Tenner S. Drug induced acute pancreatitis: Does it exist? World J Gastroenterol 2014; 20(44): 16529-16534 [PMID: 25469020 DOI: 10.3748/wjg.v20.i44.16529]
Corresponding Author of This Article
Scott Tenner, MD, MPH, Division of Gastroenterology, Department of Medicine, State University of New York, 2211 Emmons Ave, Brooklyn, NY 11235, United States. drtenner@brooklyngi.com
Research Domain of This Article
Gastroenterology & Hepatology
Article-Type of This Article
Topic Highlight
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Scott Tenner, Division of Gastroenterology, Department of Medicine, State University of New York, Brooklyn, NY 11235, United States
Scott Tenner, Downstate Medical Center, State University of New York, Brooklyn, NY 11235, United States
ORCID number: $[AuthorORCIDs]
Author contributions: Tenner S wrote the manuscript based on prior work with Nison Badalov and William Steinberg.
Correspondence to: Scott Tenner, MD, MPH, Division of Gastroenterology, Department of Medicine, State University of New York, 2211 Emmons Ave, Brooklyn, NY 11235, United States. drtenner@brooklyngi.com
Telephone: +1-51-63160830 Fax: +1-718-3682249
Received: March 10, 2014 Revised: May 8, 2014 Accepted: July 15, 2014 Published online: November 28, 2014 Processing time: 267 Days and 9.8 Hours
Abstract
As the incidence of acute pancreatitis continues to rise, establishing the etiology in order to prevent recurrence is important. Although the etiology of acute pancreatitis is not difficult in the majority of patients, almost a quarter of patients are initially labeled as having idiopathic acute pancreatitis. When confronted with a patient with acute pancreatitis and no clear etiology defined as an absence alcoholism, gallstones (ultrasound and/or MRI), a normal triglyceride level, and absence of tumor, it often appears reasonable to consider a drug as the cause of acute pancreatitis. Over 100 drugs have been implicated by case reports as causing acute pancreatitis. While some of these case reports are well written, many case reports represent poorly written experiences of the clinician simply implicating a drug without a careful evaluation. Over-reliance on case reports while ignoring randomized clinical trials and large pharmacoepidemiologic surveys has led to confusion about drug induced acute pancreatitis. This review will explain that drug induced acute pancreatitis does occur, but it is rare, and over diagnosis leads to misconceptions about the disease resulting in inappropriate patient care, increased litigation and a failure to address the true entity: idiopathic acute pancreatitis.
Core tip: While the literature has reported over 130 drugs as causing acute pancreatitis, the evidence that these drugs have a true causal role is lacking in the vast majority of drugs. While idiopathic pancreatitis is common, accounting for almost a third of patients with acute pancreatitis, drug induced acute pancreatitis is probably an uncommon, perhaps a rare disease. Before a clinician blames a drug as causing acute pancreatitis, a thorough evaluation for more common causes should be made, even a consideration that the disease is merely idiopathic.
Citation: Tenner S. Drug induced acute pancreatitis: Does it exist? World J Gastroenterol 2014; 20(44): 16529-16534
Idiopathic Acute Pancreatitis accounts for 20-40 percent of patients with acute pancreatitis[1,2]. That is, normally, approximately a third of patients who present with acute pancreatitis defy the clinician’s ability to determine what caused the disease. Idiopathic acute pancreatitis is defined as acute pancreatitis with no etiology established after initial laboratory (including lipid and calcium level) and imaging tests (trans-abdominal ultrasound, MRI and CT in the appropriate patient)[3]. These patients do not have gallstones, a significant history of alcohol use, hypertriglyceridemia and a tumor. Anatomic and physiologic anomalies of the pancreas occur in 10%-15% of the population, including pancreas divisum and sphincter of Oddi dysfunction[4]. However, it remains controversial if these disorders alone cause acute pancreatitis.
There may be a combination of factors, including anatomic and genetic, that predispose to the development of acute pancreatitis in susceptible individuals[5]. The influence of genetic defects, such as cationic trypsinogen mutation, SPINK, or CFTR mutations, in causing acute pancreatitis is being increasingly recognized. These defects, furthermore, may also increase the risk of acute pancreatitis in patients with anatomic anomalies, such as pancreas divisum[6]. The idea that acute pancreatitis may result from a combination of factors working together should not be a surprise when one considers that most patients with gallstones, hypertriglyceridemia, alcoholism and pancreas divisum will never develop acute pancreatitis.
Clinician’s caring for a patient with acute pancreatitis yearn to find a diagnosis to prevent a recurrent attack. This is compounded by the patient’s desire to understand what has happened to them to cause so much pain and suffering. In addition to endoscopic interventions, clinicians search the literature for possible causes. The profession demands it, the patient’s deserve it, and the literature provides a plethora of possibilities.
DRUGS AS A CAUSE OF ACUTE PANCREATITIS
Most patients who are admitted to a hospital are already taking a medication. Nearly 240 million Americans take at least one prescription drug weekly, and pharmacies fill over ten million prescriptions each day[7]. Over 100 drugs have been reported to cause acute pancreatitis in the scientific literature. Most reviews claim that drug induced acute pancreatitis accounts for 3%-5% of all cases of acute pancreatitis[8]. The diagnosis of drug induced acute pancreatitis is difficult to establish since drug-induced pancreatitis rarely is accompanied by clinical or laboratory evidence of a drug reaction, such as rash, lymphadenopathy, and/or eosinophilia, few ancillary data are available to help with the diagnosis.
While a few medications have been reported to cause acute pancreatitis based on a large body of evidence, most of the drugs implicated are based on case reports that suffer from a combination of inadequate criteria for the diagnosis of acute pancreatitis, failure to rule out more common etiologies, and/or lack of a rechallenge with the medication[9]. A rechallenge is a case in which a patient who develops acute pancreatitis has the medication suspected as causing acute pancreatitis withheld. After the acute pancreatitis resolves, the medication is restarted (typically as the medication was not originally suspected of causing the acute pancreatitis). Within a period of time (typically shorter), after the medication is restarted, the patient has another attack of acute pancreatitis. A valid rechallenge case report should be considered when evaluating whether a particular drug causes acute pancreatitis; however, it is not proof of causation. For example, it is clear that many patients with idiopathic pancreatitis or microlithiasis have recurrent attacks of acute pancreatitis. Therefore, stopping and restarting a drug with recurrence of pancreatitis may be a coincidence and not cause and effect[10].
Badalov et al[9] published an extensive review of published case reports in the peer reviewed literature. Using criteria based on the presence of a rechallenge, latency, and the number of case reports (Table 1), a classification system “based on the evidence” was provided. Table 2 shows the medications from the published case reports with the “most evidence” of causing acute pancreatitis. At the time the authors published the paper, none of them were aware that the United States Food and Drug Administration (“FDA”) and trial lawyers would use the classification as a partial basis for assigning blame to drugs as causing acute pancreatitis[11].
Table 1 Classification of drug induced pancreatitis.
Class Ia drugs At least 1 case report with positive rechallenge, excluding all other causes, such as alcohol, hypertriglyceridemia, gallstones, and other drugs
Class Ib drugs At least 1 case report with positive rechallenge; however, other causes, such as alcohol, hypertriglyceridemia, gallstones, and other drugs were not ruled out
Class II drugs At least 4 cases in the literature Consistent latency (75% of cases)
Class III drugs At least 2 cases in the literature No consistent latency among cases No rechallenge
Class IV drugs Drugs not fitting into the earlier-described classes, single case report published in medical literature, without rechallenge
Table 2 Summary of drug-induced acute pancreatitis.
Through the Federal Food, Drug, and Cosmetic Act (“FDCA”), the FDA is empowered to verify the safety of drugs on the market[12]. Although the FDA employs a rigorous review process to ascertain the safety and efficacy of drugs prior to approval, reports have consistently warned that pre-market research often fails to provide an accurate risk-benefit profile for marketed products[13]. Many drugs come to the market and subsequently are found to have significant side effects that pre-market trials did not reveal[14]. To rectify this problem, the FDA had developed the Adverse Event Reporting System (FAERS)[15].
Based on “MedWatch Reports”[16] filed by interested clinicians, the FDA’s reporting programs generate a “deluge of information. Annually the agency has received more than 200000 adverse event reports regarding drugs or biologic products. It is not surprising that the agency describes its analysis of this flood of data as triage[17]. The reports are typically incomplete and often, biased. Although more work on the database and system is needed to distinguish reliable findings from “variability and noise”, more resources are necessary and lacking[18].
Despite incomplete data, the FDA often relying on the FAERS will issue warnings and require manufacturers to add “black box warnings” intended to alert physicians to the importance of the adverse information learned. However, with premature data causing unsubstantiated fears, the FDA has added, modified, and often removed black box warnings from the drugs in question. The addition of these black box warnings has fueled litigation[17].
AN ILLUSTRATION OF THE FALLACY OF DRUG INDUCED ACUTE PANCREATITIS: EXENATIDE (BYETTA®)
The claim that Byetta (exenatide and other GLP-1 agonists) cause acute pancreatitis exemplifies the problem with drug induced acute pancreatitis. Based on case reports, especially following the criteria set forth in the paper by Badalov et al[9] MedWatch reports, the FAERS, resultant black box warnings, and poor science, confusion and litigation resulted as “experts” claimed the exenatide caused acute pancreatitis.
Exenatide, an incretin mimetic, was approved as Byetta by the FDA on April 25, 2005. The drug is an adjunctive therapy to improve glycemic control in patients with type II diabetes mellitus. The first published case reports of acute pancreatitis thought to be caused by exenatide appeared shortly after the drug was approved[19]. Additionally, by December 31, 2006, according to the FAERS database, there were 48 documented domestic cases of acute pancreatitis in patients taking exenatide[20]. Noting slightly more cases of acute pancreatitis than expected in the general population, the FDA asked the manufacturer, Lilly and Amylin Pharmaceuticals to strengthen the labeling of acute pancreatitis from the Adverse Reactions section to the Warnings and Precautions section of the exenatide label.
While the FDA was comparing the incidence of acute pancreatitis in the exenatide using diabetic population to the general population, it is not clear that they were aware that diabetic persons were at a significant increased risk of developing acute pancreatitis. For a variety of reasons, including increased incidence of gallstones and hypertriglyceridemia, the incidence of acute pancreatitis in patients with diabetes is higher than the general population[21]. Therefore, regardless of the drug used, if one simply compared the normal population incidence of acute pancreatitis with the diabetic population, one would find a higher incidence in the diabetics. This is a classic confounding variable rather than a drug effect.
The limitations to Medwatch reports cannot be over stated. In many reports the diagnosis of acute pancreatitis is not clearly established. Thus, there is no reason to proceed with considering the case as the adverse event suspected may be another pathology in the abdomen. Misdiagnosis of acute pancreatitis often occurs by clinicians who search for a reason for abdominal pain and merely rely on mild elevations in the amylase and lipase to reach a diagnosis. This is not appropriate, however, as any inflammatory process in the abdomen can cause a mild 2-3 fold elevation of the amylase and/or lipase[22]. Additionally, many patients with diabetes have been shown to have mild elevations, greater than three times the upper limit of normal, of amylase and/or lipase[23]. Thus, many patients with abdominal pain from other sources are falsely labeled as having acute pancreatitis. In the patients who truly have acute pancreatitis, many of the reports fail to identify if the patient has more likely causes of acute pancreatitis, such as gallstones, a history of alcoholism, hypertriglyceridemia[9].
Despite the limitations to the reports and the FDA’s position that the FAERS is for hypothesis testing, Elashoff and colleagues[24] examined the FAERS database from 2004-2009 for reported adverse events for exenatide and other medications (which served as controls) in order to determine if patients were at an increased risk of developing pancreatitis. The authors found that the risk of developing pancreatitis from exenatide was higher compared to from other therapies, but importantly the issue of reporting bias could not be entirely ruled out.
Although the FDA agreed to study the issue further, in the meantime it required Amylin and Lilly to alert health care professionals in several ways - including via industry letters, published articles, and reports of these cases in the FDA Newsletter[25]. The result was a surge of FAERS cases involving exenatide as a cause of acute pancreatitis immediately followed the FDA notification requirement. Despite the obvious reporting bias induced by the FDA notification, and the failure of the FDA to note that the population using exenatide-diabetics-inherently had a predisposition for acute pancreatitis, the FDA subsequently added a black box warning to the drug’s labeling. The black box warning stated that exenatide could cause acute pancreatitis[26].
Immediately thereafter, thousands of persons who had developed acute pancreatitis while taking exenatide initiated multiparty litigation suits. They relied on the FAERS database and resultant black box warning. The plaintiffs, diabetics already at risk for developing acute pancreatitis, claimed that the defendants Lilly and Amylin Corporations knew or should have known of the hazards associated with exenatide in causing acute pancreatitis. In addition, by claiming that the defendants actively concealed information that demonstrated the dangers of their drug and thus misled the public and prescribing physicians, the plaintiffs were granted broad access to company documents during discovery[27]. The costs of litigation skyrocketed.
Despite the persistent litigation occurring, over the last year, the FDA independently evaluated the post marketing reports that exenatide was a cause of acute pancreatitis. After an exhaustive evaluation of more than 250 toxicology studies conducted in nearly 18000 live animals, no evidence of pancreatic disease was found[28]. In addition to the laboratory data, the FDA reviewed data from 200 trials (including other GLP-1 agonists), involving 41000 patients, and found no evidence of an increased risk of pancreatic disease. The FDA has promulgated that “assertions concerning a causal relationship between incretin drugs are inconsistent with the scientific literature. Simply, despite case reports and MedWatch reports, exenatide does not cause acute pancreatitis.
RETHINKING CAUSATION
It is important to use the general scientific method in making causal claims about human health and disease[29]. The basic structure of the scientific method to determine causation includes: hypothesis generation, observable predictions, alternatives, and tests to distinguish between the causal hypothesis of interest and its alternatives. There could be competing explanations for any scientific observation. Epidemiologic methods involving human subjects are the most important means for identifying and testing hypotheses involving human disease causation. Randomized controlled trials are the strongest means and case reports are the lowest means[30] (Table 3). The use of the scientific method avoids falsely claiming causation when the truth is mere chance. Chance is not the only alternative to causation, but must be considered strongly.
The criteria of causation is best understood by the Hill criteria[31]. An “association” in this methodology is not satisfied by the existence of individuals with exposure to the putative cause and the disease of concern. Rather, an “association” from a causal perspective would only exist if a statistically-significant relationship (e.g., between the rate of acute pancreatitis in patients with diabetes mellitus patients exposed to exenatide and the rate of acute pancreatitis in similar diabetic patients not exposed to exenatide) was demonstrated in analytical epidemiological studies. Those studies should be well-designed, with careful attention to diagnostic criteria, adherence to medication, control of confounders, and avoidance (or correction) of important sources of bias. Case reports would never meet this level of evidentiary need to determine causation.
Hill’s 9 criteria evaluate the totality of evidence for causation evaluating for temporality, strength of association, consistency of the association, the presence of a biologic gradient (dose-response), biologic plausibility, specificity, coherence, experimentation and analogy (Table 4). In applying the 9 criteria to a drug like exenatide, the evidence shows no causal association. There is no temporality as the latency for exenatide causing acute pancreatitis varies among the reports. As to strength, large epidemiologic studies show no causal relationship of exenatide to acute pancreatitis. There is no consistency of the data. Results from clinical trials, epidemiology, case reports, and animal studies are inconsistent. Based on animal and clinical trial data there is no biologic plausibility (no established mechanism) or gradient. There is no evidence that increase in dosage and/or increase in time results in a linear increase in episodes of acute pancreatitis. Experimental data, in both animals and humans, do not establish that exenatide is a cause of acute pancreatitis. There is also a coherence that exenatide does not cause acute pancreatitis from laboratory, clinical, case report, and epidemiologic studies. Analogy to other anti-diabetic drugs does not strengthen the causal hypothesis as other GLP-1 agonists have also been shown not to cause acute pancreatitis from clinical trials.
Strength of association - magnitude of the relative risk estimates observed
Consistency of the association - extent to which scientific results are similar across the entire body of evidence
Biologic gradient (dose-response) - the extent to which the relative risk estimates increase in magnitude as the dose of the exposure increases
Biologic plausibility - the extent to which a mechanism of action has been proposed, studied and demonstrated in toxicological or other laboratory based studies
Specificity - refers to the precision with which the exposure and the outcome can be defined
Coherence - the extent to which the evidence and hypotheses for the results fit together into a reasonable and well-tested explanation
Experimentation - the extent to which a randomized clinical trials or cohort studies are available
Analogy - the extent that the purported exposure-disease relationship under consideration is similar to other relationships
Making reliable causal claims in pharmacovigilance is difficult if not impossible when case reports and case series are used as the primary evidentiary source[15]. While the case reports and series generate hypothesis testing, as was shown for exenatide, it is irresponsible to assign causation based on causal hypothesis[32].
DRUG INDUCED ACUTE PANCREATITIS AND IDIOPATHIC ACUTE PANCREATITIS
Although the vast majority of drugs that have been purported to cause acute pancreatitis probably do not, drug induced acute pancreatitis does exist! When evaluating drugs for causation on the basis of the evidence as described by Hill, two drugs meet the evidence of causation: Azathioprine (and its metabolite 6-mercaptopurine) and 2’3’-dideozyinosine (DDI). The strong evidence comes not from case reports but a consideration of the totality of the evidence, including randomized prospective trials, cohort trials, case reports and a molecular basis[33,34]. For example, in the National Cooperative Crohn’s Disease study, almost 6% of the 116 patients treated with 6-MP developed acute pancreatitis[35]. Similarly, Haber et al[36] treated 400 patients with inflammatory bowel disease with 6-MP and 3.25% developed acute pancreatitis. There are many more randomized trials that support the simplistic case reports.
More recently, Floyd et al[34] performed a large population based study including 1388 patients taking azathioprine and 13836 controls in a single county. The incidence rate for acute pancreatitis among all users of azathioprine was one per 659 treatment year. The crude odds ratio (OR) of having redeemed prescriptions for azathioprine within 90 d before admission for acute pancreatitis was 7.5 (95%CI: 2.6-21.6). After adjustment for gallstone disease, alcohol-related diseases, inflammatory bowel disease, and use of glucocorticoids, the OR increased to 8.4 (95%CI: 2.4-29.4). Although there was a significant risk of persons on azathioprine in developing acute pancreatitis, the population-attributable risk, which measures the proportion of all cases of pancreatitis that are attributable to the use of azathioprine in this study population, was 0.4%.
This finding of less than a half percent attributable risk of azathioprine as a cause of acute pancreatitis is extremely important when considering the claims that drug induced acute pancreatitis accounts for 3%-5% of all cases[37-39]. In the absence of data from controlled clinical trials and large pharmacoepidemiologic trials, there is little to no evidence that other drugs cause acute pancreatitis. Although similar data exists for DDI, the drug is not widely used at this time[40,41]. Therefore, drug induced acute pancreatitis probably accounts for less than 1% of cases, and maybe extremely rare in patients who are not taking obvious drugs.
Premarket approval and post-marketing surveillance has become sensitive to determining complications of drugs such as acute pancreatitis. Randomized controlled trials that evaluate for other complications, such as cardiac complications, would detect significant risks of drugs causing acute pancreatitis[42]. In addition, large pharmacoepidemiologic databasis and meta-analyses are often searched for signals to determine whether drugs cause acute pancreatitis[43].
Azathioprine (and 6-MP) and exenatide represent the two extremes of the data demonstrating a causal association for a drug and acute pancreatitis. While there are case reports in the literature and Medwatch reports on the FARS that both drugs cause acute pancreatitis, only for azathioprine (and 6-MP) have multiple randomized controlled trials and large pharmicoepidemiologic studies showing a statistically significant association. For exenatide (and the other GLP-1 agonists), the opposite is true. Multiple controlled trials, pharmacoepidemiologic databases fail to show any causal association with acute pancreatitis.
While clinicians continue to publish case reports blaming drugs as causing acute pancreatitis, it is important to consider the ideas discussed in this paper. Be critical, cynical and remember that idiopathic pancreatitis is common. Clinicians should perform a thorough workup as described to verify the absence of gallstones, alcoholism, hypertriglyceridemia, tumors. However, the struggle to identify a cause, especially in assigning blame to a drug should be done with extreme caution. When a patient asks “what caused my acute pancreatitis?” Clinicians must remember that almost a third of cases will not be clear and are labeled as idiopathic. As clinicians do not have trouble explaining to patients that “bad luck” is the cause of appendicitis, diverticulitis, cholecystitis, telling a patient that it appears simply “idiopathic” may be correct.
Footnotes
P- Reviewer: Matsuo Y, Olah A, Streba LAM S- Editor: Qi Y L- Editor: A E- Editor: Zhang DN
van Brummelen SE, Venneman NG, van Erpecum KJ, VanBerge-Henegouwen GP. Acute idiopathic pancreatitis: does it really exist or is it a myth?Scand J Gastroenterol Suppl. 2003;117-122.
[PubMed] [DOI][Cited in This Article: ]
Tenner S. Isotretinoin and inflammatory bowel disease: trial lawyer misuse of science and FDA warnings.Am J Gastroenterol. 2014;109:570-571.
[PubMed] [DOI][Cited in This Article: ]
Struve CT. The FDA and the tort system: postmarketing surveillance, compensation, and the role of litigation.Yale J Health Policy Law Ethics. 2005;5:587-669.
[PubMed] [DOI][Cited in This Article: ]
Weed DL, Gorelic LS. The practice of causal inference in cancer epidemiology.Cancer Epidemiol Biomarkers Prev. 1996;5:303-311.
[PubMed] [DOI][Cited in This Article: ]
Haber CJ, Meltzer SJ, Present DH, Korelitz BI. Nature and course of pancreatitis caused by 6-mercaptopurine in the treatment of inflammatory bowel disease.Gastroenterology. 1986;91:982-986.
[PubMed] [DOI][Cited in This Article: ]
Lambert JS, Seidlin M, Reichman RC, Plank CS, Laverty M, Morse GD, Knupp C, McLaren C, Pettinelli C, Valentine FT. 2’,3’-dideoxyinosine (ddI) in patients with the acquired immunodeficiency syndrome or AIDS-related complex. A phase I trial.N Engl J Med. 1990;322:1333-1340.
[PubMed] [DOI][Cited in This Article: ][Cited by in Crossref: 281][Cited by in F6Publishing: 288][Article Influence: 8.5][Reference Citation Analysis (0)]
Maxson CJ, Greenfield SM, Turner JL. Acute pancreatitis as a common complication of 2’,3’-dideoxyinosine therapy in the acquired immunodeficiency syndrome.Am J Gastroenterol. 1992;87:708-713.
[PubMed] [DOI][Cited in This Article: ]
