Perioperative textbook outcomes of minimally invasive pancreatoduodenectomy: a multicenter retrospective cohort analysis in a Korean minimally invasive pancreatic surgery registry

Background: The aim of this study is to investigate the perioperative composite textbook outcomes of pancreatic surgery after minimally invasive pancreatoduodenectomy (MIPD). Materials and methods: The cohort study was conducted across 10 institutions and included 1552 patients who underwent MIPD registered with the Korean Study Group on Minimally Invasive Pancreatic Surgery between May 2007 and April 2020. We analyzed perioperative textbook outcomes of pancreatic surgery after MIPD. Subgroup analyses were performed to assess outcomes based on the hospital volume of MIPD. Results: Among all patients, 21.8% underwent robotic pancreatoduodenectomy. High-volume centers (performing >20 MIPD/year) performed 88.1% of the procedures. The incidence of clinically relevant postoperative pancreatic fistula was 11.5%. Severe complications (Clavien–Dindo grade ≥IIIa) occurred in 15.1% of the cases. The 90-day mortality rate was 0.8%. The mean hospital stay was 13.7 days. Textbook outcomes of pancreatic surgery success were achieved in 60.4% of patients, with higher rates observed in high-volume centers than in low-volume centers (62.2% vs. 44.7%, P<0.001). High-volume centers exhibited significantly lower conversion rates (5.4% vs. 12.5%, P=0.001), lower 90-day mortality (0.5% vs. 2.7%, P=0.001), and lower 90-day readmission rates (4.5% vs. 9.6%, P=0.006) than those low-volume centers. Conclusion: MIPD could be performed safely with permissible perioperative outcomes, including textbook outcomes of pancreatic surgery, particularly in experienced centers. The findings of this study provided valuable insights for guiding surgical treatment decisions in periampullary disease.


Introductions
Pancreatoduodenectomy (PD) is a highly complicated abdominal surgical procedure, which procedure becomes considerably more challenging when performed using minimally invasive techniques.The first laparoscopic PD (LPD) was reported by Gagner in 1994 [1] , and robotic PD (RPD) was first reported by Giulianotti et al. in 2003 [2] .Since these first minimally invasive PDs (MIPDs) were introduced, various institutions have adopted these techniques.Furthermore, the effectiveness of MIPD has been examined in several studies, including randomized studies, yielding mixed results [3][4][5][6] .A meta-analysis of randomized controlled trials comparing LPD with open PD (OPD) also concluded that LPD had no advantage over OPD.However, this analysis noted a high risk of bias and moderate-to-low level of evidence [7] .Although it is still early to make generalized conclusions regarding MIPD, primarily due to the selection bias inherent in retrospective studies and the limited number of prospective studies, the frequency of MIPD is gradually increasing [8][9][10] .Studies on the learning curve and training for MIPD have emphasized the need for an MIPD registry to gather evidence for clinical research [11][12][13] .Korean groups have published studies on MIPD and collaborated to establish a registry in line with global trends [9,[14][15][16][17][18] .After the establishment of the Korean study group on Minimally Invasive Pancreatic Surgery (K-MIPS) in 2019 (http://kmips.or.kr/index.html),patients who underwent MIPS were retrospectively and prospectively registered.
Textbook outcomes after pancreatic surgery (TOPS) have recently emerged as a valuable approach for assessing surgical success.TOPS considers multiple quality measures representing the standard or benchmark outcomes that surgeons and medical professionals strive to achieve in pancreatic surgery, providing a comprehensive indication of overall outcomes [19][20][21][22][23] .
In this study, we aimed to investigate the perioperative standard TOPS of MIPD using nationwide databases with a large number of cases.In addition, we compared subgroups classified by hospital volume to determine whether hospital volume affects perioperative outcomes.

Study setting and population
This retrospective multicenter cohort study included 1552 patients who underwent MIPD performed by 24 pancreatic surgeons between May 2007 and April 2020 across 10 institutions in Korea.Consecutive patients who underwent intended MIPD at each institution were included.MIPD included both LPD and RPD.RPD included laparoscopic resection with robotic reconstruction or robotic resection and reconstruction.The research was conducted using data registered in the K-MIPS registry.This study was approved by the institutional review board of Asan Medical Center (Approval number: 2020-0600) and of each participating center.Patients' informed consent was waived because of the retrospective observational nature of this study.Inclusion and exclusion criteria were based on the judgment of the operator in each institution.This study followed the Strengthening The Reporting Of Cohort Studies in Surgery (STROCSS) guidelines [24] (Supplemental Digital Content 1, http://links.lww.com/JS9/C242).

Data collection
Clinical data were collected and analyzed.Postoperative complications were classified according to the Clavien-Dindo classification system [25] .Post-pancreatectomy complications, such as postoperative pancreatic fistula (POPF) [26] , delayed gastric emptying(DGE) [27] , postoperative pancreatic hemorrhage(PPH) [28] , and chyle leakage [29] , were defined and graded based on the International Study Group of Pancreatic Surgery definitions.Bile leakage was defined and graded according to the International Study Group of Liver Surgery criteria [30] .TOPS was defined based on the Dutch Pancreatic Cancer Group criteria [23] , which include the absence of POPF, bile leakage, PPH, Clavien-Dindo ≥ grade III complications, readmission within 90 days, in-hospital or 90-day mortality, or postoperative hospital stay exceeding 14 days.The classifications of malignancy and benign conditions were based on the 2019 World Health Organization classification for tumors of the digestive system [31] .Although all institutions participating in this study were tertiary referral hospitals, they were stratified to evaluate potential differences in postoperative outcomes and TOPS according to the number of MIPDs performed per year as follows: a high-volume center performing > 20 MIPDs/ year and a low-volume center performing ≤ 20 MIPDs/year.Additionally, the analysis was repeated, excluding all patients with missing postoperative outcome values for a more comprehensive assessment.

Statistical analysis
Statistical analyses involved reporting demographics, along with operative, postoperative, and pathological outcomes.Continuous variables were expressed as means standard deviations and compared using the Student's t-test, when applicable.Categorical variables were compared using the χ 2 , Fisher's exact, or linear-bylinear association test.Institutions participating in this study were divided into two groups based on the number of MIPDs performed ( > 20 MIPDs/year vs. ≤ 20 MIPDs/year).These groups were then compared in terms of demographics, perioperative outcomes, and TOPS.And to calculate the TOPS success rate more accurately, the current study performed another analysis on 1239 patients, excluding patients with missing values for postoperative outcomes.For exploratory analysis, we compared the postoperative outcomes based on the number of MIPDs ( > 50 MIPDs/year vs. 20-50 MIPDs/year vs. <20 MIPDs/year).To compare the three groups, this study utilized analysis of variance for continuous variables and the χ 2 test for categorical variables.Subsequently, a Bonferroni-corrected post-hoc analysis was Open conversion rate was 6.3% (n = 97).Among the patients, 3.5% underwent concomitant major vascular resection.The duct-to-mucosa technique (72.8%) was the most frequently employed method for pancreaticojejunostomy, followed by the dunking method.The mean estimated blood loss  1).Interestingly, when the TOPS success rate was calculated using six factors, excluding postoperative hospital stay, there was no difference between the two groups (76.8% vs. 69.8%,P = 0.052).

Comparative analysis of outcomes based on the operation method
Table 4 shows demographics and perioperative outcomes between LPD and RPD.Age and sex were not different between the two groups, but BMI was more higher in RPD groups (

Discussion
The present study represents the first comprehensive analysis of multi-institutional MIPD data from the nationwide registry in Korea, focusing on the TOPS perioperative performance rate.
Notably, this study did not collect data from all institutions that have implemented MIPD in Korea; therefore, experiences are likely to vary among institutions.However, the study included both the most representative institutions with experienced MIPD surgeons and referral institutions with less experienced MIPD surgeons.This makes the findings of this study more informative for patients undergoing MIPD.The present study revealed that perioperative outcomes in MIPD were comparable to those of other studies, with some factors showing more favorable results [32][33][34][35][36] .
In addition to conventional indicators of surgery, we investigated TOPS, which provides a multidimensional measure for assessing the quality of surgical outcomes as a comprehensive indication of overall outcomes.Previous studies have investigated textbook outcomes after hepatobiliary and pancreatic surgery [19][20][21][22][23] .Heidsma et al. [33] reported that TOPS not only signifies successful surgical outcomes but also offers potential oncological benefits.Based on an analysis of 2633 patients, van Roessel et al. [23] reported a TOPS success rate of 47.5% after PD, including a postoperative hospital stay of > 14 days.Lof et al. reported a TOPS rate of 56.4% in a cohort of 250 patients who underwent enhanced recovery after PD.They defined prolonged hospital stay as the 25th highest percentile of the total cohort [26] .The US Neuroendocrine Tumor Study Group reported a TOPS rate of 32.0% after PD [33] .Recently, a multicenter retrospective cohort study in China reported a TOPS rate of 56% after LPD [34] .In the present study, we observed a TOPS success rate of 60.4% after MIPD.
All the institutions participating in the current study are tertiary referral institutions with 800 beds or more, but the number of MIPDs performed differed among hospitals.One low-volume center performed fewer than five MIPD procedures annually, whereas the high-volume institution that was the first in South Korea to perform MIPD performed more than 100 MIPD procedures annually.The criteria for distinguishing between highvolume and low-volume centers vary among published studies [8,37,38] ; in the current study, the criterion for distinguishing between high and low volume was defined as 20 MIPD per year according to the Miami International Evidence-Based Guideline [13] .When comparing perioperative outcomes based on the annual MIPD volume, several differences were observed between the two groups.First, the demographics differed between the two groups.This difference is likely due to the differences in the locations of the centers.All four high-volume centers are located in metropolitan areas, while only one of the low-volume institutions has a metropolitan location, and the other five are located in rural areas.There are differences in the demographic structures of metropolitan versus rural regions in South Korea; for example, the proportion of older individuals is relatively higher in rural areas than in metropolitan areas [39] .The older the patient, the greater the underlying disease and the higher the operative risk.Due to the demographic differences between these patient groups, careful interpretation of the results is needed.Furthermore, regarding complicated fluid collection, ICU care, and the requirement for reoperation, there were too many missing values for the statistical results to be robust.Further multivariate analyses are needed to identify any significant differences in postoperative outcomes according to institutional volume.
High-volume centers were observed to be superior to lowvolume centers in terms of several perioperative outcomes, including transfusion, operation time, PPH, unplanned ICU care, 90-day readmission rate, 90-day readmission rate, and TOPS success rate.Differences in outcomes based on hospital volume have been published in several studies.Adam et al. [35] reported that the MIPD hospital volume is significantly associated with improved MIPD outcomes, with a threshold of 22 cases per year.The Miami International Guidelines on minimally invasive indicate the TOPS success rate, excluding cases that do not meet each TOPS item.The TOPS success rate was higher in high-volume centers than in lowvolume centers (62.2% vs. 44.7%,P < 0.001).Interestingly there was no significant difference according to hospital volume except for hospital stays over 14 days after operation (76.8% vs. 69.8%,P = 0.052).MIPS, minimally invasive pancreatoduodenectomy; POPF, postoperative pancreatic fistula; PPH, postoperative pancreatic hemorrhage.
pancreas resection reported that center volume strongly affects outcomes after minimally invasive pancreas resection.They emphasized that MIPD should be performed in high-volume centers due to reduced mortality and morbidity [13] .The differences in surgical case experience among surgeons, depending on the hospital volume, can explain this disparity.In our study, we also observed variations in the years of experience in performing MIPD based on hospital volume, indicating potential differences among individual surgeons working in those hospitals.A multicenter retrospective study identified that operator experience influenced complication and mortality rates after MIPD [36] .From this standpoint, including the current study, a cautious recommendation can be made to enhance the TOPS success rate by opting for surgery at an institution that conducts more than 20 MIPD annually.
Similar to previous reports, the current study confirmed the existence of variations in TOPS scores based on hospital volume.The primary factor contributing to this disparity was the duration of postoperative hospital stay.First, differences in hospital scale may result in variances in interventions and medical environments, influencing the provision of supportive care after complications.Second, Korea's medical environment, characterized by relatively low medical costs, may allow patients to extend their hospital stay.Nevertheless, the other factors related to TOPS, excluding this aspect, did not exhibit statistically significant differences according to hospital volume.Hence, the establishment of a registry system to disseminate countermeasures against complications and the implementation of a specialized complication response system in low-volume centers could enhance the success rate of TOPS.Recently, a retrospective multicenter analysis conducted in China in 2021 reported no differences in complications other than the length of hospital stay according to hospital volume.Interestingly, they observed differences in length of stay, POPF, operation failure, and 30-day mortality depending on the surgeon's experience [36] .In the present study, the occurrence of severe complications, classified as Clavien-Dindo grade ≥ III (a critical criterion for postoperative lethal outcomes), did not demonstrate statistically significant differences based on the hospital volume of MIPD (15.4% vs. 13.0%,P = 0.412).Similarly, no statistically significant differences were observed in the clinically relevant POPF rates according to the hospital volume of MIPD.When comparing the success rates using the six TOPS factors excluding the postoperative hospital stay, there was no statistical difference between hospital volumes (76.8% vs. 69.8%,P = 0.052).In the current study, 19 of the 24 surgeons started MIPD after experiencing at least 100 cases of open OPD.These findings suggested that surgeons with a certain level of PD experience can quickly reduce the risk of TOPS failure associated with MIPD.
In addition to PD experience, the potential dissemination of surgical knowledge and expertise on MIPD through academic society networks and academic collaborations facilitates the sharing of clinical experiences and best practices among surgeons.This, in turn, can reduce outcome disparities based on the hospital volume of MIPD.The K-MIPS convenes quarterly academic meetings to foster discussions on the clinical experiences and future directions of MIPD.Hence, we recommend all institutions conducting MIPDs participate in a prospective registry to monitor the safety and efficacy of these operations.Moreover, we suggest that sharing the registry's results could facilitate the dissemination of indirect experiences and promote knowledge exchange among healthcare professionals, enabling the adoption of best practices and ultimately improving patient outcomes.
This study additionally compared perioperative outcomes according to surgical methods.Comparative studies of LPD and RPD are currently underway worldwide.A recently published meta-analysis reported that RPD represents less blood loss and fewer wound infections [40] .Kamarajah et al. [41] also reported that RPD was associated with lower open conversion and transfusion rates than LPD.In the current study, RPD also had a lower open conversion rate, shorter operative time, and shorter postoperative hospital stay than LPD.Additionally, it was superior in DGE, PPH, chyle leakage, unplanned reoperation, readmission, and TOPS.However, these results should be interpreted with caution.A total of 84.4% of RPDs in the current study were hybrid operations that underwent laparoscopic resection followed by robotic reconstruction.Second, robotic surgery is still relatively expensive in Korean society; thus, it is targeted at patients with financial means who have private insurance, and selection bias may occur owing to the surgeon's burden of surgery.Therefore, further research is required to determine the superiority of RPD.

Strengths and limitations
The present study had several limitations.First, this study did not include a comparison to open surgery due to the retrospective nature of the data.Thus, further research utilizing prospective randomized comparative designs with larger sample sizes is needed to confirm the safety and efficacy of MIPD and guide clinical practice.Second, not all institutions performing MIPD in Korea participated in this study; therefore, caution is necessary when generalizing the study results.Third, the study data was collected retrospectively, and there may have been missing values, necessitating caution in the interpretation of the results.Despite these limitations, this study holds value as it presents the first analysis results from the MIPD registry of the K-MIPS.Furthermore, to the best of our knowledge, this is the first multicenter study to report on the success rate of achieving TOPS after MIPD.

Conclusions
The multicenter registry of K-MIPS demonstrated that MIPD is currently being performed safely, with acceptable perioperative outcomes, including TOPS, particularly in experienced centers.However, improving postoperative outcomes in low-volume centers remains an important task for the continuous advancement of minimally invasive pancreatic surgery.As the frequency of MIPD continues to gradually increase, the results of this study provide valuable insights for guiding surgical treatment decisions for periampullary disease.

Figure 1 .
Figure 1.Textbook outcome of pancreatic surgery (TOPS) categorized based on hospital volume.The textbook outcome of pancreatic surgery is assessed individually and cumulatively.Each factor's percentage was measured, excluding missing values.The orange line (high-volume center) and blue line (low-volume center)indicate the TOPS success rate, excluding cases that do not meet each TOPS item.The TOPS success rate was higher in high-volume centers than in lowvolume centers (62.2% vs. 44.7%,P < 0.001).Interestingly there was no significant difference according to hospital volume except for hospital stays over 14 days after operation (76.8% vs. 69.8%,P = 0.052).MIPS, minimally invasive pancreatoduodenectomy; POPF, postoperative pancreatic fistula; PPH, postoperative pancreatic hemorrhage.

Table 1
Postoperative complications and pathologic outcomes.

Table 2
Textbook outcome of pancreatic surgery assessed individually and cumulatively.

Table 3
Demographics and perioperative outcomes according to MIPD cases.

Table 4
Demographics and perioperative outcomes according to operation method.