Domain-general subregions of the medial prefrontal cortex contribute to recovery of language after stroke.

We hypothesized that the recovery of speech production after left hemisphere stroke not only depends on the integrity of language-specialized brain systems, but also on 'domain-general' brain systems that have much broader functional roles. The presupplementary motor area/dorsal anterior cingulate forms part of the cingular-opercular network, which has a broad role in cognition and learning. Consequently, we have previously suggested that variability in the recovery of speech production after aphasic stroke may relate in part to differences in patients' abilities to engage this domain-general brain region. To test our hypothesis, 27 patients (aged 59 ± 11 years) with a left hemisphere stroke performed behavioural assessments and event-related functional magnetic resonance imaging tasks at two time points; first in the early phase (∼2 weeks) and then ∼4 months after the ictus. The functional magnetic resonance imaging tasks were designed to differentiate between activation related to language production (sentential overt speech production-Speech task) and activation related to cognitive processing (non-verbal decision making). Simple rest and counting conditions were also included in the design. Task-evoked regional brain activations during the early and late phases were compared with a longitudinal measure of recovery of language production. In accordance with a role in cognitive processing, substantial activity was observed within the presupplementary motor area/dorsal anterior cingulate during the decision-making task. Critically, the level of activation within this region during speech production correlated positively with the longitudinal recovery of speech production across the two time points (as measured by the in-scanner performance in the Speech task). This relationship was observed for activation in both the early phase (r = 0.363, P = 0.03 one-tailed) and the late phase (r = 0.538, P = 0.004). Furthermore, presupplementary motor area/dorsal anterior cingulate activity was a predictor of both language recovery over time and language outcome at ∼4 months, over and above that predicted by lesion volume, age and the initial language impairment (general linear model overall significant at P < 0.0001; ExpB 1.01, P = 0.02). The particularly prominent relationship of the presupplementary motor area/dorsal anterior cingulate region with recovery of language was confirmed in voxel-wise correlation analysis, conducted unconstrained for the whole brain volume. These results accord with the hypothesis that the functionality of the presupplementary motor area/dorsal anterior cingulate contributes to language recovery after stroke. Given that this brain region is often spared in aphasic stroke, we propose that it is a sensible target for future research into rehabilitative treatments. More broadly, baseline assessment of domain-general systems could help provide a better prediction of language recovery.