Novel choledochojejunostomy technique “T‐shaped anastomosis” for preventing the development of postoperative cholangitis in pancreatoduodenectomy: A propensity score matching analysis

Abstract Background There have been few studies of countermeasures against postoperative cholangitis, a serious complication after pancreaticoduodenectomy (PD) that impairs quality of life. Objective To evaluate our recently developed, novel method of choledochojejunostomy with a larger anastomotic diameter, the “T‐shaped anastomosis.” Methods The study included 261 cases of PD. The T‐shaped choledochojejunostomy technique was performed with an additional incision for a distance greater than half the diameter of the bile duct at the anterior wall of the bile duct and the anterior wall of the elevated jejunum. To compensate for potential confounding biases between the standard anastomosis group (n = 206) and the T‐shaped anastomosis group (n = 55), we performed propensity score matching (PSM). The primary endpoint was the incidence of medium‐term postoperative cholangitis adjusted for PSM. Results In the PSM analysis, 54 patients in each group were matched, and the median bile duct diameter measured by preoperative CT was 8.8 mm versus 9.3 mm, the rate of preoperative biliary drainage was 31% versus 37%, the incidence of cholangitis within 1 month before surgery was 9% versus 13%, and the incidence of postoperative bile leakage was 2% versus 2%, with no significant differences. The incidence of medium‐term postoperative cholangitis was 15% versus 4%, and multivariate logistic regression revealed that T‐shaped choledochojejunostomy was an independent predictor of a reduced incidence of cholangitis (odds ratio, 0.17, 95% CI 0.02–0.81; p = 0.024). Conclusions The T‐shaped choledochojejunostomy technique was shown to be effective with a significant reduction in the incidence of medium‐term postoperative cholangitis. Clinical trial identification: UMIN000050990.


| INTRODUC TI ON
2][3][4][5] Pancreatic cancer accounts for the majority of diseases requiring pancreatectomy.7][8][9] Therefore, we need to address mid-to long-term complications after PD for stable and reliable continuation of postoperative chemotherapy.However, various complications, such as pancreatic fistula and delayed gastric emptying, can occur in 10%-23% of patients after PD 10,11 and overwhelmingly cause the most frequent and troublesome aspects of all abdominal surgeries.One of the main complications occurring in the medium term after PD is cholangitis.3][14] This is a complication that requires multiple hospitalizations and is sometimes fatal.Repeated postoperative cholangitis may impair the continuation of postoperative chemotherapy, possibly resulting in worsening of the prognosis of cancer.There have been several reports on risk factors for cholangitis after PD.
6][17][18] However, these factors are often not addressed by preoperative intervention.In this era of multidisciplinary treatment, we need specific techniques to decrease postoperative cholangitis; however, no novel anastomotic methods other than end-to-side choledochojejunostomy by interrupted suture have been reported.
We developed a novel method of choledochojejunostomy to enlarge the anastomotic diameter.In this study, we compared its usefulness for preventing postoperative cholangitis, as well as the incidence of postoperative complications, with those of conventional anastomosis methods using propensity score matching.

| Patients
A total of 272 patients underwent PD as an elective operation at Toyama University Hospital from April 2017 to June 2022.Among them, six patients were excluded because of hilar recurrence postoperatively that was diagnosed by computed tomography (CT) or magnetic resonance imaging (MRI) during postoperative followup.In addition, we excluded five patients who had postoperative complications of Clavien-Dindo Grade IVb or higher because of possible effects on the medium-and long-term postoperative course.
Note that none of the complications in these five excluded patients were related to the choledochojejunostomy.Finally, 261 patients were included in this study, none of whom underwent R2 surgery.guidelines. 19All procedures in this study were performed in accordance with the guidelines of the Declaration of Helsinki.Written informed consent for treatment was obtained from each patient prior to the start of treatment, and consent for the use of data for research was obtained on an opt-out basis.

| Surgical procedure
We performed subtotal stomach-preserving PD with a midline abdominal incision from the xiphoid to below the navel as a standard procedure.Concurrent superior mesenteric vein (SMV)/portal vein (PV) resection was performed if necessary. 20,21The bile duct was temporarily clamped with a bulldog vascular clamp to prevent potential intraabdominal infection from the bile juice.The bile duct was subsequently cut with scissors, and its stump was measured.
Bile juice was collected for intraoperative culture.We performed the modified Child reconstruction.The jejunal limb was brought up to the pancreatic and bile duct stumps via a retrocolic route.Pancreatojejunostomy was performed by the modified Blumgart method in all patients. 22In the soft pancreas, the stent tube was inserted into the main pancreatic duct, and the other side of the tube exited the body.Gastrojejunostomy was performed by hand stitching using a 3-0 synthetic absorbable blade filament and 4-0 synthetic absorbable monofilament or a mechanical anastomosis with a surgical stapling device, and Braun anastomosis was added in all patients.
The drainage tubes were placed on the ventral and dorsal sides of the pancreatojejunostomy and the dorsal side of the choledochojejunostomy.All patients received prophylactic antibiotics for 3 days and a proton pump inhibitor to prevent stress peptic ulcers. 23 usually used cefazoline as the prophylactic antibiotic; however, the selection of antibiotics was changed according to preoperative bile culture if the endoscopic nasobiliary drainage tube was placed preoperatively.

| Choledochojejunostomy
Choledochojejunostomy was performed with an interrupted suture using 5-0 synthetic absorbable monofilament sutures in an T-shaped anastomosis was exclusively performed starting in March 2021.The method of T-shaped anastomosis was as follows (Figure 1).First, the diameter of the resection stump of the bile duct is measured.The dorsal wall is sutured with a 5-0 synthetic absorbable monofilament interrupted suture as in a conventional choledochojejunostomy.Once the dorsal wall is sutured, a slit of length greater than half the diameter of the bile duct is added to the anterior median surface of the hepatic duct and the elevated jejunum.An interrupted suture is made, including the apex of the incision and the two points on either side of the incision.The number of interrupted sutures depends on the diameter of the bile duct, but usually 11 sutures are needed for the anterior wall.
Except for patients for whom the bile duct was cut to the very edge of the porta hepatis, preventing T-shaped anastomosis, and those for whom the bile duct was formed with two holes, T-shaped anastomosis has been performed for all patients since the inception of the technique whenever technically possible, even in cases of preoperative cholangitis or biliary drainage.

| Postoperative complications
All preoperative complications were graded according to the Clavien-Dindo classification. 24Postoperative cholangitis was diagnosed when the criteria for suspicion or definition were met according to the Tokyo Guidelines 2018, specifically fever, laboratory data, jaundice, biliary dilatation, and imaging such as biliary stricture. 25ver abscesses and biliary stones were diagnosed by either CT or MRI.Benign biliary stenosis was defined as suspected stenosis on CT or MRCP and endoscopically confirmed stenosis.The severity of cholangitis was also diagnosed according to the severity criteria of the Tokyo Guidelines 2018. 25In this study, we categorized cholangitis as short-(within 30 days postoperatively) and medium-term (31 days to 18 months postoperatively).

| Evaluated factors
The primary endpoint of this study was the incidence of mediumterm postoperative cholangitis adjusted for propensity score.
Clinical data were collected retrospectively for all patients and included patient demographics (age, sex, height, weight, body mass index [BMI]), preoperative and postoperative diagnosis, pathological examination, perioperative clinical information, and complications.Body temperature, white blood cell counts, Creactive protein (CRP), aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), γ-glutamyl transpeptidase (γ-GT), and total bilirubin (T-Bil) were measured before and after surgery and when cholangitis was suspected.We measured the common hepatic duct diameter by using a CT imaging scale before surgery.The postoperative observational period was terminated at 18 months (540 days) since medium-term postoperative cholangitis as the primary endpoint was defined as up to 18 months.With regard to postoperative follow-up, postoperative patients with malignant tumors were seen at the hospital once a month, and patients with low-grade tumors were seen every 3 months.When patients were being followed at other hospitals, the physicians at those hospitals were asked to carefully check for the presence of cholangitis.

| Statistical analysis
A biostatistician (K.M.) was responsible for the statistical analysis.
Differences in the nominal data between the two groups were examined using the chi-squared test or Fisher's exact test when the expected value was <5.Differences in quantitative variables were evaluated using Student's t test or the Mann-Whitney U test if the distribution was abnormal.All cutoff values for the development of medium-term postoperative cholangitis were determined using the Youden index.
To reduce the effect of selection bias, propensity score analysis was performed using the following covariates: age, BMI, previous abdominal surgery, common hepatic duct diameter on preoperative CT, primary disease (pancreatic cancer/other), preoperative biliary drainage, and preoperative cholangitis.6][17][18] The Kaplan-Meier method was used to compare the accumulated incidence of medium-term postoperative cholangitis between the standard anastomosis group and the T-shaped anastomosis group.
We used univariate and multivariate logistic regression to generate odds ratios (ORs), including 95% CIs, for clinical factors that would predict medium-term postoperative cholangitis.Variables included in the multivariate models had a p value of 0.1 or less in the univariate analysis.In addition to 1:1 matching of the standard anastomosis group and the T-shaped anastomosis group, a 2:1-matching, inverse-probability-of-treatment-weighting analytical method was performed as a sensitivity analysis.
A p value <0.05 was considered to indicate statistical significance.All statistical analyses were performed using JMP statistical software (version 15.0; SAS Institute).

| Characteristics and preoperative status of all unmatched patients
The characteristics, preoperative status, and preoperative blood test results of all patients (n = 261) are summarized in Table 1.In the comparison of the standard anastomosis group (n = 206) versus the T-shaped anastomosis group (n = 55) before propensity score matching (PSM), the median age was 72 versus 73 years, the male/female ratio was 127/79 versus 33/22, the median BMI was 22.2 versus 21.7, the median common hepatic duct diameter measured by preoperative CT was 8.9 mm versus 9.2 mm, the rate of preoperative biliary drainage was 37% versus 36%, and the incidence of cholangitis within 1 month before surgery was 10% versus 15%, with no significant differences.The two groups did not differ significantly in terms of preoperative comorbidities, including a history of diabetes mellitus.The preoperative albumin level and total lymphocyte count were also similar in both groups.
There was a significant difference in the ratio of pancreatic cancer/other as the primary disease: 68/138 versus 33/22 (p < 0.001); therefore, there was a significant difference between the two groups in the percentage of patients who received preoperative treatment (41/165 vs. 33/22, p < 0.001).

| Perioperative details and postoperative complications in unmatched patients
The proportion of patients who underwent laparoscopic surgery was 0.5% in the standard anastomosis group and 1.8% in the T-shaped anastomosis group, with no significant differences.Because pancreatic cancer was more common in the T group of unmatched patients, the median main pancreatic duct diameter (3 [1-20]   b According to Tokyo Guidelines 18.

| Analysis of propensity score-matched patients
To reduce the impact of selection bias, PSM was performed using seven selected baseline characteristics, and 54 patients in each group were matched (Figure 2).In Table 2, the primary disease ratio (pancreatic cancer/other) was 34/20 versus 32/22, the common hepatic duct diameter measured by preoperative CT was 8.8 mm versus 9.3 mm, the rate of preoperative biliary drainage was 31% versus 37%, and the incidence of cholangitis within 1 month before surgery was 9% versus 13%, with no significant differences between the standard anastomosis group and the T-shaped anastomosis group.Significant differences in operative time and blood loss disappeared after PSM.The frequency of intraoperative biliary drainage tube placement was similar.The incidence of postoperative bile leakage was 1.9% versus 1.9%, the incidence of postoperative complications (Clavien-Dindo grade IIIa or higher) was 33% versus 30%, and the median length of hospital stay was 24.5 versus 19 days, with no significant differences.The postoperative observational period was also not significantly different between the two groups (395.

| Factors predicting medium-term postoperative cholangitis
The factors predicting medium-term postoperative cholangitis in propensity score-matched patients are shown in Table 3. Univariate logistic regression with propensity scores showed that higher postoperative peak values of AST and ALT and longer postoperative hospital stay were significantly associated with the onset of medium-term postoperative cholangitis and that medium-term postoperative cholangitis was significantly less frequent in the T-shaped anastomosis group than in the standard anastomosis group.Multivariate analysis showed that T-shaped choledochojejunostomy was an independent predictor of a reduced incidence of cholangitis (odds ratio, 0.17; 95% confidence interval, 0.02-0.81;p = 0.024) (Table 4).

| Sensitivity analysis
In Table 5, logistic regression between the standard anastomosis group (n = 106) and the T-shaped anastomosis group (n = 53) using 2:1 PSM also showed that the incidence of medium-term postoperative cholangitis was significantly lower in the T-shaped anastomosis group (odds ratio, 0.26, 95% confidence interval 0.04-0.97;p = 0.045).In logistic regression of the inverse probability of treatment weighting, T-shaped anastomosis significantly reduced the frequency of medium-term cholangitis (odds ratio, 0.16, 95% confidence interval 0.06-0.35;p < 0.001).

| DISCUSS ION
Postoperative cholangitis after PD is an extremely important and sometimes fatal complication. 12,26Based on the TG13 guidelines,  in the left wall of the common hepatic duct, as a cholangioplasty to prevent postoperative cholangitis. 28They made a 5-10-mm incision in the left side of the common hepatic duct wall with an electrical scalpel, taking care not to injure the hepatic artery or the peribiliary vascular plexus.The reason for placing the incision on the left side of the hepatic duct is the anatomical factor that hepatolithiasis is more common in the left lobe. 29The left hepatic duct and the common hepatic duct meet at a sharp angle, which causes bile stasis, and by incising the bile duct wall in the area and widening the diameter of the bile duct, the biliary flow through the anastomotic portion to the intestine is smooth, preventing the occurrence of postoperative cholangitis. 30 this study, we developed a novel anastomotic tech- anastomosis, and those for whom the bile duct was formed with two holes.It cannot be denied that this may have resulted in a selection bias and reduced the frequency of cholangitis for T-shaped anastomosis.Fourth, long-term results, e.g., the incidence of cholangitis over a 5-year period, need to be examined.Therefore, more studies with a higher level of evidence, such as multicenter randomized trials, are needed in the future.
The study was reviewed and approved (ref.No. R2021142) by the institutional review board and complied with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)

K
E Y W O R D S cholangitis, choledochojejunostomy, pancreatoduodenectomy, postoperative complication, T-shaped anastomosis end-to-side fashion.Intraoperative biliary drainage tubes were not routinely used, and biliary drainage tubes were inserted only if there was postoperative concern of biliary leakage during choledochojejunostomy.
vs. 4[1-25] mm, respectively; p = 0.007) and median operative time (509.5 [230-945]   vs. 572 [220-804] min, respectively; p = 0.033) were greater in the T-shaped anastomosis group.However, the frequency of intraoperative blood transfusions was not significantly different between the two groups.Intraoperative retrograde transhepatic biliary drainage tube placement was 8.3% versus 3.8%, with no significant differences.Postoperative peak values of hepatobiliary enzymes such as AST, ALT, ALP, γ-GT, and T-Bil after surgery were not significant in either group.With regard to postoperative complications, the incidence of bile leakage was 3.9% versus 1.8%, that of clinically relevant pancreatic fistula was 25% versus 18%, that of delayed gastric emptying was 5.3% versus 10.9%, and the median length of hospital stay was 25 versus 19 days, with no significant differences.The median postoperative observational period significantly differed between the two groups (531.5 vs. 377 days, respectively; p = 0.008) since the T-shaped anastomosis was performed more recently.The incidence of short-term postoperative cholangitis was 1.9% in the standard anastomosis group and 1.8% in the T-shaped anastomosis group, and the incidence values of medium-term postoperative cholangitis in the standard and Tshaped anastomosis groups were 13% and 3.6%, respectively.The average number of times cholangitis occurred was 2 (1-6) versus 1.5 (1, 2), respectively.The severity of cholangitis was mild in 15 cases, moderate in seven cases, and severe in five cases in the standard anastomosis group and moderate in two cases in the Tshaped anastomosis group.All 15 patients with mild cholangitis in this study had a spike fever of 38°C or higher, and all required hospitalization for treatment.In addition, some patients with malignant tumors had to discontinue adjuvant chemotherapy because of cholangitis.Although cases of hilar recurrence were excluded from the present study, no cases of cholangitis developed in any patients with hilar recurrence.In addition, there were no cases of postoperative cholangitis in patients with early recurrence in the liver.Postoperative benign biliary stricture occurred in nine of 27 (33%) patients in the standard anastomosis group and zero of two patients in the T-shaped anastomosis group.Endoscopic interventions were performed in 13 of 27 (48%) patients with cholangitis in the standard anastomosis group.Of the 13 patients who underwent endoscopic intervention, only three cases did not require endoscopic intervention because there was no anastomotic stricture or intrahepatic calculus.In the other 10 cases, all patients underwent endoscopic intervention, of which five patients underwent endoscopic biliary drainage and eight patients underwent endoscopic balloon dilation.Intrahepatic stones were treated with lithotripsy in four cases, three of which required more than two lithotripsy procedures.In endoscopic intervention, there was overlap in several cases.The average number of endoscopic interventions was 2.7, with a maximum of seven interventions needed.Two patients with cholangitis in the T-shaped anastomosis group did not undergo endoscopy.TA B L E 1 Comparison of clinical characteristics between the standard anastomosis group and the T-shaped anastomosis groups before propensity score matching.
5 vs. 370.5 days, respectively; p = 0.476); however, the incidence of medium-term postoperative cholangitis was significantly lower in the T-shaped anastomosis group than in the standard anastomosis group (3.7% vs. 14.8%,respectively; p = 0.046).The cumulative incidence analysis similarly showed significantly lower cholangitis in the Tshaped anastomosis group (Figure 3).The mean number of times that cholangitis recurred was 1.6 (1-3) among the eight patients in the standard anastomosis group and 1.5 (1, 2) among the two patients in the T-shaped anastomosis group.Endoscopy intervention was performed in four of the eight cholangitis patients in the standard anastomosis group, all of whom also underwent balloon dilatation and other interventions.The mean number of interventions was two (1-4).No endoscopy was performed in two patients in the T-shaped anastomosis group.After PSM, benign biliary stenosis and endoscopic intervention were significantly less common in the T-shaped anastomosis group.

TA B L E 2
(Continued)  PD remains an invasive procedure for the patient, although in recent years, the intervention has become more universal than before.Since a certain number of patients with postoperative cholangitis require surgery or develop severe organ failure (Clavien-Dindo classification ≥ Grade IIIa),28 the establishment of cholangitis prevention methods is an urgent issue.In many reports, researchers discuss predictive factors of the development of cholangitis after PD; however, its prevention is the focus of only a few reports.To prevent postoperative cholangitis, it is important to ensure the passage of the bile duct jejunal anastomosis and elevated jejunum and to prevent the formation of sludge or stones due to bile stenosis.To avoid anastomotic stenosis, it is important to anastomose healthy bile duct tissue with adequate blood flow and without tension.Another option is to enlarge the anastomotic opening.Hiyoshi et al. reported the development of "hepaticoplasty," an incision nique called "T-shaped anastomosis" for the prevention of postoperative cholangitis, which allows all surgeons to easily and reliably ensure a wide anastomotic diameter.T-shaped anastomosis was performed after March 2021, and PD for pancreatic cancer tended to be more common during this period because there were fewer pancreatectomies for low-grade malignant disease due to the COVID-19 pandemic.To objectively evaluate the efficacy of T-shaped anastomosis as an anastomotic technique, we used PSM to equalize the preoperative factors of the control and comparison groups as much as possible.The criteria for cholangitis also complied with the latest TG18 guidelines.Since there are no references in previous reports regarding when to evaluate cholangitis in the postoperative period, the timing was defined as within 18 months postoperatively in this study.The incidence of medium-term postoperative cholangitis in the T-shaped anastomosis group was significantly lower (4%) and was suggested to be an independent factor in the prevention of postoperative cholangitis.Although the number of stitches required was increased compared to the standard anastomosis, the operative time after PSM was comparable, and the associated complications, such as bile leakage, were not significantly different.This is the first study to demonstrate the usefulness of novel choledochojejunostomy by PSM.Although the results of this study are interesting, our analysis had several limitations.First, this was a retrospective study based on single-center data, leaving open the possibility that confounding factors and selection bias were included.Second, the limited sample size may have influenced our results.Third, as noted in the Methods section, T-shaped anastomosis has been performed in all patients since March 2021, except for patients for whom the bile duct was cut up to the hilar region, which prevented creating the T-shaped F I G U R E 3 Cumulative incidence of medium-term postoperative cholangitis (A) before PSM and (B) after PSM.CI, confidence interval; PSM, propensity score matching.
Comparison of clinical characteristics between the standard anastomosis group and the T-shaped anastomosis groups after propensity score matching.
27They showed that nonoperative management for stenosis, such as percutaneous balloon dilation or endoscopic stent insertion, provided improvement in more than 80% of patients, but surgical intervention was required in 15.4% of patients.27Endoscopicinterventionat the anastomotic site of choledochojejunostomy after TA B L E 2 a Median (range).bAccording to Tokyo Guidelines 18.

Postoperative medium-term cholangitis Odds ratio (95% CI) p
Univariate logistic regression of the predictive factors for postoperative medium-term cholangitis after propensity score matching.Multivariate analysis for postoperative medium-term cholangitis in patients who underwent PD.Odds ratios of postoperative medium-term cholangitis in the T-shaped anastomosis group compared with the standard anastomosis group.Propensity score analysis was performed using the following covariates: age, body mass index, previous abdominal surgery, common hepatic duct diameter on preoperative CT, primary disease (pancreatic cancer/other), preoperative biliary drainage, and preoperative cholangitis.
TA B L E 3 TA B L E 3 (Continued)TA B L E 4Abbreviations: CI, confidence interval; IPTW, inverse probability of treatment weighting.