Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is used for advanced Parkinson’s disease treatment to improve the quality of life of patients by alleviating motor symptoms, as well as reduce dopaminergic medication requirements. Nevertheless, the success of this treatment modality depends on the correct selection of stimulation parameters, such as pulse width, the relative distribution of electric currents across contacts, the adaptation of amplitude, and stimulation frequency.
Study: Automated deep brain stimulation programming based on electrode location: a randomized, crossover trial using a data-driven algorithm. Image Credit: PopTika / Shutterstock.com
At present, the optimization of DBS parameters is based on clinical testing, which requires highly skilled medical personnel who can adjust DBS settings for certain therapeutic and negative effects. This optimization method is extremely time-consuming and subject to many factors, such as symptom fluctuations, patient fatigue, and delayed response to parameter adjustments. Thus, when following this optimization procedure, only a fraction of the vast number of parameter combinations can be assessed, which presents the risk of selecting suboptimal conditions.
For the complete utilization of modern DBS systems to achieve their maximum therapeutic potential, data-driven algorithms must be developed to guide DBS programming by introducing a subset of stimulation parameters. Electrode localization, for example, is extremely important and has been linked to both positive and adverse DBS effects across different stimulation targets and diseases.
Several commercial software programs can offer visual feedback of stimulation and electrode location data of patients’ anatomy, which is used in clinical programming procedures. The incorporation of these technologies has significantly reduced the time required for clinical programming.
Image-guided optimization of DBS parameters has been associated with some challenges, including the requirement of manual adjustments of DBS parameters within the software. Furthermore, this approach is based on many pre-assumptions, such as fiber diameters and their arrangement in the presence of an electric field, many of which are unknown.
About the study
A recent Lancet Digital Health study assesses the treatment effects of stimulation parameters suggested by a recently published data-driven algorithm (StimFit) based on neuroimaging data. StimFit was trained and tested using a large dataset of over 600 different stimulation settings applied in fifty patients with Parkinson’s disease.
The current randomized, double-blinded, 2×2 crossover, non-inferiority trial was conducted at Charité – Universitätsmedizin in Berlin, Germany. This study enrolled patients diagnosed with Parkinson’s disease, according to the British Parkinson’s Disease Society Brain Bank without neuropsychiatric symptoms, severe cognitive impairment, or severe cerebral atrophy.
Patients were treated with directional octopolar electrodes targeted at the STN or with SenSight directional leads. All participants, before being recruited, had undergone DBS programming between three months and three years prior to the start of the trial, according to the center’s standard of care (SoC) treatment.
DBS electrodes were reconstructed based on perioperative imaging data, whereas StimFit was used for optimal stimulation settings. Study participants underwent motor assessments from the Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale part III (MDS-UPDRS-III) during OFF-medication, OFF-stimulation, and ON-stimulation states, as per StimFit and SoC parameter settings.
In this study, patients were randomly assigned to receive either StimFit-programmed DBS first and SoC-programmed DBS second, or SoC-programmed DBS first and StimFit-programmed DBS second. The allocation schedule was developed based on a computerized random number generator.
A total of 35 eligible patients were included in the present study, 18 of whom were subjected to StimFit followed by SoC stimulation, whereas 17 patients first received SoC followed by StimFit stimulation. The main outcome of the study was based on the mean difference between MDS-UPDRS-III scores under StimFit and SoC stimulation, with a non-inferiority margin of five points.
Based on kinetic-rigid and axial subscores, both stimulation conditions led to considerable motor improvements of 48% and 43% by SoC and StimFit, respectively, as compared to the OFF-stimulation baseline. However, tremors responded significantly less to StimFit stimulation. These findings are in line with previous long-term SoC treatment, which reported a decrease in dopaminergic medication (LEDD) by 57% as compared to the control group.
For Parkinson’s disease, several studies have pointed out an anatomical segregation of DBS “sweet spots” for controlling tremors on one hand and rigidity on the other hand. This implies a need for personalized DBS programming procedures, which was realized through the StimFit software. In fact, StimFit starts the optimization procedure by determining the efficacy and side effects at different amplitudes and identifying ideal monopolar solutions.
Although more complex electrodes are being developed with increased numbers of contacts for a wider therapeutic window of DBS, its therapeutic potential is limited due to a smaller number of parameter combinations that can be explored in clinical trials.
A key strength of the current study is the use of a data-driven algorithm that can suggest optimal stimulation parameters in patients diagnosed with Parkinson’s disease and treated with STN-DBS based on electrode location in a fully automated fashion. In the future, more longitudinal studies are needed to determine long-term motor benefits and evaluate the impact of data-driven DBS programming on dopaminergic medication, quality of life, and programming time.
- Roediger, J., Dembek, T. A., Achtzehn, J., et al. (2023) Automated deep brain stimulation programming based on electrode location: a randomised, crossover trial using a data-driven algorithm. Lancet Digit Health, 5: e59–70. doi:10.1016/ S2589-7500(22)00214-X