Clinical

Breakdown of working memory function caused by neuropathological processes

A greater understanding of normal cognitive function will be achieved by a greater understanding of cognitive dysfunction due to neuropathological conditions. Likewise, identifying the neural mechanisms underlying normal cognitive function provides a foundation for developing theory-driven approaches toward remediating cognitive deficits. Thus, our laboratory has spent a considerable amount of effort towards characterizing working memory impairments in various neurological conditions as well as during normal aging. For example, we have documented and characterized the working memory impairments in patients with various conditions such as traumatic brain injury (e.g., McDowell et al., 1997), stroke (e.g. D’Esposito et al., 1996), multiple sclerosis (e.g., D’Esposito et al., 1996), Parkinson’s disease (e.g., D’Esposito & Postle, 2000), attention deficit disorder (e.g. Sheridan et al., 2007), addiction (e.g., Mitchell et al., 2005) and normal aging (e.g. Rypma et al., 2000; Gazzaley et al., 2007).

For example, in a study of healthy older adults (ages 60-80), we found a selective age-related neural deficit in the ability to suppress or ignore task-irrelevant information during working memory tasks documented using fMRI (Gazzaley et al., 2005). On a task in which individuals must remember faces but ignore scenes, older individuals encode scenes into memory despite being instructed not to do so. Furthermore, we observed that those older adults with a greater suppression deficit, as documented with fMRI, had a more significant working memory deficit. Also, those older adults with more difficulty suppressing or ignoring irrelevant information during the working memory task were more likely to remember this information at a later time (as compared to younger adults) when given a surprise post-experiment memory test. This finding that older adults had increased incidental long-term memory of information they were told to ignore is consistent with the neural data that task-irrelevant representations were not appropriately suppressed.

Thus, this very specific finding that precisely characterizes both the behavioral deficit and its neural underpinning provides a foundation for developing cognitive interventions for remediating cognitive deficits. We have also found a very similar deficit to this (although different in degree) in patients with more severe frontal systems dysfunction, such as those with traumatic brain injury or stroke. Thus, our approach is to identify a brain system impairment that may be affected by many disorders. This is not a traditional approach in the development of medical treatments since most focus on developing treatments that are specific to the particular disease. However, we believe this approach will have a greater impact on helping a wider range of neuropathological conditions that affect cognitive function. In this way, all disorders and conditions that exhibit frontal systems dysfunction (regardless of underlying neuropathological causes) could potentially respond to a targeted frontal systems intervention. Of course, this type of intervention can also be coupled with disease-oriented therapies aimed at specific neuropathological processes (e.g., anti-amyloid agents for Alzheimer’s disease).

Cognitive interventions to improve cognitive function

Unfortunately, there is currently no effective cognitive therapy for treating patients with cogntive deficits secondary to frontal systems dysfunction. Rehabilitation of cogntive deficits through cognitive training may be considered a process of guiding mechanisms of plasticity for the ‘re-integration’ of functional PFC networks (Ances & D’Esposito, 2000; Chen et al., 2006; D’Esposito et al., 2006). That is, mechanisms of plasticity following any neuropathological process include the possibilities of re-organization of available network components, or the generation of new network components. Mechanisms of plasticity at the cellular level that may support re-organization or regeneration include alterations in metabolism, synaptogenesis and synaptic pruning, including growth of new long-distance projections, and perhaps neurogenesis. Ultimately, for these neuronal changes to affect neurological function, they must be translated into changes in the functioning of networks of neurons. Effective cogntive training may guide these neuronal changes to achieve functionally integrated networks and coherent behavioral output.

When a brain injury (or normal aging) affects the networks involving working memory function, cognitive rehabilitation treatments may guide the neural mechanisms discussed above to enhance the functional re-integration of networks. At least two different levels of change may support re-integration of network function. First, integration of non-injurred PFC regions connected to critical posterior brain regions may be supported by synaptic re-organization and synaptogenesis. Second, with respect to functional recovery, it is possible that there is some redundancy in PFC circuitry, such that non-injured areas may be able to re-organize to take over function previously supported through other regions. Cognitive training would, in essence, help in making damaged, poorly integrated population of neurons into more efficient, better integrated functional networks for the performance of relevant tasks.

Our lab has implemented several types of cognitive interventions, one such being goal management training,  developed  by XXX. This training approach focuses on improving PFC function in the context of achieving particular goals. The content of the goals, and thus the content of the tasks, can be individualized to each patient. Neural sub-processes involved in goal management may be trained regardless of the specific content. For example, patients are asked to go through five main steps. First, patients are asked to stop, and explicitly outline the goals of their actions. Patients are guided in generating personally-relevant goals, which may include achieving everyday tasks such as planning a meal or making a doctor’s appointment. Subsequent management of goal-generated tasks would require steps that would further engage PFC networks. These steps include generation of sub-goals and listing of associated tasks; learning and recalling goals and sub-goals; and executing the goal-oriented tasks. These steps may require processes including sustaining attention, holding information in working memory, and self-evaluation of performance through comparing the intended outcomes with actual outcomes.

Over the past year, our neurorehabilitation team has been enrolling patients with traumatic brain injury and healthy older adults in our cognitive intervention programs to treat cognitive and memory deficits attributable to frontal systems dysfunction. Additionally, we have partnered with the Center for Brain Health the University of Texas to increased our enrollment in this program. This 5-week cognitive training program consists of 20 hours of group training (2 hr sessions, 2 days per week), 3 hours of individual training (1.5 hours at the beginning, 1.5 hours halfway through training), and approximately 20 hours of home practice (approximately 1 hr per day). In addition, we obtain cognitive testing and fMRI scans before training, immediately after, and 6 months after training. Thus far, the results of our interventions have been extremely promising. On neuropsychological assessments, all patients and older adults have significantly improved on various measures of attention, working memory and executive function. Our traumatic brain injury patients have reported improvement in their ability to perform tasks in daily life and, in particular, reduced distractibility. Moreover, using our fMRI biomarkers of frontal systems function (see below), we have been able to demonstrate alterations in functional brain responses following our cognitive intervention that accompanies these cognitive and functional gains. Our ultimate goal is to refine these interventions and make them widely available to clinicians. 

The pharmacology of cognitive function

The function of the cerebral cortex is clearly influenced by the diffuse inputs from brainstem neuromodulatory systems mediated by neurotransmitters such as dopamine and acetylcholine. Yet, the relationship between neurotransmitter function and cognition remains underspecified. Furthermore, few targeted pharmacological treatments for cognitive deficits are available to clinicians. A key to understanding the neural basis of cognitive function and developing effective drugs that can improve cognitive function will arise from an understanding of how cognitive function is modulated by such brainstem neurochemical systems. Based on the anatomical distribution of brainstem dopaminergic projections, there is a logical basis for proposing a role for dopamine in working memory. The mesocortical and mesolimbic dopaminergic systems project with the highest concentration to the prefrontal cortex. The link between dopamine, working memory, and prefrontal cortical function has been established in animal studies. First, in monkeys, depletion of prefrontal cortexdopamine or pharmacological blockade of dopamine receptors induces working memory impairments. This impairment is as severe as in monkeys with prefrontal cortical lesions and is not observed in monkeys in which other neurotransmitters, such as serotonin or norepinephrine, are depleted. Furthermore, dopaminergic agonists administered to these same monkeys reverse their working memory impairments.

One approach for assessing dopamine’s influence on cognitive function in humans is by testing Parkinson’s disease patients “on” and “off” their dopaminergic replacement medications. Using this approach, we have repeatedly demonstrated that Parkinson's patients improve on working memory function after taking their dopaminergic medications (e.g., D’Esposito & Postle, 2000). Our second approach is to administer dopamine receptor agonists to healthy young volunteers. Two such drugs we have used are bromocriptine (a D-2 agonist) and pergolide (a D-1 and D-2 agonist). In our first studies of each drug, we demonstrated that healthy young human subjects, when given bromocriptine (Kimberg et al., 1997) or pergolide (Kimberg and D'Esposito, 2003) perform better on working memory tasks when compared to when they are given a placebo. In these studies, the dopaminergic medication had a very specific effect on working memory since it had no effect on other cognitive abilities.

In each of these studies, we discovered that the effects of dopaminergic agonists on prefrontal cortex function were not the same for all subjects but interacted with the subject's working memory capacity. That is, subjects with a lower baseline working memory capacity exhibited cognitive improvement on the drug, while those with a higher baseline working memory capacity worsened on the drug. This relationship between working memory capacity and the effects of bromocriptine on working memory performance has been replicated many times since by us and other labs. Moreover, we have demonstrated that lower working memory capacity reflects lower dopamine synthesis capacity as measured by PET scanning (Cools et al., 2008). We have further been able to explain that individual differences in working memory function is accounted for by estrogen levels (in woman) and polymorphism of the COMT (catechol-O-methyltransferase) gene. Further characterizing these individual differences will be critical for developing rational therapies to treat individuals with memory deficits due to dopamine depletion. Our more recent efforts have been to combine the administration of dopaminergic agonists with fMRI scanning (e.g Kimberg et al., 2001; 2003; Gibbs et al., 2005; 2006). This method has allowed us to identify the neural mechanisms underlying the drug effects that we have observed. For example, in one study (Cools et al., 2007), we found that dopamine modulated striatal activity during the flexible gating of information into working memory, whereas dopamine modulated PFC activity during the stable maintenance of working memory representations.

Pharmacological interventions to improve cognitive function

Based on our findings with dopaminergic agonists in healthy individuals, we have tested the effects of these drugs in patients with neurological disorders. For example, in one study (McDowell et al., 1998), patients who suffered prefrontal cortex damage from traumatic brain injury were assessed on and off bromocriptine while performing several clinical experimental measures of prefrontal cortex function (e.g., Stroop task, the Wisconsin card sorting task, the Trail Making task, dual-task). Significant improvement in the performance of traumatic brain injury patients was observed on bromocriptine, as compared to placebo, on all tasks that required significant demands on working memory and executive control processes thought to rely on intact prefrontal cortex function. In contrast, bromocriptine did not improve performance on other cognitive measures. This finding provides evidence that dopaminergic agonists can be an effective therapy in treating the cognitive deficits found in patients who have suffered prefrontal cortex damage.

In another study, we found that children with attention-deficit disorder (ADD) perform with greater accuracy on working memory tasks while on their stimulant medications, presumably due to dopaminergic stimulation (Sheridan et al., 2010). Using fMRI, we have been able to identify the neural correlates of these behavioral effects. Specifically, similar to our studies of normal aging, ADD patients generally show a decrease in neural efficiency (i.e. increase PFC activity with poorer performance), which improves on medication.

Development of fMRI biomarkers to assess and guide cognitive and pharmacological interventions

Given that there are many potential neural mechanisms for recovery of function, it is imperative to develop biomarkers that would not only give insight into these mechanisms but could guide cognitive and pharmacological therapy and assess the effectiveness of such treatments. For example, we propose that after effective rehabilitation training aimed at enhancing PFC function (e.g. cognitive or pharmacological), PFC activity should become better integrated, and there should be evidence of increased anterior-posterior functional connectivity. As evidence of improved functional integration, there should be increased task-relevant modulation of posterior brain activity as measured by fMRI. Task-related activity in the PFC may be actually increased with training or drugs, relating to increased strength or capacity to exert modulatory control. Measurements of PFC activity and connectivity, as well as metrics that reflect the large-scale organization of the brain, can serve as potential biomarkers for recovery of function. For example, our lab has demonstrated that measurements of network connectivity can serve as imaging “biomarkers”, which can predict who will or will not respond to therapies to improve cognition (Arnemann et al., 2015; Gallen & D'Esposito, 2019).

Key References:

Arnemann KL, Chen AJ, Novakovic-Agopian T, Gratton C, Nomura EM, D’Esposito M. Functional brain network modularity predicts response to cognitive training after brain injury. Neurology, 84:1568-74, 2015.

D’Esposito M, Postle BR. Neural correlates of component processes of working memory: evidence from neuropsychological and pharmacological studies, Attention & Performance XVIII “Control of Cognitive Processes”, (Eds. S. Monsell, J. Driver), 2000.

Gallen CL, D’Esposito M. Brain Modularity: a biomarker of intervention-related plasticity, Trends in Cognitive Sciences, 23:293-304, 2019.

McDowell S, Whyte J, D’Esposito M,. Differential effects of a dopaminergic agonist on prefrontal function in head injury patients, Brain, 121:1155-1164, 1998.

Sheridan MA, Hinshaw S, D’Esposito M. Stimulant medication and prefrontal functional connectivity during working memory in ADHD: a preliminary report, Journal of Attention Disorders, 14:69-78, 2010.