As recent advances in medical treatment have decreased deaths caused by stroke, increasing demands are put on understanding, predicting, and providing up-to-date rehabilitation to improve function, independence, and quality of life among stroke survivors. One of the most prevalent cognitive symptoms following stroke is impaired working memory. However, initial and long-term working memory function improvements have been reported post-stroke. Recent reviews highlight large individual differences in clinical and cognitive outcomes following stroke. With a current lack of understanding of the neuronal mechanisms of impaired working memory post-stroke, individual prediction of outcome and the potential for rehabilitation at the individual level is still very challenging in a clinical context.
With the over-arching goal of improving clinical prediction and rehabilitation of cognitive impairment and, in particular, working memory difficulties post-stroke, the current thesis has aimed to investigate and synthesize findings from studies of healthy development, pathological mechanisms, and potential surrogate biomarkers for working memory impairment and successful rehabilitation post-stroke.
The first empirical article of the thesis (Paper I) investigates how the maturation of the
adolescent brain supports cognitive development. Using functional MRI data from 754
adolescents aged 8 to 22 years, we estimated brain network centrality during a working memory task and during rest. Comparing the cognitive states, results revealed increased centrality in frontoparietal and cingulo-opercular networks and decreased centrality in sensorimotor nodes due to task engagement across the age span. Notably, the results revealed a positive association between age and the estimated network centrality in the Insular, Cingular, and Opercular brain regions. This age-related association was found using data obtained during a resting-state condition and task performance, suggesting a relative state-independent developmental effect on hub-related brain activation and recruitment relevant to normative cognitive development.
Building on the functional connectivity findings in the first paper, the second empirical
paper (Paper II) investigated associations between the distribution and degree of disruption of structural pathways in the brain, termed disconnectivity, following stroke, and cognitive function. We used data from 102 stroke survivors, including lesion masks and cognitive status, as assessed using Montreal Cognitive Assessment (MoCA). Based on the lesion masks, we estimated probabilistic structural disconnectivity maps for each patient. Multivariate analysis by partial least squares indicated that estimated disconnectivity implicating opercular and insular brain regions, also identified in Paper I, was associated with impaired cognitive performance. This finding suggests that stroke lesions implicating brain regions or pathways projecting to opercular and insular brain regions are more likely to result in cognitive impairments than lesions not projecting to those regions.
Bridging the findings from the neuroimaging findings reported in the two initial papers
to a clinical context relevant to cognitive rehabilitation and outcome following stroke, the work presented in Paper III tested for associations between treatment response after computerized cognitive training primarily targeting working memory and longitudinal task-based fMRI. In total, 54 stroke patients were recruited from clinical hospital records. They completed a comprehensive 3-week protocol of cognitive training and cognitive and brain imaging assessments, including an experimental fMRI-based attentional task with varying demands on working memory before and after the training. The results revealed no clear association between task-related brain activation before training and treatment response and no clear evidence for training-induced alterations of functional brain activation during task performance.
In summary, the results from the empirical papers of the current thesis support the
implication of cingulo-opercular brain networks for cognitive functions and working memory in youth and highlight that stroke lesions projecting to overlapping cingulo-opercular brain network nodes may be particularly relevant for post-stroke cognitive impairment. Our results do not provide evidence that fMRI brain activation during task performance can be used for prediction or monitoring cognitive gains in response to computerized cognitive training in this sample of stroke patients.
In conclusion, the findings of the current thesis support further investigation of potential surrogate biomarkers for cognitive impairment following stroke, where altered structural and functional brain connectivity may provide clinically relevant information.