Endoscopy 2015; 47(01): 6-7
DOI: 10.1055/s-0034-1390799
Editorial
© Georg Thieme Verlag KG Stuttgart · New York

Nonsteroidal anti-inflammatory drugs for preventing post-ERCP pancreatitis: why rectal and not intramuscular?

B. Joseph Elmunzer
Division of Gastroenterology & Hepatology, Medical University of South Carolina, Charleston, South Carolina, United States
› Author Affiliations
Further Information

Publication History

Publication Date:
22 December 2014 (online)

Eight randomized, controlled trials (RCTs) and several meta-analyses have demonstrated that rectal nonsteroidal anti-inflammatory drugs (NSAIDs) reduce the risk of postendoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP) by about 50 % [1] [2]. Rectal indomethacin and diclofenac also reduce the severity of PEP [3], appear to be effective in both high- and average-risk cases [1] [2], and are recommended (as a grade A recommendation) by the European Society of Gastrointestinal Endoscopy for almost all patients undergoing ERCP [4]. Although the recent uptake of these medications in clinical practice has been rapid and widespread, there does appear to be some apprehension surrounding the rectal mode of administration.

NSAID suppositories are not available on many hospital formularies, may be somewhat challenging to obtain, and are marginally more expensive than their parenteral counterparts in certain countries. Furthermore, rectal administration is unappealing to some healthcare providers and may be undesirable to certain patients. Thus, practitioners are tempted to administer parenteral NSAIDs because of their widespread availability on anesthesia carts, their relative ease of delivery compared with suppository insertion, and the perception that the efficacy of NSAIDs is a class effect. Indeed, several prominent academic and community hospitals continue to routinely administer NSAIDs intravenously for PEP prevention despite a lack of RCT evidence supporting this route.

In this context, Park et al. have attempted to shed light on this timely clinical question by reporting the results of their double-blind, placebo-controlled RCT evaluating the efficacy of intramuscular diclofenac for the prevention of PEP [5]. They enrolled 380 eligible patients undergoing diagnostic or therapeutic ERCP, excluding patients with contraindications to NSAIDs and those deemed to be at minimal or no risk for PEP. Patients were randomized to receive either 90 mg of intramuscular diclofenac or a placebo injection immediately after ERCP. The primary end point was the development of PEP, defined somewhat atypically in this study as an amylase level more than three times the upper limit of normal, as well as new or worsening pancreatic-type abdominal pain and tenderness with associated nausea and/or vomiting lasting more than 24 hours after the ERCP.

The study sample consisted of mostly average-risk patients, with over 80 % undergoing ERCP for choledocholithiasis or malignant obstruction; only 25 % had high-risk features such as precut sphincterotomy, pancreatic sphincter ablation, or ampullectomy. Baseline characteristics were similar between the two study groups. There was no crossover of patients from one group to another, although 37 randomized patients did not undergo ERCP and were not followed up; thus the primary analysis was not conducted according to the intention-to-treat principle. In a per-protocol analysis, 22 of the 173 patients in the diclofenac arm (12.7 %) and 20 of the 170 patients in the placebo arm (11.8 %) developed PEP (P = 0.87). There was no difference in the severity of PEP between groups, and intramuscular diclofenac did not demonstrate a trend toward benefit in any of the predefined subgroups. There were no adverse events related to the study drug.

Within the limitations acknowledged by the authors, this study does help to answer an important question in current ERCP practice – that intramuscular diclofenac is not effective for preventing PEP. The study also reinforces what has emerged as an important principle in PEP pharmacoprevention – that NSAIDs administered via nonrectal routes do not appear to be protective. Indeed, prior placebo-controlled RCTs of intravascular valdecoxib [6], oral diclofenac [7], and intramuscular diclofenac [8], have also yielded negative results. As such, there are no existing clinical trial data to support administration of prophylactic NSAIDs via any nonrectal route, or to support the use of any medication other than indomethacin or diclofenac, postulated to be specifically effective because they are potent inhibitors of phospholipase A2 [9]. Thus, based on available RCT data, only rectal indomethacin or diclofenac should be prescribed for PEP prevention, as recommended by existing guidelines [4].

Furthermore, the results of this study highlight another important question – whether investigators should continue to devote time and resources toward studying nonrectal NSAIDs for PEP pharmacoprevention. As rectal NSAIDs are highly effective, safe, widely available, and easy to administer, it seems unlikely – in the absence of any compelling new preclinical evidence – that another formulation would provide enough clinical, logistical, or economic advantage over rectal NSAIDs to justify the time and cost necessary for high-quality clinical trials. Considering that the risk of PEP remains as high as 10 % despite the combination of rectal NSAIDs and prophylactic stent placement [3], other more pressing research that aims to further reduce risk or substantially limit costs should take priority. Moreover, given the highly favorable risk – benefit profile of rectal NSAIDs – and the fact that they are guideline recommended – the ethics and scientific merit of a placebo-controlled approach to pharmacoprevention research should be thoughtfully considered. Indeed, several experts in this field advocate that potential prophylactic interventions be compared with rectal NSAIDs as the standard active control. This approach renders it very challenging to evaluate nonrectal NSAID formulations due to the prohibitively large sample sizes that would be required.

Finally, this RCT also reminds us of the importance of raising our standards for the quality of PEP pharmacoprevention trials, particularly as it pertains to statistical power. From this study it can be reasonably concluded that intramuscular diclofenac was ineffective because, after analyzing 343 patients, there was truly no difference in PEP rates between the two groups (12.7 % diclofenac, 95 % confidence interval [CI] 7.8 % – 17.6 % and 11.8 % placebo, 95 %CI 7 % – 16.6 %). In this particular case, even if the investigators had intended to enroll a larger sample size, the similar PEP rates at this stage of the trial would likely have prompted a data and safety monitoring board (DSMB) to prematurely terminate the study due to futility. However, if a modest trend toward benefit associated with the study drug had been observed, a trial of this size would have likely been underpowered to detect a clinically important effect, conferring a high risk for a type 2 statistical error – failing to demonstrate a significant difference when one actually exists. This would have occurred because the assumptions used to estimate the sample size – that the baseline risk of PEP in the trial would be 17.5 % and that intramuscular diclofenac would reduce risk by nearly 60 % (down to 7.5 %) – were perhaps intended to boost practicality by limiting sample size, but actually rendered the study highly prone to being underpowered. A baseline PEP risk of 17.5 % is excessively high, especially in a study not limited to high-risk cases, and a reduction of ~ 60 % is overly ambitious, at the upper limit of the benefit that might be expected from any pharmacological agent. More conservative assumptions would have required a larger study sample, but would also have reduced the risk of an uninformative study in the event that a modest benefit was observed. Indeed, inadequate sample sizes, leading to inaccurate conclusions about the efficacy of myriad agents, have been a hallmark of PEP pharmacoprevention research and a central reason for the lack of progress in this area. For reference, assuming a two-tailed alpha of 0.05 and a power of 80 %, 1628 patients (814 per arm) would be necessary to detect a 50 % reduction in PEP from 6 % to 3 % – reasonable assumptions in an average-risk patient population – and 778 patients (389 per arm) would be necessary to detect the same reduction in high-risk cases (reduction from 12 % to 6 %).

Additional factors that have traditionally been underutilized in PEP prevention trials, but which should be embraced in all future studies, are a systematic approach to agent selection, strict patient follow-up, adherence to the intention-to-treat principle, blinding (in prophylactic stent trials), routine use of the consensus definition (until more accurate diagnostic criteria or tests are validated), and involvement of a DSMB to ensure methodological rigor and integrity of study data. In this era of diminishing research resources, improving the quality of PEP pharmacoprevention research is mandatory towards our goal of eliminating PEP. It is upon us to assimilate the lessons learned from three decades of research disappointment in order to enter a more productive era of PEP prevention trials.