ReStore: a wireless peripheral nerve stimulation system

Abstract

Background:

The growing use of neuromodulation techniques to treat neurological disorders has motivated efforts to improve on the safety and reliability of implantable nerve stimulators.

New Method:

The present study describes the ReStore system, a miniature, implantable wireless nerve stimulator system that has no battery or leads and is constructed using commercial components and processes. The implant can be programmed wirelessly to deliver charge-balanced, biphasic current pulses of varying amplitudes, pulse widths, frequencies, and train durations. Here, we describe bench and in vivo testing to evaluate the operational performance and efficacy of nerve recruitment. Additionally, we also provide results from a large-animal chronic active stimulation study assessing the long-term biocompatibility of the device.

Results:

The results show that the system can reliably deliver accurate stimulation pulses through a range of different loads. Tests of nerve recruitment demonstrate that the implant can effectively activate peripheral nerves, even after accelerated aging and post-chronic implantation. Biocompatibility and hermeticity tests provide an initial indication that the implant will be safe for use in humans.

Comparison with Existing Method(s):

Most commercially available nerve stimulators include a battery and wire leads which often require subsequent surgeries to address failures in these components. Though miniaturized battery-less stimulators have been prototyped in academic labs, they are often constructed using custom components and processes that hinder clinical translation.

Conclusions:

The results from testing the performance and safety of the ReStore system establish its potential to advance the field of peripheral neuromodulation.

INTRODUCTION

Peripheral nerve stimulation has emerged as a means to treat a variety of neurological diseases including pain, sleep apnea, urinary incontinence, stroke, and tinnitus [1–11]. Given the broad application of the potential therapy, there is a great deal of interest in developing improved technologies to deliver nerve stimulation.

The majority of commercially available implantable pulse generators (IPGs) for peripheral nerve stimulation share similar design features with a cardiac pacemaker, including a metal enclosure to house the battery and circuit components, and wire leads which conduct stimulation current from the implant to the physiological target. While these traditional IPGs can deliver therapeutic efficacy for a range of conditions [8,9,12], the technology often necessitates subsequent surgical interventions due to lead breakage and battery failure [13,14]. Each successive surgery adds cost and puts the patient at risk of complications [13,14]. Removing the battery and leads from the IPG would eliminate these two common IPG failure modes. Multiple groups have demonstrated proofs of concept of such miniaturized battery-less IPGs in animal models and in humans using various power transfer modalities [15–27]. Despite this, only two of such devices, the Bioness StimRouter and the Stimwave Freedom, have been approved for use in humans by the FDA. This barrier to translation typically results from the overwhelming cost, time, and risk associated with commercialization. Prioritizing manufacturability and low cost in the design from the beginning may increase the likelihood of translation.

In the present study, we consider a number of design strategies with the intent of eliminating battery and lead failure modes and simultaneously facilitating commercial translation. First, we avoid using a battery in the IPG by instead transferring power wirelessly from a source external to the body. Second, we avoid using leads by integrating the electrodes into the body of the IPG, and by miniaturizing the IPG to fit into confined anatomical spaces without damaging the surrounding tissue or causing cosmetic issues for the patient. Third, we avoid using custom components including application specific integrated circuits. Custom components, by nature, carry more development risk than commercial off the shelf components, and they inflate development time and cost. Fourth, we manufacture the IPGs using a microelectro-mechanical systems (MEMS) wafer level process. Moving away from the serial manufacturing of traditional IPGs not only greatly reduces cost, but also improves process reliability [28,29].

Here, we present the design of our novel device for wireless nerve stimulation: The ReStore IPG. This solution obviates the battery and lead failures seen in traditional IPGs and was designed for manufacturability by using commercially available components and processes. In this study, we demonstrate the operational performance, reliability, biocompatibility, implantation safety, and nerve stimulation efficacy of this device.

DESIGN

System Architecture

The ReStore system consists of three major components: the implantable pulse generator, the relay module, and the programming module (Figure 1). The IPG is implanted directly on the nerve of interest and delivers electrical stimulation through electrodes exposed on its enclosure. The relay module remains outside the body and is positioned directly over the IPG during stimulation. The relay module is battery powered and provides wireless power to the IPG through inductive coupling at 13.56 MHz, a frequency that results in negligible absorption in the body [30–32]. Compared to other methods of wireless power transfer like ultrasonic and capacitive coupling, inductive power transfer is a more mature technology that offers better range and data rates and ideal for designs employing off-the-shelf components [37–39]. The relay module communicates with the implant via Near Field Communication (NFC). The programming module communicates with the relay module using Bluetooth Low Energy (BLE). It controls the IPG via the relay module to set stimulation parameters and trigger stimulation.

Figure 1:

Key components of the ReStore nerve stimulator system.

Implantable Pulse Generator

The ReStore IPG uses a Near Field Communication tag IC (NXP, Eindhoven, Netherlands, part number: NT3H2211) to handle wireless communication and power harvesting (Figure 2A). The power harvesting circuit outputs a maximum of 3 V which is used to power the rest of the device. A low power 8-bit microcontroller (MCU) (Silicon Labs, Austin, TX, part number: EFM8SB1) handles the application-level communication and has a digital to analog converter that is used to control the pulse amplitude. The 3 V output from the NFC IC is stepped-up to 12V by a boost converter (Linear Technology, Milpitas, CA, part number: LT8410) in order to drive stimulation current through high tissue impedances. The stimulation circuit consists of an op amp (Texas Instruments, Dallas, TX, part number: OPA170) that amplifies the analog output range of the MCU to 0 – 1200 μA. The output of the op amp is connected to the electrodes via two single pole double throw switches (Vishay, Malvern, PA, part number: DG9636). The MCU controls the switches to deliver pulses of desired duration and frequency. Charge balancing is achieved by changing the direction of flow of current by alternating the electrode through which the charge is delivered. At the end of the pulse, the electrodes are shorted together to discharge them completely. To be able to measure the load/tissue impedance, the output of the op amp is connected to the analog to digital converter in the MCU so that the voltage waveform of a pulse can be sampled and transmitted back to the programming module. The circuit schematic of the IPG is shown in Figure 3. The stimulation parameter ranges and their step sizes are given in Table 1. The maximum pulse amplitude of 1200 μA is a hardware safety limit, chosen for a specific application, but can be modified by changing the resistor R1 within the IPG during manufacturing.

Figure 2:

Implantable Pulse Generator. (A) Block diagram of the implantable pulse generator. View of the front (B) and back side (C) of the IPG. The electrodes are visible in C. (D) The silicone cuff with the IPG inserted. (E) IPG wafer

Figure 3:

Circuit schematic of the IPG

Table 1:

Stimulation parameter ranges

Parameter	Range	Step size
Amplitude	0 – 1200 μA	20 μA
Pulse Width	100 – 250 μs	10 μs
Frequency	10 – 30 Hz	1 Hz
Duration	0.5 – 25 s	0.1 s

The IPG is principally composed of a 4-layer PCB assembly. Rogers material is used for the PCB substrate to limit moisture absorption and outgassing. A 9-turn 3-layer coil, laid out as traces on the periphery of the PCB, acts as the antenna for power reception and communication. Surface mount device components are used throughout to minimize the device size. The assembled PCB is encapsulated in biocompatible glass to provide a hermetic seal (Figure 2B). The encapsulation is performed at the wafer level with 68 IPGs per 6-inch wafer (Figure 2E) followed by dicing to singulate the IPGs, and rounding of sharp edges. Stimulation current from the PCB is conducted to platinum electrodes on the outside of the glass using hermetic through glass vias. These planar electrodes on the surface of the device that contact the nerve measure 3 mm x 2 mm. The entire glass encapsulated IPG measures 13 mm x 8 mm x 3 mm. A cuff made of medical grade silicon holds the IPG in position on the nerve (Figure 2D).

Relay Module

The relay module features a Bluetooth Low Energy system-on-chip MCU (Texas Instruments, Dallas, TX, part number: CC2650) that is connected to a Near Field Communication reader IC (Figure 4A). The NFC IC (STMicroelectronics, Geneva, Switzerland, part number: ST25R3911B) handles the protocol level encoding and decoding and drives a 6cm 2-turn 1-layer patch coil printed on a flexible PCB substrate. The output driver of this IC is impedance matched to the coil using a matching network to maximize the power transfer efficiency. The MCU handles the transcoding of commands and responses that flow between the programming module and IPG. The MCU is connected to a real time clock (STMicroelectronics, Geneva, Switzerland, part number: M41T62) and an external memory (Cypress Semiconductor, San Jose, CA, part number: FM25V20A) that are used to log the number of stimulations delivered so that daily limits can be set. The relay module is powered from a 4.2 V, 550 mAh Lithium-Polymer battery (Renata, Itingen, Switzerland, part number: ICP622540PMT) which can be recharged wirelessly using a Qi charger. The wireless charging feature is enabled by a separate coil inside the relay module that operates at around 100 – 200 kHz frequencies as part of the Qi standard (WPC V1.2). The relay module circuitry was assembled on a 4-layer PCB using surface mount components. A water-resistant plastic enclosure was molded to hold the PCB, battery, and the 2 coils (Figure 4C). The relay module is held outside the body, directly above the IPG, using harnesses appropriate for the site of application. Figure 4D shows the relay module in a collar used for holding it on the neck.

Figure 4:

Relay module (A) Block diagram of the relay module (B) Relay module PCB with the flex coil that is used to power the IPG (C) Relay module housed in a plastic enclosure (D) Relay module in a collar

Programming Module

An app to control the IPG was developed for the Android operating system. It is compatible with all android phones and tablets that support BLE. It has separate interfaces for different types of users with access controlled via a login page. On startup, the app tries to establish communication with the last connected relay module and IPG and runs self-tests to ensure that all parts of the system are functional. To be able to use this system in a closed-loop setting, the app can be run as a background service and can accept stimulation requests from other authorized apps. This enables triggering of stimulations in response to external events.

METHODS

Operational performance testing

Power consumption of the IPG was characterized using a prototype version in which the power to the individual components can be measured. The MCU in the IPG was programmed to run through the different power states and the current was measured using a sense resistor connected to an oscilloscope. The power consumption of the relay module was measured in a similar method to the IPG. In order to test the wireless charging, a Texas Instruments Qi wireless transmitter evaluation board was used to charge a fully drained relay module.

Bench top operational testing was performed by stimulating into saline (0.9%) or a resistive load. For saline stimulation, wires were attached to the IPG electrodes and a 10 Ω resistor was connected in series to measure the current. Stimulation pulses of eight different amplitudes (100 μA – 1200 μA) were delivered. The voltage and current waveforms were captured using an oscilloscope.

To test the accuracy of the stimulation parameters under different load conditions, 24 devices were each connected to resistive loads of 499 Ω, 2000 Ω and 6800 Ω and stimulated with different combinations of the minimum and maximum values of the parameters. As in the previous setup, the current through the load was captured on an oscilloscope and measurements of amplitude, pulse width, frequency, and duration were made.

In a separate set of experiments, the power and communication range of the IPG was tested. The IPG was placed in a stereotaxic rig directly above the center of the relay module NFC coil. The vertical distance between the relay module and the IPG was varied from 0 to 5 cm and at each distance, the IPG was rotated along its long axis and placed at 0° (parallel), 30°, and 45° relative to the alignment of the coil in the relay module. The relay module was programmed to send 10 stimulation commands (1.2 mA, 250 μs, 30 Hz, 0.5 s) to the IPG, and the percentage of successful stimulations at each distance and angle was calculated.

Accelerated aging

Six IPGs were placed in a condenser-beaker filled with physiological saline (0.9%, 7.4 pH). The beaker was heated to a temperature of 87 °C for a period of 150 days, which is equivalent to >10 years in the body based on the Arrhenius equation

$$A_T=e^{{}^{\left(E_a/K\right)*\left(1/T_R-1/T_A\right)}}$$

where A_T is the acceleration factor, E_a is the activation energy of the reaction in units of eV, K is the Boltzmann constant (8.617 × 10⁻⁵ eV/K), T_R is the real temperature, and T_A is the aging temperature, both in units of Kelvin. A conservative value of 0.7 eV was assumed for the activation energy based on silicon-glass hybrid packages [33,34]. Two of the IPGs were also stimulated (0.8 mA, 100 μs, 30 Hz) at a duty cycle of 0.5 s on, 5 s off for the duration of the experiment.

Chronic large animal biocompatibility tests

A biocompatibility study was conducted at NAMSA (Northwood, Ohio) where the devices were implanted on the vagus nerves of five dogs (Canis familiaris, females, 5–8 months old). The study was approved by the NAMSA Northwood Division Institutional Animal Care and Use Committee (IACUC). Histological analysis on nerve tissue was performed at Tox Path Specialists, LLC (Frederick, Maryland).

The animals were dosed with buprenorphine (0.02 mg/kg, SQ) at least one hour before surgery. In addition, maropitant (Cerenia) was given (1 mg/kg, SC) approximately one hour before surgery. The animals were pre-anesthetized with acepromazine maleate (0.2 mg/kg, SQ). The antibiotic enrofloxacin was administered intramuscularly at 10 mg/kg. General anesthesia was induced with propofol (6 mg/kg, IV). The hair on the entire neck was closely clipped with electric clippers. The animals were intubated and placed on isoflurane inhalant anesthetic for continued general anesthesia. A non-medicated ophthalmic ointment was applied to the animals’ eyes (and re-applied, as necessary) to protect the corneas from drying. The vital signs (temperature, heart rate, respiration rate, SpO2) of the animals were monitored during the procedure. Intravenous fluids (Lactated Ringer Solution) were administered during the procedure at a rate of approximately 10 mL/kg/hr through an intravenous catheter. The surgical site was scrubbed with a germicidal scrub, wiped with 70% isopropyl alcohol, painted with povidone iodine, and draped. The animals were placed in a dorsal recumbent position. Each dog had bilateral incisions created at the level of the cervical vagus nerve/sympathetic trunk. The incisions were made through the skin on the ventral surface of the neck and the fascia and musculature was dissected to expose the carotid artery and the adjacent vagus nerve on each side. One device was placed on the left cervical vagus nerve. During the study phase, this device delivered active stimulation. The second device was placed on the right cervical vagus nerve and was not stimulated to serve as a stimulation control. The cuff around the device was closed with sutures but was not attached to the muscle/muscle fascia. The base of the device was oriented to face outwards towards the skin. The site was closed in layers in a simple continuous pattern. Prior to closing the incision with sutures, the device was stimulated to verify functionality. The skin was then sutured closed.

Ten days post-implantation, all subjects began stimulation 5 days per week for 28 days. A collar containing the relay module was placed over the IPG on the left vagus nerve during daily stimulation sessions. No relay module was placed over the right control device. Stimulation consisted of current controlled biphasic pulses of 1.2 mA amplitude, 250 μs pulse width at 30 Hz frequency for 0.5 s every 5 s for approximately 2 hours per day. After the conclusion of 28 days of stimulation sessions, the subjects underwent an upper body perfusion with 10% neutral buffered formalin while under general anesthesia. After adequate fixation, the implantation sites on both the left and right vagus nerves were examined for capsule formation or other evidence of irritation. Both the left and right nerves, including the device, were carefully excised. One of the implanted devices was reserved to test nerve stimulation efficacy as described below. Vagus nerve and the tissue surrounding the device were trimmed into transverse sections, embedded in paraffin, and stained with Hematoxylin & Eosin (H&E). Additional sections of the vagus nerve were post fixed in osmium, embedded in resin, and stained with Toluidine Blue (TB). All sections were evaluated for nerve architecture and morphologic abnormalities. The implant sites were evaluated for cellular reactions to the active stimulation and control devices. Morphological analysis was performed by a trained histologist blinded to stimulation condition.

Acute nerve recruitment

Seven New Zealand white male rabbits (Charles River, 3–6 months old, 2–4 kg) were used to test nerve stimulation efficacy in vivo. All handling, housing, stimulation, and surgical procedures were approved by The University of Texas at Dallas Institutional Animal Care and Use Committee. This test was performed using two new devices, one aged device, and one explanted device from the chronic biocompatibility study. The hind leg of the rabbit was shaved and anesthesia was induced with 3% isoflurane at 3 L/min, followed by intraperitoneal injection of ketamine hydrochloride (35 mg/kg) and xylazine (5 mg/kg). Eye ointment was applied to prevent drying of the eyes and temperature and breathing were monitored. After cleaning the site with iodine and 70% ethanol, an incision was made along the axis of the femur. The biceps femoris and quadriceps femoris muscles were separated with retractors, and the sciatic nerve was exposed. The IPG was inserted into the cuff, and the cuff was fastened around the sciatic nerve with the electrode contacts facing the medial side of the nerve. The retractors were withdrawn, and a relay module was placed at 2 cm over the IPG. To record the muscle force, the foot was connected to a load cell. The resistance of the load cell was amplified and sampled at 500 Hz by custom hardware and sent to a computer. Stimulation was pseudorandomly delivered at 10 Hz in 0.5 second trains every 5 seconds, consisting of 100 μs biphasic pulses across a range of amplitudes. The threshold of nerve activation was established as the lowest stimulation intensity that induced a change in force greater than 3 times the standard deviation of the preceding 1 s of the baseline. The force data were linearly interpolated to be able to average the trials over the same amplitudes.

Hermeticity tests

Ten explanted devices from the chronic biocompatibility study and six devices from accelerated aging study were tested for hermeticity by an optical leak test (OLT). The test was performed using Norcom 2020 optical leak test system running Helium gas. The devices were placed inside the OLT chamber, and the test was run for 60 min with a pressure of 30 psig and modulation of 0.8 psig. At the end of the test, the leak rates of the devices were compared against the reject limit of 2.6 × 10–7 atm-cc/sec (based on MIL-STD-883K adjusted for Helium).

Heat generation test

To evaluate the heat generated by the IPG during normal usage, a thermocouple was affixed to an IPG such that the tip made contact with the electrodes which are the hottest parts of the outer surface. The IPG was then inserted between two pieces of pork loin and placed in an oven at 37° C. A relay module was positioned over the top of the pork with an air gap of 4–6 mm. Temperature on the IPG surface was sampled at a rate of 1 Hz. Once the temperature reached equilibrium (< 0.2 °C change in 15 min), stimulation was performed using the maximum output parameters allowed by the IPG (1.2 mA amplitude, 250 μs pulse width, 30 Hz pulse rate) at the maximum duty cycle of 25 seconds on, 250 seconds off. To examine heating from repeated stimulations, a total of 24 stimulations were performed.

RESULTS

Power characteristics

Table 2 shows the power consumption of the IPG when idle and during stimulation at 1.2 mA, 250 μs pulse width, and 30 Hz frequency. Though the peak power during a pulse is 24 mW, the average power during a stimulation train, even with maximum pulse width and frequency, is only 320 μW more than the idle power.

Table 2:

Measured power consumption of the IPG

	Idle	Stimulation (peak)	Stimulation (avg)
Near Field Communication	0.27 mW	0.27 mW	0.27 mW
Microcontroller	0.762 mW	2.274 mW	0.785 mW
Power Management	0.096 mW	0.096 mW	0.096 mW
Stimulation circuitry	1.5 mW	21.4 mW	1.798 mW
Total	2.63 mW	24.04 mW	2.95 mW

The relay module, when turned on but not powering the IPG, draws approximately 5 mA from the 4.2 V Li-ion battery. While it is connected to a programming module, its power draw increases slightly to 6 mA due to the continuous exchange of Bluetooth packets. Even in the connected state, with the 550 mAh battery, the relay module can last close to 100 hrs. When the relay module is powering the IPG, it draws the most power of about 300 mA. This state is momentary since the module only turns on the inductive field during a stimulation. The relay module battery required approximately 90 mins to fully charge from a completely drained state.

Operational performance testing

We tested the ability of the IPG to output pulses of varying amplitude through saline. Fig 5 shows the voltage and current waveforms captured on an oscilloscope for different current amplitudes. Since the IPG employs a constant current stimulation circuit, the output voltage varies to keep the current constant through the capacitive load (Fig 5B). An artifact can be observed to occur between the positive and negative phases of the pulse, when the direction of current is reversed. The mean percent errors in the pulse parameters as a result of stimulation through different resistive loads, along with their standard errors across devices are given in Table 3. The errors are low (< 10%) even at high stimulation intensities, indicating reliable operation. Fig 5C shows the minor changes in pulse waveform seen with increases in load. This demonstrates that the IPG can reliably output a controlled current pulse under varying load conditions.

Figure 5:

Results of benchtop testing. (A) Current waveforms of a 100 μs biphasic pulse at 8 different amplitudes. (B) Corresponding voltage waveforms for the 8 amplitudes. (C) Current waveforms of a 1200 μA pulse through different loads

Table 3:

Percent errors of parameters

Parameter	Value	Percent error
		499 Ω	2000 Ω	6800 Ω
Amplitude (μA)	120	5.84 ± 0.38	3.00 ± 0.22	2.59 ± 0.19
	1200	1.66 ± 0.08	1.64 ± 0.11	1.10 ± 0.05
Pulse Width (μs)	100	4.20 ± 0.18	3.09 ± 0.11	3.56 ± 0.26
	250	1.90 ± 0.14	1.58 ± 0.11	1.12 ± 0.07
Frequency (Hz)	10	Not evaluated*	Not evaluated*	0.66 ± 0.06
	30			0.61 ± 0.05
Duration (s)	0.5			0.54 ± 0.05
	25			0.65 ± 0.06

^*Frequency and duration parameters have no dependence on load

We determined the maximum range at which the IPG can successfully deliver a stimulation. At both 0° and 30° tilt, the IPG was able to complete a pulse train up to 4.3 cm from the relay module. When the IPG was tilted 45° relative to the coil of the relay module, the range dropped to 4.1 cm (Figure 6). This drop is expected since the inductive power transfer is maximum when the transmit and receive coils are parallel and declines as a function of increasing angle. The results of tilt along the short axis of the IPG were comparable to tilt along the long axis (not shown). These findings suggest the IPG can be used to stimulate peripheral nerves that are located at up to 4 cm deep and is tolerant to a tilt of up to 45°.

Figure 6:

Results of range testing (A) Percentage of successful stimulations delivered by the IPG at different distances and angles from the relay module (B) Illustration of range measurement

Chronic large animal biocompatibility tests

We performed chronic in vivo testing to evaluate the long-term biocompatibility and safety of the device. Five pairs of devices were successfully implanted in five dogs, the devices implanted on the right vagus nerve were not stimulated while those implanted on the left nerve were stimulated for 28 days. All animals received a minimum of 14,000 s of stimulation over the course of the study. The op amp voltage required to deliver the 1.2 mA of current was monitored and was found to be well within the compliance voltage of 12 V over the course of the 28 days (Figure 7A). The electrode impedance also demonstrated relatively little variance across the duration of the study. There was a trend of impedance changes stabilizing and the variance across the animals deceasing over time. Despite the constant mobility of the animals, the communication success rate was greater than 80%, barring the initial few days when the relay module was not mounted firmly (Figure 7B).

Figure 7:

Results of chronic biocompatibility testing (A) Voltage that was required to deliver 1.2 mA of current in the five animals (B) IPG communication success rate for the five animals

At the conclusion of 28 days of stimulation, the left and right vagus nerves and surrounding tissue were examined. A small amount of capsule formation was observed around the device, consistent with that expected for chronic glass encapsulated implants [40,41]. No adhesion was observed in any of the devices, indicating that the device and cuff could be explanted safely. Microscopic analysis revealed a small amount of Grade 1 thinning of myelin sheaths, likely caused by nerve manipulation during surgery. In the soft tissue surrounding the vagus nerve, all animals presented various grades and proportions of mixed infiltrates, necrosis, fibrosis, hemorrhage, and neovascularization within the perivagal capsule and adipose tissue. Both sides were randomly affected by lesions throughout the group. All the lesions found in the animals were interpreted to be caused by either the placement/presence of the device, the surgical procedure, or a combination of both. Some degree of mixed inflammatory infiltrates accompanied by granulation tissue, hemorrhage, necrosis, and fibrosis are typically seen in tissues surrounding devices implanted into soft tissue [40,41]. Some inter-site variability was expected and was present but in general, the reactions surrounding the unstimulated and stimulated implants were similar. No bacterial or other infectious organisms were seen in any of the sections evaluated, hence it is unlikely that infection of the test sites occurred. Together, these findings provide an initial demonstration of chronic biocompatibility and safety of the ReStore implant.

Effectiveness of nerve stimulation

Nerve recruitment

We next sought to confirm that the device could provide reliable, effective nerve recruitment. To do so, we evaluated fiber recruitment in the sciatic nerve in rabbits across a range of stimulation intensities. Figure 8A shows that the device demonstrated effective nerve recruitment at low stimulation intensities, exhibiting full fiber recruitment (plateauing of the force curve) at 300 μA, approximately 25% of the maximum amplitude range of the device. This demonstrates that the device is capable of producing low threshold nerve activation in the sciatic nerve.

Figure 8:

ReStore devices demonstrate reliable nerve recruitment. (A) Force measurements using normal devices (16 trials) shows the low amplitudes required to activate the sciatic nerve (B) Comparison of nerve recruitment of an aged device with a normal device shows a slight increase in activation threshold, but still well within the range of the device (C) Comparison of nerve recruitment of an explanted device from chronic safety study with a normal device shows similar activation thresholds (D) Comparison of activation thresholds of normal, aged and explanted devices

Accelerated aging

To determine whether the device would still exhibit effective nerve recruitment after sustained use, we examined the efficacy of sciatic nerve recruitment with devices subjected to an accelerated aging procedure that emulates at least 10 years at body temperature [33,34]. The results trend towards an increase in the threshold of activation compared to a normal device (Figure 8B). However, the mean threshold of 253 ± 77 μA is still well within the amplitude range of the device. These data are consistent with the notion that the device will likely provide effective nerve recruitment after sustained aging. While the 10-year target was just for a preliminary testing, further studies would be done to fully characterize the functional lifetime of the device.

Chronic implantation

The explanted devices from the chronic biocompatibility study were evaluated for nerve recruitment to look for any degradation in performance. The nerve recruitment curve of the explanted device was found to be similar to that of the normal devices with almost no change in the threshold of activation (Figure 8C). This confirms that the performance of the device is not substantially degraded with chronic implantation and stimulation.

Hermeticity tests

Ten devices explanted at the conclusion of the chronic biocompatibility study and six devices from the accelerated aging study were subjected to hermeticity testing. All devices passed optical leak testing, with a leak rate of ≤ 5.5 × 10–8 atm-cc/sec, the minimum sensitivity for a 1-hour test. This rate is less than the reject limit of 2.6 × 10–7 atm-cc/sec (based on MIL-STD-883K adjusted for Helium). These findings demonstrate that the device encapsulation is robust even after exposure to chronic conditions and aging.

Heat generation test

To ensure that the IPG does not harm the patient due to heat generation during use, the temperature of an IPG was monitored at maximum stimulation settings. The IPG electrode surface temperature is shown in Figure 9. The average temperature increase during a 25-second stimulation was 0.7°C ± 0.05°C. After each stimulation ended, the IPG temperature decreased over the 250-second inter-stimulus interval and reached the baseline temperature again by the time the next stimulation began. On two occasions, the stimulation command was not completed on the first attempt, likely due to the shielding effect of the oven. In these cases, an additional stimulation command was sent with a small delay. No cumulative heating effect was observed from multiple consecutive stimulations, as the temperature at the start of the first stimulation was equal to the temperature at the start of the last stimulation (37.25°C). This shows that the heat generated by the IPG operating at the maximum possible parameter settings is well below the amount needed to raise tissue temperature by 2°C, which is the limit set for implantable neurostimulators [35]. Device heating is expected to be negligible when delivering brief trains (i.e. the 0.5-second long trains delivered during the chronic large animal biocompatibility tests and the nerve recruitment tests)

Figure 9:

IPG temperature at the electrode surface during maximum stimulation

DISCUSSION

In this article, we describe the ReStore peripheral nerve stimulation system. The ReStore IPG is wirelessly powered and contains no battery, and thus eliminates the need for additional surgeries for battery replacement. The size of the IPG is conducive to implantation directly at the site of larger nerves, such as the cervical vagus nerve. Implanting the device directly in contact with the nerve eliminates the necessity for leads and thus obviates the possibility of lead breakage as a failure mode. The IPGs are fully assembled and encapsulated at the wafer level, similar to the process of manufacturing semiconductor and MEMS devices. Utilizing standard MEMS foundry equipment and wafer processes offers several advantages over legacy serial manufacturing techniques for medical devices, including reduced inter-device variance, greater process control and reliability, and lower cost due to parallel processing and substantial increase in yield. The entire ReStore system is manufactured using commercially available components and processes in order to minimize development time, cost, and risk of failure. One potential disadvantage of this approach is the components used may not be well suited to the application, causing the final product to have inferior performance and size metrics compared to a device utilizing custom components. In this study, we performed a variety of performance and reliability tests in order to evaluate whether the performance of the ReStore system meets the expectations for an active implantable medical device.

The measured power consumption of the ReStore IPG was 2.6 mW while powered, but idle. Although the peak power during a pulse is much higher (24 mW), the pulse durations are brief compared to the time between pulses, so the average power consumed during stimulation is only 3 mW. Energy storage capacitors in the IPG supply the momentary power required during a pulse, and recharge between pulses. The relationship between power consumed by the IPG and the heating of surrounding tissue depends on the device geometry and materials, and is therefore device specific. We empirically evaluated how tissue temperature might change during stimulation and found that the average increase was 0.7 °C for the worst-case stimulus parameters. This amount of heating is within the guidelines for active implantable medical devices based on ISO 14708–3 [35].

The results of the operational performance testing demonstrate that the ReStore IPG is able to deliver controlled-current pulses with digitally programmable parameters. The measured output of the IPG agrees with the programmed parameters to a reasonable degree of accuracy (<10% error). One consequence of using an op amp to drive the electrode current is the current spike that can be observed between the positive and negative phases of the pulse. This occurs because the op amp has a finite slew rate, so the output voltage used for the positive phase does not completely return to zero before the start of the negative phase. Since larger op amp voltages are needed for larger loads, the effect is more pronounced with larger loads. However, since the amplitudes of pulse phases are equal, this current spike will not exceed the amplitude of the second pulse phase. Moreover, the difference in charge delivered by each phase was less than 10% across the entire parameter space for all loads (not shown). This small charge imbalance is quickly dissipated since the electrodes are shorted together between pulses, compared to when the electrodes are floated. This limits any irreversible Faradaic reactions that are known to occur due to prolonged charge imbalance between electrodes [42]. The limited op amp slew rate is also the reason why the rise time of the current pulse slows with larger loads

The maximum operational range of the IPG was shown to be approximately 4 cm, which is appropriate for nerve targets that are relatively superficial. This limit is partially dependent on the power requirements of the IPG, but more importantly on the communication fidelity. NFC communication from the IPG to the relay module is performed by passive load modulation, which is power efficient but cannot be easily boosted except by increasing the coil size [36]. The operational range of the ReStore IPG is limited by the strength of this communication, and is not improved by increasing the relay module output power. Therefore, the system’s limited range makes it more appropriate for use with relatively superficial nerve targets, such as the cervical vagus nerve or tibial nerve, rather than nerves located deep within the body.

Results from chronic testing in large animals reveal that the voltage required to drive the prescribed pulse current was well within the 12 V compliance of the ReStore IPG for all animals and for the entire duration of the study. Moreover, the electrode/tissue impedance trended toward less variability across animals and became more stable over time. This suggests an absence of tissue growth between the IPG and nerve, which is supported by gross anatomical examination during explantation. The capsule formation observed around the IPG is similar to other chronically implanted, glass encapsulated devices [40,41]. The lack of inter-site variability in the histological characteristics suggest that the levels of stimulation delivered are safe. These findings provide an initial demonstration of the safety and biocompatibility of the system.

In conclusion, the ReStore peripheral nerve stimulation system was designed to eliminate the battery and lead failure modes of traditional IPGs, and to facilitate commercialization by prioritizing manufacturability. The testing performed on the ReStore system to date demonstrates it performs well under a variety of simulated conditions, and provides an initial indication of safety consistent with standards for human use.

Highlights (for review).

• The ReStore system is an implantable nerve stimulator with no battery or leads

• It is constructed using commercial components and processes for manufacturability

• Results from bench and in vivo testing demonstrate biocompatibility and reliability