APPLICATION OF HIGH FREQUENCY ELECTRICAL BLOCK ON THE EFFERENT NERVES TO THE LOWER URINARY TRACT FOR BLADDER VOIDING.

Transcription

1 i APPLICATION OF HIGH FREQUENCY ELECTRICAL BLOCK ON THE EFFERENT NERVES TO THE LOWER URINARY TRACT FOR BLADDER VOIDING By Adam Sprott Boger Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy Dissertation Adviser: Dr. Kenneth J. Gustafson Department of Biomedical Engineering CASE WESTERN RESERVE UNIVERSITY May, 2009

2 CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES We hereby approve the thesis/dissertation of candidate for the degree *. (signed) (chair of the committee) (date) *We also certify that written approval has been obtained for any proprietary material contained therein.

3 iii DEDICATION To my parents, Kenneth and Robin Boger. Your love, guidance and support made my achievements possible. And to my sisters, for being such wonderful, talented young women. And finally to my friends, for their support.

4 iv TABLE OF CONTENTS Title Page.... i Committee Signature Page.... ii Dedication...iii Table of Contents... iv List of Tables... ix List of Figures...x Acknowledgements...xiii Abstract... 1 Chapter I: Introduction... 3 I.1 Clinical Problem...4 I.2 Lower Urinary Tract Neurophysiology...5 I.2.1 Bladder...6 I.2.2 Urethra...6 I.2.3 Urethral Sphincters...7 I.2.4 LUT Innervation...7 I.2.5 Central Reflex Pathways...8 I.2.6 Bladder and EUS Reflexes During Voiding...9 I.3 Bladder Voiding Neuroprotheses I. 4 Electrical Nerve Block I.4.1 High Frequency (HF) Block...12 I.4.2 HF Stimulation Response...13

5 v I.4.3 HF Stimulation Safety...14 I.4.4 Collision Block and Quasitrapezoidal Stimulation...15 I.6 Specific Aims I.6.1 Demonstrate Bladder Voiding Using HF PN Block and Sacral Bladder Drive...18 I.6.2 Characterize Sacral Small Fiber HF Transmission Block...19 I.6.3 Demonstrate Bladder Voiding Using Sacral HF Block and Bladder Drive...20 I.7 Significance Chapter II: Bladder Voiding by Combined High Frequency Electrical Pudendal Nerve Block and Sacral Root Stimulation...29 II.1 Abstract II.2 Introduction II.3 Methods and Materials II.3.1 Experimental Preparation...33 II.3.2 Experimental Set up...34 II.3.3 Experimental Procedure...36 II.3.4 Electrical Stimulation...37 II.3.5 Data Analysis...38 II.4 Results II.5 Discussion II.6 Conclusions Chapter III: High Frequency Electrical Conduction Block of Small Myelinated Bladder Efferents...48 III.1 Abstract III.2 Introduction... 50

6 vi III.3 Methods III.3.1 Experimental Setup...52 III.3.2 Surgical Preparation...53 III.3.3 Electrical Stimulation...54 III.3.4 Experimental Design...54 III.3.5 Data Analysis...58 III.4 Results III.4.1 Transmission Block...59 III.4.2 HF Evoked Bladder Response...62 III.5 Discussion III.5.1 HF Block Amplitude and Frequency Thresholds...64 III.5.2 HF Neuroprotheses Design Implications...65 III.5.3 Validating HF Block Simulations...66 III.5.4 Chronic Pain and Bladder Voiding Neuroprostheses...67 Chapter IV: High Frequency Sacral Root Nerve Block Allows Bladder Voiding77 IV.1 Abstract IV.2 Introduction IV.3 Methods IV.3.1 Electrical Stimulation...82 IV.3.2 Trials...83 IV.3.3 Data Extraction and Analysis...85 IV.4 Results IV.4.1 Voiding...86 IV.4.2 EUS Block Quality...87 IV.5 Discussion... 88

7 vii IV.5.1 Sacral EUS Block for Voiding...88 IV.5.2 Sacral Block Extends PN Block...89 IV.5.3 Voiding Limitations...90 IV.5.4 Non Voiding Animals...90 IV.6 Conclusion Chapter V: Conclusion...99 V.1 Summary of Results V.2 Neuroprosthesis Design Choices V.2.1 Block Location V.2.6 Block Parameters V.3 Future work V.3.1 Chronic HF stimulation Safety and Stability V.3.2 What determines safe HF stimulation? V.3.3 Translation of Acute Studies V.3.4 Human Clinical Trials V.3.5 Nocioceptive and Autonomic Nerve Block V.4 Conclusion Appendix A: High Frequency Pudendal Nerve Block Parameters For Human Approved Electrodes A.1 Methods A.2 Results A.3 Discussion Appendix B: Wrap Around Activation During High Frequency Stimulation

8 viii B.1 High Amplitude Proximal Stimulation Prevented Block Appendix C: Sacral Nerve Root Identification, Length and Diameter C.1 Intra operative Sacral Root Identification C.2 Post Operative Sacral Root Verification C.3 Sacral Nerve Root Lengths and Diameters Appendix D: Experimental Setup Bibliography

9 ix LIST OF TABLES Chapter I: Introduction Chapter II: Bladder Voiding by Combined High Frequency Electrical Pudendal Nerve Block and Sacral Root Stimulation Table I: Trial Distribution by Control Variables, Initial Bladder Volumes for each Randomized Set of Trials..47 Chapter III: High Frequency Electrical Conduction Block of Small Myelinated Bladder Efferents Table I: Distribution of Trials by Control Variables..76 Chapter IV: High Frequency Sacral Root Nerve Block Allows Bladder Voiding Table IA: Block and Bladder Pressure Characteristics by Animal..98 Table IB: Distribution of Trials by Control Variables.98 Chapter V: Conclusion Appendix A: High Frequency Pudendal Nerve Block Parameters For Human Approved Electrodes Table I: HF Block Effectiveness and Parameters for Four Human Approved Electrodes Appendix B: Wrap Around Activation During High Frequency Stimulation Appendix C: Sacral Nerve Root Identification, length and Diameter Appendix D: Experimental Setup Table I: Supplier Information for Experimental Equipment

10 x LIST OF FIGURES Chapter I: Introduction Figure 1: Pelvic, Hypogastric, and Pudendal Nerves Innervate the Lower Urinary Tract Figure 2: Suprasacral Coordination of Segmental Reflexes Maintains Appropriate LUT ffunction Figure 3: Bladder and Sphincter Reflexes During Continence and Micturition,,,..25 Figure 4: Evoked Response to High Frequency Stimulation Figure 5: Electrode Placement for Voiding Using Pudendal Nerve Block and Sacral Bladder Drive Figure 6: Electrode Placement for Small Fiber Block and Sacral Root Stimulation for Bladder Voiding Chapter II: Bladder Voiding by Combined High Frequency Electrical Pudendal Nerve Block and Sacral Root Stimulation Figure 1: Bladder Pressure and Flow Recordings from Representative Trials.. 44 Figure 2: HF Stimulation Increased the Percentage of the Initial Bladder Volume Voided Figure 3: HF Stimulation Reduced Maximum Bladder Pressure During Voiding..46 Chapter III: High Frequency Electrical Conduction Block of Small Myelinated Bladder Efferents Figure 1: Electrodes Implantation Locations Figure 2: Description of Transmission Block Trials Figure 3: Validation of Conduction Block Mechanism.. 71 Figure 4: Description of HF Evoked Bladder Response Trials

11 xi Figure 5: Transmission Block was Complete, Rapid and Reversible Figure 6: Transmission Block Characteristics Improve with Increasing Stimulation Amplitude Figure 7: Increasing Frequency of Stimulation Improves HF Evoked Bladder Response Chapter IV: High Frequency Sacral Root Nerve Block Allows Bladder Voiding Figure 1: Electrodes Locations During Experiment Figure 2: Stimulus Timings and Extracted Variables Figure 3: Sacral HF Stimulation Prevents EUS Activity Figure 4: EUS HF Max Trials Achieved Clinical Effective Bladder Voiding Figure 5: High Frequency Stimulation Trials and Controls Trials Exhibit Similar Urodynamics Chapter V: Conclusion Appendix A: High Frequency Pudendal Nerve Block Parameters For Human Approved Electrodes Appendix B: Wrap Around Activation During High Frequency Stimulation Figure 1: Proximal Stimulation may Excite the Nerve Distal to the Location of HF Conduction Block Figure 2: Increasing proximal Stimulation Amplitude Decreases Block Ratio Appendix C: Sacral Nerve Root Identification, Length and Diameter Figure 1: Post operative Verification of Correct Electrode Cuff Implantation Appendix D: Experimental Setup Figure 1: Pudendal Nerve Cuff with Stimulation and Block Electrodes

12 xii Figure 2: Sacral Root Nerve Cuff with Stimulation and Block Electrodes Figure 3: Tripolar Sacral Root Nerve Cuff Electrodes Figure 4: Bipolar Sacral Root Nerve Cuff Electrodes Figure 5: HF PN Block for Voiding (Chapter II) Figure 6: HF Sacral Root Block of Small Parasympathetic Bladder Efferents (Chapter III) Figure 7: HF Sacral Root Block for Voiding (Chapter IV) Figure 8: Isolated Pulse Stretcher Schematic

13 xiii ACKNOWLEDGEMENTS I would like to thank my advisor and mentor Dr. Kenneth J. Gustafson for the guidance he has provided me during my time in his lab. Without his patience and encouragement, resources and willingness to allow me to pursue my ideas, my accomplishments would have been few and far between. I want to thank my mentor Dr. Narendra Bhadra. I am so grateful for his wise counsel and quiet support, and for time and effort he took to train me, giving me a second chance to overcome my hesitancy and develop surgical skills. His efforts and surgical skills provided the consistent, undamaged preparations that allowed my experiments to be conducted. I must acknowledge my colleagues on the P-Team, Tim Bruns, Tim Mariano, and Jaime McCoin, for all their experimental assistance. Wherever I might have been without you, I doubt I would have been getting much sleep. I would also like to thank Tina Emancipator for her experimental assistance, and for keeping everything running smoothly in the surgical suite, even when the experiments kept coming week after week. I would like to thank my committee members Drs. Dominique Durand, Kevin Kilgore, and Pedram Mohseni for their thorough and helpful criticism, feedback, and suggestions. Their input has prepared me such that I felt comfortable presenting to any audience. I must thank my family for supporting my pursuit of my doctorate. Please forgive all the missed family vacations and the weeks of phone silence vaguely explained with references to experiments and assurances that you really don t want to know.

14 xiv To all my friends in the Biomedical Engineering Department and at Case: I have enjoyed my time in Cleveland because of you. In particular I want to thank Michael Ackermann for his humor, support and late night bull sessions; Christa Wheeler for her kindness and positive effect on my GPA; Abirami Muralidharan for being such a sweetheart and for politely ignoring my inability to say her name correctly; and Natalie Brill and Aaron Hadley for being Awesome. This work was supported by The State of Ohio BRTT ONNP 03-10, Department of Veterans Affairs RR&D B3675R, NIH grants AR07505, EB004314, DK077089, EB00209, HD40298, and the Cleveland VA FES Center.

15 1 APPLICATION OF HIGH FREQUENCY ELECTRICAL BLOCK ON THE EFFERENT NERVES TO THE LOWER URINARY TRACT FOR BLADDER VOIDING Abstract By ADAM SPROTT BOGER Individuals with neurologic disease or injury such as spinal cord injury often develop dyssynergic lower urinary tract reflexes, which can prevent bladder-voiding and compromise their health. Permanent and destructive neurotomies can allow bladdervoiding, but eliminate residual sacral sensation and disrupt reflexes governing defecation and sexual function, severely affecting quality of life. High frequency electrical (HF) stimulation can immediately and reversibly block nerve impulse conduction, effectively providing a reversible neurotomy. This dissertation demonstrated effective bladder voiding in an acute feline model using HF stimulation. Complete bladder voiding was achieved using sacral root stimulation to activate the bladder and either PN or sacral root HF block to prevent activation of the external urethral sphincter (EUS). Effective bilateral HF PN EUS block was achieved, allowing complete bladder-voiding equivalent to voiding following pudendal neurotomy and reducing maximum bladder pressure during voiding. PN HF block for voiding extended previous research demonstrating unilateral HF PN EUS block. Sacral HF EUS block allowed bladder voiding comparable to voiding in the absence of EUS activation. Complete bladder voiding could be achieved through successive HF

16 2 block trials or by extending the HF block trial duration. Development of bladder-voiding neuroprostheses utilizing sacral HF block may require only modification of existing technology. Both sacral and PN HF block bladder-voiding neuroprostheses significantly improve upon existing approaches by preserving sacral reflexes and sensation. Thus HF based bladder-voiding neuroprostheses expand the population of patients willing to consider bladder-voiding neuroprostheses, potentially improving patient quality of life and decreasing the cost of patient care. HF block of small (1-3 µm) myelinated parasympathetic bladder efferents at the sacral root level was also demonstrated. Small fiber block characteristics were dependent on stimulation frequency and amplitude and HF block was localized to the region near the blocking electrode. Complete, rapid nerve block and the minimization of HF evoked bladder responses could be achieved with the appropriate choice of stimulation parameters. The demonstration of small fiber HF block supports research into HF block neuroprostheses for chronic pain or autonomic disorders. Such neuroprostheses could benefit an extremely large patient population currently lacking many treatment options.

17 3 CHAPTER I: INTRODUCTION

18 4 I.1 Clinical Problem As of 2007, there were approximately 227,000 to 301,000 patients living with spinal cord injury (SCI) in the United States (NSCISC, 2008). Approximately 12,000 new patients were injured every year. Injuries typically occurred at the cervical or thoracic level; neither complete nor incomplete injuries predominate. The typical SCI patient is young, male, and was injured in an auto accident (Nobunaga et al., 1999). Advances in care have greatly extended life expectancies for spinal cord injury patients. Society incurs significant costs treating spinal cord injury. The average present value of lifetime charges for SCI patient care, assuming a 2% discount rate, range from million dollars, depending on the severity of the injury (NSCISC, 2008). The total cost to society is likely far greater, as the direct spinal cord injury costs are likely exceeded by the indirect costs (DeVivo, 1997). Lower urinary tract (LUT) complications represent a significant component of the overall cost of SCI. Following SCI, sacral reflexes governing micturition and continence often become disordered. The External Urethral Sphincter (EUS) and bladder may contract simultaneously, a condition termed detrusor-sphincter dyssynergia (DSD). Complications arising from DSD, including urinary tract infections, autonomic dysreflexia, vesico-ureteric reflux and hydronephrosis, were once the primary cause of death for SCI patients (Hackler, 1977). The mortality associated with LUT complications has decreased (Frankel et al., 1998), but such complications remain among the most common reasons for patient re-hospitalization (Cardenas et al., 2004; French et al., 2007). The conventional first-line treatment for DSD is anti-muscarinic drugs combined with clean intermittent catheterization. This combination has proven to reduce LUT

19 5 complications (Waller et al., 1995). However, long term patient compliance with this regime is extremely poor (Yavuzer et al., 2000). The majority of male spinal cord injury patients will typically use condom catheters for bladder management, while the majority of female patients will use indwelling catheters (Sekar et al., 1997). However, indwelling catheters increase the risk of bladder cancer (West et al., 1999) while condom catheters require unobstructed voiding, which may require a sphincterotomy or a urethral stent (Reynard et al., 2003). EUS paralysis using injections of botulinum toxin can also improve voiding, but the effect is temporary (de Seze et al., 2002; Rackley and Abdelmalak, 2004). Surgical treatments, such as sphincterotomies, urethral stents, or pudendal nerve (PN) neurotomies, present an alternative should conventional treatments prove ineffective. However, these methods are destructive and may prove ineffective. Sphincterotomies may not establish low pressure and low post-void residual (PVR) micturition and often possess side effects, including recurrent urinary tract infections and urethral strictures (Ahmed et al., 2006; Reynard et al., 2003). Stents may fail due to encrustation or migration. Permanent stents can be difficult to remove should they fail, while removable stents are best considered a temporary solution (Mehta and Tophill, 2006). PN neurotomies have been shown to achieve low pressure voiding with low postvoid residual (PVR) (Engel and Schirmer, 1974), but may result in incontinence. I.2 Lower Urinary Tract Neurophysiology The lower urinary tract has two primary functions: continence and micturition. These functions are accomplished through a deceptively simple arrangement of storage container (bladder), outlet (urethra) and valve (external urethral sphincter and periurtheral

20 6 sphincter). Three nerves innervate these structures: the pelvic, hypogastric and pudendal nerves. The pelvic and hypogastric nerves are part of the autonomic nervous system, while the pudendal nerve innervates somatic muscles. This combination of autonomic and somatic innervation is unusual, as is the associated mixture of unconscious reflexes and voluntary control. I.2.1 Bladder The function of the bladder is to store urine until voiding. The bladder consists of three layers, an outer membranous layer, a middle layer of smooth muscle, and an inner layer of urothelium. The smooth muscle layer, the detrusor, is innervated primarily by post-ganglionic parasympathetic nerve fibers. The trigone is defined on the posterior surface of the bladder by the internal urethral meatus and the ureteric orifices. The trigone is mainly innervated by sympathetic efferents (Roberts, 2008). A sensory plexus is concentrated in the region of the trigone and the bladder neck. This plexus contains sensory nerve ending of afferents mediating sensations of bladder filling and pain (Andersson, 2002). I.2.2 Urethra The urethra is the outlet of the lower urinary tract system. It consists of a fibromuscular tube 3-4 cm long in females and approximately 20 cm long in males. The urethral wall consists of three primary layers: a smooth muscle layer, a striated muscle layer, and a mucosal layer. The striated muscle layer is primary responsible for maintaining continence (DeLancey et al., 2002). Both sacral parasympathetic and lumbar sympathetic neurons innervate the smooth muscle layer (Andersson, 2002), which may play a role in both continence and micturition (DeLancey et al., 2002).

21 7 I.2.3 Urethral Sphincters In males, the urethral sphincter is the EUS, which is contiguous with the membranous urethra. In females the urethral sphincter consists of three parts: the EUS, which partially encircles the proximal urethra, and the more distal compressor urethrae and urethrovaginal sphincters. These somatic muscles are innervated by nerve fibers projecting from Onuf s nucleus in sacral spinal cord through the pudendal nerve (DeLancey et al., 2002). The periurethral sphincter arises where the urethra passes through the pelvic floor and is comprised of fibers from the levator ani and the pelvic floor (Roberts, 2008). The innervation of the periurthral sphincter arises either through the pudendal nerve or directly through offshoots of the sacral nerves (Junemann et al., 1987). I.2.4 LUT Innervation Autonomic and somatic neurons innervate the lower urinary tract through the pelvic, hypogastric, and pudendal nerves (Figure 1). Parasympathetic preganglionic neurons project from the sacral parasympathetic nucleus through the pelvic nerve and synapse on ganglion cells in the pelvic plexus and bladder wall (de Groat, 2006). These second order neurons activate the detrusor. The pelvic nerve also contains the parasympathetic afferents mediating sensations of bladder distention, urgency, sexual sensation and pain (Roberts, 2008). These afferents are both Aδ and C-fibers. Bladder voiding is dependent on Aδ activation in a normal animal, but following spinal cord injury a spinal C-fiber mediated voiding reflex develops and the Aδ reflex is lost (de Groat et al., 1998).

22 8 Suprasacral sympathetic fibers project through the hypogastric nerve, while sacral sympathetic fibers reach the bladder through the pelvic nerve. The sympathetic outflow innervates the smooth muscle of the bladder (de Groat, 2006), and proximal urethra (Craggs et al., 2006). In the cat, stimulation of the sympathetic pathways inhibits the bladder and excites the smooth muscle of the bladder neck and urethra. Sympathetic fibers also synapse in the pelvic ganglion, where they may exhibit bladder volume dependent inhibition and facilitation of parasympathetic ganglionic transmission (de Groat, 2006). Somatic neurons innervate the EUS and the superficial muscles of the pelvic floor through the pudendal nerve, and innervate the levator ani through offshoots of the S3-S5 sacral roots (Roberts, 2008). Somatic afferents in the proximal and distal urethra have recently been demonstrated to be involved in frequency dependent inhibition and facilitation of bladder contractions (Boggs et al., 2005; Boggs et al., 2006b; Bruns et al., 2008a; Bruns et al., 2008b; Tai et al., 2008; Tai et al., 2006; Woock et al., 2008; Yoo et al., 2008). These reflexes are mediated through spinal and supra-spinal reflex arcs. I.2.5 Central Reflex Pathways Ensembles of segmental reflexes are coordinated at the pontine level to maintain appropriate LUT function (Figure 2). Second order projection neurons from the sacral dorsal horn synapse on pontine structures including the periaqueductal grey (PAG) and the pontine micturition center (PMC). Descending tracts from the PMC maintain the reciprocal activation of the bladder and sphincter during micturition and continence. Pontine centers also receive largely inhibitory inputs from structures above the level of the inferior colliculus (Morrison et al., 2002).

23 9 Pelvic and pudendal nerve afferents synapse on local interneurons and second order projection neurons in lamina I and lamina V-VII of the sacral spinal cord. Both descending and segmental interneurons modulate reflexes through presynaptic inhibition of the afferents and direct inhibition of second order and primary efferent neurons (Morrison et al., 2002; Shefchyk, 2001). Both EUS motorneurons and parasympathetic preganglionic efferents receive extensive inputs from these areas (de Groat, 1993). Parasympathetic neurons in the intermediolateral sacral parasympathetic nucleus (SPN) extend axons though the ventral roots to synapse on post-ganglionic bladder efferents in the pelvic plexus. Motor neurons contained in Onuf s nucleus in the ventral horn of the sacral spinal cord innervate the EUS (Morrison et al., 2002). I.2.6 Bladder and EUS Reflexes During Voiding The central pathways described above support complex, reciprocal reflexes among all the pelvic organs (Figure 3). Barrington, whose work on cats in the early 20 th century provides the framework for our understanding of LUT reflexes, described spinal and supraspinal reflexes mediating contraction of the bladder and relaxation of the urethra (Barrington, 1931; Barrington, 1941). A more recent description depicts a highly convergent spino-bulbo-spinal reflex arc selectively inhibiting and facilitating sacral reflexes. This arrangement maintains specific links between afferent input and efferent activity, while preventing inappropriate activation of reflexes (Morrison, 2002). Micturition in a non-spinalized animal is initiated by bladder tension through a supraspinal bladder-tension, bladder-contraction reflex (Barrington s first reflex). The increase in bladder tension results in a reflexive decrease in urethral pressure. This urethral relaxation reflex has supraspinal components mediated by descending inputs

24 10 from the Pontine Micturition Center (Craggs et al., 2006; Shefchyk, 2006) as well as spinal components that persist following spinalization (Galeano et al., 1986). Fluid flow in the urethra activates spinal and supraspinal reflexes facilitating bladder contractions and promoting complete bladder voiding (Galeano et al., 1986; Robain et al., 2001). Spinalization eliminates selective supraspinal inhibition and facilitation. Reflexive bladder contractions are initially absent: a segmental bladder tension- bladder contraction c-fiber reflex develops over time (de Groat et al., 1981). Lacking supraspinal input, urethral reflexes intended to promote continence are not inhibited during voiding. Fluid flow in the urethra evokes significant contractions, mediated through afferents in the PN, which can prevent voiding (Galeano et al., 1986). I.3 Bladder Voiding Neuroprotheses Neuroprostheses for bladder voiding typically utilize one of two approaches: neurostimulation or neuromodulation. Neuromodulation approaches apply peripheral stimulation to activate desired reflex arcs and achieve continence or generate voiding. Stimulation of PN afferents has been shown to evoke increased bladder pressure and to achieve voiding (Boggs et al., 2006a; Tai et al., 2007a). Bladder voiding has also been achieved using PN block of direct EUS activation during PN afferent stimulation (Tai et al., 2007b). However, approaches relying on afferent stimulation require intact reflexes and may not be as reliable as direct activation of the bladder. Neurostimulation approaches directly activate bladder efferents to evoke an increase in bladder pressure. The sacral spinal roots are the traditional location for implanted electrodes used in bladder voiding neuroprostheses (Rijkhoff et al., 1997b). However, the spinal roots contain both small diameter parasympathetic efferents

25 11 innervating the bladder and larger diameter somatic efferents innervating the EUS. Coactivation of sphincter and bladder efferents results in high intravesicular pressures and limited voiding. Unfortunately, larger fibers have lower amplitude thresholds for activation than smaller fibers: activating bladder efferents without activating EUS efferents is challenging. Furthermore, preventing direct activation of the sphincter does not prevent reflex activation. Neuroprostheses using selective stimulation to activate the bladder still require a neurotomy to achieve voiding (Bhadra et al., 2006a; Grunewald et al., 1998). The Brindley neuroprostheses combines intermittent sacral root stimulation and a dorsal rhizotomy to increase bladder capacity and generate complete post-stimulus bladder voiding on demand (Brindley et al., 1986). Incontinence, autonomic dysreflexia, urinary tract infections, and reliance on anticholinergic medications are also reduced (Creasey et al., 2001). The reduction in medication, medical supply and medical care costs following Brindley stimulator implantation are sufficient to recoup implantation and maintenance costs within five years (Creasey and Dahlberg, 2001). However, the dorsal rhizotomy eliminates residual sacral sensation and reflex erections or defecation (Brindley, 1994), severely reducing patient acceptance of the stimulator. I. 4 Electrical Nerve Block Certain types of electrical stimulation can prevent nerve impulse conduction. Two basic types of electrical nerve block have been demonstrated: direct current block and alternating current block. Direct current block can be either depolarizing or hyperpolarizing (Bhadra and Kilgore, 2004). While effectively blocking nerve impulse conduction, both depolarizing and hyperpolarizing direct current nerve block will damage

26 12 the nerve (Whitwam and Kidd, 1975). Consequently DC nerve block is not a practicable tool for bladder voiding neuroprostheses. Alternating current nerve blocks include high frequency nerve block and collision nerve block. High frequency block utilizes stimulation frequencies in the kilohertz range to generate a local conduction block under the stimulating electrode, while collision block intercepts orthodromic nerve impulses with evoked antidromic nerve impulses. I.4.1 High Frequency (HF) Block HF block has been demonstrated in a number of animal models at frequencies above 1 khz (Bhadra and Kilgore, 2004; Bhadra and Kilgore, 2005; Kilgore and Bhadra, 2004). HF block of the feline PN has been shown in isolated (Tai et al., 2004; Tai et al., 2005c) and intact preparations (Bhadra et al., 2006a). These animal studies have shown that block amplitude threshold increases with stimulation frequency, a result also seen in modeling studies (Bhadra et al., 2007; Williamson and Andrews, 2005). Block amplitude threshold has also been shown to increase with decreasing nerve fiber diameter (Bhadra et al., 2007; Tai et al., 2005a; Tai et al., 2005b; Williamson and Andrews, 2005), a characteristic which can be exploited to generate normal recruitment of electrically stimulated muscles (Solomonow, 1984). HF electrical block has also been reported at frequencies below 1 khz. Solomonow reported that monopolar current-controlled stimulation at 600 Hz provided complete block of proximally evoked impulses in the feline sciatic nerve (Solomonow et al., 1983). Effectiveness of block increased with pulse width and amplitude and decreased with stimulation frequency above 1 khz. Solomonow suggested depletion of

27 13 endplate acetylcholine and nerve conduction block under the stimulating electrode as block mechanisms. HF urethral block for voiding has been investigated in a canine model using variations on the Solomonow stimulus waveform. Monopolar voltage-controlled stimulation (.1 1 khz) achieved ineffective block of proximally evoked urethral contractions at stimulation frequencies above 300 khz (Ishigooka et al., 1994). Fatigue was the assumed block mechanism and poor block was attributed the fatigue-resistance of the EUS. Voiding was achieved using biphasic current-controlled sacral root (Abdel- Gawad et al., 2001; Shaker et al., 1998). Acute HF stimulation evoked % of the pressure evoked by supramaximal 30 Hz stimulation. Chronic selective stimulation was less effective, evoking approximately 50 % of the pressure evoked by supramaximal 30 Hz stimulation. The waveforms based on Solomonow s work may achieve reductions in evoked pressures during HF stimulation through a fatigue mechanism, rather than through true nerve conduction block. Simulations suggest that HF block cannot be achieved below 2.2 khz (Bhadra et al., 2007). Other researchers distinguish high (> 4kHz) and a low (2 4 khz) frequency ranges for electrical block (Gaunt and Prochazka, 2008) and suggest that they may have different mechanisms.. Attempts to replicate these low frequency results were only successful for the biphasic waveform and suggest a neural conduction fatigue mechanism rather than a true block mechanism (Kilgore and Bhadra, 2004). I.4.2 HF Stimulation Response The response to HF stimulation can be described in terms of three phases (Bhadra and Kilgore, 2005) (Figure 4). The first phase constitutes an onset response, similar to a

28 14 muscle twitch but of potentially greater magnitude. This is often followed by a period of slowly decreasing muscle response. Eventually a steady state pressure is reached which may or may not represent conduction block. Steady state evoked pressure may be zero without conduction block being achieved. Conversely, HF stimulation may cause some level of recurrent nerve firing, leading to a non-zero steady state pressure, while also blocking proximally evoked action potentials. The magnitude of the first two phases has been shown to decrease with stimulation frequency (Bhadra and Kilgore, 2005; Gaunt and Prochazka, 2008; Tai et al., 2005c), though this has not been seen in all studies (Bhadra et al., 2006a; Kilgore and Bhadra, 2004). I.4.3 HF Stimulation Safety Shannon presented a formula relating stimulation current, pulse width and electrode area based on McCreery s work on safe cortical surface microelectrode stimulation (Shannon, 1992). This formula relates current I (ma), pulse width T (ms), and electrode area A (cm 2 ) to an empirically derived constant k that demarcated the boundary between safe and unsafe stimulation. log(i*t/a) = k log(i*t) Stimulation for which k was greater than 2.0 were considered unsafe. 1.5 was recommended as an appropriately conservative k value. For a high frequency block electrode with dimensions 1 mm x 3 mm and an impedance of 1000 ohms, this relation limits 2 khz stimulation to 8 Vpp and 6 khz stimulation to 20 Vpp. These stimulation parameters are within the range of values used to achieve block. This model may prove to be too conservative. HF stimulation pulse widths are 2 to 20 times shorter than the 400 µs pulse width (McCreery et al., 1990). Shorter pulse

29 15 widths appear to be less damaging than longer pulse widths (McCreery et al., 1992). This is potentially because shorter pulse widths limit the generation of harmful stimulation byproducts at the nerve electrode interface (Scheiner et al., 1990). However, higher stimulation frequencies have been shown more likely to damage nerves (Agnew et al., 1989; McCreery et al., 1995). Cochlear prostheses studies of stimulus waveforms at frequencies between 400 and 1 khz have found mixed results for stimulation intensities comparable to those used in HF stimulation (Tykocinski et al., 1995; Xu et al., 1997). This damage may be due to metabolic stress induced by electrical activation, as it can be blocked by local anesthetics (Agnew et al., 1990). The mechanism of HF block is disputed (Kilgore and Bhadra, 2004; Liu et al., 2009; Zhang et al., 2006a), but the depolarization of the neuron during block will result in changes in intracellular ion concentrations. These could eventually become harmful to the neuron. I.4.4 Collision Block and Quasitrapezoidal Stimulation Collision block has been used to prevent sphincter activation (Sweeney et al., 1989) and relies on quasitrapazoidal stimulation to generate unidirectional antidromic action potentials. Quasitrapezoidal stimulation uses anodic block to selectively inhibit large fibers. The anodes are typically the flanking contacts in a tripolar electrode configuration. By distributing current unequally between the anodes or placing one anode nearer the cathode than the other, action potentials can be allowed to escape unidirectionally during stimulation (van den Honert and Mortimer, 1981). Monopolar or bilpolar electrodes can also be used to generate unidirectional propagation of action potentials. Asymmetrically placed contacts create imbalanced virtual anodes in these

30 16 electrodes to allow unidirectional action potential propagation (Sweeney and Mortimer, 1986; Ungar et al., 1986). Quasitrapezoidal stimulation can be used for selective stimulation as well as collision block. Small axons are less susceptible to anodic block than larger axons. The quasitrapezoidal waveform can be tuned to excite both large and small axons at the cathodic center contact, while blocking action potential transmission in large axons at the anodic flanking contacts (Fang and Mortimer, 1991). This method has been used to selectively activate bladder fibers without activating sphincter fibers (Bhadra et al., 2002). Other research groups have successfully pursued selective bladder activation using anodic block, though they have used rectangular rather than quasitrapezoidal waveforms (Rijkhoff et al., 1994; Rijkhoff et al., 1998). Selective quasitrapezoidal stimulation of bladder fibers has been used for bladder voiding. However, selective quasitrapezoidal stimulation of the bladder does not prevent reflex activation of the EUS. Voiding using quasitrapezoidal stimulation has been largely unsuccessful without a dorsal rhizotomy (Bhadra et al., 2006b; Grunewald et al., 1998). Achieving effective collision block or selective quasitrapezoidal stimulation is much more complicated than achieving effective HF block. A window of effective stimulation amplitudes characterizes quasitrapezoidal block (or unidirectional stimulation). Below a minimum stimulation amplitude action potential transmission is not blocked at the flanking anodes. Above a maximum stimulation amplitude virtual cathodes capable of stimulating the nerve are generated. Increasing the difference between the maximum and minimum effective stimulation amplitudes will improve reliability, but

31 17 also increases the minimum effective amplitude, requiring greater current to achieve block (van den Honert and Mortimer, 1981). In contrast, a frequency-dependent block amplitude threshold characterizes HF block. Exceeding the block amplitude threshold only improves block by reducing onset (Bhadra et al., 2007; Kilgore and Bhadra, 2004). To ensure complete block a HF block neuroprosthesis can provide stimulation at amplitudes in significantly excess of block amplitude threshold. Neuroprotheses utilizing quasitrapezoidal stimulation risk formation of a virtual cathode at high stimulation amplitudes and would have to rely on other means of ensuring complete block. Selective quasitrapezoidal stimulation and collision block do possess some advantages over HF block. Unlike HF block, quasitrapezoidal stimulation does not possess an onset response. Current requirements for collision block or selective quasitrapezoidal stimulation are also much less than for HF block. Both techniques are useful and the appropriate choice depends on the application. I.6 Specific Aims Spinal cord injury patients often develop dyssynergic lower urinary tract reflexes, which can prevent bladder voiding and endanger their health. Permanent and destructive neurotomies allow bladder voiding, but eliminate residual sensation and disrupt sacral reflex arcs governing defecation and sexual function. These sacral reflexes and sensation are extremely important for patient quality of life (Anderson, 2004), therefore patients may reject existing bladder-voiding neuroprostheses fear of losing them. Nerve sparing approaches to bladder voiding that replace neurotomies with HF stimulation would be more acceptable to patients. Such neuroprotheses would improve

32 18 the patient quality of life and decrease patient care costs. This dissertation examines HF stimulation as a replacement for destructive dorsal and PN neurotomies in an acute feline model. We will first demonstrate that PN EUS HF block allows complete bladder voiding (Aim 1). Next we will demonstrate HF block of mixed sacral nerve roots within the spinal canal (Aim 2). Finally, we will pursue voiding using HF block at the sacral root level (Aim 3). I.6.1 Demonstrate Bladder Voiding Using HF PN Block and Sacral Bladder Drive Previous researchers have shown that HF stimulation can block nerve impulse transmission in the pudendal nerve, leading to a reduction in evoked external urethral sphincter pressure (Bhadra et al., 2006a; Tai et al., 2004; Tai et al., 2005c). Aim 1 extends that research, generating bladder voiding by using PN HF block to prevent EUS activation during sacral stimulation. Our hypotheses are 1) PN HF block will allow bladder voiding during sacral root stimulation and 2) bladder voiding during combined sacral stimulation and PN HF block will be comparable to voiding in the absence of evoked external urethral sphincter tone. Our first objective is to measure voided volumes and bladder pressures over time in animals with intermittent and continuous sacral stimulation, with and without PN HF block (Figure 5). The second objective is to measure voiding following bilateral pudendal nerve neurotomy, which will denervate the external urethral sphincter. Successful completion of the first objective will allow comparisons between sacral stimulation-driven bladder voiding with and without HF PN block. Voiding following bilateral PN neurotomy will provide a benchmark for unobstructed voiding in the experimental preparation. This will allow comparisons between HF PN block voiding

33 19 and voiding unobstructed by EUS closure. As unobstructed voiding is typical of an intact individual, this comparison allows quantification of the potential clinical gain. I.6.2 Characterize Sacral Small Fiber HF Transmission Block Aim 2 demonstrates HF block within the spinal canal and HF block of small autonomic neurons. Modeling studies suggest that HF block thresholds and frequencies depend on nerve fiber diameter (Bhadra et al., 2007; Tai et al., 2005a; Zhang et al., 2006b). Previous animal work has focused on peripheral block of somatic nerves containing fibers with diameters in excess of 5 µm. Rapid, reversible, non-destructive HF conduction block of small diameter neurons (1 3 µm) would enable neuroprostheses for chronic pain and autonomic disorders. Spinal root HF block would allow bladder-voiding neuroprostheses contained entirely within the spinal canal. Our hypotheses are 1) HF stimulation will rapidly and reversibly reduce proximally evoked bladder pressures and 2) transmission block and HF-evoked bladder response characteristics will depend on HF stimulation parameters and 3) conduction block will be localized under the HF stimulation electrode. The response to proximal bladder activation and distal HF stimulation will be measured to determine effective HF block parameters and characterize transmission block. HF stimulation-evoked bladder pressures will be recorded in the absence of proximal bladder activation to determine the stimulation parameters minimizing HF onset response. Stimulation electrodes will be implanted proximal and distal to the HF blocking electrode and local conduction block will be demonstrated by continued bladder excitability during HF sacral bladder efferent block (Figure 6A).

34 20 Sacral HF parasympathetic efferent block extends HF block techniques to the sacral root level. Effective HF block parameters identified in this Specific Aim will be used in Specific Aim 3 to block EUS activation at the sacral root level and allow voiding. Sacral HF block of somatic pelvic floor and EUS efferents allows bladder-voiding neuroprostheses contained entirely within the spinal canal. Such neuroprostheses might be able to use existing human-approved sacral electrodes to speed clinical translation. Sacral HF block may also provide an alternative to PN block in patients for whom PN block fails to allow voiding. HF parasympathetic efferent block (small fiber block) extends the set of disorders HF stimulation could potentially treat and supports research into HF block for chronic pain and autonomic disorders I.6.3 Demonstrate Bladder Voiding Using Sacral HF Block and Bladder Drive The sacral spinal roots are the traditional intervention location for bladder voiding neuroprostheses. Existing clinical devices make use of this location (Brindley et al., 1986). The location within the spinal cord offers space, stability and protection for implanted electrodes (Rijkhoff et al., 1997b). Patients with existing implanted sacral electrodes may serve as research subjects, speeding clinical translation of HF block-based bladder voiding neuroprostheses. We hypothesize that 1) sacral HF stimulation allows voiding comparable to voiding unobstructed by EUS activation and 2) sacral HF stimulation allows clinically complete bladder voiding. Excitatory stimulation electrodes will be implanted on the S1 and S2 sacral nerve roots. A HF block electrode will be implanted on the S1 nerve root distal to the excitatory electrode (Figure 6B). HF stimulation will be provided at the parameters most effective at generating rapid, complete block in Specific Aim 2.

35 21 Measurement of voided volumes, residual volumes and bladder pressures in animals with or without Sacral HF block will be conducted to determine improvements in voiding efficiency during HF stimulation. Successful completion of this aim demonstrates a second nerve sparing, bladder voiding neuroprostheses. This study will assist planning of chronic animal safety studies intended to provide data on long-term HF safety as an initial step towards human trials. I.7 Significance Successful completion of this project will demonstrate two approaches for nervesparing bladder voiding neuroprostheses. Both preserve sacral reflexes and sensation, significantly improving existing approaches. The first approach uses HF PN block combined with sacral root stimulation to achieve bladder voiding. This represents a straightforward improvement over previous methods that relied on PN neurotomies to improve voiding. Like the PN neurotomy, HF PN block prevents activation of the EUS, allowing voiding during sacral stimulation. The second approach blocks EUS efferents at the sacral root level, a modification of the successful Brindley bladder voiding system. The Brindley device relies on a dorsal neurotomy to interrupt dyssynergic continence reflexes. In contrast, a sacral HF EUS block bladder voiding neuroprostheses would utilitize intra or extradural spinal root electrodes to reversibly block pelvic floor efferents. Both options should be pursued to improve the likelihood of developing an effective therapy. Sacral EUS block provides advantages in clinical translation and may prove more successful in patients with significant non-pn mediated urethral closure mechanisms. PN block is simpler and more tolerant of imperfect block. Should both

36 22 approaches to generating bladder voiding prove successful, patients would benefit from multiple treatment options. Secondary benefits of this research include increased understanding of HF block phenomena. The effect of HF stimulation on mixed nerve such a spinal sacral root is unknown: HF block studies have not examined the effect of HF stimulation on small myelinated axons. Experimental data could help refine HF block models of block thresholds and onset responses for neurons of differing diameters. HF block applications for chronic pain or autonomic disorders may be identified, potentially expanding the patient population benefiting from HF block neuroprostheses.

37 Figure 1: The principle innervation of the lower urinary tract is provided by the pelvic, hypogastric, and pudendal nerves. The pelvic nerve contains primarily parasympathetic preganglionic efferents mediating bladder contraction and afferents mediating sensations of bladder distention, urgency, sexual sensation and pain. The hypogastric nerve contains suprasacral sympathetic efferents innervating the detrusor and proximal urethra and modulating parasympathetic transmission in the pelvic ganglion. Urethral and perineal afferents are contained in the pudendal nerve, as well as somatic neurons innervating the EUS and the superficial muscles of the pelvic floor. (Figure 1, Morrison et al., 2002). 23

38 Figure 2: Segmental reflexes are coordinated at the pontine level to maintain appropriate LUT function. Pelvic and pudendal nerve afferents synapse on local interneurons and second order projection neurons in the sacral spinal cord. Both descending and segmental interneurons modulate reflexes during storage and micturition. Storage: Descending tracts from the Pontine Micturition Center (PMC) facilitate sacral reflexes maintaining EUS pressure. Sympathetic tracts in the hypogastric nerve inhibit bladder contractions. Micturition: Second order projection neurons from the sacral dorsal horn synapse on pontine structures including the periaqueductal grey (PAG) and PMC. Descending tracts from the PMC inhibit sphincter activation and facilitate bladder contraction (Figure 3, Morrison et al., 2002). 24

39 Figure 3: Bladder and sphincter reflexes operate in concert in an intact animal. Adult: Bladder filling is accompanied by an increase in external urethral sphincter pressure. Micturition is initiated by bladder tension through a supraspinal bladder-tension, bladdercontraction reflex. The increase in bladder tension results in a reflexive decrease in urethral pressure mediated by spinal and spino-bulbo-spinal reflex arcs. Fluid flow in the urethra activates spinal and supraspinal reflexes facilitating bladder contractions and promoting complete bladder voiding. Paraplegic: Spinalization eliminates selective supraspinal inhibition and facilitation. Lacking supraspinal input, urethral reflexes intended to promote continence are not inhibited during voiding. Fluid flow in the urethra evokes significant contractions, mediated through afferents in the PN, which can prevent voiding (Figure 4, Morrison et al., 2002). 25

40 Figure 4: High Frequency Stimulation evokes an onset response. The first phase of the response is similar to a muscle twitch but of potentially greater magnitude and is often followed by a period of slowly decreasing evoked pressure. Eventually a steady state pressure is reached which may or may not represent conduction block. Steady state evoked pressure may be zero without conduction block being achieved. Conversely, HF stimulation may cause some level of recurrent nerve firing, leading to a non-zero steady state pressure, while also blocking proximally evoked action potentials. (Figure 4, Bhadra and Kilgore, 2005) 26

41 Figure 5: Electrode placement for voiding using pudendal nerve block and sacral bladder drive (Specific Aim 1). Bilateral proximal continuous or intermittent stimulation evokes bladder and sphincter pressure (electrode 1). Maximal EUS pressure, mimicking worst-case dyssynergia, was evoked by continuous bilateral pudendal nerve stimulation (electrode 2). Bilateral HF stimulation (electrode 3) blocked pudendal nerve action potential conduction, effectively denervating the EUS and allowing bladder voiding. HF stimulation electrodes in black, excitatory stimulation electrodes in grey. 27

42 Figure 6: Electrode placement for small fiber block (Specific Aim 2) and sacral root stimulation for bladder voiding (Specific Aim 3). A: Proximal sacral stimulation (electrode 1) evokes bladder pressure. HF stimulation (electrode 2) blocks nerve impulse conduction, allowing bladder relaxation. Distal stimulation (electrode 3) during HF stimulation evokes bladder pressure, demonstrating a conduction block mechanism for block. B: Proximal sacral stimulation (electrode 1) evokes sphincter pressure. HF stimulation (electrode 2) blocks nerve impulse conduction, allowing the sphincter to relax. Intermittent bilateral S2 stimulation (electrode 3, shown unilaterally for clarity) evokes bladder and sphincter pressure and generates post stimulus voiding. HF stimulation electrodes in black, excitatory stimulation electrodes in grey. 28

43 29 CHAPTER II: BLADDER VOIDING BY COMBINED HIGH FREQUENCY ELECTRICAL PUDENDAL NERVE BLOCK AND SACRAL ROOT STIMULATION

44 30 II.1 Abstract Aims: Uncoordinated contraction of the external urethral sphincter is prevalent in individuals with spinal cord injury and can prevent bladder voiding. The aim of this study was to demonstrate that complete and reversible sinusoidal high frequency alternating current (HFAC) conduction block of the pudendal nerves (PN) can eliminate external urethral sphincter activation and produce low residual bladder voiding. Methods: In 4 cats, tripolar nerve cuff electrodes were implanted bilaterally on both pudendal nerves and on both extradural S2 roots. Bladder and urethral pressures, bladder volumes and flow were recorded. Bilateral HFAC was applied to determine voltage and frequency parameters resulting in bilateral PN conduction block. Sacral root stimulation provided bladder activation. Randomized sets of voiding trials were conducted with and without HFAC PN block. Additional voiding trials were conducted following bilateral PN neurotomy to eliminate somatic sphincter resistance and provide an estimate of voiding with complete block. Results: Effective bilateral PN block and voiding was obtained in three of four animals. Application of bilateral PN HFAC stimulation improved voiding from 2% ± 4% to 77% ± 18% of the initial bladder volume and significantly (p < 0.001) reduced maximum bladder pressure during voiding. Voiding in trials with PN block was not significantly different from voiding following PN neurotomy (82% ± 19%, p = 0.51). Conclusions: These results demonstrate that bilateral HFAC block of the PN can produce effective voiding. Neural prostheses using this approach may provide an alternative method for producing micturition for people with spinal cord injury.

45 31 II.2 Introduction The pudendal nerves (PN) serve as the primary conduit for somatic efferent and afferent nerve fibers between the urethra and the sacral spinal cord (Akita et al., 2003). The efferents innervate the external urethral sphincter (EUS), which acts as an outlet valve of the lower urinary tract. The afferents initiate sacral micturition reflexes modulated by supraspinal inputs leading to synchronized detrusor contraction and EUS relaxation (Blok and Holstege, 1998; de Groat et al., 2001). Following spinal cord injury (SCI) these sacral reflexes can become disordered. The EUS and bladder may contract simultaneously, a condition termed detrusor-sphincter dyssynergia (DSD). This condition can result in significant medical complications such as urinary tract infections, autonomic dysreflexia, vesico-ureteric reflux and hydronephrosis. The conventional first-line treatment for DSD is anti-muscarinic drugs combined with clean intermittent self catheterization (CISC). However, long term patient compliance with this regime is extremely poor. Certain patient populations, such as quadriplegics, are particularly likely to choose other methods of bladder management (Yavuzer et al., 2000). Alternatively, a sphincterotomy can allow low pressure, low residual bladder voiding, replacing CISC. However, potential side effects are significant and the procedure must sometimes be repeated to enable effective voiding (Ahmed et al., 2006). Neural prostheses can provide clinically effective bladder voiding using intermittent sacral anterior root stimulation in conjunction with a dorsal rhizotomy (Brindley, 1994; Creasey et al., 2001). The dorsal rhizotomy, primarily intended to reduce detrusor hyperreflexia and improve continence, also improves voiding by

46 32 eliminating DSD (Van Kerrebroeck et al., 1996) and reduces the risk of upper urinary tract damage (Brindley, 1994). Unfortunately, a dorsal rhizotomy eliminates any residual sensation, reflex erection, and reflex defecation, reducing the patient population potentially benefiting from available sacral-root stimulation neuroprostheses. Selective activation of the bladder without direct urethral sphincter activation has also been investigated using a variety of stimulus waveforms (Bhadra et al., 2002; Brindley and Craggs, 1980; Grunewald et al., 1998; Rijkhoff et al., 1997a). However, selective bladder activation does not address reflexive EUS activation; a dorsal rhizotomy to abolish DSD may still be required to allow voiding (Bhadra et al., 2006b; Kirkham et al., 2002). Reducing EUS competence can allow voiding in the absence of a dorsal rhizotomy. Denervation of the EUS by surgical PN neurotomy has been shown to achieve low pressure voiding with low post-void residual (PVR) (Engel and Schirmer, 1974). EUS paralysis using injections of botulinum toxin can improve voiding, though the effect is temporary (de Seze et al., 2002; Rackley and Abdelmalak, 2004). EUS fatigue using moderate frequency PN stimulation (< 500 Hz) in combination with sacral stimulation has successfully generated voiding in a canine model (Barada et al., 1995; Li et al., 1992; Li et al., 1995). High frequency alternating current (HFAC) stimulation can prevent nerve impulse transmission, temporarily denervating a targeted muscle. HFAC nerve block is immediate and reversible; nerve impulse transmission is prevented within milliseconds of stimulus application and resumes within seconds upon stimulus cessation (Bhadra and

47 33 Kilgore, 2004). This non-destructive approach should not affect EUS function or risk incontinence. Unilateral HFAC block has been demonstrated in a number of animal models (Bhadra and Kilgore, 2005; Kilgore and Bhadra, 2004). HFAC block of the feline PN has been shown in isolated (Tai et al., 2004; Tai et al., 2005c) and intact preparations (Bhadra et al., 2006a). However, functional voiding using bilateral HFAC PN block and direct sacral stimulation has not yet been demonstrated. The purpose of this study was to produce effective low-pressure, low-pvr bladder voiding using combined bilateral HFAC PN block and sacral root stimulation. If effective, neural prostheses using this approach could provide an alternative method for restoring bladder function for people with SCI. II.3 Methods and Materials II.3.1 Experimental Preparation This study was conducted on four male cats with a body weight of 3.8 ± 0.4 kg. All animal care and experimental procedures were approved by the institutional animal care and use committee. Data were collected under intravenous α-chloralose anesthesia (Sigma, St. Louis, MO) 65 mg kg -1. An anesthetic monitoring system (SurgiVet V9200 Advisor Monitor, Smiths Medical PM Inc, WI) was used to observe body temperature, ECG, blood oxygen saturation, blood pressure, and expired pco 2. The appropriate depth of anesthesia was maintained by monitoring blood pressure, heart rate, and withdrawal and blink reflexes. Respiration was maintained by a pressure-regulated respirator (ADS 1000, Engler Engineering Corporation, FL) and expired pco 2 measurements. Temperature was maintained between 37 and 39 C with a heating blanket. Saline 0.9%

48 34 with 8.4 mg ml -1 sodium bicarbonate and 5% dextrose was administered at 5-20 ml kg hr -1 IV. The bladder was drained and the bladder volume recorded. The bladder was then exposed via a suprapubic midline incision. A 6F dual-lumen catheter (DLC 6D, Life Tech, TX) was placed in the dome of the bladder and secured by a sero-muscular purse string suture. In two animals the ureters were ligated close to the trigone, transected proximally and drained externally. The sacral roots were accessed extradurally by a lumbosacral laminectomy of L7 and S1. Two extradural sacral roots (Right and Left) evoking the largest bladder contractions when stimulated with a custom-made bipolar hook electrode were implanted with tripolar nerve cuff electrodes. Post mortem dissection on three animals verified that the cuffs were implanted on the S2 nerve roots. The pudendal nerves were exposed through postero-lateral gluteal approaches. Two tripolar nerve cuff electrodes were implanted bilaterally on the PN trunk proximal to the origin of the deep perineal nerve. A foam stand supported the chest and pelvis without compressing the abdomen. The location of the external urethral sphincter was determined by conducting a urethral pressure profile while providing supramaximal 1 Hz sacral root stimulation. II.3.2 Experimental Set-up Bladder pressure was measured using an external pressure transducer (Deltran IV, Utah Medical Devices, UT) connected to one lumen of the 6F dual lumen suprapubic bladder catheter. Urethral pressure was measured using either a 3.5F perfusion catheter, an external pressure transducer, and an infusion pump (0.05 ml min -1, Genie YA-12,

49 35 Kent Scientific, CT) or a 3F catheter mounted microtransducer (CT3F, MMI-Gaeltech, Hackensack, NJ). Experimental data was recorded on a computer (Dell Inspiron 8600) equipped with a National Instruments data acquisition card (NI DAQ 6024E, National Instruments Corporation, TX, sampling rate 100 Hz) using a customized interface (Labview 7.1, National Instruments Corporation, TX) and also a strip chart recorder (TA-11, Gould Inc. Valley View, OH). Voided volume, as a function of time, was recorded in two animals (23 trials total) using a high-resolution single point load cell (s215, 0.9kg; Strain Measurement Devices, CT or ESP 0.6 kg, Load Cell Central, PA). Transduced physiological signals were input to the DAQ card and the chart recorder through a NI terminal block (NI SC-2345, National Instruments Corporation, TX) containing two dual-channel strain gauge signal conditioning modules and three configurable feed-through modules (NI SCC-SG24 and SCC-FT01, National Instruments Corporation, TX). A pulse train generator (Pulsar 6bp or 6bp-as, FHC, Bowdoinham, ME) provided current-controlled balanced biphasic stimulation to the sacral electrodes. A dual output universal waveform generator (Wavetek Model 195, Fluke, Everett, WA) generated the distal PN block HFAC waveform. The output of the function generator could be further amplified with an audio amplifier (KA-75, Kenwood, CA). The proximal PN stimulus, a current-controlled balanced biphasic pulse train, was generated by either a Pulsar 6bp, DS8000 Digital Stimulator and DLS100 Isolator (World Precision Instruments Inc, FL), or an isolated linear transconductance amplifier (BSI-1, BAK electronics Inc, MD).

50 36 Series capacitors in line with the outputs reduced DC current leakage from the stimulators. II.3.3 Experimental Procedure The current thresholds for stimulation-evoked EUS responses, the maximal evoked EUS pressures, and the current level required to evoke maximal EUS pressures were determined for each tripolar nerve cuff electrode. Bilateral block parameters were then determined and voiding trials conducted. The block parameters and evoked responses were periodically verified and adjusted in animals in which multiple sets of voiding trials were conducted. HFAC block parameters were determined using a method similar to that previously reported (Bhadra et al., 2006a; Bhadra and Kilgore, 2005). During HFAC block trials EUS twitches were evoked bilaterally using supramaximal current-controlled stimuli (1 ma, balanced biphasic, 100 µsec PW at 1 Hz) applied to the implanted proximal PN nerve cuff electrodes. Five to ten seconds after the initiation of proximal stimulus, bilateral HFAC stimulation (1-30 khz, 1-40 Vpp) was applied to the distal PN tripolar cuffs. HFAC stimulation continued until the EUS pressure decreased to a relatively constant residual level dependent on the choice of stimulation parameters. HFAC stimulation then ceased while proximal stimulation continued, allowing EUS muscles twitches to reappear. HFAC PN block was defined as the absence of observed EUS twitches and a residual EUS pressure during HFAC stimulation less than the sphincter pressure evoked by proximal stimulation. In the animals for which block was obtained, voiding trials

51 37 were conducted following pudendal nerve neurotomy to assess the equivalence of HFAC block and PN neurotomy. In all voiding runs, the bladder was filled to a test volume limited by the volume threshold for distension evoked contractions. The bladder was filled through the suprapubic catheter using an infusion pump at a rate of 1-2 ml min -1. The bladder was drained only after trials with significant voiding. PVR was estimated in trials with negligible voiding using infused volumes, voided volumes, and the PVR when next measured. In one Set of Trials, proximal PN stimulation ceased simultaneously with bladder stimulation. In this case the volume voided was defined as the volume voided during sacral bladder stimulation and the PVR was defined as the remaining bladder volume at the end of bladder stimulation. II.3.4 Electrical Stimulation Every voiding trial utilized current-controlled sacral root stimulation to evoke bladder pressure. Sacral stimulation in pre-neurotomy trials was either continuous (balanced biphasic, rectangular, 100 µs pulse width at 20 Hz, 1 ma amplitude) of 20 s duration or intermittent (2 s on / 2 s off, 20 Hz, rectangular biphasic, 100 µs pulse width) of 28 s duration (7 bursts of two-second pulse trains, 1 ma amplitude). Sacral stimulation in one animal was 6 ma for post neurotomy trials and one of three randomized Sets of Trials. All post-neurotomy trials used continuous stimulation. The total duration of continuous stimulus was controlled automatically, while the total duration of intermittent stimulus was controlled manually. To consistently simulate the effects of DSD we applied bilateral tetanic stimulation to the proximal PN electrode (rectangular, 100 µs pulse width at 20 Hz, 1 ma

52 38 amplitude) in all trials. Proximal PN stimulation lasted 10 s longer than sacral bladder stimulation, except in the first Set of Trials in the first animal when coincident proximal PN and sacral bladder stimulation were applied. Voltage-controlled sinusoidal HFAC, when applied, began 10 s prior to the onset of sacral and proximal PN stimulation and continued 10 s after cessation of sacral stimulation. II.3.5 Data Analysis Data recorded through the Labview interface was processed using MATLAB (Mathworks Inc., Natick, MA) and Minitab (Minitab Inc, State College, PA). For each voiding trial the maximum bladder pressure (P ves max ) and the total volume voided were measured. The percentage of the bladder volume voided (Void %) was calculated as the volume voided divided by the sum of the volume voided and the PVR. An analysis of variance was performed on the P ves max and Void % data from the randomized complete block design (RCBD). Three missing data points were interpolated using the mean of the remaining observations for that combination of factors. One duplicate observation was selected at random and discarded. The factors included in the model were Set of Trials and Bladder Stimulation Type (continuous or intermittent) and the absence or presence of HFAC PN Block. Initially, first order interactions between the control variables were included in the model. The analysis was then repeated excluding terms which were not significant at the α = 0.1 level. The sum of squares and degrees of freedom for these terms contributed to the estimation of the error. Void % and P ves max were compared between pre-neurotomy HFAC PN block trials and post-neurotomy trials by Animal. These comparisons included two pre-

53 39 neurotomy HFAC PN block trials conducted outside the RCBD. Pre-neurotomy HFAC PN block trials with both intermittent and continuous stimulation were included in the Void % comparison. Only pre-neurotomy HFAC PN block trials with continuous sacral stimulation were included in the P ves max comparison. Animal was used in place of Set of Trials as the bilateral PN neurotomy can only be performed once per animal. II.4 Results Significant voiding was obtained in three of four animals. In the 3 animals with significant voiding, 85 randomized trials (in seven Sets of Trials ), 4 non-randomized trials, and 10 trials following bilateral PN neurotomy were conducted. Three randomized trials were not recorded in error, while one trial was a duplicate. The overall distribution of trials by animal and stimulation type is given in Table I (A) and the bladder volumes at which the trials were conducted are shown in Table I (B). Representative bladder pressure and flow recordings are shown in Figure 1. Minimal voiding was achieved in one animal (0 ± 2% range 0% to 5%). In this animal sacral root stimulation during voiding trials produced bladder contractions smaller than distention evoked contractions and of minimal amplitude (45 ± 2 cm H 2 O). In addition, complete block with minimal residual EUS tone was not achieved. This animal was excluded from subsequent analysis of voiding and bladder pressures. The stimulation threshold for the tripolar cuffs was 336 ± 138 µa (range 150 µa to 600 µa). The current amplitude for maximal evoked urethral pressure was 935 ± 974 µa, (300 µa to 4000 µa). The evoked urethral pressure varied widely, 46 ± 19 cm H 2 O (Range 8 85 cm H 2 O). Effective bilateral HFAC PN block was achieved at 2 khz in

54 40 every voiding animal. Block voltages were between 2 and 25 V (8 ± 7). HFAC-evoked residual EUS pressures were low, less than 15 cm H 2 O. HFAC PN block significantly improved Void % compared to trials with no PN conduction block (p< 0.001, 77% ± 18% with HFAC block, 2% ± 4% without HFAC block or neurotomy) (Figure 2). Overall, Void % in trials following PN neurotomy (82% ± 19%) was similar to Void % in trials with HFAC block (p = 0.51). In each animal, Void % in trials following PN neurotomy was similar to Void % in trials with HFAC block (p > 0.05 for each animal). HFAC PN block trials and post PN neurotomy trials had low PVRs (3.9 ± 4.1 ml and 2.6 ± 3.5 ml respectively) while runs with no PN conduction block had high PVRs (14.9 ± 4.9 ml). The control variables with a significant effect on Void % were Set of Trials (p < 0.001), HFAC PN Block (p < 0.001), and the interaction of Set of Trials and HFAC PN Block (p < 0.001). Only Void % in trials with HFAC block differed between Set of Trials: Void % in trials without HFAC block varied little. Sacral stimulation type had no effect on Void % (p = 0.95). The analysis of variance indicated our model was highly predictive of the volume voided, accounting for nearly all of the variance in the results (adjusted R 2 of 94%). HFAC PN block significantly reduced P ves max as compared to trials with no PN conduction block (p< 0.001, 76 ± 14 cm H 2 O with HFAC block, 116 ± 26 cm H 2 O without HFAC block or neurotomy). In two animals the maximum bladder pressures following PN neurotomy were significantly lower than during HFAC block (p < for each animal). In the remaining animal the null hypothesis of equivalent means could not be rejected (p= 0.23) (Figure 3).

55 41 The control variables with a significant effect on P ves max included Set of Trials (p < 0.001), HFAC PN Block (p < 0.001), and the interaction of Set of Trials and HFAC PN Block (p < 0.001). Sacral stimulation type did not have a significant direct effect (p = 0.30). However, the interaction of sacral stimulation type with PN Conduction Block was significant (p< 0.01). As was true for the Void %, the analysis of variance showed that our model captured the nearly all of the variance in the results (adjusted R 2 of 93%). II.5 Discussion This study demonstrated effective bilateral HFAC block of the pudendal nerves. Unilateral HFAC PN block has previously been reported (Bhadra et al., 2006a; Tai et al., 2004; Tai et al., 2005c), but could be insufficient for bladder voiding neuroprostheses. Reflex activation of the EUS through one PN could be sufficient to prevent effective micturition. Achieving effective bilateral block at a single frequency with a single voltage-controlled stimulator enhances our impression of HFAC block as a useful and flexible tool. We have demonstrated that bilateral PN HFAC block combined with direct bladder drive can generate effective voiding comparable to that following bilateral PN neurotomy. Residual bladder volume percentages were similar to those in preparations utilizing sphincter fatigue mechanisms (Li et al., 1992; Li et al., 1995) or PN neurotomies (Bosch et al., 1992). The effect of HFAC PN block on voiding performance was repeatable. The coefficients of variation for Void% and P ves max were significantly lower within each Set of Trials than observed in the pooled data shown in Figures 2 and 3. Tai, et al. have also recently demonstrated voiding using afferent PN stimulation to activate the bladder, while applying distal HFAC stimulation to block the EUS (Tai et al., 2007c).

56 42 Successful HFAC block parameters were consistent within experiments and between animals. All voiding animals used 2 khz HFAC stimulation and in two of three voiding animals the stimulation amplitude remained constant throughout the experiment. Though the chronic stability of HFAC block parameters must be addressed prior to clinical implementation, this study suggests consistency of HFAC block parameters can be relied upon in an acute setting. The percentage of fluid evacuated (Void %) was independent of the sacral stimulation type. Continuous sacral stimulation was as effective as intermittent sacral stimulation in producing voiding. This is in contrast to the Brindley approach, which requires intermittent stimulation to circumvent evoked EUS pressures and generate voiding. Potential remains for improved voiding with better block. Bladder pressures were significantly lower in trials following bilateral PN neurotomy than in trials with HFAC PN block in two of the three voiding animals. Residual evoked urethral pressures were recorded when testing block. Better block could further reduce urethral resistance, lowering the maximum bladder pressure during voiding and potentially improving Void %. Proximal PN stimulation was applied to provide consistent sphincter tone during sacral root stimulus and model the clinical effect of DSD. Sacral root stimulation on roots chosen to evoke maximum bladder pressure generated variable EUS pressures that did not always prevent voiding. Reflex EUS activation during voiding was generally not substantial in this preparation. This approach reduced variability between animals and provided a worst-case scenario for EUS resistance. Producing voiding equivalent to

57 43 bilateral PN neurotomy with HFAC PN block in this worst-case model suggests that the approach is robust and may be clinically effective. This study suggests that neural prostheses for restoration of bladder function using bilateral HFAC PN block may provide a viable alternative to those requiring dorsal rhizotomy. HFAC block of the pudendal nerves, which are the final common pathway to the EUS, addresses both the issues of direct and reflex sphincter resistance during sacral root stimulation. II.6 Conclusions Reversible bilateral HFAC PN block can be achieved in an intact acute preparation with a single bilaterally-applied voltage-controlled waveform. In combination with sacral root bladder stimulation, bilateral HFAC PN block can generate clinically effective voiding equivalent to voiding following PN neurotomy. This approach does not require a dorsal rhizotomy, and may provide an alternative method for restoring bladder function for people with SCI. HFAC PN block could potentially expand the population of individuals benefiting from neuroprostheses to restore bladder function. Acknowledgements This work was supported by The State of Ohio BRTT ONNP 03-10, NIH AR07505, EB004314, DK077089, HD40298, and Department of Veterans Affairs RR&D B3675R. The authors wish to thank Timothy Bruns, Tina Goetz, and Timothy Mariano for their assistance.

58 Figure 1: Bladder pressure and flow recordings from representative trials. A and B: Extradural sacral root stimulation without pudendal nerve (PN) block. C and D: Extradural sacral root stimulation with bilateral High Frequency Alternating Current (HFAC) PN block. HFAC PN block allows voiding to occur and reduces maximum bladder pressure during voiding. A and C: Intermittent sacral root stimulation (20 Hz, 2 sec on / 2 sec off). B and D: Continuous sacral root stimulation (20 Hz). The solid trace shows the bladder pressure. The interrupted trace is the flow rate. Neither the percentage of bladder volume voided nor the maximum bladder pressure depended on sacral stimulation type. 44

59 45 Figure 2: Voided percentage of initial bladder volume versus type of pudendal nerve (PN) treatment (None = No conduction block, HFAC = High Frequency Alternating Current conduction block or Neurotomy = Bilateral surgical section of pudendal nerves). Results are shown pooled and by individual animals. Box and whiskers plots show min, max, median, 25 th and 75 th percentiles. Individual trials are shown as points. In each animal, voiding was significantly greater in trials with HFAC PN block or with neurotomy than in trials with no PN conduction block. Voiding was not significantly different in trials following PN neurotomy and trials with HFAC PN block in all animals (p> 0.05 all animals).

60 Figure 3: Maximum bladder pressure during voiding versus type of pudendal nerve (PN) treatment (None = No conduction block, HFAC = High Frequency Alternating Current conduction block or Neurotomy = Bilateral surgical section of pudendal nerves). Results are shown pooled and by individual animals. Box and whiskers plots show min, max, median, 25 th and 75 th percentiles. Individual trials are shown as points. In each animal, the maximum bladder pressure during voiding was significantly lower in trials with HFAC PN block than in trials without PN conduction block. 46

61 Table I A: Distribution of trials among the control variables. Total number of trials is given for each combination of animal and type of PN treatment (None = No PN Conduction Block, HFAC = High Frequency Alternating Current conduction block or Neurotomy = Bilateral surgical section of pudendal nerves). The number in parenthesis is the number of trials using intermittent sacral stimulation. The remaining trials used continuous sacral stimulation. B: Initial Bladder Volumes (ml, Mean ± SD) during each randomized Set of Trials. Animal 4 was not included in the bladder pressure and voiding analysis as evoked bladder pressures and HFAC block were poor. 47

62 48 CHAPTER III: HIGH FREQUENCY ELECTRICAL CONDUCTION BLOCK OF SMALL MYELINATED BLADDER EFFERENTS

63 49 III.1 Abstract High frequency (HF) nerve transmission block, which shows rapid induction and reversibility, could potentially treat many pathologies arising from inappropriate or excessive neural activity. HF nerve block has been characterized for large somatic motor neurons, but not for small (< 5µm diameter) myelinated neurons. This study demonstrates HF conduction block of small-diameter, parasympathetic bladder efferents and evaluates bladder pressures evoked by sacral root HF stimulation. Transmission block trials evaluated the dependence of HF block completeness, rapidity, and reversibility on the frequency and amplitude of HF stimulation. HF evoked bladder pressure trials characterized the bladder response to HF stimulation and the effect of HF stimulation on the subsequent excitability of the bladder. Complete, rapid nerve block and the minimization of HF onset responses could be achieved with the appropriate choice of stimulation parameters. Block validation trials demonstrated that conduction block was localized to a region near the HF stimulation electrode and that the bladder remained excitable during HF stimulation. HF small fiber conduction block may allow novel neuroprostheses for chronic pain, and sacral root block may prove useful for bladder voiding neuroprostheses.

64 50 III.2 Introduction Excessive or aberrant neural activity underlies diverse pathologies including chronic pain (D'Mello and Dickenson, 2008; Dray, 2008) and bladder sphincter dyssynergia (Craggs, 2006; Shefchyk, 2006). Eliminating or blocking nerve impulse transmission can alleviate such conditions (Chambers, 2008; Engel and Schirmer, 1974; Ko and Kim, 1997), however available treatments have many drawbacks. Chemical nerve blocks can be non-specific, incomplete, destructive or temporary, while surgical nerve transection is destructive and irreversible. Electrical stimulation at frequencies above 1 khz (HF stimulation) can immediately and reversibly block action potential transmission (Bhadra and Kilgore, 2004). HF block has been demonstrated with consistent parameters in the sciatic and pudendal nerves (Bhadra et al., 2006; Bhadra and Kilgore, 2005; Gaunt and Prochazka, 2008; Kilgore and Bhadra, 2004; Tai et al., 2004; Tai et al., 2005; Williamson and Andrews, 2005). These somatic nerves contain fibers with average diameters ranging from 6 9 µm (Adam and Friede, 1988; Martin et al., 1974; Prodanov and Feirabend, 2007). However, HF block of small mammalian myelinated axons (< 5 µm) has not been characterized. HF small fiber block would have significant clinical applications. Myelinated human nociceptors are typically small Aδ fibers with diameter ranging from 2 5 µm (Wilkinson, 2001). Demonstrating HF block in an animal model would suggest the potential for human Aδ fiber block. Reversible, immediate Aδ fiber block could permit the development of HF block-neuroprosthesis for the management of chronic pain.

65 51 Bladder voiding neuroprostheses could replace destructive permanent dorsal rhizotomies with reversible dorsal afferent nerve block. Dorsal afferent nerve block would allow voiding while preserving sacral reflexes governing erection and defecation. These reflexes are extremely important to patient quality of life (Anderson, 2004). Patients would be more likely to accept reflex-preserving bladder voiding neuroprostheses, reducing patient morbidity and cost of care. HF small fiber block can be investigated in a feline bladder model. Parasympathetic preganglionic bladder efferents are small myelinated fibers, with axon diameters ranging from 1-3 µm (De Groat et al., 1982; Morgan, 2001) and are accessible at the sacral root level. HF conduction block of these small bladder efferents causes a reduction in evoked bladder pressure. The completeness and rapidity of the bladder pressure reduction depends on the completeness and rapidity of efferent block. The onset response to HF stimulation characteristics can also be determined from bladder pressure measurements. The purpose of this study was to 1) demonstrate local HF small fiber conduction block, 2) characterize the dependence of small fiber HF conduction block on stimulation parameters, and 3) characterize HF evoked bladder responses. Sacral HF small nerve fiber block could form the basis for nerve-sparing neuroprostheses for patients with chronic pain and voiding dysfunctions. III.3 Methods This study was conducted on five male cats (4.1 ± 1.1 kg). The Institutional Animal Care and Use Committee of Case Western Reserve University approved all animal care and experimental procedures.

66 52 III.3.1 Experimental Setup Data were collected under intravenous α-chloralose anesthesia (65 mg kg -1 induction, 15 mg kg -1 maintenance, Sigma, St. Louis, MO). An anesthetic monitoring system (SurgiVet V9200 Advisor Monitor, Smiths Medical PM Inc, Waukesha, WI) was used to record body temperature, ECG, blood oxygen saturation, blood pressure, and expired pco 2. The appropriate depth of anesthesia was maintained by monitoring blood pressure, reflexes, and heart rate. Respiration was maintained by a pressure-regulated respirator (ADS 1000, Engler Engineering Corporation, Hialeah, FL). Temperature was maintained between 37 and 39 C with a heating blanket. Saline 0.9% with 8.4 mg ml -1 sodium bicarbonate and 5% dextrose was administered at 1-5 ml hr -1 IV. Data were recorded on a computer equipped with a data acquisition card (NI PCI- 6221, National Instruments Corporation, TX, sampling rate 99 Hz) using a customized interface (Labview 8.0, National Instruments Corporation, TX). Data were also recorded on a strip chart (TA-11, Gould Inc., Valley View, OH). Signal conditioning and routing was provided by strain gauge signal-conditioning modules and configurable feed-through modules (NI SCC-SG24 and SCC-FT01, National Instruments Corporation, TX) contained in an NI terminal block (NI SC-2345, National Instruments Corporation, TX). Bladder pressure was measured using an external pressure transducer (Deltran IV, Utah Medical Devices, Midvale, UT) connected to a dual lumen suprapubic bladder catheter (DLC 6D, Life Tech, Stafford, TX). The second lumen of the suprapubic bladder catheter was used to fill or drain the bladder.

67 53 III.3.2 Surgical Preparation The bladder was exposed through a suprapubic midline incision and the 6F duallumen catheter was inserted through the dome of the bladder and secured by a seromuscular purse string suture. A foam stand was used to support the chest and pelvis without compressing the abdomen. The sacral roots were accessed extradurally by a lumbosacral laminectomy (L5 to S2). Nerve cuffs containing dual tripolar electrodes (Figure 1, electrodes 2 and 3) were implanted on the two extradural sacral roots (right and left) evoking the largest bladder contractions when stimulated with a custom-made bipolar hook electrode. Post mortem dissection verified that the cuffs were implanted on the S2 nerve roots. The implanted sacral root exhibiting the best block characteristics (left or right) was chosen for experimentation. In two animals, a tripolar spiral nerve cuff electrode was intradurally implanted proximal to the extradural nerve cuff (Figure 1, electrode 1). Trials were conducted at bladder volumes evoking resting bladder pressures around 15 cmh 2 O (± 3 cmh 2 O). The bladder was periodically emptied and refilled. Physiological filling occurred during the experiment. Therefore the volume infused (19 ml ± 5 ml) was less than the volume emptied (31 ml ± 7 ml). Excitatory 20 Hz stimulation of the implanted S2 roots evoked minimal external urethral sphincter pressure in some animals. These animals required urethral obstruction to prevent voiding during excitatory stimulation. Urethral obstruction was provided by an embolectomy catheter (CM702-80, Cardio International, Inc, NY), or polyprolylene catheter (3F or 5F Sovereign, Kendall, MA) placed in the urethra.

68 54 III.3.3 Electrical Stimulation All stimuli were voltage-controlled and generated by a DS8000 Digital Stimulator and DLS100 Isolator (World Precision Instruments Inc, Sarasota, FL). HF Stimulation A biphasic square wave (HF stimulation) was used to generate a local conduction block. HF stimulation frequencies were 2.08, 6.25, and 12.5 khz (hereafter referred to as 2, 6 and 12 khz). HF stimulation amplitudes were typically 1 8 Vpp for 2 khz trials and 2 16 Vpp for 6 and 12 khz trials. Slightly higher amplitudes were used in four trials validating HF block (two trials at 2 khz, 12 Vpp and two trials at 12 khz, 20 Vpp). Excitatory Stimulation Bladder pressure was evoked using biphasic rectangular pulses (100 µs pulse width) applied at 20 Hz (excitatory stimulation). Thresholds (0.8 ± 0.2 Vpp) and maximal stimulation levels (6 ± 5 Vpp, range 1 20 Vpp) were determined for each electrode. The average evoked bladder pressure was 73 ± 19 cmh 2 O. III.3.4 Experimental Design Six experiments were conducted in five animals. An interval of 13 hours separated two experiments conducted in one animal. Each experiment consisted of several sets of trials, each including multiple trials with HF stimulation and control trials (Table I). HF waveform parameters were randomized within each set of trials and the bladder was drained and refilled between sets of trials. Transmission Block The dependence of transmission block characteristics on HF waveform parameters was characterized by both HF and Control trials. Excitatory stimulation

69 55 (Figure 1, electrode 2) was applied in both HF and Control trials. HF trials (Figure 2) also included 40 seconds of HF stimulation (Figure 1, electrode 3) at various HF waveform amplitudes and frequencies. HF trials could be separated into three intervals (Figure 2). Excitatory stimulation evoked a maximal amplitude bladder contraction during the pre-hf interval (0 20 seconds). HF stimulation blocked preganglionic action potential conduction during the HF interval (20 60 seconds), causing a measurable reduction in bladder pressure. In the post-hf interval (60 80 seconds), excitatory stimulation re-evoked a bladder contraction, the magnitude of which indicated the reversibility of block. Control trials characterized the input/output relationship between the preganglionic bladder efferents and the evoked bladder pressure. Sacral preganglionic bladder efferents were stimulated with pulse trains representing either the absence of HF stimulation (No Block trials) or perfect conduction block (Ideal Block trials). The bladder response to these stimulation patterns provided a baseline to which HF stimulationinduced reductions in bladder pressure could be compared. No Block control trials evaluated the bladder response to excitatory stimulation in the absence of HF stimulation. These trials were comprised of 80 seconds of uninterrupted excitatory stimulation at 20 Hz. Ideal Block control trials evaluated bladder response to a simulated perfect conduction block with interrupted excitatory stimulation (20 seconds on 40 seconds off 20 seconds on). The 40-second interval without excitatory stimulation mimicked perfect conduction block of proximally evoked action potentials. The maximum evoked bladder pressures before and after HF stimulus (P ves max and P ves_post ) and the minimum (P ves_min ) and average (P ves_avg ) evoked bladder pressures

70 56 were extracted from each trial (Figure 2). All extracted bladder pressure features were measured with respect to the baseline bladder pressure before the initiation of the trial. P ves max was the maximum pressure over the first 20 seconds of the trial and was used to normalize other extracted bladder pressure features, controlling for variability in the evoked bladder pressure. P ves_min was the minimum pressure and P ves avg was the average pressure over the interval of seconds from the start of the trial. P ves post was the average pressure over the final second of stimulation (79 80 seconds following the initiation of the trial). Normalizing (P ves max - P ves min ), P ves avg, and P ves post by P ves max generated ratios (Block ratio: B ratio, Rapidity ratio: A ratio, Reversibility Ratio: R ratio ) that could be better compared across experiments. B ratio provided an estimate of HF block completeness and could vary from 0 (bladder pressure does not decline at all during HF stimulation) to 1 (complete block, P ves min equals pre trial baseline). Preganglionic bladder drive was not present in Ideal Block trials, minimizing P ves min and maximizing B ratio. The reduction in bladder pressure due to prolonged excitatory stimulation, which could be mistaken for poor block, was measured in No Block control trials. HF trials B ratios ranged between No Block and Ideal Block control values, depending on the HF stimulation parameters. The rapidity ratio (A ratio ) estimated HF block rapidity and could vary from 0 (bladder pressure equals baseline during HF stimulation) to 1 (bladder pressure does not decline at all during HF stimulation). Similar HF and Ideal Block control trial A ratio s imply a minimal onset response to HF stimulation. Similar HF and No Block control trial A ratio s imply a prolonged onset response to HF stimulation. B ratio and A ratio could vary independently. A prolonged HF onset response could result in a trial with a low B ratio and

71 57 high A ratio while incomplete but rapid block could result in a trial with a low A ratio and (relatively) high B ratio. The reversibility ratio (R ratio ) measured the reversibility of conduction block and could vary from 0 (excitatory stimulation fails to evoke bladder pressure following the cessation of HF stimulation) to 1 (excitatory stimulation evokes equal magnitude contractions before and after HF stimulation). Low R ratio s imply neural fatigue or a failure of nerve impulse transmission beneath the HF stimulation electrode. Similar HF and Ideal Block control trial R ratio s imply HF stimulation minimally affects bladder excitability and conduction block is rapidly reversible. Block Validation trials, conducted in two animals, included proximal excitatory stimulation (Figure 1, electrode 1), HF block (Figure 1, electrode 2), and distal excitatory stimulation (Figure 1, electrode 3). Proximal excitatory stimulation evoked bladder contractions that were blocked by HF stimulation (Figure 3). Distal excitatory stimulation evoked bladder contractions during HF block. Bladder excitability during HF block demonstrated that block was localized to the blocking electrode and was not the result of neuromuscular transmitter depletion or junction fatigue (Figure 3). HF Evoked Bladder Response HF and control trials examined the dependence of HF evoked onset response on stimulation amplitude and frequency. HF stimulation (40 seconds) was applied in HF trials, followed immediately by excitatory stimulation (20 seconds) (Figure 4). Post-HF excitatory stimulation investigated the effects of HF stimulation on bladder excitability. Control trials included 60 seconds of excitatory stimulation and provided a benchmark for bladder contractions evoked by HF or Post-HF excitatory stimulation.

72 58 The maximum (P ves max ) and average (P ves avg ) evoked bladder pressure over the initial 40 seconds of stimulation was extracted. The average evoked bladder pressure over the final second of stimulation (P ves post ) was also measured. P ves max, P ves avg and P ves post were normalized within each set of trials using the average P ves max for control trials in that set of trials. III.3.5 Data Analysis Bladder pressure measures were extracted using Matlab (Mathworks, Inc., Natick, MA) and analyzed using Minitab (Minitab, Inc., State College, PA). Statistical analyses of Transmission Block and HF evoked bladder response trials were performed on the experimental averages of extracted variables. Control trials were averaged by experiment. HF trials were averaged by experiment, frequency and amplitude. HF Trials were not conducted at the lowest stimulation amplitudes (1 Vpp for 2 khz, 2 Vpp for 6 and 12 khz) in experiments 1 and 2; to balance the ANOVA estimated values were assigned for experiments 1 and 2 using the averages for trials at those parameters in the remaining experiments. The B ratio, A ratio, and R ratio experimental averages were regressed against experiment and HF stimulation amplitude for each HF stimulation frequency. The experimental averages for Control trials and maximum amplitude HF trials (8 Vpp for 2 khz, 16 Vpp for 6 and 12 khz) were regressed against experiment and stimulation type (Ideal Block, No Block, 2 khz, 6 khz, and 12 khz). The experimental averages P ves max, P ves avg, and P ves post from the HF evoked bladder response trials were analyzed in the same manner.

73 59 Additional analyses were performed on two subsets of data. Two-kilohertz stimulation blocked nerve impulse transmission in experiments 1, 4, and 5. The analysis of B ratio was repeated for these experiments. Excitatory stimulation (Figure 1, electrode 2) evoked minimal bladder contractions upon the cessation of HF stimulation in Animal I (Figure 1, electrode 3). The analysis of R ratio was repeated excluding this animal. III.4 Results Six experiments were conducted in 5 animals. Transmission block trials (185 total) and bladder HF response trials (114 total) were conducted in 4 animals (see Table II). Block validation trials (19 total) were conducted in 2 animals. A limited dataset was recorded from one animal due to anesthetic complications: data from this animal was not included in the analysis. III.4.1 Transmission Block HF stimulation blocked nerve impulse transmission in high-amplitude 6 or 12 khz trials in all animals. High-amplitude 2 khz HF stimulation blocked nerve impulse transmission in experiments 1, 4, and 5. The 2 khz trials in the remaining experiments resembled No Block control trials. In animals II-IV excitatory stimulation (Figure 1, electrode 2) immediately evoked a large bladder contraction upon cessation of HF stimulation (Figure 1, electrode 3). In animal 1 a minimal increase in bladder pressure followed the cessation of HF stimulation. However, in all animals a three-minute delay between trials was sufficient to allow the evoked bladder pressure to fully recover. Block Completeness (B ratio )

74 60 Block completeness for 6 or 12 khz trials was comparable to block completeness for Ideal Block trials (P > 0.57, n = 25, Figure 5) and better than block completeness for No Block control trials (P < 0.001, n = 25). Block completeness did not differ significantly between 6 and 12 khz trials (P = 1.00, n = 25). Block completeness improved with amplitude (Figure 6) and varied between experiments for 6 and 12 khz trials (P < 0.001, n = 18 for both tests). Block completeness was significantly worse for 2 khz trials than for 6 or 12 khz trials (P < 0.05, n = 25). Two-kilohertz trials were comparable to No Block trials (P = 0.05, n = 25). Block did not improve with amplitude (P = 0.14, n = 18) and varied with experiment. Two-kilohertz stimulation blocked proximally evoked impulses in experiments 1, 4, and 5 (B ratio : 0.79 ± 0.09) but not in the remaining experiments (B ratio : 0.51 ± 0.10). Two-kilohertz trials in experiments 1, 4, and 5 were not significantly different from for 6 or 12 khz trials (P > 0.35, n = 15) and had significantly greater B ratios than No Block control trials (P < 0.01, n = 15). However, block completeness remained slightly worse for these 2 khz trials than for Ideal block trials (P = 0.048, n = 15). Block completeness improved with amplitude in these trials (P < 0.05, n = 11) but did not depend on experiment. Block Rapidity (A ratio ) Block rapidity improved with increasing stimulation frequency (Figure 5) and amplitude (Figure 6). Block rapidity in 12 khz trials was comparable to Ideal Block trials (P = 0.10, n = 25). Block was more rapid in 6 and 12 khz trials than in 2 khz trials or No Block control trials (P < 0.001, n = 25, all comparisons). Block rapidity did not differ

75 61 between 6 and 12 khz trials (P = 0.94, n = 25). Block rapidity in 2 khz trials was comparable to block rapidity in No Block control trials (P = 0.15, n = 25). A ratio depended on stimulation amplitude for 6 and 12 khz trials (P < 0.001, n = 18, for both comparisons) but not for 2 khz trials (P = 0.86, n = 18). A ratio also depended on experiment. Block Reversibility (R ratio ) Block reversibility in HF trials did not depend on frequency (P > 0.85, n = 25, for all comparisons) and was comparable to block reversibility in No Block control trials (P > 0.32, n = 25, for all comparisons). HF trials at 2 and 12 khz were comparable to Ideal block trials (P > 0.05, n = 25, for both comparisons). R ratio improved with amplitude for 6 khz trials (P < 0.05, n = 18) but not 2 khz trials (P = 0.09, n = 18) or 12 khz trials (P < 0.05, n = 18, but no significant difference in means found by post-hoc comparisons). R ratio also varied between experiments. Block was reversible in Animals II-IV (R ratio : 0.70 ± 0.11) but not in Animal I (R ratio : 0.47 ± 0.12). Excluding Animal I, HF trials were more reversible than No Block control trials for all stimulation frequencies and (P < 0.005, n = 15, Figure 5) and block reversibility improved with HF stimulation amplitude (P < 0.05, n = 12, Figure 6). Block reversibility in 12 khz and Ideal Block trials was comparable (P = 0.34, n = 15), though reversibility remained significantly worse for 6 and 2 khz trials than for Ideal Block trials (P < 0.05, n = 15). R ratio did not differ between stimulation frequencies (P > 0.26, n = 15) and also varied between experiments. Block Validation Trials

76 62 Block Validation trials were conducted in two animals. Application of distal stimulation during HF block evoked bladder pressures similar to those evoked by intradural stimulation (Figure 3). III.4.2 HF Evoked Bladder Response Magnitude (P ves max ) and Persistence (P ves avg ) HF stimulation at 6 and 12 khz evoked significantly smaller and less persistent bladder contractions than 20 Hz stimulation (Figure 7). Both P ves avg and P ves max depended on stimulation type (P < 0.001, n = 16) and experiment (P < 0.05, n = 16). Both were significantly lower for 6 and 12 khz trials than Control or 2 khz trials (P < 0.001, n = 16 all comparisons). P ves max and P ves avg for 6 and 12 khz trials did not significantly differ (P > 0.58, n = 16 both comparisons). P ves max did not significantly differ between 2 khz trials and Control trials (P = 0.12, n = 16). P ves avg was significantly lower for 2 khz trials than Control trials (P <0.01, n = 16). Both P ves max and P ves avg increased with increasing amplitude to a maximum value. The maximizing amplitude differed depending on stimulation frequency (2 khz: 4 Vpp, 6 khz: 4 Vpp, 12 khz: 8 Vpp). For 2 and 12 khz, but not 6 khz, trials conducted at the minimum stimulus amplitude had significantly lower P ves max than trials conducted at the stimulation amplitude maximizing the bladder response (2 khz: P < 0.05, n = 15; 12 khz: P < 0.001, n = 15). Excitability (P ves post ) The preganglionic bladder efferents remained excitable following HF stimulation at 6 or 12 khz (Figure 7). P pves post depended on stimulation type (P < 0.01, n = 16) but not experiment (P = 0.10, n = 16). P pves post for 2 khz trials was not significantly different

77 63 from other trials (P > 0.08, n = 16, all comparisons). P pves post for HF trials at 6 khz or 12 khz were both significantly greater than P pves post for Control trials (P < 0.05, n = 16, both comparisons). P pves post for HF trials at 6 khz or 12 khz did not differ significantly (P = 0.90, n = 16). P pves post did not depend on stimulation amplitude for any HF stimulation frequency (P > 0.08, n = 15), but did depend on experiment for 12 khz (P < 0.001, n = 15) but not 2 or 6 khz (P > 0.05, n = 15). III.5 Discussion This study demonstrated and characterized reversible small fiber HF block. Completeness, rapidity and reversibility of conduction block depended on HF stimulation amplitude and frequency. A local nerve conduction block mechanism was verified through block validation trials. Block characteristics for maximal amplitude 6 or 12 khz HF stimulation trials approximated Ideal Block trial characteristics (Figure 5), demonstrating that efferent block was nearly complete with minimal onset response or recurrent firing. Lower amplitude HF trials and some 2 khz trials resembled No block trials. Preganglionic efferent conduction in these trials was either unaffected by HF stimulation or HF onset responses were significant and sustained. Previous HF block studies measuring compound action potentials (CAP) (Tanner, 1962; Woo and Campbell, 1964) also demonstrated Aδ fiber block, though at a single frequency (20 khz). HF waveforms interposed between stimulation and measurement electrodes progressively reduced the measured CAP as the HF waveform amplitude was increased. HF block of individual dorsal afferent fibers was also demonstrated. However, asynchronous activation of neurons during HF stimulation may have contributed to the

78 64 elimination of the compound action potential (Woo and Campbell, 1964) and conduction velocities and fiber diameters were not reported for the single fiber trials. This study characterized small fiber HF conduction block for a range of stimulus frequencies and amplitudes using clinically relevant outputs and an unambiguous response variable. The measured bladder pressure depended solely on parasympathetic preganglionic bladder efferents 1 3 µm in diameter. Bladder pressure provided an estimate of overall nerve block preferable to CAP or single unit recordings. Output measures of transmission block and HF evoked bladder response trials were relevant to the design of HF block neuroprotheses. III.5.1 HF Block Amplitude and Frequency Thresholds HF block characteristics exhibited frequency-specific amplitude thresholds (Figure 6). B ratio and A ratio demonstrated a sharp amplitude threshold for 6 and 12 khz: trials above and below the cutoff were significantly different. The threshold for 6 khz was 4 Vpp and the threshold for 12 khz was 8 Vpp. No obvious cutoff existed for 2 khz. The 2 khz HF block amplitude threshold was likely below the lowest test amplitude (1 Vpp). The increase in amplitude thresholds for block with increasing stimulation frequency was consistent with results reported for larger fibers (Kilgore and Bhadra, 2004; Williamson and Andrews, 2005). A frequency threshold for HF block was evident. Block characteristics improved more rapidly between 2 and 6 khz than between 6 and 12 khz. HF block was consistently achieved at 6 and 12 khz, but not at 2 khz. Frequency thresholds were more pronounced for HF evoked bladder responses. HF stimulation at 6 or 12 khz evoked small, transient contractions. HF stimulation at 2

79 65 khz stimulation evoked larger, more persistent contractions suggestive of a prolonged HF onset response. III.5.2 HF Neuroprotheses Design Implications These results have direct application to the design of implanted HF neuroprotheses. Implanted neuroprotheses are subject to stimulation power and amplitude constraints. Maximizing device efficacy requires minimizing HF stimulation amplitude while maintaining adequate conduction block. Block amplitude threshold increases with stimulation frequency. Therefore lower stimulation frequencies are desirable. However, there exists a lower bound for HF block frequency. HF stimulation at lower frequencies resulted in greater onset responses, an effect previously observed in somatic fibers (Kilgore and Bhadra, 2004). HF stimulation at 2 khz evoked a large onset response (Figure 7) and did not consistently block the small parasympathetic preganglionic bladder efferents. HF stimulation frequencies above 2 khz were preferable. Trials providing 6 khz stimulation demonstrated minimal onset response and block characteristics approximating Ideal Block trials. Increasing HF stimulation frequency to 12 khz significantly increased block amplitude threshold but did not significantly improve block characteristics. HF stimulation at 2 khz was unreliable and exhibited poor block characteristics. Stimulation frequencies at or near 6 khz may provide a good combination of reduced power consumption and desirable block characteristics for small fiber block applications.

80 66 III.5.3 Validating HF Block Simulations Computer simulations of HF block are being developed to investigate HF block mechanisms and evaluate novel waveforms for HF block neuroprostheses. HF block of 2 and 5 µm neurons has been simulated using modified Frankenhauser-Huxley (FH) (Zhang et al., 2006) and Schwarz Reid Bostock (SRB) (Schwarz et al., 1995) axon models (Liu et al., 2009). Other studies have used the McIntyre-Richardson-Grill (MRG) axon model (McIntyre et al., 2002) to investigate fibers down to 7.3 µm (Bhadra et al., 2007). These studies differ in the minimum frequency threshold for block (block frequency threshold) and the dependence of block amplitude threshold on stimulation frequency. Our results provide an opportunity to evaluate the predictions of these models. The MRG and SRB models provided better estimates of the block frequency threshold and the dependence of block amplitude threshold on frequency than the FH model. Block was consistently demonstrated at 6 and 12 khz in this study, with a subset of experiments demonstrating block at 2 khz. Block was achieved at 3 khz with the MRG model (2.2 khz for larger fibers) and 4 khz with the SRB model, while consistent block was not achieved below 10 khz with the FH model. Experimental block amplitude thresholds increased with stimulation frequency. Thresholds were the lowest for 2 khz stimulation and approximately doubled between 6 and 12 khz. A similar dependence on frequency was observed in the MRG and SRB axon models, although block amplitude threshold increased more slowly with stimulation frequency. The FH model showed no block threshold dependence on stimulation frequency.

81 67 All models demonstrated an increase in block amplitude threshold as fiber diameter decreased. However, stimulation amplitudes required to block nerve impulse conduction in this study did not differ greatly from those required to block conduction in the rat sciatic nerve (Bhadra and Kilgore, 2005) and feline pudendal nerve (Bhadra et al., 2006; Boger et al., 2008). It is unlikely that HF block amplitude thresholds for individual neurons are independent of fiber diameter. Instead, these results suggest that in vivo block amplitude thresholds are dominated by other factors, such as the distance from the neuron to the electrode. The predictions of the SRB and MRG axon models were more consistent with the experimental results of this study than those of the FH axon model. Both SRB and MRG models described an increase in block amplitude threshold with increasing stimulation frequency and both predicted block at 6 khz. Refinement of the SRB and MRG HF block models, such as the extension of the MRG model to neurons with diameters below 5 µm, may improve model accuracy and promote advances in waveform or electrode design, speeding clinical translation of HF block technology. III.5.4 Chronic Pain and Bladder Voiding Neuroprostheses This study demonstrates complete HF conduction block of small myelinated parasympathetic preganglionic bladder efferents with diameters less than human nociceptive Aδ afferents. The consistency of block parameters for somatic nerves across animal models suggests that HF block of Aδ nociceptive may be possible at similar parameters. Neuroprosthetics for relief of chronic pain could potentially utilize HF block as a nerve-sparing alternative to neurotomy or neurolysis.

82 68 Bladder voiding neuroprostheses may be able to utilize sacral HF block of somatic efferents or sensory afferents. Large-diameter somatic EUS fibers were likely blocked by the same HF stimulation that blocked small-diameter parasympathetic efferents. Sacral bladder voiding neuroprostheses could potentially block EUS activation and generate bladder pressure using electrodes implanted solely on the sacral roots. HF block of dorsal sensory afferents could directly substitute for the dorsal rhizotomy used in the Brindley system to eliminate EUS reflexes and allow post-stimulus bladder voiding. Sacral neuroprostheses could benefit from existing electrodes and surgical methods speeding clinical translation and enhancing patient acceptance. Acknowledgements This work was supported by the Department of Veterans Affairs RR&D B3675R and NIH DK077089, EB and AR07505, and the Cleveland VA FES Center. The authors wish to thank Timothy Bruns, Tina Goetz, and Timothy Mariano for their experimental assistance.

83 Figure 1: Three electrodes were implanted at two locations along the S2 sacral nerve. A spiral nerve cuff containing a tripolar electrode was implanted intradurally (1) and a molded nerve cuff containing two tripolar electrodes (proximal 2; distal 3) was implanted extradurally. 69

84 Figure 2: Transmission block trial stimulus timings and extracted bladder pressure features (12 khz, 8 Vpp). HF block trials included excitatory stimulation (Stim, dashed line, electrode 2 in Figure 1) and HF stimulation (Block, solid line, electrode 3 in Figure 1). No Block control trials included 80 seconds of excitatory stimulation. Ideal Block control trials included intermittent excitatory stimulation (20 seconds on - 40 seconds off 20 seconds on). P ves max : Maximum evoked bladder pressure preceding HF block (0 20 seconds). P ves avg : Average evoked bladder pressure during HF stimulation (20 60 seconds). P ves min : Minimum evoked bladder pressure during HF stimulation. P ves post : Average evoked bladder pressure over the final second of proximal stimulation (79 80 seconds). 70

85 Figure 3: Validation of conduction block mechanism. Evoked bladder response and stimulation timing of a Block Validation trial (12 khz, 16 Vpp). Block validation trials included 20 Hz stimulation (Proximal Stim; electrode 1 in Fig. 1, Distal Stim; electrode 3 in Fig 1) and HF stimulation (Block; electrode 2 in Fig. 1). The application of HF stimulation caused a rapid, reversible decline in proximally evoked bladder pressure. The bladder remained contractile during HF stimulation as demonstrated by the response to distal stimulation: conduction block was localized to the region near the blocking electrode. 71

86 Figure 4: Stimulus timing and extracted bladder pressure measures for HF evoked bladder response trials (12 khz, 8 Vpp). Trials used a single electrode (electrode 3 in Fig 1) to provide both HF (HF solid line) and 20Hz stimulation (Stim dashed line). P ves max : Evoked maximum bladder pressure during interval 0 40 sec. P ves avg : Average evoked bladder pressure during the interval 0 40 sec. P ves post : Average bladder pressure over the final second of stimulation (59 60 seconds). 72

87 Figure 5: Transmission block (see Figure 2) was complete, rapid and reversible. Block characteristics were significantly improved over No Block levels by the application of 6 or 12 khz HF stimulation. HF block trials at these parameters approach the block characteristics of Ideal block trials. Block characteristics for 2 khz trials varied greatly between experiments, resembling either Ideal or No Block control trials. For each frequency and block characteristic, means significantly different from the means for Ideal ( : P < 0.05; : P < 0.01) and No Block (*: P < 0.05; **: P < 0.01) control trials are labeled. Trials shown were conducted at 16 Vpp for 6 and 12 khz and 8 Vpp for 2 khz. Block reversibility results are shown for Animals II-IV. 73

88 Figure 6: Transmission block (see Figure 2) characteristics improve with increasing stimulation amplitude. A clear amplitude threshold was demonstrated for HF stimulation at 6 (4 Vpp) and 12 khz (8 Vpp). No clear amplitude threshold was observed for 2 khz trials: partial block may have been achieved at 1 Vpp. Block rapidity in trials including 2 khz stimulation was poor regardless of stimulation amplitude. Block reversibility improved with stimulation amplitude for all stimulation frequencies in Animals II-IV (shown). 74

89 Figure 7: HF-evoked bladder response (see Figure 4) improves with increasing HF stimulation frequency. The magnitude and persistence of the evoked bladder response decreased with increasing frequency. The post-hfac excitability following 6 and 12 khz HF stimulation was significantly greater than excitability following control trials. Means significantly different from the mean for Control trials for each frequency and bladder response characteristic are labeled (**: p < 0.01; *: p < 0.05). Trials conducted at 16 Vpp for 6 and 12 khz and 8 Vpp for 2 khz. All values normalized by the magnitude of the evoked bladder response for 20 Hz control trials. 75

90 76 Animal Experiment Frequency I 1 2 II 3 III 4 IV 5 Transmission Block HF Evoked Response Control 6 (3) Control 7 (3) Control 8 (4) Control 14 (6) Control 12 (6) 6 Table I: Distribution of trials by animal, experiment, and HF stimulation frequency, for transmission block and bladder HF response trials. Transmission block control trials include both Ideal Block and No Block Trials, with the number of Ideal Block trials indicated in parentheses.

91 77 CHAPTER IV: HIGH FREQUENCY SACRAL ROOT NERVE BLOCK ALLOWS BLADDER VOIDING

92 78 IV.1 Abstract 1) Aims: Spinal cord injury often results in reflexive external urethral sphincter (EUS) activity that can inhibit complete bladder voiding and is associated with significant medical complications. Neurotomies can prevent EUS activation and allow bladder voiding, but patient acceptance of irreversible neurotomies is limited. This study investigated sacral high frequency (HF) electrical conduction block s effectiveness in preventing EUS activation and allowing bladder voiding. 2) Methods: Bladder activation was provided by bilaterally implanted extradural tripolar S2 sacral nerve root electrodes in 6 cats. Two unilaterally implanted extradural S1 sacral nerve root electrodes controlled EUS pressure. Severe reflexive EUS activity was simulated using the proximal bipolar electrode. HF stimulation was applied to the distal S1 tripolar electrode to block nerve impulse conduction and reduce EUS pressure. 3) Results: HF stimulation improved single trial voiding efficiency from 3 ± 6% to 59 ± 12% of the initial bladder volume voided. Complete bladder voiding (82 ± 9% of initial bladder volume voided) was achieved through successive HF block trials or by extended duration trials. 4) Conclusions: HF extradural sacral stimulation can block EUS motor drive and alllow complete bladder voiding. Sacral HF EUS block may improve HF block neuroprostheses for voiding dysfunction.

93 79 IV.2 Introduction Following neurologic disease or injury such as spinal cord injury the external urethral sphincter (EUS) and bladder may contract simultaneously, a condition termed detrusor-sphincter dyssynergia (DSD) (Blaivas, 1982). DSD can lead to urinary tract infections, autonomic dysreflexia, vesico-ureteric reflux and hydronephrosis (Hackler, 1977). Clean intermittent self-catheterization is the conventional first-line treatment for DSD, in combination with anti-muscarinic drugs to reduce bladder activity. However, long term patient compliance with this regime is poor and treatment is not suitable for certain patient populations (Yavuzer et al., 2000). In the absence of descending supraspinal sphincter inhibition, spinal continence reflexes can activate the EUS during micturition and prevent voiding (Galeano et al., 1986; Shefchyk, 2006). Neurotomies can allow voiding by interrupting these reflex arcs. The Brindley approach combines intermittent intradural or extradural sacral stimulation for bladder voiding with a dorsal root neurotomy. The dorsal root neurotomy interrupts the afferent limb of the feedback loop, preventing EUS activation and allowing complete bladder voiding (Brindley, 1994). This approach is effective both clinically and cost-wise (Brindley, 1994; Creasey et al., 2001). Pudendal neurotomies, which interrupt the efferent limb of the feedback loops, also allow voiding (Engel and Schirmer, 1974). However, neurotomies eliminate dorsal root or pudendal nerve (PN) mediated reflexes and residual sensation. These reflexes are very important for patient quality of life (Anderson, 2004). Thus patient acceptance of neurotomies for voiding is limited, in spite of clinical effectiveness and cost of care advantages.

94 80 High frequency alternating current (HF) waveforms can mimic a neurotomy by preventing nerve impulse transmission. HF nerve block is immediate and is rapidly reversible; nerve impulse transmission resumes upon cessation of stimulation (Bhadra and Kilgore, 2004). PN HF block has been used to replace PN neurotomies and produce bladder voiding (Boger et al., 2008; Tai et al., 2007b). Voiding neuroprostheses using sacral HF EUS block could be clinical translation implemented utilizing existing human-approved electrodes and could provide a stable, protected location for electrodes within the spinal canal. Sacral HF stimulation would also provide a comprehensive block of sacral efferents, which may allow voiding in patients for whom PN EUS block is insufficient. The goals of this study were to achieve EUS block at the sacral root level; demonstrate that HF stimulation improved post-stimulus bladder voiding; show bladder voiding with HF stimulation equivalent to unobstructed bladder voiding; and achieve clinically effect bladder voiding using extended duration or successive trials. IV.3 Methods This study was conducted on six male cats. The Institutional Animal Care and Use Committee of Case Western Reserve University approved all animal care and experimental procedures. Data were collected under a combination of intravenous α- chloralose anesthesia (65 mg kg -1, Sigma, St. Louis, MO) and Isoflurane (.25 3%). Subcutaneous buprenorphine (0.01 mg kg -1, Henry Schein, Melville, NY) was provided at induction and approximately every 12 hours thereafter. An anesthetic monitoring system (SurgiVet V9200 Advisor Monitor, Smiths Medical PM Inc, Wisconsin) was used to observe body temperature, ECG, blood oxygen saturation, blood pressure, and expired

95 81 pco2. Blood pressure, heart rate, and withdrawal and blink reflexes were monitored to maintain the appropriate depth of anesthesia. Respiration was maintained by a pressureregulated respirator (ADS 1000, Engler Engineering Corporation, FL) and adjusted by expired pco2 measurements. Temperature was maintained between 37 and 39 C with a heating blanket. Saline 0.9% with 8.4 mg ml -1 sodium bicarbonate and 5% dextrose was administered at 3-20 ml kg hr -1 IV. A 6F dual-lumen catheter (DLC 6D, Life Tech, TX) was implanted suprapubicly in the dome of the bladder. Bladder pressure was measured using an external pressure transducer (Deltran IV, Utah Medical Devices, UT) connected to one lumen of the suprapubic bladder catheter. Sphincter pressure was measured using a 4F catheter-mounted dual sensor microtransducer placed in the urethra (C7C4F, MMI- Gaeltech, Hackensack, NJ). A lumbosacral laminectomy was performed from L6 to S2. Extradural tripolar cuff electrodes were implanted bilaterally on the S2 sacral roots. Two electrodes, a proximal bipolar electrode and a distal tripolar electrode, were implanted extradurally on the right S1 root (Figure 1). Signal conditioning and routing was provided by strain gauge signal-conditioning modules and configurable feed-through modules (NI SCC-SG24 and SCC-FT01, National Instruments Corporation, TX) contained in an NI terminal block (NI SC-2345, National Instruments Corporation, TX). Data were recorded on a computer equipped with a data acquisition card (NI PCI- 6221, National Instruments Corporation, TX, sampling rate 99 Hz) using a customized

96 82 interface (Labview 8.0, National Instruments Corporation, TX). Data were also recorded on a strip chart (TA-11, Gould Inc., AZ). The volume voided as a function of time was recorded using a high-resolution single point load cell (s215, 0.9kg; Strain Measurement Devices, CT or ESP 0.6 kg, Load Cell Central, PA). A foam stand was used to support the legs and elevate the pelvis. Saline (30 ml total) was infused into the bladder at a rate of 1 4 ml / minute (Genie YA-12, Kent Scientific, CT). Initial bladder volume for each trial depended on the physiological urine generation rate, which varied between animals. Baseline bladder pressures preceding trials were less than 20 cmh 2 O and bladder volumes (34 ± 7 ml) did not differ significantly between trial types. IV.3.1 Electrical Stimulation A multichannel, isolated voltage-controlled stimulator was used to generate stimulation waveforms (DS8000 and DSI 100, World Precision Instruments Inc, FL). Bladder Activation: Bladder pressures were generated by intermittent (2 s on / 4 s off) 20 Hz bilateral S2 sacral root stimulation (bladder stimulation) (Figure 1). The 4-second off time duration was chosen to allow voiding to completion before the next 2-second stimulation interval. Stimulus pulses were balanced biphasic, first phase cathodic, with a 100 us pulse width. Bladder stimulation amplitudes (7 ± 3 Vpp; range 3 8 Vpp) were supramaximal and adjusted during the experiment if necessary. Bladder stimulus duration was either 60 or 90 seconds, depending on the flow rate. EUS Activation:

97 83 Continuous stimulation (EUS stimulation) of the right S1 sacral root using the proximal bipolar electrode cuff generated external urethral sphincter pressure. This approach mimicked severely dyssynergic sphincter activity in a preparation lacking continence reflexes. EUS stimulation was 20 Hz, balanced biphasic with a 100 us pulse width and was coincident with bladder stimulation. EUS stimulation amplitude was chosen to minimize voiding during bladder stimulation. The EUS stimulation amplitude (3 ± 1 Vpp; range 2 4 Vpp) was increased during the experiment if necessary. HF Nerve Block: Biphasic square wave stimulation (EUS HF) was applied to the distal tripolar nerve cuff electrode implanted on the right S1 sacral root. EUS HF prevented transmission of nerve impulses evoked by proximal S1 stimulation. A stimulation frequency of 12.5 khz was selected based on prior results (Bhadra et al., 2006a; Boger et al., 2009) to minimize EUS HF onset response. During the experiment the EUS HF amplitude was adjusted to improve voiding while remaining below 20 Vpp (15 ± 3 Vpp; range Vpp with a single trial conducted at 24 Vpp). IV.3.2 Trials Four types of voiding trials were conducted in this study: EUS HF trials, EUS HF Max trials, Voiding control trials and EUS control trials (Figure 2). EUS HF trials quantified voiding using HF block of proximally evoked EUS activity. EUS HF Max trials attempted to maximize bladder voiding to evaluate the clinical utility of sacral block for voiding. Voiding control and EUS control trials provided upper and lower bounds on bladder voiding efficiency, allowing evaluation of the effectiveness of EUS HF trials.

98 84 EUS HF trials included bladder stimulation, proximal EUS stimulation, and distal EUS HF (Figure 1). Distal EUS HF stimulation blocked impulses evoked by excitatory proximal EUS stimulation and allowed bladder stimulation to generate post-stimulus voiding. EUS HF Max trials were either of longer duration (up to 50% longer) than standard EUS HF trials or included three summated successive trials (inter-trial interval: 69 ± 6 s), the first of which was also analyzed independently as a EUS HF trial. In both EUS HF and EUS HF Max trials, EUS HF stimulation began 10 seconds before either EUS or bladder stimulation to allow potential EUS HF onset responses time to subside. Voiding control trials demonstrated the maximum bladder voiding achievable in the preparation and included only bladder stimulation (no EUS or HF stimulation). An inability to achieve significant voiding in Voiding control trials indicated either reflex activation of the EUS, urethral obstruction or inadequate evoked bladder pressures. In contrast, efficient voiding in voiding control trials combined with poor voiding in EUS HF trials indicated either incomplete sacral root EUS block or HF EUS excitation. EUS control trials included bladder and EUS stimulation and showed that evoked sphincter pressures were sufficient to prevent voiding. Efficient voiding in HF EUS trials combined with poor voiding in EUS control trials indicated effective sacral HF block. Significant voiding during EUS control trials indicated that EUS stimulation evoked minimal sphincter pressure, rendering block completeness irrelevant. HF validation trials measured onset response magnitude and block completeness and were conducted in five experiments. Trials used either the same stimulation pattern as EUS HF trials or proximal EUS stimulation and distal EUS block.

99 85 IV.3.3 Data Extraction and Analysis The following variables were extracted from voiding trials (Figure 2): volume voided (V void ), residual volume (V residual ), maximum (Pv max ) and average (Pv avg ) vesicular pressure, maximum (Q max ) and average (Q avg ) flow rate and time to maximum flow rate (TQ max ) (Figure 2). The initial bladder volume (V initial ) was calculated as the sum of the residual and the voided volumes. The bladder was drained and V residual measured following trials with significant voiding. In all other trials, V residual was estimated based on subsequent V void and the V residual when next directly measured. The voiding efficiency (V eff ) was the ratio of V void and V initial. To extract flow data, the each voiding trial was divided into one second (1 s) bins. The end of the first bin (bin zero) was aligned with beginning of bladder stimulation. The average volume voided was found for each increment and the flow Q(i) for the ith bin was defined as V void (i+1) minus V void (i). Only V initial, V residual, and V void were calculated for EUS HF Max trials. If the EUS HF Max trial included successive trials, V initial was the initial bladder volume of the first trial and V residual was the residual bladder volume of the final trial. V void was then the total volume voided over the set of three trials. The minimum EUS pressure during HF stimulation (Pu min ) and time until the EUS pressure decreased to within 10 cmh2o of Pu min (TPu min ) were extracted from the HF validation trials (Figure 3). The EUS pressure was filtered in Matlab using an FIR filter with a cut-off frequency of 5 Hz prior to feature extraction. Data were analyzed using Minitab (Minitab Inc, State College, PA). An analysis of variance was performed for each extracted feature with the control variables animal

100 86 and trial type (Voiding Control, EUS control and EUS HF) as the fixed effects. No interactions were assumed. TQ max was compared for EUS HF trial and Voiding Control trials only, due to limited voiding during EUS control trials, Post hoc comparisons were conducted by trial type. A Student s t-test was used to compare V eff for EUS control and EUF HF Max trials. IV.4 Results Voiding during EUS HF trials was achieved in three of six animals (Table 1A). Inadequate evoked bladder pressure, urethral obstruction, and incomplete or poor HF block prevented voiding in the remaining animals. The difficulties achieving voiding in these animals are described in the discussion. IV.4.1 Voiding Forty voiding trials were conducted in three animals (Table 1B). Single EUS HF trials voided 59 ± 12% of the initial bladder volume, significantly more (P < 0.001, n = 38, Figure 4) than EUS control trials without block (Veff: 3 ± 6%). Single EUS HF trials voiding efficiency was comparable (P = 0.92, n = 38) to Voiding control trials (V eff : 61 ± 15%). V eff depended on trial type (P < 0.001, n = 38) but not the animal (P = 0.67, n = 38). Complete voiding was achieved in EUS HF Max trials (V eff : 82 ± 9%). Both EUS HF Max and Voiding control trials voided significantly more than EUS control trials (P < 0.001, n = 38 for Voiding control comparison and n = 20 for EUS HF Max comparison). Voiding control and EUS HF trials voided similar amounts (P = 0.98, n = 38) and both trial types voided more than EUS control trials (P < 0.001, n = 38 for both comparisons, Figure 5). Residual bladder volumes were significantly lower for both Voiding control and EUS HF trials than EUS control trials (P < 0.01, n = 38 for both

101 87 comparisons, Figure 5). Voided volumes did not depend on the animal (P = 0.16, n = 38) but residual bladder volumes did depend on the animal (P < 0.01, n = 38). Maximum evoked bladder pressures did not differ significantly between trial types (P = 0.67, n = 38, Figure 5). However, Pv avg was significantly greater for EUS control trials than Voiding control trials or EUS HF trials (P < 0.001, n = 38, both comparisons). Pv avg did not differ significantly between Voiding control trials and EUS HF trials (P = 0.22, n = 38). Both maximum and average evoked bladder pressures depended on the animal (P < 0.001, n = 38). The maximum and average flow rates were greatest for EUS HF trials and voiding control trials (Figure 5). Flow rates in these trials did not differ significantly (Q max : P = 0.43, Q avg : P = 1.0) and were greater than the maximum or average flow rates for EUS control trials (P < 0.001, n = 38 all comparisons). Both Q max and Q avg depended on the animal (P < 0.001, n = 38 for all factors). The time between the initiation of bladder stimulation and the maximization of the voiding flow rate depended significantly on trial type and animal (P < 0.001, n = 38) and was greater in EUS HF trials (33 ± 20 s) than in Voiding control trials (18 ± 10 s, P < 0.001, n = 38). TQ max was highly variable for EUS control trials (27 ± 30 s). Few EUS control trials demonstrated voiding. The voiding that did occur occurred as either in a single pulse at the initiation of bladder stimulation or at the very end of the trial. IV.4.2 EUS Block Quality Block onset response was examined in five animals (Figure 3). Both Pu min (22 ± 21 cmh 2 O, range 0 80 cmh 2 O) and TPu min (12 ± 14 s, range 3 43 s) varied between animals. A high residual tone during HF stimulation appeared to prevent voiding. Pu min

102 88 was minimal in three animals (mean: 3 cmh 2 O), of which two voided, but was high in the remaining two animals (18 and 80 cmh 2 O), which did not void. A faster onset response was associated with a faster time to maximum flow rate. IV.5 Discussion This study demonstrates that HF waveforms applied to the sacral roots can block efferent EUS drive and prevent EUS activation and allow low pressure bladder voiding. Low pressure voiding was achieved even in the presence of continuous EUS stimulation mimicking severe dyssynergia. HF stimulation allowed voiding equivalent to unobstructed post stimulus voiding. In EUS HF Max trials (extended duration or summated successive trials) the voiding efficiency was greater than 80% and residuals averaged less than 6 ml, comparable to maximal voiding obtained in other voiding studies conducted in this model (Boger et al., 2008; Boggs et al., 2006; Tai et al., 2007a; Tai et al., 2007b; Yoo et al., 2008). IV.5.1 Sacral EUS Block for Voiding This study is the first to demonstrate bladder voiding using complete sacral root conduction block. Previous canine studies by investigators in Montreal have generated complete bladder voiding and significant reductions in urethral pressure using lower frequency (600 Hz) sacral stimulation (Abdel-Gawad et al., 2001; Boyer et al., 2000; Shaker et al., 1998). However a neural fatigue mechanism for block has been suggested for the Sawan waveform (Kilgore and Bhadra, 2004). EUS fatigue from high rates of stimulation (Thuroff et al., 1982) seems a likely cause of the reduction of urethral pressure.

103 89 Clinical translation of the Sawan waveform may not be practicable. Urethral continence reflexes, sufficient to prevent voiding in other canine studies (Bhadra et al., 2006b; Grunewald et al., 1998), appear to have been absent in the Montreal studies. EUS block of multiple sacral roots was not required to prevent reflexive EUS activation. Instead, a single electrode was used for EUS fatigue/block. A human model with intact sacral reflexes will require multiple cuff electrodes to block reflexive EUS activation. Each electrode will contribute to an overall residual EUS pressure during stimulation. In the Montreal studies, the residual EUS pressure from the single electrode (44 ± 7.3 cmh 2 O) was more than half the evoked bladder pressure (73.5 ± 20 cmh 2 O). Residual EUS pressures during stimulation may be greater in humans than dogs, as a greater proportion of human EUS fibers are fatigue resistant (Bazeed et al., 1982; Schroder and Reske-Nielsen, 1983). It is likely that residual EUS pressures during sub-kilohertz stimulation of multiple cuff electrodes may prevent bladder voiding. IV.5.2 Sacral Block Extends PN Block Sacral HF block provides an alternative to PN HF conduction block methods that can also produce voiding (Boger et al., 2008; Tai et al., 2007b). Sacral HF block allows a potentially shorter pathway to clinical translation than PN block neuroprostheses. Existing implants could allow clinical testing of HF waveforms with human-approved electrodes. The spinal canal also provides a stable and protected location electrodes and leads (Rijkhoff et al., 1997). This may be important as a change in the relative position of nerve and cuff could result in incomplete block or a worsening of the HF onset response (Bhadra and Kilgore, 2005; Kilgore and Bhadra, 2004),

104 90 Sacral HF stimulation would block non-pn mediated continence mechanisms. Urethral pressure can depend on both the EUS, innervated through the pudendal nerve, and the pelvic floor musculature, innervated directly from the S2 and S3 sacral nerve roots (Juenemann et al., 1988; Junemann et al., 1987). Patients for whom the pelvic floor contribution to urethral pressure is pronounced may benefit more from sacral HF block than PN HF block. IV.5.3 Voiding Limitations HF stimulation onset responses reduced voiding efficiency and extended the duration of stimulation. Variations in the HF EUS onset likely drove the variability in TQmax. Onset responses appeared sufficiently sustained to delay effective voiding until the onset response had diminished in two animals. Reducing block onset would allow the maximum flow rate to be achieved earlier in the trial and likely reduce the necessary duration of voiding, allowing for shorter trials. Voiding during EUS HF trials was time limited in animal I. Extending the EUS HF trial duration in this animal 50% increased the voided volume 37%. Increasing the duration of stimulation in this animal would have improved bladder-voiding efficiency. IV.5.4 Non-Voiding Animals Voiding was achieved in only three of six animals used in this study. These results reflect the inherent difficulty of the acute preparation. A functional bladder, unobstructed urethra, and complete EUS HF block were required to achieve bladder voiding. Failure of one element of the preparation was often sufficient to prevent voiding. An unstable, low-capacity, minimally-contractile bladder prevented voiding in one animal. EUS block was excellent (Pu min : 2 ± 3 cmh 2 O, TPu min < 4 s) and voiding

105 91 occurred during bladder control trials. Voiding comparable to voiding during bladder control trials occurred in a single EUS HF trial, but was not repeated. Given the minimal evoked bladder pressure (< 20 cmh 2 O), slight elevations in urethral pressure during HF stimulation may have been enough to prevent voiding in this animal. Urethral obstruction prevented voiding in a second animal. No voiding occurred during initial bladder control trials. Voiding efficiency increased over the course of the experiment, but remained poor and was not improved by transection of the left S1 nerve root and bilateral transection of the S2 nerve roots (proximal to the nerve cuff electrodes stimulating the bladder). Block was also poor in this animal; residual EUS pressures were high (Pu min : 18 ± 3 cmh 2 O). Poor HF EUS block prevented voiding the remaining non-voiding animal. Proximal stimulation and distal block trials were attempted in this animal using 12.5 and 25 khz stimulation at voltages ranging from 16 to 30 Vpp. Residual pressure during stimulation remained high (Pu min : 80 cmh 2 O). Activation of the EUS through (unblocked) adjacent spinal roots, or wrap-around activation of the blocked nerve distal to the blocking electrode location may have contributed to elevated EUS pressures during HF stimulation. The HF cuff electrode may also have incompletely encircled the nerve, leaving a section of the nerve unblocked or stimulated. IV.6 Conclusion Sacral EUS block was achieved in four of six animals and allowed low pressure bladder voiding with voiding efficiencies in excess of 80% possible. Bladder voiding neuroprotheses utilizing sacral EUS block may be compatible with existing implants and

106 92 technology, allowing more rapid clinical development of nerve-sparing bladder voiding neuroprostheses. Acknowledgements This work was supported by NIH HD40298, AR07505, EB002091, EB004314, DK077089; the State of Ohio BRTT03-10; the Department of Veteran Affairs VA-RR&D B3675R; and the Cleveland VA FES Center. The authors wish to thank Jaime McCoin, Timothy Bruns, Tina Goetz, and Timothy Mariano for their experimental assistance.

107 Figure 1: Electrodes locations. Right proximal S1 spinal root: Continuous EUS stimulation (EUS stim) mimics dyssynergia. Right distal S1 spinal root: HF stimulation (EUS HF) blocks nerve impulses evoked by EUS stimulation. Bilateral S2 spinal root: Intermittent (2 s on / 4 s off) stimulation generates bladder pressure (bladder stim) (figure courtesy of Dr Bhadra). 93

108 Figure 2: Sacral HF stimulation blocked external urethral sphincter nerve impulse transmission and allowed voiding. Intermittent bladder stimulation evoked bladder and EUS pressure, preventing voiding. Voiding occurs in the intervals between bladder stimulation. Top panel: Maximum and average bladder pressures (Pv max, Pv avg ). were extracted over the duration of bladder stimulation. Middle panel: EUS HF Stimulation waveforms began 10 seconds before bladder Stimulation and EUS stimulation. Bladder stimulation is intermittent (2 s on 4 s off). All stimuli cease simultaneously. Bottom panel: Average and maximum flow rates (Q Max, Q Avg ) and the volume voided (V Void ) were extracted over the bladder stimulation interval. The time to the maximum flow rate (TQ max ) was also extracted. 94

109 Figure 3: Sacral HF stimulation prevented external urethral sphincter (EUS) contraction. Stimulation parameters and timing are the same as in voiding trials. The bladder was drained and a microtransducer catheter was placed in the urethra at the location of the EUS. EUS HF trial (Thick Trace): HF stimulation was applied to the distal S1 tripolar cuff (0-70 s) and evoked an onset response over the first 10 seconds of stimulation. Bilateral S2 stimulation (intermittent 2 seconds on / 4 seconds off) and continuous proximal S1 EUS stimulation were applied from seconds. Distal HF S1 nerve conduction block prevented continuous S1 stimulation, but not intermittent S2 stimulation, from evoking EUS pressure. EUS control trial (Thin, Interrupted Trace): Continuous proximal stimulation and intermittent bilateral bladder stimulation applied (10 70 s). No HF stimulation applied. Continuous proximal S1 stimulation evoked sphincter pressure in the absence of distal HF nerve conduction block. Pu min was the minimum EUS pressure during the interval of stimulation. TPu min was the time from the initiation of HF stimulation until the EUS pressure was within 10 cmh 2 O of Pu min. 95

110 Figure 4: Sacral HF Block allowed voiding. EUS HF Max trials achieved effective voiding. Voiding efficiency was significantly greater for EUS HF trials than for EUS control trials (P < 0.01). EUS HF Max trials consisted of summated successive EUS HF trials or extended duration EUS HF trials and also voided more than EUS control trials (P < 0.01). 96

111 Figure 5: High Frequency Stimulation (EUS HF) trials and trials with unobstructed poststimulus voiding (Voiding control) exhibit similar urodynamics. A) Total volume voided and residual volumes are similar for EUS HF and Voiding control trials. Voided Volumes are significantly lower for EUS control trials. B) Maximum evoked bladder pressures are similar for EUS HF, Voiding control and EUS control trials, but average bladder pressures are significantly greater for EUS control trials than EUS HF or Voiding control trials. C) Maximum and average flow rates are similar for EUS HF and Voiding control trials. Peak and average flow rates are significantly lower for EUS control trials. 97

112 98 Table 1A: EUS HF Stimulation allowed voiding in three animals. Block was measured in five animals and was complete (Pu min < 10 cmh 2 O) in three animals and rapid (TPu min < 10 s) in four animals. The maximum evoked bladder pressure exceeded 20 cmh2o in five animals. Minimal voiding was achieved in voiding control trials in animal 5, even following surgical transection of the sacral roots (proximal to the stimulating electrodes). Table 1B: Forty trials were conducted in three animals. Nine EUS HF trials in animals 2 and 3 were also the first of three successive trials in a EUS HF Max trial. Two extended duration HF EUS Max trials were conducted in animal 1.

113 99 CHAPTER V: CONCLUSION

114 100 V.1 Summary of Results This dissertation shows that high frequency (HF) electrical block can replace the neurotomies used in two methods of generating voiding: sacral stimulation combined with bilateral pudendal nerve (PN) neurotomy (Chapter II) and the Brindley bladder control system (Chapter IV). HF PN stimulation was used to directly replace the PN neurotomy. The Brindley bladder system requires an afferent rhizotomy: extradural HF stimulation blocked sacral efferents innervating the external urethral sphincter (EUS), indirectly eliminating the need for the rhizotomy. Chapter II: Specific Aim 1 extends previous demonstrations of HF nerve block at the level of the pudendal nerve (Bhadra et al., 2006a) to demonstrate that 1) the application of HF block improved bladder voiding and 2) voiding during block was equivalent to voiding following PN neurotomy. Bilateral PN block was achieved with a relatively lower stimulation frequency of 2 khz and voltage parameters were consistent across animals. Bladder voiding efficiency in trials including PN HF stimulation was comparable to voiding efficiency in trials following pudendal neurotomy though bladder pressures in PN block trials were elevated suggesting that block was not complete. These results validate development of a HF block-based bladder voiding neuroprostheses. By preserving sacral reflexes and sensation, such an approach would expand the population of patients willing to consider bladder-voiding neuroprostheses. Chapter III: Specific Aim 2 demonstrates that 1) HF stimulation rapidly and reversibly reduced evoked bladder pressure, 2) transmission block characteristics and HF evoked bladder responses depended on stimulation parameters and 3) HF stimulation generated a local conduction block. Sacral root HF block of small parasympathetic

115 101 preganglionic bladder fibers (1-3 µm diameter)(morgan, 2001) was achieved. Block of such small fibers had not been demonstrated previously and suggests novel opportunities for chronic pain and autonomic disorder neuroprostheses. The demonstration of small fiber HF block supports research into HF block neuroprostheses for chronic pain or autonomic disorders. Such neuroprostheses could benefit an extremely large patient population currently lacking many treatment options. Chapter IV: Specific Aim 3 demonstrates 1) sacral HF improves voiding during sacral bladder stimulation and 2) sacral HFAC stimulation allows voiding comparable to voiding unobstructed by EUS closure. Intermittent bilateral sacral root stimulation (20 Hz) evoked bladder and sphincter pressure (bladder drive). Reflexes were absent in this preparation and post-stimulus voiding was unobstructed by EUS closure. Bladder voiding efficiency and urodynamics in trials including HF block and unobstructed voiding trials were not significantly different. Successive or long duration HF block trials generated clinically effective voiding. The demonstration of effective sacral HF stimulation for bladder voiding further enhances our impression of HF block as a powerful and flexible tool for nervous system neuroprostheses. This sacral HF approach may be able to achieve voiding in a human model using extradural Vocare or Brindley book electrodes, speeding development of clinical bladder-voiding neuroprostheses. V.2 Neuroprosthesis Design Choices V.2.1 Block Location Two nerve sparing, bladder voiding neuroprosthetic schemes are described in this thesis. Both neuroprosthetic approaches use sacral nerve root stimulation to evoke

116 102 bladder pressure and HF stimulation to prevent EUS activation. The methods differ in the location of HF block: PN versus sacral extradural root. The significance of advantages and disadvantages associated with each block location has yet to be determined. Clinical translation of both methods should be pursued. Advantages of Pudendal Nerve Block PN block requires fewer blocking electrodes than sacral root block. Sphincter motor neurons project to the EUS through the S2 S4 sacral roots (DeLancey et al., 2002). Sacral HF block must include the bilateral S2- S4 roots (6 roots), while PN HF stimulation need only include the bilateral pudendal nerves (2 roots). With sacral HF stimulation, all six roots must block cleanly: recurrent firing or a large onset response in any root could prevent voiding. Requiring six block electrodes also increases the risk of electrode or lead failure. HF PN block allows continuous sacral bladder stimulation. Voiding during continuous bladder stimulation occurs at maximal bladder pressure. Flow rates are high and the duration of voiding is short. Neuroprostheses including HF sacral root block must use intermittent stimulation to generate post stimulus voiding. Intermittent stimulation results in higher peak bladder pressures (due to outlet obstruction during sacral stimulation), lower average flow rates, and longer duration of voiding than continuous stimulation. Longer durations voiding durations could result in bladder fatigue during voiding leading to lower evoked bladder pressures and less efficient bladder voiding. Intermittent stimulation is also potentially more limited by high residual EUS pressures (incomplete block) than continuous stimulation. Voiding with intermittent stimulation occurs post stimulus, while the bladder pressure is rapidly dropping. Voiding

117 103 ceases once the bladder pressure falls below the elevated sphincter pressure. Rapidly decreasing post-stimulus bladder pressures, or evoked bladder pressures only slightly in excess of residual sphincter pressures, may prevent voiding. Advantages of Sacral Root Block Sacral HF block may have an advantage in clinical translation. Should chronic animal studies demonstrate safety and efficacy, human studies could recruit patients with existing Brindley stimulators, shortening the path to clinical implementation. Sacral HF stimulation may block non-pudendal nerve mediated continence mechanisms and allow voiding in individuals for whom PN block in insufficient. Stimulation of the S3 roots has been demonstrated to evoke urethral pressures, even following transection of the pudendal nerve (Junemann et al., 1987). Sacral stimulation intended to evoke bladder pressure could unintentionally prevent voiding in this manner. Sacral HF stimulation would block all somatic efferents exiting the sacral roots, allowing voiding. Afferent Approaches HF block of small parasympathetic efferents (chapter III) suggests that similar sized afferents fibers mediating DSD could also be blocked. The Brindley system could be implemented with HF afferent block replacing the dorsal rhizotomy. Peripheral block of urethral afferents may also be possible. The studies in this dissertation choose efferent approaches for mechanistic reasons: efferent block was previously demonstrated, efferent activation and block are easily transduced, and afferent reflexes are rarely present in our acute animal model.

118 104 V.2.6 Block Parameters The choice of stimulation frequency used in a HF block neuroprostheses represents a trade-off between onset response, stimulation amplitude and electrode design. HF onset response has been shown to decrease with increasing stimulation frequency (Bhadra and Kilgore, 2005; Gaunt and Prochazka, 2008; Tai et al., 2005c), though this has not been seen in all studies (Bhadra et al., 2006a; Kilgore and Bhadra, 2004). This may be due to an increase in spatial selectivity due to decreasing pulse width, as seen in single pulse twitch studies (Grill and Mortimer, 1996). Simulation studies suggest individual neuron onset response decreases as HF block increases beyond the block amplitude threshold (Bhadra et al., 2007). Minimizing onset response would suggest a choice of high stimulation frequency and amplitude. However, higher stimulation frequencies require higher stimulation amplitudes to achieve block (Bhadra et al., 2006a; Bhadra and Kilgore, 2005; Bhadra et al., 2007; Kilgore and Bhadra, 2004; Tai et al., 2005b; Tai et al., 2004). High amplitude and frequency of stimulation is more likely to damage the nerve and the electrode (McCreery et al., 1995) and reduce the battery life of the implant. Two kilohertz HF stimulation was used for PN block (chapter II) in an attempt to minimize the HF stimulation amplitude. Previous PN studies had shown no variation in onset for frequencies as low as 1 khz - so onset was not considered an adjustable parameter. Instead, the effect of onset response on bladder voiding was addressed by initiating HF stimulation 10 seconds before the bladder drive. Maximizing battery life and minimizing HF stimulation amplitude requires minimizing stimulation frequency, subject to the constraint of an acceptable onset

119 105 response. In the small fiber block studies, onset response improved the most from 2 khz to 6 khz, suggesting a frequency between 2 and 6 khz (likely closer to 6 khz) would be ideal. However, simulation studies suggest that charge per phase increases rapidly for stimulation frequencies below 15 khz (Bhadra et al., 2007). A combination of high charge per phase and long pulse width may render lower stimulation frequencies more hazardous than higher stimulation frequencies (Merrill et al., 2005), even though block amplitude thresholds may be substantially lower. Lower stimulation frequencies may not block the nerve. Several lower bounds between nerve conduction block and neural fatigue have been reported, ranging from 600 Hz to 6 khz. Block at slightly higher frequencies often exhibits poor completeness and onset characteristics (Gaunt and Prochazka, 2008; Shaker et al., 1998; Tai et al., 2005c; Williamson and Andrews, 2005). These bounds depend on electrode type and may vary over time, even within the same animal (see Chapter III, also Gaunt and Prochazka, 2008). In the studies described in Appendix A and Chapter III, 2 khz stimulation appeared to cause neural fatigue rather than block. The HF evoked response did not appear to depend on stimulation amplitude. A complete block of proximally evoked pressures could not be achieved. The proximal stimulation evoked response was severely diminished for several seconds following the cessation of HF stimulation. Appendix A also describes an experiment in which noisy block was obtained at 6 khz for two electrodes and block below 25 khz was impossible for one electrode. Higher frequency stimulation (>6 khz) should be used when unconstrained by power consumption considerations or the potential for chronic damage (i.e. acute experiments with external stimulators). When pursuing sacral root block for voiding

120 106 (chapter IV), we choose to use 12.5 khz stimulation, even though block threshold amplitudes were high (8 24 Vpp) as a consequence. Controlled safety studies are required to reveal amplitude limits for HF stimulation. Implanted devices will require careful tuning using the intended electrodes in a chronic setting to find the best frequency and voltage for maximizing battery life while providing safe, minimal onset, complete HF nerve conduction block. V.3 Future work This goal of this research is the development of neuroprostheses for eventual human use. Clinical trials require human-approved stimulator/electrodes combinations capable of achieving effective block with safe HF stimulation parameters. Chronic animal studies of HF nerve block safety, stability and effectiveness must be conducted for both sacral and PN block neuroprostheses before human subject trials can begin. Additional acute studies investigating HF nocioceptor block and small fiber block should also be conducted. Neuroprostheses for chronic pain and autonomic nervous system disorders could potentially assist a patient population far exceeding the number of SCI patients. V.3.1 Chronic HF stimulation Safety and Stability Signs of nerve damage due to HF stimulation have not been evident in acute high frequency block studies (Bhadra et al., 2006a; Bhadra and Kilgore, 2005; Gaunt and Prochazka, 2008; Tai et al., 2007b). However, the chronic safety of HF stimulation has not been demonstrated. Chronic safety studies are required for development and translation of HF block based neuroprostheses.

121 107 These safety studies should have two aims 1) determining general waveform and duty-cycle limits for safe HF stimulation and 2) translating the HF block-based bladder voiding neuroprostheses described in this thesis towards clinical trials. V.3.2 What determines safe HF stimulation? Charge density, geometric charge density, pulse width, and stimulation frequency: parameters affecting stimulation safety at low frequencies are known if not completely understood (Agnew et al., 1999; McCreery et al., 1992; McCreery et al., 1995; Merrill et al., 2005; Mortimer et al., 1980; Scheiner et al., 1990; Shannon, 1992). Deriving similar relations for HF stimulation would greatly simplify HF neuroprostheses design and direct research towards applications for which HF block is best suited. For example, amplitude limits for safe somatic fiber HF stimulation could determine the viability of chronic pain neuroprostheses. Chronic pain neuroprotheses implanted on mixed nerves must not damage somatic fibers also contained in the nerve. Nocioceptive C-fibers have much higher stimulation thresholds than somatic motor neurons (Li and Bak, 1976). NEURON (Hines and Carnevale, 1997) models suggest that block amplitude threshold increases more quickly with frequency for smaller nerves than larger ones (Bhadra et al., 2007). HF block studies in NEURON, similar to Kilgore and Bhadra, 2004 can be used to estimate C-fiber block amplitude thresholds. Should HF stimulation damage somatic fibers at amplitudes below those expected to block C-fibers, committing resources to investigating C-fiber block may be unwarranted. The effect of prolonged HF stimulation on nerves must also be determined. The duration and duty cycle of HF stimulation varies greatly between potential applications. Bowel and bladder control require very low durations of application: potentially a few

122 108 minutes or less several times a day. In contrast, chronic pain applications or autonomic block applications might require continuous stimulation. Changes in intracellular ion concentrations (Stys, 2005) and metabolic demands during continuous HF stimulation (Tykocinski et al., 1995) may affect neurons independent of the choice of HF stimulation parameters. Chronic rat studies could determine general waveform and duty-cycle limits for safe HF stimulation. Following McCreery et al., stimulation and recording electrodes could be implanted on the sciatic nerve (McCreery et al., 1992). Recruitment curves would be determined and the sciatic nerve stimulated at either 12.5 khz or 25 khz using spiral nerve cuff electrodes. Each sciatic nerve would be stimulated at a single combination of stimulus amplitude and frequency. HF stimulation amplitudes would be determined as multiples of the supramaximal tetanic stimulation threshold. Stimulation would begin three-weeks post surgery. Following one week of stimulation, the rat would be sacrificed and the sciatic nerves examined for signs of axonal degeneration. Rats would also be observed for loss of function following stimulation. V.3.3 Translation of Acute Studies Chronic spinalized studies in cat or dog models could translate the HF blockbased bladder voiding neuroprostheses described in this thesis towards clinical trials. These studies would investigate stability of block characteristics and chronic safety of HF stimulation at parameters determined successful in acute studies. The HF blocked bladder voiding neuroprostheses could be used for daily bladder management. Block characteristics could be negatively affected by electrode migration or encapsulation following implantation. Changes in stimulation thresholds (Grill and

123 109 Mortimer, 1994; Grill and Mortimer, 1998) during encapsulation could result in incomplete block, worsened block onset, failure to block, or even nerve activation during HF stimulation. Electrode impedance and block thresholds would be tested several times a week until cuff impedance reached its asymptotic value, approximately 60 days (Grill and Mortimer, 1994; Malek and Mark, 1989; Stein et al., 1978). EUS pressures would be recorded prior to and following HF stimulation to evaluate any functional deficits due to electrical stimulation. Histological testing of the implanted nerve roots would be used to assess nerve damage. V.3.4 Human Clinical Trials Successful completion of chronic animal safety trials would allow for human clinical trials. Initial trials would attempt to demonstrate EUS block at either the sacral or PN levels. HF sacral EUS block testing could utilize patients receiving new Finetech bladder voiding implants or reoperations to replace leads or electrodes. PN trials could utilize patients receiving pudendal nerve decompression surgeries for (Popeney et al., 2007). Trials would attempt to demonstrate sacral HF block of the EUS as measured by a urethral microtransducer catheter. Stimulation waveforms and electrodes would be those successful in chronic animal studies. V.3.5 Nocioceptive and Autonomic Nerve Block HF block of chronic nocioceptive or peripheral neuropathic pain would greatly expand the patient population benefiting from HF block neuroprostheses. HF nocioceptive block could be investigated in an acute cat model using C-fiber dependent withdrawal reflexes. These reflexes have been extensively characterized (Clarke et al., 1989; Iwamoto et al., 1978; Wall and Woolf, 1984) and studies could be based on a

124 110 familiar preparation (Gustafson et al., 2006). Successful, safe block of C-fibers would dramatically increase the scope of HF block neuroprostheses. Potential autonomic block applications include disorders such as obesity and hyperhidrosis. Bilateral HF stimulation of the vagal nerve at 5 khz has been used to combat obesity. While this may not be true block, as a reduction in vagal nerve compound action potentials and pancreatic exocrine secretions persisted for several minutes following cessation of stimulation, it is effective in generating weight loss (Camilleri et al., 2008). Better electrode design and choice of stimulation waveforms could potential improve performance. Sympathetic nerve block for hyperhidrosis (excessive sweating) would seem an unlikely candidate for an invasive surgical intervention. However, over 1 million Americans suffer from severe hyperhidrosis that interferes with activities of daily living (Strutton et al., 2004). Treatments include thoracoscopic sympathectomy, which suffers from potentially complications (Connolly and de Berker, 2003), which can cause patients to regret the procedure (Adar, 1998). A reversible nerve-sparing alternative would be extremely valuable. V.4 Conclusion Two novel nerve-sparing, bladder-voiding neuroprosthetic procedures were demonstrated in these studies. By using HF block to prevent activation of the EUS during voiding, they preserved sacral reflexes, eliminating a major barrier to patient acceptance of bladder-voiding neuroprostheses. Nerve impulse conduction block in small parasympathetic bladder efferents was also demonstrated. Bladder efferent block suggests

125 111 that HF stimulation may have a role in the treatment of chronic pain and autonomic disorders. Further research into small fiber block is warranted. This work demonstrates the enormous potential of high frequency stimulation as a tool for interfacing with the nervous system. HF stimulation can be used to interrupt reflexes and reversibly denervate target muscles or organs. It appears safe in an acute preparation, can block nerve impulse conduction in myelinated fibers with diameters below 5 µm, and works both peripherally and within the spinal canal. Many nervous system disorders arise from abnormal nervous system activity. Immediate, reversible HF nerve conduction block has great therapeutic potential.

126 112 APPENDIX A: HIGH FREQUENCY PUDENDAL NERVE BLOCK PARAMETERS FOR HUMAN-APPROVED ELECTRODES Getting human approval for new HF block electrode will be prohibitively expensive. An existing human-approved electrode must be selected for clinical translation of High frequency (HF) PN block bladder voiding neuroprosthesis. This electrode must achieve pudendal nerve block at frequencies and amplitudes compatible with human-approved implantable stimulators. Onset response must be minimal and the frequency and amplitude threshold characteristics for block electrode must be stable in a chronic implant. The safety of the intended HF block waveform and electrode combination will be verified prior to human clinical trials. Safety testing every combination of human approved stimulation electrode and HF waveform is expensive and unnecessary. By characterizing block parameters in an acute preparation, the number of potential electrode / waveform pairs can be reduced to a manageable number. This appendix describes acute PN HF block parameter testing of the Case Western spiral electrode, the flat interface nerve electrode (FINE), the Huntington spiral electrode, and the Peterson intramuscular electrode. Successful stimulation parameters were recorded and will assist selection of a commercially available stimulator for future clinical studies. A.1 Methods HF electrical block was tested in a feline pudendal nerve model. The PN was exposed bilaterally using a postero-lateral gluteal approach and a proximal electrode cuff was implanted proximal. A microtransducer catheter (C7C4F, MMI-Gaeltech,

127 113 Hackensack, NJ) was placed in the urethra at the level of the external urethral sphincter. The human approved electrodes were implanted distally and were tested in the following order: Spiral, Fine, Huntington, IM electrode, J-cuff electrode. Proximal stimulation (20 Hz, 100 µs PW, balanced biphasic) began 10 seconds before the HF stimulation and continued for 40 seconds. HF stimulation was a voltage-controlled square wave of 20 seconds duration with frequency 2, 6, 12.5, or 25 khz and was applied to the distal HF electrode. A.2 Results Complete block could not be achieved with every electrode (Table I). The Huntington spiral electrode generated excellent block and required low stimulation amplitudes, comparing favorably to the values obtained using the molded cuff control electrode. Onset was minimal in all experiments. Poor block was achieved using the Case FINE electrode. Stimulus amplitudes required for block were high, onset response was great and the block was noisy. At stimulation parameters below 12.5 khz the animal s leg would contract rhythmically. The IM electrode achieved block using monopolar stimulation with a remote return electrode. Onset response in the sphincter was minimal. However, HF stimulation evoked violent ipsilateral leg tremors. These tremors appeared to depend on stimulation frequency and were so large that HF stimulation trials at frequencies below 25 khz could not be conducted. The Case spiral electrode could not achieve block: stimulation amplitudes appeared insufficient, though amplitudes as great as 12 V were tested.

128 114 For all electrodes, stimulation at 2 khz did not appear to generate block. The HF onset response to 2 khz stimulation was sustained and proximally evoked EUS pressures did not return immediately upon cessation of HF stimulation. The EUS 2 khz stimulation response showed little dependence on stimulus amplitude. The EUS response for higher stimulation frequencies, in contrast, showed a clear stimulus amplitude threshold for block. A.3 Discussion Electrode contact size influences electrode performance significantly. Small contacts on a big nerve may limit current spread to a portion of the cuffed nerve. This is often desirable, such as in applications using multiple small contacts to steer currents and selectively activate regions of a nerve (McNeal and Bowman, 1985; Sweeney et al., 1995; Tarler and Mortimer, 2004). In such a situation, safety constraints may limit stimulation amplitudes to levels insufficient to block all neurons. Unblocked neurons can fire repetitively during HF stimulation, contributing to the HF stimulation onset response. Such a mismatch between nerve diameter and the electrode contact size may explain why the Case spiral performed poorly. Larger, circumferential contacts, such those in Huntington electrode, may ensure even activation of the nerve. Small contacts may also become obstructed. Air bubbles can become trapped in the wells containing the contacts in the FINE, preventing current flow through the electrode. This problem was identified and resolved during, however prior to its resolution, the stimulator registered an open circuit when attempting to stimulate the FINE.

129 115 Design requirements for selective stimulation and HF block appear inconsistent. Cuff electrodes designed for selective stimulation were ineffective in achieving block, while electrode with large surface area and unconfined currents were consistently effective. Block was achieved using human approved electrodes at stimulation frequencies (>= 6 khz) and amplitudes (> 8 Vpp) similar to those causing block in previous acute experiments relying on custom-made electrodes. Continued experiments will allow a refinement of the parameter space and a selection of an appropriate election / stimulator combination for translation to clinical experiments.

130 Table I: The bipolar Huntington spiral electrode achieved the most effective conduction block with the lowest stimulation amplitude. Block was not achieved with the Case spiral electrode. The ability of the Peterson IM electrode to block nerve impulse transmission was limited and the Case FINE was largely ineffective. The custom made J cuff electrode, the nerve cuff electrode used in the HF block experiments, was included as a control. The J-cuff electrode demonstrated that block could be achieved in the preparation and provided a reference for the block parameters. *Block was also achieved at 12.5 khz and 8 Vpp for the J-cuff electrode. 116

131 117 APPENDIX B: WRAP AROUND ACTIVATION DURING HIGH FREQUENCY STIMULATION High frequency stimulation provides a local conduction block. Sufficiently large proximal stimulation may stimulate the nerve distal to the blocking cuff electrode. When studying high frequency block in trials using proximal stimulation, the proximal stimulation amplitude should not exceed the amplitude required to achieve full recruitment of the targeted end organ. B.1 High Amplitude Proximal Stimulation Prevented Block Bladder pressures were measured with a suprapubic catheter implanted in the dome of the bladder and connected to a pediatric blood pressure monitor (Deltran IV, Utah Medical Devices, UT). Tripolar spiral cuff electrodes were implanted on the intradural s2 ventral roots. Current controlled continuous 20 Hz proximal stimulation was provided (biphasic, 100 µs PW, stimulation amplitude, 3, 5 or 10 ma) for 40 seconds (Figure 1). Distal HF stimulation (12.5 khz, square wave, 10 ma) was provided for 20 seconds beginning 10 seconds after the initiation of proximal stimulation. The minimum pressure in the 5-second interval preceding the initiation proximal stimulation was extracted as Pv pre. The maximum evoked pressure (pressure in excess of Pv pre ) between the initiation of proximal stimulation and the initiation of distal HF block was extracted as Pv max. The minimum evoked pressure during the interval of HF stimulation was extracted as Pv min. The percentage of the proximal force blocked by the distal stimulation was calculated as: Bratio = (Pv max - Pv min )/ (Pv max )

132 118 A one-way ANOVA by proximal stimulation amplitude was used to analyze the data. Given a fixed level of distal HF stimulation sufficient to achieve block, the residual tone during block depends on the magnitude of the proximal stimulation. A slight increase in Pv min between 3 and 5 ma trials was observed, with a much larger increase between 5 and 10 ma trials. The HF stimulation did not appear to affect the bladder pressure during 10 ma proximal stimulation (Figure 1). The proximally evoked pressures depended on the proximal stimulation amplitude (P < 0.01). The proximally evoked pressures were not significantly different for 5 ma and 10 ma (45 ± 4 cmh 2 O, 50 ± 4 cmh 2 O, P > 0.05), but both were significantly greater than 3 ma (29 ± 4 cmh 2 O, P < 0.05 for both comparisons). Pv min also depended on proximal stimulation amplitude (P < 0.01). Each proximal stimulation amplitude evoked a different minimum level of stimulation (3mA: 4 ± 2 cmh 2 O; 5 ma: 12 ± 1 cmh 2 O; 10mA: 32 ± 1 cmh 2 O; P < 0.05 all comparisons). Differences in Bratio were significant (P < 0.01; Figure 2). Trials conducted with 3 ma proximal stimulation (n=2) had the greatest Bratio (87 ± 6 %), followed by trials conducted with 5 ma proximal stimulation (n = 3, 75 ± 3%). Trials including 10 ma proximal stimulation demonstrated the lowest Bratio (n = 3, 28 ± 3%). Each level of proximal stimulation evoked a significantly different block ratio (P < 0.05). Fiber recruitment effects can explain some gradual variation in the residual pressure during block, but not all. Increasing the amplitude of submaximal proximal stimulation recruits progressively smaller diameter nerve fibers. A given level of HF stimulation may only block some of these newly recruited fibers. In such circumstances,

133 119 small increases may be observed in Pv max and Pv min. The increment in Pv max is the contribution of the newly activated fibers to the bladder pressure, while the increment in Pv min is the contribution of unblocked subset of those fibers. For obvious reasons the increment in Pv max should be less than the increment in Pv min. For example, increasing proximal stimulation from 3 ma to 5 ma yields a 21 cmh 2 O increase in Pv max and a 9 cmh 2 O increase in Pv min. Increasing the amplitude of supramaximal stimulation does not recruit additional fibers. Consequently, neither Pv max or Pv min should increase. However, increasing proximal stimulation in these trials from 5 to 10 ma resulted in a significant increase in Pv min and generated a bladder pressure contraction resembling unblocked proximal stimulation. The proximal stimulation appears to recruit the nerve distal to the blocking electrode, however the specific mechanism for this recruitment is not clear. The parasympathetic bladder efferent fibers blocked are 1 3 µm in diameter with an approximately 0.1 mm internodal length (Morgan, 2001). Therefore even a small separation between stimulating cuffs encompasses a large number of nodes of Ranvier. The axial resistance for such a small fibers would also be relatively high, reducing intracellular current flow. In addition, HF stimulation generates a broad region of depolarization under the stimulating electrode (Bhadra et al., 2007; Kilgore and Bhadra, 2004) that would prevent activation of the nerve near the distal block electrode. It seems that wrap-around activation depends on very high stimulation amplitudes and loosely fitting electrodes to allow sufficient current leakage to stimulate the nerve significantly distal to the region of block. One solution might be to reducing proximal stimulation

134 120 amplitude or replacing the proximal electrode with a more tightly fitting one. Another solution might be to loosen the HF electrode or turn up the HF stimulation amplitude in an attempt to enlarge the area of effect of block. However this risks worsening the HF onset response and stimulating nearby nerves.

135 121 Figure 1: Proximal stimulation can excite the nerve distal to the location of HF conduction block. Proximal stim: Continuous 20 Hz stimulation (amplitudes 5 ma and 10 ma shown) evokes bladder pressure. Proximal Stim begins 10 seconds before HF stimulation and continues for 40 seconds. Distal HF: Square wave, 12.5 khz with amplitude 10 ma. Duration of stimulation is 20 seconds. Distal stimulation begins 10 seconds after proximal stimulation. Pv max is the maximum evoked bladder pressure in the 10 seconds between the initiation of proximal stimulation and the initiation of HF stimulation. Pv min is the minimum evoked bladder pressure over the interval of HF stimulation. Block Ratio (Bratio) is defined as the difference in Pv max and Pv min normalized by Pv max.

136 Figure 2: Block ratio (Bratio) decreases with increasing proximal stimulation. The maximum evoked bladder pressure during proximal stimulation (Pv max ) and the minimum pressure evoked bladder pressure (Pv min ) during HF stimulation both increase when the proximal stimulation amplitude is raised from 3 to 5 ma. The increase in Pv min is proportionately greater, reducing Bratio slightly. Increasing the proximal stimulation amplitude from 5 to 10 ma does not increase Pv max, showing that 5 ma was maximally stimulating the nerve. However, Pv min increases significantly and Bratio is greatly reduced. 122

137 123 APPENDIX C: SACRAL NERVE ROOT IDENTIFICATION, LENGTH AND DIAMETER The experiments described in this dissertation use electrodes implanted on the sacral spinal roots to generate bladder pressures and block external urethral sphincter activation. The sacral root level was identified intra-operatively prior to cuff implantation and verified post-mortem. The lengths and diameters of the sacral nerve roots were recorded during the post-mortem verification. This appendix describes the process of nerve roots identification and relates the accessible lengths and diameters of the S1 and S2 nerve roots as measured post-mortem in 11 animals. C.1 Intra-operative Sacral Root Identification The sacral roots were identified intraoperatively with reference to the lumbosacral junction. A lumbosacral laminectomy removed the spinous process and lamina from the L7 vertebrae. In some animals L6 and caudal third of L5 were removed as well. In the sacrum, the laminectomy extended laterally almost to the medial border of the dorsal foramen. The ligamentum flavum was incised and removed to visualize the spinal canal. A layer of fatty tissue surrounded the spinal cord and the nerve roots. This layer was retracted to expose the cord. Four roots were often visible: L7 S3. The L7 sacral nerve roots were noticeably larger than the sacral nerve roots and had a relatively short extradural course, proceeding anteriolaterally almost immediately upon exiting the dura. Little of the S3 root beyond the DRG was typically visible within the laminectomy. The S1 and S2 roots traveled along the anteriolateral surface of the dura before proceeding anteriolaterally and exiting

138 124 via their respective sacral foramina. At the level of the lumbrosacral junction the S2 dorsal root ganglion (DRG) was visible. The S2 root exited the dura above the S1 vertebral level, traversing the S1 vertebral segment, and exiting the spinal canal through the S2 foramen. The S2 root was mobilized by sliding a glass hook rostrocaudally along the posteriolateral edge of the dura. Light, anteriomedially-directed pressure caused the glass hook to slide between the medial surface of the S2 root and the lateral surface of the spinal cord. C.2 Post-Operative Sacral Root Verification The S1 sacral root was verified with reference to the sacroiliac joint (Figure 1). An incision was made extending laterally across tuber sacrale from the edge of the lumbosacral laminectomy. A periosteal elevator was used to retract the posterior sacroiliac ligaments and expose the tuber sacrale from the iliac crest to the greater sciatic notch. The periosteal elevator was then used to expose the sacroiliac joint and the lateral aspects of the sacrum. The medial aspects of the sacrum were previously removed during the lumbosacral laminectomy. The dorsal S1 foramen was identified medial to the sacroiliac joint. The dorsal ramus of the S1 spinal nerve exited the spinal canal though the dorsal S1 foramen and traction applied to ramus allowed identification of the S1 root inside the spinal canal. The S2 foramen was identified caudal to the S1 foramen and the S2 root identified similarly to the S1 root. In some cases the lumbosacral laminectomy was enlarged to encompass the dorsal foramen, allowing visualization of the sacral roots as they exited the foramen.

139 125 In five animals the dura was retracted. The intradural S1 an S2 roots were mobilized on glass hooks and traced centrally from the point where they exited the dura. C.3 Sacral Nerve Root Lengths and Diameters The S1 and s2 nerve roots were of equal length (1.3 ± 0.3 cm for both S1, n = 8 and S2, n = 9). However, the S2 root had a greater diameter than the S1 root (S2: 1.5 ± 0.4 mm, S1: 1.1 ± 0.1 mm, n = 6). Intradural sacral roots were accessed in 5 animals. Total intradural root length was estimated (n = 3) and measured (n = 2) as greater than 20 mm.

140 126 Figure 1: Post-operative verification of correct implantation electrode cuffs. A: Incision across tuber sacrale. B: Tuber sacrale, exposed to the level of greater sciatic notch. C: The sacroiliac joint. D: The S1 foramen (medial and lateral borders removed to visualize S1 nerve), medial to the sacroiliac joint. E: The S2 foramen (posterior and lateral borders removed to visualize S2 nerve), caudal to the S1 foramen. The S1, S2 and L7 roots are labeled. The size difference between the sacral and lumbar roots allows for convenient identification of the S1 root.

141 127 APPENDIX D: EXPERIMENTAL SETUP This section contains diagrams of the nerve cuff electrodes (Figures 1 4) and experimental setups used in the animal experiments described in Chapters II (Figure 5), III (Figure 6), and IV (Figure 7). A schematic for an isolated, battery-powered pulsestretcher designed for these experiments is provided (Figure 8). Experimental equipment used and manufacturers information is also included (Table I). All electrodes were designed by Dr Narendra Bhadra and manufactured according to his specifications. Contacts were platinum and were contained between silicone sheeting. Following manufacture, the electrode was curled around a 1mm mandrel.

142 128 Figure 1: Nerve cuff electrode design used in for PN block and proximal PN stimulation (Chapter 2). The nerve cuff electrode was typically closed using suture. This electrode was designed by Dr. Narendra Bhadra.

143 129 Figure 2: Nerve cuff electrode design used in for extradural sacral small fiber block and sacral small fiber stimulation (Chapter 3). By interweaving the slots the electrode could be secured around the nerve. This electrode was designed by Dr. Narendra Bhadra.

144 130 Figure 3: Sacral tripolar nerve cuff electrode designs. Tabs could be inserted in slots on opposite side to close cuff. Suture could be passed through and tied to close cuff. Connected design allowed for fewer leads. These electrodes were designed by Dr. Narendra Bhadra.

145 131 Figure 4: Sacral bipolar nerve cuff electrode designs (Chapter 4). Tabs could be Suture could be passed through and tied to close cuff. These electrodes were designed by Dr. Narendra Bhadra.

146 132 Table I: Model and supplier information for equipment used in the studies described in this dissertation.

147 133 Figure 1: Electrical stimulation and recording setup used study of PN HF block for voiding (chapter II). The switching box allowed easy routing of signals between the chart recorder and the DAQ. The micrograbber array provided an interface between BNC connectors and electrode wires. The Bak boxes converted voltage to current controlled stimulation. A DS 8000 was used in one experiment to provide proximal sacral stimulation.

148 134 Figure 2: Electrical stimulation and recording setup used in study of sacral HF block of small parasympathetic bladder efferents (chapter III). The Digitimer and Wavetek stimulators were not used in these experiments. Nor was the Gould amplifier. Proximal stimulation and distal block were both generated by the DS8000. The connection box is the switching box from Figure 1. New PCI DAQ card allows desktop to replace laptop used previously.

149 135 Figure 3: Electrical Stimulation and recording setup used in studies of sacral HF stimulation for voiding (Chapter IV). A new electrode connection system replaced the micrograbber array. The input cat box bundled multiple stimulation signals into a single cable. The DS8000 was used for all stimulation in this aim. The other stimulators were not used. A