(K-438) Longitudinal Changes in MRI Hippocampal Volume are Associated with Post-Stroke Cognitive Decline

Introduction:: This abstract is being submitted to the Neural Engineering Track and Neuroimaging Subtrack.

Patients who experience stroke and cerebrovascular disease are at risk for developing cognitive decline of an unknown mechanism. Neurodegenerative diseases are associated with brain atrophy, including hippocampal atrophy. The hippocampus is involved in encoding new episodic memories and changes in hippocampal volume are associated with cognitive decline. Thus, we hypothesize that there will be post-stroke changes in hippocampal volume and that these changes are associated with cognitive performance.

Materials and Methods:: Participants were enrolled in a prospective longitudinal brain MRI study if they have suffered a stroke, transient ischemic attack (TIA), or central retinal artery occlusion (CRAO) at least 3 months prior to the study, have evidence of at least early confluence of white matter lesions on MRI (Fazekas grade >=2), have preserved ambulatory and language functions, and do not have severe dementia. Follow up scans were performed every 3 months for the first year, every 6 months for the second year, and yearly thereafter for up to 5 years. The Montreal Cognitive Assessment (MoCA) was performed at each scan, and the memory sub score was obtained from the memory delayed recall component. T1 weighted images were segmented to calculate hippocampal volumes; poor-quality segmentations were excluded from further analysis (n=16). Three sub-analyses were performed: a baseline analysis testing for an association between hippocampal volume and cognitive scores, a longitudinal analysis examining hippocampal volume changes in relation to stroke hemisphere, and a trajectory analysis comparing changes in residual hippocampal fraction and change in cognitive scores. Hippocampal fraction was calculated as final hippocampal volume divided by initial. Multiple linear regression was performed to evaluate an association between baseline hippocampal volume and the baseline cognitive scores. Mixed effects linear regressions were performed to evaluate differences in hippocampal volume ipsilateral or contralateral to the stroke hemisphere, and test for longitudinal changes. Mixed effects linear regression was also used to compare residual hippocampal fraction and changes in cognitive scores. Statistical analyses were performed with STATA v13.1.

Results, Conclusions, and Discussions::

81 participants were enrolled in the study, 68 received a baseline examination, with 1 scan of insufficient quality, leaving 67 for analysis. The mean age was 70±10 years old, and 44% were female. Index stroke was in the left hemisphere for 25 participants, the right hemisphere for 31 participants, bilateral for 8 participants. An additional 3 participants had no visible acute index ischemic lesion, 2 of which were diagnosed with a TIA and 1 CRAO. Median time from index stroke to baseline imaging was 3±2 months. Of those included in the baseline analysis, 43 participants had at least one year of follow-up MRI and were included in the longitudinal and trajectory analyses, of whom the mean age was 69±9 years old and 43% were female. In the baseline analysis, no significant relationship was identified between hippocampal volume and cognitive function. Age, race, hospital site, and acute stroke severity were all associated with lower MoCA scores at baseline. In the longitudinal analysis, hippocampal volumes significantly decreased over time (B=-1.543 mm³/month; 95% CI: [-2.772 -0.313]; p< 0.05). The hippocampi ipsilateral (n=164) to the stroke hemisphere decreased at a slower rate relative to the hippocampi contralateral (n=180) to the stroke hemisphere (B= -6.5336 mm³/month; 95% CI: [-11.623 -1.045]; p < 0.05). No other covariates were significantly associated with decline in hippocampal volume. In the trajectory analysis, residual hippocampal fraction was a significant predictor of both increasing MoCA scores (B=0.504 MoCA point/month; 95% CI: [0.277 0.732]; p< 0.0001) and memory subscores (B=0.619 MoCA point/month; 95% CI: [0.342 0.896]; p< 0.0001).

We found that in patients with cerebrovascular disease who have suffered a stroke, hippocampal volumes decrease with greater loss of volume in the hemisphere contralateral to the prior stroke. We also identified an association between change in hippocampal volume and change in MoCA score and memory subscore. These findings suggest a decrease in hippocampal volume occurs after stroke and may be contributing to post-stroke cognitive decline.

Future studies will explore predictors of hippocampal volume post-stroke as well as

if longitudinal changes in hippocampal volume are related to cognition in both stroke and non-stroke patients.

 

Acknowledgements (Optional): :

This work was supported by the National Institute of Neurological Disorders and Stroke (NINDS) Intramural Research Program

References (Optional): :