Predictive factors of outcome in cervical dystonia following deep brain stimulation: an individual patient data meta-analysis

Background Deep brain stimulation (DBS) therapy has been suggested to be a beneficial alternative in cervical dystonia (CD) for patients who failed nonsurgical treatments. This individual patient data meta-analysis compared the efficacy of DBS in the globus pallidus internus (GPi) versus subthalamic nucleus (STN) and identified possible predictive factors for CD. Methods Three electronic databases (PubMed, Embase and Web of Science) were searched for studies with no publication date restrictions. The primary outcomes were normalized by calculating the relative change in TWSTRS total scores and subscale scores at the last follow-up. Data were analyzed mainly using Pearson’s correlation coefficients and a stepwise multivariate regression analysis. Results Thirteen studies (86 patients, 58 with GPi-DBS and 28 with STN-DBS) were eligible. Patients showed significant improvement in the Toronto Western Spasmodic Torticollis Rating Scale (TWSTRS) (52.5 ± 11.6 vs 21.9 ± 14.9, P < 0.001) scores at the last follow-up (22.0 ± 14.3 months), compared with scores at baseline, with a mean improvement of 56.6% ( P < 0.001) (54.9% in severity, 63.2% in disability, 41.7% in pain). There was no significant difference in the improvement (%) of the total TWSTRS scores in 3 years for the GPI and STN groups (58.1 ± 22.6 vs 47.5 ± 39.2, P > 0.05). Age at surgery and age at symptom onset were negatively correlated with the relative changes in TWSTRS scores at the last follow-up, while there was a positive correlation with preoperative TWSTRS scores. On the stepwise multivariate regression, only the age at surgery remained significant in the best predictive model. Conclusions GPi-DBS and STN-DBS both provided a common great improvement in the symptoms of CD patients in 3 years. Earlier age at surgery may probably indicate larger improvement. More randomized large-scale clinical trials are warranted in the future.