Preoperative freezing of gait predicts long-term gait outcomes after deep brain stimulation in Parkinson’s disease

Introduction: Freezing of gait (FOG) is a disabling and often treatment-refractory symptom of Parkinson’s disease (PD). Despite the motor benefits of subthalamic nucleus deep brain stimulation (STN-DBS), postoperative FOG outcomes remain heterogeneous. This study aimed to identify preoperative clinical, neuropsychological, and dopaminergic predictors of short- and long-term FOG outcomes following STN-DBS in Parkinson's disease. Methods: This retrospective study included 91 patients with PD (mean age 62.0 ± 7.6 years; 63.7 % male; disease duration 10.5 (8–13) years) who underwent bilateral STN-DBS and had a median postoperative follow-up of 36 months. Preoperative evaluation comprised clinical assessment with the Movement Disorders Society Unified Parkinson’s Disease Scale (MDS UPDRS), a standardized levodopa challenge test, and a comprehensive neuropsychological assessment including the Montreal Cognitive Assessment (MoCA), Trail Making Test Part B, Stroop Test, Judgement of Line Orientation, and Symptom Checklist-90. Levodopa equivalent daily dose (LEDD) was quantified, and patients were classified as preoperative freezers or non-freezers and according to MDS-UPDRS motor phenotype - tremor-dominant, bradykinesia-rigidity, or postural instability/gait disorder (PIGD). Early responsiveness to levodopa and DBS was defined as a ≥1-point improvement on FOG-specific items at 3 months. Long-term outcomes were assessed as the ever-occurrence of FOG in Med-OFF/DBS-ON and Med-OFF/DBS-OFF states across follow-up intervals. A supplementary video-based gait analysis including turning was available for a subset of 44 patients (Med-ON preDBS vs. Med-ON/DBS-ON at 1 year). Statistical analyses included Fisher’s exact test and logistic regression. Results: Preoperative FOG status was a strong and consistent predictor of postoperative freezing. Preoperative freezers were more likely to exhibit FOG during long-term follow-up in both DBS-OFF (OR = 4.03; 95% CI 1.44–11.29; p = 0.012) and DBS-ON states (OR = 3.80; 95% CI 1.23–10.20; p = 0.024). Notably, 47.6 % of preoperative non-freezers developed de novo FOG within 36 months. Preoperative responsiveness of FOG-specific MDS-UPDRS items 3.11 and 2.13 to levodopa challenge significantly predicted early improvement of FOG after DBS (OR = 2.55; 95% CI 1.05–6.21; p = 0.047). Neuropsychological test scores showed no predictive value. No significant differences were observed in preoperative LEDD between patients with and without FOG (p ≥ 0.673). Furthermore, the presence of a PIGD phenotype was not significantly associated with postoperative FOG status in both DBS-OFF (p = 0.660) or DBS-ON (p = 1.000) conditions. In the video-based subset, preoperative FOG remained a significant predictor of FOG persistence at 1 year: patients with preoperative FOG were seven times more likely to exhibit postoperative FOG (OR = 7.0; 95% CI (1.50, 32.72); p = 0.015). Conclusions: Preoperative FOG status, objective video-based assessment, and levodopa responsiveness are key prognostic markers of gait outcomes after DBS. Although DBS provides early symptomatic relief, the high rate of de novo FOG and the persistence of symptoms in preoperative freezers highlight the progressive, multisystem nature of gait impairment in PD. These results point to the need for therapeutic approaches that better address the underlying mechanisms of FOG.