Chronometric TMS-fMRI of personalized left dorsolateral prefrontal target reveals state-dependency of subgenual anterior cingulate cortex effects

Abstract

Transcranial magnetic stimulation (TMS) applied to a left dorsolateral prefrontal cortex (DLPFC) area with a specific connectivity profile to the subgenual anterior cingulate cortex (sgACC) has emerged as a highly effective non-invasive treatment option for depression. However, antidepressant outcomes demonstrate significant variability among therapy plans and individuals. One overlooked contributing factor is the individual brain state at the time of treatment. In this study we used interleaved TMS-fMRI to investigate the influence of brain state on acute TMS effects, both locally and remotely. TMS was performed during rest and during different phases of cognitive task processing. Twenty healthy participants were included in this study. In the first session, imaging data for TMS targeting were acquired, allowing for identification of individualized targets in the left DLPFC based on highest anti-correlation with the sgACC. The second session involved chronometric interleaved TMS-fMRI measurements, with 10 Hz triplets of TMS administered during rest and at distinct timings during an N-back task. Consistent with prior findings, interleaved TMS-fMRI revealed significant BOLD activation changes in the targeted network. The precise timing of TMS relative to the cognitive states during the task demonstrated distinct BOLD response in clinically relevant brain regions, including the sgACC. Employing a standardized timing approach for TMS using a task revealed more consistent modulation of the sgACC at the group level compared to stimulation during rest. In conclusion, our findings strongly suggest that acute local and remote effects of TMS are influenced by brain state during stimulation. This study establishes a basis for considering brain state as a significant factor in designing treatment protocols, possibly improving TMS treatment outcomes.

Fig. 1. Overview of the conducted experiment, showing both measurement sessions and the task.

The N-back task was performed at two difficulty levels (N = 0 and N = 2) in a blocked design, consisting of three blocks per difficulty. The 0-back condition was designed as an active control condition. During session 2, interleaved TMS bursts were delivered in 10 Hz triplets at 100% rMT to the individualized target site during three different states and considering two different timings for stimulation during the task.

Fig. 2. Individualized targeting and TMS during rest.

Group-averaged resting-state functional connectivity for the sgACC seed and individual functional connectivity guided target positions are shown in (A). B displays t-value group maps of TMS during rest at p < 0.05 FWE corrected at cluster level (p < 0.001 cluster defining threshold).

Fig. 3. Brain regions showing increased (hot) and decreased (blue-green) activation for TMS effective vs. ineffective timing during N-BACK (thresholded at p < 0.05, family-wise error corrected on cluster level).

Compared to TMS during 0-back as a control condition, effective vs. ineffective stimulation during 2-back leads to modulation of key regions linked to clinical TMS effects. Importantly, areas of the left DLPFC/inferior frontal gyrus (IFG) show higher activation and subgenual anterior cingulate cortex (sgACC), right inferior parietal lobule (IPL) and striatal areas lower activation if stimulation was performed in the effective timing.

Fig. 4. ROI analysis for effective and ineffective timing during N-back task.

For each ROI, subject-level mean PSCs for the effective (2-back effective > 0-back effective) and ineffective (2-back ineffective > 0-back ineffective) condition were extracted. (A) and (B) show ROI analysis of clusters showing significant differences in activation for differential timing of TMS in comparison to artificial events, i.e. same timepoints of the task but no TMS was performed. In (C) extracted mean PSCs for the predefined ROIs during TMS with different timings is displayed. P values correspond to paired t-tests between timings.