Closing the Loop: Exploring the Use of Sacral Dorsal Root Ganglia Signals for Adaptive Neuromodulation of Bladder Function

Abstract

Overactive bladder (OAB) is a highly prevalent condition which negatively affects the physical and mental health of millions of people worldwide. Sacral neuromodulation (SNM) is a third-line therapy that provides improved efficacy and less adherence issues as compared to conventional treatments. There have been ~300,000 SNM implants since the therapy was first introduced over 20 years ago. While SNM is delivered in an open-loop fashion, the therapy could have improved clinical efficacy by adopting a closed-loop stimulation paradigm that uses objective physiological feedback. One promising approach to obtain such feedback is by tapping into the nervous system that innervates the bladder. This dissertation focuses on using sacral level dorsal root ganglia (DRG) neural signals to provide sensory feedback for adaptive SNM a feline model. This work began with exploring machine learning algorithms and feature selection methods for bladder pressure decoding in an offline analysis of DRG signals. A Kalman filter delivered the highest performance based on correlation coefficient between the pressure measurements and algorithm estimation. Additionally, firing rate normalization significantly contributed to lowering the normalized error, and a correlation coefficient-based channel selection method provided the lowest error compared to other channel selection methods. Following algorithm optimization, this work implemented the optimized algorithm and feature selection method in real-time in anesthetized healthy bladder and simulated OAB feline models. A 0.88 ± 0.16 decoding correlation coefficient fit was achieved by the algorithm across 35 normal and simulated OAB bladder fill sequences in five experiments. Additionally, closed-loop neuromodulation was demonstrated using the estimated pressure to trigger pudendal nerve stimulation, which increased bladder capacity by 40% in two trials. Finally, closed-loop SNM with the DRG sensory feedback algorithm was performed in anesthetized experiments. Our approach increased bladder capacity by 13.8% over no stimulation (p < 0.001). While there was no statistical difference in bladder capacity between closed-loop and continuous stimulation (p = 0.80), closed-loop stimulation reduced stimulation time by 57.7%. Interestingly, clearly-identified bladder single units had a reduced sensitivity during stimulation, suggesting a potential mechanism of SNM. This dissertation also developed a method for chronic behavioral monitoring and neuromodulation of bladder function in a feline model. We tracked urodynamic parameters across multiple week testing intervals. We observed that animals could tolerate pudendal nerve stimulation above motor threshold. Interestingly, stimulation at 5 and 33 Hz appeared to have a modulatory effect on voiding interval and efficiency in line with prior work under anesthesia. Overall, this work demonstrated that sacral level DRG are a viable sensory feedback target for adaptive SNM. This dissertation also investigated a behavioral paradigm that will be useful for system validation in awake and chronic experiments. Behavioral experiments such as these, as well as development of low-power systems for adaptive monitoring and feedback, are a crucial step prior to clinical translation of this method. Ultimately, implementation of closed-loop adaptive SNM will lead to an improved therapy and greater potential benefit for the millions of individuals with OAB.