Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Brain Age Predicts Long-Term Recovery in Post-Stroke Aphasia

Poster E36 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall
Also presenting in Poster Slam E, Saturday, October 8, 3:00 - 3:15 pm EDT, Regency Ballroom

Sigfus Kristinsson¹, Natalie Busby¹, Chris Rorden¹, Roger Newman-Norlund¹, Dirk B. den Ouden¹, Sigridur Magnusdottir², Helga Thors², Argye E. Hillis³, Leonardo Bonilha⁴, Julius Fridriksson¹; ¹University of South Carolina, ²University of Iceland, ³Johns Hopkins University, ⁴Medical University of South Carolina

Introduction Neuroplastic properties of the brain decrease with age.1 Nonetheless, the association between age and language recovery in stroke remains unclear.2 Here, we examined the association between brain age, a neuroimaging-derived measure of brain atrophy, at stroke onset and: (1) cross-sectional language function, and (2) long-term recovery of language function, beyond chronological age. Method A total of 49 consecutive cases (age: 65.2+/-12.2 years, 25 female) of acute left-hemisphere strokes underwent routine clinical neuroimaging and a language assessment (BEST-2)3 upon hospital admission. A subsample of 30 participants returned for follow-up language assessments >2-years after stroke. Each individual’s lesion brain scan was ‘healed’ to enable automated brain age estimation. First, FLAIR/lesion maps were co-registered to participants’ own T1 scan. Next, T1s and lesion maps were used to create an enantiomorphically healed version of their T1.4 This process exploits the symmetrical nature of the brain and the fact that lesions in our sample were unilateral. Briefly, damaged tissue in the ipsilesional hemisphere was replaced with healthy tissue from homologous areas of the healthy hemisphere. We used the BrainAgeR analysis pipeline (github.com/james-cole/brainageR)5 to estimate brain age using default settings. The pretrained model was created based on images from healthy individuals (N>4,000) between 18-90 years old,5 thus serving as inherent control data here. Multiple regression models were constructed to test the effects of brain age on language outcomes. Lesion volume and chronological age were included as covariates in all models. Results Estimated brain age was on average decelerated by 3.7+/-7.5 years (range: -24.1 to 10.1 years) relative to chronological age. Accelerated brain age was associated with poorer overall language performance (F(1,48)=5.65, p=.022), naming (F(1,48)=5.13, p=.028), and speech repetition (F(1,48)=8.49, p=.006) at stroke onset. All participants who returned for a follow-up assessment >2 years after onset showed a significant improvement across all language tasks (all p<.001). At follow-up, brain age was found to be inversely associated with change in language function (F(1,26)=8.66, p=.007) and speech repetition (F(1,26)=7.10, p=.013), but its correlation with change in naming (F(1,26)=3.4, p=.078) and auditory comprehension (F(1,26)=3.3, p=.081) marginally failed to reach statistical significance. Across timepoints and assessments, chronological age was only associated with naming performance at stroke onset (F(1,48)=4.18, p=.047). Conclusion Our findings reveal for the first time that a neuroimaging-derived measure of biological brain age, as a measure of structural integrity at stroke onset, is associated with longitudinal recovery of language function. Importantly, brain age explained more variability in language function than chronological age. These findings hold substantial promise to improve clinical management of stroke as brain age was estimated based on routine clinical brain scans as opposed to the more time-consuming research-grade scans. Future research will need to investigate the extent to which brain age supplements other measures of brain health in the context of stroke recovery.

Topic Areas: Disorders: Acquired, Methods