Is Roux Loop a Solution to Delayed Gastric Emptying Following Pancreaticoduodenectomy? A Single Institute Analysis

Abstract

Delayed gastric emptying is one of the most common complications of pancreaticoduodenectomy (PD). It almost always results in delayed oral intake, prolonged hospital stays, and a delay in initiation of vital adjuvant treatment. A few earlier studies suggested that delayed gastric emptying (DGE) rates were better with the Roux-en Y reconstruction, but Indian literature regarding this is lacking. In our institutional study, we compared the traditional single-loop reconstruction (SL group), with the Roux loop reconstruction (RY group) following a subtotal stomach-preserving pancreaticoduodenectomy (SSPPD). A retrospective comparative study was conducted between the conventional single loop and a Roux-en-Y method of reconstruction following a subtotal stomach preserving pancreaticoduodenectomy (SSPPD). Sixty-three consecutive Whipple’s procedures were analyzed for multiple clinical parameters like removal of Ryles tube, tolerance of liquid diet and solid diet, delayed gastric emptying, duration of hospital stay and interval between surgery, and initiation of adjuvant treatment. Forty-one patients in the SL group were compared with 22 patients in the RY group. Ryles tube removal (POD 8.2 versus 2.25, p < 0.001), initiation of liquid diet (POD 8.43 versus 2.88, p < 0.001), post-operative hospital stay (13.5 days versus 9.63, p < 0.001), and interval between surgery and adjuvant treatment (37.75 days versus 28.88 days, p < 0.002) were all in favor of the RY group. The delayed gastric emptying was also found to be significantly better in the Roux-en-Y surgery group (p < 0.001). The Roux loop reconstruction following a stomach-preserving pancreaticoduodenectomy (SSPPD) is superior to single-loop reconstruction with respect to delayed gastric emptying. The lesser duration of hospital stay and early initiation of adjuvant therapy are an additional benefit of the Roux loop reconstruction.

Introduction

In addition to pancreatic fistula and post-operative hemorrhage, delayed gastric emptying (DGE) is one of the most common post-operative complications after pancreatic surgery, occurring in 19–57% of patients, based on the definition used and the type of reconstruction [1, 2]. DGE is almost never fatal, but adds to the morbidity of an already complex surgical procedure, resulting in prolonged hospital stay, and increasing hospital costs. The mechanism of DGE after a pancreaticoduodenectomy is however poorly understood and is best described as multifactorial [3]. The disruption of neuro-humoral pathways remains the common factor in all patients who undergo a pancreaticoduodenectomy that contributes to the occurrence of primary DGE. Gastric motility is a coordinated process of myoelectrical activities involving the gastric reservoir, antral pump, pylorus, and duodenum. These activities are controlled by both enterogastric neural reflexes and hormones released from the intestines that reach the stomach via the systemic circulation. Gastric emptying occurs during two distinct periods: the post-prandial motility period that starts after each meal and the subsequent interdigestive period, during which the gastrointestinal tract produces rhythmically recurring cycles of activity that involve four phases. The first and longest is a period of near quiescence, followed by a period of small-amplitude contractions of irregular frequency known as phase II, and then a burst of high-amplitude propulsive contractions (phase III), which move down the intestine and terminate in the distal small intestine; phase IV is sometimes used to describe the decline of activity back to baseline [4]. Phase III, known as the migrating motor complex (MMC), consists of forceful peristaltic waves that originate simultaneously in the stomach and the duodenum and propagate along the entire length of the small intestine. The main function of the MMC is to empty the stomach and small intestine from chyme residues, mucous, and bacteria [5–7]. As the MMC activity reaches the small intestine, the velocity of propagation of the peristaltic waves declines from the proximal jejunum to the distal small intestine, which is most likely related to a decreased number of motilin receptors along the gastrointestinal tract [5]. When the head of the pancreas is resected, the surgical procedure interrupts the intrinsic circuitry and removes the duodenum where the pacemaker controlling the motor activity of the bowel is located [8]. The surgical procedure also disturbs circulating hormones. Tanaka et al. reported that resection of the canine duodenum, which is known to be the principle region for motilin secretion, impaired the cyclic increase in plasma motilin concentrations and abolished the occurrence of gastric phase III activity [9].

Other operative factors may also impact the rate of DGE, such as the method of reconstruction of gastric drainage (antecolic versus retrocolic) [10, 11] and other specific techniques (Billroth I versus II type reconstruction or mechanical dilatation of the pylorus) [12–15]. However, there is evidence from a randomized trial to prove that the incidence of DGE is comparable between a standard pancreaticoduodenectomy and a pylorus-preserving pancreaticoduodenectomy (22% and 23%, respectively) [16].

In this study, we compared the DGE rates between 2 techniques of reconstruction following a PD, a standard single jejunal loop, and a Roux-en-Y jejunal loop, following a subtotal stomach-preserving pancreaticoduodenectomy.

Aim

The primary objective of the study was to evaluate the incidence and grade of delayed gastric emptying in each technique.

The secondary objective was to analyze the duration of post-operative hospital stay and the interval between surgery and initiation of adjuvant therapy between the two techniques.

Methods

A retrospective observational study was conducted between January 2015 and July 2020. Sixty-three consecutive subtotal stomach-preserving pancreaticoduodenectomy (SSPPD) procedures done by a single surgical team from the department of Surgical Oncology were analyzed. SSPPD is a pancreaticoduodenectomy wherein the stomach is transected 2 cm above the pyloric ring, preserving nearly 90% of the stomach [3].

From January 2015 to September 2018, the reconstruction method following a SSPPD in our department was a conventional single loop method, wherein the pancreaticojejunostomy, hepaticojejunostomy, and gastrojejunostomy were all done to a single jejunal loop (SL group) Fig. 1. Although the pancreatic fistula (PF) rates were acceptable, the rates of delayed gastric emptying (DGE) were very high, resulting in prolonged hospital stay, delay in initiation of adjuvant treatment, and high rates of bile reflux gastritis in the early post-operative and follow-up period.

Fig. 1.

Diagrammatic representation of the traditional single-loop technique (SL) on the left, and the Roux-en-Y technique (RY) on the right

Based on our vast experience with Roux-en-Y reconstruction following distal gastrectomy, wherein we initiate early oral intake (as early as post-operative day 1) and all bile reflux is eliminated, we modified our technique of reconstruction following SSPPD, wherein 15 cm after HJ, the jejunum is transected and a Roux gastrojejunostomy is done, with the jejunojejunostomy being 45 cm from the gastrojejunostomy (RY group) Figs. 1, 2, and 3. The gastrojejunostomy in either technique was always placed as vertically as possible, while preserving the antrum.

Fig. 2.

The point of transection of the jejunum exactly 15 cm after the HJ

Fig. 3.

Jejunojejunostomy (JJ) 45 cm from gastrojejunostomy (GJ)

Inclusion

All patients were subjected to SSPPD procedure, irrespective of tumor origin and histology. No cases were excluded.

The following parameters were defined before initiation of the study:

1. Criteria for a Ryles tube removal:

• i.Bile output through a Ryles tube < 200 ml in 24 h.
• ii.No abdominal discomfort/distension.

• 2.Criteria for a Ryles tube reinsertion:

• i.Greater than 2 episodes of bilious vomiting in 24 h.
• ii.Abdominal distension.

• 3.Patients were considered to be taking adequate oral diet if there was no need for supplementing by feeding jejunostomy in order to meet daily recommended calorie intake.
• 4.Criteria for DGE as per the International Study Group of Pancreatic Surgery (ISGPS) [17] (Table 1).

• 5.Patients were considered fit for adjuvant treatment if:

• i.ECOG 1
• ii.Meeting daily calorie requirements through per oral route only.

• 6.Criteria for pancreatic fistula as per the International Study Group of Pancreatic Surgery (ISGPS) (Table 2) [18]

In this study patients with pancreatic fistula grades B and above were excluded from final analysis, as increasing pancreatic fistula grades causes more severe delayed gastric emptying [19].

• 7.Prokinetic usage

  Prokinetic agents were used only if patient was in DGE (criteria as mentioned above).

Table 1.

The International Study Group of Pancreatic Surgery (ISGPS) criteria for DGE

DGE grade	NGT required	Unable to tolerate solid oral intake by POD	Vomiting/gastric distension	Use of prokinetics
A	4–7 days or reinsertion > POD 3	7	±	±
B	8–14 days or reinsertion > POD 7	14	+	+
C	> 14 days or reinsertion > POD 14	21	+	+

Table 2.

The revised 2016 ISGPS classification and grading of POPF

ISGPS, International Study Group on Pancreatic Surgery; POPF, post-operative pancreatic fistula. *A clinically relevant POPF is defined as a drain output of any measurable volume of fluid with amylase level greater than 3 times the upper institutional normal serum amylase level, associated with a clinically relevant development/condition related directly to the POPF. ^#Suggests prolongation of hospital or ICU stay, and includes use of therapeutic agents specifically employed for fistula management or its consequences (of these: somatostatin analogues, TPN/TEN, blood product transfusion or other medications). ^{^}Post-operative organ failure is defined as the need for re-intubation, hemodialysis, and/or inotropic agents > 24 h for respiratory, renal, or cardiac insufficiency, respectively.

Statistical Analysis

Statistical analysis was done using IBM SPSS Statistics 22.0 version. Mann-Whitney U test was applied to find the mean comparisons and significance between two independent techniques in different clinical variables. Pearson’s chi-squared test was applied to find the grade comparison between technique 1 and technique 2. If the calculated p value is below the threshold chosen for statistical significance (usually the 0.05 level or 5% level), it was considered to be statistically significant (Fig. 4).

Fig. 4.

Distribution of cases in each group

In the final analysis, 28 from SL group and 16 from RY group were included.

The following parameters were assessed in each arm:

1. Ryles tube removal

2. Initiation of oral liquids

3. Initiation of normal diet

4. Grade of DGE

5. Post-operative hospital stay

6. Duration between surgery and adjuvant therapy

Observation and Results

The study population consisted of 62% males and 38% females, with the mean age being 50.7 years. The most common site of the primary lesion was periampullary (n = 42, 66.6%), followed by head of the pancreas (n = 10, 15%), duodenum (n = 4, 6.3%), colon (n = 3, 4.7%), and common bile duct (n = 2, 3.1%). There was 1 dual malignancy (gall bladder with synchronous distal CBD malignancy), and 1 SSPPD done for a primary gall bladder malignancy. Among the 63 patients, for 4 patients (7%), the final histology turned out to be non-malignant, whereas the histology of the remaining 59 patients (93%) were all malignant.

According to the ISGPS grading of post-operative pancreatic fistula (POPF), 17 patients had no leak (30%), and there were 29 patients (52%) with biochemical leak. The clinically significant POPF was 16.3% (pancreatic fistula grade B in 7 patients (12.7%) and pancreatic fistula grade C in 2 patients (3.6%)).

The mean duration of the Ryles tube drainage in the SL group was 8.2 days versus 2.2 days in the RY group (p < 0.001). The mean time taken to tolerate oral liquids and soft diet were 8.43 days and 10.64 days in the SL group versus 2.88 days and 5.81 days in the RY group, respectively, both being statistically significant (p < 0.001). Delayed gastric emptying was absent in only 1 patient (3.6%) in the SL group, with grade A DGE seen in 12 patients (42.9%), grade B DGE in 14 patients (50%), and 1 patient had grade C DGE (3.6%). In the RY group, DGE was absent in 14 patients (87.5%), whereas grade A DGE was seen in 2 (12.5%) and no grade B and grade C DGE was observed. The p values for DGE between the 2 groups was statistically significant (p < 0.001).

The secondary objectives of the study, i.e., post-operative hospital stay were also statistically significant in favor of the RY group (mean stay in the SL group, 13.5 days versus 9.63 days in the RY group, p < 0.001). The interval between surgery and adjuvant treatment in the SL group was 37.7 days versus 28.8 days in the RY group, p < 0.002 in favor of the RY group (Tables 3 and 4).

Table 3.

Mean comparison of clinical variables between technique 1 (SL) and technique 2 (RY)

Clinical variables	Techniques	Min	Max	Mean	SD	Difference mean ± SD	P value
RT removal on post-op (days)	SL	5	16	8.21	2.81	5.96 ± 1.68	< 0.001 significant
	RY	1	5	2.25	1.13
Oral SIPS (POD) (days)	SL	5	16	8.43	2.81	5.55 ± 1.55	< 0.001 significant
	RY	1	5	2.88	1.26
Soft diet (POD) (days)	SL	6	18	10.64	3.21	4.83 ± 1.88	< 0.001 significant
	RY	4	9	5.81	1.33
Number of days in hospital	SL	9	23	13.50	3.89	3.87 ± 3.17	< 0.001 significant
	RY	8	10	9.63	0.72
Interval between surgery and ADJ RX (days)	SL	25	60	37.75	9.77	8.87 ± 6.19	0.002 significant
	RY	25	35	28.88	3.58

Table 4.

Comparison of delayed gastric emptying (DGE) between technique 1 (SL) and technique 2 (RY)

Grades	SL group	RY group	Chi-square value	P value
	n (%)	n (%)
Absent	1 (3.6)	14 (87.5)	32.559	< 0.001 significant
Grade A	12 (42.9)	2 (12.5)
Grade B	14 (50.0)	0 (0)
Grade C	1 (3.6)	0 (0)
Total	28 (100.0)	16 (100.0)

Statistical analysis: chi-square test. Statistically significant if P < 0.05

Discussion

Delayed gastric emptying has been identified as a significant morbidity of the pancreaticoduodenectomy procedure and several techniques of reconstruction have been tried to minimize it [20–24]. With the surgical procedure now being considered to be relatively safe in high-volume centers, focus has shifted to “fast tracking” the procedure, and discharging the patients at the earliest in order to reduce hospital costs [25–27]. Murakami et al. in 2007 conducted a study to identify a preferable procedure reducing the incidence of delayed gastric emptying (DGE) after pylorus-preserving pancreatoduodenectomy (PPPD) [20]. A retrocolic Billroth I type reconstruction (B-I group) and an antecolic Roux-en-Y type reconstruction (RY group) were performed for 54 and 78 patients after PPPD, respectively. Only the type of reconstruction (P < 0.001) was identified as an independent factor, which was associated with DGE. Similar to our study, in the study by Murakami et al., they changed their technique to Roux-en-Y in 2001 as against the Billroth I that they were following up until then, probably due to increased delayed gastric emptying. Our results with respect to DGE mirror theirs, with the RY group doing significantly better than the SL group (p < 0.001). The only differences between our study and the study by Murakami et al. was the pancreatic anastomosis (PJ versus PG) and resection technique (SSPPD versus PPPD). The DGE rates with a standard pancreaticoduodenectomy and a pylorus-preserving pancreaticoduodenectomy have been proven to be the same in a randomized trial by Kran et al. [16]. In a randomized study by Shimoda et al. in 2013, 101 patients who underwent SSPPD for pancreatic head or periampullary diseases were randomly allocated to the Billroth II reconstruction (n = 52) and RY reconstruction (n = 49) groups [20]. The primary endpoint was incidence of DGE, and the study concluded that the incidence of DGE after SSPPD can be decreased by using Billroth II rather than RY reconstruction for gastrojejunostomy. In this study, by Shimoda et al., the distance of the gastrojejunostomy from the hepaticojejunostomy was 40 cm in either technique and there was routine nasojejunal tube drainage until 5–7 days, which in effect makes all cases to have at least a grade A DGE. The maximal concentration of motilin receptors are in the proximal 30–40 cm of jejunum [22]. The construction of a gastrojejunostomy beyond this will lead to delayed gastric emptying irrespective of the technique, as the velocity of propagation of MMC activity decline from the proximal to the distal small intestine due to decreasing concentration of motilin receptors. In the RY technique used in our study, the jejunal loop used for the GJ is in the area of maximal motilin receptors. Another meta-analysis by Yang et al. on the effect of the Billroth II or Roux-en-Y reconstruction for the gastrojejunostomy after pancreaticoduodenectomy revealed that the incidence of DGE (grades B and C) after PD can be decreased by using the Billroth II rather than Roux-en-Y reconstruction [23]. In the meta-analysis by Yang et al., there was no exclusion of grade B and C pancreatic fistula, bile leak etc., which itself causes DGE. And also, the conclusion was a subgroup analysis for grade B and C DGE only, with the initial conclusion being that there is no significant difference in DGE (grades A, B, and C) between the Billroth II and RY group.

In the study by Barakat et al., DGE outcomes between patients who underwent pancreaticoduodenectomy with resection of the pyloric ring followed by proximal Roux-en-Y gastrojejunal anastomosis (group I, n = 90) and patients who underwent standard pancreaticoduodenectomy with the orthotopic reconstruction technique (group II, n = 118) were compared [3]. The overall and clinically relevant rates of DGE were significantly lower in the Roux-en-Y group than in the standard reconstruction (10 and 2.2% versus 57 and 24%, respectively; p < 0.05). Length of hospital stay as a result of DGE was shorter in the Roux-en-Y group. The only difference between the study by Barakat et al. and our study is the sequence of anastomosis. In the Barakat et al. study, after resection, the gastrojejunostomy was done first and the jejunum was transected 35–40 cm distal to the GJ and the pancreaticojejunostomy and hepaticojejunostomy done later (Roux-en-Y). Very similar to our study, here too, the emphasis was on use of the proximal 30–40 cm of jejunum for the gastrojejunostomy, where there is the highest concentration of motilin receptors.

The 2018 PAUDA trial, a randomized controlled trial, to compare Billroth II (single loop) and Roux-en-Y (double loop) after pancreatoduodenectomy to determine whether Roux-en-Y reconstruction is associated with a lower incidence of DGE [24]. The study concluded that the incidence and severity of DGE does not differ between single- or double-loop gastro-enteric anastomosis performed after pancreatoduodenectomy. In the Roux-en-Y group, the jejunum was transected 60 cm following the pancreaticojejunostomy and the hepaticojejunostomy, thus losing a significant amount of motilin receptor-rich jejunum segment. In our study, the jejunum is transected 15 cm following the pancreaticojejunostomy and the hepaticojejunostomy (Fig. 2 above), thus retaining a significant amount of proximal jejunum. This specific difference could be the reason for the better results in the RY group in our study.

Conclusion

The Roux-en-Y reconstruction, maximizing the use of the proximal jejunum which is rich in motilin receptors, helps in decreasing the incidence and severity of delayed gastric emptying following pancreaticoduodenectomy. The lesser duration of hospital stay and early initiation of adjuvant therapy is an additional benefit with the Roux loop reconstruction. However, a larger prospective study is required to address this question.

Drawbacks

This is a retrospective study with a small sample size, and hence, the possibility of a type II error cannot be excluded.