The Role of Simulation in Attaining Proficiency in Minimally Invasive Hepatopancreatobiliary Surgery

Introduction

For the past two decades, minimally invasive surgical techniques have infiltrated multiple surgical subspecialties. The benefits of minimally invasive surgery are broad, ranging from reduced blood loss, postoperative pain and hospital stay to improved quality of life and reductions in morbidity. Other potential benefits, however, such as cost-effectiveness and oncological efficacy remain debated.^1,2 The application of these techniques to complex hepato-pancreato-biliary (HPB) surgery has been slower than other subspecialties.

The first report of robotic HPB surgery in 2003 included pancreatic resections, hepatic segmentectomy, common bile duct resection, and hepatic cyst fenestrations among a cohort of robotic general surgery cases. ³ This study suggested that an array of robotic general surgery cases can be performed safely, with an overall morbidity of 9.3% and postoperative mortality of 1.5%. Since then, an increasing number of studies have reported on the safety, feasibility, and advantages of the robotic platform, but its adoption in HPB surgery has been slower than other subspecialties. Several reasons contribute to this slow adoption. HPB procedures are inherently long and complex, with significant learning curves even for the open approach. Learning and adopting a technology perceived to be more technically challenging in an already challenging field has been met by hesitancy by surgeons. In addition, postoperative complications after these procedures, particularly pancreatic operations, remain high. Finally, since many of these operations are performed for cancer, ensuring adequate oncological resection, lymph node sampling, and expediting return to adjuvant therapy has not been conclusively demonstrated.

Despite these limitations, several high-volume centers have demonstrated improved outcomes in select patients. Many benefits, however, have only been realized after surmounting long curves. Recently, programs have investigated the role of simulation in shortening the proficiency curve for HPB surgery. Although this arena remains to be fully explored, this article will review the current evidence for simulation in HPB surgery, focusing on the robotic training paradigm developed by the University of Pittsburgh.

The Learning Curve for Robotic HPB Surgery

Proficiency in open pancreatic surgery, particularly the pancreaticoduodenectomy (PD), is associated with an estimated 50–60 cases for surgical trainees and new faculty.^4,5 Based on operative log data from the Accreditation Council for Graduate Medical Education (ACGME) The average resident trainee performs only 11.5 pancreatic resections and 9.4 liver resections during general surgery training, with the International Hepato-Pancreato-Biliary Association requiring 25 liver and 30 pancreatic resections as part of fellowship. Both requirements fall short of the cases need for operative proficiency, particularly for PD. ⁶ Although many of these programs are beginning to incorporate minimally invasive pancreatic and hepatic resections into their training, HPB experience varies widely with no current codification or specific requirements for minimally invasive caseloads.

Recent reports suggest that the learning curve for robotic pancreatic resections is strikingly similar to open pancreatic resections. Based on a series of the first 200 consecutive robotic PD (RPD) at the University of Pittsburgh, inflexions in outcomes were noted within the first 80 cases for several metrics. ⁷ Decreases in blood loss and conversion were noted after 20 cases (600 mL versus 250 mL, P = .002 and 25% versus 3.3%, P < .001). Pancreatic fistula decreased (27.5% versus 14.4%, P = .04) and lymph node harvest significantly increased (17–26, P < .001) after 40 cases, reflecting improved dissection and reconstruction skills. Operative time significantly decreased from 581 to 417 minutes after 80 cases (P < .001). ⁷ By completion of the study, the operative time was comparable with open PDs (OPDs) with clinically relevant fistula rates falling within the range cited in OPD (6.9%).^7,8,9 Expectantly, the learning curve was shorter for patients undergoing more straightforward purely ablative procedures such as robotic distal pancreatectomies (40 cases required for proficiency).

The learning curve for hepatectomies is similar, ranging between 15 and 60 cases, likely related again to the complexity and extent of resection.^10,11,12 Importantly, these learning curves reflect initial experiences as operative techniques were being developed in real time, and appear to be shorter as trainees now benefit from refined surgeries by their experienced robotic mentors. We believe this learning curve can be further streamlined with the implementation of structured robotic curricula that include simulation, among other elements.

Outcomes of Minimally Invasive HPB Surgery Within the Learning Curve

The impact of the learning curve on outcomes needs to be carefully accounted for when comparing two surgical HPB approaches. Initial studies comparing minimally invasive PD (MIPD) and OPD suggested higher postoperative mortality for MIPD. A review of the National Cancer Database (NCDB) nearly a decade ago found a higher 30-day mortality rate in patients undergoing MIPD versus OPD (odds ratio [OR] = 1.87, 95% confidence interval = 1.25–2.80, P = .002). Importantly, 92% of hospitals in the study were low volume (≤ 10 cases/2 years) and most surgeons were likely working through the proficiency curve of MIPD. ¹³ Similarly, outcomes of laparoscopic PD (LPD) in the LEOPARD-2 trial, which compared LPD with OPD, suggest that the learning curve period may be associated with inferior outcomes. This study was terminated prematurely due to concern for increased mortality in the laparoscopic arm. ¹⁴ Notably, although higher volume centers were selected, surgeons were required to have completed only 20 LPD before participation; a requirement well below the number needed for proficiency in RPD.

These data stand in contrast to outcomes published beyond the learning curve. In a recently published study using the NCDB during three time periods (2010–2012, 2013–2014, and 2015–2016), use of the robotic platform to perform RPD significantly increased during the study period. ¹⁵ In this study, improvements in mortality (6.7%–1.8%, P = .013) and lymphadenectomy (18–21 nodes, P = .035) were observed compared with OPD, with no changes in conversion to open surgery, negative margin resections, or readmissions for RPD.

The Role of Simulation in a Formal HPB Robotics Curriculum

Given the considerable effort and time needed to surmount the robotic HPB learning curve, and the relative paucity of complex HPB cases compared with other relatively common procedures, several groups have investigated the use of a structured simulation curriculum, to shorten the learning curve while maintaining acceptable postoperative outcomes. A recent review noted six studies that examined simulation training—using a variety of animal or cadaver models—in laparoscopic liver surgery. ⁹ Many of these studies focused either on participant feedback or attempts at finding a specific model system best able to simulate true laparoscopic liver resections. The robotic daVinci system (Intuitive Surgical- Sunnyvale, CA, USA) however is unique, in that it provides for a built-in simulation experience that allows the trainee to practice skills specific to the robot including while using the same operative consol. Many of these benefits are though to contribute to improved dissection, decreased blood loss and improve precision when performing anastomoses that are unique to the robot platform. ¹⁶

In 2019, a consensus article was published by Fong et al. discussing the need for HPB focused robotic surgery curricula. ¹⁷ The panel felt this should include the acquisition of basic robotic skills, which may not be unique to HPB surgery, as well as an advanced curriculum that focuses on specific skills unique to the challenges of robotic HPB surgery. Although the specific curriculum may be modified for novice and experienced surgeons, both were likely to need a combination of basic training in and out of the operating room, as well as laboratory skills development, including simulation. For advanced use, an additional advanced curriculum was also needed to navigate the complexity of HPB surgery. Current curricula exist through the fundamentals of robotic surgery that utilizes a four module step approach: introduction to robotic systems, didactic instructions for robotic surgery systems, psychomotor skills curriculum, and finally team training and communication skills. ¹⁸ This curriculum is recommended to all surgeons pursuing a robotic surgery practice, but it is not tailored to the challenges of HPB surgery. Although a current curriculum exists through the Fundamentals of Robotic Surgery (FRS), utilizing a four module step approach (introduction to robotic systems, didactic instructions for robotic surgery systems, psychomotor skills curriculum, and team training and communication skills), this curriculum is not tailored to the challenges of complex HPB surgery. ¹⁸

To address this need for a structured HPB curriculum, Hogg et al. at the University of Pittsburgh created and validated a five-step proficiency-based curriculum. Robotic simulation represents the first of those five steps, followed by biotissue reconstruction drills, video library review of HPB cases, intraoperative assessment of technical skills, and didactics focused on quality assessment. ¹⁹ The simulation segment consists of 24 virtual reality modules scored on the Intuitive Surgical Si Backpack simulator, requiring a score of 90% or above in all included fields (e.g., economy of motion, instrument collision, and instrument force) or at least 10 attempts. ²⁰ A pre- and post-test is administered that includes grading 4 of the most difficult simulator tasks on the Backpack (match box 3, ring and rail 2, tubes and continuous suture) and 3 inanimate tasks on the robot. The latter was scored by graders using the objective structured assessment of technical skills (OSATS), which allows not only for assessment of more concrete skills such as time and errors, but also qualitative metrics of instrument handling and tissue exposure.

To evaluate this initial curriculum, 16 surgical oncology trainees completed the curriculum and performed a pre- and post-test. Notably, the median time to completion of the curriculum was 4.2 hours, which allows for broader application of this step of the curriculum to busy surgeons and trainees. Although all fellows showed improvement on their virtual post-test and most improved on their time to task completion and inanimate post-test, completion of the curriculum did not demonstrate statistically significant improvement in errors. ²⁰

The second step in the curriculum focuses on simulating the complex anastomosis required to successfully complete a robotic pancreatoduodenectomy (RPD), namely performing a successful hepaticojejunostomy (HJ), gastrojejunostomy (GJ), and pancreaticojejunostomy (PJ) on bioartificial material. When trainees underwent step 2 of the curriculum, a significant improvement in errors and OSATS was noted after five attempts of GJ and PJ anastomosis, both of which reached proficiency as established by attending surgeons performing the procedure. ²¹ Completion of more than five HJ significantly improved errors and OSATS as well, however, did not reach attending level competency. ²¹ This suggests that biotissue simulation is a crucial part of the HPB simulation experience, and complements the virtual experience by not only improving metrics such as time and OSATS but also decreasing errors. The ability to obtain attending level proficiency during simulation is also notable here, and may ultimately result in a shorter learning curve when trainees then move to formal operative experience. The final steps of the curriculum occur outside of simulation, and include video review, mentored operative curriculum, and skill maintenance. ²⁰

The next evolution of robotic HPB simulation surgery is yet to be seen, but much remains to be done to improve the training of and dissemination of these complex procedures. One area of possible expansion would is patient-specific virtual reality, which has been introduced in other surgical fields. This method utilizes patient imaging to create a simulated operative experience. ²² This may allow advanced HPB surgeons to tailor their training to a specific procedure, challenging case, or anatomic variation, and be well suited for the challenges of learning complex HPB surgery, and may be particularly well suited for the challenged faced in HPB surgery.

Conclusions

The development and implementation of robotic surgery in the field of HPB has been a slow but significant process. HPB surgeries offer a unique challenge due to their complexity and long learning curves. Yet the benefits of the robotic approach for this patient population are notable, decreasing length of stay, blood loss, postoperative complications, and improving quality of life. The use of robotic simulation focused curriculum plays a crucial role in mentoring experienced surgeons and surgical trainees. Although further study remains, early studies suggest a structured simulation curriculum decreases time, technical errors, and improves proficiency, ultimately leading to safe dissemination of minimally invasive HPB surgery safe robotic.