State-of-the-art surgery for pancreatic cancer

Several substantial modifications have been introduced since Allan Whipple promoted partial pancreatoduodenectomy which had been originally described by Walther Kausch more than a century ago [19]. The initial two-staged procedure has been modified by Whipple himself to a one-stage operation, anecdotally caused by the visit of a group of European surgeons coming to Columbia University in New York to learn about the new procedure and stimulating Whipple to perform the operation in one step. After the classical Whipple operation had been performed for four decades with a resection of the stomach, Traverso and Longmire introduced the modification of a pylorus-preserving procedure without compromising oncological radicality [20]. Though being regarded as the more physiological procedure, pylorus preservation has been suspected to be a promoting factor for the complication of delayed gastric emptying (DGE)—a still poorly understood and unsolved clinical problem. However, recent randomized controlled trials on this topic have investigated the resection of the pylorus to reduce DGE without confirming a beneficial effect of pyloric resection [21]. Consequently, today’s standard of care remains the pylorus preserving modification. The same observation with regard to DGE has been made for the route of reconstruction, namely, the question if an ante- or retrocolic route may be superior in order to reduce DGE. Meanwhile, high-level evidence has shown that both options are possible and do not significantly impact the incidence of DGE, leaving the decision up to the surgeon’s discretion [22].

Other specific aspects in pancreatic cancer surgery include the techniques of “artery first,” “uncinate first,” “triangle resection,” “arterial divestment,” “RAMPS,” “DP-CAR,” and venous as well as arterial vascular resections. “Artery first” described the technique of approaching the superior mesenteric artery as the initial step of the procedure to evaluate a possible arterial tumor involvement. The mesenteric artery can be approached from different directions with at least six pathways having been described [23]. These include left or right infracolic as well as superior transmesocolic approaches. These approaches should be routinely used especially in case of any doubt of an arterial involvement. A recent meta-analysis comparing artery-first vs. standard resection with more than 1400 patients showed less blood loss and transfusion as well as a higher rate of R0 resections and a shorter hospital stay when an artery-first approach was used [24]. In contrast, these findings were not confirmed by a randomized trial including 153 patients that failed to confirm the advantage in terms of a radical R0 resection [25]. However, it has to be kept in mind that the initial step of approaching the artery first does not automatically imply a true radical resection technique afterwards as the number of harvested lymph nodes was similar in this study, but there was a considerably higher rate of R1 resections in positions other than the medial margin (i.e., pancreatic transection margin 7% standard vs. 28% artery first). This underlines that an artery first approach needs to be combined with a radical soft tissue removal (see below). The “uncinate first” approach describes a technique where the resection is carried out in a caudo-cranial technique, thereby starting with the dissection of the uncinate process from the transverse mesocolon and especially the superior mesenteric vein and artery—in contrast to the classical dissection starting with the division of the cranial portion of the retropancreatic soft tissue [26]. The uncinate first technique allows a meticulous control of the vessels from the beginning of the resection phase on and requires a division of the first jejunal loop followed by skeletonization and transposition of the loop to the right side of the mesenteric axis in the beginning of the resection phase. After lifting the specimen on the right side, the mesenteric vessels are clearly exposed and guide the further resection until the skeletonization is completed. It can be combined with an artery first approach and other techniques and be performed laparoscopically or robotically. Regarding procedure-specific outcomes, lymph node harvest, R0 resection rates, operative time, and median length of hospital stay at least equal conventional approaches, some studies even report superior outcomes with less blood loss and a shorter time to oral intake [27, 28].

With the increasing application of neoadjuvant therapy in pancreatic cancer, especially in locally advanced disease, surgical strategies and concepts have been developed for cases which have been downstaged or shown a stable course. The “triangle” operation was initially described as a method of radical resection after neoadjuvant therapy in locally advanced pancreatic cancer [29]. The rationale of the procedure is the observation that after neoadjuvant therapy, conventional imaging fails to differentiate between actual tumor abutment/encasement and fibrotic residual tissue around the vascular structures. The technique comprises a frozen section proof of no viable tumor tissue around the arterial structures (celiac axis, hepatic artery, superior mesenteric artery) allowing a sharp dissection without an arterial resection or reconstruction with the removal of all lymphatic and neural tissue structures to the basis of celiac axis and superior mesenteric artery. The venous porto-mesenteric vessels—in contrast to the arterial structures—have to be resected and replaced during the procedure in a high proportion of patients. This radical artery-sparing approach results in an anatomic triangle bordered by the superior mesenteric artery, celiac axis, and portal vein. It is essential that the arteries are reached on the adventitial layer which opens longitudinally and allows for complete lymphadenectomy and soft tissue removal of the respective area. Besides its use in locally advanced pancreatic cancer, a triangle operation can—and potentially should be—carried out in all resectable as well as borderline resectable tumors in the light of a truly radical surgical approach [30, 31].

Another related technical approach in patients after neoadjuvant therapy has been described as the “periarterial divestment” technique [32]. This technique comparably aims at a radical tumor clearance without arterial resection and is characterized by entering the adventitial layer between the arterial wall and remnant tumor/fibrotic tissue. Once this layer is opened, it is the guiding plane for dissection and allows the surgeon to avoid an arterial resection in a considerable proportion of patients.

Radical antegrade modular pancreatosplenectomy (RAMPS) describes the systematic radical surgical approach in distal pancreatectomy [33]. It is characterized by anatomic landmarks including the left-sided portal vein as well as the aorta with the celiac axis and superior mesenteric artery as the left-sided borders and the left kidney vein and diaphragm as the inferior and superior borders, respectively. Posteriorly, the location and extension of the tumor defines the choice of procedure. This is either an anterior RAMPS including Gerota’s fascia, the prerenal fat on the surface of the adrenal gland and upper half of the kidney or a posterior RAMPS which includes the left adrenal gland, all retroperitoneal fat tissue and leaving the muscle layer of the abdominal wall as the posterior border of dissection. This procedure can lead to R0 resection rates as high as 90%’ however, its influence on recurrence-free and overall survival remains controversial.

Distal pancreatectomy with celiac axis resection (DP-CAR) is a modification of the Appleby procedure, originally developed for gastric cancer surgery [34]. It is suitable for a subgroup of pancreatic cancer patients in whom only the celiac axis is involved, whereas aorta, superior mesenteric artery, and gastroduodenal artery are not involved. In this patient group, DP-CAR allows a radical tumor removal and can lead to a margin-negative resection with a median overall survival ranging between 16 and 32 months comparable to resectable pancreatic cancer [35]. The crucial point in this procedure is the arterial liver perfusion via retrograde inflow from the superior mesenteric artery, inferior pancreatico-duodenal artery, and the gastroduodenal artery. Some centers have described the technique of preoperative hepatic artery embolization to precondition arterial flow and finally allow DP-CAR in a larger proportion of patients than without embolization [36]. Despite such modifications, DP-CAR has traditionally been burdened by high rates of morbidity (50–80%) as well as mortality, varying between 3.5 and 17% [35]. As shown for many surgical pancreas procedures, these outcomes are depending on the center’s case load which result in three- to five-fold differences in mortality (18% in low volume vs. 3.5–6% in high volume centers, respectively). Furthermore, oncological outcomes of DP-CAR can be improved when embedded in a neoadjuvant therapy concept.

Overall, artery first, uncinate first, triangle operation, periarterial divestment, RAMPS, and DP-CAR are complementary modern techniques in pancreatic cancer surgery which are vessel-oriented and allow not only resections in patients that historically were regarded as unresectable, but also the removal of all putatively tumor-infiltrated soft tissue with the aim of an increased rate of R0 resection combined with a decreased risk of local recurrence.

Vascular resection during pancreatic cancer surgery has been described since the 1950s [37]. However, until the 1990s, there has not been a widespread acceptance for such procedures, despite reports describing en bloc pancreatoduodenectomy with vein resection with reasonable results [38]. In the meantime, venous resections have become a routine surgical procedure in high volume centers and have been applied for all types of resections including pancreaticoduodenectomy, distal, or total pancreatectomies. Depending on tumor location and extent of resection, four types of venous reconstruction can be differentiated and have been defined [39]. Surgical outcomes show that pancreatectomy with venous resection requires longer operative time and leads to an increased blood loss compared to standard resections although overall postoperative morbidity is comparable [40]. As venous resections are performed for more advanced tumors, they may be burdened by lower R0 rates and more positive lymph nodes. Based on these findings, from the oncological point of view, patients with venous resection may show lower 1-year, 3-year, and 5-year overall survival rates [41]. Yet, this remains controversial as a recent propensity score-matched analysis showed similar survival in pancreaticoduodenectomy with venous resection compare to standard pancreaticoduodenectomy after adjustment for baseline characteristics. Also, a Japanese study demonstrated a 30-month median survival time in these borderline resectable patients with venous resection in combination with adjuvant therapy, which is comparable to the survival of patients without venous resections [42].

This underlines the need to perform a venous resection whenever required to achieve negative resection margins and not to compromise radicality by avoidance of vascular resection. However, the effects of neoadjuvant therapy in this setting cannot be finally answered until results from larger high-level evidence studies are available.

In contrast to DP-CAR as an arterial resection where no vascular reconstruction is required, pancreatic resections with major arterial reconstruction have had—and partially still have—a questionable reputation due to both a high morbidity and mortality as well as potentially poor oncological outcomes [43]. However, in recent years, several large studies have demonstrated that these problems can be overcome by passing a learning curve with a respective case load and complication management facilities as well as a multimodal treatment approach. Single-center series including between 34 and 195 patients show good surgical outcomes with morbidity and mortality rates of approximately 50% and mortality of as low as 3–5% in experienced hands [44‐46]. Furthermore, median survival times of 29 months and up to 25% 5-year survival justify this approach as an important tool in modern pancreatic cancer surgery after proper patient selection.

Open surgery remains the standard of surgical care for partial and total pancreatoduodenectomy. Some recent studies show that minimally invasive techniques (laparoscopic/robotic) can reduce the morbidity of pancreatectomies without having a negative impact on cancer outcome. After two randomized controlled trials showed potential advantages of laparoscopic pancreatoduodenectomy, these results were challenged by the Dutch LEOPARD 2 trial which reported an increased mortality in the laparoscopic group and was—therefore—stopped prematurely [47]. A subsequent meta-analysis, however, did not confirm the potential inferiority of the laparoscopic approach, but rather showed that applying laparoscopic techniques is a matter of learning curve and case load [48]. In line with this, a recent Chinese multicenter randomized trial on laparoscopic vs. open partial pancreatoduodenectomy including 656 patients has shown similar perioperative outcomes in terms of morbidity, mortality, and radicality of resection [49]. In the laparoscopic group, hospital stay was reduced by one day as the only significant result. Thus, as this difference is marginal, long-term results are not available, and this procedure is associated with a long learning curve; it cannot be recommended as a routine approach.

In contrast, minimally invasive distal pancreatectomy can be recommended as the standard approach for tumors of the body and the tail of the pancreas. There is level I evidence for a superiority in terms of postoperative delayed gastric emptying, time to functional recovery, and length of hospital stay [50, 51].

Robotic approaches have to date only been reported in observational patient cohorts. Therefore, no evidence-based recommendation can currently be given. In centers with the respective expertise, robotic approaches seem to be not only feasible but also potentially advantageous in terms of blood loss, time to recovery, and length of hospital stay without compromising oncological radicality [52]. Also, extended robotic approaches in pancreatic cancer surgery, including pancreatoduodenectomy with venous resection and DP-CAR, have been reported with good outcomes. These findings remain to be confirmed in larger series and are—as all advanced surgical procedures—depending on an extensive experience of the surgeon as well as the center. Current randomized and thereby high-level evidence studies on the impact of robotics in pancreatic cancer surgery are ongoing and focus not only on perioperative but also on long-term oncological results.