Importance Acute pancreatitis has significant morbidity and mortality. Previous studies have raised the possibility that glucagonlike peptide 1 (GLP-1)–based therapies, including a GLP-1 mimetic (exenatide) and a dipeptidyl peptidase 4 inhibitor (sitagliptin phosphate), may increase the risk of acute pancreatitis.
Objective To test whether GLP-1–based therapies such as exenatide and sitagliptin are associated with an increased risk of acute pancreatitis. We used conditional logistic regression to analyze the data.
Design Population-based case-control study.
Setting A large administrative database in the United States from February 1, 2005, through December 31, 2008.
Participants Adults with type 2 diabetes mellitus aged 18 to 64 years. We identified 1269 hospitalized cases with acute pancreatitis using a validated algorithm and 1269 control subjects matched for age category, sex, enrollment pattern, and diabetes complications.
Main Outcome Measure Hospitalization for acute pancreatitis.
Results The mean age of included individuals was 52 years, and 57.45% were male. Cases were significantly more likely than controls to have hypertriglyceridemia (12.92% vs 8.35%), alcohol use (3.23% vs 0.24%), gallstones (9.06% vs 1.34), tobacco abuse (16.39% vs 5.52%), obesity (19.62% vs 9.77%), biliary and pancreatic cancer (2.84% vs 0%), cystic fibrosis (0.79% vs 0%), and any neoplasm (29.94% vs 18.05%). After adjusting for available confounders and metformin hydrochloride use, current use of GLP-1–based therapies within 30 days (adjusted odds ratio, 2.24 [95% CI, 1.36-3.68]) and recent use past 30 days and less than 2 years (2.01 [1.37-3.18]) were associated with significantly increased odds of acute pancreatitis relative to the odds in nonusers.
Conclusions and Relevance In this administrative database study of US adults with type 2 diabetes mellitus, treatment with the GLP-1–based therapies sitagliptin and exenatide was associated with increased odds of hospitalization for acute pancreatitis.
The glucagonlike peptide 1 (GLP-1)–based therapies include a GLP-1 mimetic, exenatide, and a dipeptidyl peptidase 4 inhibitor, sitagliptin phosphate. These therapies are used in patients with type 2 diabetes mellitus to lower blood glucose levels. Exenatide is a subcutaneous injectable GLP-1 agonist that enhances glucose-mediated insulin secretion. Sitagliptin is an oral dipeptidyl peptidase 4 inhibitor that prevents the proteolysis of endogenous GLP-1.1
Sitagliptin and exenatide have been shown to cause acute pancreatitis in rodent models via amplification of ductal replication and induction of acinar to ductal metaplasia.2-5 Mouse models have also shown that GLP-1–based therapy may induce focal proliferation in the exocrine pancreas and accelerate formation of dysplastic lesions and pancreatitis.6 Reports of spontaneous acute pancreatitis in humans have been published7-9; however, the strength of this association and causality cannot be inferred from these reports.
Previous observational studies of this association have yielded inconsistent results owing to limited statistical power, inadequate duration of follow-up, or inadequate adjustment for confounders such as obesity and diabetes disease severity.10-14 The clinical trials of sitagliptin also had inconsistent findings owing to poor ascertainment and limited statistical power.15,16 Acute pancreatitis is more prevalent among patients with type 2 diabetes mellitus compared with a nondiabetic cohort.17 Thus, clinical uncertainty remains about whether the association between sitagliptin or exenatide and acute pancreatitis is causal. Given the significant risk of morbidity and mortality with acute pancreatitis, studies should determine whether the GLP-1–based therapies increase this risk.
Our objective was to determine the association of acute pancreatitis with the use of exenatide or sitagliptin in adults with type 2 diabetes mellitus. We aimed to test the hypothesis that 2 GLP-1–based therapies, one a GLP-1 mimetic (exenatide) and the other a dipeptidyl peptidase 4 inhibitor (sitagliptin), may increase the risk of acute pancreatitis.
We used administrative claims data from 7 Blue Cross Blue Shield Association (BCBS) plans. These are BCBS of Tennessee, Hawaii, Michigan, and North Carolina; Highmark, Inc, and Independence Blue Cross of Pennsylvania; and Wellmark, Inc, of Iowa and South Dakota.
Study design and population
We assembled a population of BCBS beneficiaries with type 2 diabetes mellitus who filled at least 1 prescription for any drug used to treat type 2 diabetes from February 1, 2005, through December 31, 2008. We classified patients as having type 2 diabetes mellitus if they had 1 relevant inpatient code from the International Classification of Disease, Ninth Revision (ICD-9) or 2 outpatient ICD-9 codes separated by at least 30 days (relevant codes included 250.xx, 648.0 [diabetes mellitus with pregnancy], 362.0 [diabetic retinopathy], and 266.41 [diabetic cataract]). Individuals were also classified as having diabetes mellitus if they filled a prescription for a medication to treat hyperglycemia (eTable 1). If the prescription was for metformin hydrochloride alone, the individual was also required to have an ICD-9 code for diabetes for inclusion in this group. Individuals who had ICD-9 codes for type 1 diabetes mellitus (250.x3) or gestational diabetes (648.83) only were excluded. To be eligible for inclusion, individuals had to be aged 18 to 64 years on the date of their first code for diabetes, contribute at least 6 months of medical or pharmacy coverage in the calendar year with a diabetes code, and of known sex. We excluded participants older than 64 years because their information concerning the use of health care resources was incomplete owing to insurance in addition to BCBS coverage (typically Medicare).
We determined eligibility for the study from computerized encounter data, including enrollment files for administrative data; benefits information to determine medical and pharmacy coverage; and inpatient, outpatient, and pharmacy claims records containing Current Procedural Terminology codes, ICD-9 codes, and National Drug Codes, or diagnosis related group codes and costs and charges (submitted, allowed, and paid).
Inclusion/exclusion criteria for cases and controls
Briefly, we identified presumptive cases by using a validated algorithm of ICD-9 and Current Procedural Terminology codes for acute pancreatitis18 (eTable 2). Cases were defined as individuals with an inpatient code for acute pancreatitis at any time after the first exposure to the drugs of interest. For patients with multiple episodes of acute pancreatitis, we included only the first episode. We excluded pancreatitis occurrences within 3 months of enrollment.
We randomly selected 1 control subject for each case from the eligible source population matched on age within 10 years, sex, insurance plan site, Diabetes Complication Severity Index (0, 1, 2, or ≥3),19-21 and enrollment pattern or duration of follow-up (incidence density sampling). Acute pancreatitis cases were not eligible to undergo resampling as controls.
Prescription data were used as an indicator of drug exposure. We obtained information about drug use from a computerized pharmacy database containing the date the prescription was filled and the supply (measured as number of days). We used these data elements to determine the dates that a patient was exposed to the drug before the first observed diagnosis of acute pancreatitis. Drug exposure was defined as having filled a prescription for sitagliptin or exenatide before the first observed diagnosis of pancreatitis. An individual with exposure to sitagliptin or exenatide after the index diagnosis of acute pancreatitis was counted as unexposed. We defined 4 categories of exposure. Any users were exposed to sitagliptin or exenatide after the diagnosis of diabetes mellitus and before the index date of pancreatitis. Current users were exposed to sitagliptin or exenatide within 30 days before the index date of onset of pancreatitis. Recent users had a claim for sitagliptin or exenatide ranging from 30 days to 2 years before the index date of onset of pancreatitis. Nonusers had no sitagliptin or exenatide prescription more than 2 years before the index date of pancreatitis. Current users, recent users, and nonusers were mutually exclusive categories of exposure. However, any users included current and recent users.
We calculated proportions for categorical variables and means or medians for continuous variables according to case or control status. We used conditional logistic regression to estimate the McNemar odds ratio (OR) for exposure to sitagliptin or exenatide to account for the matched design of the study. We conducted analyses for recent and current sitagliptin and exenatide users and for any users in a series of statistical models.
We adjusted statistical models for the matching variables, potential confounders specified a priori and identifiable in claims data, and metformin exposure during the same period. The confounders were hypertriglyceridemia, alcohol use, gallstones, tobacco abuse, obesity, biliary and pancreatic cancer, cystic fibrosis, and an indicator of general morbidity level (the resource utilization band from The Johns Hopkins adjusted clinical group case-mix system)20 (eTable 3). In a sensitivity analysis, we excluded the cases and controls exposed to a combination of sitagliptin and metformin to test the robustness of our results.
We used commercially available statistical software (SAS, version 9.2; SAS Institute, Inc) for analysis. Statistical significance was set at 2-sided P < .05. We obtained ethical approval from the institutional review board at The Johns Hopkins University. This study was approved by the BCBS advisory board.
We identified 1 100 899 patients with type 2 diabetes mellitus who were potentially eligible during the study period. Within this group, 3004 participants experienced an episode of inpatient pancreatitis during the study period. We identified 1708 individuals with pancreatitis as our cases, after we excluded beneficiaries older than 64 years or younger than 18 years (n = 700), those who did not have 6 or more months of coverage in the calendar year of the starting date (n = 256), those with missing information on sex (n = 159), those with a diagnosis of diabetes after the date of the first pancreatitis diagnosis (n = 443), and those with onset of pancreatitis earlier than May 1, 2005 (n = 1) (some patients met more than 1 exclusion criterion). We matched 1680 controls selected from a pool of 631 779 potentially eligible controls. Ninety-three case-control pairs were excluded because they had no verifiable drug coverage during the study period, and 318 case-control pairs were excluded because the pancreatitis diagnosis was within 90 days of the index date (date of diabetes mellitus diagnosis or May 01, 2005, whichever was later). Thus, the analysis included 1269 each cases and controls (Figure). The mean age of included beneficiaries was approximately 52 years, and 57.45% were male (Table 1).
Cases with pancreatitis were statistically significantly more likely than controls to have hypertriglyceridemia (12.92% vs 8.35%), alcohol use (3.23% vs 0. 24%), gallstones (9.06% vs 1.34%), tobacco abuse (16.39% vs 5.52%), obesity (19.62% vs 9.77%), biliary and pancreatic cancer (2.84% vs 0%), cystic fibrosis (0.79% vs 0%), and neoplasms (29.94% vs 18.05%). Cases and controls also differed significantly in the use of health care resources.
The use of the studied GLP-1–based therapies associated with acute pancreatitis in cases and controls in various exposure windows is shown in Table 2.
Odds of hospitalization for acute pancreatitis
After adjusting for available confounders, including metformin use, current use within 30 days (adjusted OR [AOR], 2.24 [95% CI, 1.36-3.69]; P = .01) and recent use (2.01 [1.37-3.18]; P = .01) were associated with statistically significantly higher odds of acute pancreatitis. Any use was also associated with statistically significantly higher odds of acute pancreatitis (AOR, 2.07 [95% CI, 1.36-3.13]; P = .01).
The results of the sensitivity analysis with 1260 each cases and controls after exclusion of 9 pairs who were exposed to metformin (along with sitagliptin) were consistent with the primary analysis. After adjusting for confounders and metformin use, current use within 30 days (AOR, 2.01 [95% CI, 1.19-3.38]; P = .01) and recent use (1.95 [1.21-3.14]; P = .01) were associated with a statistically significant odds of hospitalization for acute pancreatitis. Any use was also associated with statistically significantly higher odds of acute pancreatitis (AOR, 2.02 [95% CI, 1.31-3.01]; P = .01).
Our findings suggest a significantly increased risk of hospitalization for acute pancreatitis associated with the use of sitagliptin or exenatide among adult patients with type 2 diabetes mellitus. Our results support findings from mechanistic studies and spontaneous reports submitted to the US Food and Drug Association that such an association may be causal.5
Our findings need to be interpreted in the context of other studies. A cross-sectional analysis of spontaneous adverse events reported that the use of sitagliptin or exenatide increased the odds for reported pancreatitis 6-fold compared with the use of other antidiabetic drugs.10 However, that study did not have data on confounders. An administrative claims–based study reported no difference in acute pancreatitis among exenatide users, sitagliptin users, and controls but could only adjust for a limited set of confounders.11 Another administrative claims study reported no difference in acute pancreatitis among exenatide users, sitagliptin users, and controls owing to limited statistical power and limited duration of follow-up.12 Another cohort study reported no significantly increased risk of acute pancreatitis with exposure to exenatide relative to other antihyperglycemics after propensity score adjustment but had limited statistical power and a limited exposure window.13 In that study, the use of exenatide beyond 32 days was associated with a significantly increased risk of pancreatitis compared with the use of other antihyperglycemics. The study did not evaluate the role of sitagliptin. A descriptive database study reported no difference in the rates of acute pancreatitis among exenatide users (0.44%), sitagliptin users (0.28%), or controls (0.39%) after a follow-up of 540 days without any statistical tests or adjustment for confounders.14 A meta-analysis of clinical trials reported no difference for sitagliptin use compared with placebo or other oral hypoglycemics in the incidence rates of pancreatitis (0.1% and 0%, respectively), acute pancreatitis (0% and 0.1%, respectively), and chronic pancreatitis (0.1% and 0%,respectively).15 However, that study was limited by inadequate statistical power and incomplete ascertainment of adverse events in clinical trials.16
Although we adjusted for diabetes severity, residual confounding is possible, and we recognize that tobacco abuse, obesity, and alcohol abuse are markedly undercoded in claims. Misclassification of case and control status is possible because we did not have access to the clinical records or diagnostic markers, such as serum amylase and serum lipase levels and/or radiological images. However, we used the best-performing algorithm with a positive predictive value of 60% to 80% and a negative predictive value of greater than 90% for acute pancreatitis.18 Misclassification of exposure is possible because information on exposure was derived from pharmacy claims, and we cannot guarantee that the pills were ingested. The number of GLP-1–based therapy users was small, limiting our statistical power to detect an effect for specific drugs, resulting in imprecise estimates.
Our findings are not generalizable to patients older than 64 years, who were excluded from the analysis. We did not impute for missing data. We did not evaluate the risk of chronic pancreatitis because administrative databases are not suitable for outcomes detectable only by histological examination. We had limited duration of follow-up in our database, which was insufficient to evaluate the potential long-term risk of pancreatic cancer. Other GLP-1–based therapies, such as saxagliptin hydrochloride and liraglutide, were not available during the study period.
Our study also has several strengths. We minimized the possibility of type II error by selecting a large number of cases and focusing on the available class of GLP-1–based therapies during the study period rather than a single agent. We adjusted for a large number of available confounders, including metformin use. Our analytic plan allowed for adequate exposure windows because the timing of the risk of acute pancreatitis associated with GLP-1–based therapies was unknown.
Unanswered questions and future research
We suggest that a self-controlled case series design that allows control for individual-level confounding by disease severity may offer additional insight. Further studies using new user designs should clarify the exact timing of these risks and determine whether susceptible subgroups, such as those with genetic mutations, or pancreatitis risks, such as obesity, may be at the highest risk. A recent study noted elevation of serum lipase and serum amylase levels in patients with GLP-1–based therapies.22 Future studies should determine whether monitoring of serum enzyme levels can be used to predict the occurrence of acute pancreatitis among patients using GLP-1–based therapies. Long-term prospective studies should examine other outcomes, such as chronic pancreatitis and pancreatic cancer.
In summary, acute pancreatitis has significant morbidity and mortality. In this administrative database study of US adults with type 2 diabetes mellitus, treatment with the GLP-1–based therapies sitagliptin and exenatide was associated with an increased risk of hospitalization for acute pancreatitis.
Correspondence: Sonal Singh, MD, MPH, Department of Medicine and the Center for Public Health and Human Rights, Johns Hopkins University Schools of Medicine and Public Health, 614 N Wolfe St, Ste E7144, Baltimore, MD 21287 (sosingh@jhsph.edu).
Accepted for Publication: October 13, 2012.
Published Online: February 25, 2013. doi:10.1001/jamainternmed.2013.2720
Author Contributions: Dr Singh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Singh and Chang. Acquisition of data: Richards, Weiner, and Clark. Analysis and interpretation of data: Chang, Richards, Clark, and Segal. Drafting of the manuscript: Singh and Richards. Critical revision of the manuscript for important intellectual content: Singh, Chang, Weiner, Clark, and Segal. Statistical analysis: Singh and Chang. Obtained funding: Clark. Administrative, technical, and material support: Singh, Chang, Richards, Weiner, and Clark. Study supervision: Singh and Segal.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was supported by the Johns Hopkins Clinical Research Scholars Program (Dr Singh); by grant 1KL2RR025006-03 from the National Center for Research Resources (NCRR), a component of the National Institutes of Health (NIH); and by the NIH Roadmap for Medical Research.
Role of the Sponsors: The NCRR, NIH, and BCBS had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Disclaimer: The data set used in this study was created for a research project on the patterns of obesity care within selected BCBS plans. The original development of the data set was funded by unrestricted research grants from Ethicon Endo-Surgery, Pfizer, and GlaxoSmithKline. Data and database development support were provided by BCBS (Tennessee, Hawaii, Michigan, and North Carolina), Highmark, Inc, and Independence BlueCross (Pennsylvania), and Wellmark, Inc (Iowa and South Dakota). The BCBS plans were invited to review the manuscript. Statistical protocol is available from the authors on request. The contents are solely the responsibility of the authors and do not necessarily represent the official view of the NCRR or the NIH.
Additional Information: Information on NCRR is available at http://www.nih.gov/about/almanac/organization/NCRR.htm.
1.Amori RE, Lau J, Pittas AG. Efficacy and safety of incretin therapy in type 2 diabetes: systematic review and meta-analysis.
JAMA. 2007;298(2):194-20617622601
PubMedGoogle ScholarCrossref 2.Butler PC, Dry S, Elashoff R. GLP-1–based therapy for diabetes: what you do not know can hurt you.
Diabetes Care. 2010;33(2):453-45520103562
PubMedGoogle ScholarCrossref 3.Drucker DJ, Sherman SI, Gorelick FS, Bergenstal RM, Sherwin RS, Buse JB. Incretin-based therapies for the treatment of type 2 diabetes: evaluation of the risks and benefits.
Diabetes Care. 2010;33(2):428-43320103558
PubMedGoogle ScholarCrossref 4.Nachnani JS, Bulchandani DG, Nookala A,
et al. Biochemical and histological effects of exendin-4 (exenatide) on the rat pancreas.
Diabetologia. 2010;53(1):153-15919756486
PubMedGoogle ScholarCrossref 5.Matveyenko AV, Dry S, Cox HI,
et al. Beneficial endocrine but adverse exocrine effects of sitagliptin in the human islet amyloid polypeptide transgenic rat model of type 2 diabetes: interactions with metformin.
Diabetes. 2009;58(7):1604-161519403868
PubMedGoogle ScholarCrossref 6.Gier B, Matveyenko AV, Kirakossian D, Dawson D, Dry SM, Butler PC. Chronic GLP-1 receptor activation by exendin-4 induces expansion of pancreatic duct glands in rats and accelerates formation of dysplastic lesions and chronic pancreatitis in the Kras
G12D mouse model.
Diabetes. 2012;61(5):1250-126222266668
PubMedGoogle ScholarCrossref 7.Ahmad SR, Swann J. Exenatide and rare adverse events.
N Engl J Med. 2008;358(18):1970-197218456920
PubMedGoogle Scholar 10.Elashoff M, Matveyenko AV, Gier B, Elashoff R, Butler PC. Pancreatitis, pancreatic, and thyroid cancer with glucagon-like peptide-1–based therapies.
Gastroenterology. 2011;141(1):150-15621334333
PubMedGoogle ScholarCrossref 11.Garg R, Chen W, Pendergrass M. Acute pancreatitis in type 2 diabetes treated with exenatide or sitagliptin: a retrospective observational pharmacy claims analysis.
Diabetes Care. 2010;33(11):2349-235420682680
PubMedGoogle ScholarCrossref 12.Dore DD, Seeger JD, Arnold Chan K. Use of a claims-based active drug safety surveillance system to assess the risk of acute pancreatitis with exenatide or sitagliptin compared to metformin or glyburide.
Curr Med Res Opin. 2009;25(4):1019-102719278373
PubMedGoogle ScholarCrossref 13.Dore DD, Bloomgren GL, Wenten M,
et al. A cohort study of acute pancreatitis in relation to exenatide use.
Diabetes Obes Metab. 2011;13(6):559-56621320263
PubMedGoogle ScholarCrossref 14.Herrera V, Ronald Aubert R, Tully L,
et al. Pancreatitis in patients treated with exenatide or sitagliptin [abstract 10-LB]. Abstract presented at: 69th Scientific Session of the American Diabetes Association; June 6, 2009; New Orleans, LA
15.Williams-Herman D, Round E, Swern AS,
et al. Safety and tolerability of sitagliptin in patients with type 2 diabetes: a pooled analysis.
BMC Endocr Disord. 2008;8:1418954434
PubMedGoogle ScholarCrossref 16.Singh S, Loke YK. Drug safety assessment in clinical trials: methodological challenges and opportunities.
Trials. 2012;13:13822906139
PubMedGoogle ScholarCrossref 17.Noel RA, Braun DK, Patterson RE, Bloomgren GL. Increased risk of acute pancreatitis and biliary disease observed in patients with type 2 diabetes: a retrospective cohort study.
Diabetes Care. 2009;32(5):834-83819208917
PubMedGoogle ScholarCrossref 18.Moores K, Gilchrist B, Carnahan R, Abrams T. A systematic review of validated methods for identifying pancreatitis using administrative data.
Pharmacoepidemiol Drug Saf. 2012;21:(suppl 1)
194-20222262606
PubMedGoogle ScholarCrossref 19.Young BA, Lin E, Von Korff M,
et al. Diabetes Complications Severity Index and risk of mortality, hospitalization, and healthcare utilization.
Am J Manag Care. 2008;14(1):15-2318197741
PubMedGoogle Scholar 20.Health Services Research and Development Center. The Johns Hopkins ACG Case-Mix System Reference Manual, Version 7.0. Baltimore, MD: Johns Hopkins University Bloomberg School of Public Health; 2005
21.Chang HY, Weiner JP, Richards TM, Bleich SN, Segal JB. Predicting costs with diabetes complications severity index in claims data.
Am J Manag Care. 2012;18(4):213-21922554010
PubMedGoogle Scholar 22.Lando HM, Alattar M, Dua AP. Elevated amylase and lipase levels in patients using glucagonlike peptide-1 receptor agonists or dipeptidyl-peptidase-4 inhibitors in the outpatient setting.
Endocr Pract. 2012;18(4):472-47722440997
PubMedGoogle ScholarCrossref