Effects of different types of deep brain stimulation on gait disorders in patients with Parkinson’s disease: a network meta-analysis of randomized controlled trials

Objective To systematically compare and rank the relative efficacy of different types of deep brain stimulation (DBS) for gait disorders in patients with Parkinson’s disease (PD), with particular emphasis on three outcomes: motor function (MDS-UPDRS III), Freezing of Gait Questionnaire (FOG-Q), and normal gait velocity (cm/s). The goal is to provide evidence-based guidance for individualized neuromodulation strategies. Methods Following a preregistered protocol (PROSPERO CRD420251074368), a comprehensive literature search was conducted from January 1, 2000, to June 15, 2025, across PubMed, Embase, Cochrane Library, and CNKI. Randomized controlled trials (RCTs) meeting predefined PICOS criteria were included, and continuous outcome data for MDS-UPDRS III, FOG-Q and normal gait velocity in the Off-Dopa state were extracted. A frequentist network meta-analysis framework (R, netmeta package) was employed, with mean difference (MD) and 95% confidence intervals (CI) as effect estimates. Heterogeneity was quantified using τ 2 and I 2 statistics. Global and local inconsistency tests were applied, and treatment ranking was performed using P-scores. Robustness was examined through fixed- vs. random-effects comparisons, leave-one-out sensitivity analyses, and, where feasible, meta-regression and subgroup analyses. Results Twenty-five RCTs were included, involving 324 patients in intervention groups and 324 in control groups. (1) MDS-UPDRS III: Network meta-analysis showed that Low-PPNa-DBS produced the greatest improvement in motor function (MD = −31.20, 95% CI –53.25 to −9.15, p = 0.0055), followed by PPN-DBS (MD = −26.00, 95% CI –38.92 to −13.08, p ≤ 0.0001) and CuN-DBS (MD = −23.00, 95% CI –42.65 to −3.35, p = 0.0218). Additional significant but moderate effects were observed for Posterior-STN-dDBS (MD = −16.55, 95% CI –28.43 to −4.68, p = 0.0063), IL-IL-DBS, 60 Hz-STN-DBS, 80 Hz-STN-DBS, STN-DBS, STN + SNr HF-DBS, and STN + SNr LF-DBS. Overall heterogeneity was moderate to high ( τ 2 = 12.1151, I 2 = 68.7%), and heterogeneity testing indicated significant heterogeneity ( Q total = 14.48, df = 5, p = 0.0128). (2) FOG-Q: None of the DBS modalities yielded statistically significant improvements in FOG-Q scores (all 95% CIs crossed zero, p > 0.05). This outcome exhibited high heterogeneity ( τ 2 = 1.3074, I 2 = 88.3%) and significant inconsistency ( Q total = 25.61, p < 0.0001), suggesting that current evidence is insufficient to confirm a definitive DBS benefit for freezing of gait. (3) Gait velocity (cm/s): Network analysis demonstrated that spDBS [MD = 18.06, 95% CI (10.42, 25.70), p < 0.0001] and STN-DBS [MD = 17.58, 95% CI (13.20, 21.95), p < 0.0001] provided the most pronounced improvements. Conventional aDBS [MD = 14.80, 95% CI (7.71, 21.89)] and GPi-DBS [MD = 13.91, 95% CI (2.40, 25.42)] were associated with moderate benefits. Heterogeneity for this outcome was moderate to high ( τ 2 = 12.1151, I 2 = 65.5%), indicating persistent between-study variability. Conclusion This network meta-analysis systematically compared the effects of different DBS modalities on gait disorders in PD. For MDS-UPDRS Part III, Low-PPNa-DBS, PPN-DBS, and CuN-DBS yielded the most pronounced and statistically robust improvements. In contrast, analysis of FOG-Q scores revealed no statistically significant effects for any DBS intervention. Regarding gait speed, spDBS and STN-DBS showed clear and significant enhancements. Systematic review registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420251074368 , Identifier: CRD420251074368.