Rectal indomethacin reduces the risk of post-endoscopic retrograde cholangiopancreatography pancreatitis in low-risk patients

Introduction

Post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP) is the most frequent major complication of ERCP. PEP occurs in 1-9% of average-risk patients [1] and from 11-40% of high-risk patients, with a 0.1% mortality risk related to pancreatitis and an annual healthcare expenditure totaling $199 million in the United States [2-9].

Freeman et al showed that prophylactic pancreatic duct stents (PDS) could reduce the incidence of PEP from 23% to 4% in high-risk patients [10]. Multiple studies have confirmed that PDS placement could effectively minimize pancreatitis in high-risk patients [11-19]. One meta-analysis reported an 88% success rate for PDS placement and a low complication rate (1.8%). Moreover, placing a PDS reduced PEP from 15.5% to 5.8% (odds ratio [OR] 3.2, 95% confidence interval [CI] 1.6-6.4) and reduced the incidence of severe PEP to 11.5% [15,20,21]. The number needed to treat showed that PDS should be placed in 10 patients to prevent one episode of PEP [22].

Nevertheless, there has been inconsistent adoption of this technique among endoscopists [10], perhaps because PDS placement has an unacceptable failure rate, especially in patients with small and stenotic ducts [13,23]. It was noted that failure to successfully place a PDS could cause more harm than if PDS placement was not attempted [12,24]. Secondly, complications can be seen with PDS in up to 4.9% of patients, including misplacement, proximal migration and subsequent duct injury, and stricture that can lead to chronic pancreatitis [16,25]. Thirdly, PDS placement requires follow-up imaging and potentially repeat endoscopy to remove the retained stent, thus increasing costs.

Given the substantial burden from PEP, multiple pharmacological agents have been studied as prophylaxis against PEP. With the exception of rectal non-steroidal anti-inflammatory drugs (NSAIDs), all other pharmacological prophylactic agents have shown disappointing results. A prospective controlled study conducted by Elmunzer et al reported a 46% relative risk reduction of PEP with rectally administered indomethacin (RI) compared to placebo alone in high-risk patients [26]. The study was criticized later because 80% of the patients with high suspicion for sphincter of Oddi dysfunction (SOD) had PDS placement, whereas lower-risk patients had indomethacin therapy [27] Two meta-analyses, including data from Elmunzer et al, have shown that rectal NSAIDs prevent PEP [28,29]. Shortly after these studies were published, the European Society for Gastrointestinal Endoscopy (ESGE), followed by the Japanese guideline, recommended RI in all patients undergoing ERCP without complications [30,31]. A retrospective study of 4017 patients by Thiruvengadam et al showed that RI reduced the risk of PEP in low-risk patients and patients with malignant obstruction [32].

In contrast, in a prospective randomized trial of a cohort where 70% of patients were at average risk for PEP, Levenick et al showed that RI does not prevent PEP in consecutive patients [33]. Two other randomized trials [34,35], followed by 2 meta-analyses by Inamdar and Feng et al concluded that RI does not prevent PEP in average-risk patients [36,37]. Further complicating the issue, Rainio et al published a retrospective study of 2000 consecutive patients, demonstrating that rectal diclofenac did not affect the incidence or severity of PEP [38]. Given the above evidence, the role of RI in consecutive patients is not clear. The current recommendation of the ASGE is RI should be given to high-risk patients and may be considered in average-risk patients [39].

Pharmacological agents could offer the ideal prophylaxis measures, preferable to PDS because of their safety profile and low cost. However, RI does not eliminate the risk of PEP in high-risk patients, and the role of RI in low-risk patients is still considered controversial. Furthermore, it is not known whether the combination of RI with PDS can offer additional benefit in decreasing the incidence and severity of PEP. Accordingly, the objective of our study was to evaluate the impact of RI in preventing PEP in consecutive patients and to determine whether the combination of RI and PDS decreases the incidence of PEP compared to RI alone.

Patients and methods

After approval by the University and Medical Center Institutional Review Board at East Carolina University/Vidant Medical Center, we conducted a retrospective study of all ERCPs performed at Vidant Medical Center from January 2007 to October 2016. ERCP procedures performed before January 2009 were excluded because electronic medical records were not used in our institution at that time. The study included all patients who underwent diagnostic or therapeutic ERCP from January 2009 to October 2016. RI was available starting in November 2012. The indomethacin group consisted of patients who received 100 mg RI during or immediately after the ERCP while the unexposed group consisted of patients who did not receive RI.

The procedures were performed by 4 experienced endoscopists, no trainees were involved in any of the procedures. All patients underwent ERCP with a standard duodenoscope (TJF-160F and TJF-Q180V; Olympus Optical Co., Ltd., Tokyo, Japan). Patient characteristics and procedural details were collected. PEP was defined by consensus criteria. Patients noted to have had acute pancreatitis within 72 h before the procedure were excluded from the study.

Patients were observed in the recovery area for at least 2 h after the procedure and were assessed by the endoscopy nurse and the endoscopist before discharge. Patients who developed symptoms of PEP were monitored closely and received intravenous hydration after confirmation of the diagnosis.

We reviewed patients’ demographic information, relevant medical history and home medications. Likewise, physician medication orders and nursing notes were reviewed to confirm that the patient received RI before or immediately after the ERCP. We reviewed available laboratory blood work, pre- and post-procedure imaging studies, and follow-up documentation. Procedural notes were reviewed with careful attention to diagnosis and indications for ERCP and the use of anesthesia or sedation. Procedural details were recorded, including sphincterotomy, biopsy, stent placement/removal, cannulation of the common bile duct or the main pancreatic duct, cholangiogram or pancreatogram. In addition to noting the type and size of the stent used, we also gathered information on other therapeutic interventions during the ERCP.

Results

We evaluated 2238 patients who underwent ERCP and had adequate follow up (1055 in the RI group and 1183 in the control group). The average patient age was 60.4±18.7 years; 1247 (56%) patients were female and 991 (44%) were male. The average BMI was 29±8.2 kg/m2. Procedure details and patient characteristics are compared in Table 1.

Table 1 Patient baseline characteristics for the control group and rectal indomethacin group (n=2238)

Study outcome

Overall, PEP was diagnosed in 107 of the 2238 (4.5%) patients who underwent ERCP, 48/1055 (4.5%) in the RI group and 59/1183 (5%) in the control group. There was no difference in the rate of bleeding between the 2 groups.

In the univariate analysis, the following factors were associated with PEP: females less than 40 years of age with a native papilla, contrast injection into the pancreatic duct, difficult cannulation, pancreatic sphincterotomy, failed PDS placement, native papilla, and a history of recurrent acute pancreatitis (Table 2).

Table 2 Incidence of post-endoscopic retrograde cholangiopancreatography pancreatitis according to risk factors in consecutive patients (n=2238)

In the multivariate model, contrast injection of the main pancreatic duct, difficult biliary cannulation, females less than 40 years of age, native papilla, pancreatic sphincterotomy, and failed PDS placement were associated with a higher incidence of PEP. RI reduced the incidence of PEP in consecutive patients by 55% (OR 0.45, 95%CI 0.36-0.94; P=0.03) and reduced the incidence of moderate/severe PEP by 47% (OR 0.53, 95%CI 0.27-0.97; P=0.04) (Table 3).

Table 3 Factors associated with post-endoscopic retrograde cholangiopancreatography pancreatitis in consecutive patients: multivariate analysis

RI impact on low-risk patients

Of the 2238 patients, 364 (16%) were considered at high-risk for PEP and 1874 (84%) were at low-risk, according to the consensus definition in the Methods section. In the univariate model, PEP was associated with any contrast injection into the pancreatic duct, pancreatic sphincterotomy, biliary sphincterotomy, failed PDS, and female sex with native papilla. In the multivariate model, pancreatic sphincterotomy and biliary sphincterotomy were associated with a higher incidence of PEP. RI reduced the incidence of PEP in low-risk patients by 62% (OR 0.38, 95%CI 0.19-0.74; P=0.004) (Table 4). RI reduced the incidence of moderate/severe PEP by 57% (OR 0.43, 95%CI 0.18-0.10; P=0.049).

Table 4 Factors associated with post-endoscopic retrograde cholangiopancreatography pancreatitis in low-risk patients: multivariate analysis

Discussion

Our results showed that RI decreased the incidence of PEP by 42% and reduced the rate of moderate/severe PEP by 55% in consecutive patients. Analysis of only low-risk patients demonstrated that RI reduced the incidence of PEP by 62% and reduced the incidence of moderate/severe PEP by 57%. Our results are consistent with those of Thiruvengadam et al, who reported a reduction in the incidence of PEP by 65% and the incidence of moderate/severe PEP by 83%. Previous trials showed no effect of RI in decreasing the incidence of PEP in low-risk patients [33-35]. The different outcomes could be due to the small sample sizes in these trials for detecting the differences in PEP, and the rarity of PEP in low-risk patients.

There were some differences in patient- and procedure-related risk factors. Compared to the control group, the RI group had a slightly higher percentage of patient-related risk factors: native papilla (53% vs. 42%), history of recurrent pancreatitis (4% vs. 2%), and SOD (14 vs. 6 patients). Likewise, the RI group had slightly higher prevalence of procedure-related risk factors: pancreatic sphincterotomy (6% vs. 1%), main pancreatic duct cannulation (41% vs. 20%), and contrast injection into the pancreatic duct (27% vs. 9.8). The RI group had a higher incidence of PDS placement (24% vs. 6%). The multivariate analysis model that combined all possible patient- and procedure-related risk factors demonstrated that RI truly reduced the rate of PEP. Given the higher prevalence of risk factors in the RI group, we further confirmed our multivariate model with a propensity-matched group of low-risk patients with native papilla. Likewise, RI reduced the incidence of PEP by 61% (OR 0.39, 95%CI 0.18-0.8; P=0.009).

In our study, we found that females less than 40 years of age with a native papilla, contrast injection of the pancreatic duct, difficult cannulation, pancreatic sphincterotomy, failed PDS placement, native papilla, and a history of recurrent pancreatitis were associated with PEP. This is consistent with previous reports [6].

In our study, 83% of the patients considered low-risk for PEP. Our overall reported rate of PEP was 4.5%, close to the reported rate of 4.2% in a large systemic review by Andriulli et al [4]. However, it is worth mentioning that the positive effect of RI in low-risk patients could be explained by the relatively high incidence of PEP in both our study and that of Thiruvengadam et al (4.5% and 7.53%, respectively). Patients were considered low-risk for ERCP if they did not meet the high-risk criteria defined in prospectively validated trials [26,40-42]. Procedure- and patient-independent risk factors for PEP were studied extensively, but did not show potential for predicting every case of PEP. Occasionally, meeting a single minor criterion for high-risk ERCP could markedly increase the risk of PEP without fulfilling the high-risk ERCP criteria (1 major or 2 minor criteria); such criteria included any injection into the pancreatic duct, and female sex with native papilla having wire cannulation into the pancreatic duct. Thus, we performed an analysis of consecutive patients as well as only those at low risk for ERCP, to confirm that the effect on PEP applies to any patients undergoing ERCP, regardless of their risk factors.

The mechanism leading to PEP has not been fully established. It has been suggested that local injury to the papilla from instrumentation, sphincterotomy, or forceful and repetitive injection of contrast causes papillary edema or spasm. This is thought to result in ductal hypertension followed by reduced pancreatic duct drainage. The poor drainage initiates an inflammatory cascade, promoting intraluminal activation of proteolytic enzymes and autodigestion of the pancreas. Significant release of cytokines (interleukins 1, 6 and 8) results in a systemic inflammatory response with multiorgan involvement [8,43].

PDS maintains pancreatic drainage to the ampulla through the pancreatic duct after ERCP, without providing prophylaxis for other proposed mechanisms of PEP, including chemical, allergic, enzymatic and infectious mechanisms [44]. Indomethacin, an NSAID, is a potent inhibitor of cyclooxygenase, neutrophil-endothelial interactions and phospholipase A2. Phospholipase A2 regulates the pro-inflammatory mediators, including arachidonic acid products and platelet-activating factors. This inhibition is thought to be key in aborting the early steps in the inflammatory cascade process and preventing PEP [45,46].

Thiruvengadam et al reported the maximal reduction of PEP when using RI in patients with malignant obstruction compared to the control group (2.3% vs. 7.5%) [34]. We did not see a significant reduction in PEP in patients with a malignant obstruction compared to the control group. This may have been due to the overall low incidence of PEP in our malignant obstruction patients (2.7%). It is worth noting that more patients in the RI group received prophylactic ciprofloxacin before the procedure compared to the control group. Our subgroup analysis did not show any impact of ciprofloxacin on the incidence of PEP.

One of the main strengths of our study was the large sample size with adequate power to detect the impact of RI on preventing PEP in a primarily low-risk endoscopy unit. This mixed study population should mirror and be applicable to many ERCP practices. In addition, the unique geographic location of our hospital as the only tertiary referral center serving 29 counties helps to ensure adequate follow up of any possible ERCP complications. Nevertheless, our study has several limitations. First, its retrospective nature, comparing 2 heterogeneous groups, is subjective to confounder bias for the baseline patient risk of PEP. We tried to control for this limitation by collecting all possible risk factors and by performing univariate and multivariate analyses to explore all potential confounders. Second, there was a difference in time period and individual endoscopists between the RI group and control group. We did not find any differences in the incidence of PEP in relation to the ERCP year or the endoscopists. Third, large-volume hydration as a prophylaxis for PEP was not adopted during the study period and could not be reliably examined as a confounder. Finally, the number of PDS placements in the study population was small and lacked uniform criteria.

In conclusion, RI reduced the incidence and severity of PEP in consecutive and low-risk patients. The combination of RI with PDS may not offer any additional benefit to RI alone. Given the high safety profile and the low cost, RI should be considered for every patient undergoing ERCP.