Abstract
BACKGROUND: Targeted cognitive training (TCT) for schizophrenia is hypothesized to improve global cognitive functioning by driving plasticity in distributed frontal and temporal systems. While previous work has demonstrated that TCT evokes functional changes in prefrontal and temporal cortex, less is known about whether treatment response coincides with structural changes in these brain areas. In the current study, we examined whether changes in cognition in response to TCT (compared to a computer games control condition) correspond to changes in cortical thickness (CTh) in frontal and temporal cortices among patients with recent onset schizophrenia (SZ). Additionally, we determined whether baseline CTh predicted treatment response. METHODS: SZ patients were randomized to 40 hours of TCT of the auditory system (N=21) or computer games (CG; N=22). Before and after training, participants underwent MRI and cognitive testing. Freesurfer’s longitudinal pipeline was used to calculate CTh at baseline and follow up. Patients averaged 4.5 months to complete training. Using regions from the Desikan-Killiany Atlas, we created regions of interest (ROIs) from left and right frontal and temporal cortices and calculated CTh change scores. Next, we examined slope differences to determine whether the relationship between change in CTh and change in cognition differed between TCT and CG. Last, we determined whether baseline CTh in any sub-regions of the ROIs was predictive of treatment response in the TCT group. RESULTS: The relationship between change in CTh and global cognition differed between groups in the left frontal cortex (t=2.42; p=.02), characterized by a correlation in TCT (r=.50 p=.02), but no relationship in CG (r=-.18; p=.42). The same pattern was observed in the right frontal cortex (t=3.29 p=.002; TCT: r=.59; p=.004; CG: r=-.29 p=.18). We observed a similar pattern in the temporal cortex, with a significant slope difference in the left hemisphere (t=2.38 p=.02), characterized by a correlation in TCT (r=.58 p=.006) but not CG (r=-.13 p=.55). The same trend was observed in the right temporal cortex (t=1.93 p=.06), with a strong TCT correlation (r=.72 p=.0002), and no relationship in CG (r=.11 p=.61). We observed no significant group x time interactions on change in CTh. We conducted post-hoc analyses on sub-region ROIs that may be driving these relationships and identified numerous frontal and temporal regions showing a significant slope difference, including superior and middle frontal gyrus, as well as temporal pole, inferior, and middle temporal regions (all p’s<.05 – uncorrected). Finally, we investigated whether CTh in sub-region ROIs at baseline predicted response to treatment. The right caudal middle frontal gyrus (r=-.45, p=.04 - uncorrected), the left entorhinal cortex (r=-.57, p=.007 - uncorrected), and right inferior temporal gyrus (r=-.62, p=.003 - uncorrected), all showed lower baseline CTh relating to gains in global cognition. DISCUSSION: The current results demonstrate that response to TCT in recent onset SZ coincides with changes in CTh in both the frontal and temporal lobes. These changes appear to be driven by neuroplasticity in regions critical for auditory processing and cognitive control. Taken with related research, these findings support the hypothesis that TCT of the auditory system influences cognition by way of coordinated plasticity in distributed temporal and prefrontal regions. Additionally, we demonstrate that lower CTh in sub-lobar frontal and temporal ROIs corresponds to greater subsequent gains in cognition, suggesting that SZ patients with greater pathophysiological impairments may be especially responsive to targeted training interventions.