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

384M. Luciana and C. A. Nelson

two-alternative forced-choice discrimination of two lined drawings using immediate feedback provided by the computer (Fig. 22.2). The child is told to touch one of two Wgures presented on the screen, and if his/her choice is correct, the computer will Xash green; if incorrect, the computer will Xash red. The child is told that s/he will learn a rule to determine which choice is correct but that the rule might change as the task continues. The child is told that s/he should try to make as many correct choices as possible. The learning criterion is six consecutive correct responses. After achieving this criterion on the Wrst stage, the conditions are reversed so that the stimulus that was Wrst correct is now incorrect, and the stimulus that was at Wrst incorrect is now correct (simple reversal). Again, there are six trials to criterion.

Next, a second dimension (shapes) is introduced together with the lined drawings so that each stimulus now contains two drawings: one lined and one shaped. To succeed on this compound discrimination condition, the subject must continue to respond to the previously relevant dimension (ªrespond to linesº) while ignoring the presence of the new irrelevant dimension (shapes). Two compound discrimination conditions (stage III and stage IV) are administered, one where the lined and shaped drawings are presented separately (e.g., side by side) and one where the stimuli are superimposed upon each other. Following successful completion of these two conditions, there is a compound discrimination reversal (i.e., the correct stimulus becomes incorrect and vice versa).

The next (VIth) stage involves the Wrst demand for an attentional shift using novel stimuli. Termed the intradimensional shift stage, novel or never-seen exemplars of each of the two dimensions (line and shape) are introduced without further instructions, and the subject's responses to the previously relevant dimension (lined drawings) are correct. Success on this stage requires that the subject generalize the rule from previous learning (e.g.,ªlined drawings are correctº) to new stimuli. Following another intradimensional reversal, a demand for a second type of attentional shift is imposed. This stage is termed the extradimensional shift. Once again, novel exemplars of each stimulus dimension are presented without further instructions. To succeed at the extradimensional shift stage, the subject must shift response set from the previously relevant dimension (lines) to the previously irrelevant dimension (shapes). This shift requires that the subject learn and respond to a new rule (e.g., ªlines are no longer correct; shapes are correctº). This stage is presumably analogous to the types of category shift that are required by more diYcult tests of set-shifting ability such as the WCST. The Wnal task stage is a reversal of the extradimensional shift.

Each response made by the subject is presumably

inXuenced by the feedback (correct versus incorrect) that s/he receives on the previous trial.Variables coded for each subject include the stage reached, the trials to criterion, and number of errors for each stage completed. Testing is automatically discontinued if a subject fails to reach criterion (i.e., six consecutive correct responses) after 50 trials.

Patients with frontal lobe lesions generally proceed without diYculty until an extradimensional (between-cat- egory) shift is required, at which point they experience diYculty. Patients with Parkinson's disease demonstrate similar impairments. In contrast, patients with temporal lobe lesions, relative to normal controls, proceed through all stages without diYculty (Owen et al., 1991). A functional dissociation between prefrontal regions is suggested by a recent lesion study in marmosets (Dias et al., 1996) that has implicated the dorsolateral prefrontal cortex in betweencategory set shifting and the orbitofrontal cortex in withincategory reversal shifts.

Planning and behavioral inhibition: the Tower of London

The Tower of London task was Wrst described by Shallice (1982) and measures planning and behavioral inhibition. It is described in detail in Chapter 21 and illustrated in Fig. 21.1. The task has a minimum number of moves in which it can be completed successfully, and the solution can be reached after a minimum of two, three, four, or Wve moves. For each trial, the number of moves to complete the trial, as well as execution times, are recorded.

Additionally, the child completes a ªyoked controlº condition in order to provide baseline measures of motor initiation and execution times. On each trial of the control condition, the computer executes a series of moves using the stimuli on the top half of the screen. The subject is required to imitate this sequence of movements using his/her balls on the bottom half of the screen. The movements are designed to require visuomotor integration and to occur one ball at a time; that is, the computer moves one ball, following which the child copies the computer's move. This task is yoked to the Wrst half of the test in that for each trial of the yoked condition, the movements of the balls are exact replicas of moves made by the child in the corresponding test trials. The measurement of the selection and execution latencies in the yoked condition provides baseline estimates of motor initiation and execution times (Veale et al., 1996). These movement times are then used to derive measures of planning or ªthinkingº times in the experimental task. In each problem, the initial thinking time is the time between the initial presentation of the problem and the Wrst touch minus the corresponding

motor initiation time as calculated from the yoked control task. The subsequent thinking time is the time between the Wrst move and the completion of the problem minus the total motor execution time derived from the control task.

In all, the following variables are computed for each problem set representing four levels of problem diYculty (2-, 3-, 4-, and 5-move problems): average number of moves to complete each set, initial thinking time, and subsequent thinking time. For the purpose of data analysis, two variables are of primary interest: the average number of moves to complete each problem set and initial thinking (i.e., planning) times.

Adults with either unilateral or bilateral frontal lobe lesions require more moves to solve Tower of London problems than do normal controls (Owen et al., 1990). Their initial planning times (the time before starting a given problem) do not diVer from those of normal control subjects but their subsequent thinking times are increased. Owen et al. (1990) interprets this pattern as evidence that patients with frontal lesions, unlike patients with temporal lobe lesions, frequently initiate a response before fully thinking it through.

This task has been modiWed for use in a neuroimaging environment. BrieXy, subjects are required to solve the problems mentally and then report the number of moves required. PET imaging has been conducted in normal adults during the problem-solving phase of the task. A distributed network of cortical areas including bilateral prefrontal, cingulate, and parietal cortices was activated (Baker et al., 1996). Of signiWcance is the fact that increased problem diYculty was associated with activation in right rostral prefrontal cortex. Concordantly, Morris et al. (1993), using single photon emission computed tomography (SPECT), reported increased regional cerebral blood Xow in the left prefrontal cortex during planning compared with that in a motor control task. This lateralized activation was particularly associated with increased planning time, as well as increased eYciency of task performance.

Other abilities measured by CANTAB

Although the strength of the CANTAB lies in its use as a tool for the assessment of frontostriatal dysfunction, several other tasks included in the battery rely on other neural circuitry underlying basic neuropsychologic functions. Among these functions are psychomotor speed and accuracy, memory-storage capacity, and recognition memory. In our normative studies, we have included three other CANTAB tasks as controls for the frontal lobe-mediated behavioral functions that are of primary interest.

Motor screening task

The motor screening task measures psychomotor speed and accuracy. The child's task is to touch visual targets that are presented one at a time on the computer screen quickly and accurately. Accuracy and response latency are recorded.

Spatial span task

The spatial span task is based on the Corsi block task (Milner, 1971) and measures memory for a spatial sequence. BrieXy, on each trial, the child views an array of 10 white boxes displayed on the computer screen. At the start of the task, two boxes change color one at a time, in a sequence. After completion of the sequence, the child hears a beep that signals him/her to reproduce the sequence (e.g., touch the box that changed color Wrst, then the box that changed color second, and so on). If the child reproduces the two-item sequence correctly by touching the boxes in the correct order, the computer advances to the next trial, which includes a sequence of three boxes. The maximum sequence length (which deWnes the length of the child's memory span) that can be achieved is nine. The child has three attempts at each diYculty level to pass that level. If the child fails all three attempts at a diYculty level, the test is terminated. Prior to the start of test trials, the child is given two practice trials to assure his/her understanding of the task.

The verbal counterpart to this test is the digit span test of the Wechsler Intelligence Scales and/or the Wechsler Memory Scale.

This task diVers from the spatial working memory task described above in that it yields a measure of nonverbal memory capacity rather than the extent to which memory can be utilized to reach a self-directed goal. Typically, we have found that an individual's memory span is negatively correlated with the number of forgetting errors on the spatial working memory task (Luciana and Nelson, 1998). However, the spatial working memory task uniquely measures executive skills that are nonmnemonic in nature. For example, the spatial working memory strategy score is not necessarily related to memory span but is dependent upon behavioral organization.

Pattern recognition

Pattern recognition employs a DMTS paradigm to measure recognition memory for visual patterns. (CANTAB also includes a more-complicated DMTS task that was not used in our normative study because of its increased time demands.) For each of two blocks, the subject is told to

attend to the computer screen. At the center of the screen, a series of geometric patterns is presented one after the other for a 3 s viewing interval. Following the presentation of the pattern series, the screen ªpausesº for a duration of 5 s, after which the subject is presented with two geometric patterns. One of the two designs is from the previously viewed list. The other is a completely novel stimulus. The two designs diVer in shape but not in color. The child is told to touch the design that s/he remembers having seen. Twelve trials, one containing each of the 12 target stimuli, are presented in each of the two blocks for a total of 24 trials. Accuracy and response latency are recorded. The percentage of correct responses across both blocks is used to represent the subject's pattern recognition score. Notably, performance on this task is highly sensitive to posterior brain lesions in adult patients and appears to be unaVected by frontal lobe pathology (B. J. Sahakian, personal communication 1998).

In addition to the speciWc brain±behavior correlates described above for these subtasks, the CANTAB appears to be sensitive to Xuctuations in ascending cholinergic and monoaminergic neurotransmitter levels, as is apparent from several recent treatment studies in patients with dementia and in acute pharmacologic challenges with normal volunteers (Robbins et al., 1994; Robbins, 1996).

Developmental Wndings

Our data consistently indicate that planning and working memory skills have not reached adult levels by the age of 12 years. This conclusion is illustrated based on analysis of two CANTAB tasks that are reliable correlates of frontal lobe function in adults: the Tower of London and the spatial working memory task. As can be seen in Fig. 22.3, the number of total forgetting errors on the spatial working memory task is con-

sistently high from 5 to 12 years of age. Although sample sizes are small (13) for children above the age of 11, it appears that forgetting errors have declined signiWcantly by 20 years of age. Additionally, when performance on the most diYcult problems of the Tower of London is examined, a similar proWle is evident (Fig. 22.4). Figure 22.4 indicates the average number of moves used to solve the most diYcult 5-move problems by children between the ages of 5 and 18 years. As can be seen, 12-year-old children are still making a number of ªexcessº moves to solve the problems relative to young adults, indicating that their ªlook ahead planning skillsº are not fully developed. These Wndings are consistent with the literature demonstrating a protracted course of neurodevelopment of prefrontal, relative to other, brain regions through early adolescence (Huttenlocher, 1994; Huttenlocher and Dabhholker, 1997).

In contrast, abilities that rely on posterior brain regions, including pattern recognition memory (Fig. 22.5), appear

to plateau by the age of 8. Although the pattern recognition memory test requires a forced response to one of two choices, leading to the possibility that these Wndings may indicate a ceiling eVect on performance in older children, these Wndings are consistent with a large body of literature indicating early maturity of the temporal lobe hippocampal system that supports visual recognition memory (see Nelson (1995) for a review). Conversely, abilities that involve interactions between posterior brain regions and the frontal lobe, such as spatial memory span, show an intermediate developmental pattern, whereby there appear to be linear increments in memory span from 4 years of age (Fig. 22.6).

It is unclear how performance on the CANTAB relates to more general measures of intellectual function (e.g., IQ) in normally developing children. To estimate verbal and performance IQ, respectively, we have included two subtests from the Wechsler Intelligence Scale for Children-third

edition (WISC-III). These are the vocabulary and the block design subtests, both of which are highly representative of full-scale IQ scores (Sattler, 1992). Preliminary data indicate that there are not signiWcant correlations between Wechsler vocabulary scores and CANTAB performance in school-aged children (see Table 22.2). However, correlations between Wechsler block design scores and CANTAB performance are moderate with respect to several subtasks. These data further validate the CANTAB as primarily a measure of nonverbal information processing skill.

Having developed a normative database from which to infer the natural trajectory of functional development of memory and problem-solving skills in children, two goals remain. The Wrst is to assess whether CANTAB is a useful tool for identifying neuropsychologic impairments in pediatric populations and whether diVerential patterns of performance will be reliably observed in distinct clinical subsamples relative to normal controls. The second goal is to link both the normative and clinical performance para-

meters to brain functioning using electrophysiology, psychopharmacologic manipulations, and functional brain imaging.

Clinical validation of CANTAB in children with neurologic disorders

Adolescents with phenylketonuria

Phenylketonuria (PKU) is an inborn disorder of metabolism whereby an individual cannot metabolize phenylalanine. Because phenylalanine is present in a typical diet, toxic levels build up in the bodies and nervous systems of aVected individuals, resulting in mental retardation and other disturbances. Successful treatment requires early identiWcation of the disorder and dietary restriction of phenylalanine. Although restriction of phenylalanine appears to prevent mental retardation, there have been

several reports of deWcits in executive functions in individuals treated early and continuously for PKU. These deWcits may be a consequence of decreased dopaminergic activity in frontal cortex resulting from low tyrosine levels. Tyrosine is a necessary precursor for dopamine synthesis and may be decreased when phenylalanine levels are high (see Diamond et al. (1997) for a review). In a study directed at cognitive outcomes in adolescents and young adults with PKU (M. Luciana, J. Sullivan and C. A. Nelson, unpublished data), the CANTAB was administered to 18 individuals who had been treated since birth at the PKU clinic at the University of Minnesota Hospital, and the data were compared with those of several groups of age-matched controls. This study is unique in that only individuals with IQ values in the normal range were selected for study. Additionally, it is one of the few studies of long-term cognitive outcome in PKU. Data from young children treated

from birth for PKU have suggested deWcits in executive functions (Diamond et al., 1997), raising the concern of signiWcant cognitive deWcit even in individuals without impaired general intelligence.

Our data indicate that it is not necessarily the case that young adults with PKU show stable (i.e., trait) deWcits on tests of executive function relative to age-matched controls. Rather, performance on the Tower of London, spatial working memory, and set-shifting tasks appears to be moderated by phenylalanine levels, even after controlling for IQ levels. Those PKU probands with high phenylalanine levels, presumably resulting from inadequately controlled diets, demonstrate impaired executive function, while those with lower phenylalanine levels achieve scores that are comparable to those of normal controls. In contrast, performance on the pattern recognition memory, motor screening, and spatial memory span tasks appears to be

unrelated to phenylalanine levels and does not diverge from that observed in the normative sample. These preliminary Wndings using CANTAB support the view that subtle alterations in neurotransmission to the frontal cortex (see Diamond et al., 1997) may be present in young adults who have been treated early for PKU, and that these alterations exert subtle, but measurable, eVects on frontal lobe-mediated working memory and planning skills (M. Luciana, J. Sullivan and C. A. Nelson, unpublished data).

Children with histories of neonatal neurologic injury

Cognitive deWcits are believed to be common sequelae of high-risk births (Volpe, 1995; Ross et al., 1996). Yet the impact of early adverse neurologic events on school-aged children who are placed in mainstream classrooms has not been widely studied. To investigate these eVects, the CANTAB was administered to school-aged children (age 7±9 years) who were preterm infants treated at birth in the neonatal intensive care unit (NICU) at the University of Minnesota Children's Hospital. Perinatal events that place infants at risk for neurocognitive sequelae were quantiWed by chart review and included degree of prematurity, hypoxia, the presence of intraventricular hemorrhage, intrauterine growth retardation (birth weight z-scores #2),

development of bronchopulmonary dysplasia (supplemental oxygen requirement at 28 days), apnea, and neonatal seizures (Luciana et al., 1999).

Findings for the NICU survivors, relative to age-matched normal controls differ from those obtained from the PKU sample. In addition to demonstrating slower and lessaccurate responses on the motor screening task, as well as more errors in pattern recognition memory and decreased length of memory span (which would suggest subcortical and/or temporal lobe pathology), NICU survivors showed strikingly deWcient spatial working memory performance, making approximately 25% more forgetting errors than age-matched subjects in the normative sample (Fig. 22.7).

Additionally, the NICU group had a higher mean strategy score (indicating lower use of organized searching) than the normative group (Fig. 22.7). With respect to planning skills, NICU and normative groups did not diVer in their total number of moves made to complete Tower of London problems, but the average planning time across problem sets was longer in NICU survivors. Motor initiation times did not diVer between groups for yoked control problems, indicating that these latency diVerences are not a result of group diVerences in simple motor initiation processes in response to Tower of London task stimuli.

Several factors were considered as potential moderators

392M. Luciana and C. A. Nelson

of the deWcits found in the NICU survivors, including gender, handedness, birth weight, gestational age, and the presence of speciWc neurologic conditions. Although there was a signiWcant diVerence in the incidence of nonrighthandedness between the normative (17%) and NICU (28%) groups, handedness was not signiWcantly related to performance on any of the cognitive task measures in either group. Additionally, classiWcation of NICU survivors by birth weight (#1500, 1500±2499, 2500 g) did not signiWcantly distinguish cognitive performance on any task.

All but one child in the NICU group was placed either at or above their expected grade levels in school, but with some parental reports of academic diYculties. Since these diYculties were not assessed in a standardized manner, we conservatively identiWed each child with a reported academic problem (typically in the domain of reading skill) as ªpotentially learning disabledº. The NICU group was then divided into those children with potential learning disabilities and those without reported diYculty. No signiWcant diVerences in cognitive performance on any task measure were found between these subgroups. Similarly, when the group with potential learning disabilities was excluded from the overall analyses comparing the NICU and normative groups, signiWcant between-group diVerences remained with respect to pattern recognition memory, spatial memory span, slow planning on Tower of London problems, spatial working memory forgetting errors, and spatial working memory strategy score.

The results of this study suggest the presence of working and recognition memory impairments in school-aged children with histories of neonatal intensive care treatment. On the spatial working memory task, NICU survivors exhibited a high number of forgetting errors and failed to organize eYcient problem-solving strategies. The items of the spatial working memory task from which the strategy score is derived are those that demand a maximal degree of ªmulti-taskingº. To perform these problems most eYciently, one must remain highly attentive, recruit memory skills to remember previously selected and targeted locations, develop strategies to organize each search, and maintain use of a strategy once it is initiated. With increased task diYculty, initiation of organized search strategies is less likely to occur in the NICU group, and if strategies are initiated, they do not appear to be maintained, as indicated by a high degree of forgetting errors. These Wndings may not indicate a localized lesion process in NICU survivors but rather may indicate that the entire brain performs less eYciently under high demand. This interpretation would appear to be most consistent with our inability to associate any one neonatal risk factor with

cognitive performance. Continued follow-up assessment is necessary to determine whether these Wndings indicate a developmental lag in brain maturation that will resolve with age or whether executive problems will become increasingly apparent in these children as the demands of school and life become more rigorous. This latter pattern has been observed in other conditions that demonstrate a neurodevelopmental course (Weinberger, 1988).

CANTAB performance in children with early deprivation

Another group that we are studying includes foreign-born children between the ages of 4 and 10 who were institutionalized as young infants before subsequent adoption to the USA. Most of these children endured conditions of severe deprivation, in terms both of emotional nurturing and of environmental stimulation. The details of their histories are diverse, and their presentation is heterogeneous. Three children were adopted from the Philippine Islands and four were adopted from Romania. All had been in the USA for at least a year when tested and were referred for testing through the University of Minnesota's International Adoption Clinic. All adoptees were functioning in mainstream classrooms at the time of testing although not necessarily at their age-appropriate grade levels. The parents of children whom we have tested to date (seven children, ages 4±10 years) universally report that their children have diYculties with attention and with inhibitory control over behavior with respect to both cognitive and social functions. Data on general levels of intellectual function and other cognitive skills are unavailable. The absence of such information should be considered a serious consideration in guiding the design of future studies but does not detract from the severity of deWcits observed in this small pilot sample. To evaluate performance relative to age-matched children in our normative sample, the seven children were evaluated individually against norms generated by the appropriate age group in the normative sample. Performance was considered impaired if it was two or more standard deviations worse than the control mean. As indicated in Table 22.3, individuals with early deprivation display a variety of diYculties with CANTAB tasks, including increased motor response latencies, low memory spans, low accuracy of pattern recognition memory, and a high number of forgetting errors on the spatial working memory task. It is interesting to note that these children are similar in their performance to the children with neonatal neurologic injuries. Unfortunately, the adoptive parents rarely knew details of the birth and prenatal histories of their children, so the relative contributions of pre-

and perinatal complications verus early postnatal deprivation cannot be determined. However, we hope to be able to obtain this information in the form of a more comprehensive neurologic assessment as our study progresses.

Focal brain lesions

A body of literature has demonstrated that adults with focal lesions display distinctive patterns of performance on the CANTAB (Owen et al., 1990, 1991, 1995, 1996b,c). Whether CANTAB can be used to identify similar problems in children is unknown, as a comparable literature does not exist for children with acquired focal lesions. To begin to investigate such relationships, we have repeatedly tested one child, an 11-year-old girl, during the course of treatment for temporal lobe epilepsy that began after an episode of status epilepticus when she was 18 months old.

This patient was tested at three intervals: at age 11 years, several days prior to neurosurgery; 2 weeks following a right temporal lobe excision; and 1 year postsurgery. Functional and structural MRI and a complete neuropsychologic workup were also done. In adult patients with temporal lobe lesions, deWcits in visual memory are apparent (Owen et al., 1995). Relative to other children in her age group prior to surgery, this patient had a spatial memory span of Wve items (corresponding to the 50th percentile

relative to others (n5 23) in her age group). She performed at the 85th percentile with respect to spatial working memory error scores. Her performance on the Tower of London planning task was generally within the normal range. On the set-shifting task, she successfully completed all stages of the task but made a high number of errors when there was a demand for the Wrst within-category attentional shift. Her pattern recognition performance was at the 60±75th percentile.

Immediately postsurgery, several functions were noted to have markedly improved, while there were transient decrements in selective frontal lobe functions, namely in spatial working memory (where she demonstrated a high number of forgetting errors and decreased strategy use). Her spatial memory span increased from 5 to 7 items (70±90th percentile). Her pattern recognition memory performance was 100% correct, unchanged from her performance prior to surgery. Tower of London performance was unchanged. On the set-shifting task (where diVerent items were used from the original test session), she made negligible errors when demands in set-shifting were required. One year later, this patient was completely seizure free for the Wrst time since infancy. Memory span had dramatically increased to eight items, comparable with the 95th percentile for her age. Spatial working memory errors dropped by 45% (relative to immediate postsurgery values), and a