(45) Personalized Models Predict Perception Thresholds and Activation Patterns in Spinal Cord Stimulation for Chronic Pain

Introduction: Spinal cord stimulation (SCS) is an electric stimulation therapy widely used for chronic pain management. However, the underlying mechanisms of SCS remain unclear, posing challenges in optimizing treatment. In our randomized, double-blind, placebo-controlled clinical trial (NCT04732325), we assessed how different SCS paradigms (i.e., 50Hz, 1kHz, and Burst SCS) affect sensory processing. We hypothesized that patient-specific computational models can be used to quantitatively describe the neural response to SCS, thereby addressing inter-patient variability in neural activation and promoting the development of personalized SCS therapies.

Methods: To describe the direct neural response to SCS, we built individualized computational models that considered variations in anatomy, lead placement, and stimulation parameters. To calculate the extracellular potentials generated by SCS, we built patient-specific finite element method models derived from preoperative magnetic resonance imaging and postoperative X-ray computed tomography. Next, we calculated the neural response to SCS by applying these model-based extracellular potentials to multi-compartment axon models distributed throughout the spinal cord.

Results: We compared the neural activation patterns predicted by computational models to regions of patient-reported paresthesias (e.g., tingling) in 12 participants. Our patient-specific models accurately predicted clinical perception amplitudes (i.e., the stimulation amplitude at which participants first feel paresthesia) with a small percentage of ≥ 8.7 microns axon activation in the spinal cord, achieving mean absolute percent errors of 22%, 21%, and 27% for 50Hz, Burst, and 1kHz SCS, respectively. Our models accurately predicted activated dermatomes based on the somatotopic organization in the dorsal columns, suggesting that 1kHz SCS activates a smaller dorsal column area at the clinical perception threshold and demonstrates patient-specific models' potential to capture unique neural activation profiles across settings and patients.

Conclusion: Existing literature indicates that 1kHz SCS causes rapid asynchronous firing [1,2], and both spatial and temporal summation are thought to play a role in influencing the perceived paresthesia intensity. [3] Higher firing rates generated in dorsal column axons may lead to a stronger input (for a given number of axons) to dorsal nuclei neurons; thus, fewer activated axons are needed to generate paresthesia perception. However, these differences in spatial activation at the perception threshold may also be related to the different pulse widths used for each paradigm: 300 μs for 50Hz, 1000 μs for Burst, and 180 μs for 1kHz SCS. [4,5] Moving forward, we will continue to develop patient-specific models to characterize neural recruitment during different forms of SCS across additional participants.