**4. Procedural simulation for renal cancer**

Competently performing a whole procedure requires knowledge of surgical anatomy, procedural steps and the ability to perform each surgical component. Whole procedure simulation is challenging and at present time in renal surgery, it is largely limited to cadaveric and animal models. As a result of these limitations part-procedural simulation, where a particular procedural step is simulated (i.e. tumour excision or renorraphy), has advanced significantly over the last decade. The majority of these models are bench-top, either wet or dry, and have the advantage of being able to be utilised for open, laparoscopic and robotic platforms. The following section aims to explore the models available for radical and partial nephrectomy.

#### **4.1. Radical nephrectomy**

Radical nephrectomy remains the most utilised treatment approach for renal malignancy [41, 42]. Traditionally performed as an open procedure, laparoscopic radical nephrectomy has become widespread due to the benefits of shorter convalescence and less procedural morbidity [42]. The initial experience with laparoscopy was technically challenging, and the learning curve and associated complication rates for novice surgeons were a significant barrier to uptake [43]. Developments in training and simulation subsequently followed in an attempt to provide an adjunct for skill development outside of the operating theatre [44, 45]. At present there are a vast array of simulators available for acquiring laparoscopic skills with extensive validation ranging from box trainers to develop basic skills, to whole procedural simulation on live animals and VR platforms.

#### *4.1.1. Physical simulation*

The first clinical laparoscopic radical nephrectomy (LRN) was performed in 1990 by Clayman and colleagues [46] after extensive experimentation on porcine models. The benefits of animal models for teaching dissection, tissue handling, haemostasis and vascular control are significant, and subsequently this simulation modality remains central to the development and dissemination of minimally invasive surgical techniques [47].

Molinas and colleagues [17] demonstrated the validity of live simulation in LRN using a rabbit model. Ten gynaecologists and 10 medical students each performed 20 laparoscopic nephrectomies over a 20-day training course. The overall time required to perform the LRN decreased from 44 ± 18 to 11 ± 2 minutes for the first and the last procedure, respectively, and complication rates similarly decreased. Despite the rabbit's smaller size compared to pigs for example, pneumoperitoneum was able to be established, and conventional instruments were used for all procedures. Reduction in acquisition and handling costs associated with the rabbits allowed the authors to provide a more prolonged period of training demonstrating the impact of repetition on learning curves and complication rates.

Cruz and colleagues [48] assessed the impact of repeated LRN in the porcine model on overall surgical performance among established surgeons. Six urologists with limited laparoscopic experience were recruited to perform a live porcine LRN weekly for 10 weeks. Surgical performance was judged quantitatively including total operative time and estimated blood loss. Qualitative measures were also assessed using the Global Operative Assessment of Laparoscopic Skills (GOALS) including depth perception, dexterity, efficiency, tissue handling and autonomy. Over the course of the study, blood loss, depth perception and dexterity showed statistically significant improvements. The remaining domains including operative time showed no statistical improvement.

#### *4.1.2. Virtual reality*

(Simulated Surgical Systems, USA) incorporates a real surgical procedure into the virtual reality framework and guides the user through an enhanced version of the operation, with audio-visual illustration, haptic cues and guided movements [24]. The HoST system currently does not offer procedural simulation for nephrectomy or partial nephrectomy. Maestro AR (Mimic Technologies, USA) provides procedure-specific 3D video and interaction via virtual reality robotic instruments. This includes a module on partial nephrectomy that demonstrates

Competently performing a whole procedure requires knowledge of surgical anatomy, procedural steps and the ability to perform each surgical component. Whole procedure simulation is challenging and at present time in renal surgery, it is largely limited to cadaveric and animal models. As a result of these limitations part-procedural simulation, where a particular procedural step is simulated (i.e. tumour excision or renorraphy), has advanced significantly over the last decade. The majority of these models are bench-top, either wet or dry, and have the advantage of being able to be utilised for open, laparoscopic and robotic platforms. The following section aims to explore the models available for radical and partial nephrectomy.

Radical nephrectomy remains the most utilised treatment approach for renal malignancy [41, 42]. Traditionally performed as an open procedure, laparoscopic radical nephrectomy has become widespread due to the benefits of shorter convalescence and less procedural morbidity [42]. The initial experience with laparoscopy was technically challenging, and the learning curve and associated complication rates for novice surgeons were a significant barrier to uptake [43]. Developments in training and simulation subsequently followed in an attempt to provide an adjunct for skill development outside of the operating theatre [44, 45]. At present there are a vast array of simulators available for acquiring laparoscopic skills with extensive validation ranging from box trainers to develop basic skills, to whole procedural simulation

The first clinical laparoscopic radical nephrectomy (LRN) was performed in 1990 by Clayman and colleagues [46] after extensive experimentation on porcine models. The benefits of animal models for teaching dissection, tissue handling, haemostasis and vascular control are significant, and subsequently this simulation modality remains central to the development and

Molinas and colleagues [17] demonstrated the validity of live simulation in LRN using a rabbit model. Ten gynaecologists and 10 medical students each performed 20 laparoscopic nephrectomies over a 20-day training course. The overall time required to perform the LRN decreased from 44 ± 18 to 11 ± 2 minutes for the first and the last procedure, respectively, and

face, content, construct and concurrent validity [22].

**4. Procedural simulation for renal cancer**

**4.1. Radical nephrectomy**

80 Evolving Trends in Kidney Cancer

on live animals and VR platforms.

dissemination of minimally invasive surgical techniques [47].

*4.1.1. Physical simulation*

Despite the obvious benefits of high-fidelity animal models, the costs and associated ethical issues restrict access which is often limited to several day courses. A high-fidelity virtual reality LRN simulation platform has obvious advantages in overcoming some of these barriers. The LAP Mentor (Simbionix, USA) and LapSim (Surgical Sciences, Sweden) are two commercially available laparoscopic simulators, which provide VR laparoscopic training including a full nephrectomy module. While both simulators have been validated in terms of basic laparoscopic skills [49, 50], the nephrectomy modules remain to be formally scientifically assessed. Despite this, these simulators provide full procedure simulation that is reproducible and able to provide feedback on performance metrics such as economy of motion, procedure time and error rates. These metrics have potential utility in assessing progression and setting benchmarks for training curriculums.

#### **4.2. Partial nephrectomy**

With the advent of widespread cross-sectional imaging, there has been a surge in incidental detection of small renal masses. This has subsequently led to increased utilisation of partial nephrectomy (PN) in order to preserve normal renal parenchyma in these otherwise well patients [51]. PN is a technically challenging operation with a significant learning curve and variability unrivalled by almost any other frequently performed kidney procedure [52]. Perhaps, most challenging, however, is the *time-critical* nature of PN. The vast blood supply to the kidney means bleeding is a significant intraoperative risk and efficient excision, and renorrhaphy is therefore crucial. Furthermore, prolonged warm ischaemia is deleterious to healthy renal tissue and can impact post-operative renal function [53, 54]. Finally, each tumour is highly variable in size, location and relation to critical structures, making oncological excision a persistent challenge even for experienced surgeons. For these reasons, training in PN is subsequently fraught with complexity, and mentors must try and negotiate sometimes the discordant goals of training with patient safety. Simulation for PN has rapidly progressed in response to this dilemma, and the availability of PN models is becoming more widespread.

#### *4.2.1. Physical simulation*

Tumour-mimic models for PN rose to prominence in the initial laparoscopic era in response to the technically challenging nature of the procedure and associated learning curve. Taylor and colleagues [55] described one of the earliest models in 2004, whereby a pigmented mixture was injected into a series of ex-vivo and in-vivo porcine kidneys. The authors were able to create a variety of lesions both endo- and exophytic with a mean size of 10 mm. This model was not formally assessed as part of a training programme but established the feasibility of artificial tumour creation. Hidalgo et al. [15] similarly described the creation of an in-vivo porcine PN model through the percutaneous injection of a liquefied plastic solution into the subscapular renal space to create exophytic lesions. This model was evaluated as part of a laparoscopic training programme and found to enhance the learning experience in 96% of participants. While advantageous for the novice, the inability of these techniques to create large endophytic or central lesions may limit the utility to more advanced surgeons.

(no robotic cases), 9 intermediates (1–100 robotic cases), and 13 experts (>100 robotic cases). Among expert surgeons, the model demonstrated excellent face and content validity. Experts rated the applicability for advanced surgeons as lower, however, which likely reflects the lack

Simulation and Training in Kidney Cancer Surgery http://dx.doi.org/10.5772/intechopen.85683 83

The recent advent of rapid prototyping (3D-printing) has allowed the formation of synthetic surgical renal tumour models. Several groups have already demonstrated that high-fidelity 3D printed renal models can be created using specialised software to import diagnostic crosssectional imaging [26, 58]. Monda and colleagues [14] recently developed and validated a silicone tumour model from a 3D printed cast of a kidney with a tumour. A medium complexity tumour was selected from a patient who had previously undergone RAPN at the authors' institution, and a 3D printed negative-volume mould was created. Following this, tumour models could be repeatedly cast with silicone using this mould. The model was validated by surgeons of different training levels and demonstrated face, construct, and content validity. Through the use of a 3D printed mould, the authors were able to subsequently reproduce

Von Rundstedt et al. [26] used advanced 3D printing to create a high-fidelity, patient-specific, synthetic renal tumour model for the purposes of surgical rehearsal prior to actual RAPN. Surgical models were created for 10 patients and the same surgeon performed all rehearsals and actual RAPNs. The resection times and resection volumes were compared between rehearsal and live procedure and found to be predictive. Being able to predict, excision time has significant implications and could be utilised in assessing the feasibility of more complex masses for PN within an acceptable warm-ischaemia time. Furthermore, the authors reported altering their actual surgical approach in several patients based on difficul-

Maddox and colleagues [13] used a slightly different process to construct patient-specific tumour models by 3D printing an outer polymer 'shell' which was subsequently filled with an agarose gel solution to resemble normal renal parenchyma. The renal mass of interest, as well as critical structures such as renal vasculature and collecting system, was able to be pigmented to distinguish them from the normal parenchyma. It is very conceivable that 3D-printed bench models may ultimately decrease the learning curve and potentially improve surgical outcomes; however, further studies are needed to fully elucidate this effect. Current limitations include the lack of 'real-life' confounders such as perinephric fat and an active blood supply; however, it is very possible that these could be overcome in the future.

No PN specific whole procedure VR simulation is commercially available at present. In an attempt to bridge the gap, Hung and colleagues [22] developed and validated an augmented reality platform now commercially available as Maestro AR (Mimic Technologies, USA). In this 'hybrid' model, augmented reality and virtual reality were combined to create a procedural specific platform that aimed to teach surgical anatomy, procedural steps and operative skills. High-definition actual surgical video of a full length RAPN was embedded with

of renorrhaphy and haemostasis component associated with the simulation.

multiple models reliably with minimal cost.

*4.2.2. Virtual reality*

ties encountered with tumour excision in the simulated rehearsal.

Yang et al. [27] described an ex-vivo porcine model, whereby the kidney was secured to a specifically designed box for use with a laparoscopic trainer. The renal vessels were preserved, and simulated vascular perfusion was achieved through infusion of red-dyed water through the artery. Urology trainees were requested to excise a 2 cm spherical piece of renal parenchyma and then complete renorrhaphy. The model was validated by five urology trainees, each of whom completed 10 attempts at the LPN model over a 20-day period. Trainees demonstrated a decrease in the total operative and renorrhaphy times with progressive attempts, as well as increase in the quality of the PN as assessed by two blinded experts. Trainees also reported an improvement in their confidence to perform a LPN, particularly with respect to tissue manipulation, intra-corporeal suturing and knot tying.

The proliferation of robotic-assisted surgery has helped overcome many of the barriers associated with LPN, resulting in shorter learning curves and subsequent growth in this area [56]. Eun and colleagues [57] described a novel technique for creating renal tumour mimics for RAPN in addition to a renal vein/inferior vena cava (IVC) tumour model for tumour thrombectomy. A tumour-mimic mixture was percutaneously injected into eight live pigs and one human cadaver in order to create 33 renal pseudotumours. A renal vein thrombus model was also created by injecting the material into the renal vein while clamped and allowing this to solidify. In addition, a renal-vein thrombus with extension into the IVC was created through partial clamping of the IVC with a long, curved bulldog clamp. Subsequent robotic radical nephrectomy with excision of the involved IVC cuff and IVC reconstruction was performed. This model was not validated by the authors but was the first demonstration of the feasibility of artificial renal vein and IVC tumour thrombus creation. While all procedures in this paper were performed robotically, such a model could be beneficial in both laparoscopic and open surgery.

Hung and colleagues [16] devised a novel robotic specific model for RAPN using an ex-vivo porcine kidney embedded with a 3.8 cm Styrofoam ball to mimic an exophytic renal tumour. The model task included tumour excision with a parenchymal margin but did not incorporate renorrhaphy. Forty-six participants were classified into 3 groups for validation, 24 novices (no robotic cases), 9 intermediates (1–100 robotic cases), and 13 experts (>100 robotic cases). Among expert surgeons, the model demonstrated excellent face and content validity. Experts rated the applicability for advanced surgeons as lower, however, which likely reflects the lack of renorrhaphy and haemostasis component associated with the simulation.

The recent advent of rapid prototyping (3D-printing) has allowed the formation of synthetic surgical renal tumour models. Several groups have already demonstrated that high-fidelity 3D printed renal models can be created using specialised software to import diagnostic crosssectional imaging [26, 58]. Monda and colleagues [14] recently developed and validated a silicone tumour model from a 3D printed cast of a kidney with a tumour. A medium complexity tumour was selected from a patient who had previously undergone RAPN at the authors' institution, and a 3D printed negative-volume mould was created. Following this, tumour models could be repeatedly cast with silicone using this mould. The model was validated by surgeons of different training levels and demonstrated face, construct, and content validity. Through the use of a 3D printed mould, the authors were able to subsequently reproduce multiple models reliably with minimal cost.

Von Rundstedt et al. [26] used advanced 3D printing to create a high-fidelity, patient-specific, synthetic renal tumour model for the purposes of surgical rehearsal prior to actual RAPN. Surgical models were created for 10 patients and the same surgeon performed all rehearsals and actual RAPNs. The resection times and resection volumes were compared between rehearsal and live procedure and found to be predictive. Being able to predict, excision time has significant implications and could be utilised in assessing the feasibility of more complex masses for PN within an acceptable warm-ischaemia time. Furthermore, the authors reported altering their actual surgical approach in several patients based on difficulties encountered with tumour excision in the simulated rehearsal.

Maddox and colleagues [13] used a slightly different process to construct patient-specific tumour models by 3D printing an outer polymer 'shell' which was subsequently filled with an agarose gel solution to resemble normal renal parenchyma. The renal mass of interest, as well as critical structures such as renal vasculature and collecting system, was able to be pigmented to distinguish them from the normal parenchyma. It is very conceivable that 3D-printed bench models may ultimately decrease the learning curve and potentially improve surgical outcomes; however, further studies are needed to fully elucidate this effect. Current limitations include the lack of 'real-life' confounders such as perinephric fat and an active blood supply; however, it is very possible that these could be overcome in the future.

#### *4.2.2. Virtual reality*

*4.2.1. Physical simulation*

82 Evolving Trends in Kidney Cancer

surgery.

Tumour-mimic models for PN rose to prominence in the initial laparoscopic era in response to the technically challenging nature of the procedure and associated learning curve. Taylor and colleagues [55] described one of the earliest models in 2004, whereby a pigmented mixture was injected into a series of ex-vivo and in-vivo porcine kidneys. The authors were able to create a variety of lesions both endo- and exophytic with a mean size of 10 mm. This model was not formally assessed as part of a training programme but established the feasibility of artificial tumour creation. Hidalgo et al. [15] similarly described the creation of an in-vivo porcine PN model through the percutaneous injection of a liquefied plastic solution into the subscapular renal space to create exophytic lesions. This model was evaluated as part of a laparoscopic training programme and found to enhance the learning experience in 96% of participants. While advantageous for the novice, the inability of these techniques to create

large endophytic or central lesions may limit the utility to more advanced surgeons.

tissue manipulation, intra-corporeal suturing and knot tying.

Yang et al. [27] described an ex-vivo porcine model, whereby the kidney was secured to a specifically designed box for use with a laparoscopic trainer. The renal vessels were preserved, and simulated vascular perfusion was achieved through infusion of red-dyed water through the artery. Urology trainees were requested to excise a 2 cm spherical piece of renal parenchyma and then complete renorrhaphy. The model was validated by five urology trainees, each of whom completed 10 attempts at the LPN model over a 20-day period. Trainees demonstrated a decrease in the total operative and renorrhaphy times with progressive attempts, as well as increase in the quality of the PN as assessed by two blinded experts. Trainees also reported an improvement in their confidence to perform a LPN, particularly with respect to

The proliferation of robotic-assisted surgery has helped overcome many of the barriers associated with LPN, resulting in shorter learning curves and subsequent growth in this area [56]. Eun and colleagues [57] described a novel technique for creating renal tumour mimics for RAPN in addition to a renal vein/inferior vena cava (IVC) tumour model for tumour thrombectomy. A tumour-mimic mixture was percutaneously injected into eight live pigs and one human cadaver in order to create 33 renal pseudotumours. A renal vein thrombus model was also created by injecting the material into the renal vein while clamped and allowing this to solidify. In addition, a renal-vein thrombus with extension into the IVC was created through partial clamping of the IVC with a long, curved bulldog clamp. Subsequent robotic radical nephrectomy with excision of the involved IVC cuff and IVC reconstruction was performed. This model was not validated by the authors but was the first demonstration of the feasibility of artificial renal vein and IVC tumour thrombus creation. While all procedures in this paper were performed robotically, such a model could be beneficial in both laparoscopic and open

Hung and colleagues [16] devised a novel robotic specific model for RAPN using an ex-vivo porcine kidney embedded with a 3.8 cm Styrofoam ball to mimic an exophytic renal tumour. The model task included tumour excision with a parenchymal margin but did not incorporate renorrhaphy. Forty-six participants were classified into 3 groups for validation, 24 novices No PN specific whole procedure VR simulation is commercially available at present. In an attempt to bridge the gap, Hung and colleagues [22] developed and validated an augmented reality platform now commercially available as Maestro AR (Mimic Technologies, USA). In this 'hybrid' model, augmented reality and virtual reality were combined to create a procedural specific platform that aimed to teach surgical anatomy, procedural steps and operative skills. High-definition actual surgical video of a full length RAPN was embedded with interactive VR exercises and virtual instruments in five modules: colon mobilisation, kocherisation of duodenum, hilar dissection, kidney mobilisation, tumour resection and renorrhaphy. In the final module, an embedded VR exercise was developed, whereby a mobile sponge could be manipulated around a central pivot point (renal hilum) and sutured. This platform was internally validated throughout development, and concurrent validity was assessed by comparison to an in-vivo porcine model. Expert surgeons rated the platform a useful tool for training residents and fellows particularly with respect to teaching the steps of the procedure and surgical anatomy. Performance in the VR renorrhaphy task correlated with that of the invivo porcine model in the intermediate and expert groups. While this platform is a significant progression towards procedure-specific VR simulation, further advances are needed before this could feasibly replace wet lab training. Allowing the user to alter the surgical view and perform embedded tasks for each step of the procedure would likely increase validity.
