BRAIN INJURY, VOL.
18,
NO.
10 (OCTOBER 2004), 957–974
Investigating the neurobiological basis of cognitive rehabilitation therapy with fMRI L . K . L A A T S C H y, K . R . T H U L B O R N z, C . M . K R I S K Y }, D . M . S H O B A T z a n d J. A. SWEENEY} y Department of Neurology and Rehabilitation z Center for Magnetic Resonance Research } Center for Cognitive Medicine, University of Illinois, College of Medicine, Chicago, Ill, USA (Received 10 June 2003; accepted 20 January 2004) The neurobiological changes occurring during cognitive rehabilitation therapy (CRT) have yet to be systematically studied. In the present study, functional magnetic resonance imaging (fMRI) was used to demonstrate brain plasticity in response to CRT (n ¼ 5) following mild traumatic brain injury. Neuropsychological tests and two fMRI activation tasks, a visually guided saccades and a reading comprehension task, were employed pre- and post-CRT. CRT was used to systematically address the identified deficits in visual scanning and language processing. As hypothesized, changes in the pattern and extent of activation within expected neuroanatomical areas occurred post-CRT. Changes in fMRI activation are discussed for each subject and related to changes on neuropsychological measures. This study demonstrates how fMRI can illustrate the neurobiological mechanisms of recovery in individual subjects. The variability in subject responses to CRT supports the notion of tailoring rehabilitation strategies to each subject in order to optimize recovery following brain injury.
Introduction Research has demonstrated that cognitive rehabilitation therapy (CRT) assists recovery from cognitive deficits associated with acquired brain injury [1–3]. A broad, generally accepted definition of CRT is: systematic use of well-defined structured activities designed to improve higher cerebral functioning in a subject with brain injury or to help the individual accommodate for their deficits by teaching methods of compensation [4]. Although CRT is a standard treatment following acquired brain injury, understanding the biological basis of CRT is essential to maximize treatment efficacy. There is limited research relating CRT to basic neurobiological principles of recovery and mechanisms of plasticity [5]. CRT provides an enriched environment designed to promote recovery in known areas of cognitive deficit. Research in
Correspondence to: L. K. Laatsch, PhD, Associate Professor of Psychology, Department of Neurology and Rehabilitation, University of Illinois, College of Medicine, Chicago, Ill, USA. e-mail:
[email protected] Brain Injury ISSN 0269–9052 print/ISSN 1362–301X online # 2004 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/02699050410001672369
958
L. K. Laatsch et al.
animals has demonstrated that exposure to enriched environments is associated with a beneficial effect on cognitive function post-lesion [6], locomotion after bilateral sensorimotor cortical lesions [7] and glial elaboration of brain derived growth factors [8]. Human research has also demonstrated experience-induced functional changes occur rapidly in motor, language and visual systems after CVA [9, 10]. A range of mechanisms including unmasking of existing circuits, functional reorganization, modification of synaptic connectivity and inter-hemispheric competition underlie neuroplasticity after brain injury and are believed to be involved in recovery [3]. Early in the development of CRT, a distinction was made between (1) Direct restoration; involving reducing the effects of the deficit through systematic exposure to specific cognitive activities and (2) Compensation strategy training: involving substitution of function using compensation through instruction concerning application of mental strategies to improve the cognitive activity [11]. Whereas CRT efforts aimed at direct restoration generally are believed to be associated with restitutive reconnection, compensation training is thought to be associated with reorganization/ redistribution and use of adjacent and remote neuronal circuits. Although it is understood that there is overlap in the types of CRT during treatment, particular CRT techniques are recommended following deficits in specific cognitive domains. For example, when the focus of CRT is sustained attention and visual scanning, repetitive practice with graded visual stimuli is generally utilized [3, 12]. CRT guidelines suggest that rehabilitation of memory is most beneficial when there is a focus on compensative memory strategies [2, 13]. In contrast, rehabilitation of expressive language skills involves both restorative therapy and compensative strategy training depending on the specific type of language deficit [2]. fMRI may be an appropriate tool to further understanding of the neurobiological basis of CRT. In a large volume of systematic studies, fMRI has been shown to provide information concerning the network of discrete brain areas involved in cognitive activities in normal subjects [14–18]. In contrast, published use of fMRI with subjects who have non-progressive acquired brain injury has been limited to case studies of stroke [5, 9, 10, 18, 19] and traumatic brain injury (TBI) [20–22]. No studies to date have involved fMRI pre- and post-CRT with subjects who have a history of TBI. Support for the use of fMRI as a technique for understanding the neurobiological basis of CRT is provided below through a review of published fMRI studies involving subjects with TBI. The first publication demonstrating the use of fMRI with TBI subjects was McAllister et al. [20]. Using the N-Back working memory paradigm with recently injured mild traumatic brain injury (MTBI) subjects, the researchers examined changes in brain activity during exposure to an auditory N-back task that had three levels of difficulty. Even when the subject’s performance accuracy was equal to the matched, healthy control group, they exhibited increased bilateral frontal and parietal activation in response to the cognitive challenge. This group of researchers expanded their work with MTBI subjects, including another level of difficulty [21]. As expected on this very difficult task, there was a significant decline in accuracy for both the controls and subjects with MTBI. While during the moderate difficulty task there was an increase in overall activation, in response to this new level of difficulty the subjects with MTBI responded with less of an increase in activation during the new level of difficulty compared to controls.
Cognitive rehabilitation therapy with fMRI
959
McAllister et al. [21] hypothesized that limitations in the allocation of processing resources, as illustrated at the highest level of N-back difficulty, might underlie the memory complaints common after MTBI. Christodoulou et al. [22] studied subjects with severe TBI who were many years post-injury using a working memory task that required sustained attention to digits. While the healthy controls’ activation on fMRI was primarily within the left cerebral hemisphere, the TBI subjects’ activation was more extensive and generally within the right hemisphere. Current fMRI research suggests that subjects following TBI need to recruit more and possibly diverse brain resources in response to memory tasks during imaging. While sufficient support is provided for the use of fMRI with subjects who have TBI, Price and Friston [23] advise caution with subjects with known neuropsychological impairments. Subjects with brain injury will generally have reduced performance relative to controls on neuropsychological tasks. In order to examine the activation pattern elicited by a similar cognitive probe in the MR scanner, the subjects will need to demonstrate near-normal performance on the matched neuropsychological test. Without this criteria, it would be difficult to characterize what mental activity is being performed during fMRI in response to the cognitive challenge using block activation tasks and, in addition, frustration with the task may result in aberrant activation patterns and movement. Therefore, in the present study, a full neuropsychological evaluation was completed with all potential subjects prior to imaging to determine the subject’s specific level of neuropsychological impairment and appropriateness for the chosen fMRI tasks. In addition, fMRI tasks with low complexity and difficulty were used, which subjects were expected to be able to perform adequately. In this study, fMRI was used to characterize activation patterns longitudinally in response to CRT in a series of clinical subjects with neuropsychological impairments following TBI. Two fMRI block activation tasks, extensively researched and established in control subjects, were employed in this study [24–26]. A reading comprehension (RC) task, similar to the task used with subjects with stroke and matched control subjects [10], and a visually guided saccade (VGS) task that has been studied extensively in healthy and clinical populations [24–26] were employed. The extensive control subject data allowed for the definition of expected areas of activation in both the RC and VGS tasks. Additionally, the two activation paradigms were of interest because they can be related to functional cognitive activities, reading and visual scanning, which are often impaired post-TBI. Although an individualized CRT plan was established for all subjects based on the neuropsychological evaluation, all five subjects participating in this study were found to have deficits in language processing and visual scanning. The subjects were selected so that there would be relative consistency in the CRT intervention across all subjects to facilitate comparison. Imaging and neuropsychological testing occurred before and after CRT. It was hypothesized that changes in the fMRI activation network pattern and extent would occur during the RC and VGS tasks following systematic rehabilitation of language and visual processing skills. Furthermore, it was hypothesized that initial cognitive deficits in language processing and visual scanning would demonstrate improvement upon re-testing at the completion of CRT.
L. K. Laatsch et al.
960 Methods Subjects
Five subjects are presented in this study. All subjects were referred by their physician for outpatient CRT at variable times following injury. After the initial neuropsychological evaluation, consent was obtained using standardized procedures required by the Office for Protection of Research Subjects at our institution. All subjects met the criteria established by the American Congress of Rehabilitation Medicine for MTBI [27]. Demographics and mode of injury of the five subjects are provided in Table 1. Details concerning the nature of each subjects’ MTBI are provided under each individual subject in the Results section. Measures Neuropsychological testing measures A brief, standardized neuropsychological evaluation was used as part of the research protocol and administered pre- and post-CRT. This brief battery included target neuropsychological measures that were similar to the two fMRI activation tasks administered. The battery was part of a full neuropsychological evaluation designed to highlight subject’s cognitive strengths and weaknesses. In addition, a broad measure of psychiatric status, Symptom Checklist 90-R [28], was administered to rule out current psychiatric symptoms that might interfere with progress in rehabilitation. Three neuropsychological measures of visual processing were administered. The Performance Intelligence Scale from the Wechsler Adult Intelligence Test -III [29] provided a summary measure of visual processing. Two tests of visual attention and scanning were given in the research battery, Trails A and B [30] and the Digit Vigilance Test [31], because a visual scanning activation task was used as an activation task. While Trails A and B provide measures of brief visual scanning with and without a distracter present, the Digit Vigilance Test, which requires scanning for a specific digit, provides a measures of extended visual scanning speed and accuracy. Expressive language processing measures included the verbal subtests from the Wechsler Adult Intelligence Test-III [29] and the FAS Verbal Fluency Test [32]. Also administered was a measure of oral reading [33] to assure sufficient reading for the RC activation task administered at a 4th grade level. All subjects
Table 1.
Demographics of subjects with MTBI
Subject
Age
Handed
Years Education
Race
Gender
Injury
Mos Post
Hx
CRT hours
1 2 3 4 5
27 45 52 20 20
R R L R R
16 12 16 13 14
W AA AA W W
F F F M M
MVA Fall MVA Sport Inj. Sport Inj.
29 2 12 2 2
BI (1) Seiz NS NS BI (5)
24 12 20 21 12
Mos Post ¼ number of months between current MTBI and initial neuropsychological evaluation; Hx ¼ prior neurological history; BI ¼ prior brain injury (number of prior brain injuries); Seiz ¼ hx of seizures; NS ¼ no reported neurological history.
Cognitive rehabilitation therapy with fMRI
961
performed at the 12th grade or above on the reading test. Age, education and gender normative data were used to evaluate neuropsychological status pre- and post-CRT [34]. fMRI activation measures Both the VGS and RC tasks were presented using a block design in which periods of central fixation were alternated with periods of task stimuli every 30 seconds. During the VGS task, subjects were asked to visually track a white dot on a black background moving every 500 ms in 4 steps to one of five potential locations (0 , 4 and 8 ) along the horizontal plane. The direction of target movement (right or left) was unpredictable except after the 8 locations, when the target always stepped back toward the centre of the screen. In the fixation condition, subjects were instructed to fixate on a white cross presented in the centre of the screen. Blocks of fixation and saccades were alternated every 30 seconds (starting with fixation) for a total task duration of 4.5 minutes. Eye movement was monitored throughout the task using a remote camera [24]. This task is well established in healthy control subjects as a measure of visual attention and visually directed eye movement [25], older adults and Alzheimer’s disease [26]. A distributed activation pattern in neocortical regions has been consistently demonstrated in healthy control subjects including the frontal eye fields, supplemental eye field, posterior parietal areas bilaterally and the cerebellum. The RC task has been previously used with stroke subjects and control subjects [10, 24]. This task involves six cycles of two conditions, a central fixation condition (white cross) and a condition in which a simple sentence is presented with a true– false question concerning the sentence. The subject was asked to read the sentence silently and respond true or false by pressing a finger switch, which was previously placed in the subject’s dominant hand. An example of the stimuli used in this task is, ‘The fox chased the rabbit. Did the fox chase the rabbit?’ The task requires 9 minutes for completion. The paradigm was chosen because in it activates Broca’s and Wernicke’s areas, the intraparietal sulcus, the frontal eye fields and vision areas in control subjects [10, 24]. The tasks described above were administered pre- and post-CRT in the identical format, but with different language stimuli post-CRT. In both tasks, control subject data were used to establish expected areas of activation and to determine if the pre- or post-CRT activation pattern was different from established control subject data. Intervention Cognitive rehabilitation therapy An individualized Developmental Metacognitive Approach to CRT was used in this study [35]. The technique involves an hierarchical skill development approach and a metacognitive focus, involving thinking about one’s own thinking, throughout treatment. All CRT sessions took place in an isolated office space weekly for 50–60 minutes with the first author. This specific CRT approach was used to systematically address the identified deficits in visual scanning and language processing in the five subjects included in the study. Therefore, the intervention was consistent in terms of rehabilitation procedures. Target criterion tasks, described below, were used to identify competence in each cognitive domain. Target criterion tasks were all computerized CRT tasks with speed and accuracy measures
L. K. Laatsch et al.
962
and they were used to measure progress in each cognitive domain. The number of CRT sessions varied across subjects and was dependent on the individual subject’s progress on the rehabilitation tasks and ability to complete assigned home exercises. Rehabilitation of visual processing deficits involved systematic, repeated presentation of increasingly difficult visual tasks to enhance restoration of visual processing speed, visual attention, visual perception and visual scanning. Restoration was assisted though repetitive use of graded tasks involving diverse visual stimuli. Visual processing speed, visual scanning, visual perception, detail perception and visual spatial analysis were emphasized using graded-difficulty computerized and non-computerized tasks. When appropriate, computer programs were utilized to demonstrate the use of a specific strategy to improve cognitive function [36, 37]. Home exercises involved timed visual scanning activities involving visual-spatial exercises, books, puzzles or specifically-chosen computer programs. The target criterion tasks in this domain involved timed visual-perception tasks; (1) Visual Processing Speed: Flasher [36], >75% accurate, best speed
