Introduction Pallidal deep brain stimulation (DBS) has been shown to be effective in cervical dystonia (CD) with an improvement of about 50–60% in the Toronto Western Spasmodic Torticollis Rating (TWSTR) Scale. However, predictive factors for the efficacy of DBS in CD are missing with the anatomical location of the electrodes being one of the most important potential predictive factors.

Methods In the present blinded observational study we correlated the anatomical localisation of DBS contacts with the relative clinical improvement (CI %) in the TWSTR as achieved by DBS at different pallidal contacts in 20 patients with CD. Localisations of DBS contacts were derived from postoperative MRI-data following anatomical normalisation into the standard Montreal Neurological Institute stereotactic space. The CIs following 76 bilateral test stimulations of 24 h were mapped to stereotactic coordinates of the corresponding bilateral 152 active contacts and were allocated to low CI (<30%; n=74), intermediate CI (≥30%; <60%; n=52) or high CI (≥60%; n=26).

Results Euclidean distances between contacts and the centroid differed between the three clusters (p<0.001) indicating different anatomical variances between clusters. The Euclidean distances between contacts and the centroid of the cluster with high CIs correlated with the individual level of CIs (r=−0.61; p<0.0001). This relationship was best fitted with an exponential regression curve (r²=0.41).

Discussion Our data show that the clinical effect of pallidal DBS on CD displays an exponential decay over anatomical distance from an optimised target localisation within a subregion of the internal pallidum. The results will allow a comparison of future DBS studies with postoperative MRI by verifying optimised (for instance pallidal) targeting in DBS-treated patients.