From Jul 2012 to May 2016, 218 consecutive LT patients with duct-to-duct anastomosis who underwent ERCP due to post-LT suspected biliary complications for the first time were included in the LT group. From Nov 2015 to May 2016, 495 consecutive patients who underwent ERCP for the first time were enrolled into non-LT group. ERCP procedures were performed by seven experienced endoscopists. For prevention of PEP, prophylactic rectal indomethacin and/or pancreatic stent were provided based on the individual endoscopist’s discretion. Medical records were reviewed one-by-one and demographics, serum amylase and lipase values before and after LT, procedure details related with PEP including pancreatic stent placement and pancreatic duct (PD) opacification, provision of rectal indomethacin, immunosuppressants taken at the time of ERCP in LT group, and occurrence of PEP were collected. As this study was conducted as a retrospective analysis, the risk or preventive factors related with post-ERCP pancreatitis that could be determined by the review of ERCP reports were selected: PD stent placement [11,12], PD opacification [13], and use of rectal indomethacin [2].

For comparison of serum amylase and lipase before and after LT, 96 LT patients who had serum amylase (n = 96) and/or lipase (n = 53) values before and after LT were identified among 218 LT patients. The serum amylase and lipase were drawn at the discretion of their liver transplant managing physicians. If multiple amylase and lipase values were present, the lowest ones were used for analysis. The serum amylase and lipase values before and after LT were compared with those from the patients with non-calcific chronic pancreatitis (NCCP) and calcific chronic pancreatitis (CCP) enrolled in our previous studies [14,15].

The primary outcomes in this retrospective study were (1) the frequency of PEP between LT and non-LT group, and (2) the comparison of serum pancreatic enzyme values before and after LT.

The standard immunosuppression protocol at Indiana University Hospital includes anti-thymocyte globulin and tacrolimus (FK506). Anti-thymocyte globulin (1.5 mg/kg of actual body weight) is given intravenously on post-operative day (POD) 1, 3, and 5. Methylprednisolone is given on POD 1 (250 mg) and POD 3 (125 mg) and prednisone (20 mg) is given on POD 5. Rituximab (150 mg/m2 of body surface) is given on POD 2. All patients begin to receive oral tacrolimus on day 2 (2 mg twice daily) and the dosage is adjusted to maintain serum trough concentration of 8 ng/mL by POD 7 through 3 months and 6 ng/mL thereafter.

PD opacification was defined as occurring when the contrast was injected and opacified any portion of the PD. Post-ERCP pancreatitis was diagnosed when new-onset or increased abdominal pain caused a hospital admission or prolonged hospital stay and was associated with an increase in amylase or lipase levels of at least 3-fold above normal 24 hours after the procedure [16]. Pancreatitis severity was graded as mild for patients with up to 3 additional hospital days and moderate for those staying between 4 and 10 days. Pancreatitis was graded as severe in patients who spent more than 10 days in the hospital or if any of the following occurred: hemorrhagic pancreatitis, pancreatic necrosis, development of a pancreatic pseudocyst, or the need for percutaneous drainage or surgery [16].

Because this was a retrospective cohort study, patients were not randomized before the interventions. Therefore, we used a propensity score matching method to reduce the bias due to confounding [17]. A logistic regression model was created to calculate the propensity score. Patients in each group were matched and allocated according to a propensity score, which reflected their possibility of undergoing LT. The following variables were tested to determine the propensity score: age, sex, use of PD stent, rectal indomethacin, PD opacification [18].

The patients in each group were matched in a 1:1 ratio using a propensity score matching method. A patient in group LT was matched with one in group non-LT based on the similarity of their propensity scores. Patients whose caliper radius was over 0.001 at the nearest available matching were excluded from this study.

The standardized differences for covariates were tested to estimate the relevance of balance between the matched groups. Standardized difference is the difference in means between both groups in units of standard deviation. If the standardized difference between both groups was < 20%, their comparability was considered to be good [19].

For continuous variables, data distribution was first evaluated for normality using the Shapiro-Wilk test. Normally distributed data were then compared using parametric methods while non-normally distributed data were analyzed using non-parametric methods. Before the matching, an unpaired t test or Mann-Whitney U test was used to compare the continuous variables, and the chi-square test or Fisher’s exact test was used to compare the descriptive variables.

After the matching, statistical differences between non-LT and LT groups were evaluated using the paired t, the Wilcoxon signed-rank, or the McNemar test. The continuous variables were expressed as the mean ± standard deviation and the descriptive variables were expressed as absolute numbers (%).

p values < 0.05 were considered statistically significant. All the analyses were conducted using the SPSS software suite version 23 (IBM Corp., Armonk, NY, USA).

Mean and median serum amylase and lipase values are summarized in Table 1. Median serum amylase values (normal range, 19 to 86 U/L) were 49.0 U/L (38.0 to 68.0) before LT and 27.0 U/L (19.3 to 36.8) after LT, and significantly decreased after LT (immunosuppression) (p < 0.001). On the other hand, when the median serum amylase value after LT was compared with the median values among NCCP (34.0 U/L; interquartile range [IQR], 24.5 to 49.0) and CCP (31.7 U/L, IQR, 20.0 to 40.8), the median serum amylase value after LT decreased to a level lower than NCCP (p < 0.05) or comparable to CCP (p = 0.793).

The median serum lipase value (normal range, 7 to 59 U/L) was 40.0 U/L (26.5 to 54.0) before LT and 10.5 U/L (6.0 to 21.0) after LT, and significantly decreased after LT (immunosuppression) (p < 0.001). Similar to the results with amylase, when the median serum lipase value after LT was compared with the median value among NCCP (22.3 U/L; IQR, 9.0 to 30.0) and CCP patients (10.0 U/L; IQR, 3.0 to 19.0), the median serum lipase value after LT decreased to a level lower than NCCP (p < 0.05) or comparable to CCP (p = 0.676). Therefore, serum amylase and lipase values after LT (immunosuppression) decreased to a level of chronic calcific pancreatitis. Fig. 1 shows the distribution of serum amylase and lipase values among the four groups.

We reviewed the records of 713 patients. The number of patients who underwent LT and did not undergo LT was 218 and 495, respectively. The characteristics of patients are shown in Table 2.

Four of the five confounding variables (sex, PD stent, rectal indomethacin, and PD opacification) showed statistical difference and three of the five confounding variables showed poor standardized difference scores before propensity score matching, i.e., the following confounding variables differed significantly between non-LT group and LT group: sex, placement of PD stent, administration of rectal indomethacin and PD opacification (Table 2).

We generated 146 matched pairs using the propensity score matching. After propensity score matching, there were 408 patients in non-LT group and 115 in LT group. All five confounding variables had an acceptable standardized difference (i.e., < 20%), indicating that the matching procedure was efficient in creating a balance between the non-LT and LT groups (Table 2).

There was marginal significant difference between both non-LT and LT groups before the propensity score matching with respect to frequency of post-ERCP pancreatitis (16 [3.2%] in non-LT group vs. 2 [0.9%] in LT group, p = 0.069). After propensity score matching, marginal significant difference still existed with respect to frequency of post-ERCP pancreatitis (7 [4.8%] in non-LT group vs. 1 [0.7%] in LT group, p = 0.067).