(K-440) Modeling Evoked Neural Activity in the Subthalamic Nucleus: Implications for Deep Brain Stimulation Electrode Targeting and Mechanism of Action

High-frequency Deep Brain Stimulation (DBS) in the Subthalamic Nucleus (STN) is an effective treatment for late-stage Parkinson's disease. Precise targeting of the dorsolateral STN is crucial for therapeutic benefits, but accurate electrode implantation is challenging due to the nucleus’s small size. DBS surgeries are typically performed on awake patients, allowing surgeons to evaluate behavioral responses to acute electrical stimulation for target confirmation. This approach offers a reliable method to guide targeting, alongside microelectrode electrophysiology and magnetic resonance imaging. While many patients prefer asleep DBS surgery over traditional awake procedures, challenges arise from the inability to measure behavioral responses and the suppression of intrinsic neural activity due to anesthesia. STN DBS is known to elicit Evoked Neural Activity (ENA) and/or Evoked Resonant Neural Activity (ERNA) in the nuclei. These responses are most prominent in the dorsolateral STN and remain present in asleep patients, suggesting their potential to enhance electrode targeting. Analyzing these responses may also provide insights into the mechanism of action of DBS.

The purpose of this work was to understand the origin of ENA. To achieve this, we reconstructed ENA using a computational model of STN Local Field Potential (LFP). The LFP model system comprised of two main components: 1) the volume conductor, which was a finite element model of the human head including the DBS lead, and 2) the neural sources, which were multi-compartment STN neuron models. Fig 1A shows the model of the DBS lead surrounded by approximately 220,000 STN neurons (each represented by a green or blue dot). The neural compartments received excitatory (AMPA) and/or inhibitory (GABAa) synaptic inputs (Fig. 1B), mimicking the corticosubthalamic (hyperdirect pathway) and pallidosubthalamic inputs, respectively. While the globus pallidus externa (GPe) and the hyperdirect pathway were not explicitly modeled, we simulated their synaptic inputs to the STN.

Transmembrane currents from each compartment were simulated in the Neuron simulation environment and coupled with the volume conductor using a reciprocity-based solution in MATLAB. This integration enabled the simulation of evoked potentials (electrical voltages) recorded at the DBS electrodes.

The model successfully generated ENA that closely resembled the experimental recordings (Fig. 1C and 1D). ENA primarily consists of two positive peaks (P1 and P2) and one negative peak (N1). P1 emerges around 4 ms, followed by N1 at approximately 5 ms, and finally, P2 at around 7 ms. It is important to note that the first positive peak observed immediately after the DBS pulse in the experimental ENA is attributed to the stimulation artifact.

The model suggests that P1 is generated through the direct activation of pallidosubthalamic fibers (Fig. 2), leading to pronounced inhibition of the STN. This inhibition causes negative ions to flow into the STN cells, resulting in a positive extracellular potential observed as P1 (Fig. 2Bi). Furthermore, the activation of pallidosubthalamic fibers also leads to antidromic activation of the GABAergic GPe. The GPe's recurrent connections trigger self-inhibition, subsequently disinhibiting the STN. This combined disinhibition, along with some excitation through the hyperdirect pathway collaterals, induces a net excitatory effect on the STN, resulting in a negative evoked potential observed around 5 ms, marked as N1 (Fig. 2Bii). Lastly, the slower orthodromic activation of the GPe through subthalamopallidal fibers inhibits the STN again around 7 ms, resulting in a second positive peak, P2 (Fig. 2Biii).