Clinically Relevant Postoperative Pancreatic Fistula rates in two methods of pancreatico-jejunal anastomosis compared: Evolution of surgical technique in a single centre over time

Pancreaticoduodenectomy remains the gold standard for management of patients with pancreatic head and periampullary neoplasms. Despite the low mortality, overall morbidity remains high, principally due to the development of a clinically relevant postoperative pancreatic stula (CR-POPF). The aim of the present study is to compare rates of CR-POPF in two groups of patients treated with two different pancreaticojejunostomy techniques.


Background
Pancreaticoduodenectomy (PD) is considered the standard treatment for malignant and benign lesions of the pancreatic head and periampullary region. It is a technically challenging procedure requiring a high level of surgical experience with regard to resection and reconstruction.
Whereas the rate of mortality associated with PD has dramatically decreased in the last two decades to below 5%, the morbidity rate remains high, from 30-50%, indicating the complexity of pancreatic surgery [1]. The most important and potentially life-threatening complication after PD is the occurrence of a postoperative pancreatic stula (POPF), traditionally perceived as the Achilles heel. From the literature, its incidence varies from 2% to more than 20% [2][3][4]. In 2005, the International Study Group of Pancreatic Fistula (ISGPF) reached a consensus, de ning a clinical system of 3 grades of POPF: A, B and C [5].
Afterwards, criticism was published until the International Study Group of Pancreatic Surgery (ISGPS) in 2017 introduced a new de nition on the grading of POPF [6][7][8]. Grade A is now rede ned as "biochemical leak"; grades B and C (clinically relevant POPF) are con rmed but de ned more strictly: patients who required percutaneous drains are classi ed as grade B, while those classi ed as grade C are associated with organ failure, reoperation or death [8].
The aim of the present study is to compare the rate of clinically relevant POPF in two groups of patients before and after the introduction of a standardized technique of pancreaticojejunal anastomosis described by Z'graggen, Shrikhande and Buchler [9,10].  Table 1.

Surgery
All patients underwent open surgical procedure.
In group 1, a pancreaticojejunostomy described by Longmire and Traverso [11] "as a direct mucosa-tomucosa anastomosis of the pancreatic duct to the jejunum" was performed. Posterior and anterior layers were constituted of 4.0 interrupted stitches and mucosa to mucosa with 6.0 poliglicolic mono lament splinted in 47% of cases with an internal stent.
Then, an end-to-side hepaticojejunostomy was splinted (91% of cases) with a T tube inserted through the bile duct.
The surgical technique of Group 2° was previously described by Z'graggen, Shrikhande and Buchler "as end to side duct to mucosa anastomosis with an outer seromuscolar and inner full thickness layer" [9,10]. The cut end of the left hemipancreas was mobilized for 2 cm to allow the placement of a posterior external suture. All stitches were mono lament absorbable PDS 5 − 0 (C-1, 13 mm 3/8c, visi-black, or TF 13 mm 1/2c) for interrupted single sutures. The rst step was the placement of at least three ductal sutures, rst anterior and then posterior, including the cut edge of the pancreas remnant; the ductal stitches were retained with clamps and organized in order.
The anastomosis consisted of two posterior row, external and inner sutures, and two anterior, inner and external sutures. External posterior suture was performed with interrupted suture every 4-5 mm placed from the dorsal pancreatic capsule and parenchyma to the jejunum end-to-side; all stitches were retained and organized, and the knots were gently tied. The jejunum was then opened on its antimesenteric side, next to the mesentery, and the mucosa was sutured on the cut edge of the pancreas with interrupted sutures with integration of previously posterior placed ductal sutures; all knots were tied after completion of the inner posterior suture. The anterior inner suture between the anterior cut edge of pancreas and the opened jejunum was performed utilizing the previously placed anterior ductal suture. The second row of anterior suture covered the inner suture line without tension and invaginated the anterior cut edge into the jejunum by 1 cm.
Then, an end-to-side hepaticojejunostomy with interrupted single sutures PDS 5.0 was performed.
In selected cases, we used the round ligament as a patch around the PJ anastomosis, as described by Iannitti et al. [12].
In both group 1° and 2°, surgical loupes were routinely used at 2.5 × magni cation for pancreaticojejunal anastomosis; in two cases, a microscope at 12.5 × was used.
Digestive continuity was restored by an ante-colic duodenal/gastro-jejunostomy in classic Whipple with a Braun anastomosis or a Roux-en-Y in ve patients. One (or two) 4 × 10 mm uted at drainages were placed behind the pancreatic and biliary anastomosis, and the postoperative secretion was routinely monitored with respect to amylase contents. A nutritional jejunostomy was placed 20 cm next to the last jejunal anastomosis and utilized from the second postoperative day [13].
In case of cancer, lymph node dissection along the hepatoduodenal ligament, common hepatic artery, vena cava, superior mesenteric vein and the right side of the superior mesenteric artery was a standard part of the procedure. Octreotide was started during the induction of anaesthesia and given for some days after operation (three doses of 0.1-0.2 mg per day subcutaneously).

Endpoint
The aim of the present study was to compare rates of clinically relevant POPF using the last classi cation proposed by ISGPS [8] in two groups of patients, before and after introduction of a different technique for pancreaticojejunal anastomosis. Morbidity and mortality were also evaluated.

Statistical analysis
Statistical analysis was done by addressing differences in categorical variables between the two groups with Chi square test for 2 × 2 contingency tables. Fisher's exact test was used when the number of an observed value was < 5. For continuous variables, a two-tailed t-test was used, and P values < 0.05 were considered statistically signi cant.

Results
Results were retrospectively analysed in terms of postoperative mortality and morbidity ( Table 2).
Mortality was de ned as death within 30 days or in hospital, irrespective of the duration of stay. The de nition of CR-POPF was based on the consensus classi cation of the International Study Group of Pancreatic Surgery (ISGPS) [8]; for the delayed gastric emptying (DGE) and Postpancreatectomy haemorrhage (PPH), we referred to ISGPS classi cation [14,15].

Mortality:
In group 1°, the mortality rate was 2,8% (2/71). One patient died of perioperative haemorrhagic shock, and the second died three months after PD after delayed sentinel haemorrhage from POPF C, reoperation with total pancreasectomy, discharge and readmission for uncontrolled diabetes and herpes virus disease. Zero mortality was observed in group 2°.

Morbidity:
As reported by a grading system applied to pancreaticoduodenectomy [16], we also assessed grade II-V Clinically relevant POPF occurred in 12 patients in group 1° (17%) and in 4 patients in group 2° (5,6%), p = 0.06. For POPF B (10 in group 1° and 3 in group 2°), an intraoperative drain was left for more than 3 weeks, except for one patient in the rst group, for whom interventional drainage was used.
Two patients had POPF C in group 1°; both underwent reoperation for sentinel bleeding from stump erosion of the gastroduodenal artery in post operative day (POD) 7 and POD 20, and one of these patients died.
In group 2°, one case of POPF C (1,4%) was observed. Because of multiorgan failure, the patient was admitted to the Intensive care unit (ICU) and then discharged after 7 days.
PPH occurred in ve patients (7%) in group 1°: two grade B and three grade C; 2 early and 3 delayed haemorrhage; and 1 intraluminal (treated with blood transfusions) and 4 extraluminal. Four subjects were submitted to re-laparotomy, in two cases for sentinel haemorrhage and POPF C. In group 2°, PPH occurred in two patients (2,8%): one grade B and one grade C; one early submitted to re-laparotomy, and one had delayed intraluminal haemorrhage treated with angiography and embolization.
Five patients suffered from DGE in both groups (7%); one DGE C case in group 2° was re-operated at POD 25 with Roux-en-Y. Prolonged enteral nutritional support through routinely placed jejuneostomy was performed and well tolerated; prolongation of hospital stay did not cause any delay of adjuvant therapy, as the patient was classi ed as TNM stage I.

Discussion
A standardized approach to the pancreatic anastomosis and a consistent practice of a single technique appear to be related to decreased postoperative complications [18].
The ISGPS in 2010, improving uniformity and standardization in pancreatic anastomosis (PA), formulated a new classi cation that incorporates factors related to the pancreatic remnant, such as pancreatic duct size, length of mobilization, and gland texture, as well as factors related to the pancreatoenteric anastomosis. The use of duct-to-mucosa anastomosis or invagination (dunking) of the remnant into the jejunum, the use of a stent (internal or external) across the anastomosis, and the use of intraperitoneal drains were other important issues discussed [19].
In 2017, an ISGPS position statement failed to provide de nitive, consistent and convincing level 1 evidence that any PA technique was better than the others. Pancreaticojejunostomy (PJ) remains the most common type of reconstruction, practised by 88.7% of surgeons. [20] Another issue of interest concerns the placement of trans-anastomotic stents. [21] Some authors reported that a normal soft pancreas does not require a stenting tube [22][23][24]. Ductal stents in pancreatic anastomosis were frequently used (47%) in our group 1° and in only one patient in group 2°. A biliary Ttube was inserted in 91% of patients in group 1° and rarely (7%) in group 2°, with the rationale that the diversion of biliary secretions away from pancreatic drainage may avoid activation of pancreatic enzymes and therefore protect the healing process of the PJ [25]. Other authors subsequently suggested reconstruction on dual loop (Roux-en-Y) [26,27].
The use of routine drains at the site of the anastomosis has been widely debated [28,29]; a randomized, prospective, multicenter trial demonstrated an increase number of complications and mortality in PD without intraperitoneal drainage [30]. Nevertheless, drains may provide access for bacteria, potentially leading to super infected uid, bile, chyle, and pancreatic juice collections [31]. Lastly, a recent dual-centre randomized trial indicated that selective drainage might be a better concept; abandoning intraperitoneal drains did not increase the reintervention rate, mortality, and overall morbidity. Moreover, the clinically relevant POPF rate was lower in the no-drain group (CR-B/C 11,9% vs 5,7%) [32].
Furthermore, management decisions involve the postoperative day of drain removal. Even the timing of drain removal is still under discussion. In a report on 1.507 patients with PD, drain removal occurred at median POD 7 [6]; in contrast, in patients with a low risk of pancreatic stula, intra-abdominal drains can be safely removed on POD 3 or POD 1 after standard pancreatic resections [33][34][35]. A prolonged period of drain insertion was associated with a higher rate of postoperative complications with increased hospital stay and costs [36,37]. We routinely used drains in group 1° patients without POPF, removing the drain on POD 7; in low risk group 2° patients, we removed drain earlier, on POD 3.
Preoperative biliary drainage (PBD) does not have a bene cial effect on postoperative outcome; intraoperative bile samples had demonstrated that contamination of bile was signi cantly increased in patients who underwent stenting, as recently reported [38]. The increase of postoperative overall complications and wound infections urges precise indications for preoperative biliary drainage [39,40]. Our experience showed many infections (biliary and wound) in patients with preoperative biliary drainage (nearly half of patients).
Grade B POPF requires a change in the postoperative management: drains should be left in place for more than 3 weeks or repositioned through endoscopic or percutaneous procedures. Grade C POPF refers to those conditions that require reoperation or lead to single or multiple organ failure or death [8].
The decrease in CR-POPF from 17-5.6% in our experience is probably due to standardization and meticulous attention to technical details. The use of ne mono lament sutures with ne needles could also reduce tissue damage, improving postoperative outcomes.
Certainly the absence of routine detection of pancreatic texture constituted a lack of the study.
The inability to tolerate a solid diet by POD 7 should be considered as DGE, occurring in 14% − 21% of patients after PD with prolonged hospitalization [1,3,4,45]. The consensus de nition of DGE after pancreatic surgery without mechanical obstruction was reached in 2007 by ISGPS and classi ed into Grades A, B and C [17]. As others author reported, in PPPD we always performed a dilatation of the pylorus with farabeuf until calibration to 26 mm [54].
PPH occurs in 1-8% of all pancreatic resection and represents from 11-18% of overall mortality [45,47,55]. In the literature, it was de ned on the basis of 3 criteria: time of onset (early PPH occurring in the rst 24 hours and late PPH occurring after 24 h postoperative), location (intra or extraluminal), cause and severity. In 2007, ISGPS proposed classifying PPH as A, B and C [15].
Several authors reported operation time, blood loss and length of stay as features of surgery [32,45,48]. In our experience, operative time was longer in group 2° (P < 0,001), perhaps for more time to perform pancreaticojejuneal anastomosis and probably due to vascular resection. Length of stay was decreased in group 2° (P < 0,001) because of lower numbers of complications, especially POPF.

Conclusion
In conclusion, the comparison between two groups of patients with different pancreatico-jejuneal anastomosis showed a lower rate of clinically relevant POPF and lower morbidity, mortality and length of stay in the second group of patients.
Although it is a long period of study observation, it seems that the reduction of stulas is mainly dependent on the type of anastomosis rather than other factors.
We consider this technique to be safe and reproducible by other surgeons through an educational surgical program.
The standardization of technical surgical details and careful perioperative care are crucial aspects, in our experience.
Furthermore, we believe, like Traverso, that "surgery alone is not su cient" for pancreatic cancer [57]. The goal is to avoid morbidity, as the best candidate for adjuvant treatment will be that patient who has no delay in postoperative recovery. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This retrospective study did not need the approval of the ethics committee. Written informed consent was obtained from all patients before surgery.

Consent for publication
Not applicable Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request

Competing interests
The authors declare that they have no competing interests.

Funding
No funding has been received for this study.