Participants
This study included 47 participants with chronic post-stroke aphasia, primarily resulting from left hemisphere strokes, with some participants exhibiting bilateral lesions. Lesion information for each participant is shown in Supplementary Table S1. MRI scans were not available from some patients; thus, lesion details were only available for 21/22 tDCS patients and 19/25 sham patients. The participants were divided into two groups: the cerebellar tDCS condition (n = 22) and sham condition (n = 25). We initially recruited 25 patients for cerebellar tDCS, but three dropped out due to either COVID-19 infection, schedule conflict, or lack of English proficiency. The cerebellar tDCS dataset were acquired between 2020-2022 with a non-randomized, double-blind design. The sham tDCS data were obtained from a previous study conducted between 2018-2020, which employed a randomized, double-blind design. Table 1 summarizes the participants’ demographic and clinical information. The inclusion criteria for participants were as follows: (1) a clinical diagnosis of aphasia resulting from ischemic or hemorrhagic stroke, (2) at least 12 months post-stroke, (3) aged 18 years or older, and (4) English proficiency. Participants with multiple strokes were also included. Exclusion criteria consisted of aphasia resulting from any neurological conditions other than stroke. All participants provided written informed consent prior to participating in the study. The study was approved by the Institutional Review Board of Casa Colina Hospital. The trial was registered on ClinicalTrials.gov as NCT03699930.
Study design
Following a double-blind design, both participants and experimenters who conducted the interventions and assessments were blinded to the allocation of participants. To incorporate the pre-existing control data while preserving the study’s blinded nature, participants in the active tDCS group were told they could receive either real cerebellar tDCS or sham tDCS, paired with speech and language therapy. However, all participants received real tDCS. The P.I. was the only person aware of the allocation. Allocation concealment was accomplished by assigning a 6-digit code per participant.
For both the sham and cerebellar tDCS groups, the primary outcome measure was collected at two time-points, within one week before the intervention (pre-intervention) and one week after the intervention (post-intervention). Secondary outcome measures were collected at three time points: pre-intervention, post-intervention, and during a 3-month follow-up.
Outcome measures
The primary outcome measure was change in language scores as assessed by the Western Aphasia Battery-Revised (WAB-R). This evaluation encompassed four distinct subscales, namely spontaneous speech (SS), auditory verbal comprehension (AVC), name and word finding (NWF), and repetition (REP). The scores from these four subscales were consolidated to compute the aphasia quotient (AQ [range 0-100]), providing a measure of aphasia severity, with lower scores reflecting worse language performance. Secondary outcome measures (SOMs) were patient and/or caregiver self-reported and comprised the Communication Outcomes after Stroke (COAST), Carer COAST, Stroke and Aphasia Quality of Life Scale-39 (SAQOL-39), and Patient-Reported Outcomes Measurement Information System (PROMIS-Global). Higher scores indicate better outcomes. The COAST and Carer-COAST surveys consist of 20 items evaluating post-stroke communication effectiveness. The PROMIS-Global (10 items) and SAQOL-39 (39 items) evaluate patients' quality of life and overall well-being. The SAQOL-39 assesses functioning in four domains: physical, psychosocial, communication, and energy.
Transcranial direct current stimulation
TDCS stimulation was paired with speech and language therapy and was delivered for 20 minutes using a constant current stimulator (Soterix Medical 1 x 1 clinical trials device). Consistent with other studies on cerebellar tDCS (Sebastian 2021; DeMarco et al., 2022), the current study utilized 2 mA tDCS generated between two 5cm x 5cm saline-soaked sponges. The anode (active electrode) was placed on the right cerebellar cortex, 1cm under, and 4cm lateral to the inion (approximately over the cerebellar lobule VII) (Pope and miall, 2012). The reference electrode (cathode) was placed on the right deltoid. The current was ramped up to 2 mA over 30 seconds, maintained for 20 minutes, then ramped down over 30 seconds.
In the sham condition (data previously collected), after the initial ramp-up period of 30 seconds, the intensity was promptly reduced to 0 mA in the subsequent 30-second ramp-down phase and maintained at this level throughout the session. Each stimulation condition, whether real or sham, was coupled with a 20-minute speech and language therapy session. Participants underwent this procedure for five consecutive days.
Speech and language therapy
Participants underwent approximately 20 minutes of personalized speech and language therapy, tailored by the therapist’s clinical judgment, and guided by standardized assessments. The primary goal of the therapy was to enhance expressive language skills, employing a range of evidence-based treatment techniques, such as Melodic Intonation Therapy, Language Stimulation Approach, Modified Response Elaboration Training, Semantic Feature Analysis Treatment, Verb Network Strengthening Treatment, Sound Production Treatment, Integral Stimulation, and Word Retrieval Cuing Strategies (e.g., phonological and semantic cuing). For participants with severe auditory comprehension deficits that impeded their ability to understand tasks, the therapy incorporated targeted auditory comprehension interventions. The therapy sessions prioritized areas that would yield the most significant improvements in the participant's functional communication skills.
Statistical Analysis
To identify potential confounding variables, we compared the demographic and clinical characteristics of the two groups, including age, gender, time since injury (TSI), stroke type (ischemic or hemorrhagic), fluency (fluent or non-fluent), baseline AQ, handedness, and lesion hemisphere. Independent samples t-tests were used for continuous variables, while chi-square tests were employed for categorical variables. TSI was found to be significantly different between the two groups and was therefore included as a covariate in subsequent analyses.
Mixed-design analyses of variance (ANOVAs) were conducted to examine the treatment effects, with Group (tDCS or sham) as the between-subjects factor and Time (pre-intervention or post-intervention) as the within-subjects factor. The primary outcome measure consisted of scores from WAB-R, including AQ, SS, AVC, NWF, and REP. Secondary outcomes included scores from COAST, Carer COAST, PROMIS-Global, and SAQOL-39.
After finding a significant Time X Group interaction for SAQOL-39 scores, we conducted a follow-up analysis (Mixed ANOVAs with TSI as a covariate) to determine which subdomains (Physical, Psychosocial, Communication, Energy) of SAQOL-39 were driving the significance. Due to the limited number of questions (four) in the Energy subdomain, which may have resulted in a restricted range of scores and reduced variability, we combined the Energy and Physical subdomain scores. Three participants were missing SAQOL-39 data, resulting in n = 20 for tDCS and n = 24 for sham.
Due to challenges faced during the COVID-19 pandemic, many participants were unable to complete the 3-month follow-up assessment, resulting in insufficient data for group analysis at this time point. Consequently, only pre-intervention and post-intervention scores were included in the statistical analyses.