Pancreatoduodenectomy with venous vascular resection (PDVR) is a surgical procedure performed for the margin-negative resection of pancreatic and periampullary tumors that involve the surrounding superior mesenteric and portal vein (PV) system. The extent of combined venous vascular resection depends on the size and location of the tumor, and the degree of vascular involvement. After removal of the tumor and the involved structures, vascular reconstruction should be performed for adequate blood supply to the liver to maintain proper circulation.

Current guidelines [1] recommend that patients with borderline resectable pancreatic cancers should undergo neoadjuvant therapy before resection, because they might have a higher probability of incomplete tumor resection [2-5]. Considering that margin negative pancreatectomy is essential for the long-term survival of pancreatic cancer, combined venous vascular resection is thought to increase the chance of curative resection in the case of suspicious venous vascular involvement on preoperative imaging, or during surgical intervention. PDVR is widely accepted for R0 resection; however, whether this approach is a safe and feasible option remains controversial, due to possible higher mortality and complication rates.

With the advance of minimally invasive surgical experiences, clinical application of laparoscopic and robotic pancreatoduodenectomy (MI-PD) is increasing, showing that MI-PD is feasible, safe, and effective in treating periampullary cancers, even for pancreatic head cancer. In particular, regarding laparoscopic PDVR, Croome et al. [6] first reported the technical feasibility of major venous resection during laparoscopic PD in 2011. Subsequently, they successfully demonstrated that not only is PD with major vascular resection feasible and safe, it can achieve similar short-term and long-term oncologic results to patients with open PD with major vascular resection [6]. However, since then, most research papers related to this topic have been only on technical skills in a selected case. Therefore, minimally invasive pancreatoduodenectomy with venous vascular resection (MI-PDVR) still needs more investigation of its safety and its long-term oncologic effectiveness in its clinical application.

In this study, we evaluated the short-term and long-term outcome of MI-PDVR compared with open pancreatoduodenectomy with venous vascular resection (O-PDVR), to discuss the indication, safety, and oncologic efficacy of MI-PDVR.

Study population and data collection

This retrospective comparative analysis was conducted on a cohort of patients who underwent pancreaticoduodenectomy for a period spanning from January 1, 2016, to December 31, 2023, at Severance Hospital, Seoul, Korea. During this study period, a total of 1,418 patients underwent pancreaticoduodenectomy, performed by a team of 9 surgeons. Among these patients, 139 cases (9.8% of the total) included major vascular resection as a part of their surgical procedure. The indication for vascular resection in MI-PDVR requires that patients must be in sufficient general condition to endure long-term pneumoperitoneum. Ideally, the tumor should have minimal vascular involvement (less than 2 cm in length), and there should be no arterial invasion. Patients who underwent surgery for indications other than cancer or neoplasm, and those who had vascular resections other than those involving the PV and superior mesenteric vein (SMV), were excluded from the analysis. This study was approved by the Institutional Review Board of the Yonsei University College of Medicine (registration number: 4-2024-0892). The requirement to obtain informed consent was waived.

Subsequently, the study cohort was further categorized into patients who underwent PDVR specifically due to periampullary cancer. Among these patients, there were 72 cases of O-PDVR, and 52 cases of MI-PDVR, which included both laparoscopic and robot-assisted procedures. Additionally, within the MI-PDVR group, 7 cases required conversion to O-PDVR, and these were subsequently classified under the O-PDVR category for analysis (Fig. 1).

Fig 1. Study design. PD, pancreatoduodenectomy; PV, portal vein; SMV, superior mesenteric vein; O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection.

All patient data were obtained through electronic medical records (EMR). To compare perioperative variables between two groups, we investigated the age, sex, American Society of Anesthesiologist (ASA) score (divided at a cutoff of 3 points), preoperative body mass index (BMI), preoperative carcinoembryonic antigen (CEA), preoperative carbohydrate antigen 19-9 (CA19-9), presence of hypertension (HTN), presence of diabetes mellitus (DM), and pre-biliary drainage. For analysis of the perioperative variables, we examined the operation duration, estimated blood loss, intraoperative transfusion, length of hospital stay, and postoperative complications. Complications were categorized using the Clavien–Dindo classification, with grade III as the cutoff point.

To compare oncologic outcomes, the following factors were investigated exclusively in pancreatic cancer patients. We examined whether neoadjuvant chemotherapy was administered, whether adjuvant chemotherapy was given, tumor size, number of retrieved lymph nodes, presence of positive lymph nodes, histological cell differentiation, tumor-node-metastasis (TNM) stage, lymphovascular invasion, perineural invasion, resection margin status, and recurrence. Disease-free survival (DFS) and overall survival (OS) were also compared between the two groups.

Advanced techniques in vascular reconstruction and criteria for open conversion

The procedure for performing vascular resection and reconstruction due to vascular invasion at the SMV-splenic vein (SV)-PV confluence involves several steps. First, to establish inflow control, the PV, SMV, SV, left gastric vein, or the inferior mesenteric vein (if it drains directly into the SMV) are completely isolated, and clamped using laparoscopic bulldog forceps. For a tangential wedge resection of the SMV-SV-PV confluence, a 5-0 Prolene (PROLENE Polypropylene, Ethicon) running suture is used. In cases of segmental resection of the SMV-SV-PV confluence, a 5-0 Prolene suture is also employed, with anterior and posterior vessels sutured with a running stitch, ensuring alignment with the axis of the SMV-SV-PV confluence to prevent twisting. Finally, the restoration of flow is confirmed using laparoscopic ultrasound following vascular reconstruction (Fig. 2).

Fig 2. Combined venous vascular resection during MI-PD. (A) Primary repair of resected venous system. (B) End-to-end anastomosis of resected SMV-SV-PV confluence. MI-PD, minimally invasive pancreatoduodenectomy; SMV, superior mesenteric vein; SV, splenic vein; PV, portal vein.

The criteria for open conversion are as follows: typically, the extent of vascular involvement was greater than preoperative imaging suggested, or the resection faced challenges such as bleeding or achieving R0 margins. Additionally, severe pancreatitis could halt dissection progress for over an hour. In cases where unexpectedly long-segment resections were required, or when arterial involvement was suspected during surgery, conversion during the laparoscopic procedure was necessary.

Statistical analysis

Demographic characteristics, perioperative parameters, pathological findings, and oncological results were subjected to comparative analysis between the MI-PDVR and O-PDVR cohorts. Continuous variables were summarized using either the mean and standard deviation or the median and range, depending on the data distribution, and compared using Student’s t-test. Categorical variables underwent comparison through suitable statistical methods, such as the chi-square test, Fisher’s exact test, or linear-by-linear association test. All statistical tests were conducted as two-tailed tests, with statistical significance defined at a p-value ≤ 0.05. Survival curves were constructed employing the Kaplan–Meier approach, and the comparison of survival outcomes between the MI-PDVR and O-PDVR groups was executed using the log-rank test.

Patient demographics

During the study period, 79 patients underwent O-PDVR, and 45 received MI-PDVR. Table 1 summarizes the basic patient demographics. Preoperative variables, including age, sex, ASA score, preoperative BMI, preoperative CEA, preoperative CA19-9, HTN status, DM status, and the implementation of preoperative biliary drainage, were compared, and showed no significant differences between the two groups.

Table 1 . Patient demographics

	O-PDVR (n = 79)	MI-PDVR (n = 45)	p-value
Preoperative variables
Age (yr)	63.51 ± 8.7	64.27 ± 8.9	0.645
Sex (male:female)	48:31	19:26	0.061
ASA score
< 3	33 (41.8)	13 (28.9)	0.179
3 ≤	46 (58.2)	32 (71.1)
Preoperative BMI (kg/m2)	23.56 ± 3.6	22.50 ± 3.0	0.079
Preoperative CEA (ng/mL)	4.34 ± 5.2	6.13 ± 18.6	0.531
Preoperative CA19-9 (U/mL)	340.50 ± 917.6	197.28 ± 487.2	0.271
HTN
No	44 (55.7)	30 (66.7)	0.258
Yes	35 (44.3)	15 (33.3)
DM
No	41 (51.9)	19 (42.2)	0.352
Yes	38 (48.1)	26 (57.8)
Pre-biliary drainage
No	32 (40.5)	15 (33.3)	0.449
Yes	47 (59.5)	30 (66.7)
Disease entity
Pancreatic cancer	60 (75.9)	34 (75.6)	> 0.99
Distal CBD cancer	17 (21.5)	10 (22.2)
Ampulla of Vater cancer	1 (1.3)	1 (2.2)
Duodenal cancer	1 (1.3)	0 (0)
Types of vascular resection
PV	36 (45.6)	27 (60.0)	0.286
SMV	21 (26.6)	10 (22.2)
PV-SV-SMV confluence	22 (27.8)	8 (17.8)
Resection type
Segmental resection	58 (73.4)	10 (22.2)	< 0.01
Tangential resection	21 (26.6)	35 (77.8)

Values are presented as mean ± standard deviation or number (%).

O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection; ASA, American Society of Anesthesiologists; BMI, body mass index; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; HTN, hypertension; DM, diabetes mellitus; CBD, common bile duct; PV, portal vein; SMV, superior mesenteric vein; SV, splenic vein.

When comparing the disease entities between the two groups, there were no differences in the types of periampullary cancer. In both groups, pancreatic cancer was the most prevalent, followed by distal common bile duct cancer. Although there was no difference in the resected vascular types between the two groups, there was a notable distinction in the distribution of segmental and tangential resections. Specifically, in the O-PDVR group, segmental resection accounted for 73.4%, while tangential resection constituted 26.6%. In the MI-PDVR group, segmental resection was 22.2%, and tangential resection was 77.8% (p < 0.01).

Perioperative outcomes: O-PDVR vs. MI-PDVR

Table 2 shows a comparison of perioperative outcomes between the O-PDVR and MI-PDVR groups. The median operation time for MI-PDVR was significantly shorter at 452.69 ± 149.7 minutes, compared to the O-PDVR group’s (543.91 ± 194.9 minutes; p = 0.004). In terms of estimated blood loss, the MI-PDVR group had a median of 410.44 mL, significantly less than the O-PDVR group’s 747.59 mL (p < 0.01). Consequently, intraoperative transfusions were also fewer in the MI-PDVR group with 2 cases, compared to the O-PDVR group’s 18 cases (p = 0.01). Regarding hospital stay, the MI-PDVR group showed a shorter median duration of 18.16 days, compared to 23.91 days in the O-PDVR group (p = 0.008). When investigating complications until the postoperative discharge day, based on the Clavien–Dindo classification with a cutoff value set at class III, no significant differences were observed between the two groups (p = 0.809).

Table 2 . Perioperative variables

	O-PDVR (n = 79)	MI-PDVR (n = 45)	p-value
Operation time (min)	543.91 ± 194.9	452.69 ± 149.7	0.004
Estimated blood loss (mL)	747.59 ± 638.9	410.44 ± 404.4	< 0.01
Intraoperative transfusion	18 (22.8)	2 (4.4)	0.01
Hospital stay (day)	23.91 ± 12.2	18.16 ± 10.9	0.008
Complication
< III	65 (82.3)	38 (84.4)	0.809
III ≤	14 (17.7)	7 (15.6)

Values are presented as mean ± standard deviation or number (%).

O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection.

Long-term oncological outcomes: O-PDVR vs. MI-PDVR

Fig. 3 illustrates the Kaplan–Meier curves for OS and DFS between the O-PDVR and MI-PDVR groups. There were no statistical differences in OS (49.92 months, 95% confidence interval [CI]: 40.97–58.87 vs. 51.55 months, 95% CI: 35.95–67.14; p = 0.340) (Fig. 3A) and DFS (median 38.77 months, 95% CI: 29.80–47.75 vs. median 35.06 months, 95% CI: 21.47–48.65; p = 0.585) (Fig. 3B) between the two groups.

Fig 3. Overall survival (OS) and disease-free survival (DFS) of O-PDVR and MI-PDVR. There were no statistical differences in (A) OS (49.92 months, 95% CI: 40.97–58.87 vs. 51.55 months, 95% CI: 35.95–67.14; p = 0.340) and (B) DFS (median 38.77 months, 95% CI: 29.80–47.75 vs. median 35.06 months, 95% CI: 21.47–48.65; p = 0.585) between the two groups. O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection; CI, confidence interval; MIS, minimally invasive surgery.

Subgroup analysis focusing on pancreatic ductal adenocarcinoma: short-term oncologic outcomes

To facilitate clear comparison of oncologic and pathologic outcomes, we conducted a comparative analysis specifically within the subset of patients diagnosed with pancreatic ductal adenocarcinoma (PDAC) among the periampullary cancer cases. Among a total 124 patients, of 94 (75.8%) PDAC patients of the PDVR cohort, 60 patients belonged to the O-PDVR group, while 34 belonged to the MI-PDVR group (Table 3).

Table 3 . Pathologic outcomes in pancreatic cancer patients

	O-PDVR (n = 60)	MI-PDVR (n = 34)	p-value
Neoadjuvant chemotherapy
No	24 (40.0)	7 (20.6)	0.054
Yes	36 (60.0)	27 (79.4)
Tumor size (cm)	2.5 ± 0.9	2.7 ± 1.0	0.285
Retrieved lymph node	22.1 ± 10.8	19.7 ± 10.3	0.290
Positive lymph node	1.9 ± 2.6	1.1 ± 1.7	0.064
Differentiation
Well	7 (11.7)	5 (14.7)	0.511
Moderate	46 (76.7)	22 (70.6)
Poor	7 (11.7)	5 (14.7)
N stage
N0	25 (41.7)	19 (55.9)	0.365
N1	23 (38.3)	11 (32.4)
N2	12 (20.0)	4 (11.8)
TNM stage
IA	9 (15.0)	12 (35.3)	0.227
IB	12 (20.0)	5 (14.7)
IIA	4 (6.7)	2 (5.9)
IIB	22 (36.7)	11 (32.4)
III	13 (21.7)	4 (11.8)
IV	0 (0)	0 (0)
Lymphovascular invasion
No	35 (58.3)	24 (70.6)	0.238
Yes	25 (41.7)	10 (29.4)
Perineural invasion
No	11 (18.3)	7 (20.6)	0.789
Yes	49 (81.7)	27 (79.4)
Resection margin
Negative (R0)	44 (73.3)	21 (61.8)	0.243
Positive (R1, R2)	16 (26.7)	13 (38.2)
Adjuvant chemotherapy
No	8 (13.3)	6 (17.6)	0.296
Yes	52 (86.7)	26 (76.5)
Recurrence
No	27 (45.0)	19 (55.9)	0.311
Yes	33 (55.0)	15 (44.1)

Values are presented as number (%) or mean ± standard deviation.

O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection; TNM, tumor-node-metastasis.

No significant differences were observed between the two groups concerning neoadjuvant chemotherapy. Additionally, there were no discernible distinctions in tumor size, retrieved lymph nodes, positive lymph nodes, or differentiation. Consequently, no statistically differences were noted in the N or TNM stage. Moreover, there were no disparities in lymphovascular invasion, perineural invasion, or resection margin, leading to a lack of significant differences in adjuvant chemotherapy and recurrence rates.

Subgroup analysis focusing on pancreatic ductal adenocarcinoma: long-term oncologic outcomes

Fig. 4 depicts Kaplan–Meier curves illustrating the OS and DFS for PDAC patients, similar to the previously presented figures. There were no statistical differences in OS (median 48.48 months, 95% CI: 38.16–58.59 vs. median 40.86 months, 95% CI: 34.45–47.27; p = 0.270) (Fig. 4A) and DFS (median 34.35 months, 95% CI: 25.44–43.27 vs. median 24.42 months, 95% CI: 17.03–31.85; p = 0.740) (Fig. 4B) between the two groups.

Fig 4. Overall survival (OS) rate and disease-free survival (DFS) of O-PDVR and MI-PDVR in PDAC patients. There were no statistical differences in (A) OS (median 48.48 months, 95% CI: 38.16–58.59 vs. median 40.86 months, 95% CI: 34.45–47.27; p = 0.270) and (B) DFS (median 34.35 months, 95% CI: 25.44–43.27 vs. median 24.42 months, 95% CI: 17.03–31.85; p = 0.740) between the two groups. O-PDVR, open pancreatoduodenectomy with venous vascular resection; MI-PDVR, minimally invasive pancreatoduodenectomy with venous vascular resection; PDAC, pancreatic ductal adenocarcinoma; CI, confidence interval.

Recent evidence has shown oncological benefits from radical pancreatectomy performed after preoperative chemotherapy in cases of advanced pancreatic cancer [3,4]. Accordingly, neoadjuvant chemotherapy is recommended first, rather than upfront surgery, as a treatment guideline for borderline resectable pancreatic cancer [1]. In the treatment of pancreatic cancer, achieving a margin-negative resection is regarded as the most effective standalone therapy, and is essential to ensure long-term patient survival. This is particularly important in cases of borderline resectable pancreatic head cancer, which has a high probability of invading nearby vascular structures, such as the SMV-PV system. Therefore, PDVR can lead to potential R0 resection. This approach is complex and challenging, but several promising short-term and long-term oncologic outcomes of PDVR are reported [7-9].

Regarding venous resection for pancreatic head cancer, recent systemic review and meta-analysis [7] based on 32 studies describing 2,216 patients with PDVR showed PDVR to be a safe and feasible option in patients with pancreatic cancer and suspicion of venous involvement by demonstrating similar 90-day mortality, and OS to patient without PDVR. However, PDVR was found to be associated with more frequent postpancreatectomy hemorrhage, 30-day mortality, and advanced stages of cancer.

Another recent meta-analysis [10] based on 36 retrospective observational studies, including 2,986 patients with periampullary cancers, also demonstrated that PDVR was related to higher postoperative complications, mortality, blood loss, positive margin, and operation time, compared with PD. Considering publication bias, the actual risk of PDVR is thought to be higher when determining indication of surgery. Therefore, PDVR is a complex and technically challenging procedure that requires a skilled surgical team experienced in specialized centers, with expertise in pancreatic surgery.

In recent years, many retrospective studies [11-13] and some prospective randomized control studies have been published on the safety and benefit of MI-PD [14-18]. In addition, clinical experiences with MI-PDVR for margin-negative resection in advanced periampullary cancers, including pancreatic cancer, have been reported, suggesting this approach is feasible, safe, and effective. However, since most of these reports focus on technical feasibility and are based on a limited number of cases, further research is necessary to evaluate the technical feasibility, safety, and oncological effectiveness of MI-PDVR. This is particularly important given the increasing clinical practice of MI-PD in the treatment of periampullary cancers, including pancreatic cancer.

To date, there is not much concrete evidence on the safety and long-term oncologic effects of MI-PDVR, and it is not yet a common procedure in daily clinical oncology. Till now, there are a few literatures reporting more than 10 cases of MI-PDVR for periampullary cancers [6,11,19-25]. Croome et al. [6] reported 31 cases of MI-PDVR, showing very favorable short-term and long-term oncologic outcomes. Ouyang et al. [22] recently published a Chinese multicenter collaborating study based on 142 patients for pancreatic head cancer, which had the largest study population regarding this issue, demonstrating quite acceptable perioperative outcomes, and long-term survival outcome as well. However, the number of patients, key perioperative outcomes, and long-term oncologic results reported in various studies are too limited to draw definitive conclusions regarding the safety and oncologic efficacy of this technique.

In addition, the unintended intraoperative conversion during MI-PD has a negative impact on the prognosis of patients [26-28], so special attention should be paid to the selection of appropriate indications. Most of the reasons for conversion were cases that were difficult to dissect, or with bleeding during the tumor blood vessels during the dissection process. Therefore, it is inevitable that there are hurdles that must be overcome when attempting MI-PDVR. Considering the significant learning curves to safely implement MI-PD [29,30] it is doubtful that MI-PDVR can generally become a standard approach in the future.

It is believed that the treatment of cancer should focus on whether a radical approach is possible and can be safely recovered, rather than a minimally invasive approach, or not. Given that so far, no dramatic clinical benefit of MI-PD has been observed in previous prospective randomized control studies [14-18], the debate is likely to continue for some time, until there is more strong evidences to support MI-PD over open pancreatoduodenectomy (O-PD). Therefore, at present, depending on the patient’s condition, surgical extent for margin negative resection, patient safety, and the degree of surgeons’ technique, it seems that O-PD or MI-PD should be carefully selected, especially if combined venous vascular resection is highly expected.

Clinical effort to achieve a high level of evidence will continue, but randomized controlled trials (RCTs) comparing O-PDVR with MI-PDVR in periampullary cancers (especially, pancreatic head cancer) are considered to be very difficult. Rather than that approach, it is necessary to perform multicenter collaborative studies regarding the safety and feasibility of MI-PDVR based on more clinical experience. At the same time, when the application of neoadjuvant treatment in pancreatic cancer is increasing, efforts to consider and try MI-PD are also expected to increase, so a practical surgical education system [31-33] needs to be well established, so that surgeons could more easily overcome the learning curve for the successful clinical settlement of MI-PDVR. Additionally, further research is needed to understand who will be benefit from MI-PDVR and potential advantage of minimally invasive approach in treating pancreatic cancer.

This study has several limitations. It relies on retrospective data due to the inherent challenge of selecting patients for vascular resection using MI-PDVR, which involves ethical considerations. As noted in the manuscript, the data presented exhibits selection bias. Ideally, a RCT would be needed to overcome this limitation; however, conducting such a study poses practical challenges. Despite these limitations, the research suggests that it is possible to perform vascular resection and achieve R0 margins using MI-PDVR, thus offering the benefits of MI-PD over O-PD.

In summary, MI-PDVR might be safe and provide similar oncologic impact to an open approach when performed in a patient who has been well selected by an experienced surgeon. However, since this conclusion is based on limited data, it is believed that further study is mandatory based on appropriate selection criteria, improved surgical techniques, and long-term follow-up in near future.

Some part of the contents are presented in HBP Surgery Week 2024 (March 21th, Room B, WalkerHill, Seoul, Korea).

None.

Chang Moo Kang is the Executive Editor of the journal but was not involved in the review process of this manuscript. Otherwise, there is no conflict of interest to declare.

Conceptualization: CMK. Data curation: SYR, SSH, SHK, HKH. Methodology: DHS, MC, CMK. Visualization: DHS, MC. Writing - original draft: DHS. Writing - review & editing: MC.