Traumatic brain injury (TBI) is a common phenomenon, characterized by a variety of causes, severities, anatomic distributions, and consequences. Despite this variability, however, certain behavioral consequences are commonly observed, including deficits in memory, attention, and executive function. To better understand the pathophysiological relationship between impaired memory and executive dysfunction following TBI, our TBIMS team conducted a functional magnetic resonance imaging (fMRI) project wherein we measured brain activity in participants with TBI and matched controls as they performed word list learning tasks.

Our project had two primary goals. First, we sought to assess the effects of a traumatic brain injury (TBI) on the ability to activate particular regions in the brain associated with verbal learning and semantic organization (based on category, e.g., foods, tools, etc.) of verbal materials. Second, we wished to determine whether or not brain activity as measured by fMRI could be used to help predict outcomes from a novel cognitive memory group therapy. Three groups of participants were included in the study: (1) fifty-four individuals with chronic TBI (greater than one year post-injury), (2) twenty healthy controls matched to a subset of group 1 participants on age, sex, and education, and (3) another group of individuals with TBI (currently numbering thirty-two) who did not undergo fMRI or therapy sessions ("TBI controls").

The study consisted of five phases. During the initial phase, each participant underwent a series of cognitive evaluations, including tests of memory and executive functioning. This data served as baseline information for all three groups of participants.

The second phase consisted of fMRI scanning (for the first two groups only). Participants were scanned while learning word lists under three conditions differing in terms of the likely use of a strategy, semantic grouping of the words, which reflects one aspect of executive function. Semantic grouping is the strategy of categorizing the words according to their meaning, e.g, foods, tools, etc. Individuals were scanned while encoding (that is, memorizing) word lists, with recall and recognition of the words assessed after each scanning run. To vary the amount of semantic processing, participants learned: 1) semantically unrelated words (Unrelated condition), which could not be grouped; 2) semantically related words (such as dog, fish, bird, turtle, which could be grouped), but with no instructions other than to remember as many words as possible (Spontaneous condition); and 3) semantically related words following training on the use of the semantic grouping or clustering strategy (Directed condition). In the Directed condition, individuals were specifically requested to try to group the words together into semantic categories as they learned the words. Thus, the unrelated condition was expected to result in the least amount of reorganization of list items during learning, the Spontaneous condition somewhat more, and the Directed condition the most.

The third phase of the study consisted of therapy sessions (group 1 participants only). Individuals with TBI participated in 12 group-based therapy sessions, 90 min. each, twice per week for six weeks. These sessions emphasized semantic organization and other internal memory strategies (i.e., elaboration and imagery) from encoding, storage, and retrieval perspectives.

The fourth and fifth phases were cognitive post-testing sessions with all participants with TBI (groups 1 and 3). The first such session was conducted immediately following therapy sessions or 6 weeks after the first test in the TBI controls, who did not attend the therapy group. The second post-test session was conducted one month after therapy sessions ended or 10 weeks after the first test in the TBI controls. During these two sessions, participants re-took the same tests they had taken in the first phase of the study. Test scores from these post-tests served as outcome data points.

For our first project goal, we sought to identify differences in brain activity while participants with TBI learned word lists as compared to matched controls (group 1 versus 2). As already mentioned, verbal learning and strategic processing deficits are common sequelae of traumatic brain injury (TBI), though the exact mechanisms underlying such deficits remain poorly understood. We obtained fMRI scans in 25 individuals from group 1 with TBI at least 1 year post-injury and compared them with scans in our 20 matched healthy controls (group 2). Memory test scores for performance on recall, recognition, and semantic clustering improved significantly for all participants groups according to the condition in which they memorized the words: In the Unrelated condition they memorized the fewest and in the Directed condition they memorized the most words. Memory tests performance was in-between for all groups in the Spontaneous condition. This is consistent with previous research and reflects an improved ability to remember words when they are encoded using additional semantic processing. Individuals with TBI exhibited impaired yet parallel memory test performance relative to control participants.

While the overall brain activity during verbal learning was quite similar between the two groups, fMRI measures of brain activity during verbal encoding revealed decreased activity in participants with TBI relative to controls in left anterior dorsolateral prefrontal cortex (DLPFC) and in a region spanning the left angular and supramarginal gyri in the parietal lobe. Functional connectivity analysis between these two regions revealed evidence of a dramatic functional breakdown in the connectivity between the DLPFC and other regions specifically when participants with TBI were directed to utilize the semantic encoding strategy. That is, these two brain regions showed highly coupled activity in the Directed condition in controls, but no coupling at all in the Directed condition in the individuals with TBI. It would thus appear that, following TBI, the DLPFC appears to decouple from other brain regions specifically when strategic control is required. The regions do remain anatomically connected, since they are coupled during the other conditions (Unrelated and Spontaneous). The reason for this decoupling in only this condition is not fully understood. However, based on this finding we hypothesize that therapeutic approaches designed to enhance coupling between these regions under such conditions may aid cognitive rehabilitation following TBI.

Our second project goal was to evaluate the ability of fMRI to provide predictive value for rehabilitation outcomes over and above standard predictors of outcome. In this case, we utilized the fMRI scan data from all fifty-four individuals with TBI (all of group 1). We selected our primary outcome measure to be the Hopkins Verbal Learning Test (HVLT) delayed recall score, which is a test of word list learning that is sensitive to improvements in semantic organization abilities. When setting up models to predict outcome, we used the following standard predictors of outcome: pre-therapy HVLT, age, education, injury severity, plus task-related fMRI activation. Based on previous research implicating left frontal cortex in such tasks, we investigated two regions in particular: left DLPFC and (separately) left ventrolateral prefrontal cortex (VLPFC).

As expected, baseline HVLT was a significant predictor of outcome, as was having a severe injury. However, we also found a significant quadratic (inverted-U) effect of fMRI in the VLPFC such that both high and low activations on fMRI were associated with poorer outcomes. Importantly, this effect was significant after accounting for all the other predictors. This provided initial evidence suggesting that regional brain activity may help predict who might benefit from a memory rehabilitation program over and above standard predictors of outcome. It also suggests that nonlinear phenomena may play an important role in helping select and guide individual rehabilitation therapies. In addition, it is notable that acute injury severity appears to provide predictive value in rehabilitation outcome greater than ten years post-injury.

An important secondary goal of our project was to assess the efficacy of our therapy. For this evaluation, we compared the pre- to post-therapy scores on the HVLT as well as the Rivermead Behavioral Memory Test (a test of general memory function), and Trailmaking B (a test of executive function). Following therapy in our participants with TBI (group 1), we found a 43% improvement in HVLT and 15-20% improvements in the more general tests. Preliminary analysis of our TBI control subjects (group 3), who did not participate in therapy sessions, has revealed no change in performance in all of the test scores.

Overall, we have good evidence that our therapy was effective for improving memory function, at least up to one month after the end of the therapy sessions. We have also gained initial information about alterations in brain activity that may underlie differences in memory and semantic processing in participants with TBI. And finally, we have new evidence that fMRI may provide predictive value about rehabilitation outcomes.

While relatively broad in scope, this represents a single, initial study on the topic of neuroimaging in TBI. At this stage of the research, the immediate implications for survivors and their families are limited. But, these findings suggest that, eventually, it may be possible to tailor memory and other cognitive therapies to match the individual strengths of people with TBI. That is, clinicians may be able to use functional imaging to predict whether or not a certain therapy will be effective for a particular individual. Some day, it may even be possible for clinicians to use functional imaging as a biofeedback method to give patient information about whether or not he or she is using the brain in a way that is likely to bring success.

More immediately, we have additional evidence that a group therapy that teaches ways of using the meanings of words to help with memorization can improve memory in people with TBI. More research needs to be done on this kind of therapy, but it looks very promising.

back to top