The goal of this study was to integrate detailed computer modeling with clinical outcomes analysis to enhance understanding of the effects of DBS of the subthalamic region. We developed 10 patient-specific models of unilateral DBS based on neuroimaging, neurophysiology, neuroanatomy, and neurostimulation data. Our results suggest that direct stimulation of the STN is only one of multiple neuroanatomical territories in the STN region that may play a role in the therapeutic benefit achieved with DBS.
Our theoretical models show that direct activation of ~70 mm3
of axonal tissue dorsal, lateral, and posterior to the geometric center (or centroid) of the atlas defined STN generated therapeutic benefit from DBS (). This general area includes the sensorimotor territory of the STN that is believed to be involved in motor control, and whose physiological activity is altered in Parkinson’s disease [38
]. Patients with the best clinical outcomes also tended to have a higher percentage of direct stimulation of axonal tissue outside of, and dorsal to, the STN. Similar conclusions have been reached by several previous investigations examining the anatomical location of therapeutic electrode contacts [e.g. 16
], while others have suggested that optimal DBS contacts were located in the dorsolateral sensorimotor STN, but not the white matter dorsal to the STN [e.g. 17
Debate on the “optimal” implantation location for DBS electrodes will undoubtedly continue over the next decade as new techniques enable more detailed analysis of the anatomical, electrical, and behavioral variables of DBS. The evolutionary addition of this study to the previous literature is the quantitative integration of clinical outcomes analysis and electrically accurate models of the spread of stimulation [24
]. This process allowed us to critically examine the interaction between the volumes of tissue activated (VTA) and the underlying neurophysiology and neuroanatomy.
Given that the fundamental purpose of DBS is to modulate neural activity with electric fields, it is imperative that scientific analyses of DBS attempt to account for the variables associated with clinical stimulation parameter selection (contact, impedance, voltage, pulse width, frequency) and the resulting spread of stimulation relative to the anatomy. Our patient-specific DBS modeling system uses some of the most advanced neurostimulation prediction techniques currently available. However, it should be noted that there are several limitations in this study. First, we selected patients with monopolar stimulation to simplify calculations of the neural response to DBS. This selection criterion may have biased our analysis away from patients with lateral electrodes who have capsular spread limiting benefit, as these patients are typically reprogrammed to bipolar stimulation. Second, the co-registration of multiple images and atlas representations of the patient creates spatial variability that can not be ignored. We attempted to minimize co-registration error by using easily identifiable landmarks such as the AC / PC and widely accepted co-registration algorithms. Third, while we extended great effort to place all of our data into the stereotactic coordinate system of the patient to utilize MER data in the most accurate way possible, one caveat was the inherent uncertainty in the intra-operative electro-physiologist’s anatomical designation of the recordings. However, we used established criteria (e.g. increased background activity followed by the presence of neuronal activity with discharge patterns similar to that previously described by the STN, along with the presence of sensorimotor responses) to make the MER designations used in our study [38
]. Fourth, outside of histological reconstruction it is impossible to know the exact size, shape, and location of the STN in a given patient [44
]. We relied on a 3D atlas model fit to match boundaries defined by the recorded neurophysiology and neuroanatomy visible on the MRI. Fifth, due to signal-to-noise considerations, the DTI brain atlas used in this study was acquired with relatively large voxel sizes [37
]; therefore, the 3D tissue conductivities used in the model only represent a gross estimate. Sixth, the VTA prediction functions used in the model were derived from the activation of straight, relatively large diameter myelinated axons, and may not be representative of the response of other neuron types surrounding the electrode (local projection neurons, local interneurons, afferent inputs, etc.). Given that myelinated axons are considered the most excitable neuron type to extracellular electrical stimulation [47
]; our VTA predictions should be considered an upper limit in terms of stimulation spread.
Experimental validation of the VTA predictions is a difficult task. We are actively pursuing research studies that link our DBS models with electrophysiological recordings in humans and non-human primates [26
]. The results of these studies show that our models can accurately predict stimulation spread into the corticospinal tract during STN DBS, and the synergistic evolution of our modeling technology and experimental analysis will allow for continuous improvement in their accuracy and validity. Nonetheless, we believe that the patient-specific DBS modeling system used in this study is capable of making quantitative, clinically relevant, predictions.
Our results suggest that stimulation of axonal tissue dorsal, lateral, and posterior to the centroid of the STN maximizes therapeutic benefit from DBS. However, every patient’s disease pathology, electrode location, and behavioral response to stimulation are different. For example, patients 2, 7, and 9 all had therapeutic VTAs with similar sizes and anatomical locations, but their therapeutic outcomes showed substantial differences. In turn, maximizing therapeutic benefit for an individual DBS patient involves more than just electrode placement and a VTA calculation, as many variables unaccounted for in this study could impact the behavioral response to DBS. For example, it is possible that based on the patient’s symptoms one electrode location may be preferential to another, or stimulation spread into one anatomical region may be preferential to another. In turn, the interplay between the patient and clinician performing the DBS parameter selection is critical in defining the balance between therapeutic benefit and side effects. However, this clinical process is typically done without the opportunity to visualize the regional spread of stimulation and its location with respect to the surrounding anatomy. This could be an important issue in patients like 6 and 8 where the model suggests that the electrodes are in a good location, but the stimulation parameter settings may not be optimal because of stimulation spread into the internal capsule. And the converse is suggested with patient 10 where a lateral electrode location limits the allowable size of the VTA to avoid spread into internal capsule. Therefore, the next step along this line of research is to couple patient-specific DBS model predictions with prospective clinical evaluations to develop new and improved techniques to optimize the clinical efficacy of DBS. For example, methodology from this study may find utility in augmenting DBS surgical placement planning [46
], and the post-operative stimulation parameter selection process [52