Persistent cognitive dysfunction significantly impacts the quality of life for many chronic stroke survivors. Translating therapeutic strategies to improve cognitive function is hampered by the disparate cognitive assessments used in animals and humans. Current assessments in animal models, such as the Morris water maze and novel object recognition, have limited similarities to clinical assessments in stroke survivors and often suffer from high variability due to aversive task nature and experimenter interference. The Cambridge Neuropsychological Test Automated Battery (CANTAB), a validated touchscreen-based tool, is commonly used in humans to assess multi-domain cognitive function, but its application in chronic stroke is limited. This study aimed to address the lack of comparable cross-species cognitive assessments in the chronic phase after stroke by employing a touchscreen-based visuospatial object-location PAL task in both chronic stroke survivors and a mouse model of photothrombotic stroke. The primary research question was whether mice and humans in the chronic phase post-stroke respond similarly when evaluated using this touchscreen-delivered PAL task. Successful demonstration of comparable cognitive deficits in both species would represent a significant advance in translational stroke research, providing a more robust framework for evaluating potential therapies.

Chronic stroke survivors frequently experience persistent memory and learning impairments across multiple cognitive domains. The challenge in advancing research lies in establishing and validating parallel cognitive assessment platforms for humans and animal models. This need was highlighted in recent recommendations from the Stroke Recovery and Rehabilitation Roundtable. While CANTAB is widely used in assessing cognitive function in neurodegenerative diseases, its application in chronic stroke patients has been limited, primarily focusing on sub-acute phases. Rodent models typically employ tasks like the Morris water maze and novel object recognition to assess spatial memory and learning, but these have limited comparability with human clinical assessments. Previous studies have demonstrated the feasibility of aligning touchscreen-based cognitive tests in mice and humans using comparable tasks, particularly in the context of schizophrenia and Dlg gene mutations, indicating the potential for cross-species translation using this approach.

This cross-sectional study recruited 140 participants (70 chronic stroke survivors and 70 age-matched controls) between November 2017 and February 2019. Stroke survivors were recruited through the Hunter Stroke Research Volunteer Register, while controls were recruited from the HMRI control registry and social media. Exclusion criteria included a history of pituitary and adrenal gland diseases. Participants completed a CANTAB Motor Screening task before the cognitive assessment. Cognitive performance was measured using the CANTAB visuospatial PAL task, comprising four stages (2, 4, 6, and 8 patterns). PAL performance was assessed based on various metrics (correct choices at first attempt (PAL FAMS), total errors (adjusted), total attempts, patterns reached). A retrospective power calculation yielded an effect size of 0.55 for PAL FAMS. A separate mouse model utilized 45 C57BL/6 male mice, randomly assigned to sham or stroke groups. Photothrombotic stroke was induced, and cognitive function was assessed at 20 weeks post-stroke using the rodent touchscreen visuospatial PAL task. The sample size of seven mice per group was determined using power calculation, considering the percentage of correct rate (primary outcome) in sham mice. Data analysis for humans involved non-parametric tests and linear regression adjusting for confounders (age, gender, diabetes, blood pressure, waist circumference). Mouse data was analyzed using two-way ANOVA and Sidak multiple comparisons. Histological analysis confirmed stroke in mice. Ethical approvals were obtained for both human and animal studies.

In the human cohort, 58.6% had ischemic stroke and 37.1% had hemorrhagic stroke. The median time post-stroke was 38.5 months. Stroke survivors showed significantly lower scores on PAL FAMS (correct choices on first attempt, p=0.002), requiring more attempts to pass easier levels (p<0.001 and p=0.016 for 2 and 4 patterns, respectively), and making significantly more errors overall (p=0.044). In mice, stroke resulted in a significantly lower correct rate in the object-location PAL task (p=0.032), longer time to complete sessions (p<0.0001), and fewer tasks completed (p<0.0001). Histological analysis confirmed significant tissue loss in the ipsilesional hemisphere and corpus callosum in stroke mice, along with neuronal loss and reactive astrogliosis. There was no significant difference in the number of correction trials between sham and stroke mice.

This study demonstrates, for the first time, comparable cognitive deficits in chronic stroke survivors and mice using an analogous touchscreen-based PAL task. The findings support the use of this parallel assessment platform to improve our understanding of post-stroke cognitive decline and facilitate the translation of novel therapies. Both humans and mice exhibited impairments in visuospatial PAL, characterized by reduced correct choices and increased errors, demonstrating the cross-species validity of the task. The consistency of findings across species strengthens the translational potential of the PAL task, suggesting that it could serve as a reliable and sensitive measure for evaluating the effectiveness of new therapies aimed at enhancing cognitive function after stroke. Further research can explore whether these findings extend to other cognitive domains and stroke subtypes.

This study provides valuable evidence for the utility of parallel cross-species cognitive assessment platforms. The comparable cognitive deficits observed in humans and mice in the chronic phase post-stroke using the touchscreen-based PAL task highlight its potential for preclinical testing and translation of novel therapeutic strategies to improve long-term cognitive outcomes. Future research should investigate the applicability of this approach across other cognitive domains and stroke subtypes and should include assessment of motor function post-stroke in mice.

The cross-sectional nature of the human study limited the assessment of pre-stroke cognitive function, a known confounder. The wide range of stroke subtypes and time post-stroke in the human cohort might have influenced individual cognitive performance. Lack of detailed clinical demographics (infarct characteristics, education level) could have affected the analysis. In mice, motor function was not explicitly assessed post-stroke, though differences in response latency were not detected. Future studies should address these limitations by incorporating longitudinal designs, detailed clinical data, and motor function assessments.