of patients with OCD have found positive correlations between druginduced decreases in platelet serotonin concentrations (Flament et al., 1985) and CSF 5-HIAA levels (Thoren et al., 1980) in adult patients with OCD; however, other studies have failed to detect such a relationship (Pandey et al., 1993). Studies of platelet serotonin and CSF, however, provide a poor index of brain function and chemistry.

Recently, Rosenberg et al. (1998) compared seven children with OCD (aged 8 to 13 years) with seven age-matched controls using PET and '-[11C]-methyl-l-tryptophan ([11C]- AMT), an analog of tryptophan and a tracer for serotonin synthesis (Diksic et al., 1991). Tryptophan is converted to

serotonin by tryptophan hydroxylase, the rate-limiting enzyme in serotonin synthesis. Preliminary data indicated a reduction in serotonin synthesis in the caudate nucleus of treatment-naive, pediatric patients with OCD, relative to controls (Fig. 13.6, p. 242). Five children were scanned twice: before and after a 3-week treatment with the SSRI paroxetine. Treatment response was associated with increased serotonin synthesis in the caudate nucleus in three children (Fig. 13.6, p. 242).

Because PET scanning involves exposure to ionizing radiation, the study of healthy children as controls is problematic. The risk±value attached to radiation exposure at the doses used in brain imaging studies is diYcult to address because of the emotional impact at the mention of radiation. In fact, health hazards associated with low-level radiation have not been detected in any of the large studies conducted to date. Despite the lack of evidence of excess risk of low-level radiation exposure above the expected risks of daily life, most Institutional Review Boards (IRBs) consider low-level radiation exposure to be above minimal risk. In the aforementioned study (Rosenberg et al., 1998), comparison subjects for the patients with OCD included developmentally normal siblings of patients with OCD or autism. The IRB of the Children's Hospital of Michigan at Wayne State University School of Medicine, where the study was performed, did not allow the study of healthy children without a Wrst-degree relative with a signiWcant neuropsychiatric disorder. On the one hand, the use of siblings of aVected patients as comparison groups may not be ideal because they may carry a genetically determined brain abnormality predisposing them to the disorder, but without expression of the symptoms. Indeed, the IRB approved these protocols in siblings because of their increased risk for developing neuropsychiatric disorders. For example, if brain serotonergic abnormalities are found to constitute a pathophysiologic mechanism in OCD, then it is possible that their high-risk siblings may also have serotonergic abnormalities. On the other hand, the use of siblings as a comparison group can reduce intersubject variability, increasing the chance to detect diVerences related to diagnostic status.

Glutamatergic neurotransmitter

MRS permits investigators to monitor directly and noninvasively brain chemistry without radiation exposure (see Chapter 4). Proton MRS can identify compounds that include the neuronal marker N-acetylaspartate (NAA) (Birken and Oldendorf, 1989), glutamate/glutamine, gamma-aminobutyric acid (GABA) (Glx), creatine/phosphocreatine (Cr), choline compounds (Cho), and myo- inositol (myo-I).

The ability to measure Glx concentrations by MRS may be especially relevant, because glutamatergic±serotonin modulation may be involved in the pathogenesis of OCD (Rosenberg and Keshavan, 1998). Indeed, glutamate plays a critical role in the striatum (Becquet et al., 1990), which receives dense glutamatergic projections from the prefrontal cortex (Taber and Fibiger, 1993, 1994). The caudate nucleus, in particular, receives a very large glutamatergic innervation from the cerebral cortex such that if the frontal cortex or the hemicortex is ablated, there is a marked reduction of glutamate concentrations in the rat caudate nucleus (Kim et al., 1977). In addition, Becquet et al. (1990) have demonstrated a potent presynaptic inhibitory glutamatergic control of serotonin release, possibly via GABA interneurons, in the cat caudate nucleus.

Preliminary MRS studies have shown elevated glutamatergic concentrations in the caudate nucleus of pediatric patients with OCD compared with controls (Rosenberg et al., 1998). Following 12 weeks of SSRI treatment, the level of glutamatergic concentrations in the caudate nucleus decreased in patients with OCD (Moore et al., 1998) (Fig. 13.7). This reduction was associated with a reduction in OCD symptom severity (Fig. 13.8).

Using 1H MRS, Ebert et al. (1997) demonstrated reduced NAA levels, suggestive of neuronal dysfunction, in the striatum and anterior cingulate of 12 patients with OCD that correlated with symptom severity, but not with illness duration. No abnormalities were observed in the parietal cortex. A reduction of NAA levels in the ventral prefrontal cortex and striatum may be indicative of underlying metabolic abnormalities in OCD (Ebert et al., 1997). These results demonstrate how 1H MRS can be used for the in vivo monitoring of brain chemistry and the impact of psychotropic medication on brain neurochemistry, as it relates to the medication's therapeutic eVect. Moreover, this technology involves no radiation risks or blood sampling, both of which are best avoided in pediatric research studies. It should be noted, however, that considerable development and reWnement of methods is necessary before determination of the precise meaning of the MRS signals is possible.

Summary

To date, functional neuroimaging has most consistently identiWed the caudate nucleus, ventral prefrontal cortex, and anterior cingulate as neural substrates of OCD. These regions have shown abnormal serotonergic and/or glutamatergic function. These Wndings evidence the great potential of brain imaging for unraveling the mechanisms

underlying OCD. Functional networks, as well as neurobiochemical systems, need to be examined in a systematic fashion. This work needs to be interactive with genetic research, which requires the identiWcation of homogeneous phenotypes (which brain imaging may provide) but which can also deWne subgroups to be studied using neuroimaging.

Generalized anxiety disorder

Epidemiologic catchment area studies have found a high lifetime prevalence rate of 3.7% for GAD (called overanxious disorder in DSM-III-R) in 5596 nonreferred children aged 14±17 years (Whitaker et al., 1990), a prevalence of 2.4% in 1869 children 12±16 years (Bowen et al., 1990), and 2.9% for 792 11 year olds (Anderson et al.,

1987). In a pediatric primary care setting, a sample of 300 children aged 7±11 years demonstrated a 15.4% prevalence of anxiety disorders, with simple phobia, separation anxiety disorder, and GAD being the most prevalent (Benjamin et al., 1990). The prevalence of GAD in this population was 4.6%. Median age of onset of GAD was found to be 10 years (Keller et al., 1992). To date, there have been only three published brain imaging studies in adults with GAD, and none in children or adolescents with GAD.

Benzodiazepine cerebral eVects

Wu et al. (1991) utilized PET to measure brain glucose metabolism in 18 patients with GAD and 15 healthy control subjects. Measurements in patients with GAD were taken at three diVerent times: baseline, pretreatment (or baseline

Fig. 13.8. Summary of 1H MRS and obsessive-compulsive symptomatology Wndings in a 9-year-old boy during a 12-week trial with the selective serotonin reuptake inhibitor paroxetine. Glx, glutamate/glutamine concentrations (- - -); CY-BOCS, Children's Yale±Brown Obsessive Compulsive Scale (Ð). (Permission granted from the Williams & Wilkins Publishing House; Moore, G. J., MacMaster, F. P., Stewart, C. and Rosenberg, D. R. (1998). Caudate glutamatergic changes with paroxetine therapy for pediatric obsessive compulsive disorder, Journal of the American Academy of Child and Adolescent Psychiatry, 27(6), 663±7.)

2), and following a 21-day double-blind treatment with clorazepate (8) or placebo (10). During the Wrst baseline, subjects (GAD and controls) were required only to observe the degraded stimulus of a continuous performance task (CPT) (passive task). In the pretreatment and posttreatment studies, subjects actively had to identify target stimuli during a CPT (active, vigilance task) before and after receiving either a benzodiazepine (clorazepate) or placebo for 21 days. The use of two diVerent cognitive tasks during baseline permitted the identiWcation of the regions of the brain involved in attention, determined through comparisons of activity during the passive versus active tasks. Attention is likely to activate regions involved in anxiety, which is characterized by a heightened state of vigilance. The identiWed regions were then targeted in the assessment of treatment eVects.

During the passive visual task, patients with GAD showed lower basal ganglia (particularly putamen and globus pallidus) metabolic rates and higher glucose metabolism in occipital cortex, right posterior temporal cortex, and the right precentral frontal gyrus relative to controls. Comparison of the two baseline studies (active

task, passive task) showed that patients had higher metabolic rates in the basal ganglia and right parietal lobe and lower rates in right temporal and occipital cortices during the active task than during the passive task. The absence of comparison with controls makes the interpretation of these latter Wndings diYcult. Treatment with a benzodiazepine reduced absolute rates of glucose metabolism, particularly in occipital cortex, but did not normalize regional glucose metabolic patterns. The authors hypothesized that the changes in occipital metabolic rates reXected anxiety-related alteration of visual information processing (increased scanning of the environment).

Neurochemical brain imaging studies

Tiihonen et al. (1997) used SPECT with NNC 13-8241, a new 123I-labeled speciWc benzodiazepine receptor radioligand. Ten medication-naive female patients with GAD were compared with 10 ageand sex-matched healthy controls. Results showed that benzodiazepine receptor binding was signiWcantly reduced in the left temporal pole in patients with GAD relative to controls.

Conclusions

The studies described above are diYcult to interpret, in part because of the complex designs that involve the interaction of various factors such as the use of diVerent conditions and pharmacologic intervention. Issues of primary versus secondary eVects of the biochemical changes, as well as of the changes in symptoms, are diYcult to sort out in pharmacologic studies and require the use of careful study designs. The serotonin receptor SPECT study is of great interest but, like most brain imaging studies, suVers from low statistical power because of the small sample size; it requires replication.

It is clear that abnormalities in integrated brain activity, indexed by measurements of glucose metabolism, need to be characterized neurobiochemically. An interesting proposition is that benzodiazepine receptor dysfunction may be secondary to GABAergic abnormalities (Allen et al., 1995). The 1H MRS technique at high magnetic Welds (4 T and higher) may provide a means with which to examine this hypothesis (Rothman et al., 1993; Keltner et al., 1997).

Panic disorder

As described in DSM-IV (American Psychiatric Association, 1994), the diagnosis of panic disorder can be made with or without agoraphobia (fear of crowds and

being unable to escape from places where panic attacks may occur). Panic disorder is characterized by recurrent panic attacks or the fear of developing a panic attack. A panic attack is characterized by feelings of impending doom and/or fear of dying, with typical physical and psychologic symptoms. Retrospective studies of adults with panic disorder indicate that panic disorder typically begins in adolescence or early adulthood (Moreau and Follett, 1993) and is uncommon before puberty (Black and Robbins, 1990; Klein et al., 1992). The peak age of onset of panic disorders is 15 to 19 years (von KorV et al., 1985). In a study of 754 pubertal sixthand seventh-grade girls in the USA, 5.3% of the girls reported a history of having had at least one panic attack (Hayward et al., 1992). Interestingly, panic attacks increased signiWcantly during sexual maturation, with rates of 8% for females at Tanner stage 5 (sexually mature) and 0% for the 94 girls at Tanner stages 1 or 2 (sexually immature). This increased frequency of panic attacks with sexual maturation could not be explained by diVerences in chronologic age (Bernstein et al., 1996). This suggests the inXuence of sexual steroid hormones on the induction of panic attacks.

Pollack et al. (1996) found that over 54% of 194 adults with panic disorder had a history of childhood anxiety disorders, and patients with a history of childhood anxiety disorders had signiWcantly more comorbid mood disorders during adulthood. Furthermore, over 60% of adults with panic disorder who had a history of childhood anxiety disorder had been diagnosed with two or more anxiety disorders during childhood. The comorbidity among anxiety and mood disorders must be considered in designing brain imaging studies. Neuroimaging studies in adults with panic disorder have begun to delineate the neurobiology of this disorder (Sallee and Greenwald, 1995; Goddard and Charney, 1997).

Structural abnormalities

To identify more homogeneous subtypes of panic disorder, Dantendorfer et al. (1996) divided a sample of 120 patients with panic disorder (53 males, 67 females) into two groups: one with nonepileptic electroencephalographic (EEG) abnormalities (35) and one without EEG abnormalities (85). Twenty eight patients with EEG abnormalities were able to complete an MRI successfully. These 28 subjects were sexand age-matched with 28 patients with GAD without EEG abnormalities and with 28 healthy subjects, all with normal EEG. The EEG was evaluated clinically by a psychiatrist. Although EEG abnormalities were more frequent in patients with panic disorder than rates reported for healthy subjects, their role in panic disorder is unclear,

and the choice of this discriminating factor not well justiWed. Subjects were compared on structural MRI, read clinically by three experienced neurologists who were blind to all subject information. MRI abnormalities were present in over 60% of the patient group with abnormal EEG, 17.9% of the patient group with normal EEG readings, and 3.6% of the controls. Patients with panic disorder had signiWcantly more septohippocampal abnormalities than controls, which included smaller right hippocampi and the presence of a cavum septum pellucidum. The implication of hippocampal abnormalities is consistent with functional neuroimaging Wndings.

Functional neuroimaging

Utilizing PET and H215O with 16 patients with panic disorder and 25 control subjects, Reiman et al. (1986) reported that those patients who were most susceptible to lactateinduced panic attacks (eight) had abnormally high global brain oxygen metabolism and deviant asymmetries of parahippocampal regional cerebral blood Xow (rCBF) and oxygen metabolism.

Using PET and FDG during the performance of an auditory discrimination attention task, Nordahl et al. (1990) compared 12 patients with panic disorder (Wve males, seven females) with 30 healthy volunteers (16 males, 14 females). Similar to Reiman's results (1986), the patient group showed abnormal hippocampal asymmetry. However, global brain glucose metabolism was not altered. In addition, regional glucose metabolism was abnormally low in the left inferior parietal lobe and anterior cingulate cortex and abnormally high in the orbitofrontal cortex.

More recently, Dager et al. (1995) used 1H MRS to compare the eVects of hyperventilation on brain lactate in seven patients with panic disorder (four males, three females) and seven healthy comparison subjects (Wve males, two females). Hyperventilation normally increases brain lactate concentrations. This study was stimulated by previous observations of abnormally large increases in brain lactate in patients with panic disorder during lactate administration (Dager et al., 1994). Prior to hyperventilation, these patients had brain lactate levels similar to those of controls, but they demonstrated a signiWcantly greater increase in brain lactate than controls in response to the same degree of hyperventilation (Dager et al., 1995). Blood levels of lactate measured before and after hyperventilation did not diVer signiWcantly between the patients and the healthy controls, underscoring the importance of direct evaluation of the brain regions of interest as opposed to relying on peripheral indices of brain function and chemistry. Here again, the mechanism of increased lactate