Clinical phenotypes associated with outcomes following deep brain stimulation for childhood dystonia

OBJECTIVE Although deep brain stimulation (DBS) is an accepted treatment for childhood dystonia, there is significant heterogeneity in treatment response and few data are available to identify ideal surgical candidates. METHODS Data were derived from a systematic review and individual patient data meta-analysis of DBS for dystonia in children that was previously published. Outcomes were assessed using the Burke-Fahn-Marsden Dystonia Rating Scale for movement (BFMDRS-M) and for disability (BFMDRS-D). The authors used partial least squares, bootstrapping, and permutation statistics to extract patterns of contributions of specific preoperative characteristics to relationship with distinct outcomes, in all patients and in patients with primary and secondary dystonia separately. RESULTS Of 301 children undergoing DBS for dystonia, 167 had primary dystonia, 125 secondary dystonia, and 9 myoclonus dystonia. Three dissociable preoperative phenotypes (latent variables) were identified and associated with the following: 1) BFMDRS-M at last follow-up; 2) relative change in BFMDRS-M score; and 3) relative change in BFMDRS-D score. The phenotype of patients with secondary dystonia, with a high BFMDRS-M score and truncal involvement, undergoing DBS at a younger age, was associated with a worse postoperative BFMDRS-M score. Children with primary dystonia involving the trunk had greater improvement in BFMDRS-M and -D scores. Those with primary dystonia of shorter duration and proportion of life with disease, undergoing globus pallidus DBS, had greater improvements in BFMDRS-D scores at long-term follow-up. CONCLUSIONS In a comprehensive, data-driven, multivariate analysis of DBS for childhood dystonia, the authors identified novel and dissociable patient phenotypes associated with distinct outcomes. The findings of this report may inform surgical candidacy for DBS.