Impact of Obesity on Perioperative Outcomes in Robotic Pancreaticoduodenectomy: A Propensity Score‐Matched Study Authors: Bor‐Uei Shyr, Bor‐Shiuan Shyr, Shih‐Chin Chen, Shin‐E Wang, Yi‐Ming Shyr Published: 2024 ABSTRACT ABSTRACTBackgroundThe aim of this study is to assess the impact of obesity on the perioperative outcomes in robotic pancreaticoduodenectomy (RPD), rarely documented.MethodsA total of 886 patients undergoing RPD or open pancreaticoduodenectomy (OPD) were enroled. These patients were categorised into the obese RPD, non‐obese RPD and obese OPD groups.ResultsOperation time was longer in obese RPD compared with non‐obese RPD and obese OPD (median: 9.5 vs. 7.5 and 8.0 h). Blood loss was lower in both RPD groups than in the obese OPD group (median: 200 and 120 vs. 500 c.c.). Surgical mortality and morbidity were comparable between these three groups. Delayed gastric emptying (DGE) occurred less frequently in both RPD groups than obese OPD (7.5% and 4.4% vs. 19.1%).ConclusionsObesity does not exert a negative impact on the perioperative outcomes in RPD, except for longer operation times. RPD has emerged as a safe and viable surgical approach for obese patients. Introduction Obesity is a rapidly escalating global health concern, elevating the risk of various ailments, including coronary artery disease, diabetes, sleep apnoea, osteoarthritis, and cancer [1, 2, 3]. Notably, pancreaticoduodenectomy, one of the most intricate abdominal surgeries, has historically exhibited a substantial morbidity rate ranging from 40% to 50%. Nevertheless, advancements in surgical techniques and the centralisation of cases have led to a reduction in mortality rates to less than 3%–5% in highly specialised centres [4, 5, 6]. It is widely acknowledged that visceral fat mass can obstruct the surgical process, amplifying surgical challenges, prolonging operative time, and potentially increasing surgical morbidity and mortality in traditional open pancreaticoduodenectomy [3, 6, 7, 8]. The introduction of the da Vinci Robotic Surgical System (Intuitive Surgical, Sunnyvale, CA, USA) has addressed the limitations of laparoscopic surgery, marking a significant breakthrough in the realm of minimally invasive surgery (MIS), including complex pancreaticoduodenectomy. This system features EndoWrist instruments that enhance the range of motion, effectively replicating open surgery, while providing high‐definition, three‐dimensional views of the surgical field. Moreover, it ensures tremor‐free motion for both the camera and instruments, thereby enhancing ergonomics, increasing surgical dexterity, and reducing surgeon fatigue. Notably, one key advantage of this technology is its ability to perform smaller incisions, resulting in reduced wound pain and improved cosmetic outcomes [9]. However, the adoption of the robotic approach for pancreaticoduodenectomy has been gradual, primarily due to the technical challenges involved and the relatively low surgical volume associated with this complex procedure. Despite the slow uptake of the robotic approach for pancreaticoduodenectomy, several studies have demonstrated the feasibility and safety of robotic pancreaticoduodenectomy (RPD) when compared to the traditional open pancreaticoduodenectomy [4, 9, 10]. Obesity is known to heighten surgical risks in various abdominal procedures, yet its impact on RPD remains uncertain [7]. To date, only a limited number of studies have explored the outcomes of RPD in obese patients [1, 7, 10]. Considering the paucity of conclusive data in this area, our study aimed to elucidate the influence of obesity on perioperative outcomes in robotic pancreaticoduodenectomy (RPD). Specifically, we sought to compare the study group of obese RPD with the other two control groups, non‐obese RPD and the obese OPD. It's important to note that all procedures were carried out by the same surgical team employing identical surgical techniques. To mitigate potential selection bias inherent in this retrospective study, we implemented a propensity score matching (PSM) approach, ensuring comparability in relevant confounding factors among the obese RPD study group and the two control groups, comprising non‐obese RPD and obese OPD patients. Materials and Methods We selected patients with periampullary lesions who underwent either RPD or OPD between July 2012 and May 2023 from a prospectively‐collected computer database. To ensure the consistency of surgical techniques, patients who underwent RPD during the learning curve period were excluded from this study [9]. For the purpose of this study, obesity was defined based on the World Health Organization (WHO) criteria for Asian and South Asian populations, with a body mass index (BMI) ≥ 25 kg/m2 [11]. Subsequently, patients were categorised into three groups for propensity score matching (PSM): 1. The study group comprised obese RPD patients with a BMI > 25 kg/m2, group B; 2. The negative control group consisted of non‐obese RPD patients with a BMI < 25 kg/m2, group A; 3. The positive control group included obese OPD patients with a BMI > 25 kg/m2, group C. Initially, a total of 886 patients undergoing pancreaticoduodenectomy were included in this study before PSM, including 183 (33.1%) obese RPD and 370 (66.9%) non‐obese RPD among the 553 RPD group, and 70 (21.0%) obese OPD among the 333 OPD group. Because the study interest was the study group, ‘group B: obese RPD group’, and ‘group A: non‐obese RPD’ and ‘group C: obese OPD’ were negative and positive control groups, only comparisons of study group B with control groups A and C were shown, but not those between negative (group A) and positive (group C) control groups. The entire cohort with original data of RPD and OPD before PSM were also shown and compared. This study received approval from the Institutional Review Board (IRB) of Taipei Veterans General Hospital (IRB‐TPEVGH No.: 2023‐10‐001AC) and was conducted in accordance with the IRB regulations and guidelines. Given the retrospective nature of the study and data anonymisation, the requirement for informed consent was waived. Patient selection for RPD was based on individual preferences following comprehensive counselling regarding the innovative nature of RPD, its advantages and disadvantages, and the availability of robotic technology. Patients with long vascular encasement (> 2 cm) and severe intraabdominal adhesions were excluded from consideration for RPD. We conducted an assessment of demographic and clinical characteristics, encompassing: gender, age, BMI, presence of jaundice, diabetes mellitus (DM) status, American Society of Anaesthesiologists physical status classification system (ASAPS), pancreatic parenchymal texture categorised as soft or hard by the surgeon, pancreatic duct diameter, presence of malignancy, tumour size, and diagnosis of periampullary lesions. In addition to these factors, we also evaluated intraoperative and pathological parameters, which comprised: operation time, mesopancreatic dissection levels, blood loss, vascular resection, resection radicality, lymph node status, perineural invasion, lymphovascular invasion, tumour stage. The postoperative surgical outcomes under scrutiny included: surgical mortality, various complications, including delayed gastric emptying (DGE), post‐operative pancreatic fistula (POPF) and Length of hospital stay (LOS). Study Endpoints The objective of this study was to assess the impact of obesity, defined as BMI ≥ 25 kg/m2, on RPD. This was achieved by comparing the study group of obese RPD (group B) patients with two control groups, namely non‐obese RPD (group A) as negative control and obese OPD (group C) patients as positive control, following propensity score matching (PSM). The primary study endpoint was to elucidate the impact of obesity on perioperative outcomes in RPD, achieved through a comparative analysis of the obese RPD, non‐obese RPD, and obese OPD groups. Furthermore, as a secondary endpoint, we sought to identify the independent predictors of DGE and POPF within these three patient groups. Propensity Score Matching (PSM) Strategy To ensure balanced comparisons and minimise selection bias while estimating the causal treatment effects among the obese RPD, non‐obese RPD, and obese OPD groups, we employed PSM. This approach generated individual propensity scores via logistic regression modelling, utilising six commonly employed covariates to predict the occurrence of DGE or POPF: 1. Type of operation (RPD vs. OPD), 2. age (< 65 vs. > 65 years old), 3. Pancreatic parenchyma texture (soft vs. hard), 4. Pancreatic duct diameter (< 3 mm vs. ≥ 3 mm), 5. Presence of pancreatic cancer/chronic pancreatitis (negative vs. positive), 6. Blood loss (< 400 vs. ≥ 400 c.c.) [5]. Patients from the study group of obese RPD, along with the two control groups of non‐obese RPD and obese OPD, were then paired in a 1:1:1 ratio. Matching began with cases having the largest propensity scores. A specific calliper width of 0.02 standard deviations of the logit of the estimated propensity score was applied during this PSM process. Surgical Technique For all patients in these three groups, we conducted pancreaticojejunostomy reconstruction using the modified Blumgart technique, as previously described in our study [4, 12]. The procedure was consistently performed by the same pancreatic surgical team led by Dr. Shyr YM and Dr. Wang SE. In the RPD group, we employed the Si or Xi da Vinci Surgical System (Intuitive Surgical, Inc., Sunnyvale, CA, USA) to facilitate the pancreaticoduodenectomy. The Harmonic scalpel, an energy device, was utilised for the division of small vessels, while Hem‐o‐lok systems (Teleflex Inc., Chelmsford, MA, USA) were selectively employed for the management of larger vessels. In contrast, for the OPD group, most of the vascular pedicles were either cauterised or selectively ligated with suture‐ties. We did not perform subtotal gastrectomy in all procedures. In the RPD group, the final anastomosis involved a hand‐sewn gastrojejunostomy, performed through an extracorporeal approach with meticulous downward positioning of the stomach. This gastrojejunostomy was placed antecolic, antiperistaltic, and inframesocolic, near the umbilicus region, resulting in a relatively vertical positioning of the stomach following RPD. In the OPD group, the jejunal limb was elevated for the gastrojejunostomy, and the final position of the gastrojejunostomy was antecolic, antiperistaltic, and supramesocolic. Definitions of Surgical Complications In our study, we defined delayed gastric emptying (DGE) as clinically significant, categorised as grade B or C, following the criteria established by the International Study Group of Pancreatic Surgery (ISGPS) [13]. Similarly, post‐operative pancreatic fistula (POPF) was identified as clinically relevant when it met grade B or C criteria, as per the 2016 grading system devised by the International Study Group for Pancreatic Fistula (ISGPF) [14]. Additionally, we classified post‐pancreatectomy haemorrhage (PPH) and chyle leak using the standardized criteria proposed by the ISGPS [15, 16]. Surgical mortality, in our study, was defined as any death occurring within 90 days after surgery, including both the period of admission for the initial operation and any subsequent hospital readmission. Statistical Analysis We conducted our statistical analysis using version 26.0 of Statistical Product and Service Solutions (SPSS Inc., IBM, Armonk, NY, USA). Continuous data were described using median, range, mean, and standard deviation (SD). To compare the means of the two groups, we employed the Student's t‐test. To assess any statistically significant differences between the means of three or more independent (unrelated) groups, we used one‐way analysis of variance (ANOVA). In cases where continuous variables did not adhere to a normal distribution, the Wilcoxon rank‐sum test was applied. Categorical variables were presented as counts and percentages and were compared using either Pearson's χ2 test or Fisher's exact test in contingency tables. For multivariate analysis, binary logistic regression was utilised to identify independent factors. An odds ratio of one means that both groups had the same odds of exposure ; therefore, the exposure probably does not increase the risk of event. An odds ratio of greater than one means that the exposure may increase the risk of event and an odds ratio of less than one means that the exposure may reduce the risk of event. We established confidence intervals at 95%. A p‐value of less than 0.05 was considered statistically significant. Results Following propensity score matching (PSM) at a 1:1:1 ratio (Figure 1), each group, namely non‐obese RPD, obese RPD, and obese OPD, consisted of 68 patients. The demographic characteristics of all patients (RPD 553 and OPD 333) before PSM are shown in Table 1, and those three groups after PSM are summarised in Table 2. The median BMI was notably different among the groups, measuring 27.5 kg/m2 in the obese RPD group, 22.3 kg/m2 in the non‐obese RPD group, and 27.6 kg/m2 in the obese OPD group, demonstrating statistical significance (group A vs. B, p < 0.001). The prevalence of diabetes mellitus (DM) was higher in the obese groups, with 38.2% in the obese RPD group and 32.4% in the obese OPD group, compared to 13.2% in the non‐obese RPD group (group A vs. B, p = 0.001). Nausea and vomiting were notably more common in the obese OPD group, with a prevalence of 38.2%, compared to 16.2% in the non‐obese RPD group and 13.2% in the obese RPD group (group B vs. C, p = 0.001). Additionally, the obese RPD group included a higher proportion of patients with an ASAPS score ≥ 3, standing at 38.2%, in contrast to 19.1% in the non‐obese RPD group and 18.2% in the obese OPD group (group A vs. B, p = 0.014; group B vs. C, p = 0.015). However, there were no significant differences observed with regard to sex, age, jaundice, tumour size, pancreatic parenchyma texture, pancreatic duct diameter, periampullary lesions, malignant disease, and the presence of pancreatic cancer or chronic pancreatitis among these three groups. Propensity score matching with a 1:1:1 ratio based on six commonly employed covariates was used to predict the occurrence of DGE or POPF after robotic pancreaticoduodenectomy (RPD) or open pancreaticoduodenectomy (OPD). A specific calliper width of 0.02 standard deviations of the logit of the estimated propensity score was applied during this PSM process. TABLE 1Demographic characteristics for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy.RPDOPD p valuePatients, n (%)553333BMI< 0.001Median (range)23.6 (15.4–36.2)22.3 (14.5–36.1)Mean ± SD23.8 ± 3.522.9 ± 3.5Sex0.667Female254 (45.9%)148 (44.4%)Male299 (54.1%)185 (55.6%)Age, year old0.068Median (range)67 (13–97)65 (23–88)Mean ± SD65.8 ± 12.664.3 ± 11.3DM164 (29.7%)87 (26.1%)0.259Jaundice268 (48.5%)193 (58.0%)0.006Nausea/vomiting123 (22.2%)94 (28.2%)0.045Body weight loss147 (26.6%)110 (33.0%)0.040ASAPS0.009< 3353 (63.8%)199 (72.9%)≥ 3200 (36.2%)74 (27.1%)Tumour size, cm< 0.001Median (range)3.0 (0.6–11.0)3.0 (1.5–33.5)Mean ± SD3.1 ± 1.43.7 ± 3.1Pancreas parenchyma< 0.001Soft367 (66.4%)177 (53.2%)Hard186 (33.6%)156 (46.8%)Pancreatic duct0.350Non‐dilated ≤ 3 mm277 (50.1%)156 (46.8%)Dilated > 3 mm276 (49.9%)177 (35.2%)Periampullary lesions< 0.001Pancreatic head adenocarcinoma187 (33.8%)158 (47.4%)Ampullary adenocarcinoma136 (24.6%)57 (17.1%)Distal CBD adenocarcinoma29 (5.2%)14 (4.2%)Duodenal adenocarcinoma46 (8.3%)24 (7.2%)Other malignancy91 (16.5%)51 (15.3%)Other benign lesions49 (8.9%)12 (3.6%)Chronic pancreatitis15 (2.7%)17 (5.1%)Malignancy489 (88.4%)304 (91.3%)0.178Pancreatic cancer/pancreatitis202 (36.5%)175 (52.6%)< 0.001Abbreviations: ASAPS, American Society of Anaesthesiologists physical status classification system; BMI, body mass index; CBD, common bile duct; DM, diabetes mellitus; OPD, open pancreaticoduodenectomy; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Demographic characteristics for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy. Abbreviations: ASAPS, American Society of Anaesthesiologists physical status classification system; BMI, body mass index; CBD, common bile duct; DM, diabetes mellitus; OPD, open pancreaticoduodenectomy; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. TABLE 2Demographic characteristics for propensity score matched patients after robotic and open pancreaticoduodenectomy.A: Non‐obese RPD (BMI < 25)B: Obese RPD (BMI ≥ 25)C: Obese OPD (BMI ≥ 25) p valueA versus BB versus CPatients, n (%)686868BMI< 0.0010.955Median (range)22.3 (15.9–24.9)27.5 (25.1–35.0)27.6 (25.1–36.1)Mean ± SD23.8 ± 3.327.9 ± 2.328.0 ± 2.6Sex0.7260.592Female28 (41.2%)26 (38.2%)23 (37.7%)Male40 (58.8%)42 (61.8%)45 (65.2%)Age, year old0.5350.639Median (range)67 (34–95)64 (27–89)64 (35–86)Mean ± SD64.8 ± 13.163.6 ± 12.062.6 ± 10.4DM9 (13.2%)26 (38.2%)22 (32.4%)0.0010.473Jaundice38 (55.9%)29 (42.6%)37 (54.4%)0.1230.170Nausea/vomiting11 (16.2%)9 (13.2%)26 (38.2%)0.6280.001Body weight loss14 (20.6%)12 (17.6%)17 (25.0%)0.6630.295ASAPS0.0140.015< 355 (80.9%)42 (61.8%)45 (81.8%)≥ 313 (19.1%)26 (38.2%)10 (18.2%)Tumour size, cm0.5860.216Median (range)3.0 (0.6–8.5)3.0 (1.5–7.0)3.0 (0.4–13.0)Mean ± SD3.1 ± 1.53.3 ± 1.43.6 ± 2.4Pancreas parenchyma0.8570.592Soft44 (64.7%)45 (66.2%)42 (61.8%)Hard24 (25.3%)23 (33.8%)26 (38.2%)Pancreatic duct0.5850.293Non‐dilated < 3 mm47 (69.1%)44 (64.7%)38 (55.9%)Dilated > 3 mm21 (30.9%)24 (35.3%)30 (44.1%)Periampullary lesions0.9780.682Pancreatic head adenocarcinoma25 (35.8%)30 (44.1%)24 (35.3%)Ampullary adenocarcinoma15 (22.1%)12 (17.6%)15 (22.1%)Distal CBD adenocarcinoma6 (8.8%)6 (8.8%)5 (7.4%)Duodenal adenocarcinoma3 (4.4%)3 (4.4%)7 (10.3%)Other malignancy11 (16.2%)10 (14.7%)12 (17.6%)Other benign lesions6 (8.8%)6 (8.8%)5 (7.4%)Chronic pancreatitis2 (2.9%)1 (1.5%)0Malignancy60 (88.2%)61 (89.7%)63 (92.6%)0.7840.545Pancreatic cancer/pancreatitis27 (39.7%)31 (45.6%)24 (35.3%)0.4880.221Abbreviations: A, Non‐obese RPD; ASAPS, American Society of Anaesthesiologists physical status classification system; B, Obese RPD; BMI, body mass index; C, Obese OPD; CBD, common bile duct; DM, diabetes mellitus; OPD, open pancreaticoduodenectomy; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Demographic characteristics for propensity score matched patients after robotic and open pancreaticoduodenectomy. Abbreviations: A, Non‐obese RPD; ASAPS, American Society of Anaesthesiologists physical status classification system; B, Obese RPD; BMI, body mass index; C, Obese OPD; CBD, common bile duct; DM, diabetes mellitus; OPD, open pancreaticoduodenectomy; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Intraoperative and pathological parameters for all patients (RPD 553 and OPD 333) before PSM are shown in Table 3. After PSM analysis, the operation time was notably longer in the obese RPD group, with a median duration of 9.5 h, compared to 8.0 h in the obese OPD group and 7.5 h in the non‐obese RPD group, indicating statistical significance (group A vs. B, p = 0.006; group B vs. C, p = 0.047) (Table 4). Intraoperative blood loss was significantly lower in the RPD groups, with a median of 200 c.c. In the obese RPD group and 120 c.c. In the non‐obese RPD group, in contrast to 500 c.c. In the obese OPD group (group B vs. C, p < 0.001). The conversion rate in the obese OPD group was 11.8%, higher than that (11.8%) in the non‐obese RPD group (p = 0.049). Lymph node yield was also notably higher in the RPD groups, measuring a median of 17 in both the obese RPD and non‐obese RPD groups, compared to 13 in the obese OPD group (group B vs. C, p = 0.001). On the other hand, other surgical and pathological parameters did not demonstrate significant differences among these three groups. These parameters included mesopancreas dissection, vascular resection, radicality of resection, perineural invasion, lymphovascular invasion, and tumour stage. The vascular resection rate in non‐obese RPD was lower (4.4%) compared with 16.2% in obese OPD (group A vs. C, p = 0.024). TABLE 3Intraoperative and pathological parameters for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy.RPDOPD p valuePatients, n 553333Operation time, hour< 0.001Median (range)8.3 (3.3–16.0)7.5 (3.8–15.0)Mean ± SD8.5 ± 2.37.8 ± 2.0Mesopancreas dissection< 0.001Level 1308 (55.7%)235 (70.6%)Level 1I152 (27.5%)41 (12.3%)Level 1II93 (16.8%)57 (17.1%)Blood loss, c.c.< 0.001Median (range)150 (2–3000)417 (0–7000)Mean ± SD235 ± 395581 ± 637Conversion45 (8.1%)NANALymph node yield< 0.001Median (range)17 (10–49)15 (8–40)Mean ± SD17 ± 615 ± 7Vascular resection48 (8.7%)59 (17.7%)< 0.001Radicality0.791R0507 (91.7%)301 (90.4%)R124 (4.3%)16 (4.8%)R222 (4.0%)16 (4.8%)Lymph node involvement245 (44.3%)174 (52.3%)0.022Perineural invasion283 (51.2%)204 (610.3%)0.003Lymphovascular invasion254 (45.9%)150 (45.0%)0.798Stage0.490I and II324 (71.7%)208 (74.0%)III and IV128 (28.3%)73 (26.0%)Abbreviations: NA, not available; OPD, open pancreaticoduodenectomy; R0, curative resection without residual cancer > 1 mm margin; R1, microscopic residual cancer < 1 mm margin; R2, gross residual cancer; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Intraoperative and pathological parameters for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy. Abbreviations: NA, not available; OPD, open pancreaticoduodenectomy; R0, curative resection without residual cancer > 1 mm margin; R1, microscopic residual cancer < 1 mm margin; R2, gross residual cancer; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. TABLE 4Intraoperative and pathological parameters for propensity score matched patients after robotic and open pancreaticoduodenectomy.A: Non‐obese RPD (BMI < 25)B: Obese RPD (BMI ≥ 25)C: Obese OPD (BMI ≥ 25) p valueA versus BB versus CPatients, n 686868Operation time, hour0.0060.047Median (range)7.5 (4.0–14.0)9.5 (4.7–14.0)8.0 (5.0–14.0)Mean ± SD8.0 ± 2.29.0 ± 2.38.3 ± 2.2Mesopancreas dissection0.6510.232Level 143 (63.2%)41 (60.3%)50 (73.5%)Level 1I15 (22.1%)13 (19.1%)10 (14.7%)Level 1II10 (14.7%)14 (20.6%)8 (11.8%)Blood loss, c.c.0.301< 0.001Median (range)120 (2–950)200 (5–1500)500 (66–7000)Mean ± SD174 ± 168277 ± 275769 ± 944Conversion2 (2.9%)8 (11.8%)NA0.049NALymph node yield0.9190.001Median (range)17 (7–35)17 (12–30)13 (10–40)Mean ± SD17 ± 717 ± 612 ± 7Vascular resection3 (4.4%)9 (13.2%)11 (16.2%)0.0700.628Radicality0.5450.597R061 (89.7%)64 (94.1%)66 (97.1%)R14 (5.9%)3 (4.4%)1 (1.5%)R23 (4.4%)1 (1.5%)1 (1.5%)Lymph node involvement33 (48.5%)30 (44.1%)36 (52.9%)0.6060.303Perineural invasion38 (55.9%)35 (51.5%)35 (51.5%)0.6061.000Lymphovascular invasion34 (50.0%)30 (44.1%)33 (48.5%)0.4920.606Stage0.6910.588I and II40 (70.2%)39 (73.6%)46 (78.0%)III and IV17 (29.8%)14 (26.4%)13 (22.0%)Abbreviations: A, Non‐obese RPD; B, Obese RPD; C, Obese OPD; NA, not available; OPD, open pancreaticoduodenectomy; R0, curative resection without residual cancer > 1 mm margin; R1, microscopic residual cancer < 1 mm margin; R2, gross residual cancer; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Intraoperative and pathological parameters for propensity score matched patients after robotic and open pancreaticoduodenectomy. Abbreviations: A, Non‐obese RPD; B, Obese RPD; C, Obese OPD; NA, not available; OPD, open pancreaticoduodenectomy; R0, curative resection without residual cancer > 1 mm margin; R1, microscopic residual cancer < 1 mm margin; R2, gross residual cancer; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. The surgical outcomes for all patients (RPD 553 and OPD 333) before PSM were listed in Table 5 and for the three groups after PSM were summarised in Table 6. Surgical mortality and morbidity did not exhibit significant differences among these groups. Notably, there was no substantial increase in major complications (Clavien–Dindo ≥ III), wound infections, chyle leakage, intraabdominal abscesses, PPH, or bile leakage in either the obese RPD or obese OPD groups when compared with the non‐obese RPD group. TABLE 5Surgical outcomes for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy.RPDOPD p valuePatients, n 553333Surgical mortality8 (1.4%)4 (1.2%)0.759Surgical morbidity324 (58.6%)196 (58.9%)0.937Postoperative complications0.197Clavien–Dindo 0229 (41.4%)137 (41.4%)Clavien–Dindo I185 (35.3%)127 (38.1%)Clavien–Dindo II52 (9.4%)38 (11.4%)Clavien–Dindo III62 (11.2%)28 (8.4%)Clavien–Dindo IV6 (1.51%)0Clavien–Dindo V (death)9 (1.6%)3 (0.9%)Major complicationClavien–Dindo ≥ III77 (13.9%)31 (9.3%)0.042Wound infection27 (4.9%)29 (8.7%)0.023Chyle leakage141 (25.5%)69 (20.7%)0.105DGE23 (4.2%)55 (16.5%)< 0.001POPF48 (8.7%)36 (10.8%)0.294Pancreas parenchymaSoft, n = 13141 (11.2%)32 (18.1%)0.027Hard, n = 737 (3.8%)4 (2.6%)0.531Pancreatic ductNon‐dilated ≤ 3 mm, n = 12934 (12.3%)28 (17.9%)0.106Dilated > 3 mm, n = 7514 (5.1%)8 (4.5%)0.789PPH33 (6.0%)17 (5.1%)0.590Bile leakage1 (0.2%)1 (0.3%)0.717LOS, day< 0.001Median (range)19 (5–118)24 (7–136)Mean ± SD23 ± 1427 ± 14Abbreviations: DGE, delayed gastric emptying; LOS, length of stay; OPD, open pancreaticoduodenectomy; POPF, postoperative pancreatic fistula (All of the POPF were grade B and no grade C occurred in this series.); PPH, postpancreatectomy haemorrhage; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Surgical outcomes for the entire cohort before propensity score matching after robotic and open pancreaticoduodenectomy. Abbreviations: DGE, delayed gastric emptying; LOS, length of stay; OPD, open pancreaticoduodenectomy; POPF, postoperative pancreatic fistula (All of the POPF were grade B and no grade C occurred in this series.); PPH, postpancreatectomy haemorrhage; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. TABLE 6Surgical outcomes for propensity score matched patients after robotic and open pancreaticoduodenectomy.A: Non‐obese RPD (BMI < 25)B: Obese RPD (BMI ≥ 25)C: Obese OPD (BMI ≥ 25) p valueA versus BB versus CPatients, n 686868Surgical mortality1 (1.5%)1 (1.5%)3 (4.4%)1.0000.310Surgical morbidity39 (57.4%)42 (61.8%)49 (72.1%)0.6090.202Postoperative complications0.7860.679Clavien–Dindo 029 (42.6%)26 (38.2%)29 (42.6%)Clavien–Dindo I25 (36.8%)28 (41.2%)25 (36.8%)Clavien–Dindo II4 (5.9%)7 (10.3%)4 (5.9%)Clavien–Dindo III8 (11.8%)6 (8.8%)8 (11.8%)Clavien–Dindo IV1 (1.5%)00Clavien–Dindo V (death)1 (1.5%)1 (1.5%)2 (2.9%)Major complication0.4371.000Clavien–Dindo ≥ III10 (14.7%)7 (10.3%)7 (10.3%)Wound infection3 (4.4%)3 (4.4%)6 (8.8%)1.0000.301Chyle leakage23 (33.8%)16 (23.5%)13 (19.1%)0.1840.530DGE5 (7.5%)3 (4.4%)13 (19.1%)0.4660.008POPF3 (4.4%)6 (8.8%)14 (20.6%)0.4930.044Pancreas parenchymaSoft, n = 1313 (6.8%)5 (11.1%)13 (31.0%)0.4790.022Hard, n = 7301 (4.3%)1 (3.8%)0.3020.929Pancreatic ductNon‐dilated ≤ 3 mm, n = 1293 (6.4%)5 (11.4%)13 (34.2%)0.4020.013Dilated > 3 mm, n = 7501 (4.2%)1 (3.3%)0.3440.872Intraabdominal abscess3 (4.4%)8 (11.8%)5 (7.4%)0.1160.382PPH4 (5.9%)4 (5.9%)4 (5.9%)1.0001.000Bile leakage01 (1.5%)1 (1.5%)0.3161.000LOS, day0.2130.028Median (range)18 (7–95)22 (7–80)27 (8–99)Mean ± SD23 ± 1526 ± 1728 ± 16Abbreviations: A, Non‐obese RPD; B, Obese RPD; C, Obese OPD; DGE, delayed gastric emptying; LOS, length of stay; OPD, open pancreaticoduodenectomy; POPF, postoperative pancreatic fistula (All of the POPF were grade B and no grade C occurred in this series.); PPH, postpancreatectomy haemorrhage; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. Surgical outcomes for propensity score matched patients after robotic and open pancreaticoduodenectomy. Abbreviations: A, Non‐obese RPD; B, Obese RPD; C, Obese OPD; DGE, delayed gastric emptying; LOS, length of stay; OPD, open pancreaticoduodenectomy; POPF, postoperative pancreatic fistula (All of the POPF were grade B and no grade C occurred in this series.); PPH, postpancreatectomy haemorrhage; RPD, robotic pancreaticoduodenectomy; SD, standard deviation. DGE rates, however, were lower in both the non‐obese RPD and obese RPD groups compared to the obese OPD group. Specifically, DGE occurred in 7.5% of the non‐obese RPD group, 4.4% of the obese RPD group, and 19.1% of the obese OPD group (group B vs. C, p = 0.008). Subsequent multivariate analysis revealed that only RPD was the independent factor to be associated with low DGE (Figure 2). Forest plot and multivariate analysis by binary logistic regression for independent factors associated with delayed gastric emptying (DGE). The control group is marked with an asterisk *. CI, confidence interval; DM, diabetes mellitus; OPD, open pancreaticoduodenectomy; POPF, post‐operative pancreatic fistula; RPD, robotic pancreaticoduodenectomy. All of the POPF were grade B, and no grade C occurred in this series. POPF rates were significantly lower in both the non‐obese RPD and obese RPD groups, standing at 4.4% and 8.8%, in contrast to 19.1% in the obese OPD group (group B vs. C, p = 0.044). This substantial reduction in POPF was consistently observed in patients with soft pancreatic parenchyma (6.8%, 11.1%, and 31.0%, respectively, group B vs. C, p = 0.022) and those with non‐dilated pancreatic ducts (6.4%, 11.4%, and 34.2%, respectively, group B vs. C, p = 0.013). Multivariate analysis identified the independent factors associated with a lower risk of POPF, which included undergoing RPD, being under the age of 65 years, and having a dilated pancreatic duct > 3 mm (Figure 3). Forest plot and multivariate analysis by binary logistic regression for independent factors associated with post‐operative pancreatic fistula (POPF). The control group is marked with an asterisk *. CI, confidence interval; OPD, open pancreaticoduodenectomy; RPD, robotic pancreaticoduodenectomy. Additionally, length of stay (LOS) was shorter in both the non‐obese and obese RPD groups compared to the obese OPD group, with a median of 18 and 22 days, as opposed to 27 days (group B vs. C, p = 0.028). Discussion Obesity has evolved into a substantial global health concern, with projections indicating that by 2030, approximately 58% of the population may be obese or overweight [10]. As our study demonstrates, obese individuals often present with comorbidities, such as DM and an ASAPS score ≥ 3. Historically, obese patients undergoing OPD have faced increased surgical risks, including prolonged operation time, a higher likelihood of major postoperative complications such as POPF and DGE, elevated intraoperative blood loss, surgical site infections, wound disruption, extended hospital stays, and even readmission [1, 2, 10, 17, 18, 19, 20, 21]. Surgery in obese patients is often technically more challenging due to the presence of excess adipose tissue, resulting in prolonged operative time and increased blood loss. Furthermore, the fatty degeneration of the pancreas in obese patients can pose difficulties in reconstruction and represents a well‐established risk factor for the development of POPF [2, 8]. Performing robotic surgery in obese patients also presents unique challenges. The thicker subcutaneous fat necessitates precise placement of Trocar orifices, which can be prone to slip and can limit the pneumoperitoneal space. Additionally, obese patients tend to have increased visceral fat in the omentum and mesentery, rendering them more susceptible to bleeding and bowel injury during RPD [10]. Despite these challenges, RPD offers significant advantages, including the potential for smaller incisions, more precise manipulations, and faster postoperative recovery [19]. These benefits can help mitigate some of the difficulties associated with obesity in pancreaticoduodenectomy patients. To the best of our knowledge, no published comparisons exist regarding the outcomes of patients undergoing RPD between obese and non‐obese individuals. In this PSM study, we compared the study group (obese RPD group) not only with the positive control group (obese OPD group) but also with the negative control group (non‐obese RPD group). Our study, when comparing the non‐obese RPD group in this study, demonstrates that perioperative outcomes in both obese and non‐obese RPD groups are largely comparable across various parameters. In this PSM study, we compared the study group (obese RPD group) not only with the positive control group (obese OPD group) but also with the negative control group (non‐obese RPD group). Our study, when comparing the non‐obese RPD group in this study, demonstrates that perioperative outcomes in both obese and non‐obese RPD groups are largely comparable across various parameters. These parameters include intraoperative blood loss, lymph node yield, surgical radicality, surgical mortality, surgical morbidity, major complications, delayed gastric emptying (DGE), postoperative pancreatic fistula (POPF), and length of hospital stay (LOS), with the exception of a longer operation time and higher conversion rate in obese RPD group. Although LOS is shorter in RPD groups than in the obese OPD group (p < 0.001), LOS is still long in both non‐obese and obese RPD groups with a mean of 23 and 26 days, respectively. Most of the patients would like to stay late in hospital until the drains are removed since they do not need to pay the charge of hospitalisation which is paid by our national insurance. Our findings indicate that obesity does not have a negative impact on perioperative outcomes in RPD. Meanwhile, compared with the obese OPD group, our PSM study revealed several advantages associated with obese RPD. Specifically, obese RPD was linked to lower intraoperative blood loss, a higher lymph node yield, lower rates of DGE, higher rates of POPF, and shorter hospital stays. The higher lymph node yield in the RPD groups could be attributed to the mediculous dissection under a magnified view of the robotic scope. It's important to note that POPF is a particularly concerning complication following pancreaticoduodenectomy, while DGE is a common postoperative issue encountered after OPD, affecting a range of patients from 4.5% to 56.1% [5, 22, 23]. These complications are often associated with extended hospitalisation [22]. Previous research has identified overweight/obesity as a risk factor for both POPF and DGE in various studies [1, 2, 6, 10, 17, 18, 19, 20, 21]. However, our multivariate analysis reveals that only RPD is independently associated with a reduced risk of DGE, while independent factors linked to a lower risk of POPF are RPD, age younger than 65 years, and a pancreatic duct diameter greater than 3 mm. Additional studies, such as the one by Rosso et al. [24], have found that pancreatic fatty infiltration exceeding 10% constitutes a risk factor for POPF after pancreaticoduodenectomy, with age and BMI serving as useful predictors of the percentage of pancreatic fat. Another PSM study conducted by Li et al. [10] also concluded that RPD showed advantages in terms of reduced blood loss and a lower incidence of Clavien–Dindo III‐V complications when compared with OPD in overweight and obese patients. Furthermore, Girgis et al. [1] demonstrated that a robotic approach to pancreaticoduodenectomy can lead to a reduction in major complications, wound infections, and POPF in obese patients. They suggested that the robotic approach was an independent predictor of fewer wound infections and less pancreatic fistula formation. Based on the findings of our study and literature reports, RPD seems to be a safe and feasible surgical approach for obsess patients, with comparable perioperative outcomes with the non‐obese RPD group. Limitations This retrospective comparison conducted at a single institution using a prospectively maintained database comes with certain limitations. One of the unavoidable challenges is the potential for selection bias, despite our efforts to address this issue through PSM, which aims to emulate some of the specific characteristics of a randomized controlled trial. Conclusion In conclusion, our findings indicate that obesity does not exert a negative impact on the perioperative outcomes in patients with RPD, with the exception of longer operation times when compared to patients with non‐obese RPD. Both obese and non‐obese RPD groups exhibit favourable outcomes, including reduced blood loss, higher lymph node yield, lower rates of DGE, lower rates of POPF, and shorter hospital stays, in comparison to obese patients undergoing OPD. Multivariate analyses also confirm that RPD is independently associated with a lower risk of DGE and POPF. Consequently, RPD emerges as a safe and viable surgical approach for obese patients, demonstrating comparable perioperative outcomes to the non‐obese RPD group and improved surgical outcomes compared to obese patients undergoing OPD. Author Contribution Bor‐Uei Shyr and Bor‐Shiuan Shyr: Conception of the study, acquisition, formal analysis and interpretation of data, writing original draft, and critical revision of the manuscript. Shih‐Chin Chen: Acquisition, validation, curation and visualisation of data and critical revision of the manuscript. Shin‐E Wang: Conception of the study, acquisition, formal analysis and interpretation of data, critical revision of the manuscript. Yi‐Ming Shyr: Resources, formal analysis and interpretation of data, critical revision of the manuscript. All authors have read and approved the final version of the manuscript. Ethics Statement The study was approved by the Institutional Review Board (IRB) of Taipei Veterans General Hospital (IRB‐TPEVGH No.: 2023‐10‐001AC). Consent Given the retrospective nature of the study and data anonymisation, the requirement for informed consent was waived. Conflicts of Interest The authors declare no conflicts of interest.