Comparative efficacy of non-invasive brain stimulation for post-stroke aphasia: A network meta-analysis and meta-regression of moderators.

In recent years, non-invasive brain stimulation (NIBS) interventions for post-stroke aphasia have received increasing attention, but their effects across different language domains and the influence of targeted locations and moderators remain unclear. Randomized controlled trials (RCTs) on NIBS in patients with post-stroke aphasia were searched. Standardized mean differences (SMDs) for pre-post language changes were pooled in Bayesian network meta-analyses. Moderators were examined using meta-regression. Of the 2105 records identified, 69 RCTs involving 1670 patients were included. Low-frequency (LF)-transcranial magnetic stimulation (rTMS) (SMD 0.84 [0.65,1.03]) was superior to anodal-transcranial direct-current stimulation (a-tDCS) (SMD 0.38 [0.05,0.71]) for global severity. Dual-tDCS was the leading option for naming and repetition. For spontaneous speech, both a-tDCS and dual-tDCS resulted in greater effects than LF-rTMS. As stimulation targets, the right inferior frontal gyrus ranked higher in global severity and spontaneous speech, while the temporoparietal region ranked higher in comprehension. Meta-regression demonstrated that therapeutic effects in the naming domain were moderated by the mean period of each therapy condition and the first language, while significant associations with age, therapy period, and number of sessions were observed for spontaneous speech. Overall, LF-rTMS is the most prioritized NIBS mode to alleviate global severity. Dual and anodal tDCS outperform rTMS for naming and repetition. The optimal stimulation region varies across different domains.