Comparison of three reconstruction techniques performed after pancreaticoduodenectomy: Using external, internal, or no stent

Abstract

BACKGROUND

Postoperative pancreatic fistula (POPF) is the most frequent cause of morbimortality after pancreaticoduodenectomy, but the best technique to use to prevent its development is unclear. The choice of drainage method external duct stent (EDS), internal duct stent (IDS), or non-ductal stent (NDS) is also controversial.

AIM

To compare the three groups (EDS, IDS and NDS), analyzing the patient characteristics, perioperative examinations and survival.

METHODS

Patients who underwent pancreaticoduodenectomy and pancreaticojejunostomy between 2012 and 2020, were divided into the EDS, IDS and NDS groups.

RESULTS

Of the 244 patients included, 129 were in the EDS group, 71 in the IDS group, and 44 in the NDS group. Except for preoperative pancreatitis in the NDS patients, comorbidities were similar among the groups. Patients in the NDS group had a high caliber of the Wirsung duct and frequently presented with a hard pancreas (P < 0.001). A lower rate of grade C POPF was observed in the EDS (1.6%) compared to the NDS (9.1%) and IDS group (14.1%) (P = 0.009). The groups showed similar findings for delayed gastric emptying, postoperative hemorrhage, reoperation, and 5-year survivals. Ninety-day mortality rate was significantly higher in the IDS group (5.6%) compared to the EDS (1.6%) and NDS (4.5%) groups (P = 0.046). Multivariate analysis showed that the use of EDS was a protective factor for grade B/C POPF (P = 0.034), and 90-day mortality (P = 0.018). Additionally, a Wirsung duct diameter < 3 mm was the only risk factor for grade B/C POPF (P = 0.001), and 90-day mortality (P = 0.031).

CONCLUSION

The use of the EDS was a protective factor for grade B/C POPF and 90-day mortality, and the Wirsung duct < 3 mm was a risk factor for grade B/C POPF and 90-day mortality.

Key Words: Pancreaticoduodenectomy; Pancreaticojejunostomy; External duct stent; Internal duct stent; No ductal stent; Postoperative pancreatic fistula

Core Tip: Postoperative pancreatic fistula is the most frequent cause of morbimortality after pancreaticoduodenectomy. Controversy exists concerning the best reconstruction technique to reduce morbimortality. Comparison of the three techniques of Wirsung duct drainage (external, internal or non-ductal stent) after pancreaticoduodenectomy demonstrated that external stent was a protective factor for grade B/C postoperative pancreatic fistula and 90-day mortality, and the Wirsung duct < 3 mm was a risk factor for grade B/C postoperative pancreatic fistula and 90-day mortality.

INTRODUCTION

Pancreaticoduodenectomy is universally accepted as the therapy of choice for tumors of the pancreatic head and periampullary region. In most institutions with high case volume, the mortality rate after pancreaticoduodenectomy has been significantly reduced to below 5%; however, the incidence of post-pancreaticoduodenectomy complications remains between 32.4% and 61.1%[1-6]. The main reason for this high morbidity is the development of postoperative pancreatic fistulas (POPFs), at rates of 21.6% to 37.7%[1,3,7,8]. The most frequent complications associated with POPFs are peritonitis, intra-abdominal abscess, sepsis, or hemorrhage[9,10] and the mean POPF-related mortality is about 1%, with 37% to 43% of postoperative deaths being directly related to POPF[11].

Currently, no agreement concerning the best reconstruction technique to use after pancreaticoduodenectomy has been reached. Pancreaticojejunostomy and pancreaticogastrostomy are the most common techniques used for pancreatic diversion, but no significant differences have been found between the two techniques in terms of occurrence of POPFs or global morbidity[11-13].

To reduce the rate of POPF, and consequently morbimortality, placement of an external duct stent (EDS) or internal duct stent (IDS) is widely used in pancreaticojejunostomy[1,3-6,14-16]. Currently there is no consensus regarding the best method of Wirsung duct drainage. Thus, the results comparing IDS and EDS have been mixed, although better results have been reported using EDS in most studies[2,6,8,11,14,15] finding similar results in other series[1,5,6,16]. Complementary therapies such as the administration of somatostatin analogs or use of fibrin glue to reinforce pancreaticojejunostomy have failed to reduce the occurrence of POPFs[17,18]. To our knowledge, no previous studies have compared the three most common complementary techniques, EDS, IDS, and non-ductal stent (NDS) after pancreaticoduodenectomy and pancreaticojejunostomy. The aim of this retrospective study was to compare the outcomes of these three groups of patients using EDS, IDS, or NDS after pancreaticoduodenectomy.

MATERIALS AND METHODS

Study design and patients

A series of patients who underwent pancreaticoduodenectomy for tumors located in the pancreatic head and periampullary region at our institution were included in our study. End-to-side or end-to-end pancreaticojejunostomy was performed after pancreaticoduodenectomy. The patient sample was divided into three groups according to the type of pancreaticojejunostomy performed. Patients in the EDS group received an EDS across the pancreaticojejunostomy, those in the IDS group an IDS across the pancreaticojejunostomy, and in those of NDS group no stent was placed across the pancreaticojejunostomy.

The decision to perform a pancreaticojejunostomy, with or without stent placement, was based on the preferences of the surgeon, consistency of the pancreas, and diameter of the Wirsung duct of the pancreas.

We retrospectively collected clinical data from medical records of patients treated with a pancreaticoduodenectomy, including the following variables: Patient demographics, comorbidities, clinical presentation, laboratory parameters, diagnostic procedures, perioperative variables, histological features, tumor staging, postoperative morbidity/mortality, and actuarial patient survival. After discharge all surviving patients were routinely reviewed every three months in the outpatient clinic performing hematological and biochemical tests, tumoral markers [carcinoembryonic antigen and carbohydrate antigen (CA) 19-9] and computed tomography scan. The follow-up was closed in June 2023. No patients were lost during follow up.

Informed consent for surgical treatment was obtained from all patients before surgery. This retrospective cohort study was approved by our institutional review board (approval number of IRB: 23/056) and was conducted in accordance with the 1975 Helsinki Declaration. This study is conducted and reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology criteria[19]. The need for local clinical research ethics committee approval was waived due to the retrospective nature of the research.

Surgical technique

Pylorus-preserving pancreaticoduodenectomy was performed when the tumor was distant from the duodenum and antrectomy was performed when it was close to the proximal portion of the duodenum. Tumor invasion of the portomesenteric venous axis was treated with segmental resection and end-to-end anastomosis.

Each surgeon chose the anastomosis technique (with or without ductal stent) based on consideration of their experience, preference, and the presence of risk factors such as a Wirsung duct diameter of < 3 mm and/or a soft consistency to the pancreas. One (classical Child´s reconstruction) or two jejunal loops were used for pancreaticojejunostomy, at the surgeon’s discretion.

The pancreaticojejunostomy was usually performed using an end-to-side duct-to-mucosa anastomosis through the two-layer technique. For the inner layer, suturing was performed to join the Wirsung duct and jejunal mucosa using 6-8 interrupted sutures of absorbable monofilament (5/6-0 polyglyconate). For external layer interrupted suturing with polypropylene 3/0 sutures were used.

In patients treated with an EDS, the tip of the device [polyvinyl cylindrical urethral catheter (Drenoplex-DICLISA^R)], with 3 distal holes measuring 2.0 mm-3.90 mm, gauges 6 or 8, depending on the diameter of the Wirsung, was placed approximately 4 cm into the Wirsung duct, passing into the jejunal lumen, and externalized through the ascending proximal afferent jejunal loop. The external end of the stent was closed using the Witzel technique. The distal end of the catheter was externalized through the left anterior abdominal wall, where it was anchored to the skin using 3/0 silk sutures. The external stent was closed and covered with a dressing at the time of discharge and removed 5-6 weeks after surgery at an outpatient clinic. In patients treated with an IDS, the stent was placed approximately 4 cm into the duct of Wirsung and approximately 6 cm into the intestinal lumen. Both internal and external stents were fixed to the inner layer of the anastomosis with a single 3/0 Vicryl suture to prevent the stent from dislodging. The NDS technique was performed either by duct-to-mucosa pancreaticojejunostomy or invagination pancreaticojejunostomy (Figure 1).

Figure 1 Schematic techniques of pancreaticojejunostomy after pancreaticoduodenectomy. A: External ductal stent: Duct-to-mucosa anastomosis and outer layer joining pancreatic parenchyma and jejunum; B: Internal ductal stent: Duct-to-mucosa anastomosis and outer layer joining pancreatic parenchyma and jejunum; C: No ductal stent: Invagination of the transected pancreas by pancreaticojejunostomy (inner layer) and outer layer joining pancreatic parenchyma and jejunum.

Gastrojejunostomy or duodenojejunostomy (pylorus-preserving) was performed approximately 60 cm to 65 cm distal from the end-to-side hepaticojejunostomy. Two drains were placed superior and inferior to the pancreaticojejunostomy. Antibiotic prophylaxis and parenteral nutrition were administered. Intra-abdominal drains were usually removed on 4^th or 5^th postoperative day in the absence of abdominal complications (fistula, hemorrhage, or infection) with drainage amylase levels < 400 IU. In the absence of POPF, the nasogastric tube was usually removed on 3^rd postoperative day and sips of water were progressively administered, starting with a regular diet on 6^th or 7^th postoperative day.

Definitions

Preoperative pancreatitis was defined as the presence of abdominal pain, more than 3 times the normal value of serum amylase and radiological confirmation by ultrasonography or computed tomography.

Pancreatic fistulas were classified according to the updated definition formulated by the International Study Group on Pancreatic Surgery[20]: Grade A, biochemical fistulas; Grade B, requirement of change in therapy or percutaneous drainage of collections; and Grade C, patients with clinical instability, requiring drainage of collections or reoperations, leading to organ failure and/or mortality. Post-pancreaticoduodenectomy bleeding was classified according to the International Study Group on Pancreatic Surgery definition[21], as was post-pancreaticoduodenectomy delayed gastric emptying (DGE)[22]. Complications were reported using the Clavien-Dindo classification[23].

The primary endpoint of this study was to investigate the rate and risk factors for clinically relevant grade B/C POPF development. Secondary endpoints included morbidity rate, lengths of intensive care unit and postoperative hospital stays, rate and risk factors for 90-day mortality, and patient survival.

Statistical analysis

Qualitative variables are expressed as absolute numbers and relative frequencies are expressed as percentages. Associations were analyzed using the χ² or Fisher’s exact test, as appropriate. Most quantitative variables did not have a normal distribution according to the Kolmogorov-Smirnov test; therefore, all the quantitative variables were expressed as medians and percentiles and expressed as 0 and 100. The relationship between quantitative variables was analyzed using the Mann-Whitney U test. All clinically significant variables and all variables with P < 0.05 in the univariate analysis were subsequently investigated using a binary logistic regression model to evaluate the association between variables at baseline evaluation and 90-day mortality. We also investigated the relationship with the presence of grades B or C fistulas. Results are expressed as odds ratios and 95% confidence intervals. Survival analysis was performed using the Kaplan-Meier estimator and the log-rank test. A P value of < 0.05 was considered statistically significant. Statistical analysis was performed using SPSS Statistics, version 25 (SPSS, Inc., Chicago, IL, United States).

RESULTS

From January 2012 to December 2022, 244 patients underwent pancreaticoduodenectomy for tumors located in the pancreatic head and periampullary region. The EDS group contained 129 patients, the IDS group 71 patients, and the NDS group 44 patients (Figure 2). The median age, sex frequency, American Society of Anesthesiologists (ASA) score, and body mass index were similar among the groups, with no significant differences in comorbidities and treatment with neoadjuvant therapy. Antecedents of preoperative acute pancreatitis (diagnosis < 2 months before pancreatoduodenectomy) were significantly more frequent in the NDS group; however, no significant differences were observed among the other clinical findings. The frequency of use of diagnostic radiological procedures, and tumor and bile duct sizes were similar among the groups. No significant differences were observed among the groups with regards to the placement of the preoperative biliary stents. Comparative analysis of laboratory parameters only showed significantly higher levels of the tumor marker CA19-9 in the EDS group (P = 0.004), as compared to the other two groups (Table 1).

Figure 2 Flowchart of patient selection. PD: Pancreatoduodenectomy; EDS: External duct stent; IDS: Internal duct stent; NDS: Non-ductal stent.

Table 1 The preoperative characteristics of patients in the three groups who underwent pancreaticoduodenectomy, n (%).

Characteristics	EDS group (n = 129)	IDS group (n = 71)	NDS group (n = 44)	P value
Age, years	68 (45-86)	64 (17-85)	66 (16-89)	0.091
Sex, male/female	65/64 (50.4/ 49.6)	43/28 (60.6/39.4)	24/20 (54.5/45.5)	0.384
ASA score				0.652
I-II	62 (48.1)	38 (53.6)	22 (50)
III-IV	67 (51.9)	33 (46.4)	22 (50)
Body mass index, kg/m2	24.6 (16-41)	25.2 (14-43.3)	24.5 (18-36.3)	0.760
Comorbidities
Cardiovascular	68 (52.7)	32 (45.1)	22 (50)	0.124
Hypertension	59 (45.7)	31 (43.7)	16 (36.4)	0.556
Smoker	28 (21.7)	17 (23.9)	8 (18.2)	0.887
Diabetes	27 (20.9)	23 (32.4)	14 (31.8)	0.109
Pulmonary	24 (18.6)	18 (25.3)	8 (18.2)	0.096
Tumor	19 (14.7)	9 (12.7)	8 (18.2)	0.549
Alcohol use	18 (14)	10 (14.1)	4 (9.1)	0.886
Neoadjuvant chemotherapy	19 (14.7)	7 (9.9)	2 (4.5)	0.165
Clinical
Jaundice	84 (65.1)	42 (59.2)	24 (54.5)	0.412
Weight loss	72 (55.8)
Abdominal pain	62 (48.1)	25 (35.2)	18 (40.9)	0.204
Cholangitis	24 (18.6)	13 (18.3)	8 (18.2)	0.997
Preoperative pancreatitis	10 (7.8)	6 (8.5)	12 (27.3)	0.001
Duodenal obstruction	9 (7)	4 (5.6)	0 (0)	0.203
Upper gastrointestinal bleeding	8 (6.2)	3 (4.2)	2 (4.5)	0.811
Diagnostic examinations/findings
CT scan	126 (97.7)	69 (97.2)	43 (97.7)	0.973
MRI	91 (70.5)	53 (74.6)	30 (68.2)	0.728
Endoscopic ultrasound	73 (56.6)	49 (69)	27 (61.4)	0.226
ERCP	64 (49.6)	38 (53.5)	23 (52.3)	0.886
PTHC	19 (14.7)	8 (11.3)	3 (6.8)	0.368
Tumor size, cm	2.2 (0.1-6)	2.1 (0.2-7.1)	2 (0.2-5.5)	0.653
Bile duct size, cm	1.2 (0.3-2.5)	1.2 (0.2-2.4)	1 (0.2-3.1)	0.303
Preoperative biliary stent				0.127
No stent	78 (60.5)	37 (52.1)	23 (52.3)
Stent	51 (39.5)	34 (47.9)	21 (47.7)
Laboratory findings
Hemoglobin, g/dL	13.1 (9.1-17.2)	12.6 (10-16)	13 (11.2-16)	0.401
Platelet count, × 103	238 (59-431)	236 (130-487)	227 (80-390)	0.103
INR	1 (0.7-1.34)	1.02 (0.89-1.26)	1 (0.87-1.47)	0.664
Creatinine, mg/dL	0.78 (0.38-1.62)	0.76 (0.52-2.48)	0.77 (0.53-2.15)	0.379
Glycemia, mg/dL	105 (56-304)	92 (77-107)	120 (76-267)	0.401
Bilirubin, mg/dL	2.7 (0.2-26)	1.2 (0.2-2.4)	4.2 (0.2-28.9)	0.475
GOT, IU/L	60 (14-571)	63 (17-372)	79 (12-1140)	0.658
GPT, IU/L	226 (10-2246)	281 (9-1877)	228 (8-1407)	0.877
GGT, IU/L	76 (13-584)	117 (13-933)	89 (9-1682)	0.554
Alkaline phosphatase, U/L	274 (39-2262)	191 (42-1378)	359 (31-1441)	0.799
CA19-9, U/mL	75.4 (3-13950)	28.1 (21-950)	16.5 (10-1052)	0.004
CEA, ng/mL	2.7 (0.1-831)	0.53 (0-33.7)	2.8 (0.7-48)	0.313
Cytology	23 (17.8)	10 (14.1)	5 (11.4)	0.545

EDS: External duct stent; IDS: Internal duct stent; NDS: Non-ductal stent; ASA: American Society of Anesthesiologists; CA: Carbohydrate antigen; CEA: Carcinoembryonic antigen; CT: Computed tomography; ERCP: Endoscopic retrograde cholangiopancreatography; GGT: Gamma-glutamyl transferase; GOT: Glutamic oxaloacetic transaminase; GPT: Glutamic pyruvic transaminase; INR: International normalized ratio; MRI: Magnetic resonance imaging; PTHC: Percutaneous transparietohepatic cholangiography.

Surgeries were significantly shorter in the EDS group (P = 0.001), and patients in the NDS group were significantly more frequently transfused (P = 0.036). The caliber of the pancreatic duct was significantly larger in the NDS group (P < 0.001) than they were in the other groups. Histological variables and R0 resection were similar among the groups (Table 2).

Table 2 Perioperative variables and histological features among the three groups, n (%).

Variable/feature	EDS group (n = 129)	IDS group (n = 71)	NDS group (n = 44)	P value
Duration of surgery, minutes	370 (190-590)	430 (300-748)	415 (240-710)	0.001
Patient transfused	14 (10.6)	19 (26.8)	6 (13.6)	0.036
Surgical findings/techniques
Caliber of Wirsung duct, mm	3.5 (1-13)	3 (2-10)	5 (1-17)	0.001
Consistency of the pancreas				< 0.001
Hard	60 (46.5)	20 (28.2)	31 (70.5)
Soft	38 (29.5)	16 (22.5)	5 (11.4)
Normal	31 (24)	35 (49.3)	8 (18.1)
Pylorus preservation	55 (42.6)	41 (57.7)	26 (59.1)	0.051
Vascular resection	12 (9.3)	6 (8.5)	5 (11.4)	0.872
Type of reconstruction				0.119
One intestinal loop	99 (76.7)	64 (90.1)	37 (86)
Two intestinal loops	30 (23.3)	7 (9.9)	6 (14)
Histological features
Type of tumor				0.558
Adenocarcinoma	89 (68.9)	50 (70.4)	29 (65.9)
Cholangiocarcinoma	10 (7.7)	4 (5.6)	3 (6.8)
Neuroendocrine	6 (4.6)	4 (5.6)	4 (9.1)
Other	24 (18.6)	13 (18.3)	8 (18.2)
Degree of differentiation				0.991
Good	32 (24.8)	17 (23.9)	11 (25)
Moderate	45 (34.9)	30 (42.3)	20 (45.5)
Poor	19 (14.7)	12 (16.9)	7 (15.9)
Not valuable	33 (25.6)	12 (16.9)	6 (13.6)
Tumor invasion				0.339
In situ	2 (1.5)	1 (1.4)	2 (4.5)
< 2 cm	20 (15.5)	14 (19.7)	10 (22.7)
> 2 cm, limited to pancreas	36 (27.9)	28 (39.4)	10 (22.7)
Peripancreatic invasion	44 (34.1)	20 (18.2)	17 (38.6)
Invasion at CA or SMA	6 (4.6)	1 (1.4)	0 (0)
Lymph node invasion	71 (55)	38 (53.5)	25 (56.8)	0.938
Number of resected nodes	15 (11-38)	15 (10-53)	13 (8-78)	0.602
Number of positive nodes	1 (0-11)	1 (0-14)	1.5 (0-8)	0.707
Macrovascular invasion	18 (13.9)	12 (16.9)	8 (18.2)	0.961
Microvascular invasion	48 (37.2)	27 (38)	16 (36.4)	0.961
Neural invasion	61 (47.3)	35 (49.3)	16 (36.4)	0.228
R0 resection	84 (65.1)	48 (67.6)	24 (54.5)	0.361
Tumor staging
IA	7 (6.7)	8 (13.1)	4 (10.3)
IB	8 (7.6)	11 (18)	0 (0)
IIA	11 (10.5)	5 (8.2)	7 (17.9)
IIB	41 (39)	15 (24.6)	17 (43.6)
III	15 (14.3)	11 (18)	3 (7.7)
IV	4 (3.8)	2 (3.3)	2 (5.1)

EDS: External duct stent; IDS: Internal duct stent; NDS: Non-ductal stent; CA: Celiac artery; SMA: Superior mesenteric artery.

The median values of C-reactive protein levels and overall frequency of medical complications and wound infection showed no significance among the groups. Significant differences were found among the groups in terms of the overall rates of POPF development. Specifically, grade C POPFs occurred more frequently in the IDS group than in the NDS and EDS groups (P = 0.009).

The rates of DGE and postoperative hemorrhage were similar among the three groups; however, the rate of reoperation was higher, but not significantly so, in the NDS group. All patients with grade C POPF were managed by relaparotomy, performed according to the following: (1) The abscess was washed and drained in one patient in the EDS group, two patients in the IDS group, and 6 patients in the NDS group; (2) The pancreaticojejunostomy was disassembled, the jejunal loop closed, and drainage performed in one patient in the EDS group and three patients in the NDS group; or (3) Total remnant pancreatectomy was performed in two patients in the IDS group and one patient in the NDS group.

The rate of grades I and II, but not grades III and IV, complications-graded according to the Clavien-Dindo classification-was significantly different among the three groups; however, a lower rate was observed in the EDS group. There were no significant differences among the groups regarding lengths of intensive care unit and hospital stays. A significant difference was demonstrated among the groups concerning mortality at 90-days after pancreaticoduodenectomy, which coincided with hospital and 90-day mortalities. The overall rate of grade C POPF was 6.6% (16 patients) and eight out of the 16 (50%) patients died because of this complication due to intra-abdominal sepsis and/or hemorrhage. At follow-up, morbidity and frequency of adjuvant chemotherapy administration were similar among the groups (Table 3). No significant differences among the groups were seen regarding 1-, 3-, and 5-year actuarial patient survival: 84.6%, 53.2%, and 40.2%, respectively, in the EDS group; 76.9%, 45.2%, and 39.9%, respectively, in the IDS group; and 81.8%, 48.4%, and 29%, respectively, in the NDS group (P = 0.573).

Table 3 Postoperative morbimortality among the three groups, n (%).

	EDS group (n = 129)	IDS group (n = 71)	NDS group (n = 44)	P value
C-reactive protein, mg/dL, 3rd day	14.6 (0.1-39)	20.4 (2-57)	10.3 (0.6-33.6)	0.054
Medical complications	39 (30.3)	24 (33.8)	17 (38.6)	0.307
Surgical complications
Wound infection	20 (15.5)	12 (16.9)	4 (9.3)	0.515
Postoperative pancreatic fistula				0.009
Grade A/non-POPF	108 (83.7)	50 (70.5)	36 (81.8)
Grade B	19 (14.7)	11 (15.5)	4 (9.1)
Grade C	2 (1.6)	10 (14.1)	4 (9.1)
Delayed gastric emptying				0.195
Grade A	20 (15.5)	12 (16.9)	10 (22.7)
Grade B	20 (15.5)	19 (26.8)	6 (13.6)
Grade C	7 (5.4)	4 (5.6)	0 (0)
No delayed gastric emptying	82 (63.6)	36 (50.7)	28 (63.6)
Postoperative hemorrhage				0.460
Grade A	6 (4.7)	4 (5.6)	2 (4.5)
Grade B	7 (5.4)	3 (4.2)	4 (9.1)
Grade C	3 (2.3)	5 (7)	4 (9.1)
No postoperative hemorrhage	113 (87.6)	59 (83.1)	34 (77.3)
Reoperations	16 (12.4)	14 (19.7)	9 (20.5)	0.063
Hemoperitoneum	3 (2.3)	4 (5.6)	4 (9.1)
Pancreatic fistula	2 (1.6)	10 (14.1)	4 (9.1)
Jejunostomy	3 (2.3)	0 (0)	0 (0)
Other	8 (6.2)	0 (0)	1 (2.3)
Complications, Clavien-Dindo classification
Grade I-II	74 (57.3)	18 (40.9)	25 (35.2)	0.007
Grade III-IV	26 (20.1)	13 (29.5)	23 (32.5)	0.149
Intensive care unit stay, days	2 (1-15)	2 (1-30)	2 (1-30)	0.157
Hospital stay, days	17 (5-78)	20 (7-97)	18 (5-59)	0.089
Follow-up
Mortality, 90 days	2 (1.6)	4 (5.6)	2 (4.5)	0.046
Morbidity, 90 days	90 (70.3)	42 (59.1)	30 (73.2)	0.436
De novo endocrine insufficiency	16 (12.4)	15 (21.1)	6 (13.6)	0.293
De novo exocrine insufficiency	53 (41.1)	30 (42.3)	19 (43.2)	0.217
Venous thrombosis	4 (3.1)	5 (7)	3 (6.8)	0.142
Pulmonary embolism	4 (3.1)	2 (2.8)	0 (0)	0.204
Adjuvant chemotherapy	62 (48)	34 (47.9)	22 (50)	0.975

EDS: External duct stent; IDS: Internal duct stent; NDS: Non-ductal stent; POPF: Postoperative pancreatic fistula.

Multivariate analysis showed that the use of EDS was a protective factor for grade B/C POPF (P = 0.034), and Wirsung duct diameter < 3 mm was the only risk factor for grade B/C POPF (P = 0.001) (Table 4). Multivariate analysis also demonstrated that the use of an EDS was a protective factor for 90-day mortality (P = 0.031), and a Wirsung duct < 3 mm was a risk factor for 90-day mortality (P = 0.018) (Table 5).

Table 4 Results of the logistic binary regression for postoperative pancreatic fistula of grades B/C after pancreaticoduodenectomy.

Factor	Odds ratio	P value	95% confidence interval
External duct stent	0.466	0.034	0.231-0.943
Age > 80 years	0.530	0.349	0.140-2.001
ASA III/IV	1.404	0.345	0.695-2.838
Preoperative pancreatitis	1.461	0.529	0.449-4.757
Wirsung duct < 3 mm	3.794	0.001	1.761-8.173
Preoperative biliary drainage	0.924	0.827	0.453-1.883
Soft consistency of pancreas	1.057	0.803	0.683-1.637

ASA: American Society of Anesthesiologists.

Table 5 Results of the logistic binary regression for 90-day mortality after pancreaticoduodenectomy.

Factor	Odds ratio	P value	95% confidence interval
External duct stent	0.289	0.018	0.019-0.670
Age > 80 years	1.420	0.760	0.150-13.479
ASA III/IV	0.569	0.404	0.152-2.142
Preoperative pancreatitis	0.532	0.583	0.056-5.052
Wirsung duct < 3 mm	4.223	0.031	1.122-15.252
Preoperative biliary stent	0.615	0.463	0.168-2.253
Soft consistency of pancreas	1.320	0.517	0.565-3.104

ASA: American Society of Anesthesiologists.

DISCUSSION

Even though the three groups of patients in our study were relatively homogeneous, a significantly higher preoperative level of CA19-9, reflecting more advanced tumors, was found in the EDS group, and a significantly higher rate of preoperative pancreatitis was found in the NDS group coincident with a higher mean value of bilirubin. The relationship between preoperative acute pancreatitis and pancreatic cancer is controversial, because a significant number of patients with this tumor have presented acute pancreatitis shortly before pancreaticoduodenectomy[24]. However, other authors have reported that the history of pancreatitis ≥ 2 years before a diagnosis of pancreatic cancer accelerates carcinogenesis, with patients being diagnosed 4.7 years earlier than patients without a history of pancreatitis[25]. Additionally, in our series all patients with acute pancreatitis were diagnosed < 2 months before pancreaticoduodenectomy, ruling out its carcinogenic effect.

Wirsung duct stenting is preferred in patients with soft pancreas and nondilated pancreatic ducts[7,26]. Independent of the surgeon’s preferences, the NDS technique was more frequently used in our patients who showed a hard pancreas and a Wirsung duct > 3 mm, as these two factors facilitate a pancreaticojejunostomy being performed without a stent. Consistent with other reports[7,26,27], our team prefers to use EDS or IDS in patients with a soft pancreas and nondilated pancreatic duct that allows duct-to-mucosa anastomosis. In patients with a minimal or invisible Wirsung duct, the only possible procedure to perform is an invaginating pancreaticojejunostomy.

Currently, the best reconstructive technique after pancreaticoduodenectomy is the one showing the lowest rate of POPF. The great disparity in the results of the use, or not, of duct stents prevents a consensus being reached on the best method to use to prevent POPF after pancreaticoduodenectomy[1,3,4,6,7,28]. However, the controversy continues, and some studies have found similar rates of POPF development both with and without ductal stents[26,28,29].

Most studies that compared the use of IDS or EDS did not show significant differences between the groups regarding grade B/C POPFs, morbidity, and mortality[1,3,5,6]. Exceptionally, one series revealed a significantly higher rate of grade B/C POPF and overall complications in the IDS group than those in the EDS group[4]. A recent meta-analysis of seven series that compared IDS and EDS observed that the use of EDS reduced the total incidence of POPF, and the incidence in cases with a soft pancreas and DGE; no significant morbidity and mortality was found between the groups[15]. An older meta-analysis of five randomized and eight observational studies also showed that the use of EDS was associated with a significantly lower overall morbidity and risk for development of POPFs as well as overall morbidity and risk for development of POPFs in cases with normal pancreas[14]. In contrast, a randomized trial comparing the IDS and NDS procedures demonstrated that an IDS did not protect against POPF[27]. However, several other studies[2,8,30], as well as four meta-analyses[31-34], have compared the use of EDS vs NDS demonstrating that the use of EDS was associated with a significantly lower rate of POPF development. The most recent meta-analysis to compare the use of IDS or EDS vs NDS concluded that pancreatic duct stents did not reduce the risk for POPF, but in comparison to NDS, EDS was associated with a reduced POPF rate[34].

The theoretical advantage of using an IDS is that it diverts the pancreatic juice away from the pancreaticojejunostomy. In so doing, it allows the patency and decompression of the Wirsung duct and enables a period in which the anastomosis is protected and can heal. Thus, the intention of using an IDS is to decrease the rate of POPF development[1,27].

In comparison between IDS and EDS, we agree with other authors, that the use of an EDS has additional advantages, in that it enables a more precise placement for duct-to-mucosa anastomosis, complete drainage of the pancreatic juice to the outside of the body, radiological monitoring, decreased tension of the pancreaticojejunostomy and jejunal loop, and reduced contact between the pancreatic juice and pancreaticojejunostomy[8,30]. These advantages can facilitate the healing of the anastomosis and even prevent the development of POPF[4,31].

Multiple risk factors have been attributed to the development of POPFs, such as a soft consistency of the pancreas[3,4,6], a body mass index of > 23 kg/m²[2-4], a Wirsung duct diameter < 3 mm[2,4,30], use of a NDS technique[2,30], pancreatic steatosis[35], a significant amount of blood transfused[9], and the presence of a non-pancreatic disease[3,4]. The rate of grade C fistula development is between 4.4% and 9.9%[6,8,36] and has been related to a mortality rate of between 25.7% and 35%[36,37].

Analysis of our series revealed a significant difference in the rates of overall POPF development among the three groups. In particular, the rate of grade C POPF development was lower in the EDS group than in the IDS and NDS groups. Consistent with other studies[2,4,8,14,15,30], our multivariate analysis confirmed that the use of an EDS is a protective factor for grade B/C POPF, whereas a Wirsung duct diameter < 3 mm was a risk factor for the development of POPF. Other variables analyzed in our study, such as older age of patients (≥ 80 years), a patient with ASA grade III/IV, having preoperative pancreatitis, or having a soft pancreas were not demonstrated as risk factors for the development of grade B/C POPF.

Like other studies comparing stent and non-stent groups, we found not significantly differences in the rates of DGE and postoperative hemorrhage[6,14,26,31,34,38], and our rate of reoperation was higher in the NDS group[28,32,34] than in the other groups, although it did not reach statistical significance. Another ingenious technique, such as the binding pancreaticojejunostomy technique has significantly decreased the rate of POPF and hospital stay when compared with conventional NDS pancreaticojejunostomy[39].

Among several studies in which EDS and IDS[1,3-6], EDS and NDS[2,8,29,33], or IDS and NDS[27] were compared, only two studies reported significantly lower overall morbidity in patients treated with EDS[4,8]. Among the three groups, our patients showed similar rates of occurrence of Clavien-Dindo grade IV to V complications, lengths of stay in the intensive care unit and hospital, and overall 90-day morbidity. No comparative studies have shown significant differences in hospital mortality, the reported incidence of which ranges from 0% to 5%[1-6,8,30]. However, a recent meta-analysis concluded that the use of an EDS significantly reduces POPF-related mortality[11]. Similarly, our 90-day mortality was significantly lower in the EDS group than in the IDS and NDS groups, demonstrating in multivariate analysis that EDS was a protective factor. The presence of a Wirsung duct < 3 mm is a risk factor for POPF that subsequently contributes to increasing the mortality, as has been demonstrated in our 8 patients’ deaths for grade C POPF. Ninety-day mortality was significantly higher in IDS group and was attributed to both significantly lower caliber of Wirsung duct and higher incidence of grade C POPF. A low 5-year patient survival rate was found in patients treated with NDS; however, the difference among the groups was not significant.

No consensus has been reached about when to perform conservative therapy or relaparotomy for grade B/C POPF. The lack of consensus arises from variations in the clinical condition of patients and surgeon´s experience[40]. Thus, several procedures are performed with the intention of managing patients who present with B/C grade POPFs; these procedures include catheter drainage, relaparotomy and drainage of the abscess, disassembly of the pancreaticojejunostomy and closure of the jejunal loop and drainage, wirsungostomy maintained for 4 months, and partial or total removal of the remnant pancreas and placement of a drain near the resection margin of the pancreas[36,41-45]. Catheter drainage has been firstly proposed in cases of severe pancreatic fistula: It shows significantly lower morbimortality than reintervention as the primary step. Relaparotomy is reserved for patients with progressively worsening disease[43]. Wirsungostomy is a pancreas-preserving procedure that is recommended for grade C POPFs with uncontrolled hemorrhage or peritonitis[40-42]. Complications with wirsungostomy include dislodgment, occlusion, and rupture of the external tube[38]. Our patients with EDS did not show any discomfort, although 2 patients developed transient moderate abdominal pain coinciding with the catheter removal that needed hospitalization for observation for 24 hours.

Recent studies concluded that pancreas-preserving procedures seem preferable to total pancreatectomy in patients who have undergone relaparotomy as a necessary indication for either grade B/C[44] or grade C[45] POPF due to the high rate of mortality associated with total pancreatectomy. In our study, to manage grade C POPFs, all patients underwent relaparotomy, and washing and draining of the intra-abdominal abscess was the most common treatment (nine cases), followed by the exclusion of pancreaticojejunostomy (four cases), and removal of the remnant pancreas (three cases).

This study presents some limitations, such as its retrospective nature, small sample size, and being a single institution study, consequently, it is subject to bias. In addition, the preference of the surgeons for a particular reconstruction technique can be another limitation of this study.

CONCLUSION

Our study revealed a low rate of grade B/C POPF in the EDS group, although the remaining surgical complications did not show significant differences among the groups. Thus, the use of an EDS resulted in good outcomes. Multivariate analysis showed that the use of EDS was a protective factor for grade B/C POPF and 90-day mortality, whereas a Wirsung duct diameter < 3 mm was the only risk factor for grade B/C POPF and 90-day mortality. According to the advantages and our improved results with EDS compared with IDS and NDS techniques, we recommend the use of EDS for pancreaticojejunostomy after pancreaticoduodenectomy, except in patients with invisible or very small Wirsung duct diameters, where an invaginating technique should be performed. The NDS technique could be used in patients with larger Wirsung duct and a hard pancreas. Future prospective randomized studies, performed by experienced surgeons in high volume centers, comparing different technical Wirsung drainages are necessary to demonstrate the optimal technique after pancreaticoduodenectomy to reduce the rate of POPF.