SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Background

Drug toxicity is a well-known cause of acute pancreatitis (AP). Although many drugs have been associated with AP, the magnitude of the risk of most of them remains largely unknown.

Aim

To determine the pancreatotoxic risk of a wide range of drugs.

Methods

The hospital-based Berlin case–control surveillance study, including all 51 Berlin hospitals in a hospital network, ascertained 102 cases with idiopathic AP (IAP) and 750 controls between 2002 and 2011. Patients with IAP were thoroughly validated using anamnestic, clinical or laboratory data. Drug exposure was obtained in a face-to-face interview. Possible drug aetiology was assessed in individual patients through a standardised causality assessment applying the criteria of the World Health Organization. Drug risks were further quantified [odds ratios (OR) with 95% confidence intervals (CI)] in a case–control design with unconditional logistic regression analysis.

Results

The pancreatotoxic risk of several drugs, including azathioprine (OR 5.1; 95% CI 1.9–13.5), fenofibrate (OR 12.2; 95% CI 2.3–69.1), mesalazine (OR 3.3; 95% CI 1.1–9.5) or angiotensin-converting enzyme inhibitors, was corroborated by case–control analysis and causality assessment. Causality assessment suggested a pancreatotoxic potential, among others, for mercaptopurine or the seldom reported leflunomide, and alluded to a novel risk for tocilizumab. Case–control analysis showed an increased risk for two phytotherapeutics: harpagophytum and valerian radix.

Conclusions

Our study quantified the pancreatotoxic risk of different drugs and phytotherapeutics. The findings corroborate previous results from the literature but also indicate risks for substances not previously reported, highlighting the need for further controlled studies on pancreatic toxicity.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Drug toxicity is a common cause of non-alcohol, nonbiliary acute pancreatitis (AP). Its incidence varies among different studies between 0.1% and 5.3% of all AP cases.[1] As early as in the 1990s, Bergholm et al. reviewed thoroughly cases of drug-induced acute pancreatitis (DIAP) that had been either published or were coming from several spontaneous reporting systems, including, for example, the database of the World Health Organization (WHO). The comparatively scarce data deriving from relevant controlled trials were also taken into consideration.[2] More than 500 different drugs from various classes were found associated with AP, but as a result of the general lack of formal epidemiological studies, the magnitude of the risk of most of these medications remained unknown.

Spontaneous reporting systems and published case reports are a valuable source to provide signals of drug risks, but drug risks cannot be further quantified due to underreporting and lack of a denominator, i.e. lack of information on the drug use in the population.[3] Some other aspects also limit this information: firstly, events regarding newer drugs are more likely to be reported than events regarding established medications, or medications considered ‘harmless’ (e.g. dietary supplements or herbal medications); and secondly, AP cases following treatment with drugs known for their potential pancreatic toxicity will be rather published than AP cases associated with medications lacking prior reports. Furthermore, diagnosing DIAP is a challenging matter considering the several differential diagnoses that have to be excluded, a procedure not always strictly followed in case reports.[1, 2]

Here, we present the results of the hospital-based Berlin case–control surveillance study FAKOS on pancreatic toxicity. Two different methods of identifying drugs that may cause pancreatic damage were used. Firstly, we applied a standardised causality assessment on all individual patients diagnosed with idiopathic AP (IAP) using defined clinical criteria to detect a possible drug-relation of the disease. Secondly, a case–control study was performed to quantify drug risks.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Case identification and recruitment

The hospital-based Berlin case–control surveillance study FAKOS was initiated in 2000 to study serious toxicity of drugs.[4-6] In this study of pancreatic toxicity, it comprised more than 180 Departments of Internal Medicine, Neurology, Psychiatry and Anaesthesiology of all 51 Berlin hospitals from October 2002 to December 2011. The physicians in the respective departments were contacted regularly to identify potential cases of IAP. Cases identified between these intervals were actively reported to the study centre. A trained member of FAKOS contacted eligible patients to obtain their informed consent. In addition, a standardised face-to-face interview was conducted, collecting information on all previous drug intakes, co-morbidity, and other possible risk factors such as chemicals, solvents or smoking. To minimise recall or information bias, which both illustrate potential problems in case–control studies,[7] the standardised questionnaire first asked cases and controls for previous and current diseases and then for drugs taken for these diseases in a second step to increase recall also in the controls. Interviewers were regularly trained and several interviews were thoroughly reviewed afterwards. The study region was Berlin, with 2.8 million adult inhabitants as source population.

Case definition

Patients with a minimum age of 18 years and a new diagnosis of IAP within the last 6 months were included in the study. At least two of the following three criteria had to be met: elevation of lipase or amylase at least threefold above the upper limit of normal (ULN), characteristic upper abdominal pain or signs of pancreatitis in imaging. Exclusion criteria were chronic pancreatitis, biliary aetiology including choledocholithiasis, prior history of biliary colic, dilated biliary tract or concomitant rise of transaminases and bilirubin, other obstruction-related aetiologies of AP such as pancreatic tumours or pancreatic malformations, alcoholic, ischaemic or trauma-induced AP, hyperparathyroidism, massive hypertriglyceridaemia (>11.2 mmol/L) or an endoscopic retrograde cholangiopancreatography in the last 48 h before the onset of symptoms or the appearance of respective laboratory or imaging abnormalities. Due to the difference in drug use in hospital from that in ambulatory setting, there was a distinction between patients who developed IAP in hospital (‘inpatient cases’), and patients who developed IAP in the out-patient setting (‘outpatient cases’) and who were then hospitalised because of their pancreatitis. The index date for both in-patient and out-patient cases was defined as the date of the elevation of pancreatic enzymes as mentioned above or the onset of clinical or imaging signs of AP, whichever occurred earlier.

Case validation and characterisation

Idiopathic AP was validated as certain, probable, possible or unlikely based on clinical information as well as information about laboratory and other diagnostic tests provided on a standardised form by the treating physician to the study centre. The grading of pancreatitis as oedematous or necrotising was based on imaging findings.

Control selection

Control selection was conducted from January 2002 to December 2011 at the same hospitals where cases were recruited to evaluate drug risks also in a case–control approach. For this ascertainment, 80 Departments of Surgery and Orthopaedics additionally participated in the study. The aim was to raise a multiple of controls compared with cases to increase study power. The control/case ratio was approximately 7:1, as higher ratios lead only to marginal increases in power.[8] An extensive list of possible control diseases was set up for the selection of controls to represent the drug use in the population at large. Incident and not prevalent diseases were preferred to exclude control patients with chronic administration of specific drugs. Moreover, a wide mixture of orthopaedic, neurological or internal control diseases was pursued. The prevalences of the different groups of control diseases are illustrated in Table S2. Informed consent was obtained and control patients were subsequently interviewed in the same standardised way as the IAP patients. As the number of in-patient cases was low (Figure 1), no case–control analysis was conducted for these cases and thus no controls for in-patient cases were recruited. For the case–control analysis in out-patient cases, hospitalised controls were recruited and out-patient drug use was documented; the index date was defined as the date of hospitalisation or the date of the onset of the control disease if this preceded hospitalisation. Patients with a diagnosis of any pancreatic disease were excluded as controls.

image

Figure 1. Study design. IAP, idiopathic acute pancreatitis; WHO, World Health Organization.

Download figure to PowerPoint

Standardised assessment of drug causality in individual cases

A possible drug aetiology was assessed for each case in a standardised causality assessment according to the criteria of the WHO assessment method, which comprises the categories ‘certain’, ‘probable’, ‘possible’, ‘unlikely’, ‘unclassified’ and ‘unclassifiable’.[9] A drug reaction was evaluated as ‘certain’, when the time relationship with drug intake was plausible, other causes could be ruled out, a clinically reasonable response on drug withdrawal (‘positive dechallenge’) was observed and AP was observed on re-exposure to the same drug (‘positive rechallenge’). The causality assessment was ‘probable’, when AP occurred with a reasonable time sequence to administration of the drug and it was unlikely to be attributed to other causes, and a positive dechallenge reaction was observed on drug withdrawal. The drug reaction was ‘possible’, when there was a plausible time sequence to drug intake; however, another cause could not be ruled out and information on dechallenge was lacking, not clear or negative. Suspected drugs were grouped according to the Anatomical Therapeutic Chemical Classification System[10] and analysed by descriptive statistics.

Case–control analysis

The case–control analysis included all cases validated as certain or probable irrespective of the result of causality assessment. Out-patient cases and controls were considered as exposed to a drug, if the drug had been taken in the last 7 days before the index date. Odds Ratio (OR) and 95% confidence interval (CI) of IAP associated with exposure to specific drugs was calculated with unconditional logistic regression analysis adjusting for age and sex (‘single drug assessment’) and for age, sex and all other drugs, which were significant in the single drug assessment in the first analysis (‘joint drug assessment’). All risk calculations were performed with SAS statistical software package (version 9.1, sas Institute Inc, Cary, NC, USA). A two-tailed P value < 0.05 was considered significant. The Ethics Committee of the Charité-Universitätsmedizin Berlin approved the study.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Overall, 284 patients with possible IAP were notified to the study centre during the study period. Of those, 168 patients were not included in the study for the reasons outlined in Figure 1, the most common among them being alcohol-induced pancreatitis. In four cases, IAP was validated only as possible, so these cases were not further considered in the case–control analysis. As the number of in-patient cases was low (n = 10), a case–control analysis was not conducted for these cases. The final population for the case–control analysis thus included 102 out-patient cases with certain or probable IAP and 750 controls.

Drug causality assessment in individual cases

In a total of 64 cases, the drug relationship was judged as at least possible. A total of 68 different drugs were identified in these patients. Of those, three drugs/drug combinations were assessed as certainly related to AP (azathioprine, mercaptopurine and the combination of olmesartan medoxomil plus amlodipin). Azathioprine also showed three probable associations with AP, while mesalazine and the combination of dienogest plus ethinylestradiol showed two probable associations with AP and 16 further drugs showed one probable association with AP. Altogether, antineoplastic and immunomodulating agents (21%), drugs of the cardiovascular system (18%) and anti-infectives (16%) were the groups represented the most (Table 1; Table S1). In 38 cases, there was only one offending drug, with the duration of use strongly varying from few days to several months (data not shown). Drug dosage also spanned a wide range; for example, azathioprine-induced AP was observed in patients receiving from 25 to 150 mg of this agent per day (data not shown).

Table 1. Drugs with certain or probable causal relationship in suspected drug-induced acute pancreatitis according to the individual causality assessment categorised by main groups of the Anatomical Therapeutic Chemical (ATC) Classification System
ATC groupCertainaProbable
  1. a

    Positive rechallenge.

Alimentary tract and metabolism (A)
Budesonide 1
Sulfasalazine 1
Mesalazine 2
Exenatide 1
Cardiovascular system (C)
Lisinopril 1
Olmesartan medoxomil plus amlodipine1 
Fenofibrate 1
Ezetimibe plus simvastatin 1
Systemic hormonal preparations (G/H)
Dienogest plus ethinylestradiol 2
Antiinfectives (J)
Cefuroxime 1
Lopinavir plus ritonavir 1
Raltegravir 1
Antineoplastic and immunomodulating agents (L)
Mercaptopurine1 
Mycophenolic acid 1
Leflunomide 1
Tocilizumab 1
Azathioprine13
Nervous system (N)
Propofol 1
Paracetamol plus codeine 1
Valproic acid 1
Olanzapine 1

Case–control analysis

For the case–control analysis, we used the final population of 102 out-patient cases, which were compared with 750 out-patient controls. The mean age of out-patient cases and controls was 49 and 55 years respectively. Forty-eight per cent of the out-patient cases were females; the respective number in the out-patient controls was 51.5% (Table 2). Table 2 also depicts further characteristics of out-patient cases, including grading of AP, associated laboratory findings and clinical symptoms, complications, imaging findings and outcome.

Table 2. Characteristics of acute pancreatitis in out-patient cases
CharacteristicsOut-patient cases; n = 102
  1. a

    Remaining cases either negative or unclassifiable.

  2. b

    A biliary aetiology was ruled out by endoscopic ultrasound of the biliary tract in two cases.

  3. c

    A biliary aetiology was ruled out by endoscopic ultrasound of the biliary tract.

Sex, n (%)
Male53 (52)
Female49 (48)
Mean age (standard deviation), years49.3 (18.3)
Grading, n (%)
Oedematous64 (62.7)
Necrotising14 (13.7)
Unclassifiable24 (23.5)
Lipase × ULN; median (interquartile range)19 (7.3–36.1)
C reactive protein × ULN; median (interquartile range)12.8 (2.2–39.2)
Leukocytosisa (n, %)66 (64.7)
Characteristic abdominal symptomsa (n, %)97 (95.1)
Infectious complicationsa (n, %)17 (16.7)
Organ failurea (n, %)10 (9.8)
Biliary sludgea, b (n, %)9 (8.8)
Asymptomatic cholecystolithiasisa,c (n, %)1 (1)
Recoverya (n, %)90 (88.2)
Deatha (n, %)3 (3)
Death with acute pancreatitis as certain causea2 (2)

Next, the prevalence of diabetes mellitus (DM), as well as the percentage of current or past smokers, was examined in out-patient cases and controls. We found no significant differences between the two groups, as the prevalence of DM was 18.6% in the out-patient cases and 16.3% in the out-patients controls (P > 0.05), and the percentage of current or past smokers was 63.8% in the out-patient cases and 69.5% in the out-patient controls (P > 0.05).

Risk estimates for drugs obtained from the case–control analysis are shown in Table 3. The highest adjusted risk estimates were calculated for fenofibrate (OR 12.2; 95% CI 2.3–69.1), harpagophytum (devil's claw) (OR 12.0; 95% CI 1.9–74.3), valerian radix (OR 10.3; 95% CI 1.7–53.4), the combination of formoterol and budesonide (OR 6.3; 95% CI 1.1–32.1), the combination of ramipril and hydrochlorothiazide (OR 5.8; 95% CI 1.7–18.0), lisinopril (OR 5.4; 95% CI 1.4–17.5) and azathioprine (OR 5.1; 95% CI 1.9–13.5).

Table 3. Odds ratios of drug-induced acute pancreatitis for different drugs based on out-patient cases and controls (numbers in bold indicate significant results)
ATC Group102 cases; n (%)750 controls; n (%)ORa (95% CI)ORb (95% CI)
  1. a

    Single drug assessment (adjusted for age and sex).

  2. b

    Joint drug assessment (adjusted for age, sex and all drugs with a significant odds ratio in single drug assessment).

Alimentary tract, metabolism/Blood (A/B)
Proton pump inhibitors17 (16.7)94 (12.5)1.6 (0.9–2.8) 
Omeprazole7 (6.9)27 (3.6)2.2 (0.9–5.1) 
Pantoprazole6 (5.9)42 (5.6)1.2 (0.4–2.7) 
Esomeprazole3 (2.9)25 (3.3)0.9 (0.2–2.8) 
Mesalazine7 (6.9)10 (1.3) 4.5 (1.6–12.1) 3.3 (1.1–9.5)
Short-acting insulin3 (2.9)6 (0.8)4.7 (0.96–10.4) 
Metformin3 (2.9)25 (3.3)1.1 (0.3–3.3) 
Calcium carbonate plus cholecalciferol3 (2.9)23 (3.1)1.1 (0.3–3.3) 
Magnesium3 (2.9)61 (8.1)0.4 (0.1–1.0) 
Acetylsalicylic acid14 (13.7)124 (16.5)1.1 (0.6–2.1) 
Cardiovascular system (C)
Hydrochlorothiazide4 (3.9)16 (2.1)2.5 (0.7–7.1) 
Furosemide5 (4.9)36 (4.8)1.2 (0.4–3.0) 
Torasemide5 (4.9)48 (6.4)1.0 (0.3–2.4) 
Metoprolol11 (10.8)112 (14.9)0.8 (0.4–1.6) 
Bisoprolol7 (6.9)34 (4.5)2.0 (0.8–4.6) 
Nebivolol3 (2.9)18 (2.4)1.7 (0.4-5.2) 
Amlodipin5 (4.9)33 (4.4)1.4 (0.5–3.4) 
Angiotensin-converting enzyme inhibitors (monopreparations)22 (21.6)127 (16.9) 1.9 (1.08–3.3)  
Enalapril (monopreparation)7 (6.9)42 (5.6)1.7 (0.7–3.7) 
Lisinopril (monopreparation)4 (3.9)10 (1.3) 3.9 (1.04–12.3) 5.4 (1.4–17.5)
Ramipril (monopreparation)10 (9.8)59 (7.9)1.6 (0.7–3.2) 
Ramipril plus hydrochlorothiazide6 (5.9)11 (1.5) 5.8 (1.9–16.3) 5.8 (1.7–18.0)
Simvastatin8 (7.8)58 (7.7)1.3 (0.6–2.8) 
Atorvastatin3 (2.9)26 (3.5)1.0 (0.2–3.0) 
Fenofibrat4 (3.9)3 (0.4) 12.4 (2.6–64.6) 12.2 (2.3–69.1)
Systemic hormonal preparations (G/H)
Levonorgestrel plus ethinylestradiol (monophasic combinations)3 (2.9)7 (0.9)2.3 (0.5–8.9) 
Prednisolone8 (7.8)38 (5.1)1.7 (0.7–3.6) 
Levothyroxine9 (8.8)69 (9.2)1.1 (0.5–2.3) 
Levothyroxine incl. combinations12 (11.8)79 (10.5)1.3 (0.7–2.5) 
Antiinfectives (J)
Cephalosporins2 (2.0)5 (0.7)3.5 (0.5–17.1) 
Antineoplastic and immunomodulating agents (L)
Azathioprine9 (8.8)11 (1.5) 5.8 (2.3–14.7) 5.1 (1.9–13.5)
Musculo-skeletal and nervous system (M/N)
Diclofenac4 (3.9)25 (3.3)1.4 (0.4–3.7) 
Ibuprofen7 (6.9)33 (4.4)1.6 (0.6–3.6) 
Allopurinol3 (2.9)42 (5.6)0.6 (0.1–1.8) 
Harpagophytum (devil's claw)3 (2.9)3 (0.4) 10.0 (1.8–56.5) 12.0 (1.9–74.3)
Acetylsalicylic acid incl. combinations4 (3.9)21 (2.8)1.3 (0.4–3.7) 
Metamizole4 (3.9)17 (2.3)2.0 (0.6–5.7) 
Valerian radix incl. combinations3 (2.9)4 (0.5) 8.0 (1.5–38.3) 10.3 (1.7–53.4)
Rest
Formoterol plus budesonide3 (2.9)4 (0.5) 7.0 (1.4–32.9) 6.3 (1.1–32.1)
Budesonide3 (2.9)8 (1.1)3.1 (0.7–11.3) 
Timolol3 (2.9)5 (0.7) 6.1 (1.2–25.9) 5.5 (0.9–27.2)

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Although more than 500 different medications have been reported to be associated with AP, the risk of most of these drugs remains unclear, as many data derive from case reports, case series or summaries of them.[1] In 2003, Lancashire et al. conducted one of the first case–control studies in this field using the UK General Practitioner Research Database (GPRD). Their results indicated that a few drugs commonly reported as suspected to cause AP either did not have an increased risk at all (statins), or did not have an increased risk compared with other, more seldom reported drugs of the same class (valproic acid),[11] highlighting the need for further epidemiological studies. Possible limitations of this study were that it could not verify the diagnosis of AP in the cases and that it did not exclude alternative aetiologies of AP such as alcohol or gallstones, whereas our study included cases of IAP based on laboratory, clinical or imaging findings in which alternative aetiologies had been ruled out.

Based on the case–control analysis or the individual causality assessment, we corroborated known risks for a number of drugs, including azathioprine,[12] mesalazine,[11] mercaptopurine[13] and ACE inhibitors.[14] Furthermore, the results indicated a pancreatic toxicity also for substances rarely being reported associated with AP, such as fenofibrate[15] or leflunomide.[16] Moreover, this study is the first one to the best of our knowledge alluding to a pancreatotoxic potential for the herbal remedies harpagophytum and valerian radix as well as for tocilizumab. As more than 90% of the patients included in the case–control analysis were recruited until the end of 2010, the pancreatotoxic risk of newer drugs such as the antidiabetic agents gliptins or glucagon-like peptide-1 (GLP-1) agonists could not be quantified. However, we did find one case of exenatide-induced AP with probable causality. Similarly, some of the older drugs not commonly used nowadays may not have been assessed well.

Analysis of the DIAP cases with only one offending drug regarding duration of drug use led to the detection of a wide range spanning from few days to several months. Hence, we corroborate previous findings that suggest an absence of fixed duration of drug use in DIAP.[17]

Data on drug dosage in the same DIAP cases imply that pancreatic toxicity relies rather on an idiosyncratic and not a dose-dependent mode, but the small amount of respective data makes further conclusions challenging.

In all three patients with a positive rechallenge, the offending medication was re-administered as it was initially unclear whether drug toxicity had been the cause of AP. Thus, the respective medical therapy was re-initiated to treat the underlying diseases.

The thiopurines azathioprine and mercaptopurine, as well as the derivative of salicylic acid mesalazine, have been repeatedly reported being associated with AP in case reports in the past. Several times, a positive rechallenge was documented, and, therefore, a certain causality was indicated.[18, 19] Azathioprine has also shown an increased risk in the population-based case–control study conducted by Floyd et al.[12] Moreover, in a retrospective cohort study including only patients with inflammatory bowel disease (IBD), thiopurine medication was the main aetiological factor for the development of AP.[20] Concerning mesalazine, its association with AP remains controversial. Muff Munk et al. found no increased risk for mesalazine using data from a Danish hospital discharge registry.[21] In contrast, the UK study on AP mentioned above showed a nine-fold increased risk in patients receiving mesalazine up to 3 months before the onset of the disease.[11]

Our case–control analysis showed significantly increased risks for angiotensin-converting enzyme (ACE) inhibitors as a group, for lisinopril, as well as for the ramipril/hydrochlorothiazide combination. Based on the respective 95% CI, increased risk estimates cannot be ruled out also for ramipril and enalapril. Furthermore, lisinopril was considered once probably and once possibly associated with AP in the standardised causality assessment of individual cases. These findings corroborate the results of a European, multi-centre, population-based case–control study, which has shown a slightly increased risk for ACE inhibitors.[14] Moreover, this is the first study to our knowledge detecting a significant risk estimate for lisinopril; in recent reviews on DIAP based on case reports, this ACE inhibitor was either not considered as a possible cause at all[1] or it has been ascribed a comparably low risk.[18]

Fenofibrate illustrated the highest risk estimate in our case–control analysis, whereas the two statins included (simvastatin, atorvastatin) showed either no increased risk or failed to reach statistical significance. These findings are consistent with a large retrospective cohort study, showing that fenofibrate has a several fold increased risk compared with statins.[15] On the other hand, a recent meta-analysis depicted no significantly increased risk for fibrates. However, as the studies recruited were randomised controlled trials evaluating the effects of fibrates on cardiovascular events, incidence of pancreatitis was not documented in a standardised way, posing the question of possible variation in results among trials.[22]

Interestingly, several cephalosporins, including cefuroxime, cefotaxime and cefixime, showed a possible or probable causal relationship in our causality assessment of individual cases. To this day, ceftriaxone has been the only antibiotic of this class to be reported associated with AP,[23-28] possibly via formation of gallstones.[24] As cases of biliary AP were excluded from our study, it seems that this pathomechanism of ceftriaxone-induced AP may not apply to all other cephalosporins. In the case–control analysis, we demonstrated a nonsignificantly increased risk estimate for cephalosporins as a group – however, the low number of exposed cases makes an interpretation of this result challenging.

A potential risk for AP was found for several herbal medications or dietary supplements as well. In the case–control analysis, high-risk estimates were demonstrated for harpagophytum (devil's claw), an herb mostly used as an analgesic,[29] and valerian radix, an herb commonly taken by patients with sleeping disorders.[30] In addition, the individual causality assessment revealed one possible association with AP for harpagophytum, for cinnamon powder, and for hypericum perforatum (St. John's wort) respectively. Interestingly, the secondary metabolites iridoids, which represent harpagophytum's main ingredients and are also an important constituent in valerian radix, have a well-established choleretic activity.[31] This increases the risk of gallstone formation, and could eventually lead to the development of AP. With AP of biliary origin illustrating an exclusion criterion in our study, a possible explanation could be the development of asymptomatic biliary sludge. The US prescribing information of harpagophytum contraindicates this herb for patients suffering from gallstones, but does not indicate a contraindication for patients with pancreatic disease.[32] Furthermore, several cases of hepatotoxicity associated with valerian radix have been published in the past,[33-35] indicating a broader toxic potential of this herbal remedy. In contrast to herbal hepatotoxicity, reports of AP associated with either herbal medications or dietary supplements are rare,[36-40] a possible cause being the less stringent surveillance of such remedies. Taken together, this is the first controlled study showing an increased risk of AP for two well-known herbals. However, as the possibility of recall bias especially regarding the use of phytotherapeutics cannot be excluded in a case–control study, the results at hand should be interpreted with caution. To assess possible risks of other herbal medications or dietary supplements, further effort is needed.

DM and smoking have been associated with an increased risk for AP.[41, 42] However, it appears unlikely that difference in the prevalence of these important risk factors between cases and controls affected the current results, as the prevalence of DM and the frequency of current or past smokers did not significantly differ between out-patient cases and controls.

Several weaknesses and strengths of this study should be addressed. Biliary sludge or asymptomatic cholecystolithiasis were no exclusion criteria, as such findings can be also seen in AP patients with probable or certain drug causality.[37, 43, 44] A biliary aetiology was excluded via endoscopic ultrasound of the biliary tract in two of the nine patients with biliary sludge and in the single patient with asymptomatic cholecystolithiasis, whereas no further signs of biliary aetiology were found in the remaining seven patients with biliary sludge (for example choledocholithiasis, prior history of biliary colic, dilated biliary tract or concomitant rise of transaminases and bilirubin). However, as biliary sludge depicts a possible cause of AP in cases initially labelled as idiopathic,[45] this remains a potential limitation of the study. We used hospital-based controls, although they may not represent a community setting, as they are usually associated with higher response rates and lower costs compared with community-based controls.[7]

Regarding the drugs for the treatment of IBD (mesalazine, azathioprine and mercaptopurine), confounding by indication cannot be ruled out, as IBD is an established risk factor for AP.[21, 46] However, another case–control study found an increased risk for azathioprine after adjusting for IBD, indicating a ‘genuine’ risk for this drug,[12] and most probably for its active metabolite mercaptopurine. Regarding mesalazine, a quantification of its pancreatotoxic risk in non-IBD patients was not possible, as ulcerative colitis and Crohn's disease are the only indications for its use in Germany.[47] Therefore, all patients exposed to mesalazine in our study had an IBD diagnosis.

The thorough validation of the AP cases as well as the strict inclusion and exclusion criteria reduced the number of cases considerably, resulting in decreased statistical power for the case–control analysis. Therefore, precaution in the interpretation of results is needed. Moreover, it cannot be ruled out that recall and information bias, and unmeasured confounding may explain the findings of this study.

The increased risk estimate for the combination inhaler formoterol/budesonide is possibly a statistical artefact, although systemic adverse effects of inhaled corticosteroids are well established,[48] and corticosteroid-induced pancreatitis for oral preparations has been already described in the 1960s.[49]

FAKOS is one of the few controlled studies on pancreatic toxicity and merely the second one with a case–control design not narrowing its focus to specific drugs. This allowed the risk quantification of various drugs or herbal medications, including also substances that had been hardly reported or not reported at all so far. It is not uncommon for case–control studies on drug toxicity to select their ‘target’ medications based on the quantity of published case reports. However, drugs that are considered ‘harmless’ (e.g. herbal medications) are usually not monitored as thoroughly as substances with known serious adverse effects. Due to the high number of participating hospitals and the variety of medical departments, we were able to assess a wide range of drugs.

Regarding the individual causality assessment, it was always carried out by the study physician in a standardised way. Furthermore, it was not based on a suspicion of the attending physician, thus precluding a potential bias because of prior knowledge and attention to drugs as a possible cause of disease.

In summary, our study identified a few drugs as well as herbal medicines as possible causes of AP. Our findings corroborate previous results from the literature, but also indicate risks for substances not reported so far, highlighting the need for further controlled studies on pancreatic toxicity. In case of characteristic abdominal pain or an elevation of the respective laboratory parameters, drugs should be part of the differential diagnosis, especially when alcohol intake or gallstones can be ruled out as possible causes.

Authorship

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Guarantors of the article: Edeltraut Garbe, Antonios Douros.

Author contributions: Edeltraut Garbe, Frank Andersohn, Elisabeth Bronder, Andreas Klimpel and Johann Ockenga contributed to study concept and design. Elisabeth Bronder, Frank Andersohn and Michael Thomae acquired the data. Andreas Klimpel and Antonios Douros analysed and interpreted the data. Antonios Douros drafted the manuscript. All authors revised critically the manuscript. Edeltraut Garbe, Elisabeth Bronder and Reinhold Kreutz supervised the study. All authors approved the final version of the article, including the authorship list.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Declaration of personal interests: Edeltraut Garbe had consultant arrangements with Bayer AG, Novartis and Teva in the past and currently with Schwabe and has received grant/research support from Bayer AG and Novartis. Reinhold Kreutz has consultant/advisory arrangements with Bayer Pharma, Berlin-Chemie, Daiichi and BMS. Frank Andersohn has consultant arrangements with Abbot Products and AstraZeneca.

Declaration of funding interests: This study was funded fully by the Federal Institute for Drugs and Medical Devices in Germany (BfArM), grant number: V-5238/68605/2012. The BfArM played no role in the study design, collection, analysis, and interpretation of the data and in the writing of the report.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Authorship
  8. Acknowledgements
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
apt12461-sup-0001-TableS1.docWord document69KTable S1. Drugs with possible causal relationship in suspected drug-induced acute pancreatitis according to the individual causality assessment categorised by main groups of the Anatomical Therapeutic Chemical (ATC) Classification System.
apt12461-sup-0002-TableS2.docWord document33KTable S2. Classification of the control diseases (n = 750) based on the International Statistical Classification of Diseases and Related Health Problems (ICD 10) with prevalences and most frequent diagnoses of each group.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.