Methods
We performed a systematic literature review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes (PRISMA) guidelines [
8]. PubMed/MEDLINE, EMBASE, Scopus, Cochrane Library (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials-CENTRAL) and Web of Science (Science and Social Science Citation Index) databases were used to search all related literature, by combining the following non-MeSH / MeSH terms: ((duodenal stump fistula OR duodenal stump leakage OR duodenal stump leak) AND (management OR treatment) AND (gastric cancer OR gastric tumor OR gastric neoplasm OR stomach cancer OR stomach neoplasm)) OR (duodenal stump AND “Fistula”[Mesh] AND “Stomach Neoplasms”[Mesh]).
Our literature review was restricted to articles published over the past 30 years (January 1988–November 2018). Only English-written scientific papers were selected, including case reports, case series, case-control studies, cohort studies, controlled clinical trials, and randomized clinical trials. Prior systematic reviews and meta-analyses were excluded. The selected articles included adult patients treated for DSF after total or subtotal gastrectomy for gastric cancer. For those patients, DSF treatment methods and DSF treatment-related outcome (DSF resolution vs no resolution/mortality, and/or healing time) had to be reported, while articles not reporting DSF treatment methods and/or post-management outcome were excluded. In addition, references of relevant articles were searched, in order to identify cases of interest.
Two independent reviewers (MZ and LU) selected and identified papers based on title, abstracts, keywords, and full-text, then collecting following information from the selected papers: author’s surname and year of publication, study period, study type, DSF patient, timing of DSF diagnosis, neoadjuvant chemotherapy administration, stage of gastric cancer, DSF output, DSF-related complications, therapeutic strategy (conservative, endoscopic, percutaneous, surgical), clinical outcome, length of hospitalization, healing time, DSF-related mortality rate. Eventually, all collected results were reviewed by a third independent reviewer (AM).
Discussion
DSF following total or subtotal gastrectomy for gastric cancer represents a rare complication with a reported incidence of 1.6–5% [
2]. Despite the relatively low incidence rate, mortality rate remains high (from 7 to 67%) with a reported spontaneous closure rate of 28–92% [
2,
5,
6,
22].
DSF pathogenesis remains unknown [
2]. Main risk factors may be devascularization of duodenal stump or its inadequate surgical closure, inflammation of duodenal wall, local hematoma, neoplastic involvement of resection line, incorrect abdominal drain placement, and postoperative distension of the duodenum due to distal obstruction [
2,
23].
Clinical DSF presentation time is variable with a mean 10-day diagnosis time [
2]. Low fistula output may delay diagnosis, making it difficult to define fistula occurrence time [
2]. Therefore, possible late clinical presentation must be kept in mind.
Many risk factors are related to DSF occurrence [
2]. These may be related to patient characteristics (advanced age, cirrhosis, diabetes, heart disease, bio-humoral nutritional status impairment - preoperative albumin < 35 g / L and/or preoperative lymphocyte count < 2000 / mm3, preoperative chronic anemia, presence of chronic ulcer or ectopic pancreas in the duodenal bulb, previous hepatobiliary surgery), primary gastric cancer-related conditions (gastric outlet obstruction before gastrectomy, pylorus cancer invasion) and intraoperative procedures (blood loss > 300 ml, absence of manual suture line reinforcement, excessive vascular or pancreatic dissection around the duodenal stump, direct thermal damage of the duodenal stump) [
3‐
6,
22].
Some studies underlined the importance of suture line reinforcement in DSF prevention [
2]. In a recent prospective phase II study, Kim et al highlighted DSF absence in 100 patients undergoing laparoscopic reinforcement suture (LARS) with barbed suture during laparoscopic gastrectomy for gastric cancer [
24]. Other authors suggested application of coated sutures, fibrin glues or resorbable reinforcements [
2]. In a retrospective study on 2034 patients undergoing gastrectomy for gastric cancer, Shao et al analyzed three different techniques of duodenal stump closure [
25]. They concluded that purse-string suture gave better outcomes in DSF rate when compared to duodenal stump treated with linear cutting stapler plus seromuscular layer reinforcement suture or full-thickness suture plus seromuscular layer reinforcement suture [
25]. Orsenigo et al reported absent manual suture line reinforcement as an independent prognostic factor for DSF occurrence [
3]. However, suture line reinforcement is not always easily performed as it happens in distal gastric lesions invading the pylorus or duodenum, where extended ultrapiloric resections are needed, as Ramos et al suggested [
2]. Prospective randomized studies might help us determine effectiveness of suture line reinforcement, which is difficult to be performed due to small sample size related to low DSF incidence.
For subtotal gastrectomy, Marincas et al recently suggested the use of an intraoperatively introduced duodenal decompression probe, with the aim of reducing DSF risk [
23]. However, the results were unsatisfactory [
23].
DSF treatment can be classified into nonsurgical (conservative, endoscopic, percutaneous) and surgical. Nonsurgical treatment includes adequate fistula drainage, infection source control, and patient nutritional support. It represents the cornerstone of DSF management [
2‐
7]. Instead, surgical treatment should be reserved only to those cases when nonoperative management does not allow an adequate fistula drainage leading to secondary complications such as intra-abdominal bleeding, sepsis, other fistulas, and intestinal obstruction [
2‐
7,
22].
Scientific literature well describes impact of parenteral and enteral nutrition in preventing major complications after upper gastrointestinal, hepatobiliary, and pancreatic surgery [
2]. Therefore, aggressive parenteral and / or enteral nutritional therapy, can significantly reduce DSF risk in addition to promoting its repair [
2‐
5]. Analysis of published papers allowed to collect neither detailed data on DSF patient nutritional status nor indications and characteristics of nutritional support, except for what Garden et al reported [
9]. However, malnutrition represents a key issue in patients with gastrointestinal fistulas, as it is closely associated to site and fistula output and it represents a major concern in patients affected by upper gastrointestinal fistulas such DSF [
26]. A previous study identified a 53% malnutrition rate in patients with gastric or duodenal fistulas [
26]. An “optimal nutritional support”, defined as a < 3000 Kcal or more per day and a positive nitrogen balance through a combination of oral, enteral and parenteral nutrition, was recommended in patients with gastrointestinal fistulas [
26].
Patients with low-output fistulas should receive basal energy requirement and 1–1.5 g of protein / kg of body weight / day, with a minimum 30% caloric intake provided as lipids [
26]. On the contrary, patients with high-output fistulas should receive 1.5–2 equivalent of their basal energy expenditure plus 1.5–2.5 g protein / kg body weight / day [
26].
As patients often fail to achieve caloric support goals through enteral route for several days after starting feeding, immediate introduction of parenteral and enteral supports is strongly recommended for those patients, with the aim of interrupting parenteral support when enteral nutritional goals are met [
26‐
29].
Equally important is that broad-spectrum antibiotics are administered and hydroelectrolytic and acid-base disorders corrected [
2]. Effectiveness of treatment with somatostatin analogues was largely debated, although many authors suggested the administration of somatostatin analogues based on their potential efficacy in reducing intestinal secretion [
3,
5]. On the contrary, the role of oral diet still appears uncertain, although it seems to be better than fasting, excluding patients with diffuse peritonitis and / or ileus [
5].
Placement of abdominal drains in surgery for gastric cancer is under discussion [
2]. It did not prevent DSF formation, although it could allow early DSF diagnosis thus avoiding other invasive diagnostic / therapeutic procedures [
2]. Patients without abdominal drains or presenting DSF after their removal may be treated by a percutaneous approach: fluoroscopy, computed tomography, or ultrasound-guide drainage with pigtail catheter placement; transhepatic biliary drainage; fistula obliteration by cyanoacrylate or prolamine; occlusive balloon or Foley catheter placement [
5,
15,
16,
18].
Biliary diversion with choledocostomy or percutaneous transhepatic biliary drainage and occlusive balloon were useful procedures in high-output fistulas described by literature [
2]. Cozzaglio et al reported effectiveness of percutaneous transhepatic biliary drainage and occlusive balloon with from 500 to 100 ml / day reduced output in 6 patients [
16]. However, complete resolution of DSF was achieved in half treated cases [
16].
Conservative and / or endoscopic and / or percutaneous approach is / are considered first choice for DSF treatment and should be extended for at least 4–6 weeks, unless patient’s clinical situation worsens, thus requiring prompt surgery [
2,
4,
5].
Surgery aims at draining multiple localized abscesses or treating a diffuse peritonitis (from severe abdominal sepsis or active bleeding) [
2]. However, authors recommended to avoid surgery on fistulas occurring between 10 days and 6 weeks of initial gastric surgery [
5]. During surgical reintervention, DSF can be managed / closed in different ways: washing of peritoneal cavity and abdominal drainage; closure of fistula (simple suture or re-stapling); biological glue; repair with rectus abdominis muscle flap; Roux-en-Y duodenojejunostomy; biliogastric diversion; laparostomy [
2,
4]. However, effectiveness of these procedures is limited and it includes high risk of duodenal stump re-leakage due to postoperative edema and inflammation [
4]. Therefore, treating acute setting with a duodenostomy would be more appropriate [
4]. Ali et al suggested duodenostomy in order to avoid complex surgical interventions, concomitant increased morbidity and longer hospitalization, allowing future surgery where possibility for transfer exists or subspeciality expertise might be required [
7]. Following duodenostomy, leakage site might close spontaneously within 2–6 weeks [
18]. Other authors suggested pancreatoduodenectomy [
4]. In addition to any surgical procedure for DSF, some authors recommended a prophylactic cholecystectomy, due to high risk of acute cholecystitis [
5]. However, need for such additional procedure has not been confirmed yet.
An analysis of cases reported by literature did not allow to identify the most appropriate surgical strategy, probably due to high number of performed surgical procedures and low number of events [
4]. However, patient outcome would seem better if peritoneal lavage and abdominal drainage were associated to surgical or percutaneous procedure on the biliary tree [
4].
Despite improvements in nonsurgical diagnostic and therapeutic procedures, and surgical techniques, DSF-related mortality rate remains high, particularly during the first weeks following onset [
5]. In small series, literature declared DSF and old age as independent factors associated to risk of surgical death (Clavien V) [
2]. On the contrary, Cozzaglio et al found that DSF alone did not lead to patient’s death [
5]. Development of new complications represented the real issue [
5]. Moreover, the risk of death appeared to be closely related to the number of arising complications [
5]. Therefore, best effort in preventing and treating septic complications is mandatory [
5].
Some authors discussed the impact of laparoscopy on the risk of DSF development. Minimally-invasive surgery, laparoscopic gastrectomy in particular, is gaining increasing popularity in gastric cancer management [
30]. Overall, relevant literature mainly stem from East Asia, while Western countries rarely performed randomized studies [
30]. Currently, as early gastric cancer is concerned, in particular when it is located in distal stomach, different randomized trials proved laparoscopic gastrectomy superiority/noninferiority, in particular in reducing surgical trauma and enhancing postoperative recovery, with no compromise on surgical safety and oncologic efficacy [
30]. Conversely, in advanced gastric cancer, multicenter large-scale randomized evidence endorsed laparoscopic gastrectomy safety and feasibility by experienced hands, while long-term survival outcomes, whose clarification requires support by several ongoing trials, remain pivotal in determining whether a wider applicability can be accepted [
30,
31]. Cozzaglio et al estimated a 5 times higher risk in laparoscopic gastrectomies [
4]. However, risk would seem related to specific learning curve, as suggested by other authors [
3]. Another possible explanation could be nonroutine execution of suture line reinforcement in laparoscopic approach, although such assumption was not confirmed by Cozzaglio et al’s large multicenter study [
4].
Limitations
Our systematic review presents several limitations: i) reported events were mainly case reports or small retrospective series; ii) populations under analysis presented heterogeneity; iii) many relevant data were not described by the authors in detail, as reported in Tables
1 and
2; iv) number of reported procedures was higher than number of patients, given frequent association of different therapeutic approaches; therefore, some patients were simultaneously taken into consideration in different groups; v) data on timing of DSF diagnosis, healing time and length of hospitalization were reported in median days or mean days; therefore, direct confrontation of results appeared difficult.