(33) Kilohertz-frequency conduction block in models of unmyelinated nerve fibers

Introduction: Kilohertz-frequency block (KHF) is a neuromodulation method in which a high-frequency signal is applied to rapidly and reversibly block nerve conduction. Many applications of KHF would benefit from conduction block of small unmyelinated fibers, which carry many of the pain and autonomic signals that must be arrested to improve outcomes. However, existing KHF signals are almost exclusively designed based on the response of large somatic fibers, and there are no computational models of unmyelinated fibers designed to accurately represent KHF conduction block, making it impossible to computationally test signal waveforms. Here, we evaluated KHF blocking signals across a range of unmyelinated fiber models.

Methods: We used Python + NEURON to simulate widely used computational models of unmyelinated axons, including a warmed Hodgkin-Huxley model (Rattay), models with more complex ion dynamics (Sundt, Tigerholm, Schild 1994, Schild 1997) and a model of a sympathetic preganglionic neuron (Briant). Each model was simulated as a single axon of length 8.3 mm and 1 μm diameter. We applied a 10 Hz test pulse train intracellularly at 10% of the length of the axon to evoke action potentials, then measured action potential transmission at 90% of the axon length. We applied a sinusoidal blocking signal at 20 kHz using an extracellular point source located 100 μm away from the axon at 50% of its length.

Results: Sinusoidal KHF signals blocked conduction in the Rattay, Sundt, Tigerholm, and Briant models, but did not produce blocking effects in either Schild model. The signal amplitude necessary to arrest nerve conduction varied widely depending on model, with blocking thresholds of 30 mA (Rattay), 110 mA (Sundt), 18 mA (Tigerholm), and 5 mA (Briant).
During KHF block, myelinated fibers commonly produce an onset response, in which the KHF signal evokes increased activity before the blocking effect sets in. An onset response was present in the Briant and Sundt models, but not the Tigerholm or Rattay models. Additionally, myelinated fiber models often exhibit re-excitation at signal amplitudes well above block threshold; we observed no re-excitation in the unmyelinated fiber models.

Conclusion: The variance in blocking thresholds and presence of onset response across these models indicates the need for models of unmyelinated fibers that are calibrated to accurately reproduce the characteristics of KHF block. Tuning unmyelinated fiber models to experimental measurements has the potential to improve KHF signal waveforms and generate block of unmyelinated fibers that may be required for some clinical applications.