The Brescia Internationally Validated European Guidelines on Minimally Invasive Pancreatic Surgery (EGUMIPS)

Objective: To develop and update evidence-based and consensus-based guidelines on laparoscopic and robotic pancreatic surgery. Summary Background Data: Minimally invasive pancreatic surgery (MIPS), including laparoscopic and robotic surgery, is complex and technically demanding. Minimizing the risk for patients requires stringent, evidence-based guidelines. Since the International Miami Guidelines on MIPS in 2019, new developments and key publications have been reported, necessitating an update. Methods: Evidence-based guidelines on 22 topics in 8 domains were proposed: terminology, indications, patients, procedures, surgical techniques and instrumentation, assessment tools, implementation and training, and artificial intelligence. The Brescia Internationally Validated European Guidelines on Minimally Invasive Pancreatic Surgery (EGUMIPS, September 2022) used the Scottish Intercollegiate Guidelines Network (SIGN) methodology to assess the evidence and develop guideline recommendations, the Delphi method to establish consensus on the recommendations among the Expert Committee, and the AGREE II-GRS tool for guideline quality assessment and external validation by a Validation Committee. Results: Overall, 27 European experts, 6 international experts, 22 international Validation Committee members, 11 Jury Committee members, 18 Research Committee members, and 121 registered attendees of the 2-day meeting were involved in the development and validation of the guidelines. In total, 98 recommendations were developed, including 33 on laparoscopic, 34 on robotic, and 31 on general MIPS, covering 22 topics in 8 domains. Out of 98 recommendations, 97 reached at least 80% consensus among the experts and congress attendees, and all recommendations were externally validated by the Validation Committee. Conclusions: The EGUMIPS evidence-based guidelines on laparoscopic and robotic MIPS can be applied in current clinical practice to provide guidance to patients, surgeons, policy-makers, and medical societies.

(Ann Surg 2024;279:45-57) M inimally invasive pancreatic surgery (MIPS) has become increasingly popular in the last decades and is now considered an important part of current pancreatic surgery practice.][10][11][12][13] Nevertheless, MIPS-including laparoscopic and robotic surgery-is complex and technically demanding surgery associated with a long learning curve and high postoperative morbidity rates, and as such, it calls for stringent implementation of evidence-based guidelines to minimize patient harm.For this reason, in 2019, the Miami International Evidence-based Guidelines on Minimally Invasive Pancreas Resection (IG-MIPR) were established, 14 aiming to guide the safe adoption of MIPS covering many relevant topics such as indications, patient selection, learning curves, training, and center volumes.However, since then, new literature has become available, and a significant expansion in robotic procedures has been seen.The IG-MIPR mainly included general guidelines on MIPS without distinguishing between laparoscopic and robotic surgery.The guidelines concluded that laparoscopic, robotic, and open pancreas resection each have their role, and future research should focus on the utility and (technical) advantages of each approach.][17][18] Therefore, an update of the previous guidelines, which includes separate guidelines for laparoscopic and robotic pancreatic surgery, is needed.The First Internationally Validated European Guidelines on Minimally Invasive Pancreatic Surgery (EGUMIPS) finally achieved consensus in Brescia (September 2022), aiming to provide new terminology on surgical approaches and separate guidelines on robotic and laparoscopic surgery based on the most recent available body of evidence.The Brescia EGUMIPS guidelines were developed by a large faculty of experts and researchers following an unique combination of a validated and novel methodology covering 8 relevant domains: terminology, indications, patients, procedures, surgical techniques and instrumentation, assessment tools, implementation and training, and artificial intelligence.

METHODS
The guideline development followed 3 validated methodologies, which had previously been used in the development of the Miami guidelines.The Scottish Intercollegiate Guidelines Network (SIGN) methodology was used to assess the evidence and develop guideline recommendations by working groups consisting of Experts and Researchers, 19 the Delphi method to establish consensus on the recommendations among the Expert Committee, 20 and the AGREE II-GRS tool for methodological guideline quality assessment and external validation by the Validation Committee. 21The Validation Committee functioned independently, as it did not participate in formulating the recommendations and did not receive any information regarding the specific details of the guidelines before the meeting.They were only regularly updated on the ongoing overall process by the Chairman.To validate and assess the public voting process, to evaluate the interaction between the Expert and Validation Committee, and to ensure all methodologies were followed correctly, an independent Jury Committee was appointed.The Jury Committee completed a specifically designed form after each meeting day to assess quality aspects of the guidelines development process.
Before the development of the guidelines, these different Committees were established.First, a Steering Committee of 6 members was established on the grounds of their clinical and scientific expertise and knowledge in MIPS.This committee and the congress Chairman identified an Expert Committee of 21 European and 6 international experts, a Validation Committee of 22 members including 17 pancreatic surgeons, 2 methodologists, and 3 patients' representatives, a Jury Committee of 11 members, and a Research Committee of 18 members dedicated to research in MIPS (Supplementary Appendix S1-6, Supplemental Digital Content 1, http://links.lww.com/SLA/E730).In the Expert Committee and Validation Committee, a geographically balanced selection was ensured between surgeons practicing only open surgery, surgeons practicing mainly minimally invasive surgery, and surgeons practicing both.After the group selection, 8 key domains for guideline development were identified by the congress Chairman and the Steering Committee, which included: terminology, indications, patient selection, procedures, surgical techniques and instrumentation, assessment tools, implementation and training, and artificial intelligence.All domains were subsequently subdivided into 22 relevant topics with a total of 29 clinical questions on laparoscopic, 29 on robotic, and 18 on general MIPS, created and reviewed by the Chairman and the Steering Committee.The Expert Committee, the Steering Committee, and the Research Committee members were divided into working groups over the 8 different domains.Steering Committee members participated in 2 domains.Experts were allocated to a domain based on their expertise, and an equal number of laparoscopic-focused surgeons and roboticfocused surgeons within each domain was assured.Experts were allocated to either the laparoscopic or the robotic questions based on their common practice approach to achieve reliable and separate evidence on both approaches.Eventually, each working group consisted of 4-6 experts (2-3 laparoscopic and 2-3 robotic) and 3-5 surgical researchers.
For every domain, systematic reviews of available literature on robotic and laparoscopic pancreatic surgery, including comparative studies with open pancreatic surgery, were performed by the working groups using the PubMed, Embase, and Cochrane databases (the overall PRISMA diagram is shown in Supplementary Figure 1, Supplemental Digital Content 1, http:// links.lww.com/SLA/E730).Studies with a minimum sample size of 20 patients and published in the English language were included.All studies found eligible after screening were reviewed and summarized in separate evidence tables.The SIGN methodology was used to score the quality of each study and assign a level of evidence (Supplementary Appendix S7, Supplemental Digital Content 1, http://links.lww.com/SLA/E730). 19ased on the evidence and their quality, recommendations were formulated for each clinical question by the experts of the working group.A form of recommendation, based on a GRADE rating (ie, strong or weak, Supplementary Appendix S8, Supplemental Digital Content 1, http://links.lww.com/SLA/E730), 22 was given for every recommendation as well.Each group delivered their final recommendations with a GRADE rating and the included evidence to the Chairman.In total, 4 online meetings were held with the Research Committee and the Chairman to evaluate the literature review process (dates March 8, 2022, April 2, 2022, May 4, 2022, June 7, 2022).The recommendations of all domains were merged into a questionnaire and circulated to the experts for a first voting round, per the Delphi methodology. 20Experts had the voting option to agree or disagree with a particular recommendation and could also comment.Recommendations that achieved an agreement rate of 80% or higher were approved; otherwise, recommendations were returned to the original working group to be revised.Revised recommendations were entered into a second Delphi voting round, after which the identical procedure was repeated.Results of the voting rounds were only accessible by the Chairman and were kept anonymous.On June 23 and August 4, 2022, the first and second Delphi questionnaires respectively, were sent to all experts.On September 28, 2022, before the official guideline meeting, a premeeting was held with the Chairman, Steering Committee, Expert Committee, and Research Committee.During this meeting, a final third Delphi round was held where comments on all recommendations were discussed and minor changes were made.On September 29 and 30, 2022, an in-person meeting took place.
At the start of the plenary meeting, a professional oath ensuring the commitment to an unbiased process was sworn publically by each Committee leader to the Chairman.During the 2-day meeting, all evidence-based recommendations were presented by each domain working group.After each statement, the attending audience (n = 170, consisting of residents, fellows, or surgeons who registered for the conference through the EGUMIPS website and Expert and Jury committee members) voted using a digital voting device for an agreement or disagreement on the given statement.The final audience vote was immediately shown on the presentation screen for transparency and to encourage the discussion.For each topic, the Validation Committee assessed the guideline process and quality according to the AGREE II-GRS instrument 21 and reviewed the language of the recommendations.This was done during private Validation Committee sessions after each domain presentation.At the end of the 2-day meeting, the Validation Committee presented a report including the quality scores on each topic and suggestions for adjustments or eliminations.The audience revoted on the recommendations revised by the Validation Committee and that initially received an audience agreement rate below 80%.All other adjustments and suggestions were reviewed and accepted by the Chairman, Steering Committee, and Expert Committee.
The Brescia guidelines were endorsed by the International Hepato-Pancreato-Biliary Association, European-African Hepato-Pancreato-Biliary Association, Society of American Gastrointestinal and Endoscopic Surgeons, Società Italiana di Chirurgia, International Consortium on MIPS, European Consortium on MIPS, Associazione Italiana per lo Studio del Pancreas and Women in Surgery (Supplementary Appendix S9-14, Supplemental Digital Content 1, http://links.lww.com/SLA/E730).Representative members of each of those societies were among the experts participating in the guideline process.

RESULTS
The 8 domains consisted of 22 topics, including 76 clinical questions; 29 on laparoscopic, 29 on robotic, and 18 on general MIPS.Eventually, 98 evidence-based recommendations were established; 33 for laparoscopic, 34 for robotic, and 31 for general MIPS.A flowchart of the process is shown in Supplementary Figure 2, Supplemental Digital Content 1, http://links.lww.com/SLA/E730.The complete set of laparoscopic (L), robotic (R), and general (G) questions, recommendations, and GRADE rating per domain and topic is provided in Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/E730.A more extensive document, including the expert agreement rate, audience agreement rate, topic quality score, comments, and literature, is provided as Supplementary File A, Supplemental Digital Content 1, http://links.lww.com/SLA/E730.
For many topics, the recommendations for laparoscopic and robotic approaches were similar, although the laparoscopic recommendations generally received a stronger GRADE due to a higher level of evidence.Differences between the laparoscopic and robotic recommendations were mainly observed in the topics of learning curves, cost-effectiveness, and artificial intelligence.Compared with the Miami Guidelines, 3 new domains were introduced: terminology, assessment tools, and artificial intelligence.

Domain 1: Terminology
In the domain terminology, definitions were established for the different types of surgical approaches and conversions.The recommendations are shown in Table 1.The new set of agreed definitions of surgical approaches is shown in Table 2.

Domain 2: Indications
In the domain indications, both laparoscopy and robotassisted were considered alternative approaches to distal pancreatectomy and pancreatoduodenectomy in the treatment of benign, premalignant, and malignant lesions when performed by experienced surgeons in high-volume centers.The strengths of the recommendations were higher for those related to distal pancreatectomy.The recommendations were not profoundly different from the Miami Guidelines and are shown in Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/E730.

Domain 3: Patients
In the domain patients, no contraindications were identified for laparoscopic and robotic pancreatic resections regarding age, obesity, previous abdominal surgery, and size of the lesion (see Supplementary Table S1, Supplemental Digital Content 1, http://links.lww.com/SLA/E730), as also stated in the Miami Guidelines.Scarce evidence exists regarding the use of vascular resection and neoadjuvant therapy before laparoscopic and robotic pancreatic resections.Further investigation into this topic is warranted.

Domain 4: Procedures
The recommendations of the domain procedures are shown in Table 3.Both the laparoscopic and robot-assisted approaches were considered appropriate alternatives for enucleation, total pancreatectomy, and vessel-sparing and vesselresecting spleen-preserving distal pancreatectomy.The role of both approaches in central pancreatectomy has yet to be determined.Also, there is insufficient evidence to define a superior anasomostic technique during robotic and laparoscopic pancreatoduodenectomy, and it remains at the surgeon's preference.In laparoscopic pancreatic surgery, the unplanned use of a laparotomy to complete the procedure must be defined as a nonurgent conversion.

Expert opinion Strong (upgraded by experts)
R7 How should we define an unintended conversion in RAS (I.g.gastrojejunostomy performed open, even though it was initially planned laparoscopically)?
In robot-assisted pancreatic surgery, the unplanned use of a laparotomy to complete the procedure must be defined as a nonurgent conversion.
Expert opinion Strong (upgraded by experts)

Domain 5: Surgical Techniques and Instrumentation
In the domain of surgical techniques and instrumentation, a wide set of recommendations is provided for techniques in pancreatoduodenectomy and distal pancreatectomy, surgical devices, vessel and hemorrhage control, stump closure after distal pancreatectomy, and drain management.The recommendations are shown in Tables 4-6.

Domain 6: Assessment Tools
In the domain assessment tools, the following core parameters in the assessment of laparoscopic and robotic MIPS were defined: severe morbidity, mortality, postoperative pancreatic fistula, conversion rate, and patient-reported outcomes.R0 resection rate, 3-year overall survival, and disease-free survival were considered core outcomes for PDAC.Several outcome measurements such as Benchmarks, Textbook Outcome, Comprehensive Complication Index, and Clavien-Dindo classification and patient outcomes as Patient Reported Outcome Measures (PROMs) and Quality-Adjusted Life Year (QALY) were considered suitable to assess the validity and efficacy of laparoscopic and robotic pancreatic resections, but it was deemed necessary to develop a multidimensional composite outcome measure to assess

TABLE 2. Terminology and Definitions of Surgical Approaches
Approach Definition

Laparoscopic
The procedure is fully performed through laparoscopic ports.Roboscopic (This applies to PD/TP/CP) The procedure is performed minimally invasively, using both laparoscopic and robot-assisted approaches.It is characterized by the placement of 3-4 robotic ports and 1 or more laparoscopic ports.The robot can be docked at any time during the surgery.The resection is performed laparoscopically and reconstructive phase by combining the laparoscopic and robotic techniques.At least 1 anastomosis is performed using the robot-assisted technique.

Pure Robotic
The procedure is performed through 3-4 robotic ports and 1 or more laparoscopic ports.The robot is docked at the beginning of the surgery.Both the pancreas resection and reconstructive phase (when expected) are carried out using robotic instruments.No laparoscopic energy device is used in purerobotic procedures.Robot-assisted (This applies to PD/TP/ CP/DP) The procedure is performed through 3-4 robotic ports and 1 or more laparoscopic ports.The robot is docked at the beginning of the surgery.The pancreas resection phase is carried out using both robotic and laparoscopic instruments.When expected, the reconstructive phase is carried out using exclusively robotic instruments.

Open
The procedure is fully performed through a laparotomy incision without the use of any minimally invasive technique.

Hand-assisted Laparoscopic
The procedure is performed through laparoscopic ports and an auxiliary hand port.The procedure is performed laparoscopically, 1 surgeon's hand is placed through the hand port and mostly used for retraction and palpation.

Hand-assisted Robotic
The procedure is performed under robotic assistance and through robotic ports.An auxiliary hand-port is also used.The procedure is performed robotically, 1 surgeon's hand is placed through the hand port and mostly used for retraction and palpation.

Single-port Laparoscopic
The procedure is performed through a single glove port using several either standards or specific laparoscopic instruments.

Single-port Robot-assisted
The procedure is performed using a single specific robotic access with or without an additional robotic/ laparoscopic port.Combined Robot-assisted/Open It is a combined robotic/open procedure.The resection phase of the procedure is performed with a robotassisted approach.During the reconstructive phase, at least 1 of the anastomoses is performed by a minilaparotomy.This procedure is further classified according to the number of anastomoses performed with open approach as follows: Type I: only While there was already a lot of evidence on laparoscopic distal pancreatectomy and laparoscopic pancreatoduodenectomy (LPD), robotic distal pancreatectomy and robotic pancreatoduodenectomy (RPD) have increasingly been studied in recent years due to the latest emergence of robotic platforms.After conflicting results from the multicenter LEOPARD-2 trial 10 that raised concerns regarding the safety of LPD, on the one hand, and the PLOT trial 8 and the PADULAP trial 9 that reported positive results after LPD, on the other, interest in the role of RPD increased.The recently completed German monocenter EUROPA trial (DRKS00020407), the ongoing European multicenter DIPLOMA-2 trial (ISRCTN27483786), and the Chinese multicenter PORTAL trial (NCT04400357) are expected to shed more light on this.Initially, the Validation Committee and audience disagreed with some of the experts' recommendations on the domain "indications."Wording such as "should be considered" was perceived as overly firm given the limited currently available and debatable evidence.A constructive discussion within the Validation Committee led to the proposal to soften and clarify the statements by changing the wording.When the public voted again on those revised statements, all of them achieved at least 80% agreement.Moreover, the domain "indications" was further divided into specific indications such as benign, premalignant, and malignant lesions to reflect the different levels of severity or potential for harm associated with each type of condition.Studies that reported outcomes for all indications only were therefore excluded during the literature review phase.As a result, the current established guidelines are based on studies that reported outcomes separately for benign, premalignant, or malignant indications.In our opinion, this makes the current guidelines more reliable.
Besides indications, surgeon learning curves and minimum center volumes were the most debated topics.Defining center volumes and learning curves was considered crucial to guarantee the safety of MIPS.However, it also raised the fear that a universal minimum number of cases per year considered acceptable for maintaining skills cannot be reached worldwide, mostly because of low volume in some countries or the lack of centralization.In addition, cultural work dynamics have their impact.For example, even in countries where centralization is strongly implemented, a unit with a larger number of surgeons working independently would still not allow one surgeon to perform 20 MIPDs.While in other countries, multiple surgeons may be involved in the same procedure, each performing and registering one part of the procedure as the first operator, thus making it difficult to assess the true number of procedures performed per surgeon and, herewith, the real learning curve.Moreover, several confounding factors can strongly affect the learning curve assessment and definition.Those include previous surgical experience, previous MIPS experience, and previous MIS experience in other fields, such as minimally invasive liver surgery, which is common considering that most pancreatic surgeons are also hepatobiliary surgeons.This reflects the difficulty to achieve firm and generalized agreements on those topics, which translated into a lack of audience support during the voting rounds.After the revisions made by the Validation Committee implying that centers with a lower volume but with  a well-trained multidisciplinary pancreas team should not be excluded from MIPS when acceptable outcomes are guaranteed, the audience accepted all final recommendations.Furthermore, the Brescia guidelines introduce 3 important novel domains: terminology, assessment tools, and artificial intelligence (AI).Despite the 2017 IHPBA guidelines on terminology, there is still a large variety in how different approaches are reported as surgeons adopt the robot differently during their surgical procedures.Some report using the robot and only robotic instruments throughout the whole procedure, while others report using a combination of laparoscopic and robotic instruments.Consequently, numerous terms are currently interchangeably used in literature, such as "robot-assisted," "robotic," "roboscopic," and "hybrid," without clear definitions.However, as those approaches have different surgical, clinical, and economic implications, they would affect the precision and homogeneity of research studies and outcome comparisons.Including a domain on terminology in the Brescia guidelines has now resulted in a new set of terminology and definitions, agreed on with high consensus rates by the experts and the public."Robotic" should be used as an umbrella term to describe the general use of robotics in MIPS, while "robot-assisted" and "pure-robotic" should be used as procedure-specific terms to differentiate the precise use of the robot during a surgical procedure.The new terminology could not retrospectively be adjusted in the questions and recommendations according to the guideline methodology, but it is agreed that these definitions should now be adopted in databases, registries, and studies to ensure a homogenous language for the whole surgical community.We would support this new terminology being adopted by other surgical specialties as well for the sake of standardization of surgical terminology and clarity in reporting.
Similarly, the domain of AI was included as it is increasingly being used in medical practice and has demonstrated growing applicability.This domain was welcomed by the pancreatic society.Although there is still little evidence available on this topic, consensus could be achieved on 4 recommendations regarding its role and position in future MIPS.Lastly, assessment tools were incorporated as a new domain to comply with the increasing importance of accurate outcome measurement in surgical care.In this domain, all outcome metrics were elaborated, and their applicability was clarified to enable future accurate outcome comparisons.It was, however, stressed that the available outcome metrics do not tell the whole surgical story, leaving a great need for a more holistic end point that takes into account both surgical and in particular, patient aspects.This was also emphasized by the 3 patient representatives who were included in the Validation Committee, since patients' perspective, in addition to surgical outcomes, is an important indicator of surgical quality.Their suggestions received considerable room and attention during the private Validation Committee deliberations, which enriched the current guidelines with a different point of view.
Compared with the previous guidelines, new evidence has emerged on various topics, including the role of MIPS in cancer, spleen preservation, learning curves, drain management, and center volumes.Still, limited evidence is available on the best anastomotic techniques in MIPD, central pancreatectomy, quality of life, and cost-effectiveness of the robot-assisted approach.Future research is encouraged to explore the advantages of both approaches and address the aforementioned knowledge gaps.

CONCLUSION
The 2022 EGUMIPS meeting in Brescia has resulted in a large number of evidence-based recommendations on laparoscopic and robotic pancreatic surgery, established by a group of recognized international and European experts in the field of minimally invasive and open pancreatic surgery.The Brescia guidelines provide the most up-to-date evidence and can provide evidence-based guidance to pancreatic surgeons, policy-makers, and patients.

TABLE 1 .
Domain 1 Terminology; Questions and Recommendations on Laparoscopic (L), Robotic (R) and General (G) MIPS R4 Do all conversions to open in RAS have the same impact on patients' outcomes?In robot-assisted pancreatic surgery, urgent conversions are usually associated with an adverse impact on patients' outcomes compared with nonurgent conversions.An effort should be made to perform an elective conversion before getting into an emergency conversion.

TABLE 5 .
Domain 5 SurgicalTechniques and Instrumentation; Questions and Recommendations on Laparoscopic (L), Robotic (R), and General (G) MIPS When appropriate, dissection between the pancreas and splenic vessels should be carefully performed with a combination of blunt dissection and energy devices after complete mobilization of the colonic splenic flexure.When dividing the pancreas at the level of the neck, clear visualization of the splenic/portal vein junction should be obtained before ligation and division of the splenic vein.When dividing the pancreas to the left of the celiac trunk, the splenic vessels could be individually ligated or incorporated in the pancreatic division according to surgeon preference.