You are on page 1of 8

Surgical Endoscopy and Other Interventional Techniques

https://doi.org/10.1007/s00464-021-08437-7

Propensity score‑matched comparison of the oncological feasibility


and survival outcomes for pancreatic adenocarcinoma with robotic
and open pancreatoduodenectomy
Bor‑Uei Shyr1 · Bor‑Shiuan Shyr1 · Shih‑Chin Chen1 · Yi‑Ming Shyr1 · Shin‑E Wang1

Received: 16 December 2020 / Accepted: 5 March 2021


© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021

Abstract
Background  This study is to clarify the feasibility of and justification for robotic pancreaticoduodenectomy (RPD) in patients
with pancreatic adenocarcinoma.
Methods  A 1-to-1 propensity score-matched comparison of RPD and open pancreaticoduodenectomy (OPD) was performed
based on six covariates commonly used to predict the survival outcome for pancreatic adenocarcinoma.
Results  A total of 130 patients were enrolled, with 65 in each study group after propensity score matching. The median
operating time was longer for RPD (8.3 h vs. 7.0 h, P = 0.002). However, RPD was associated with less blood loss, lower
overall surgical complication rate, and lower incidence of delayed gastric emptying. The resection radicality was oncologi-
cally similar between these two groups, but the median lymph node yield was higher for RPD (18 vs. 16, P = 0.038). Before
propensity score matching, the 5-year survival was better in RPD (27.0% vs. 17.6%, P = 0.006). After matching, there was
still a trend towards improved overall survival in the RPD group; however, the difference in 5-year survival between RPD
and OPD was not significant (24.5% vs. 19.7%, P = 0.088).
Conclusion  RPD is not only technically feasible with no increase in surgical risk but also oncologically justifiable without
compromising survival outcome. However, unlike randomized control trials, the limitations in this propensity score-matched
analysis only accounted for 6 observed covariates commonly used to predict the survival outcome in patients with pancreatic
adenocarcinoma, and confounders not included in this study could also affect our results.

Keywords  Oncological · Pancreaticoduodenectomy · Propensity score · Robotic · Survival

Pancreatic adenocarcinoma is a malignancy with a notori- but its application has been limited by its technical complex-
ously dismal outcome, mainly due to its aggressive biology ity and the high level of surgical skill required. Pancreatic
[1, 2]. Surgical resection remains the only hope of cure in reconstruction requires precise placement of suture needles
patients with this type of cancer. Pancreaticoduodenectomy into small lumens followed by intracorporeal knot tying in
requires a high level of technical skill, is time consuming, delicate and often friable pancreatic parenchyma. Mastering
and is traditionally performed via an open approach. Open these complicated operative steps requires advanced lapa-
pancreaticoduodenectomy (OPD) requires a long abdominal roscopic skills and involves a steep learning curve. Since
incision, which is very painful for patients. Minimally inva- the da Vinci Robotic Surgical System (Intuitive Surgical,
sive surgery has become a worldwide trend in many surgi- Inc., Sunnyvale, CA, USA) was introduced into the pancre-
cal fields, including pancreatic surgery [3, 4]. Laparoscopic atic surgery, several limitations related to the laparoscopic
pancreaticoduodenectomy was introduced early in 1994 [5], approach have been overcome. Providing high-quality three-
dimensional vision, an optical magnification of 10–15,
* Shin‑E Wang articulation of instruments with 540 degrees of motion,
sewang0408@gmail.com greater precision with suture targeting, and elimination of
surgeon tremor, the robotic approach makes it possible to
1
Division of General Surgery, Department of Surgery, Taipei perform more delicate and complex procedures, such as
Veterans General Hospital and National Yang Ming Chiao
Tung University, 201 Section 2 Shipai Road, Taipei 112, pancreaticoduodenectomy [6]. However, robotic pancreati-
Taiwan coduodenectomy (RPD) is not widely performed because of

13
Vol.:(0123456789)
Surgical Endoscopy

concerns about the cost of the instruments and the associated of RPD for pancreatic adenocarcinoma by comparing the
technical challenges and surgical risks [3]. surgical outcomes of RPD and OPD.
Although some retrospective studies have claimed that
RPD is technically feasible and safe [7–9], there have been Propensity score‑matching strategy
concerns about the lack of randomized trials demonstrat-
ing the survival and oncologic outcomes when a robotic Propensity score matching was performed to balance poten-
approach is used to work inside the closed and limited space tial confounders between the two types of surgical approach
of the abdominal cavity in patients with pancreatic adeno- (OPD and RPD) and to minimize selection bias when esti-
carcinoma. To minimize the inevitable selection bias that mating causal treatment effects in this retrospective non-
occurs in retrospective research, we performed a propensity randomized study. An individual propensity score was devel-
score-matched analysis to clarify the feasibility of and justi- oped by logistic regression modeling based on six covariates
fication for RPD in pancreatic adenocarcinoma by compar- commonly used to predict survival outcomes in pancreatic
ing the survival and oncologic outcomes of RPD and OPD. adenocarcinoma, including tumor size, lymph node metasta-
sis, disease stage, lymphovascular invasion, perineural inva-
sion, and tumor cell differentiation. The patients with pan-
creatic head adenocarcinoma in the OPD and RPD groups
Materials and methods
were then paired using the exact matching method and in
a 1-to-1 ratio, with the matches starting from cases with
Patients with periampullary lesions who underwent pancrea-
the largest propensity score. A specific caliper width of 0.2
ticoduodenectomy with a modified Blumgart pancreaticoje-
standard deviation of the logit of the estimated propensity
junostomy (PJ) procedure via a robotic approach (n = 322)
score was used [3, 4].
or open approach (n = 230) between 2012 and 2019 were
identified. Among these PD patients, only those with pancre-
Surgical technique
atic head adenocarcinoma were recruited for study. All data
were prospectively collected and entered into a computer
All procedures were performed using the same surgical tech-
database. The study was approved by our institutional review
nique by the same team led by the same surgeon (Shyr YM).
board (IRB-TPEVGH no. 2020-08-009BC). The patient con-
Shyr YM and Wang SE have extensive experience with both
sent was waived in view of the retrospective nature of the
surgical approaches, with more than 500 cases of pancrea-
research and the anonymity of the data.
ticoduodenectomy, and all the procedures were under the
The da Vinci Si or Xi Surgical System was used for RPD.
supervision of these 2 experienced surgeons. There would be
Patient selection for RPD was decided mainly by the avail-
no significant difference in the quality of lymphadenectomy.
ability of a robotic machine and patient preference after
Pancreaticoduodenectomy with a modified Blumgart PJ has
detailed counseling about the innovative nature of RPD and
been described previously in detail [10, 11]. Briefly, after
the possible advantages and disadvantages of this surgical
pancreaticoduodenectomy, the jejunal limb was brought up
approach, which is presently not funded in our country. To
through mesocolon of transverse colon. Blumgart PJ was
avoid technique bias, only pure RPD and OPD cases were
performed using 2 or 3 transpancreatic U-sutures with 3–0
included in the final analysis; cases needing conversion to
monofilament synthetic absorbable sutures made of poly-
open or hybrid surgery were excluded. The study included
dioxanone (PDS™), usually 1 placed cranial and 1 or 2
only those with a pathologic diagnosis of pancreatic ductal
caudal to the pancreatic duct. Each of the U-sutures was
adenocarcinoma, with exclusion of other pancreatic pathol-
placed at a distance of about 5 mm to the next one. These
ogies and periampullary lesions. All perioperative data,
sutures with needles on them were left untied at this time,
including patient demographics, pathologic data, periop-
but instead kept separately and held with clamps or robotic
erative parameters, and oncologic and survival outcomes
instruments until all of the inner duct-to-mucosa interrupted
were compared between RPD and OPD. A propensity score-
sutures were placed and tied. A series of simple interrupted
matched analysis based on six factors predictive of survival
sutures with 4–0 absorbable synthetic monofilament suture
outcomes in pancreatic adenocarcinoma was conducted.
made of polydioxanone (MonoPlus®) were then carefully
and accurately placed for duct-to-mucosa anastomosis. After
Endpoints the duct-to-mucosal interrupted sutures were tied, the outer
anterior horizontal mattress sutures on the jejunum using
The primary aim of the study was to clarify the justifica- previously held U-sutures were completed and tied one by
tion for RPD in patients with pancreatic adenocarcinoma one on the anterior surface of the pancreas. Thus, the pan-
by comparing the survival and oncologic outcomes of RPD creatic remnant was completely covered and compressed
and OPD. The secondary aim was to evaluate the feasibility by jejunal serosa. Pancreatic duct stents were not routinely

13
Surgical Endoscopy

used except for a small pancreatic duct using a short internal or Fisher’s exact test contingency tables. Actuarial survival
stent. Pylorus was preserved whenever possible. The extent rates were estimated using the Kaplan–Meier method. The
of lymphadenectomy included dissections around superior log-rank test was used to compare differences between the
mesenteric vein, the right side of superior mesenteric artery, survival curves. The statistical analyses were carried out
common hepatic artery, and hepatoduodenal ligament. using SPSS for Windows version 21.0 software (IBM Corp.,
The techniques for lymph node and nerve plexus dissec- Armonk, NY, USA). For all analyses, a P value < 0.05 was
tion applied in RPD and OPD were same. There were some considered statistically significant.
differences in the technique used between the OPD and RPD
groups. An energy device with a Harmonic® scalpel was
used to divide the small vessels, and the Hem-o-lok® system Results
(Teleflex Inc., Chelmsford, MA, USA) was selectively used
for the large vessels in RPD, whereas most of the vascular One hundred and thirty patients were enrolled in the study,
pedicles were selectively ligated or cauterized in OPD with- with 65 in each study group after propensity score matching
out use of a Harmonic scalpel. The nasogastric tube placed (Fig. 1). Before matching, there were 190 patients with pan-
during the operation was usually removed on postoperative creatic adenocarcinoma who had undergone RPD (n = 82)
day 1 to 3 if the patient had no significant abdominal dis- or OPD (n = 108) with a modified Blumgart PJ (Table 1);
comfort. Oral intake was allowed as early as postoperative patients with other pancreatic pathology or periampullary
day 2 to 4 when tolerated. Postoperative chemotherapy was lesions who underwent RPD (n = 240) or OPD (n = 122)
our routine adjuvant therapy, but no neoadjuvant chemo- were excluded. Body mass index was higher in the RPD
therapy was administrated for all patients. group than in the OPD group before (P = 0.003) and after
(P = 0.012) propensity score matching. After matching, there
Definitions of surgical risk was no significant between-group difference in any of the
other demographic data, including sex, age, chemotherapy,
Surgical mortality was defined as death within 90 days of vascular resection, tumor size, lymph node metastasis, lym-
surgery, including hospital readmission or during the same phovascular invasion, perineural invasion, tumor cell differ-
admission as the operation. Postoperative pancreatic fistula entiation, disease stage, and borderline resectability.
(POPF) was referred to as grade B and C based on the new The median operating time was longer for RPD than for
definition and grading system devised by the International OPD before (8.5 h vs. 7.3 h, P < 0.001) and after (8.3 h
Study Group for Pancreatic Fistula [12]. Delayed gastric vs. 7.0 h, P = 0.002) propensity score matching. RPD was
emptying (DGE), postpancreatectomy hemorrhage, and associated with less blood loss, a lower overall surgical
chyle leak were identified and classified using the standard- complication rate, and a lower incidence of DGE than
ized criteria proposed by the International Study Group of that of OPD before and after propensity score matching
Pancreatic Surgery [13–15]. Postoperative complications (Table 2). There was no significant difference between
were graded according to the Clavien-Dindo classification RPD and OPD regarding other surgical risks, including
[16]. Resection was stratified into three categories based on surgical mortality, severity of complications based on the
the resection margin status: R0, a resection without both Clavien-Dindo classification, POPF, postpancreatectomy
gross and microscopic evidence of cancer at the resection hemorrhage, chyle leak, bile leak, and wound infection,
margin, under a definition of margin > 0 mm instead of 1 mm
defined by the National Comprehensive Cancer Network;
R1, a resection with grossly negative but microscopically
positive cancer at the resection margin; R2, a resection with
grossly positive cancer at the resection margin [17].

Statistical analysis

All continuous data are presented as the median or as the


mean and standard deviation, and frequencies are presented
as appropriate according to the type of data. Nonparametric
statistical tests were used if the variables did not follow a
normal distribution. Categorical variables are presented as
the number and percentage. The mean values for continuous Fig. 1  Propensity score-matching study focusing on survival out-
comes after robotic pancreaticoduodenectomy (RPD) or open pan-
variables were compared with a two-tailed Student’s t test. creaticoduodenectomy (OPD) in patients with pancreatic head adeno-
Categorical variables were compared using Pearson’s χ2 test carcinoma

13
Surgical Endoscopy

Table 1  Demographics of Before propensity score P value After propensity score P value


patients with pancreatic head matching matching
cancer undergoing robotic and
open pancreaticoduodenectomy RPD OPD RPD OPD
with modified Blumgart
pancreaticojejunostomy Patients, n 82 108 65 65
Gender 0.108 0.218
 Female 38 (46.3%) 61 (56.5%) 31 (47.7%) 39 (60.0%)
Age, year old 0.850 0.871
 Median (range) 66 (29–89) 66 (23–86) 66 (29–89) 66 (23–86)
 Mean ± SD 66 ± 12 66 ± 11 66 ± 13 66 ± 11
BMI 0.003 0.012
 Median (range) 24 (17–33) 22 (14–33) 24 (17–33) 22 (14–29)
 Mean ± SD 24 ± 4 22 ± 3 24 ± 4 22 ± 3
Chemotherapy (+) 65 (79.3%) 89 (82.4%) 0.582 51 (78.5%) 54 (83.1%) 0.657
Vascular resection (+) 9 (11.0%) 25 (23.1%) 0.036 9 (13.8%) 11 (16.9%) 0.809
Tumor size, cm 0.152 0.827
 Median (range) 3 (1.3 –6.0) 3 (1.0–33.5) 3 (1.3–5.7) 3 (1.5–5.0)
 Mean ± SD 3.1 ± 0.9 3.7 ± 3.2 3.1 ± 0.8 3.1 ± 0.7
Lymph node metastasis (+) 49 (58.8%) 77 (71.3%) 0.121 46 (70.8%) 46 (70.8%) 1.000
Lymphovascular invasion (+) 68 (82.9%) 66 (61.1%) 0.001 56 (86.2%) 56 (86.2%) 1.000
Perineural invasion (+) 76 (92.7%) 104 (96.3%) 0.333 62 (95.4%) 62 (95.4%) 1.000
Tumor cell differentiation 0.865 0.979
 Well 1 (1.2%) 2 (1.9%) 1 (1.5%) 1 (1.5%)
 Moderate 61 (74.4%) 77 (71.3%) 48 (73.8%) 49 (75.4%)
 Poorly 20 (24.4%) 29 (26.9%) 16 (24.6%) 15 (23.1%)
Stage 0.951
 I 17 (20.7%) 8 (7.4%) 8 (12.3%) 8 (12.3%)
 II 48 (58.5%) 79 (73.1%) 43 (66.2%) 44 (67.7%)
 III 15 (18.3%) 20 (18.5%) 12 (18.5%) 12 (18.5%)
 IV 2 (2.4%) 1 (0.9%) 2 (3.1%) 1 (1.5%)
Resectability 0.631 1.000
 Resectable 79 (91.5%) 96 (88.9%) 59 (90.8%) 60 (92.3%)
 Borderline resectable 7 (8.5%) 12 (11.1%) 6 (9.2%) 5 (7.7%)

RPD robotic pancreaticoduodenectomy, OPD open pancreaticoduodenectomy, BMI body mass index

before or after propensity score matching. The radicality Discussion


of resection was oncologically similar between RPD and
OPD; however, more lymph nodes were harvested dur- Pancreaticoduodenectomy is among the most technically
ing RPD, with a median lymph node yield of 18 vs. 16 challenging of surgical procedures. It involves not only
before (P = 0.012) and after (P = 0.038) propensity score extensive and protracted dissection but also meticulous
matching. suturing technique for complex reconstructions, which
Before propensity score matching, the 5-year sur- usually makes the inherent technical limitations of the
vival outcome was better in the RPD group than in the laparoscopic approach more evident [4, 18, 19]. Recently,
OPD group (27.0% vs. 17.6%, P = 0.00; Table 3). After robotic surgery has been gaining momentum in many sur-
propensity score matching based on six factors predic- gical fields. Given its innovative nature, there are pres-
tive of survival outcome, there was still a trend towards ently few reports in the literature on RPD when performed
improved overall survival in the RPD group, although the for pancreatic adenocarcinoma [20, 21]. Thus far, RPD
between-group difference was not significant (P = 0.088). has been compared with OPD only in small, retrospec-
The 1-year, 3-year, and 5-year overall survival rates were tive, non-randomized cohort studies of variable quality
79.9%, 42.8%, and 24.5% for RPD and 67.4%, 26.3%, and [20, 22–27]. Moreover, most of the studies have included
19.7% for OPD (Fig. 2).

13
Surgical Endoscopy

Table 2  Surgical and oncological outcomes for pancreatic head cancer after robotic and open pancreaticoduodenectomy with modified Blumgart
pancreaticojejunostomy
Before propensity score matching P value After propensity score matching P value
RPD OPD RPD OPD

Patients, n (%) 82 108 65 65


Operation time, hour  < 0.001 0.020
 Median (range) 8.5 (4.5–14.0) 7.3 (4.2–12.5) 8.3 (4.5–14.0) 7.0 (4.2–12.5)
 Mean ± SD 8.7 ± 2.0 7.6 ± 1.9 8.5 ± 2.0 7.7 ± 2.0
Blood loss, c.c  < 0.001  < 0.001
 Median (range) 200 (0–1000) 485 (100–1900) 200 (0–1000) 500 (120–1900)
 Mean ± SD 243 ± 192 536 ± 350 251 ± 210 562 ± 368
Surgical Mortality (+) 1 (1.2%) 0 0.423 0 0 1.000
Surgical complication (+) 35 (42.7%) 63 (58.3%) 0.023 28 (43.1%) 41 (63.1%) 0.035
Complication classification 0.027 0.079
 Clavien–Dindo I 20 (24.4%) 50 (46.3%) 18 (27.7%) 32 (49.2%)
 Clavien–Dindo II 7 (8.5%) 8 (7.4%) 6 (9.2%) 5 (7.7%)
 Clavien–Dindo III 7 (8.5%) 5 (4.6%) 4 (6.2%) 4 (6.2%)
 Clavien–Dindo IV 0 0 0 0
 Clavien–Dindo V (death) 1 (1.2%) 0 0 0
POPF (grade B and C) 9 (11.0%) 5 (4.6%) 0.159 6 (9.2%) 2 (4.6%) 0.429
Delayed gastric emptying 1 (1.2%) 17 (15.7%) 0.001 1 (1.5%) 11 (16.9%) 0.004
PPH 3 (3.7%) 5 (4.6%) 1.000 3 (4.6%) 4 (6.2%) 1.000
Chyle leakage 19 (23.2%) 27 (25.0%) 0.865 15 (23.1%) 18 (27.7%) 0.687
Bile leakage 0 2 (1.9%) 0.507 0 2 (3.1%) 0.496
Wound infection 2 (2.4%) 6(5.6%) 0.470 2 (3.1%) 4 (6.2%) 0.680
Radicality
 R0 71 (91.5%) 96 (88.9%) 0.776 60 (92.3%) 61 (93.8%) 0.843
 R1 3 (3.7%) 4 (3.7%) 2 (3.1%) 1 (1.5%)
 R2 4 (4.9%) 8 (7.4%) 3 (4.6%) 3 (4.6%)
Lymph node harvested, n 0.012 0.038
Median (range) 18 (7–43) 16 (2–40) 18 (7–43) 16 (2–30)
Mean ± SD 18 ± 7 16 ± 7 18 ± 7 16 ± 6

RPD robotic pancreaticoduodenectomy, OPD open pancreaticoduodenectomy, POPF postoperative pancreatic fistula, PPH postpancreatectomy
hemorrhage, R0 curative resection with a margin > 0 mm, R1 microscopic residual cancer, R2 gross residual cancer

Table 3  Survival outcomes Before propensity score P value After propensity score P value
for pancreatic head cancer matching matching
after robotic and open
pancreaticoduodenectomy RPD OPD RPD OPD

Patients, n (%) 82 108 65 65


Survival time, month 0.006 0.088
 Median 17.7 14.3 18.6 13.2
 Range 1.7–57.4 0.7–93.2 1.7–57.4 0.7–93.2
 Mean ± SD 19.1 ± 12.7 20.0 ± 18.0 20.4 ± 13.5 18.0 ± 15.9
1-year survival 83.7% 68.6% 79.9% 67.4%
3-year survival 47.2% 27.7% 42.8% 26.3%
5-year survival 27.0% 17.6% 24.5% 19.7%

RPD robotic pancreaticoduodenectomy, OPD open pancreaticoduodenectomy

13
Surgical Endoscopy

Fig. 2  Survival curves for patients with pancreatic head adenocarci- dict survival outcomes for pancreatic head adenocarcinoma, includ-
noma after robotic pancreaticoduodenectomy (RPD) and open pan- ing tumor size, lymph node metastasis, disease stage, lymphovascular
creaticoduodenectomy (OPD). A Before propensity score matching invasion, perineural invasion, and tumor cell differentiation
and B after matching based on six covariates commonly used to pre-

both pancreatic head adenocarcinoma and other periamp- patients in RPD group and 24 in OPD. They concluded that
ullary lesions; therefore, the survival outcomes could not RPD and OPD achieved the same rate of R1 resections in
be addressed due to heterogeneous pathology. resectable pancreatic cancer. RPD was also non-inferior to
In view of the lack of high-quality randomized con- OPD with respect to all secondary study endpoints. How-
trolled trials, evidence for the feasibility of and justifica- ever, the limitation of these two studies is small number of
tion for RPD in pancreatic adenocarcinoma is still limited, patients in each arm.
although it was confirmed to be feasible and safe by some While the early reports focused mainly on technical safety
pioneer surgeons [4, 7, 22–24]. A propensity score-matched and feasibility in selected patients, oncologic and survival
study could be an alternative way of balancing OPD and outcomes would be the prime concerns after RPD. Our pre-
RPD for potential confounders and to minimize selection vious propensity score-matched study has already demon-
bias when estimating causal treatment effects in retrospec- strated the technical feasibility of RPD [7]. In that study,
tive non-randomized studies. So far, there were two studies our main purpose was to evaluate POPF and the surgical
using a propensity score-matched analysis for pancreatic risk; therefore, the propensity score-matched analysis was
adenocarcinoma [21, 28]. Baimas-George et al. [28] con- based on four covariates commonly used in the Callery fis-
ducted a propensity-matched study comparing operative tula risk scoring system [29]. To our knowledge, the present
and oncologic outcomes between RPD and OPD in patients report is the first to compare the oncologic feasibility and
with pancreatic adenocarcinoma, with 38 cases in each arm. survival outcomes between RPD and OPD when performed
They demonstrated that RPD not only afforded a similar if exclusively in pancreatic adenocarcinoma by the same sur-
not superior immediate postoperative benefit by decreasing gical team led by the same surgeon using the same surgical
DGE but also, importantly, that it might have an improved technique. Compared with OPD, RPD was associated with
oncologic outcome. They found a trend towards improved similar surgical risks, including surgical mortality, compli-
median overall survival in their RPD group (30.4 months cations, POPF, PPH, chyle leakage, bile leak, and wound
vs. 23.0 months; P = 0.1105) and a longer time to recur- infection, and the radicality of resection was comparable
rence (402 days vs. 284 days; P = 0.7471). They concluded between the two groups before and after propensity score
that while the feasibility and safety of RPD had been dem- matching. Moreover, RPD had the benefits of less blood
onstrated, the impact on oncologic outcome had yet to be loss, a lower DGE rate, and a higher lymph node yield. The
confirmed. The significantly higher lymph node yield and lower DGE rate could be due to less inflammation inside the
decreased inflammatory response demonstrated for robotic abdominal cavity by minimally invasive robotic approach.
surgery might improve overall survival for pancreatic adeno- This study confirms both the technical and oncologic feasi-
carcinoma. Kauffman et al. [21] reported a propensity score- bility of RPD for pancreatic adenocarcinoma.
matched analysis of robotic versus open pancreatoduodenec- To minimize selection bias in this study, which focused
tomy for pancreatic cancer based on margin status, with 20 mainly on survival outcome, the propensity score-matched

13
Surgical Endoscopy

analysis was based on six measurable covariates com- References


monly used to predict the survival outcome in patients
with pancreatic adenocarcinoma, including tumor size, 1. Jang JY, Kang MJ, Heo JS, Choi SH, Choi DW, Park SJ, Han SS,
lymph node metastasis, disease stage, lymphovascular Yoon DS, Yu HC, Kang KJ, Kim SG, Kim SW (2014) A prospec-
tive randomized controlled study comparing outcomes of standard
invasion, perineural invasion, and tumor cell differentia- resection and extended resection, including dissection of the nerve
tion. However, there were other potentially confounders plexus and various lymph nodes, in patients with pancreatic head
that were not taken into account, such as emotional sta- cancer. Ann Surg 259(4):656–664
tus of the patient and family support. The 5-year survival 2. Inoue Y, Saiura A, Yoshioka R, Ono Y, Takahashi M, Arita J,
Takahashi Y, Koga R (2015) Pancreatoduodenectomy with sys-
outcome was better in our RPD group than in our OPD tematic mesopancreas dissection using a supracolic anterior
group before propensity score matching (27.0% vs. 17.6%, artery-first approach. Ann Surg 262(6):1092–1101
P = 0.006), and there was still a trend towards improved 3. McMillan MT, Zureikat AH, Hogg ME, Kowalsky SJ, Zeh HJ,
overall survival following RPD after matching, although Sprys MH, Vollmer CM Jr (2017) A propensity score-matched
analysis of robotic vs open pancreatoduodenectomy on incidence
the between-group difference was not significant (24.5% of pancreatic fistula. JAMA Surg 152(4):327–335
vs. 19.7%, P = 0.088). These findings could reflect the sur- 4. Napoli N, Kauffmann EF, Menonna F, Costa F, Iacopi S, Amorese
gical radicality of RPD and its higher lymph node yield. G, Giorgi S, Baggiani A, Boggi U (2018) Robotic versus open
In conclusion, this study showed that the survival out- pancreatoduodenectomy: a propensity score-matched analysis
based on factors predictive of postoperative pancreatic fistula.
come was better after RPD than after OPD before pro- Surg Endosc 32(3):1234–1247
pensity score matching. There was still a trend towards 5. Gagner M, Pomp A (1994) Laparoscopic pylorus-preserving pan-
improved overall survival for RPD after matching, creatoduodenectomy. Surg Endosc 8(5):408–410
although the difference between RPD and OPD was not 6. Boggi U, Signori S, De Lio N, Perrone VG, Vistoli F, Belluomini
M, Cappelli C, Amorese G, Mosca F (2013) Feasibility of robotic
significant. The oncologic outcomes after RPD were pancreaticoduodenectomy. Br J Surg 100(7):917–925
not inferior to those after OPD, with similar radicality 7. Wang SE, Shyr BU, Chen SC, Shyr YM (2018) Compari-
of resection and a higher lymph node yield. Moreover, son between robotic and open pancreaticoduodenectomy with
RPD had the benefits of less blood loss and a lower DGE modified Blumgart pancreaticojejunostomy: a propensity score-
matched study. Surgery 164(6):1162–1167
rate when compared with OPD. RPD is not only techni- 8. Beane JD, Zenati M, Hamad A, Hogg ME, Zeh HJ 3rd, Zureikat
cally feasible with no increase in surgical risk but also AH (2019) Robotic pancreatoduodenectomy with vascular resec-
oncologically justifiable without compromising survival tion: outcomes and learning curve. Surgery 166(1):8–14
outcome. However, there are some limitations of pro- 9. Zhang YH, Zhang CW, Hu ZM, Hong DF (2016) Pancreatic can-
cer: open or minimally invasive surgery? World J Gastroenterol
pensity score-matched analysis. One of the limitations in 22(32):7301–7310
this study is that propensity score-matched analysis only 10. Wang SE, Chen SC, Shyr BU, Shyr YM (2016) Comparison of
accounted for 6 observed covariates commonly used to Modified Blumgart pancreaticojejunostomy and pancreaticogas-
predict the survival outcome in patients with pancreatic trostomy after pancreaticoduodenectomy. HPB 18(3):229–235
11. Shyr BU, Chen SC, Shyr YM, Wang SE (2020) Surgical, survival,
adenocarcinoma. Unlike randomized control trials, the and oncological outcomes after vascular resection in robotic and
remaining unobserved confounders might still be present, open pancreaticoduodenectomy. Surg Endosc 34(1):377–383
thus, leading to biased results. Confounders not included 12. Bassi C, Marchegiani G, Dervenis C, Sarr M, Abu Hilal M,
in the propensity score-matched analysis could also affect Adham M, Allen P, Andersson R, Asbun HJ, Besselink MG, Con-
lon K, Chiaro MD, Falconi M, Fernandez-Cruz K, Fernandez-Del
our results. Large prospective randomized controlled tri- Castillo C, Fingerhut A, Friess H, Gouma DJ, Hackert T, Izbicki
als with long-term follow-up are needed to confirm the J, Lillemoe KD, Neoptolemos JP, Olah A, Schulick R, Shrikhande
oncologic feasibility and survival outcomes of RPD for SV, Takada T, Takaori K, Traverso W, Vollmer CR, Wolfgang CL,
pancreatic adenocarcinoma. Yeo CJ, Salvia R, Buchler M (2017) The 2016 update of the Inter-
national Study Group (ISGPS) definition and grading of postop-
erative pancreatic fistula: 11 years after. Surgery 161(3):584–591
Acknowledgements  The authors would like to acknowledge the sup- 13. Wente MN, Bassi C, Dervenis C, Fingerhut A, Gouma DJ, Izbicki
port of the Biobank of Taipei Veterans General Hospital. This work JR, Neoptolemos JP, Padbury RT, Sarr MG, Traverso LW, Yeo CJ,
was financially supported by grants from the Taipei Veterans General Büchler MW (2007) Delayed gastric emptying (DGE) after pan-
Hospital (V110B-023, V110C-010 and V110C-011), the Ministry of creatic surgery: a suggested definition by the International Study
Science and Technology (MOST 108-2314-B-075-051-MY3), and the Group of Pancreatic Surgery (ISGPS). Surgery 142(5):761–768
Ministry of Health and Welfare (MOHW107-TDU-B-212-114026A). 14. Wente MN, Veit JA, Bassi C, Dervenis C, Fingerhut A, Gouma
DJ, Izbicki JR, Neoptolemos JP, Padbury RT, Sarr MG, Yeo CJ,
Declarations  Büchler MW (2007) Postpancreatectomy hemorrhage (PPH): an
International Study Group of Pancreatic Surgery (ISGPS) defini-
Disclosures  Dr. Bor-Uei Shyr, Bor-Shiuan Shyr, Shih-Chin Chen, Yi- tion. Surgery 142(1):20–25
Ming Shyr, and Shin-E Wang have no conflicts of interest or financial 15. Besselink MG, van Rijssen LB, Bassi C, Dervenis C, Montorsi
ties to disclose. M, Adham M, Asbun HJ, Bockhorn M, Strobel O, Buchler MW,
Busch OR, Charnley RM, Conlon KC, Fernández-Cruz L, Finger-
hut A, Friess H, Izbicki KD, Lillemoe JR, Neoptolemos JP, Sarr

13
Surgical Endoscopy

MG, Shrikhande R, Sitarz SV, Vollmer CM, Yeo CJ, Hartwig 23. Napoli N, Kauffmann EF, Menonna F, Perrone VG, Brozzetti S,
W, Wolfgang CL, Gouma DJ (2017) Definition and classification Boggi U (2016) Indications, technique, and results of robotic pan-
of chyle leak after pancreatic operation: a consensus statement creatoduodenectomy. Updates Surg 68(3):295–305
by the International Study Group on Pancreatic Surgery. Surgery 24. Zureikat AH, Moser AJ, Boone BA, Bartlett DL, Zenati M, Zeh
161(2):365–372 HJ 3rd (2013) 250 robotic pancreatic resections: safety and feasi-
16. Clavien PA, Barkun J, de Oliveira ML, Vauthey JN, Dindo D, bility. Ann Surg 58(4):554–559 (discussion 559-562)
Schulick RD, de Santibanes E, Pekolj J, Slankamenac K, Bassi 25. Zureikat AH, Postlewait LM, Liu Y, Gillespie TW, Weber SM,
C, Graf R, Vonlanthen R, Padbury R, Cameron JL, Makuuchi Abbott DE, Ahmad SA, Maithel SK, Hogg ME, Zenati M, Cho
M (2009) The Clavien-Dindo classification of surgical complica- CS, Salem A, Xia B, Steve J, Nguyen TK, Keshava HB, Cha-
tions: five-year experience. Ann Surg 250(2):187–196 likonda S, Walsh RM, Talamonti MS, Stocker SJ, Bentrem DJ,
17. Evans DB, Farnell MB, Lillemoe KD, Vollmer C Jr, Strasberg Lumpkin S, Kim HJ, Zeh HJ 3rd, Kooby DA (2016) A multi-
SM, Schulick RD (2009) Surgical treatment of resectable and bor- institutional comparison of perioperative outcomes of robotic and
derline resectable pancreas cancer: expert consensus statement. open pancreaticoduodenectomy. Ann Surg 264(4):640–649
Ann Surg Oncol 16(7):1736–1744 26. Peng L, Lin S, Li Y, Xiao W (2017) Systematic review and meta-
18. Gagner M, Palermo M (2009) Laparoscopic Whipple proce- analysis of robotic versus open pancreaticoduodenectomy. Surg
dure: review of the literature. J Hepatobiliary Pancreat Surg Endosc 31(8):3085–3097
16(6):726–730 27. Esposito A, Balduzzi A, De Pastena M, Fontana M, Casetti
19. Boggi U, Amorese G, Vistoli F, Caniglia F, De Lio N, Perrone V, L, Ramera M, Bassi C, Salvia R (2019) Minimally invasive
Barbarello L, Belluomini M, Signori S, Mosca F (2015) Laparo- surgery for pancreatic cancer. Expert Rev Anticancer Ther
scopic pancreaticoduodenectomy: a systematic literature review. 19(11):947–958
Surg Endosc 29(1):9–23 28. Baimas-George M, Watson M, Murphy KJ, Iannitti D, Baker E,
20. Nassour I, Tohme S, Hoehn R, Adam MA, Zureikat AH, Alessan- Ocuin L, Vrochides D, Martinie JB (2020) Robotic pancreati-
dro P (2020) Safety and oncologic efficacy of robotic compared coduodenectomy may offer improved oncologic outcomes over
to open pancreaticoduodenectomy after neoadjuvant chemother- open surgery: a propensity-matched single-institution study. Surg
apy for pancreatic cancer. Surg Endosc. https://​doi.​org/​10.​1007/​ Endosc 34(8):3644–3649
s00464-​020-​07638-w 29. Callery MP, Pratt WB, Kent TS, Chaikof EL, Vollmer CM Jr
21. Kauffmann EF, Napoli N, Menonna F, Iacopi S, Lombardo C, (2013) A prospectively validated clinical risk score accurately
Bernardini J, Amorese G, Cacciato Insilla A, Funel N, Campani predicts pancreatic fistula after pancreatoduodenectomy. J Am
D, Cappelli C, Caramella D, Boggi U (2019) A propensity score- Coll Surg 216(1):1–14
matched analysis of robotic versus open pancreatoduodenec-
tomy for pancreatic cancer based on margin status. Surg Endosc Publisher’s Note Springer Nature remains neutral with regard to
33(1):234–242 jurisdictional claims in published maps and institutional affiliations.
22. Polanco PM, Zenati MS, Hogg ME, Shakir M, Boone BA, Bart-
lett DL, Zeh HJ, Zureikat AH (2016) An analysis of risk factors
for pancreatic fistula after robotic pancreaticoduodenectomy:
outcomes from a consecutive series of standardized pancreatic
reconstructions. Surg Endosc 30(4):1523–1529

13

You might also like