Neural activity changes following cognitive behavioural therapy in major depressive disorder: a preregistered functional magnetic resonance imaging meta-analysis

Major depressive disorder (MDD) is among the most prevalent and disabling psychiatric conditions worldwide, affecting approximately 4.7% of the global population, with disrupted fronto-limbic functioning central to its pathophysiology. Cognitive behavioural therapy (CBT) is an effective first-line treatment, putatively working by modifying maladaptive cognitive patterns and reducing negative affect. Seminal models propose that these changes are neurally instantiated through strengthened prefrontal top-down regulation of limbic reactivity - a shift from ventral, affect-driven processing towards greater dorsal cognitive control. However, empirical neuroimaging studies have yielded heterogeneous findings, attributable to small sample sizes, methodological variability, and the frequent inclusion of mixed diagnostic groups, such that a systematic, disorder-specific synthesis of neural CBT effects in MDD is still lacking.

This meta-analysis aims to identify consistent patterns of neural activity change associated with CBT in MDD and to test whether these changes align with predicted increases in prefrontal cognitive control and normalisation of limbic hyperactivation. We conducted a preregistered coordinate-based meta-analysis using anisotropic effect-size seed-based d mapping (AES-SDM), integrating reported peak coordinates and unthresholded whole-brain statistical maps to maximise available evidence and improve spatial sensitivity over activation likelihood estimation approaches. We systematically searched PubMed, Embase, PsycINFO, and Web of Science for longitudinal task-based fMRI studies examining BOLD signal changes from pre- to post-CBT in adolescent and adult patients with MDD (DSM-IV/DSM-5), covering cognitive-emotional and reward-processing paradigms. Effect-size maps were pooled in a random-effects model with statistical significance assessed via non-parametric permutation testing and family-wise error correction. Preliminary results, based on coordinate data from eight independent samples (128 patients), are reported at p < .005 (uncorrected), minimum cluster extent 10 voxels, peak SDM-Z > 1.0.

Preliminary analyses identified increased activation following CBT in bilateral insula (right: MNI 44,18,0; SDM-Z = 2.08, p < .001, 609 voxels; left: MNI −32,18,2; SDM-Z = 1.52, p = .002, 323 voxels), medial cingulate and paracingulate gyri (MNI 0,4,44; SDM-Z = 1.85, p < .001, 600 voxels), bilateral precentral gyri, and right medial orbitofrontal cortex (MNI 8,40,−12; SDM-Z = 1.45, p = .003, 12 voxels). A large cluster of decreased activation was observed in the corpus callosum (MNI 26,−38,52; SDM-Z = −1.31, p < .001, 980 voxels).

Post-CBT increases in the insula and medial cingulate cortex suggest enhanced integration of interoceptive and affective signals with action-relevant processing, consistent with more adaptive salience evaluation. Increased medial orbitofrontal activity may reflect improvements in affective valuation and self-referential processing. Together, these findings provide partial support for the hypothesis that CBT strengthens top-down cognitive control, while prominent activations in ventral and integrative regions indicate a more distributed neural reorganisation than a simple limbic-to-prefrontal shift. As current results are based on coordinate data alone, integration of unthresholded whole-brain maps, currently underway, is expected to increase statistical power and spatial resolution. Ultimately, identifying consistent neural correlates of CBT response may support treatment personalisation and help predict which patients are most likely to benefit.

References

No references