This study recruited 18 patients diagnosed with subacute ischemic stroke from May 2019 to February 2022 at the Nagoya City Rehabilitation Center. Inclusion criteria involved patients: (1) between 40 and 80 years; (2) first ischemic stroke-associated regions of the CST occurred within 3 months and were confirmed by a neurologist using MRI; (3) independent in activities of daily living (ADLs) before onset; (4) right dominant hand; and (5) without cognitive impairment that makes decision-making difficult (Mini-Mental State Examination ≥ 24 points). Exclusion criteria were (1) major health problems or poor physical condition that might limit rehabilitation; (2) contraindication to MRI; (3) presence of other neurological or psychiatric diseases; (4) disturbance of consciousness or neurological symptoms suggesting a cortical lesion (hemispatial neglect, agnosia, aphasia, and apraxia); and (5) has received a surgical treatment for ischemic stroke. Table 1 presents the patient background information.
This study also recruited 30 right-handed healthy control participants (15 males and 15 females; mean age, 65.87±8.43; range, 44–79 years). The healthy control group had no cognitive dysfunction (Mini-Mental State Examination ≥ 24 points), lived independently, and had no neurological or psychiatric disorder history.
The Institutional Review Board of the Nagoya City Rehabilitation Center approved this study (approval number 2018012). Informed consent was obtained from all participants in writing prior to their enrollment in the study.
Patients received physical, occupational, and speech therapies 7 days a week for 6 weeks. Each therapy was delivered for 40–80 minutes daily. We conducted clinical evaluations, and a radiological technologist performed MRI examinations before (pre-treatment) and after (post-treatment) rehabilitation.
Patients received interventions for upper extremity hemiplegia based on guidelines (Winstein et al., 2016). Robotic-assisted therapy and modified Constraint-induced movement therapy were mainly provided for UE functional training following the protocol from the developer and promoted real-world arm use in the affected hands (Morris et al., 2006; Anmoto et al., 2023).
DTI acquisition
DTI was performed using a 3.0-T magnetic resonance scanner (Trio; Siemens AG, Erlangen, Germany) with a 16-channel head coil. FA images created by the MRI scanner were used for the analysis. FA, one of the most robust metrics for quantifying diffusion anisotropy, is computed and displayed on a voxel-by-voxel basis (Alexander et al., 2007).
Within the constraints of the in-plane resolution, some regions of WM (such as the CST) should normally have very high FA. In contrast, others should have considerably lower FA, even though they are fully volumed, which probably represents architectural differences in fiber tract organization at the intravoxel level (Le Bihan et al., 2001; Alexander et al., 2007). A high FA indicates a strong signal on the image, whereas a small FA indicates a weak signal.
The three-dimensional coefficients of anisotropic diffusion are expressed as λ1, λ2, and λ3 (Pfefferbaum et al., 2000; Hagmann et al., 2006). Each FA was calculated using the following formula from these diffusion coefficients, and the FA image was generated(Pfefferbaum et al., 2000; Hagmann et al., 2006).
A single-shot echoplanar imaging sequence was used to obtain 1 non-diffusion weighted image (b=0 s/mm2) and 12 images with noncollinear diffusion gradients (b=1000 s/mm2) for 64 axial slices per patient (field of view, 230.4 mm × 230.4 mm; acquisition matrix, 128 × 128; gapless slice thickness, 3 mm; echo time, 75 ms; and repetition time, 8700 ms)
We also obtained T1-, T2-, and diffusion-weighted MRI images for structural information and lesion localization (Supplement Ⅰ).
DTI processing and analysis
Brain images were analyzed using the FMRIB Software Library (FMRIB Center, Oxford, United Kingdom; www.fmrib.ox.ac.uk/fsl); various brain image analysis tools, including FMRIB's Diffusion Toolbox, FMRIB's Linear and Nonlinear Image Registration Tool, Atlases and FSL eyes, was used for image processing. They were converted linearly and non-linearly to the standardize FA images. Subsequently, these FA were mapped to the standard stereotaxic space (International Consortium of Brain Mapping DTI-81 Atlas). Visual comparison with images generated by the FSL eyes confirmed spatial transformations of the FA brain maps.
We hypothesized that the CST, corpus callosum, and fibers connecting the limbic system, which have been related to UE functional recovery and behavioral learning, are involved in functional and behavioral changes (Takebayashi et al., 2018; Takenobu et al., 2014; Fan et al., 2013; Sampaio-Baptista et al., 2013; Schaechter et al., 2009). Unpaired or paired neural pathways were specified as regions of interest (ROI). The ROIs were defined as unpaired neural pathways, including (1) the genu of the corpus callosum (GCC), (2) the body of the corpus callosum (BCC), (3) the splenium of the corpus callosum (SCC), and (4) the column and body of the fornix (CBF). The pairs of neural pathways included affected and unaffected (5) cerebral peduncles (CP), (6) posterior limb internal capsule (PLIC), (7) cingula of the cingulate gyrus (CgC), and (8) hippocampal cingulum (HIP CgC), which were also defined as the ROIs (Fig 1). We used templates (Mori et al., 2008) and mean FA values within the ROIs were estimated. In addition, when comparing the FA in patients at admission with the FA in healthy control participants, the ratio of FA on the left and right sides (rFA) within the ROI were calculated to eliminate individual differences in brain morphology. The rFA was defined as the inferior hemisphere (right hemisphere) or superior hemisphere (left hemisphere) for healthy participants because they were right-handed. In these patients, rFA was defined as the affected or unaffected hemispheres. When comparing the FA before and after the intervention within patients, we confirmed the changes in the paired fibers on the affected and unaffected sides. In addition, we had a similar observation when confirming the correlation between clinical evaluation and FA.
Behavioral assessment
All patients were evaluated for real-world arm use using MAL. The MAL is a structured interview that measures how well (11-point quality of movement scale) and how much (11-point amount of use scale) patients use their more affected arm to accomplish 14 activities of daily living. The score ranges from 0 (no use of the more-affected arm) to 5 (normal use of the more-affected arm). The total scale scores are the means of the item scores. Previous studies have indicated that MAL is a reliable, stable, and valid measure of real-world arm behavior (van der Lee., 2004; Uswatte et al., 2005; Uswatte & Taub, 2005). In this study, the quality of movement score was adopted because it is more internally consistent and reliable than the amount of use, and it captures components of the amount and quality of arm use outside the laboratory (Uswatte et al., 2005; Uswatte & Taub, 2005).
Functional motor impairment assessment
We assessed UE motor impairment using FMA-UE. The FMA-UE comprised 33 UE items that measure the movement and reflexes of the shoulder, elbow, forearm, wrist, and hand, and coordination. The reliability, validity, and responsiveness of the FMA have been established (Hsieh et al., 2009). Each tested movement was scored as 0 (movement cannot be performed), 1 (reduced strength, speed, amplitude, or precision), or 2 (movement performed completely). The FMA-UE determined the severity of paralysis (25 points or less is severe paralysis, 25–50 points are moderate paralysis, and 50 points or over is mild paralysis) (Luft et al., 2004).
Statistical analysis
A two-sample t-test was used to compare the differences between the rFA of healthy control participants and those of patients at admission. A paired Wilcoxon test was used to compare differences between pre-treatment and post-treatment scores for FMA-UE and MAL. Whereas, a paired t-test was used to compare differences between before and after the intervention in each FA. For all analyses, statistical significance was set at p<0.05.
Pearson’s correlation coefficients were calculated for the relationships between changes in FA in each ROI and changes in MAL. We also confirmed the relationship between FA changes in each ROI and FMA-UE changes. Single regression analyses were used to detect linear relationships. Statistical significance was defined as p ≤ 0.05, and statistical tendencies as 0.05 < p < 0.10.
Furthermore, the severity of paralysis significantly impacts motor function recovery and may also affect WM structural changes. Therefore, we also performed a sub-analysis of 11 patients with moderate-to-severe paralysis (FMA-UE < 50 points) (Luft et al., 2004).
All analyses were performed using IBM SPSS Statistics for Windows (version 25.0; IBM Inc., Tokyo, Japan).