Abstract
Although sacral neuromodulation (SNM) is an established treatment for faecal incontinence, stimulation parameters have been derived empirically and only one frequency (14 Hz) is employed clinically. The aim of this study was to test a range of stimulation frequencies to establish an optimal frequency of SNM for maximum augmentation of anal canal cortical evoked potentials (EPs) in an animal model.
In female Wistar rats, anal canal EPs were recorded over the primary somatosensory cortex using a flexible multielectrode array, and the effect of SNM was studied. SNM was applied at 0·1–100 Hz and a frequency response curve plotted. The data were fitted to a quadratic equation.
The magnitude of potentiation of anal canal EPs caused by SNM depended significantly on stimulation frequency (P < 0·001). The frequency–potentiation relationship was parabolic in form, with a clear optimum at 2 Hz. The SNM must be applied for at least 3 min. The theoretical maximal potentiation predicted by the model was not found to be statistically different to actual data recorded (P = 0·514–0·814). The response depended on stimulation amplitude in an ‘all-or-nothing’ fashion. EPs were augmented when the SNM intensity was 0·5 times the motor threshold to tail twitch or greater, but values below this intensity failed to affect the EPs.
The effect of SNM in this animal model is governed principally by frequency, with an optimum of 2 Hz. If animal data can be translated to humans, optimization of SNM frequency may offer a clinically relevant improvement in the efficacy of SNM.
Surgical relevance
Sacral neuromodulation (SNM) for faecal incontinence currently employs stimulation parameters that have been derived empirically and may not be optimal. This study used an animal model of SNM and focused on its acute effect on anal canal cortical evoked potentials (EPs). It was found that SNM potentiated EPs, with a clear optimum at a frequency of 2 Hz. If this finding is applicable to the mechanism of action of human SNM, this suggests that there may be a clinically relevant improvement by reducing stimulus frequency from its typical value of 14 Hz to 2 Hz.
Introduction
Faecal incontinence is a common, embarrassing and socially disabling condition1–4. Sacral neuromodulation (SNM) is now a well established treatment option for patients with severe symptoms of faecal incontinence unresponsive to conservative therapies. SNM was developed initially to treat urinary incontinence and then applied to faecal incontinence5. Successful stimulation parameters used in urinary applications were transferred without adaptation to the colorectal field. These include a frequency of 14 Hz and a pulse width of 210 µs, while the intensity is set to a fraction of the motor threshold.
Many large prospective case series have reported success of chronic SNM in 74–86 per cent of patients6–10, and long-term results have been favourable11. This success is less impressive if expressed on an intention-to-treat basis. with systematic review data12 showing that only 58 per cent of patients had a greater than 50 per cent reduction in faecal incontinence episodes in the medium term. In addition, all studies acknowledge that some patients lose efficacy over the medium and long term13. These patients have their devices reprogrammed by changing stimulus amplitude and/or reversal of electrode polarity13. In contrast, changing frequency and pulse width has not been routine. Although two studies14,15 have demonstrated that changes in stimulation frequency (to higher frequency) and reductions in pulse width can have a short-term positive impact on SNM outcome, to the present authors' knowledge no study has systematically explored a range of frequencies to determine whether an optimum exists.
The mechanism of action of SNM is still unclear, although evolving human and animal data support improved cortical awareness of the anorectum16,17 via modulation of afferent somatosensory fibre pathways18. Specific changes in cortical activation following SNM have been demonstrated in patients suffering from urinary urge incontinence using positron emission tomography19 and pudendal evoked potentials (EPs)20. In patients with faecal incontinence, previously elongated pudendal EP latencies have been shown to decrease slightly after SNM, but only in patients with a successful outcome21.
The present authors have developed an animal model of SNM showing that 14-Hz SNM potentiates anal canal inputs to the rat somatosensory cortex18. Electrophysiological findings were supported by immunocytochemical evidence of increased levels of neuronal cell adhesion molecule, a marker of neuronal plasticity, after SNM18. In light of the relative ease at which neuromodulation can be studied in this animal model, the aim of the present study was to test a range of stimulation frequencies to establish whether there is an optimal frequency of SNM that augments anal canal cortical EPs.
Four objectives were addressed: exploration of the effect of stimulation frequency and duration in SNM through application of a fixed number of pulses at various frequencies (range 0·1–100 Hz) in order to isolate a pure frequency effect; investigation of the effect of SNM frequency applied for a fixed duration with the goal of expressing the frequency–response relationship as a mathematical model and calculating an optimal stimulation frequency; comparison of predicted and actual responses at the theoretical optimal frequency in order to validate the model; and investigation of the effect of different stimulus amplitudes at the calculated optimal frequency.
Methods
All experiments were carried out in accordance with protocols approved by the University College Dublin animal ethics research committee and licensed by the Irish Department of Health and Children (licence no. B100/4435). A total of 72 female nulliparous Wistar rats (weight 175–275 g) were used. Animals were kept at a 12-h/12-h light/dark cycle, and had free access to water and rodent standard diet.
General surgical preparation
Rats were anaesthetized using urethane (Sigma, Arklow, Ireland) in a 20 per cent solution at a dose of 1·5 g/kg intraperitoneally. The level of anaesthesia was monitored regularly throughout the procedure by means of the pedal withdrawal reflex to toe pinch and the corneal reflex. The femoral vein and artery were cannulated to administer fluids/urethane, to monitor the arterial blood pressure and for arterial blood gas analysis (RAPIDPoint® 400; Siemens Healthcare Diagnostics, Tarrytown, New York, USA). A tracheostomy and intubation were performed to ensure airway patency before the animal was placed in a stereotactic frame. On conclusion of each experiment, the rat was killed by anaesthetic overdose.
Recording of evoked potentials
Anal canal EPs were recorded over the primary somatosensory cortex. The head was placed in a stereotactic frame and the position of the anal canal representation localized (anteroposterior coordinate −0·6 mm, mediolateral coordinate +2 mm measured from the bregma18). A small craniotomy, approximately 4 × 4 mm, was performed over the right hemisphere using a small dental drill (RS 472–2776; Radionics, Dublin, Ireland).
EPs were recorded with a 32-channel multielectrode array (MEA). The MEA consisted of polyimide 2611 foil base material with gold contacts and track material, and titanium nitrate electrodes (FlexMEA; Multi Channel Systems, Reutlingen, Germany). The miniature circuit contained 32 recording electrodes, two reference electrodes and two ground electrodes. The recording electrodes were arranged in rows: a row of four electrodes followed by four rows of six and another of four. Each electrode was 30 µm in diameter with 300-µm interspacing (total area 1830 × 1830 µm); electrode impedance was approximately 50 kΩ. The MEA was placed extradurally over the cortical representation of the anal canal, and the positioning was guided stereotactically. Before placement, the MEA was washed with deionized water and dipped in 0·9 per cent saline.
Recordings were processed through a tenfold miniature preamplifier (MPA32I; Multi Channel Systems) and the MEA head stage, a combined amplifier, filter and data acquisition system (USB-ME-FAI System; Multi Channel Systems); MC_Rack 4·3·0 (Multi Channel Systems) was used to display and record data. Anal canal stimulation and the MEA head stage were both triggered by a programmable stimulator unit (Master 8; Grass, Slough, UK) to synchronize stimulation and recording. A sampling frequency of 10 kHz was used. For each trial, a sweep average over 400 EPs was constructed.
Anal canal stimulation
For anal canal stimulation, a gold-plated plug cathode (diameter 2 mm) was placed in the anal canal and a silver wire anode (diameter 500 µm) was introduced subcutaneously lateral to the external anal sphincter on the left side. The following stimulation parameters were used: amplitude 10 V, frequency 1 Hz, pulse duration 1 ms.
Sacral nerve stimulation
A small medial incision was made over the sacrum and the first sacral foramen visualized. A concentric needle electrode was placed in the left first sacral foramen to stimulate the first sacral nerve root (26-G disposable concentric electrode; SLE, Croydon, UK). Whereas in humans the pudendal nerve arises from S2–S4, which in turn gives rise to the inferior rectal nerve innervating the anal canal, in rats L6 and S1 contribute to the inferior rectal nerve. Afferent fibres from the anal canal travel mainly through the first sacral nerve root22. Therefore, S1 was used to simulate SNM in the rat, as described previously18. Correct placement of the electrode was confirmed by the observation of a tail twitch. Stimulation was applied at 15 V, 1-ms pulse duration, and varied frequencies and durations. The voltage appears high in this study compared with common clinical values because the electrode tip is very small. Application of an InterStim™ II (Medtronic, Minneapolis, Minnesota, USA) and a customized rodent lead, which more accurately mirrors the clinical situation, produced equivalent results at very small voltages (less than 1 V) (Fig. S1, supporting information).
Effect of stimulation frequency
The total number of impulses at each test frequency was clamped to 180. Table 1 shows the resulting frequency–duration combinations. A sham stimulation group was also created by placing the concentric needle electrode into the S1 foramen but without turning on the current. Four animals were studied in each group. EPs were recorded before SNM was applied, and recordings were repeated every 10 min for 70 min. To facilitate the principle of reduction in animal use, the study was designed in such a way that groups with the same stimulation parameters were reused to meet the other objectives. Only animals with a complete set of recordings over the whole time frame were used in analysis. The exact allocation of animals to the specific objectives and groups is shown in Fig. S2 (supporting information).
Frequency–time combinations used in the frequency study
| Frequency | Time | No. of pulses |
|---|---|---|
| No. of pulses fixed at 180 | ||
| 0·1 Hz | 30 min | 180 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 18 s | 180 |
| 100 Hz | 1·8 s | 180 |
| Stimulation time fixed at 3 min | ||
| 0·1 Hz | 3 min | 18 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 3 min | 1800 |
| 25 Hz | 3 min | 4500 |
| 100 Hz | 3 min | 18 000 |
| Frequency | Time | No. of pulses |
|---|---|---|
| No. of pulses fixed at 180 | ||
| 0·1 Hz | 30 min | 180 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 18 s | 180 |
| 100 Hz | 1·8 s | 180 |
| Stimulation time fixed at 3 min | ||
| 0·1 Hz | 3 min | 18 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 3 min | 1800 |
| 25 Hz | 3 min | 4500 |
| 100 Hz | 3 min | 18 000 |
Frequency–time combinations used in the frequency study
| Frequency | Time | No. of pulses |
|---|---|---|
| No. of pulses fixed at 180 | ||
| 0·1 Hz | 30 min | 180 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 18 s | 180 |
| 100 Hz | 1·8 s | 180 |
| Stimulation time fixed at 3 min | ||
| 0·1 Hz | 3 min | 18 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 3 min | 1800 |
| 25 Hz | 3 min | 4500 |
| 100 Hz | 3 min | 18 000 |
| Frequency | Time | No. of pulses |
|---|---|---|
| No. of pulses fixed at 180 | ||
| 0·1 Hz | 30 min | 180 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 18 s | 180 |
| 100 Hz | 1·8 s | 180 |
| Stimulation time fixed at 3 min | ||
| 0·1 Hz | 3 min | 18 |
| 1 Hz | 3 min | 180 |
| 10 Hz | 3 min | 1800 |
| 25 Hz | 3 min | 4500 |
| 100 Hz | 3 min | 18 000 |
Creation of a frequency–response curve and calculation of optimal frequency
A 3-min burst of stimulation using one of the following frequencies was administered: 0·1, 1, 10, 25 or 100 Hz. As only one of the values can be fixed at a time, the number of pulses was varied (Table 1). Each group consisted of six animals.
Model validation
The model developed in the objective above predicted the size of effect at the theoretical optimal frequency. Four animals were used to derive the actual effect at the optimal frequency, and this result was compared statistically with the predicted value. A new sham stimulation group containing four animals was also created to ensure stability of signals. In total, eight animals were used. Recordings were taken for 120 min after SNM.
Role of stimulus amplitude
For this experiment the calculated optimal frequency was used. The motor threshold to tail movement served as a baseline, and stimulation at 0·25, 0·5 and 0·75 times the motor threshold was investigated. There were four animals in each group. Recordings were taken before SNM and at 10-min intervals for 70 min after SNM.
Statistical analysis
Data were displayed, recorded and reviewed using MC_Rack 4·3·0 (Multi Channel Systems). The channel showing the maximum amplitude was selected. Using Spike 2·06 (Cambridge Electronic Design, Cambridge, UK), amplitudes and latencies were measured. For the analysis of maximum amplitude the percentage change was calculated in relation to the initial value before SNM treatment. A two-way repeated-measures ANOVA and a Bonferroni post-test were used for further analysis. For calculation of the frequency–potentiation relationship, non-linear regression was performed. Validation of the optimal frequency model involved a Student's t test of actual data against a single predicted value. The statistical software package GraphPad Prism® 4 (GraphPad Software, La Jolla, California, USA) was used for all statistical analyses and non-linear curve fitting. Values are given as mean(s.e.m.). Statistical significance was set at P < 0·050.
Results
Cardiorespiratory indices
Mean blood pressure was 99(2) mmHg, measured in the left femoral artery at the beginning of the experiment (94(2) mmHg at the end of the experiment; n = 52). The haematocrit was 42(1) per cent. Arterial blood gases were measured after completion of the operations and after the last recording in 11 animals: partial pressure of carbon dioxide (Pco2) 44(1) mmHg, partial pressure of oxygen (Po2) 84(2) mmHg and pH 7·4(0·01) after surgery, and Pco2 43(1) mmHg, Po2 78(2) mmHg and pH 7·4(0·01) after the last recording. All parameters were within the physiological range for the duration of the experiment.
Effect of stimulation frequency
EPs were recorded in each animal (total 21 rats), although one animal was excluded from analysis owing to cessation of recordings after 20 min. EPs consisted of one upward deflection with a latency of 13·5(0·8) ms and amplitudes of 25·9(4·7) μV, followed by an inconstant small downward deflection with amplitudes of 15·6(3·2) μV. Recordings taken after death showed the stimulation artefact and a flat line. Group traces before SNM, 30 min after SNM, and after death are shown in Fig. 1.
Group traces (n = 4) before the application of sacral neuromodulation (SNM), 30 min after the application of SNM, and after death. Note the moderate increase in amplitude after 0·1-Hz stimulation for 30 min (a) and the marked increase after 1-Hz stimulation for 3 min (b). There is no change in amplitude for 10-Hz (c), 100-Hz (d) and sham (e) groups. Post-mortem traces show only the stimulation artefact
The coefficient of variation ((s.d./mean) × 100) for an animal over seven trials was 10·6(1·4) per cent. Because of a high interanimal variation of amplitude, data are expressed as the percentage of the initial value before SNM was applied. A marked increase of 130 per cent of the initial value could be observed for 1-Hz stimulation, whereas 0·1-Hz stimulation yielded a moderate increase of 50 per cent. Neither 10-Hz nor 100-Hz stimulation resulted in a potentiation of EPs. The amplitude in the sham group was stable.
Fig. 2 shows the course of the maximum amplitudes over time. A two-way repeated-measured ANOVA showed that the frequency factor was highly significant (P < 0·001) and that time was not significant (P = 0·169). There was no interaction between the curves (P = 0·625). A Bonferroni post-test showed that, compared with the sham stimulation group, the increase was significant at 30 and 70 min in the 0·1-Hz group (P < 0·050) and at every time point for the 1-Hz group (P < 0·010). Onset latencies before and after SNM were not significantly different (2-way repeated-measures ANOVA; frequency factor, 0·20).
Group data (n = 4) illustrating the effect of frequency and time. The amplitude is shown as a percentage of the initial value. There is a marked increase in the 1-Hz and a moderate increase in the 0·1-Hz group. Note that 10-Hz stimulation for 18 s did not result in an increase in amplitude
Creation of a frequency–response curve and calculation of optimal frequency
EPs were recorded in each animal (total of 29 additional rats). However, recording could not be continued after SNM was applied in one animal, which was excluded from further analysis. EPs consisted of one upward deflection with latencies of 11·6(0·6) ms and amplitudes of 49·6(6·1) μV, followed by an inconstant small downward deflection with amplitudes of 25·8(3·2) μV (Fig. 3). The coefficient of variation for an animal over seven trials was 11·6(1·2) per cent. A marked increase of 120 per cent of the initial value was observed for 1- and 10-Hz stimulation, whereas 0·1- and 25-Hz stimulation yielded a moderate increase of 50 per cent. No potentiation could be observed for 100-Hz stimulation. No change in amplitude was seen in the sham group.
Group traces (n = 6) at baseline before and 30 min after the application of sacral neuromodulation (SNM). Note the moderate increase in amplitude after 0·1-Hz stimulation for 3 min (a) and 25-Hz stimulation for 3 min (d). There is a marked increase after 1-Hz stimulation for 3 min (b) and 10-Hz stimulation for 3 min (c). There is no change in amplitude for the 100 Hz (e) and sham (f) groups
Fig. 4 shows the course of the maximum amplitudes over time. A two-way repeated-measures ANOVA showed that the frequency factor was highly significant (P < 0·001) and that time was significant (P = 0·038). There was no interaction between the curves (P = 0·462). A Bonferroni post-hoc test showed that, compared with the sham stimulation group, the increase was significant for every time point of the 1- and 10-Hz groups (P < 0·001), and for 70 min in the 0·1-Hz group (P < 0·050). Onset latencies were not statistically significantly different between groups (2-way repeated-measures ANOVA; frequency factor, P = 0·208).
Group data (n = 6) illustrating the effect of frequency. The amplitude is shown as a percentage of the initial value. There is a marked increase in 1- and 10-Hz groups, and a moderate increase in 0·1- and 25-Hz groups
The relationship between stimulation frequency and magnitude of potentiation was also investigated; percentage increases were expressed as a function of stimulation frequency. Because EPs evened out 20 min after SNM application, time points from 20 min onwards were analysed. When the frequency was expressed as a log2 value, the increase in amplitude could be expressed by a polynomial function of the second order (y = c + bx − ax2). The graph was parabolic in shape, with the inflection point showing the maximum increase in amplitude and the corresponding frequency. The graphs for each time point are shown in Fig. 5, and were all similar. The calculated optimal stimulation frequency range was 1·7–2·1 Hz, and corresponded to an increase of 120–140 per cent. The theoretical increase for 14-Hz stimulation, which is used clinically, was 89–104 per cent.
Graphs resulting from non-linear curve-fitting for each time point. The curve describing the frequency–potentiation relationship is a downward opening parabola and curves are similar for the different time points (a–f). The graphs can be described by the formula c + bx − ax2. The optimum was at 1·7–2·1 Hz and the predicted maximum increase at the optimum was 120–140 per cent. R2 values (from non-linear regression) ranged between 0·42 and 0·57
Model validation
To validate the model constructed in the above objective, 2-Hz stimulation (15 V, 1-ms pulse duration) was studied in four animals, and a further four animals were used as a sham stimulation group. EPs had amplitudes of 37·9(10·1) μV before the application of SNM. Acute SNM yielded a marked (100–150 per cent) increase in maximum amplitude in the 2-Hz stimulation group. The potentiation was stable for 120 min. In the sham group, EPs were stable throughout this interval (Fig. 6). A two-way repeated-measures ANOVA showed that the frequency factor was highly significant (P = 0·002) and the time factor was not (P = 0·131). These experimentally measured values were compared with the predicted increase. The calculated percentage increase for 2-Hz stimulation was 120–140 per cent, depending on the time point. Amplitudes increased by 113–152 per cent in this experiment. If the actual increase was compared with the predicted maximum, no significant difference was noted (1 parameter Student's t test; P = 0·514–0·814).
Group data (n = 4) of the 2-Hz test and sham groups. The amplitude is shown as a percentage of the initial value. The 130 per cent potentiation in the 2-Hz group and the amplitude of evoked potentials in the sham group were stable for 120 min after sacral neuromodulation
Role of stimulus amplitude
Five groups with four animals each were used: 2-Hz stimulation for 3 min was applied at 1·0, 0·75, 0·5 and 0·25 times the motor threshold, and 0·0 times as sham stimulation. EPs had amplitudes of 41·3(5·6) μV and a latency of 11·3(0·8) ms before the application of SNM. The coefficient of variation for amplitudes in any one animal was 10·5(0·9). The motor threshold was 16(0·9) V. The response was approximately 130 per cent at 1, 0·75 and 0·5 times motor threshold, whereas EPs for 0·25 times motor threshold did not change (Fig. 7). A two-way repeated-measures ANOVA with the two factors (stimulus amplitude and time) showed that the stimulus amplitude factor was highly significant (stimulus amplitude factor, P < 0·001; time factor, P = 0·001; interaction, P = 0·792). Bonferroni post-tests were used to compare each group with the 1·0 times motor threshold group; 0·75 and 0·5 times motor threshold groups were not significantly different from this group, but both 0·25 times motor threshold and sham stimulation groups were significantly different for all time points (P < 0·001). Onset latencies were not statistically significant between groups (2-way repeated-measures ANOVA; frequency factor, P = 0·058).
Group data (n = 4) illustrating the effect of stimulation intensity. The amplitude is shown as a percentage of the initial value. Note the marked increase in amplitude for 1, 0·75 and 0·5 times motor threshold groups. There is no potentiation in the 0·25 times motor threshold and sham groups
Discussion
Stimulation frequency and amplitude in SNM have acute effects in healthy rats. Stimulation frequency greatly affects the magnitude of potentiation seen after application of SNM; the optimal frequency is 2 Hz. SNM of EPs is an ‘all or nothing’ phenomenon relayed via sensory afferents; even a short burst of SNM results in a long-term potentiation in the primary somatosensory cortex and, for this effect, a critical burst duration is required. Each of these points is discussed in detail below.
The frequency that yielded maximum potentiation was calculated to be 2 Hz, and was confirmed experimentally. Compared with the clinically used frequency of 14 Hz, a 35 per cent greater potentiation (130 per cent instead of 95 per cent) could be attained. The effect of stimulation frequency on patients' outcomes has been studied by two groups14,15. Duelund-Jakobsen and colleagues15 tested different frequency–pulse width combinations (14, 6·9 and 31 Hz with 210, 90 and 330 µs) for 4 weeks each in a double-blind randomized cross-over trial in 15 patients with faecal incontinence who experienced a loss of efficacy after initially good performance. The patients' preferred setting was permanently programmed. More than half of the patients preferred higher-frequency stimulation (31 Hz, 210 µs). Although patient satisfaction and physiological parameters (resting anal pressure, maximum anal squeeze pressure and rectal sensitivity to volume distension) showed no changes at the preferred setting, bowel scores and the total number of bowel movements improved with the higher-frequency and longer pulse width settings15. The study by Dudding and co-workers14 also tested different frequencies (6·9, 14 and 31 Hz) and pulse widths (90, 210 and 450 µs), in 12 patients who had improved only partially after implantation. Rectal compliance measured with a rectal barostat and an infinitely compliant rectal intraluminal bag was used as a surrogate marker to determine the optimal setting for individual patients. The greatest rectal compliance, measured at half the maximum tolerated volume, was reached with a frequency of 31 Hz in seven patients and a 90-µs pulse duration in five patients. With the optimized setting, incontinence episodes, soiling, faecal urgency and quality of life improved14.
In contrast to the present findings, both of the above studies14,15 showed a trend towards higher frequencies. This may be explained by selection of patients with chronic implantation, who were already experiencing a loss of efficacy. These patients may differ in pathophysiology from those with functioning implants, or may have developed some measure of fibrosis or displacement of the implanted electrode. However, the limitations of the present study need to be considered. First, results derived from an animal model may not be directly applicable to humans. The use of an animal model was chosen because it made the systematic examination of a wide range of frequencies possible. Second, although the authors have developed animal models of pudendal nerve injury22, the present study focused on healthy animals and an optimal therapeutic frequency may be different to an optimal physiological frequency. Third, only a brief time period was examined in comparison with the chronic nature of SNM therapy. However, the initial plasticity changes that lead to long-term potentiation happen within the first 2 h23. Fourth, only one possible mechanism of action of SNM has been explored in this study.
The choice of appropriate stimulus intensity is important for efficacy. This study has shown an ‘all or nothing’-type phenomenon when the voltage of SNM is varied. Once a stimulation intensity of half motor threshold was exceeded, a potentiation of anal canal EPs occurred and further voltage increase did not increase the magnitude of this potentiation. A voltage decrement below half motor threshold resulted in no response.
The stimulation voltages used in the present study may appear high compared with the clinical values. However, a comparison of the absolute intensity of human and rat SNM is difficult. The concentric electrode design used here differs from the human neurostimulator system in which a large cathode and anode are spaced widely. The injection of current into the nerve depends on the relationship between the surface area of the electrode and the current applied. The concentric design was chosen deliberately, because the small surface area makes for precise current injection and minimizes the stimulus artefact in the cortical recordings. To make the animal model more relevant to the clinical setting, a customized rodent SNM lead was designed and built by Medtronic. This was connected to a human InterStim™ II device (Medtronic) and tested in two additional animals. The motor threshold was as low as 0·4 V and stimulation at 2 Hz for 3 min resulted in a 160 per cent increase in EPs (Fig. S1, supporting information).
The intensity of 15 V applied via the concentric electrode corresponds closely to the mean motor threshold (16(0·9) V), because the visible tail twitch was required for confirmation of the right electrode position. This intensity cannot be used in therapy, but this study has shown that potentiation of the same strength is preserved when using voltages below motor threshold. Increasing stimulation intensity above half motor threshold does not seem to provide further benefit.
In terms of the effect of stimulation duration, the present study indicates that a critical burst duration is required to trigger long-term potentiation in the somatosensory cortex. The minimum time was not defined in this study, although it was shown that 180 s was sufficient and 18 s was not. It follows from this observation that an intermittent, rather than continuous, pattern of stimulation may be effective. This has important implications for the battery life of neuromodulator devices.
The present body of work shows that the frequency of SNM governs the efficacy of its effect on anal cortical EPs in an animal model. The optimal stimulation frequency is 2 Hz, and SNM must be applied for at least 3 min and at a voltage above half motor threshold.
Acknowledgements
The authors gratefully acknowledge the generous support of the following agencies: Science Foundation Ireland (11/RFP/3115), Medtronic, and the Bowel Disease Research Foundation.
Disclosure: The authors declare no other conflict of interest.
