Invasive Neuromodulation for Treatment-Resistant Depression: a Systematic Review and Meta-Analysis

Abstract

Introduction: Catastrophes exacerbate mental health crises, including Major Depressive Disorder, of which 30% become Treatment-Resistant Depression (TRD). When traditional pharmacotherapy fails, invasive neuromodulation techniques like Deep Brain Stimulation (DBS) and Vagus Nerve Stimulation (VNS) have emerged as promising alternatives, but efficacy and safety remain uncertain.

Objective: This systematic review and meta-analysis evaluates existing evidence on DBS and VNS usage in TRD patients.

Methods: A systematic search in PubMed, Embase, and Cochrane Library was performed on 17^th October 2025. Randomised controlled trials and observational studies were included. The primary outcome was the change in Montgomery–Åsberg Depression Rating Scale (MADRS). Meta-regressions examined age, sex, duration of depression, intervention length and DBS target. Study-level data were pooled using random-effects meta-analytic models.

Results: Of 338 screened studies, 29 met inclusion criteria (10 VNS, 19 DBS). The pooled mean MADRS reduction was 14.0 points (95% CI 12.2–15.8), indicating a robust antidepressant effect, with all studies exceeding the minimum clinically important difference of ≥ 1.9 points. Subgroup analyses showed mean improvements of 14.8 (95% CI 12.6–17.0) for DBS and 12.6 (95% CI 10.2–15.0) for VNS. Meta-regression found no significant moderating effects of age, sex, or depression duration (p > 0.05). Among DBS studies, stimulation site (vALIC/NAcc, SCC, sIMFB) did not affect efficacy (p = 0.547), whereas longer stimulation duration predicted greater improvement (−0.11 MADRS points/month, p = 0.004). Considerable heterogeneity was present overall (I² = 87.1%) and within subgroups (DBS: I² = 79.4%; VNS: I² = 86.0%).

Conclusion: Invasive neuromodulation confers clinically meaningful antidepressant effects in TRD, with both DBS and VNS comparably reducing MADRS scores irrespective of demographic or illness factors. Nevertheless, benefits must be balanced against procedural risks. Developing reliable predictors of response will optimise patient selection and inform future trials.