Convergent functional effects of antidepressants in major depressive disorder: a neuroimaging meta-analysis

Abstract

Background: Neuroimaging studies have provided valuable insights into the macroscale impacts of antidepressants on brain functions in patients with major depressive disorder. However, the findings of individual studies are inconsistent. Here, we aimed to provide a quantitative synthesis of the literature to identify convergence of the reported findings at both regional and network levels and to examine their associations with neurotransmitter systems.

Methods: Through a comprehensive search in PubMed and Scopus databases, we reviewed 5258 abstracts and identified 36 eligible functional neuroimaging studies on antidepressant effects in major depressive disorder. Activation likelihood estimation was used to investigate regional convergence of the reported foci of antidepressant effects, followed by functional decoding and connectivity mapping of the convergent clusters. Additionally, utilizing group-averaged data from the Human Connectome Project, we assessed convergent resting-state functional connectivity patterns of the reported foci. Next, we compared the convergent circuit with the circuits targeted by transcranial magnetic stimulation therapy. Last, we studied the association of regional and network-level convergence maps with selected neurotransmitter receptors/transporters maps.

Results: No regional convergence was found across foci of treatment-associated alterations in functional imaging. Subgroup analysis in the Treated > Untreated contrast revealed a convergent cluster in the left dorsolateral prefrontal cortex, which was associated with working memory and attention behavioral domains. Moreover, we found network-level convergence of the treatment-associated alterations in a circuit more prominent in the frontoparietal areas. This circuit was co-aligned with circuits targeted by "anti-subgenual" and "Beam F3" transcranial magnetic stimulation therapy. We observed no significant correlations between our meta-analytic findings with the maps of neurotransmitter receptors/transporters.

Conclusion: Our findings highlight the importance of the frontoparietal network and the left dorsolateral prefrontal cortex in the therapeutic effects of antidepressants, which may relate to their role in improving executive functions and emotional processing.

Fig. 1. Study selection flowchart.

MDD major depressive disorder, LLD late-life depression, ROI region of interest, SVC small volume correction.

Fig. 2. Treatment-induced increase of voxel-based physiology in the left dorsolateral prefrontal cortex.

A Peak coordinates of the included experiments in Treated > Untreated (red) and Untreated > Treated (blue) comparisons. Each dot represents a peak coordinate. B Activation likelihood estimation showed significant convergence of Treated > Untreated comparisons in the left dorsolateral prefrontal cortex (DLPFC) after family-wise error correction at cluster level (p_cFWE < 0.05).

Fig. 3. Connectivity mapping of the left dorsolateral prefrontal cortex cluster.

The left DLPFC convergent cluster identified in the activation likelihood estimation meta-analysis on Treated > Untreated experiments was used as the seed (outlined patch) to map its meta-analytical co-activation (A) and resting-state functional connectivity (B).

Fig. 4. Convergent connectivity mapping of antidepressant effects.

left: The cortical and subcortical map represent convergent connectivity map of the foci from all experiments. right: Mean resting-state functional connectivity (RSFC) of the observed foci across canonical resting-state networks (denoted by “-“) compared against null mean values calculated based on 1000 permutations of randomly selected foci (half-violin plots). In frontoparietal network (FPN) the observed mean RSFC was significantly more extreme than the null distribution in a two-tailed test. VIS visual network, SMN somatomotor network, DAN dorsal attention network, SAN salience network, LIM limbic network, FPN frontoparietal network, DMN default mode network.

Fig. 5. Association of antidepressants meta-analytic effects and transcranial magnetic stimulation targets.

A The locations of three different TMS targets is shown in comparison to the left DLPFC convergent cluster identified in the activation likelihood estimation meta-analysis on Treated > Untreated experiments. B The convergent connectivity map of antidepressant effects, the RSFC map of the left DLPFC cluster, and the RSFC maps of the different transcranial magnetic stimulation (TMS) sites as well as their cross-correlations are shown. In the correlation matrix within each cell the Pearson correlation coefficient is reported. Asterisks denote p_{variogram, FDR} < 0.05.

Fig. 6. Association of meta-analytic findings with neurotransmitter receptor/transporter maps.

A The parcellated and Z-scored PET maps of neurotransmitter receptor/transporter (NRT) are shown. Red outline indicates the left dorsolateral prefrontal cortex (L DLPFC) convergent cluster identified in the activation likelihood estimation meta-analysis on Treated > Untreated experiments. B Median normalized density of NRTs in L DLPFC cluster. None of the NRTs showed significantly more extreme normalized density in this cluster compared to a null distribution created using variogram-based surrogate maps and after false discovery rate adjustment at 5%. C Pearson correlation of parcellated convergent connectivity map with the NRT maps. None of the correlations were significant using variogram-based surrogates and after false discovery rate adjustment at 5%.