Pancreatoduodenectomy (PD) is currently considered the standard surgical approach for cancers of the head of the pancreas, duodenum, and distal bile duct. It was originally performed and described by Alessandro Codivilla from the Hospital of Imola (Italy) in 1898 [1], but was named after the American surgeon Allen Oldfather Whipple who proposed the definitive version of this operation in 1935 [2]. The Whipple procedure entails resection of the pancreatic head along with the duodenum, gastric antrum, gallbladder, and common bile duct. In 1978, Traverso and Longmire [3] demonstrated that the pylorus can be spared if there is no infiltration of the stomach or the perigastric lymph nodes, to reduce the side effects of gastric resection and maintain a normal gastric motility. However, the real advantages of the Traverso-Longmire procedure over the Whipple operation are yet to be demonstrated. In 2010, Giulianotti et al. [4] published the first significant series of PD performed by robotic surgery with positive results. In recent years, the attention of pancreatic surgeons has been drawn at the few critical–and still unresolved–points of PD, either by open or mini-invasive approach: (1) approach to the superior mesenteric artery (SMA) and the mesopancreas to verify the resectability and obtain an R0 resection in case of tumors of the uncinate process, (2) preparation of the superior mesenteric vein (SMV) while reducing the risk of the lesion and subsequent bleeding of the gastrocolic trunk on the anterior aspect of the head of the pancreas, and (3) risk of anastomotic complications and mainly of postoperative pancreatic fistula (POPF).

The approach to the superior mesenteric pedicle is a crucial step in PD. Classically, the SMV is approached after the incision of the pancreatic fascia at the inferior border of the pancreatic neck. This step can be done early, just after the division of the gastrocolic ligament, to rule out an eventual infiltration of the portal vein. However, venous infiltration is no longer considered a contraindication to radical surgery as portal resection and repair are routinely performed in dedicated pancreatic centers. Therefore, many surgeons prefer to postpone this step until after the Kocher maneuver and sometimes after the mobilization of the hepatic flexure of the colon, the dissection of the second and third parts of the duodenum (D2 and D3), the mobilization of the Treitz’s loop and the transection of the first jejunal loop (J1). Sometimes, this step can be particularly difficult due to peripancreatic adhesions (in patients with a history of pancreatitis) and hemorrhage due to a short and fragile gastro-colic trunk (GCT). To reduce these risks, this can be done by dissecting the origin of the superior mesenteric pedicle from the inframesocolic space, preparing the Fredet’s fascia, and the inferior knee of the duodenum through the “duodenal window” [5]. We hereby present our “duodenal-window first pancreatoduodenectomy” technique.

The dissection phase of PD starts from the inframesocolic area. The transverse colon and mesocolon are lifted upwards and ventrally to expose its root. The duodenal window (DW) and the Treitz’s foramen (TF) are easily identified (Fig. 1A). The DW is a small area at the root of the proximal transverse mesocolon where a thin layer of visceral peritoneum covers the inferior knee of the duodenum. Usually, only a minimal amount of fat, if any, can be found between the peritoneum and the duodenum and it is usually easily identified. The margins of the DW are the right wall of the SMV medially, the ileocolic pedicle inferiorly, the right colic pedicle superiorly, and the marginal artery laterally. Opening the peritoneum of the DW (Fig. 1B) gives access to the Fredet’s fascia covering D2 and D3. The anterior aspect of the duodenopancreatic block can be bluntly dissected free from the root of the mesocolon. The dissecting finger (or instrument) gently develops an avascular plane between the SMV and the anterior aspect of D3 and D4, almost reaching the TF and the duodeno-jejunal junction (DJJ) from right to left. During this maneuver, the uncinate process can be easily palpated and possibly detached from the posterior aspect of the SMV. Subsequently, the lateral ligaments of distal D2, inferior knee, and D3 are divided and the duodenopancreatic block is gently detached from the aortocaval plane, again addressing the dissection towards D3 and D4. At this stage, the initial Kocher maneuver can be extended to D2, if possible, but the difficult access to the superior knee usually prevents a complete mobilization of the duodenum from the inframesocolic access. However, D3 and D4 can also be mobilized posteriorly up to J1. The dissection then moves to the Treitz’s area. The Treitz’s ligament is divided and the DJJ is mobilized. The peritoneal folds and the loose ligaments fixing the DJJ anteriorly and posteriorly are divided, and a blunt dissection is carried out on the anterior and posterior aspects of J1 and D4 to reach the previous plane of dissections. The posterior dissection is carried out as far as possible from left to right, in particular on the posterior plan, to reach the lateral aspect of D2. This step resembles a “retrograde Kocher maneuver,” as at the end the duodenum should be almost completely detached from the posterior plane. An eventual SMV and SMA infiltration from the uncinate process can be ruled out at this stage by bimanual palpation with the index fingers embracing the superior mesenteric pedicle between the vascular pedicle and the anterior aspect of D3-D4 (Fig. 1C). At this stage, the mesopancreas can be easily palpated with the tip of the fingers between the uncinate process and the posterior aspect of the SMA. This whole maneuver is carried out dorsally to the superior mesenteric pedicle, without any risk of bleeding from the GCT that is not stretched. A blunt and prudent dissection at the inferior knee and D3 should avoid the risk of bleeding from the inferior pancreatoduodenal artery (IPDA) and the first jejunal artery (J1A). These two small vessels should be easily seen and controlled, the first from the DW and the second from the TF. This should allow an easy mobilization and division of D1, should the surgeon choose to do this at this stage. Moreover, the DW-first dissection reduces the traction on the root of the transverse mesocolon and therefore the risk of injuring the Henle’s trunk.

Fig 1. Preparation of the duodenal window in open pancreatoduodenectomy. (A) The duodenal window can be found at the root of the transverse mesocolon, at the right side of the superior mesenteric pedicle, and mirrors the site of the Treitz’s foramen that is located on the left side of the superior mesenteric pedicle. (B) The duodenal window can be easily opened to obtain access to the Fredet’s fascia and distal duodenum. (C) After dissection of the Treitz’s foramen, the virtual spaces dorsal and ventral to D3 and D4 are easily developed from right to left and from left to right.

At the end of this step, the surgeon should achieve the following: (1) the tip of the right index finger touching the tip of the left index finger on the anterior aspect of D3, (2) the tips of the two fingers touching each other on the posterior aspect of D3, and (3) both index fingers swiping the posterior aspect of the duodenopancreatic block.

During robotic PD, the fingers can be easily replaced by the robotic instruments (Fig. 2). The 3D robotic vision allows a better vision of the DW and the TF with respect to the open approach, due to the specific imaging features of the surgical robot. The enhanced vision and finer movements allow more precise control of the small vessels of the mesopancreas and around the uncinate process. The dissection continues with the division of the gastrocolic ligament and access to the lesser sac, to easily complete the Kocher maneuver and prepare the anterior and lateral aspects of the SMV, the portal vein, and the spleno-mesenteric confluence behind the neck of the pancreas. The resection then proceeds as usual.

Fig 2. Preparation of the duodenal window in robotic pancreatoduodenectomy. (A) Robotic access to the duodenal window. (B) Preparation of the distal duodenum and division of the Treitz’s ligament from right to left.

The DW access leaves a large defect at the basis of the proximal transverse mesocolon. We use this access to transpose the biliogastric limb of the double-loop reconstruction [6], while the pancreatic limb has easy access through the enlarged TF.

Case series

We have applied this technique on 15 cases, of both open (10 cases) and robotic (5 cases) PD, from March 2022 to June 2023. Indication for PD was pancreatic head carcinoma in 10 cases, distal cholangiocarcinoma in 3 cases, and intraductal papillary mucinous neoplasm of the pancreas in 2 cases. Robotic PD was performed with the DaVinci XI surgical robot (Intuitive Surgical) within the Robotic Surgery Development Program of the S. Maria Hospital Trust, Terni, Italy. Please see our robotic PD technique in the Supplementary Video 1 of this manuscript.

The DW-first access is a well-known surgical maneuver, therefore ethical committee approval was deemed unnecessary. All patients gave full informed consent to the operation.

Results

The duodenal-window-first technique was performed as described before in all cases. There were no specific complications of the DW-first access in this group. One patient died of hypovolemic shock three days after open PD, although there was no bleeding during reoperation. Two patients developed Grade B POPFs, which delayed their discharge but did not require reintervention. One patient required relaparotomy and drainage due to Grade C POPF.

PD is a long and complex surgical procedure, as it entails fine dissection around delicate abdominal structures. In particular, the preparation of D2, D3, and D4 represents a challenging step. It can be achieved through the lesser sac, after the division of the gastrocolic ligament and detachment of the root of the transverse mesocolon. Unfortunately, this maneuver can be particularly challenging in patients with a history of pancreatitis or bulky lesions of the pancreatic head. Moreover, stretching the transverse mesocolon at its root may increase the risk of bleeding from the GCT. Another option could be to perform the Cattell-Braasch maneuver [7] and achieve a complete lateral-to-medial derotation of the right and proximal transverse colon to expose the whole duodenum and pancreatic head along with the superior mesenteric pedicle, up to J1 [8]. Although this is an interesting maneuver that can be performed in bulky lesions of the pancreatic head with possible invasion of the SMV/SMA and enlarged lymph nodes at station n.14 (origin of the SMA), its risks and benefits have not been fully evaluated and we wonder if it could be suitable for a mini-invasive approach.

Another option, never really explored for PD, is to access D2/D3 through the DW.

The DW is an easy-to-find anatomical landmark to access the Fredet’s fascia and the duodenum from the inframesocolic space. During right or extended right hemicolectomy, it offers an easy starting point for the dissection, to identify and protect the duodenum since the first stages of the procedure and to begin the posterior dissection (of the Fredet’s and Toldt’s planes) from a safe starting point [5]. If approached at the first stages of PD, it allows a large mobilization of the duodenopancreatic block and easy access to the root of the SMV and SMA to confirm that the tumor is resectable before entering the lesser sac.

The DW mirrors on the right side of the TF site (Fig. 1A). Once both the DW and the TF are opened, easy bilateral access is granted to two virtual spaces: (1) the anterior plan between the anterior aspect of D3/D4 and the posterior wall of the SMV/SMA and (2) the posterior plane between D2/D3 and the major abdominal vessels (Fig. 1C). These two spaces can be dissected easily in an almost bloodless way if only the two small arterial branches (IPDA and J1A) are identified and either ligated and divided or protected to be divided at a later stage. The early control of these two arteries should reduce the congestion of the pancreas head, thus facilitating further maneuvers [8]. The dissection of the DW/TF offers clear advantages with respect to the Cattell-Braasch maneuver. In fact, it allows the same preparation, that is, access to the inferior knee of the duodenum–but with a less extensive dissection, as the right colon does not get fully mobilized, with a reduced risk of bleeding or volvulus, and can also be easily performed in laparoscopic or robotic surgery, whereas the extensive dissection of the Cattell-Braasch derotation may be challenging with a mini-invasive approach. Moreover, with the DW/TF maneuver through the inframesocolic access, it is possible to accomplish an almost complete retrograde Kocher maneuver, which greatly facilitates the subsequent supramesocolic steps.

The duodenal-window-first PD is based on solid anatomical principles and can offer several advantages with respect to the classic technique of PD. Unfortunately, these advantages are hardly measurable, as the “easiness” of dissection is not a quantitative variable. However, a straightforward dissection of D2 to J1 should reduce blood loss and possibly the operative time. The reported case series is not comparative, and therefore it does not present evidence of the superiority of this technique compared to the traditional initial approach through the gastrocolic ligament. Similarly, our initial experience does not allow us to state that the DW-first approach is more useful in mini-invasive vs open surgery. We developed this technique in open surgery and used it for a while before shifting to robotics and we did not see any difference in bleeding. Comparing the estimated blood loss (and other variables) in our open vs robotic cases (10 vs. 5, respectively) would not be sensible due to the small sample size. While the advantages of the DW-first technique in open surgery are immediately evident, accessing and mobilizing D2, D3, D4, and J1 before opening the gastrocolic ligament is equally advantageous in robotic surgery.

This paper should be considered as a feasibility report to introduce the DW-first PD technique and show its anatomical basis, but the real efficacy of the DW-first technique should be further evaluated in a larger series. Moreover, the technique must be further developed and fully validated in robotic surgery.

Supplementary data related to this article can be found at https://youtu.be/kYstONlTdqA.

None.

No potential conflict of interest relevant to this article was reported.

Conceptualization: GDT, RC. Data curation: All authors. Methodology: GDT, RC. Writing - original draft: GDT, RC. Writing - review & editing: All authors.