Stimulation of a Nerve Supplying the Spleen

Abstract

Stimulation of neural activity in a nerve supplying the spleen, wherein the nerve is associated with a neurovascular bundle, can re-programme immune cells in the spleen, modulate pro- and anti-inflammatory molecules levels, and induce disease-resolution pathways system-ically thereby reducing inflammation and providing ways of treating inflammatory disorders. The invention provides improved ways of treating inflammatory disorders which minimize off-target effects.

Claims

1.-51. (canceled)

52. A system for stimulating neural activity of a nerve the system comprising: at least one electrode in signalling contact with the nerve; and at least one controller electrically coupled to the at least one electrode, the at least one controller configured to control operation of the at least one electrode to apply an electrical signal to the nerve, wherein the electrical signal comprises a pulse train having a pulse width >1 ms.

53. (canceled)

54. (canceled)

55. The system of claim 52, wherein the pulse width is ≤5 ms.

56. The system of claim 52, wherein the pulse width is between 1.5 and 2.5 ms.

57. The system of claim 52, wherein the pulse width is ≤3 ms.

58-71. (canceled)

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0239] Embodiments of the invention will be described, by way of example, with reference to the following drawings, in which:

[0240] FIG. 1 illustrates a neural stimulation system.

[0241] FIG. 2 illustrates a wider system including the neural stimulation system.

[0242] FIG. 3 shows histological and electrophysiological characterization of a pig splenic nerve. FIG. 3A is a photomicrograph of a semi-thin section (0.5 μm thickness) of the SpA (splenic artery)/SpN (splenic arterial nerve) stained with Toluidine blue. No myelinated axons can be observed in the image. FIG. 3B representative traces of evoked compound action potential (eCAP) recorded from fascicles of the pen-arterial splenic nerve dissected off the artery when stimulating at 1 Hz with a pen-arterial cuff (around the entire SpN plexus) or with a small cuff around few fascicles of the SpN bundle. The traces are the average of 10 responses. FIG. 3C shows the range of conduction velocities of the different components of the eCAP. FIGS. 3D and 3E show the strength-duration curve of the SpN obtained by stimulating the whole plexus (FIG. 3D) or few dissected fascicles (FIG. 3E). The graphs show also the relative charge density to obtain threshold eCAP at different stimulation amplitudes. All stimulations were performed at 1 Hz to limit stimulation-indiced action potential conduction slowing in the nerve.

[0243] FIG. 4 shows transient changes in mSpA BF, mSpV BF, sMABP and HR that are stimulation intensity dependent caused by SpN stimulation. FIG. 4A shows the mean (n=8) change in mSpA BF (from −30 to +180 s, relative to start of stimulation) during a 1 minute stimulation (symmetric square biphasic pulses, 400 μs PW at 10 Hz) of the SpN plexus at different current amplitudes (between 3.5 and 20 mA). FIG. 4B shows the maximum reduction in mSpA BF reached during a 1 minute stimulation (symmetric square biphasic pulses, 400 μs PW at 10 Hz) of the SpN plexus at different current amplitudes. Each line represent an animal tested. FIG. 4C shows the mean (n 3) maximum reduction in mSpA BF reached during a 1-minute stimulation (symmetric biphasic pulses, 400 μs or 200 μs PW at 10 Hz) of the SpN plexus at different current amplitudes and with two different PW: 400 (black circles) and 200 (black squares)μs. FIG. 4D shows the change in mSpV BF (from −30 to +180 s, relative to start of stimulation) during a 1-minute stimulation (symmetric biphasic pulses, 400 μs PW at 10 Hz) of the SpN plexus at different current amplitudes (between 3.5 and 12 mA). FIG. 4E shows the mean (n=3) change in sMABP and HR (from −30 to +180 s, relative to start of stimulation) during a 1-minute stimulation (symmetric biphasic pulses, 400 μs PW at 10 Hz) of the SpN plexus at different current amplitudes (between 3.5 and 20 mA). FIGS. 4F and 4G summarize the mean (n=3) maximum changes in mSpA BF, sMABP, HR and RR during a 1-minute stimulation (symmetric biphasic pulses, 400 μs PW at 10 Hz) of the SpN plexus (FIG. 4F) or some dissected SpN fascicles (FIG. 4G) at different current amplitudes. Both graphs show the amplitude (measured as area under the curve of the response) of the recorded eCAP (expressed as % over the maximal response). SpA BF changes are expressed as maximum reduction from baseline in %, HR changes are expressed as beats per minute (bpm), sMABP changes are expressed as mmHg, RR changes are expressed as breaths per minute (bpm). The two graphs also reports the charge density per phase relative to the stimulation amplitude used.

[0244] FIG. 5 shows that changes in mSpA BF, mSpV BF, sMABP and HR during SpN stimulation were frequency dependent. FIG. 5A shows the mean (n=3) change in mSpA BF (from −30 to +180 s, relative to stimulation) during a 1 minute stimulation (symmetric biphasic pulses, 400 μs PW at about 36.9 μC/cm2/phase) of the SpN plexus at different frequencies (between 0.25 and 100 Hz). FIG. 5B shows the mean (n=3) maximum reduction in mSpA BF observed during a 1 minute stimulation (symmetric biphasic pulses, 400 μs PW at about 36.9 μC/cm.sup.2/phase) of the SpN plexus at different frequencies (between 0.25 and 100 Hz). In FIG. 5C to 5D, the graphs show the changes in mSpV BF, sMABP, HR (expressed as % over prestimulation baseline) during a 1 minute stimulation (symmetric biphasic pulses, 400 μs PW at about 36.9 μC/cm.sup.2/phase) of the SpN plexus at different frequencies (between 0.25 and 100 Hz). Data in FIG. 5A is expressed as mean±s.d. In FIGS. 5A and 5C to 5D, the box represents the stimulation time window.

[0245] FIG. 6 shows local and systemic effects of few dissected SpN fascicles at different frequencies. In particular, FIG. 6 shows a representative experimental recording of local and systemic changes associated with the stimulation of few SpN fascicles dissected off the artery with different frequencies. HR sMABP, Stimulation input, eCAP, SpA BF raw and mSpA BF data are shown from a representative experiment where frequency ranges from 3 to 300 Hz.

[0246] FIG. 7 shows SpA blood flow changes monitored via intra-operative splenic ultrasonography. The images of FIG. 7 were obtained from two different animals during SpN stimulation. Note the reduced Doppler trace during stimulation (middle panels) versus pre-stimulation and post-stimulation (top—and bottom panels, respectively).

[0247] FIG. 8 shows activity-dependent variations of SpN eCAP amplitude and conduction velocity. FIG. 8A shows the eCAP recorded from the SpN during a 1 minute stimulation at different frequencies (1, 10 and 30 Hz, from left to right). Each image show the superimposition of all the evoked responses. For 1 Hz stimulation there are 60 responses superimposed; for 10 and 30 Hz, each trace represents the average of 5 consecutive responses. Note that the responses at 10 and 30 Hz shift to the right over time and the amplitude is reduced over time. FIG. 8B shows the eCAP recorded from the SpN during a 1 minute stimulation delivered at 10 Hz in burst of 5 or 10 pulses separated by an off period of 5 or 10 s. Each image show the superimposition of all the evoked responses. Note that the responses neither shift to the right nor are reduced in amplitude over time. FIG. 8C shows the quantification of the area under the curve (AUC) of each recorded eCAP of the different stimulation paradigms. In particular, FIG. 8C shows the comparison of the values between 1 and 60 pulses delivered with the different paradigms.

[0248] FIG. 8D shows eCAP latency (expressed as % over the latency of the first response) over 600 consecutive pulses with the different patterns of stimulations shown in FIGS. 8A-C. Data are shown as mean (N≥3). Dotted lines represent the 95% confidence interval. FIG. 8E shows eCAP amplitude (expressed as % over the amplitude of the first response) over 600 consecutive pulses with the different pattern of stimulations shown in FIGS. 8A-C. Data are shown as mean (N 3). Least squares regression curves were fitted against the latency (as shown in FIG. 8D) and amplitude data (as shown in FIG. 8E). Dotted lines represent the 95% confidence interval. FIG. 8F shows change in SpA mBF (circles) and sMABP (triangles) during a 60s stimulation of the porcine splenic neurovascular bundle (NVB) using 400 μs PW and 12 mA (symmetric, biphasic square pulses) delivered at 10 Hz (in black), 1 Hz (light grey) or burst stimulation (10 Hz, 5 pulses every 10 seconds, in grey). Data are from representative stimulations within the same animal. FIG. 8G shows SpA mBF and sMABP max changes recorded during a 60s stimulation delivered with the different stimulation patterns shown in FIG. 8F. Values are expressed as % over the max change obtained at 10 Hz. Data are shown as mean (N=4) ±s.d. Statistical analysis in FIG. 8C was performed using One-way ANOVA and Tukey post-hoc correction for multiple comparison. *, P≤0.05; **, P≤0.005; ***, P≤0.001; ****, P≤0.0001.

[0249] FIG. 9 shows burst and 1 Hz stimulation produced the lowest changes in mSpA BF. In particular, FIG. 9 shows the maximum change in mSpA BF expressed as % over the change obtained during a 60 s stimulation with biphasic, symmetric pulses delivered at 10 Hz (in black). Different stimulation paradigms, delivered with the same current amplitude, are compared: continuous 10 Hz, continuous 1 Hz and burst stimulation (5 pulses at 10 Hz every 5 s) with either symmetric or asymmetric biphasic pulses.

[0250] FIG. 10 is a schematic illustration of the porcine left abdomen highlighting the anatomical features of the splenic plexus (spleen, nerves, artery and veins). The location for cuff placement during the experiments of pen-arterial SpN stimulation is shown. Nerves are represented in black, and arteries and veins in grey.

[0251] FIG. 11 shows that stimulation of the SpN in pigs causes reduction in the LPS-induced cytokine production in vivo. FIGS. 11A and 11B show the dynamic change in TNFα and IL-6 measured directly from plasma collected by terminally anesthetized pigs administered with 0.25 μg of E. coli LPS. FIGS. 11C and 11E show the peak value of TNFα and IL-6 measured in the plasma after LPS administration in the different groups. FIGS. 11D and 11F show the quantification of the Area under the curve (AUC) of both TNFα and IL-6. Data are expressed as mean (Sham: N=6, Dexamethasone: N=2, LVNS: N=5, eLVNS: N=5, SpNS: N=6)+/−SD.

[0252] FIG. 12 shows anatomical and histological analysis of the SpN along the main SpA and the short gastric and epiploic arteries. FIG. 12A is a schematic representation of the splenic neuroanatomy highlighting (dashed lines) the regions where the histological analysis was performed. FIGS. 12B to 12D show sections of the SpN at different levels, main splenic artery (FIG. 12B), short gastric (SG) arteries (FIG. 12C) and gastroepiploic (GEP) artery (FIG. 12D), stained with Haematoxylin and Eosin (H&E). Nerves in FIG. 12C and FIG. 12D are indicated by the arrowheads. In FIG. 12D, the insert shows a high magnification caption of one nerve fascicle. FIG. 12E shows a box plot reporting quantification of the number of SpN fascicles at different locations (top panel) and the mean diameter distribution of the same fascicles in the different locations (bottom panel). FIG. 12F shows the number of fascicles at different locations and their relative mean diameter.

[0253] FIG. 13 shows that the human splenic nerve is a plexus of pen-arterial fascicles containing slow conducting axons. FIG. 13 includes the following subsections: A) Human splenic splenic neurovascular bundle (NVB) containing the SpA, the SpN, connective tissue, sections of pancreas and lymph nodes freshly isolated from a donor. Two small cuff electrodes (650 μm in diameter) were placed on a select few dissected fascicles. The schematic of the preparation indicates the position (a and b) of the stimulating and recording cuffs. The dotted lines indicate the areas in which the sections shown in B and C were taken; (B) Section of the human NVB stained with Haematoxylin and Eosin (H&E). The SpN fascicles are encircled; (C) Section of the stimulated fascicles that were isolated for electrophysiological study. The section was stained with H&E and shows the nerve fascicles (encircled) and fat/connective tissue; (D) eCAP recorded when applying monopolar, monophasic stimulation of the human SpN at 1 Hz and 400 μs PW prior (top panel) and after (bottom panel) crushing the nerve between the stimulating and recording cuff. The left box indicates the stimulation artefact while the larger on the right indicates the area in which eCAP should be observed, with the arrows indicating the eCAP; (E) Recruitment curve of the human SpN quantifying the eCAP amplitude (expressed as % of the maximum response) vs the stimulation amplitude. Each point represents the average amplitude of 8 consecutive monopolar, monophasic pulses delivered at 1 Hz and 400 μs PW; (F) Conduction velocities of all the eCAP components recorded from the human, porcine (pig) and rat SpN; (G) Strength-duration relationship (black circles) of the human SpN obtained by stimulating the dissected fascicles. The data represent the minimum current needed to trigger a detectable eCAP at the different PW tested. The graphs also show the corresponding charge density (black triangles) of the different stimulations (referred to the right Y axis). Least squares regression curves were plotted against the strength-duration and charge density data.; and (H) Charge densities required to stimulate the SpN of the three different species at different PW. The data were fitted with linear regressions. Scale bars: B=2 mm; C=100 μm.

[0254] FIG. 14 shows A) Example of a human splenic sample with suture indicating the proximal end close to celiac, (B) Conceptual representation of slicing of tissue in blocks for histology, (C) Haematoxylin and Eosin (H&E) stained slide from one of the blocks, and (D) methodology for histomorphometric estimations.

[0255] FIG. 15 shows (Left) Fascicle diameter, (Middle) Fascicle spread around adventitia (outer splenic arterial wall) for proximal, middle and distal parts of the splenic neurovascular bundle (NVB), and (Right) Percentage of fascicles vs distance from adventitia.

[0256] FIG. 16 shows In-vivo data from porcine splenic neurovascular bundle stimulation; (A) population recruitment curve, (B) Strength-duration curve.

[0257] FIG. 17 shows (A) Recruitment curve from in-silico modelling in porcines with x-axis representing charge injection estimates at 400 us pulses, (B) same with x-axis reflecting stimulation amplitude, (C) Recruitment curve from in-silico modelling in humans with x-axis representing charge injection estimates at 400 us (blue) and 1 ms pulses (red), (D) same with x-axis reflecting stimulation amplitude (mA).

[0258] FIG. 18 shows (A) an example of the human splenic tissue. The dark stained spots on the sample indicate the splenic artery with aorta towards the left end, and spleen on the right end of the sample (for orientation). (B) shows placement of a pen-arterial cuff around the neurovascular bundle (I) and placement of a smaller diameter cuff around a few nerves (III). The nerve is dissected, placed in a bath with Kreb's solution, and traced all along till the end of the sample, where the hooks are placed to record compound action potentials (C, III). (D) shows a conceptual sketch of tissue with the cuff, and hook placement, and (E) shows an example of an eCAP observed on the oscilloscope.

[0259] FIG. 19 shows results from an ex-vivo electrophysiological study of the human splenic samples. (A) shows current amplitude-pulse width and charge density-pulse width curves. The error bars demonstrates the range, and the lower bar of the range is not presented on the graph. (B), (C), and (D) show recruitment graphs for 0.4 ms, 1 ms and 2 ms pulse widths respectively.

[0260] FIG. 20 shows predictions of recruitment curves for a human splenic nerve in chronic scenarios based on human ex-vivo data at 2 ms pulse width. The y-axis represents the eCAP amplitude as a percentage of maximum response and the x-axis represents the total charge (μC) injected into the human splenic nerve.

[0261] FIG. 21 shows comparisons of recruitment curves calculated for the human model for acute and chronic stimulations with different parameterisations of biphasic pulse waveforms, in particular different pulse widths (0.4 ms, 1 ms) and different interphase delays (0 ms, 0.1 ms, 0.2 ms). In the key (e.g. ‘Chronic1mOms’), the word represents the type of stimulation (e.g. ‘Chronic’), first number represent the pulse width in ms (e.g. ‘1’ ms), and the second number represents the interphase delay in ms (e.g. ‘0’ ms).

[0262] FIG. 22 shows the charge required to stimulate neural activity per pulse width in a human splenic nerve based on in-silico modelling data. Simulations are based on electrical signals with pulse trains having biphasic pulses with a 0 ms interphase delay (“Biphasic”), biphasic pulses with a 0.1 ms interphase delay (“Biphasic (0.1 ms interp. delay”), and monophasic pulses (“Monophasic”).

[0263] FIG. 23 shows unmyelinated fiber pulse height thresholds verses interphase delay normalised to a 100 μs interphase delay. The y-axis represents the threshold relative to an interphase delay of 100 μs and the x-axis represents the interphase delay (μs).

[0264] FIG. 24 shows comparison of frequency. An increase in frequency from 1 Hz to 10 Hz indicates a reduction in eCAP amplitude and is indicative of nerve fatigue, thus re-confirming porcine data assumptions on frequency.

MODES FOR CARRYING OUT THE INVENTION

[0265] Porcine Data

[0266] Electrical Stimulation of the Splenic Arterial Nerve in Pig

[0267] Materials and Methods

[0268] A total of 8 pigs (body weight between 40-50 Kg) were used for the histological and electrophysiological characterization of the splenic nerve.

[0269] On the experimental day, the animal was sedated with ketamine (1.5 mg/kg) and midazolam (0.5 mg/kg) administered by intramuscular injection. An intravenous catheter was placed in one auricular vein, and anesthesia was induced by propofol (2 mg/Kg) administered intravenously. An endotracheal tube was placed, and anesthesia was maintained with sevoflurane inhalant combined with continuous rate infusion (CRI) of fentanyl (0.2 μg/Kg/min).

[0270] After induction of general anesthesia, the animal was positioned in dorsal recumbency for placement of bilateral indwelling jugular vein catheters and one femoral arterial catheter under ultrasonographic guidance. Animals undergoing SpN cuff implantation were then repositioned into right lateral recumbency.

[0271] The surgical approach to SpN cuff implantation was as follows. The thoracolumbar junction was supported and slightly elevated using a sand bag. After appropriate surgical preparation (clipping and aseptic scrub with chlorhexidine gluconate and alcohol), the left flank was aseptically draped exposing a 20×25 cm area centered on the second to last rib. A 15 cm skin incision was made in the second to last intercostal space using monopolar electrocautery. The incision was continued through the subcutaneous tissues and intercostal musculature until the peritoneum was exposed. Two Finochietto rib retractors were placed retroperitoneal, taking care to engage the ribs. Over the next few minutes, the retractors were gradually opened, resulting in exposure of the left lateral abdomen measuring approximately 10×8 cm. The retractor blades were covered with gauze sponges soaked in carboxymethyl cellulose (CMC). The peritoneum was longitudinally incised and sutured to the skin (Vicryl 2-0; Ford interlocking suture pattern) covering the retractors blades in order to minimize risk of splenic tears during handling. Using careful digital manipulation, the spleen was exteriorized and the splenic artery (SpA) was identified along its visceral surface. At the mid portion of the spleen, proximal to the SpA branching into the left gastroepiploic artery, a short segment of the SpA was carefully dissected free of surrounding soft tissue for placement of a 1 mm ultrasonic flow probe (Transonic). After probe placement, the spleen was repositioned into the abdomen.

[0272] By slight rotation of the splenic visceral base towards the operator, and placing gentle ventral traction on the spleen, the gastrosplenic ligament at the splenic hilum was incised using Metzembaum scissors, exposing the SpA. The artery was followed in a dorsal direction to its origin (i.e. the bifurcation of the celiac artery into the left gastric artery (LGA) and the SpA). Immediately distal to this bifurcation, an approximately 1 cm segment of the SpA with the peri-arterial SpN network intact, was circumferentially isolated by blunt dissection using Metzenbaum scissors. A curved Mixter artery forceps was inserted under the artery from caudal to cranial, grasping one flap of the 2.5 mm diameter CorTec cuff introduced into the surgical field using straight Microdissection forceps. The cuff was placed around the SpA and the intact pen-arterial SpN network by reversing the motion of the Mixter forceps, taking care to appose the two flaps of the cuff when properly placed. The tension on the spleen and artery was then released. SpA blood flow readings were tested and finally the rib retractors were partially closed and the exposed incision covered with saline-soaked gauze sponges.

[0273] Electrophysiological experiments generally entailed dissecting and cuffing (using a 500 μm diameter bipolar or tripolar CorTec cuff) one or several discrete SpN fascicles few centimeters distal (closer to the spleen) to the stimulating cuff to enable evoked compound action potential (eCAP) recording during stimulation of the whole SpN plexus or of few fascicles (see FIG. 3). Also, different combinations of blocking neural signaling (e.g. using topical administration of local anesthesia, or transection of the SpN fascicle) either upstream or downstream of the stimulation site were performed.

[0274] Recorded eCAP were amplified and filtered (100-1000 Hz) using an 1800 2-Channel Microelectrode AC Amplifier (A-M system). Nerve activity was monitored continuously using an oscilloscope and recorded to a computer using a 16 channels PowerLab (AD Instruments) acquisition system and LabChart 8 software using a sampling rate of 20 kHz. eCAP were generally averaged (8-10 pulses) and peak to peak or area under the curve (AUC) of the averaged response quantified. The conduction velocity of the eCAP components of the SpN were calculated from the distance between stimulation and recording site and the latency of the eCAP signal.

[0275] Electrocardiogram (ECG), Heart rate (HR), arterial blood pressure, respiratory rate (RR), pulse oximetry, capnography, spirometry were monitored throughout the surgery. Body temperature was recorded continuously with an intranasal probe. Arterial blood gasses were analyzed throughout the experiment to monitor pH, Glucose, pO2 and pCO2, K+levels. All physiological parameters as well as the level of used sevoflurane were recorded (every 5-10 minutes) on the record sheet. Physiological data were also digitalized using Powerlab acquisition system and LabChart software. All parameters were generally sampled at a frequency between 0.1 and 2 kHz.

[0276] The depth of anesthesia was assessed by palpebral reflex, corneal reflex, medioventral eye ball position, and jaw tone.

[0277] Moreover, physiological parameters as well as a bispectral index monitoring system (levels between 30 and 60) were used to adjust anesthetic levels. In some cases, boluses of propofol were used.

[0278] In some cases intra-operative ultrasonography of the spleen was used for real-time monitoring of SpA blood flow changes during SpN stimulation. For this procedure, an intra-operative probe (i12L-RS linear intraoperative transducer 4-10 MHz, 29×10 mm footprint, 25 mm field of view; GE Vivid-i) was used.

[0279] SpA blood flow changes was assessed by color Doppler and continuous wave spectral tracing. After color Doppler identification of the SpA within the splenic parenchyma 2-3 cm distal to the splenic hilum, continuous wave spectral tracing of the SpA flow was obtained by directing the windowing cursors to the center of the SpA lumen. After obtaining a representative signal, the ultrasonography probe and cursor window was left in position while SpN stimulation commenced.

[0280] All statistical analyses were performed with commercially available statistical software (JMP Pro 13.0.0 or GraphPad Prism 5.0).

[0281] Results

[0282] Recording of the eCAP generated during SpN stimulation, either of the whole SpN plexus with the pen-arterial cuff, or stimulation of few fascicles with a smaller cuff, generated an eCAP with a specific latency dependent on the distance between stimulating and recording sites (FIG. 3B). The range of conduction velocities of the different components of the eCAP is shown in FIG. 3C. The stimulation of either the whole plexus or few fascicles generated an eCAP with an average speed below 1 m/s (FIG. 3C). This conduction velocity is in line with histology findings in the characterization data below that describe the SpN being an unmyelinated nerve. The relationship between current amplitude and pulse duration necessary to elicit an eCAP either stimulating the whole plexus or few fascicles in shown in FIGS. 3D and 3E (respectively). When stimulating the whole plexus with a peri-arterial cuff the threshold of the nerve response was found between 7.692 and 15.58 μC/cm.sup.2/phase. When stimulating few dissected fascicles with a smaller cuff the threshold was found to be between 5.796 and 11.594 μC/cm.sup.2/phase. In both cases the threshold value of current density for eCAP recording was lower at shorter pulse width (PW).

[0283] SpN biphasic stimulation for 1 minute at 10 Hz and 400 μs PW above a specific current threshold consistently caused transient blood flow reduction within the distal SpA as measured via a perivascular flow probe. There was a clear dose-response relationship between delivered current and flow reduction: the higher the amplitude the stronger was the observed reduction in blood flow (FIG. 4A). The blood flow change threshold, defined as a 5% change in mean SpA blood flow (mSpA BF) compared to pre-stimulation baseline, was observed around 4.5 mA (with a 400 μs PW) and around 12 mA (with 200 μs pulse width) (FIGS. 10B and 10C). When calculating the charge density per phase of the threshold to cause blood flow changes the value was very similar: about 13.8 μC/cm.sup.2/phase at 400 μs and 18.46 μC/cm.sup.2/phase at 200 μs. Stimulation with 12 mA and 400 μs PW (36.9 μC/cm.sup.2/phase) caused a mean maximum BF reduction in the SpA of about 40% from baseline values.

[0284] In parallel, recording of the blood flow within the splenic vein (SpV) was recorded by using a Doppler flow probe placed at the splenic base, where the vein leaves the splenic hilum. Interestingly, stimulation (symmetric biphasic pulses, 400, 10 Hz for 1 minute) caused an increase in the mean SpV blood flow (mSpV BF) that was current amplitude dependent. Stimulation with 12 mA and 400 μs PW (36.9 μC/cm.sup.2/phase) caused a maximum increase of about 200% when compared to baseline mSpV BF. The transient reduction of mSpA BF was also accompanied by a transient increase in systemic mean arterial blood pressure (sMABP). This increase (in average between 1-6 mmHg) from baseline correlated again with the stimulation intensity (FIG. 4E). Consistent sMABP changes were observed with stimulations causing a 20-30% drop in the SpA flow. In contrast, HR was only minimally affected (<3 bpm changes, either increase or decrease), but more consistently only with high stimulation amplitudes (>45 μC/cm.sup.2/phase causing 3-10 bpm changes) (FIG. 4G). SpN stimulation did not affect respiratory rate (RR) in the conditions tested.

[0285] The changes observed in mSpA BF, sMABP, HR, RR during a 1-minute stimulation (symmetric biphasic pulses, 10 Hz, 400 μs PW) at different current amplitudes (1-50 mA, corresponding to 3.076-153.8 μC/cm.sup.2/phase) are summarized in FIG. 4F. In FIG. 4F, it is possible to observe how the magnitude of these changes was correlated with the recording of an eCAP (black line and circles) from the SpN. The higher was the number of fibers recruited (measured as eCAP % over the maximum recorded response) the stronger was the reduction in mSpA BF and the other associated changes.

[0286] Direct stimulation of discrete SpN bundles dissected off the SpA (using a 500 μm diameter cuff) evoked similar changes in the mSpA BF, sMABP and HR. These changes, occurring during a 1 minute (symmetric biphasic pulses, 1 Hz, 400 μs PW) and different current amplitudes (0.1-2.5 mA, corresponding to 3.86-96.61 μC/cm.sup.2/phase), are summarized in FIG. 4G. Even in this case the associated changes were dependent on the proportion of fibers (eCAP shown in black) recruited by the stimulation. The maximum eCAP (and therefore maximum changes) was obtained at about 153 μC/cm.sup.2/phase when stimulating the whole plexus and at about 70 μC/cm.sup.2/phase. The magnitude of the changes when stimulating few fascicles were lower than those obtained when stimulating the whole plexus, as expected since the total number of fibers stimulated was lower and the frequency was lower.

[0287] Blood flow changes in the mSpA were also affected by different frequencies of stimulation. When stimulating (symmetric biphasic pulses, 400 μs PW for 1 minute at about 36.9 μC/cm2/phase) at different frequencies (between 0.25 and 100 Hz), 30-50 Hz reliably caused the strongest blood flow reduction in the SpA (FIG. 5A). Above 50 Hz (between 70 and 100 Hz) the reduction in BF was in fact smaller, in the range of reductions obtained with a 10 Hz stimulation (FIG. 5B). The changes in mSpV BF, sMABP and HR were also found to be dependent on the frequency of the stimulation applied. The strongest effects were again observed between 30 and 50 Hz (FIGS. 5C to 5D).

[0288] This was once again observed when maximally (around 70 μC/cm.sup.2/phase) stimulating only few fascicles dissected off the artery. A stronger reduction in mSpA BF occurred already at lower frequencies (1 Hz and below), because of the higher recruitment of nerve fibers compared to the stimulation amplitude used for the whole plexus during the frequency analysis. Consistently however, the maximal reduction was observed between 30-50 Hz (FIG. 5D).

[0289] In order to further confirm that the observed changes in SpA BF were due to direct neuronal activation (rather than stimulation of smooth muscles) Lidocaine (2% lidocaine hydrochloride solution) was applied locally around the implanted SpN cuff (either the pen-arterial cuff or the cuff for dissected fascicles). Lidocaine is a specific blocker of fast voltage gated Na+channels. Lidocaine was able to block the changes in SpA BF. Further, mechanical occlusion of the SpA, able to reduce the BF up to 80%, did not cause any change in sMABP or HR. In addition, transection of the central end of the SpN (proximal to the cuff) did not abolish stimulation effects on SpA blood flow, sMABP and HR. Also the transection of the SpN within the GEP and SG segments did not prevent these changes. Interestingly, all these effects were only abolished when the peripheral end of the SpN (distal to the cuff) was cut. All these data suggest that the changes in SpA BF and SpV BF were neuronal driven and related to the constriction of the SpA as well as the contraction of the spleen capsule. On the other hand, the changes in sMABP and HR were probably not due to the activation of a neuronal pathway towards the brain but to the increase outflow of blood from the spleen towards the heart.

[0290] In few animals, SpA blood flow changes during stimulation was also monitored using intra-operative ultrasonography at the splenic hilum. After identifying the SpA by color Doppler, the change in BF was monitored as Doppler signal as shown in FIG. 7. During stimulation at 10 Hz, a reduction in BF could be easily observed as indicated by the changed amplitude and shape of the flow traces.

Discussion

[0291] Splenic nerve stimulation was associated with transient local changes in mSpA BF and mSpV BF as well as splenic contraction. These changes were due to the direct activation of the SpN, rather than direct stimulation of the smooth muscles of the SpA. Spleen contraction during SpN stimulation has been previously reported also in other species [16]. The observed change in mSpA BF was very consistent between animals. The variation was probably mainly due to different fitting of the cuff around the SpN plexus in different animals. Changes in SpA BF could be easily monitored via non-invasive ultrasound and therefore could be used as a marker to assess effective stimulation of the SpN also in a clinical setting.

[0292] The transient changes observed during SpN stimulation were shown to be amplitude and frequency dependent. During a minute of stimulation at different current amplitudes, the strongest mSpA BF reduction was observed at the highest current amplitude tested that also corresponded to the peak of the recorded eCAP. This was true when stimulating the whole SpN plexus (with a peri-arterial cuff) or when stimulating only few fascicles placed within a smaller cuff. The difference in the total charge density needed to obtain maximum eCAP from the SpN plexus and from SpN fascicles could be explained by the partial coverage of the plexus with the 2.5 mm cuff used. In most of the pigs in fact this cuff resulted only in a 270-300 degrees of circumferential coverage. When cuffing only few fascicles of the SpN dissected off the artery the coverage was almost total. Therefore, in order to limit charge density needed to obtain optimal recruitment of SpN fascicles, optimal circumferential coverage of the artery will be needed.

[0293] The strongest changes (in mSpA BF and sMABP) were observed at frequency between 30 and 50 Hz. Although the total number of pulses delivered could be an important factor in determining the magnitude of this changes, it is true that when comparing changes occurring with the same number of pulses delivered at different frequencies, 30-50 Hz range still caused the strongest changes. This could be explained with previously reported data showing that maximum release of NA from the cat spleen was observed at 30 Hz [17,18]. Higher release of NA could explain the higher magnitude of the changes observed in this stimulation range.

[0294] Optimization of the Signal Parameters

[0295] Materials and Methods

[0296] In order to develop an optimized stimulation paradigm, several signal parameter settings were tested in the pigs mentioned above using the materials and methods described above.

[0297] The optimization of parameters was focused at reducing off target effects and increasing efficiency of the SpN response. In particular, since the systemic changes caused by SpN stimulation were related to the local constriction of the SpA and contraction of the spleen, parameters able to minimize these changes could represent optimal paradigm to be transferred in chronic studies and clinical studies where systemic effects (e.g. changes in sMABP and HR) are not ideal.

[0298] Results

[0299] Repetitive stimulation of the SpN was found to cause fatigue in the nerve fibers at certain frequencies. This effect consisted of two characteristics: i) a reduction of SpN eCAP amplitude and ii) slowing of SpN conduction velocity. Both these characteristics were observed at frequencies higher than 1 Hz, with fatigue effects increasing with increasing frequency. Stimulation of the SpN at 10 Hz continuously for 1 minute caused, in fact, adaptation in the recorded response, resulting in a decreased amplitude of the eCAP over time (FIG. 8A), and a reduction of conduction velocity of each of the eCAP recorded peaks. This effect was stronger (in magnitude) and faster at higher frequencies. Stimulation of the SpN at 30 Hz continuously for 1 minute, for example, caused a faster and stronger reduction in both eCAP amplitude (FIGS. 8A, 8B) as well as conduction velocity. After 60 s of stimulation, 10 Hz (total of 600 pulses) pulses caused a reduction of about 60% of the eCAP maximum amplitude while 30 Hz pulses (total of 1800 pulses) resulted in about 80% reduction (FIGS. 8C, 8E). When the SpN was stimulated for 1 min with 1 Hz, instead, the reduction in eCAP over time (FIGS. 8A and 8C) was very small and no significant reduction in conduction velocity was observed. When comparing the reduction of eCAP amplitude over the same number of pulses, 10 Hz and 30 Hz still produced the faster and stronger reduction (FIG. 8C). This fatigue effect over repetitive stimulation of the SpN could be reduced by periodically switching OFF and ON the stimulation. When stimulating the SpN with a burst paradigm, for example giving 5 pulses delivered at 10 Hz, every 5 s, the reduction of eCAP and conduction velocity was abolished (FIGS. 8C to 8F).

[0300] In addition, when stimulating at 10 and 30 Hz continuously, a rapid increase in eCAP amplitude was observed within the first 5-20 pulses. This phase preceded the successive reduction in amplitude and velocity (FIGS. 8C and 8E).

[0301] As described previously the frequency of the stimulation not only impacted the response of the nerve but it also affected differently the physiology. Frequencies between 30 and 50 Hz caused the strongest changes in mSpA BF that drives the changes in sMABP. Frequencies 1 Hz caused, in comparison, little changes in mSpA BF. Frequencies between 1 and 30 Hz caused changes of mSpA BF of increasing magnitude. By selecting a high current amplitude (in order to recruit most of the SpN fibres), a 10 Hz, biphasic, symmetric 60 s stimulation is sufficient to cause, at least, a 50% maximum reduction in the mSpA BF. The stimulation of the SpN with the same biphasic symmetric pulses and current amplitude but with 1 Hz frequency, caused a reduction in mSpA BF about 40% lower (decrease from around 70% to around 50%) than the one generated with the 10 Hz stimulation (FIG. 9, FIGS. 8F and 8G).

[0302] The same was observed when applying a burst stimulation (5 pulses at 10 Hz every 5 s) using biphasic pulses and the same current amplitude (FIG. 9). Importantly, when the stimulation was applied as biphasic, asymmetric pulses (and same current amplitude) each of the stimulation paradigm resulted in a lower reduction of mSpA BF when compared to their respective biphasic, symmetric paradigm (FIG. 9). Even in this case 1 Hz and burst stimulations produced the lowest changes in mSpA BF.

[0303] Herein lies a description of the opportunity space to provide optimised neural activation, in the absence of ‘nerve fatigue’, while provide options for effectively stimulating physiological changes (high frequency), or to avoid them (low frequency or burst frequency), depending on the target profile of the therapy. During surgery it may be very effective to use blood flow changes to induce easily visualised target engagement profiles, to prove the device and therapy are appropriately positioned and of suitable amplitude. Then a switch to burst or low frequency stimulation will ensure ideal nerve engagement for efficacy in splenic engagement, while avoiding clinically consequential effects of continual changes in blood flow in awake patients.

Discussion

[0304] Parameters for optimal stimulation (and therefore efficacy) should in principle i) generate a very consistent and maintained amplitude response of the nerve, ii) deliver as many pulses as needed (to obtain the therapeutic effect) in the shortest possible time window in order to reduce energy requirement and discomfort to patients and iii) have the lowest spectrum of off target effects. The stimulation of the SpN at frequencies higher than 1 Hz showed a clear activity-dependent variations in response (eCAP) amplitude and conduction velocity. This effect has been previously observed in other unmyelinated nerves, in both rodents and humans [19,20,21,22]. During regular stimulation at 0.25 Hz, conditioning pulses intermittently interposed at varying inter-stimulus intervals, unmyelinated fibers showed a progressive reduction of conduction velocity dependent on the number of interposed stimuli [20]. The same effect was observed with stimulation of the SpN, where repetitive stimulation caused reduction in conduction velocity as well as reduction in eCPA amplitude. This effect has been previously called “subnormality” of action potential conduction.

[0305] When the SpN was stimulated continuously at 10 Hz, a short period of increased response has been observed followed by the slowing and reduction of the response phase. This other period has been also described before and termed “supernormality”. Supernormality and subnormality of action potential conduction are probably due an after-depolarization and a subsequent period of hyperpolarization of the membrane when pulses are delivered within a short time window (below 1000 ms from one another) [20,21,22]. When recording eCAP, this membrane changes in the axons that cause an increase in the current activation threshold, resulted in a lower amplitude of the recorded signal. This was not simply an effect of the slowing of the conduction velocity: the recorded eCAP shifted to longer latencies and decreased in amplitude without spreading in total width. In fact the measured AUC was smaller over time. However, it still possible that the observed reduction in eCAP amplitude was due to a de-synchronization of the action potentials that then cancel each other when recording a compound response. Single unit recording would be ideal to demonstrate that the SpN is really subjected to this activity-dependent changes.

[0306] It has been shown here that low frequency stimulation (1 Hz or below) or burst stimulation (5 pulses at 10 Hz, every 5 s) caused limited or no fatigue on the SpN as well as caused the lowest off target effects. To date these two paradigms, delivered with biphasic, asymmetrical pulses, represent the optimal stimulation pattern for the SpN for treatment of immune-mediated conditions (e.g. inflammatory disorders). Frequencies higher than 1 Hz, induce changes in blood flow and pressure, and could be useful in identifying nerve-target engagement during electrode positioning.

[0307] Effects of Continuous Electrostimulation in In Vivo LPS Porcine Model

[0308] Materials and Methods

[0309] Animals

[0310] A total of 23 pigs (weight 65-70 Kg) were used for this section of the study.

[0311] General Design

[0312] Pigs were terminally anesthetized and split into the following five groups: sham (implanted with electrodes but not stimulated), Dexamethasone (the SpN was accessed and then animals were injected with Dexamethasone at −2 and 0h), LVNS (pigs were implanted on the cervical LVN), eLVNS (pigs were implanted on the cervical LVN, that was ligated and cut distal to the cuff electrode and the efferent stump only stimulated), and SpNS (pigs were implanted on the pen-arterial SpN).

[0313] The implanted devices of the LVNS, eLVNS and SpNS pigs were stimulated continuously from −2h to +1h (relative to the injection of LPS) at 1 Hz. E. coli-derived LPS was administered at a dose of 0.25 mg/Kg to all groups at 0 h. Dexamethasone was used as positive control.

[0314] Peripheral venous blood was collected for 2 h prior to LPS injection (baseline), and then every half hour up to 4 hours post injection. For all of these time points, cytokine analysis (TNFα and IL-6), was performed by commercially available ELISAs.

[0315] Results

[0316] Cytokine quantification. For all groups, LPS injection resulted in a significant increase in TNFα levels in all post-injection samples compared to baseline (pre-LPS injection), with the peak response observed at approximately 1 hour post injection (FIG. 11A). A similar trend was observed for IL-6. IL-6 response peaked approximately 2.5 hours post injection.

[0317] Cytokine levels were compared between groups by calculating the area under the curve (AUC) from between −2h to +4h post-injection. TNFα levels were reduced around 30% at 1.5 hours in SpN as well as LVNS and eLVNS groups when compared to Sham. For IL-6, there was a reduction in both the peak response value (FIG. 11E) and the AUC (FIG. 11F) in SpNS group. Similar reduction was also observed for LVNS and eLVNS groups compared to the sham control.

Discussion

[0318] The administration of LPS in vivo to mimic an inflammatory response provided a good model to test the efficacy of SpN stimulation. The administration of LPS (0.25 mg/Kg) in 65-70 kg pigs caused upregulation of cytokines (TNFα and IL-6) in the blood of all the animals tested. In particular, TNFα reached a peak value of about 5 ng/ml at 1h post injection while IL-6 peaked at around 0.5 ng/ml at 2.5h post LPS injection.

[0319] Therefore, this model provides the proof that SpN stimulation is able to modulate the response to an inflammatory stimulus, and shows that prolonged stimulation of SpN reduces the levels of pro-inflammatory cytokines, as seen in particular by the reduction of IL-6. This is likely to be beneficial for reducing inflammatory responses in subjects, particularly in light of recent evidence showing the vagus nerve stimulation is beneficial in the treatment of autoimmune disorders.

SUMMARY

[0320] In summary, it was found that neural stimulation of a nerve supplying the spleen, and in particular, the splenic arterial nerve, showed pro-survival effects in an in vivo LPS animal model. Hence, stimulation of the neural activity of splenic nerves can be particularly useful for treating inflammatory disorders.

[0321] Characterization of the Splenic Arterial Nerves in Pig

[0322] Materials and Methods

[0323] Gross anatomical studies of the spleen with related organs were performed in 12 female pig cadavers (body size 22 to 120 kg) within 1 hour of euthanasia. The following measurements were made: length and width of the spleen; length of the celiac artery (from the aorta to the branching in to the left gastric and splenic arteries); length of the splenic artery (SpA) (from the branching of the celiac artery to entering the splenic parenchyma); SpA diameter measured 1 cm distal to the celiac artery and at the splenic hilum; distance from pancreas to the spleen; distance from pancreas to the splenic lymph nodes. Also, the number and course of the abdominal vagal branches, celiac ganglion, splanchnic nerves and splenic nerves were recorded. The SpA with associated splenic nerves were processed for Haematoxylin and Eosin (H&E) histology.

[0324] The spleen with intact vasculature and innervation was harvested from 12 female pig cadavers (body weight 22 kg, n=6; body weight 45 kg, n=6). All tissues were harvested within 1 hour of euthanasia, and were immediately fixated in 10% neutral-buffered formalin. The SpA with an intact perivascular neuronal network was sectioned every 5 mm from the origin at the bifurcation of the celiac artery, to the splenic hilum. This resulted in 5 sections, defined as the Bifurcation; the Proximal SpA; the Middle SpA; the Distal SpA and the Hilum location. The proximal SpA section corresponds to the location for cuff placement in the following electrical stimulation study discussed above.

[0325] At each of these five locations, sections were processed for routine H&E staining. The Proximal, Middle and Distal SpA sections were also processed for immunohistochemistry and for semi-thin sectioning and staining with osmium tetroxide and toluidine blue.

[0326] Digital images of the H&E stained sections were acquired at 2× magnification and appropriate software (Image J 1.50i) was used for histomorphometric analysis as detailed below. After manually selecting every single nerve fascicle by using the ROI manager function, the number of pen-arterial nerve fascicles were counted and the fascicle sizes assessed by measuring minimum Feret's diameter (μm).

[0327] The total nerve area (in μm.sup.2) was calculated, and the pen-arterial fascicle distribution was quantified by assessing the percentage of the arterial circumference in which fascicles were identified, defining 360 degree distribution as 100%. The distance from each fascicle to the external arterial wall was measured by drawing the shortest possible perpendicular line from each fascicle to the arterial wall. Splenic artery external and internal diameters were measured at the proximal, middle and distal SpA locations.

[0328] Double staining with tyrosine hydroxylase (TH) and acetylcholine transferase (ChAT) was used for assessing neuronal phenotype. By counterstaining with neurofilament 200 (NF200) and the nuclear stain 4′,6-diamidino-2-phenylindole (DAPI), NF200-TH double positive nerves were considered sympathetic, while NF200-ChAT double positives were considered parasympathetic nerves. In order to determine the proportion of efferent versus afferent nerves, the same locations were double stained with the efferent marker TH and the afferent marker calcitonin gene-related peptide (GCRP). Two different digital images were randomly captured at 20× magnification from each nerve, and pseudocolored composites generated using appropriate software (AxioVision LE64).

[0329] Myelination of SpN axons was assessed by immunofluorescent staining as well as from semi-thin sections. Different portions of the SpA and SpN were stained with antibodies against Neurofilament and β-III Tubulin and Myelin Basic Protein (MBP). Pseudocolored composite images were generated using appropriate software as described above. Semi-thin sections were stained with osmium and toluidine blue. Digital images were acquired at 100× magnification and the number of myelinated and unmyelinated axons were manually counted in an area of 100×100 μm. This procedure was repeated 3 times per nerve, and the mean of these were used for further analysis. Also, this procedure was used for deriving axon density (number of axons/mm.sup.2).

[0330] All statistical analyses were performed with commercially available statistical software (JMP Pro 13.0.0). Due to non-normal distribution, all histomorphometric measurements were compared between the different pig sizes and SpA locations using the Wilcoxon rank-sum test. Statistical significance was defined as P<0.05.

[0331] Results

[0332] Neurovascular structures enters and leaves the spleen along the visceral surface only. Specifically, the first major abdominal branch of the aorta, the celiac artery, bifurcates into the hepatic artery, the SpA and the LGA (FIG. 10). The SpA enters the spleen at the hilum, which is located a few centimeters distal to the splenic base. At the hilum, the SpA immediately bifurcates into one dorsal branch coursing towards the splenic base, and one ventral branch running along the visceral surface towards the splenic apex. The left gastroepiploic artery arises from this ventral SpA branch approximately at the transition between the middle and the distal ⅓ of the spleen.

[0333] At the splenic base, the dorsal SpA branch divide into several smaller arteries identified as the short gastric arteries, which courses towards the greater curvature of the stomach. Although these arteries are considered terminal branches of the SpA, they are capable of providing collateral blood supply to the spleen by anastomoses with branches of the LGA and the left gastroepiploic arteries. The SpV runs parallel to the SpA along the visceral surface of the spleen, from the apex to the hilum. After leaving the splenic hilum, the SpV courses closely adhered to the SpA for a short distance until it travels in a medial direction to drain into the hepatic portal vein, which in turn drains into the caudal vena cava. This leaves a small space in which the artery and the vein run separated by a few millimeters of soft tissue. This area, which is immediately distal to the bifurcation of the celiac artery into the SpA and LGA, has been identified as the optimal interface point for the following functional studies. At this location, the SpA diameter is 1.5-3 mm in the 30 kg animal; 2-4 mm in the 60 kg animal and 5-8 mm in the 110 kg animal.

[0334] The SpN consist of a plexus of fibers running along the SpA towards the splenic hilum. It is difficult to establish the origin of these nerves, although fibers can be seen arising from the CG which is located immediately caudal to the bifurcation of the celiac artery into the SpA and the LGA. Data from previous studies conducted mainly in rodents, established that most of the SpN originates from the celiac and suprarenal ganglia. This has yet to be proven in large animal species.

[0335] It has also been hypothesized that a direct connection exists between the Vagus nerve (VN) and the SpN, though the celiac ganglion. The subdiaphragmatic VN was seen crossing the lateral aspect of the celiac artery, and coursing towards the CG and the left adrenal gland. Some fibers coming from the subdiaphragmatic VN was also seen merging with fibres within the SpN plexus. Although these fibers might continue on to innervate the pancreas, a direct connection between the VN and the spleen cannot be excluded based on these observations. In rodents, it has been demonstrated that a direct connection exists between the VN and the SpN, and also between the splanchnic nerves and the SpN via the CG. In the pig, part of the splanchnic nerves was seen running towards the adrenal gland and CG.

[0336] In rodent species, other nerves have been described to innervate the spleen in addition to the peri-arterial SpN; more specifically, an apical nerve has been described within the gastro-splenic ligament of rats and mice. This is a sympathetic nerve (TH+) possibly originating from the paravertebral sympathetic nerves, and runs towards the apex of the spleen within the gastrosplenic ligament.

[0337] All histological measurements are presented in Table 1. The SpN-SpA distance was the only measurement significantly larger in the 45 kg pigs versus the 22 kg pigs (at the middle SpA and distal SpA locations; P<0.001); therefore, for all the other measurements, data from all pigs were combined for statistical analysis. There was a reduction in number of pen-arterial nerve fascicles along the SpA from proximal to distal; there were statistically significantly more fascicles at the bifurcation versus all other locations (P<0.0001). At the splenic hilum, nerve fascicles were significantly larger than at the other locations (P<0.0001). The SpA external diameter was significantly larger at the proximal SpA location versus the middle and the distal SpA locations (P=0.0162 and P=0.0158, respectively). The SpN/SpA distance also decreased from proximal to distal; in the 45 kg pigs, the distance was significantly larger at the Bifurcation versus all other locations (P<0.001). Also in the 45 kg pigs, the SpN/SpA distance was significantly larger at the Hilum versus the Proximal, Middle and Distal SpA locations (P<0.008).

TABLE-US-00001 TABLE 1 Histological measurements of SpN and SpA in 12 female pigs. Location Proximal Middle Distal Bifurcation SpA SpA SpA Hilum SpN-SpA 22 kg, N/A 437.5 ± 344.3 180.3 ± 111.6 161.4 ± 105.4 N/A distance n = 6 (μm ± SD) 45 kg, .sup. 1185 ± 616.2* 476.9 ± 334.1 .sup. 284.6 ± 166.4.sup.¥ .sup. 382.9 ± 247.4.sup.¥ 592.7 ± 354.2 n = 6 Mean no. of 105.8 ± 32.7* 41.6 ± 16.5 29.5 ± 5.1  27.7 ± 5.6  23.8 ± 1.4  fascicles ± SD Mean Feret's diameter 144.8 ± 100.6 160.3 ± 108.0 142.8 ± 89.7  157.7 ± 98.7   228.2 ± 157.9* (μm) ± SD SpA Internal diameter 1020.0 ± 440.2  1163.8 ± 351.9  904.2 ± 304.1 690.7 ± 201.6 (μm) ± SD SpA External diameter 2020.7 ± 560.0  2255.4 ± 479.sup.Δ  1791.6 ± 386.8  1574.2 ± 296.9  (μm) ± SD Neuronal circumferential 93.6 ± 9.8.sup.Δ 76.6 ± 19.0 73.8 ± 16.1 distribution (% ± SD) .sup.¥Significantly larger in the 45 kg vs. the 22 kg pigs. *Significantly different from all other locations. .sup.ΔSignificantly different from the Middle and Distal SpA. Significance P < 0.05. N/A: Not available.

[0338] The circumferential SpN distribution was significantly higher at the Proximal versus the Middle and Distal SpA locations (P=0.02 and P=0.15, respectively). Also, fascicles were more uniformly circumferentially distributed around the SpA at the proximal location whereas at the middle and distal SpA, the distributional pattern was more bimodal with fascicle clustering on opposite sides of the artery.

[0339] In the pig nerves are found along both the short gastric and gastro-epiploic arteries within the gastrosplenic ligament (FIG. 12). These nerves seem to be a continuum of the main pen-arterial SpN plexus and runs towards (or from) the stomach. At this location immunohistochemical analysis was performed and it was found that the SpN at any location is TH+ and ChAT−. Interestingly along the main SpA nerve fibers positive to Calcitonin Gene-Related Peptide (CGRP) were identified, commonly used as afferent neuronal marker.

[0340] The number of nerve fascicles and fascicle size observed in these two regions is much smaller compared to those observed along the main SpA. The quantification of the number and relative diameter of the nerve fascicles along the main SpA and along the other different anatomical locations in 45-50 Kg farm pigs is shown in FIGS. 12E and 12F.

[0341] Further histochemical and immunohistochemical analysis showed that the SpN is composed by >99.9% of unmyelinated fibers. Toluidine blue staining of semi-thin sections, in fact, did not show myelinated axons. In line with this, staining for Myelin Basic Protein (MBP) revealed a very little number of positive axons (<0.01%). Both of the techniques assessing myelination revealed almost complete absence of myelin in the investigated sections of the SpN as illustrated in FIG. 9.

Discussion

[0342] The histological analysis performed here showed that the SpN constitutes a neurovascular plexus along the main SpA as well as short gastric and gastroepiploic arteries. The number of fascicles is unexpectedly high. Considering the average size of a SpN axon (ca. 2 μm in diameter) it is possible to calculate that the SpN plexus should contain (at maximum) a total of about 150K axons at the level of the main SpA (middle section). Part of these axons will innervate the SpA endothelium and part of these axons will instead enter the spleen and forms synaptic connections with either smooth muscles or immune cells at the level of the marginal zone between white and red pulp as well as within the white pulp as previously described in other species [23,24,25,26,27]. The number of axons seems high if it is considered that the human vagus nerve (that has the same size of the pig vagus nerve), which targets several organs in the body, is supposed to contain about 100 k axons. The high number of axons in the SpN could be related to the size of the spleen in the pig, which has a volume approximately 2-3 times bigger than the human spleen, and the length of the artery that the SpN is supposed to innervate. The number of fascicles and axons in the human SpN might be different.

[0343] The spleen of pigs (and other mammals, such as dogs) is also thought to contain a higher proportion of smooth muscle cells compared to the human spleen [28]. However, several papers have also shown that the human spleen is able to contract during stressful conditions, such as apnea and physical exercise [29,30].

[0344] The vascular organization of the splenic artery and vein is slightly different between pigs and humans. In the pig the SpA and SpV run in close approximation towards and from the spleen. Moreover, SpV and SpA do not present loops or convolutions like those observed in humans. Therefore, only a short (approximately 1-1.5 cm) segment of the SpA, close to the trifurcation point of the celiac artery, is better separated from the SpV. This segment of the artery was chosen as best intervention point in the stimulation studies above. The access to the neurovascular bundle at this location is, in fact, safer, thus reducing the chances to damage the nerves as well as artery and vein during dissections.

[0345] Human Data

[0346] Electrophysiological Characterization of Human Splenic Nerves:

[0347] Materials and Methods

[0348] Human SpN specimens One fresh harvested tissue from a donor patient containing the splenic neurovascular bundle NVB was preserved in organ transplant-suitable solution on ice for transportation. Upon arrival the specimen was placed in ice-cold Kreb's solution under a dissecting microscope, and a minimum of one discrete SpN fascicle per sample was carefully separated from the SpA and subsequently instrumented with two bipolar circumferential cuff electrodes (0.65 mm diameter, 5.5 mm length; CorTec GmbH) placed approximately 10 mm apart, to evoke and record CAPs. Fascicle electrode coverage was estimated to be 100% in all implantations.

[0349] Recordings Nerve activity was continuously monitored using an oscilloscope, and digitally recorded via a 1401 digital acquisition system and Spike2 software (Cambridge Electronic Design Ltd), with the sampling rate set at 20 kHz. Evoked CAPs were averaged (8 pulses) and the peak-to-peak amplitude of the averaged response quantified. The conduction velocity of the eCAP components was calculated from the measured distance between the stimulation site and the recording site and the latency of the eCAP signal (measured from the peak of the stimulation artefact to the peak of the eCAP).

[0350] Results Compared to the porcine samples, the human SpA presented with a more convoluted course as previously described (Michels 1942). Furthermore, the splenic NVB was embedded in extensive amounts of connective tissue and fat (FIG. 13A), making recordings from the entire circumference of the structure challenging. However, using a dissecting microscope, several nerve fascicles were visible and later confirmed as such by histological sections of the specimens (FIG. 13B). After instrumenting some of these fascicles with stimulating and recording cuff electrodes (FIG. 13A, upper and lower image), stimulation generated clear eCAPs (FIG. 13D, upper trace). To confirm the validity of the recording at the end of the experiment the fascicles were crushed between the stimulating and recording electrodes and attempts to re-record were made (FIG. 13D, lower trace). Typical recruitment curves were obtained when applying stimulations at specific pulse durations (e.g. 100, 200, 400, 800 and 1000 μs; PW) and increasing amplitude (FIG. 13E).

[0351] Calculated conduction velocities demonstrated typical values for unmyelinated fibres, where the range and average conduction velocity was 0.49 m/s, compared to porcine (0.7 m/s) and rat (0.72 m/s) SpN (FIG. 13F). In addition, the eCAP recordings of the human SpN showed a typical strength-duration relationship between current amplitude for nerve recruitment and pulse duration (FIG. 13G). Linear regression of the calculated charge density value for eCAP threshold recording showed slopes significantly different from zero (P=0.0084), with the lowest PW (100 μs) requiring 13.44 μC/cm.sup.2, and the longest PW (2000 μs) requiring 14.7 μC/cm.sup.2. Importantly, the slope in the charge density for the human SpN fascicles was found to be similar to the slope of the charge density for the porcine fascicles (FIG. 13H). In addition, the charge density requirement for nerve activation of the dissected human fascicles was about 1.5-2 times higher than the charge density required for activation of the porcine SpN fascicles at any PW (FIG. 13H).

Discussion

[0352] The human SpN has anatomical, morphological and electrophysiological characteristics similar to other mammals (porcine and rodent). The human SpN are composed of unmyelinated axons as confirmed by conduction velocities. It is therefore appropriate to assume that the stimulation parameters (frequency and waveform) optimized in the pig will be also suitable for the human splenic nerve. However, requirements for charge need to be calculated from the entire NVB.

[0353] Histomorphometric Characterisation of Human Splenic Anatomy

[0354] The objective of this study was to develop an understanding of the human splenic anatomy and estimate the approximate values of splenic neurovascular bundle (NVB) using histology (see Table 2). The study was performed on the splenic tissue received from transplant patients. Histomorphometric estimations for lumen diameter, arterial wall, fascicle diameter (mean Feret diameter) and the approximate distance of each fascicle from adventitia (outer splenic arterial wall) were calculated.

[0355] Materials and Methods

[0356] Five human splenic NVBs were provided from transplant patients at Addenbrooke's hospital, Cambridge, UK. The tissue was immersed in 10% neutral buffered formalin (NBF) as soon as possible post-excision. Photographs of the tissue were taken, with a ruler present for gross measurements (see FIG. 14A). The samples were divided in sequential blocks of 0.5 cm-1.5 cm for histology (see FIG. 14B). The tissue around the artery was retained for inclusion in the block. The sections were embedded and sectioned such that the same face of each block (i.e. proximal or distal to spleen) was sampled each time. The sections were usually 4-5 um thick and were stained with hematoxylin and eosin stain (H&E) (see FIG. 14C). Finally, a quality check of the tissue was performed by a pathologist and the glass slides were scanned at ×20. It should be noted that, as per literature, 10% of tissue shrinkage is accounted for. However, the artery diameter is representative of zero pressure. High amounts of adipose tissue was noted in all the samples received from transplant patients and the fascicles were found to be buried in a thick layer of adipose tissue.

TABLE-US-00002 TABLE 2 Estimated range for human splenic neurovascular bundle (~7 mm to 10 mm) Range of extravascular Total NVB Accounting for tissue (*Does not Lumen Wall + shrinkage of (based on account for Sample Lumen Arterial the tissue middle splenic pulsatile nature Number Diameter wall (+10%) arterial loop) of the artery) Sample 3.01 mm 5.02 mm 5.5 mm 3.5 mm 9 mm 308B X91165 Sample 3.92 mm 5.2 mm 5.72 mm 2.4 mm 8.12 mm 359B X91252 Sample 3.3 mm 4.93 mm 5.42 mm 3.8 mm 9.2 mm 377C X91287 Sample 2.76 mm 4.72 mm 5.192 mm 4.9 mm 10 mm 380C X91291 Sample 2.57 mm 4 mm 4.4 mm 2.5 mm 6.9 mm 382B X91299

[0357] For quantification purposes, the splenic tissue was divided into three parts: proximal, middle and distal. Each of these parts consisted of several sections. The proximal end is close to the celiac indicated with a suture in FIG. 14A and distal is close to the spleen. Both of these are unlikely to be the intervention site for neural interface placement. The middle part with loops would be the likely intervention site.

[0358] To summarise, as shown in FIG. 15, fascicle diameters are in the range of 20-400 um. For the fascicle spread approximately half of the nerve fibres were found in 0-1 mm region, 30% in 1-2 mm, 15% in 2-3 mm and the remaining in about 3-4 mm region.

[0359] Translational Charge Requirements from Porcine to Human Splenic Neurovascular Bundle

[0360] Materials and Methods

[0361] 3D Finite Element Model computer simulations were created using histology data from porcine and human splenic histology. This essentially comprised of splenic artery (lumen+arterial wall) and extravascular tissue. The ‘extravascular tissue’ is composed of ‘adipose tissue’ and ‘connective tissue’, with nerves embedded in the tissue. For porcines, a model with a split in the Cortec cuff (representing the in-vivo cuff) was used. For human models, cuffs with three arms structure were used. The diameter of the used cuff was 9 mm.

[0362] Considering the differences between porcine and human histology: the fascicles in porcine are evenly distributed around the artery and are in close proximity, whereas the fascicles in humans appear more dispersed; and b) the histology in porcine indicates negligible adipose tissue extravascularly, converse to substantial amounts in humans.

[0363] To translate the estimation of stimulation parameters from porcine to human, modeling was performed in the following two phases:

[0364] Phase (a): Development of 3D Finite Element Models (FEM) in Sim4Life simulation tool.

[0365] Sim4Life was used to develop representative nerve and artery models (based on histology and image quantification), cuff and electrodes (specifications defined by CAD) and 3D voltage fields.

[0366] Phase (b): Analysis of FEM solutions in the same tool. Sim4Life was used to interpolate voltage along axons using Sundt nerve model [31], and axon simulations estimated the strength-duration and population recruitment curves.

[0367] Results

[0368] FIG. 16A represents the in-vivo acute data from porcine splenic neurovascular bundle from five animals. The range from five animals for charge requirements is estimated to be approximately 20-160 uC/cm.sup.2 at <50 mA, 400 us and 10 Hz. For the third animal represented in grey the charge requirements are approximately 100 uC/cm.sup.2 at 30 mA, 400 us and 10 Hz, which correlates well with the simulated data in-silico (see FIG. 17A). Using the correlation of in-silico vs in-vivo as a validation for the computational model in porcine, the charge requirements were translated to human splenic neurovascular bundle using histology sections for two pulse widths. The data is presented in FIGS. 17C-D and Table 3.

TABLE-US-00003 TABLE 3 Charge estimates for human models for two pulse widths i.e. 400 us and 1 ms pulses Pulse Width Charge estimates Charge estimates 400 us 1000 us (μC/phase/cm.sup.2) (μC/phase/cm.sup.2) % recruited (Approx.) (Approx.) Threshold 79 70 10 130 110 30 170 150 50 225 200 80 422.8 335 80-100 450-1300 350-1100

[0369] It is estimated that the charge requirements in human acute models for a recruitment of 100% can potentially vary from approximately 80-1300 μC/cm.sup.2 (using 400 uS pulse widths, 12 mm.sup.2 surface area) and 70-1100 μC/cm.sup.2 (using 1 ms pulse widths). Approximately 70% of the recruitment is indicated under 350 μC/cm.sup.2. The additional 30% recruitment requires exponential increase in charge requirements beyond what is likely accommodated for by an implantable device. For example, it can be seen that a recruitment of 100% can potentially vary between 70-1300 μC/cm.sup.2, between 70-450 μC/cm.sup.2 for 80% recruitment, between 70-250 μC/cm2 for 50% recruitment, and between 70-170 μC/cm.sup.2 for 30% recruitment.

Discussion

[0370] The nerves fibres in the humans are more dispersed in comparison to porcines. The range of the fascicle spread around splenic artery as indicated by histology profiling can be in the range of approximately 1-3 mm. The histomorphomteric data was further used to optimise the stimulation parameters and translate the charge requirements from porcines to humans using computatational modelling tools. Using Sundt c-fibre model the charge requirements for humans is indicated to be in range of approximately 70-1000 μC/cm.sup.2 for hundred percent recruitment.

[0371] Ex-vivo electrophysiological study of human splenic nerves

[0372] The objective of this study was to estimate indicative stimulation parameters of human splenic nerves in order to de-risk and optimize the biological efficacy and reproducibility of stimulation parameters of the electrical signal for use in humans, in particular for stimulation of a human splenic nerve. The study was performed using ex-vivo using human splenic samples.

[0373] Materials and Methods

[0374] FIG. 18A shows an example of fresh splenic sample from a 63-year-old female donor (it is noted that the range of age of donors making up the data described below is 23-63 years). The sample, approximately 15 cm in length, was placed in a petri dish, and the splenic neurovascular bundle (SNVB) was then carefully surgically isolated from excess adipose tissue and splenic vein under a microscope. The dots on the sample indicates the top part of the splenic artery used in order to maintain the orientation of the sample. The sample was tortuous and seemed to have loops. A few splenic nerves were carefully isolated distally for the purpose of recording eCAPs.

[0375] An isolated fascicle was used as a control and cuffed with a smaller diameter Cortec Cuff electrode (500 μm diameter) for recording and stimulation, as shown in FIG. 18B, (II). A bigger periarterial cuff of approximately 6 mm diameter was placed on the neurovascular bundle (see FIG. 18B, (I)). Subsequently, the tissue with the cuff was moved into the recording chamber which was constantly circulated with fresh, oxygenated and warm Kreb's solution (34-36 degrees Celsius). The stimulation cuffs were connected to a DS5 instrument (current stimulator) and recording cuff was connected to a bioamplifier (CWE, USA) as indicated in the schematics (see FIG. 18C, FIG. 18D). For stimulation, a bipolar configuration with monophasic pulses were used. The schematics of the evoked compound action potential is represented in FIG. 18E.

[0376] Results

[0377] The nerve viability on isolated nerves was verified with a smaller 500 μm cuff electrodes, used as a control. The current strength-pulse width results from stimulation in eight human SNVB samples stimulated with 6 mm cuff demonstrates that the use of a 2 ms pulse width permits a 2.5- to 3-fold reduction of the stimulation threshold of pulse height for a 2.5-fold increase of pulse width i.e. from 0.4 to 2 ms (see FIG. 19A).

[0378] Interestingly, 400 μs pulse width, which seems to be an optimum stimulation parameter in the porcine in-vivo study, did not experimentally prove optimum in the case of human ex-vivo and in-silico tissue preparations. The mean pulse height from N=6 in acute porcine study was approximately 3.5 mA (see FIG. 3D), whereas in humans it was found to be at an average seven-eight times higher at approximately 25 mA. The reason why trade-off between pulse width and pulse height is important is to inform an optimum output level for implantable stimulator design and electrode charge injection capacities. With reference to FIG. 19A, 3 ms also seems a suitable pulse width, however, there is an increase in charge density with negligible decrease in pulse-duration. A significant increase in charge density is observed at and above 5 ms.

[0379] An increase in frequency from 1 Hz to 10 Hz indicates a reduction in eCAP amplitude and is indicative of nerve fatigue (see FIG. 24). Thus in this instance re-confirming porcine data assumptions on frequency. Nerve recruitment curves from individual donor samples at different pulse width of 0.4, 1 and 2 ms are illustrated in FIGS. 19B, 19C, and 19D respectively. The compound action potentials are normalised with respect to the maximum eCAP amplitude response recorded on the oscilloscope. DS5 instrument has a limitation of 50 mA in amplitude, which was not enough to recruit 100% nerves at 0.4 ms (as seen in FIG. 19B). Thus, moving to 1 m and 2 ms pulse width effectively proves to be a more ideal trade-off. It is estimated that the charge requirements in human ex-vivo sample for 100% can be as high as 400 μC/cm.sup.2 (assuming a 0.12 cm.sup.2 total electrode surface area) as can be seen in FIG. 19D. Based on assumptions of fibrotic encapsulation modelling, and the effects we have seen in pre-clinical animal models, a right shift effect is observed (as also seen in literature such as in [32]) by factors of ×1.5, ×2 and ×3, for example, on the charge requirements in chronic. This can be seen in FIG. 20, where our estimation of charge requirements in chronic clinical scenario could be as high as approximately 100 μC (850 μC/cm.sup.2). A similar trend of charge requirements is observed from in-silico results for both 0.4 and 1 ms pulse width.

Discussion

[0380] It was found that for increasing pulse width, particularly pulse widths greater than 1 ms, a decrease in the pulse height threshold needed to trigger an action potential in a human splenic nerve is observed.

[0381] This is a surprise based on the porcine model which showed the optimum pulse width to be far lower, at 0.4 ms. Lower pulse height thresholds are generally desirable because the biological efficacy and reproducibility of the stimulation parameters for use in humans is improved.

[0382] It has also been found that at a pulse width of 3 ms or above (3-5 ms shown in data) there is no further decrease in pulse height, whereas there is an increase in charge density. Therefore, the strain of the electrodes outweighs the benefits seen in the IPG beyond a pulse width of 3 ms. Between 2 ms and 3 ms, there is a negligible decrease in pulse height threshold but the amount of charge density required increases. Therefore it may be desirable to use a pulse width of less than 3 ms in humans. Pulse width around 2 ms offer an optimal trade-off between ensuring a low charge density being required, and a low pulse height being required for the stimulation of a human splenic nerve.

[0383] It is estimated that the charge density per phase requirements in human ex-vivo sample for 100% nerve recruitment can be as high as 400 μC/cm.sup.2. However, it is expected that for chronic stimulation, the formation of scar tissue may reduce the nerve recruitment by a factor of between 1.5 and 3. FIG. 20 shows the 2 ms pulse width human ex-vivo data multiplied by a factor of 1.5×, 2× and 3×, and the change in recruitment based on the charge injected into the human splenic nerve. FIG. 20 suggests that up to 100 μC charge may need to be injected for recruitment of 100% nerves in humans in chronic scenario. This equates to a charge density per phase of approximately 850 μC/cm.sup.2 based on a 0.12 cm.sup.2 total electrode surface area. Accordingly, the charge density per phase required in order to achieve 100% recruitment of the human splenic nerve is expected to be up to approximately 850 μC/cm.sup.2 for a pulse width of 2 ms.

[0384] Human Chronic Model Stimulations

[0385] The purpose of this study was to determine the biological effect varying of interphase delay and pulse width. The study was conducted using a human chronic model simulation.

[0386] Materials and Methods

[0387] Hybrid electromagnetic (EM) and neuronal simulations were used to predict axonal recruitment in two representative image-based and 3D computational neurostimulation models of human and porcine splenic neurovascular bundle, for multiple variations of dielectric parameters of the nerve bundles, stimulus waveforms (0.4 ms, 1 ms and 2 ms biphasic pulses), and fibre diameters (0.5-1 mm). One representative cross section histological image of splenic neurovascular bundle for each species was segmented using iSEG within Sim4Life platform. Tissues were differentiated to identify vessel wall, blood, extra fascicular medium—internal and external to the electrode—and the endoneurium tissue within fascicles. The segmented tissue surfaces were extruded in 3D using extrusion functionalities. The bundle models were combined with cuff electrodes geometries, were surrounded by saline solution tissue to mimic experimental conditions, and fascicles were populated with multiple parallel axonal trajectories randomly distributed within each fascicle cross section.

[0388] EM simulations were performed using a FEM solver in the quasi-static approximation that handles anisotroμC electric tensors conductivity and support thin layer settings. FEM calculations were executed on unstructured meshes created on the model geometries, built within Sim4Life using adaptive criteria and mesh quality adjustment. The meshes were edited to extract patches at the electrode surface to assign flux density boundary conditions, and at the interfaces between fascicles and interfascicular tissues to define thin layers mimicking the perineurium. In order to execute transient neuroelectric simulations for a given set of stimulation conditions (fibre diameters, pulse waveform, temperature), the range of parametrised axon electrophysiology in Sim4Life was extended by a c-fibre model (Sundt Model) completing the functionality required to stimulate nerves featuring distribution of unmyelinated c-fibres with arbitrary fibre diameters. Sim4Life functionalities such as the automatic sweeping and titration procedure were used to quantify stimulation thresholds (e.g. the pulse height threshold), investigate strength-duration (SD) curves and perform sensitivity analysis e.g. with respect to dielectric properties of tissues or pulse parameters. The creation of neuroelectric models, the creation and the setup of hybrid EM-neuronal simulations, and the post-processing of the results was assisted by 1) Python scripts facilitating the flexible, parametrised generation of functionalised nerve models, 2) the assignment of heterogeneous tissue properties and anisotroμC electrical conductivities, 3) the creation of mesh and its editing, 4) the distribution of fibre models within fascicles, 5) the assignment of electrophysiological behaviour as well as for automised post-processing analysis, e.g. the quantification of stimulation thresholds, extraction of recruitment curves, identify location of spike initiation and latencies (time of first spikes) with respect to stimulus pulse-shape.

[0389] The image-based models of neurovascular bundles developed were adapted to include fibrotic tissue surrounding the electrodes and the insulating silicone to mimic the presence of a post-implantation fibrotic tissue. Hybrid EM-neuronal simulations were used to calculate the neuroelectric responses of electrophysiological models of individual unmyelinated C-fiber axons inserted within the fascicles of the bundles to quantify the stimulation thresholds (e.g. pulse height threshold) for initiation of the action potentials. From the calculated thresholds, recruitment curves were plotted for both acute and the chronic scenarios based on biphasic waveforms with different pulse durations (rdur) and interphase delays (Tinter). The results are based on the following principal assumptions: (i) the dielectric properties, the structure, and the composition of the fibrotic tissue are uniform across all simulations; (ii) the fibrotic tissue is homogeneous and isotropic; (iii) there is no distinction between the fibrotic tissue formed around the electrodes vs. the silicone; (iv) the position of the fascicles is kept constant moving from acute to chronic scenario. The diameter of the neurovascular bundle is also kept constant and 0.5 mm of interfascicular tissue has been replaced by fibrotic tissue layer.

[0390] Results

[0391] FIG. 21 shows comparisons of the recruitment curves calculated for the human model for acute and chronic stimulations with different parameterisations of the biphasic pulse waveforms. For the chronic case, it was found that the presence of the fibrotic encapsulation increases the pulse height threshold required to trigger the creation of an action potential, with the increase for a fixed pulse duration being smaller for larger interphase delays. The increase in pulse height threshold is dependent on the specific parameters of the biphasic pulse waveform. For instance τdur=1 ms, the pulse height threshold increase is 37% when τinter=0 ms (simulations Acute1msOms vs. Chronic1msOms) but is 29% when τinter=0.2 ms (simulations Acute1msOms vs. Chronic1 ms02 ms). Similar results were found for τdur =0.4 ms: the pulse height threshold increase is 49% (simulation Acute04 ms0 ms vs. Chronic04 ms0 ms) vs. 27% with τinter=0.2 ms (Acute04 ms0 ms vs. Chronic04 ms02 ms). The results for 0.1 ms interphase have also been demonstrated in the graph for both the pulse durations (Chronic04 ms0.1 ms and Chronic1 ms0.1 ms). The impact of the pulse duration on pulse height threshold increase is large, ranging from 133% for the comparison of biphasic pulses of 0.4 ms vs. 1 ms in the acute case (Acute1msOms vs. Acute04 ms0 ms). Importantly, these results are for fibre diameter 1 μm. The variations in pulse height threshold due to acute vs. chronic stimulations were also investigated for dependence on fibre diameter for fibers of 0.5 μm vs. 1 μm. It was found for the acute scenario, thresholds increase by approximately 80-90% for a fiber of diameter 0.5 μm compared to one of 1μm fiber. The studies have indicated that the pulse height threshold increases with decreasing fiber diameters and the pulse height threshold may be decreased by increasing the pulse duration. In particular, the interphase delay of 0.2 ms demonstrated a potential advantage of 5-10% over a 0 ms interphase delay. FIG. 23 shows the ex-vivo validation of these in-silico calculations, and beyond 0.3 ms no further improvement in threshold reduction is noted, thereby further illustrating 0.2 ms as an optimal interphase parameter.

[0392] The findings on pulse width in the ex-vivo preparations are further supported by this in-silico modelling data, as shown in FIG. 22. In particular, this figure shows that as the pulse width increases beyond 1 ms for a biphasic pulse train, the charge required to stimulate neural activity is reduced. Then, for pulse widths of 3 ms or higher, the charge required significantly increases.

Discussion

[0393] It was found that effects of interphase delay and pulse width are prominent. In particular, the interphase delay of 0.2 ms demonstrated a potential advantage of 5-10% over 0 ms interphase delay.

[0394] It is noted that these findings are supported by in-silico modelling data, as shown in FIG. 22. In particular, FIG. 22 shows that as the interphase delay of a biphasic pulse train is increased from 0 ms to 0.1 ms, the charge required to stimulate neural activity is reduced. It is further expected that as the interphase delay is increased beyond 0.1 ms, that the charge required to stimulate neural activity will reduce further and become closer to that required by a monophasic pulse train. Since it is not desirable to stimulate the nerve with a monophasic pulse train, a biphasic pulse train with an interphase delay greater than 0.1 ms is preferable.

[0395] Other ex-vivo studies in unmyelinated fibers have found that for interphase delays greater than 300 μs, no further reduction in pulse amplitude threshold is found. This is depicted in FIG. 23. Accordingly, the optimum interphase delay for stimulation of a human splenic nerve is likely to be between 100 μs and 300 μs, more particularly between 200 μs and 250 μs.

REFERENCES

[0396] [1] Medzhitov, Nature 454, 428-435 (24 Jul. 2008). [0397] [3] D. M. Nance, V. M. Sanders, Brain Behav Immun 21, 736. [0398] [4] H. H. Dale, H. W. Dudley, J Physiol 68, 97. [0399] [5] C. Cailotto et al., Neurogastroenterol Motil 24, 191. [0400] [6] M. Rosas-Ballina, K. J. Tracey, Neuron 64, 28. [0401] [7] G. Vida, G. Pena, E. A. Deitch, L. Ulloa J Immunol 186, 4340. [0402] [8] B. O. Bratton et al., Exp Physiol 97, 800. [0403] [9] D. Martelli, S. T. Yao, M. J. McKinley, R. M. McAllen, J Physiol 592(7), 1677. [0404] [10] D. Martelli, S. T. Yao, J. Mancera, M. J. McKinley, R. M. McAllen, Am J Physiol Regul Integr Comp Physiol 307, R1085. [0405] [11] D. Martelli, M. J. McKinley, R. M. McAllen, Auton Neurosci. 182, 65. [0406] [12] Koopman FA et al., Proc Natl Acad Sci USA, 19;60(29):8284. [0407] [12] Greenway et al., J. Physiol. (1968), 194, 421-433. [0408] [13] US 2006/0287678.[14] US 2005/0075702.[15] US 2005/0075701. [0409] [16] Schafer, E. A. and Moore, B., J Physiol, 1896. [0410] [17] G. L. Brown, J. S. Gillespie JS, J Physiol, 138:81-102, 1957. [0411] [18] A. G. Garcia, et al., J Physiol, 261:301-317, 1976. [0412] [19] Gee, M. D., Lynn, B., Cotsell, B., Neuroscience, 1996, 73, 3, 667-675. [0413] [20] Weidner, C., et al., J of Physiol 2000, 527, 185-191. [0414] [21] Lin, C. S., Mogyoros, I., Burke, D., Muscle Nerve, 2000, 23, 5, 763-770. [0415] [22] Stohr, M., J of Neurol Sci 1981, 49, 1, 47-54. [0416] [23] F. Tischendorf F, Biol Lat 9:307-342, 1956. [0417] [24] M. Fillenz, Proc R Soc Lond B Biol Sci 174:459-468, 1970. [0418] [25] I. G. Zelenova, Arkh Anat Gistol Embriol 60:88-90, 1971. [0419] [26] F. D. Reilly, Experientia 41:187-192, 1985. [0420] [27] D. L. Felten et al., J Neurosci Res 18:28-36, 118-121, 1987. [0421] [28] D. G. Onkar, J Morphol Sci 30:p. 16-20, 2013. [0422] [29] D. Bakovic et al., Clin Exp Pharmacol Physiol 32:944-951, 2005. [0423] [30] A. Lodin-Sundstrom, E. Schagatay, Aviat Space Environ Med 81:545-549, 2010. [0424] [31] Sundt D, et al., Journal of neurophysiology. 114:3140-53, 2015. [0425] [32] W M Grill J T Mortimer. Electrical properties of implant encapsulation tissue. Annals of biomedical engineering, 22:23-33, 1994