Risk assessment of neuromuscular stimulation by energy-based transurethral resection devices: an ex vivo test standard

The surgical application of RF current can lead to adverse NMS. This is commonly attributed to the so called faradic effect created by unwanted frequencies below 20 kHz (pp. 518–519 in [1]). In urology, adverse NMS events have been reported during transurethral resection of bladder (TURB) [2‐5] and prostatic tumours (TURP) [4, 6]. Moreover, in gynaecology, the hysteroscopic transcervical resection should be mentioned. TURB is the most common surgical technique (gold standard) for the definitive diagnosis and initial treatment of bladder cancer [5], which is one of the most common urinary tract malignancies [7]. Potential intraoperative complications of TURB include bleeding and bladder perforation, typically as a consequence of stimulation of the obturator nerve, which may be associated with sudden, strong contractions of leg adductors. These contractions can lead to uncontrolled movements of the instruments introduced into the bladder, possibly leading to a bladder perforation, which requires conversion to open surgery. Current in vivo or clinical techniques for evaluating NMS typically include documenting the incidence of adverse NMS events such as obturator nerve reflex/adductor muscle contraction or bladder injury/perforation rate [2, 3, 5, 8, 9]. In some studies, the severity of intraoperative NMS events is evaluated on a rating scale, such as “no”, “weak/moderate” and “strong/severe” [10, 11]. Moreover, compound muscle action potentials from adductor muscles were examined in patients undergoing TUR surgery [12, 13]. To reduce the risk of NMS, several measures were described: First, a pharmacological block of the obturator nerve [8, 11‐14]. Second, intubation anaesthesia and relaxation of skeletal muscles of the patient (chpts. 11.3 & 11.3.4 in [15]). Third, a low-power setting, e.g. 50 W for cutting and 40 W for coagulation instead of a typically higher setting [2]. Fourth, the use of bipolar instead of monopolar RF current: Some studies reported advantages of bipolar over monopolar TURB, including a lower recurrence rate, better quality of biopsy specimens for the pathologist, less haemorrhage, lower incidence of the TUR syndrome and also reduced risk of NMS [2, 10]. A recent meta-analysis including seven clinical trials suggests that the incidence of obturator reflex and bladder wall perforation for monopolar vs. bipolar TURB is 14.8% vs. 3.7% (p < 0.0001) and 4.3% vs. 0.3% (p = 0.003) [3]. However, there are also studies showing that bipolar is not superior to monopolar TURB with respect to these complications [5, 9, 16, 17]. Consequently, the EAU guideline (Update 2016) to “Non-muscle-invasive Bladder Cancer” states that the results concerning resection techniques for TURB remain controversial (chpt. 5.11.3 in [18]). Overall, the measures described to reduce the incidence of NMS do not really solve the problem, as they either pose a different risk to the patient or impair performance of the RF application. A satisfactory solution would be RF generators which, due to technical advances, trigger less NMS. Therefore, we developed an ex vivo test system to assess the unfavourable NMS risk of different RF generators. This was previously only possible with in vivo animal testing or clinical trials.

A two-stage ex vivo approach was pursued: First, during the application of RF energy in an ex vivo TURB model, we recorded voltage curves at possible positions of the obturator nerve behind a porcine bladder wall (Fig. 1a). Second, these voltage records were used as stimuli in another ex vivo model with peripheral porcine nerves (Fig. 1b). The influence of these stimuli on nerve activity, i. e. the generation of evoked nerve compound action potentials (CAP) was recorded. A CAP is the spatial sum of action potentials generated by all stimulated axons of a whole nerve or a large population of axons recorded extracellularly. Thus, in this study, the firing rate and strength of CAPs represent a surrogate measure for the risk of NMS. We refer to this measure as “NMS potential”, which is defined mathematically in subsection “Data analysis”. The tested RF generators (250–450 kHz base frequency) were used with the control algorithm/mode for cutting with bipolar instruments in saline solution. Activation is divided into an ignition phase, in which spark formation is triggered at relatively high current flow (e.g. 8 A), and a cut phase, in which sparking is maintained at relatively low currents (e.g. 4 A). The ignition phase is fixed, whereas during the cut phase the intensity can be adjusted via the generator setting. Since the hardware components and modes differ, the exact electric parameters and instrument-specific adaptations are quite different between manufacturers. Thus, the recorded voltage curves represented a spectrum of different manufacturers and generator settings.

The basis for the validity of the ex vivo TURB model was to mimic the essential components of this clinical application. The usage of normal saline in compartment C1 is equivalent to the clinical setting of TURB. The usage of normal saline in compartment C2 is a simplification based on multiple considerations: First, the electrical conductivity of normal saline solution is quite similar to that of extracellular human blood plasma or interstitial fluid outside the bladder wall (~ 1.9 S/m and 1.7 S/m at 25 °C). Second, exactly matching the conductivities is not mandatory because the extracellular fluid conductivity may vary, e. g. as a consequence of various pathological states [28]. Third, assuming the relatively high conductivity of normal saline solution within C2 is an appropriate worst-case estimate, in which a fluid behind the bladder wall makes direct contact to the obturator nerve without further tissue in between. In comparison, muscle or connective tissue is approximately five times less conductive (0.3–0.4 S/m). The usage of porcine bladder is the most appropriate substitute for human bladder since pigs have been established as a model for research of human diseases due to similarities, which includes size, organ structure and function, physiology and organ development [29, 30]. A high sampling rate (5 MS/s) for the differential probe was chosen to assure faithful reproduction of the signal in the nerve model. Extracellular recording of CAPs from peripheral nerve trunks is a well-established method to study nerve activity that is utilised both in teaching at universities and in preclinical studies [22‐25]. A similar test method is used clinically e. g. by recording CAPs of the auditory nerve in cochlear implant patients (for a review see [31]). Another potential ex vivo test method suitable for this study could be a nerve-skeletal muscle preparation that encompasses the sciatic nerve and the associated gastrocnemius muscle [32‐34]. This kind of preparation includes the neuromuscular junction and the effector or muscle allowing for a more direct measure of the output variable, i. e. muscle contraction, by means of electromyography or force measurement. However, this increases the overall complexity of the test method and the effort of dissection compared to the usage of an isolated peripheral nerve trunk. It would be much more challenging to standardise measurements at this level of complexity and there is no well-established ex vivo methodology incorporating a mammalian nerve-skeletal muscle preparation. Moreover, porcine nerves are very similar to human nerves regarding anatomy and physiology [29, 30]. Employing fresh tissue samples spares use of live animals, does not require animal testing application for the study, nor ethical animal husbandry or other additional efforts. Taken together, the porcine nerve trunk model seems to be the most parsimonious yet reliable ex vivo test method to address the aims of this study. A high inter-data set correlation (Fig. 3) demonstrates reproducibility of the outcome measure, which is due to the standardisation of the test method.

The justification for the two-stage setup is as follows: First, CAP recording during RF activation is not possible due to RF interference. Second, voltage recording enables multiple tests with the same stimulus in the nerve model, thereby demonstrating reproducibility. Regarding the position of the voltage measurement in the TURB model, we have pre-tested several positions. Only at a distance of ~ 3 mm from the RF application was the signal strength sufficient, and it was also far enough away not to damage the measuring electrode and theoretically a nerve [19]. The choice of position thus represents a clinical worst-case for nerve stimulation without nerve damage. In addition, the voltage recordings were processed (100 kHz low pass & scaling) to serve as input for the nerve model while minimising damage to the nerve samples and maximising nerve response. Therefore, the original strength of the voltage recordings was not the decisive parameter for the obtained NMS potentials. Instead, the scaling of the stimulation signals to adapt to the individual responsiveness and impedance of each nerve sample was crucial to obtain comparable NMS potentials across the entire measurement series. In parallel, scaling compensated for the reduced nerve excitability that occurs ex vivo during prolonged and repeated stimulations.

Another point of discussion is why generator voltage should not be used as input for a nerve model or for the immediate evaluation of NMS. At first glance, the voltage recordings of the TURB model look like an attenuated version of the generator voltage (data not shown). However, there are important differences: First, different tissue layers of the bladder attenuate the signal depending on frequency and thus act as a complex filter. Second, NMS may be caused by DC loops around the high-density or “active” electrode, which are caused by sparks and cannot be detected in the outer chain of generator, electrodes and connecting wires [35]. Thus, an application model with the relevant tissue components is necessary for the evaluation of NMS.

Regarding the outcome measure, we observed obvious differences between the RF generators in terms of their NMS potential (Fig. 4). Thus, the percentage of significant comparisons between the generators (~ 87%, Table 1) is an indicator for the selectivity of our test method. The differences in NMS potential have at least two reasons: First, differences of the effect settings, i. e. the minimum, default and maximum effect settings of each vendor are not in complete correspondence with respect to basic electrical parameters such as applied voltage and maximum power output. Second, differences in the programming of the RF generators and the components used in these generators could be the cause. Our test method is suitable for quantifying the NMS potential of different electrosurgical systems ex vivo but it does not reveal causes behind the differences observed. An additional study limitation is that the NMS potential does not translate into a relevant clinical parameter such as the incidence of obturator reflex or bladder perforation. However, the observed differences in NMS potential should clinically lead to more or less adverse NMS events. Of course, other influencing factors that are not represented by our model, such as the experience of the surgeon, patient-specific differences such as weight, pathological changes, etc. must also be taken into account. Furthermore, clinically investigated influencing factors, such as the monopolar versus bipolar technique or the use of a pharmacological block, were not addressed in our ex vivo model. These system-level questions are likely to remain a task of in vivo testing. Nonetheless, the comparison of new electrosurgical systems with well-established and safe equipment allows for a meaningful safety evaluation. To our knowledge, the test method described here is the first to address the ex vivo risk assessment of NMS for transurethral resection devices. In the future, this test method can easily be adapted to test bipolar TURP as well as monopolar TURB and TURP, e.g. the TURB model can be adapted to emulate TURP using prostatic instead of bladder tissue.

Our test method is suitable for quantifying the NMS potential of different electrosurgical systems ex vivo with high selectivity at a reasonable degree of standardisation and with justifiable effort. Our results suggest that the clinical incidence of NMS is considerably influenced by the type of RF generator. The development of future generations of RF generators takes advantage from the proposed test standard through higher safety and by replacement or reduction of the extent of animal testing. Most important, health professionals and treated patients will benefit from improved RF surgery using generators with a low risk of NMS.