Amylase, Lipase, and Acute Pancreatitis In

Diabetes Care 1

William M. Steinberg,1 John B. Buse,2 Amylase, Lipase, and Acute Marie Louise Muus Ghorbani,3 David D. Ørsted,3 Michael A. Nauck,4 and Pancreatitis in People With Type 2 the LEADER Steering Committee, on behalf Diabetes Treated With Liraglutide: of the LEADER Trial Investigators* Results From the LEADER Randomized Trial https://doi.org/10.2337/dc16-2747

OBJECTIVE To evaluate serum amylase and lipase levels and the rate of acute pancreatitis in patients with type 2 diabetes and high cardiovascular risk randomized to liraglutide or placebo and observed for 3.5–5.0 years.

RESEARCH DESIGN AND METHODS A total of 9,340 patients with type 2 diabetes were randomized to either liraglutide or placebo (median observation time 3.84 years). Fasting serum lipase and amylase were monitored. Acute pancreatitis was adjudicated in a blinded manner. PATHOPHYSIOLOGY/COMPLICATIONS

RESULTS Compared with the placebo group, liraglutide-treated patients had increases in serum lipase and amylase of 28.0% and 7.0%, respectively. Levels were increased at 1George Washington University Medical Center, 6 months and then remained stable. During the study, 18 (0.4% [1.1 events/1,000 Washington, DC 2 patient-years of observation] [PYO]) liraglutide-treated and 23 (0.5% [1.7 events/ University of North Carolina School of Medicine, ﬁ Chapel Hill, NC 1,000 PYO]) placebo patients had acute pancreatitis con rmed by adjudication. Most 3Novo Nordisk A/S, Bagsvaerd, Denmark acute pancreatitis cases occurred ‡12 months after randomization. Liraglutide-treated 4St. Josef-Hospital, Ruhr University, Bochum, patients with prior history of pancreatitis (n = 147) were not more likely to develop Germany acute pancreatitis than similar patients in the placebo group (n = 120). Elevations of Corresponding author: William M. Steinberg, amylase and lipase levels did not predict future risk of acute pancreatitis (positive [email protected]. predictive value <1.0%) in patients treated with liraglutide. Received 23 December 2016 and accepted 3 April 2017. CONCLUSIONS Clinical trial reg. no. NCT01179048, clinicaltrials In a population with type 2 diabetes at high cardiovascular risk, there were numer- .gov. ically fewer events of acute pancreatitis among liraglutide-treated patients (regard- This article contains Supplementary Data online at http://care.diabetesjournals.org/lookup/ less of previous history of pancreatitis) compared with the placebo group. Liraglutide suppl/doi:10.2337/dc16-2747/-/DC1. was associated with increases in serum lipase and amylase, which were not pre- *The full list of LEADER Trial Investigators dictive of an event of subsequent acute pancreatitis. and Committee Members can be found in the Supplementary Data online. Glucagon-like peptide 1 (GLP-1) receptor agonists are established glucose-lowering © 2017 by the American Diabetes Association. Readers may use this article as long as the work drugs for treating type 2 diabetes (1). However, incretin-based therapies (GLP-1 re- is properly cited, the use is educational and not ceptor agonists and dipeptidyl peptidase 4 [DPP-4] inhibitors) are associated with for proﬁt, and the work is not altered. More infor- increased levels of serum lipase and amylase, and a potential for an increased risk of mationisavailableathttp://www.diabetesjournals acute pancreatitis has previously been raised (2–6). Current assessments from regula- .org/content/license. tory agencies in Europe and the U.S. indicate that pancreatitis should be considered a See accompanying article, p. XXX. Diabetes Care Publish Ahead of Print, published online May 5, 2017 2 Amylase, Lipase, and Pancreatitis in LEADER Diabetes Care

risk associated with incretin-based drugs were repeated annually until the trial the liver enzyme profile alanine ami- until further data are available (7). Sub- ended. No measurement was performed notransferase/aspartate transaminase) sequently, analyses of pooled data from after the planned end-of-treatment visit. (11). In addition, characteristics of the shorter phase III clinical trials of GLP-1 re- Serum pancreatic lipase and total amylase events not confirmed by adjudication ceptor agonists have noted an associa- were measured in the fasting state in all were evaluated in a post hoc review by tion with 38 cases of acute pancreatitis patients using an enzymatic colorimetric the sponsor. among 17,775 patient-years of observa- assay (Roche Diagnostics, Mannheim, Ger- tion (PYO) (event rate 2.1/1,000 PYO) many) performed by a central laboratory Statistical Analysis compared with 9 cases after 5,863 PYO (ICON PLC, Dublin, Ireland). No patients All analyses were prespecified (before da- (event rate 1.5/1,000 PYO) (8). For DPP-4 had symptoms suggesting acute pancrea- tabase lock) unless marked otherwise. inhibitors, though meta-analysis of shorter titis at baseline blood draw. The upper Due to a right-skewed distribution, ob- regulatory trials failed to show an associ- limit of normal (ULN) for the assays was served geometric means for lipase and ationwithpancreatitis(8),meta-analysis 63 units/L for lipase and 100 units/L for amylase levels are presented. The differ- of the longer cardiovascular outcomes tri- amylase. Neither limit was adjusted for ences between treatment groups in als does suggest a small increased risk the distribution expected in patients with lipase and amylase were estimated (9,10). type 2 diabetes. More frequent tests for using a mixed effects model for repeated Due to the previous results and the lack lipase and amylase were allowed at the measurements adjusted for baseline co- of long-term data, pancreatitis continues discretion of the investigator. Baseline li- variates. Comparisons between groups to be of interest in studies of GLP-1 re- pase activity and amylase activity among were performed at 36 months, as this ceptor agonists, particularly in studies the study population have previously been was the last annual visit with laboratory where long-term safety data can be col- reported (14). Per protocol, participants testing for the majority of the patients, lected. Furthermore, serial amylase and diagnosed with pancreatitis should be given the minimum treatment period of lipase measurements have been sug- withdrawn from study medication by 42 months. An analysis using a linear gested to predict the development of their care team. model with repeated measures (Mixed acute pancreatitis (11). effect Model Repeat Measurement In the Liraglutide Effect and Action in Evaluation of Acute Pancreatitis [MMRM]) assuming an unstructured co- Diabetes: Evaluation of cardiovascular Pancreatitis was predefined by the trial variance matrix was performed to evalu- outcome Results (LEADER) trial, 9,340 protocol as a medical event of special in- ate the interaction between explanatory patients with type 2 diabetes and high terest. Revised Atlanta criteria were used factors at baseline and treatment in re- cardiovascular risk were randomized to for a confirmed diagnosis of acute pancre- lation to change from baseline for log- the GLP-1 analog liraglutide or match- atitis and classified by severity (15) transformed levels of lipase and amylase, ing placebo, both in addition to stan- (Supplementary Table 1). For the diagno- respectively, at each visit. Effects of each dard of care, and followed for 3.5–5.0 sis of acute pancreatitis, two or more of level in the factors were presented as the years. The aim of the current study was to the following three criteria had to be ful- treatment ratio between liraglutide ver- evaluate results from LEADER regarding filled: 1) severe acute upper abdominal sus placebo at the 36-month visit. Explan- the effects of liraglutide treatment on se- pain, 2) amylase and/or lipase threefold atory factors were age, sex, BMI, smoking rum lipase and amylase and the number or more above the ULN ($33 ULN), and status, diabetes duration, glycated hemo- of cases of acute pancreatitis confirmed 3) characteristic findings on imaging (ul- globin (HbA1c), baseline lipase and amy- by adjudication. trasound, computed tomography [CT], or lase levels, medical history (including MRI) of the pancreas. Chronic pancreatitis history of pancreatitis), estimated glo- RESEARCH DESIGN AND METHODS was defined by characteristic imaging merular filtration rate, lipid levels, and Detailed methods have previously been findings (ultrasound, CT, and MRI) with use of certain medications. The interac- published (12,13). Briefly, 9,340 patients abnormal pancreatic function tests or tion between baseline characteristics were enrolled at 410 sites in 32 countries. characteristic histological findings. All po- and allocated treatment (liraglutide vs. Patients with type 2 diabetes at high risk tential cases of pancreatitis were adjudi- placebo) on the development of acute for cardiovascular events were random- cated by an independent committee of pancreatitis was examined using logistic ized 1:1 double-blind to either subcutane- experts, which was blinded to treatment. regression. Correction for multiple com- ous liraglutide 1.8 mg daily (or maximum To ensure that potential events of pancre- parisons in the analyses of baseline char- tolerated dose) or placebo with a treat- atitis were not missed for adjudication, a acteristics was performed using the ment period of 3.5–5.0 years and a 30 day Medical Dictionary for Regulatory Activi- Bonferroni method. follow-up period. Patients with a history ties search of the clinical database was The analyses for risk of acute pancrea- of pancreatitis were not excluded. The performed to identify events for adjudi- titis were based on a Cox proportional primary composite outcome was the first cation not already identified by study in- hazards model with treatment as a cova- occurrence of death from cardiovascular vestigators. In those who developed riate. All randomized patients were in- causes, nonfatal myocardial infarction, or acute pancreatitis as confirmed by adju- cluded in the analyses, and patients nonfatal stroke. Other prespecified out- dication, data were further analyzed post without acute pancreatitis were cen- comes included acute pancreatitis. hoc as to whether gallstone disease was sored on the day of their follow-up visit/ Serum lipase and amylase (pancreas present at the time of pancreatitis event withdrawal date or date of death. Events specific) were measured at baseline and either on imaging (sonography, CT, MRI) occurring after follow-up visit were not months 6 and 12. Thereafter, assessments or blood testing (threefold elevation of included. All analyses were performed care.diabetesjournals.org Steinberg and Associates 3

using SAS, version 9.3 (SAS Institute, Cary, and considering baseline characteris- positive association with subsequent ele- NC). tics, at 36 months liraglutide was associated vation of amylase. with an estimated relative 28.0% increase in lipase (estimated treatment ratio 1.28 fi fi RESULTS Events Con rmed or Not Con rmed as [95% CI 1.25–1.30; P , 0.001). Observed Acute Pancreatitis by the Adjudication A total of 9,340 patients were random- mean lipase changed from 40.0 units/L Process ized (4,668 to liraglutide and 4,672 to pla- to 55.2 units/L for liraglutide. Similarly, Supplementary Table 3 presents details cebo, both added to standard of care). Of an estimated 7.0% (1.07 [95% CI 1.06– of the adjudication process for Event Ad- fi these, 96.8% completed a nal visit, died, 1.09]; P , 0.001) increase in amylase was judication Committee confirmed acute or had a primary outcome. The me- seen with an observed amylase change pancreatitis cases and which of the three dian and mean times of exposure to from 59.4 units/L to 70.9 units/L for diagnostic criteria (typical pain, elevated study medication were 3.52 and 3.07 liraglutide. Results for both lipase and am- serum pancreatic enzymes $33 ULN, years, respectively. The mean proportion ylase were similar when only patients and characteristic imaging findings) were of time on study drug was 84.0% for with an on-treatment laboratory value met. In all cases, a minimum of two of these liraglutide and 82.0% for placebo, and were included (data not shown). three criteria were met. – 71.3% of patients were exposed for 3 5 Figure 2 shows the categorical maxi- Supplementary Tables 4–6present years. The median observation time was mum values of lipase and amylase over similar data for patients with events sug- 3.84 years in both groups. The median the course of the study. Of patients ran- gestive of acute pancreatitis but in whom daily dose of liraglutide was 1.78 mg domized to liraglutide, 51.3% experi- the diagnosis of acute pancreatitis was not – (interquartile range 1.54 1.79) including enced at least one elevated lipase level confirmed by the adjudication process. This off-treatment periods. Of patients enter- (.ULN) over time compared with 31.8% information is based on post hoc review by ing the study, 267 had a history of pan- on placebo. In the liraglutide group, 8.3% the trial sponsor. The diagnosis was not creatitis (147 and 120 in the liraglutide of patients and in the placebo group 5.3% confirmed by the Event Adjudication Com- and placebo groups, respectively). experienced increases of threefold or mittee in 50 liraglutide-treated or 21 pla- more in lipase at some point during cebo patients. For the majority of events in Baseline Characteristics the trial (Fig. 2). For amylase, 29.0% of both treatment groups, patients had been Baseline population characteristics have liraglutide-treated patients and 22.9% of clinically evaluated due to any abdominal previously been published (13). The the placebo group had at least one ele- pain or elevation in lipase or amylase levels. mean age of the population was 64.3 vated amylase level during the study Information on presence or absence of years, 64.3% of the population were (1.0% and 0.8%, respectively, had levels pain, lipase and amylase levels, and imag- male, and mean duration of diabetes elevated by threefold or more). ing was available in the majority of cases was 12.9 years. but did not support a diagnosis of acute Effect of Baseline Data on Lipase pancreatitis. There was no imbalance in di- Lipase and Amylase Levels Over Time and Amylase Levels Over Time agnostic procedures performed and infor- fl The time course and magnitude of change The in uence of baseline characteristics on mation available for the adjudication in lipase and amylase over time are de- increases in lipase and amylase over time is process between patients administered picted in Fig. 1 (13). In the liraglutide arm, shown in Supplementary Table 2. With liraglutide or placebo. elevated levels of both enzymes were adjustment for multiple comparisons, seen at 6 months and persisted for the du- only increased baseline lipase level was Acute Pancreatitis fi ration of the study. Compared with placebo found to have a statistically signi cant Acute pancreatitis events confirmed by adjudication were identified in 18 of the 4,668 liraglutide-treated patients (0.4%) and 23 of the 4,672 placebo-administered patients (0.5%) (Supplementary Table 7). One of 18 patients in the liraglutide group had 2 acute pancreatitis events (19 total events), and 8 patients in the placebo group had .1 event (31 total events) (Sup- plementary Table 7). The corresponding event rates were 1.1/1,000 PYO for the liraglutide group and 1.7/1,000 PYO for the placebo group (Supplementary Table 7). When we analyzed first acute pancreatitis events, there was a nonsignifi- cant 22% reduction in the incidence (hazard Figure 1—Lipase and amylase over time. Data are observed geometric means (full analysis ratio 0.78 [95% CI 0.42–1.44]) (Fig. 3). Re- set) (12). Estimated treatment ratios were calculated using a mixed model for repeated measure- sults were similar (hazard ratio 0.73 [95% ments. For amylase, 135 patients equally distributed between the liraglutide and placebo arms had – missing information and were not included in the analysis. For lipase, 177 patients had missing CI 0.35 1.54]) when only patients with information and were not included in the analysis. ETR, estimated treatment ratio; Lira, liraglutide; events that occurred during days on U, units. treatment (+1 day) were included. 4 Amylase, Lipase, and Pancreatitis in LEADER Diabetes Care

positive predictive values for elevated serum levels of both enzymes was ,1.0%. Supplementary Fig. 1 plots lipase and amylase over time in patients who subsequently developed acute pancreatitis, showing minimal elevations in levels measured at scheduled visits prior to the diagnosis of acute pancreatitis. Fi- nally, in both the liraglutide and placebo groups there was no association between onefold or threefold elevated lipase and — Figure 2 Maximum lipase and amylase levels by category. Data are presented as proportion of the subsequent risk of acute pancreatitis subjects with maximum lipase and amylase levels at scheduled measurements within the stated categories (.ULN, $23 ULN, or $33 ULN [full analysis set]). in patients without acute pancreatitis at thetimeofmeasurement(Supplementary Fig. 2). In fact, relatively more patients on Most acute pancreatitis events were In both groups, most cases of acute pan- placebo had categorical increases in en- mild according to modified Atlanta crite- creatitis developed .12 months after be- zyme levels prior to the attack of acute ria (89.5% and 83.9% in the liraglutide and ginning the trial (Supplementary Table 8). pancreatitis than for the liraglutide group, placebo groups, respectively) (Supple- Of the patients who entered the trial with albeit numbers were small. mentary Tables 7 and 8). Six of the 18 pa- a history of pancreatitis, 2 of 147 (1.4%) in tients in the liraglutide group had been the liraglutide group and 6 of 120 (5.0%) CONCLUSIONS off liraglutide for extended periods of in the placebo group had another event This randomized, double-blind trial com- time when the acute pancreatitis event during the study. paring liraglutide and placebo added to occurred (28, 30, 103, 146, 634, and Effect of Baseline Characteristics on standard of care in patients with type 2 di- 637 days, respectively) (Supplementary Subsequent Development of Acute abetes and high cardiovascular risk is the Table 8). Gallstone disease was observed Pancreatitis largest and longest study to date exposing (by imaging and/or $33 ULN elevations In Supplementary Table 9, the interaction patients to the GLP-1 analog liraglutide. In of alanine aminotransferase/aspartate between baseline characteristics and the present report, liraglutide treatment transaminase) at the time of acute pan- treatment on subsequent development was associated with a 28.0% increase in creatitis event for 7 of 18 (38.9%) com- of acute pancreatitis is shown. There mean serum lipase and 7.0% increase in pared with 10 of 23 (43.5%) patients in was no statistically significant interaction mean serum amylase compared with pla- the liraglutide and placebo groups (P = between liraglutide or placebo adminis- cebo. There were fewer acute pancreatitis 0.81) (Supplementary Table 8). tration and baseline characteristics on events in the liraglutide group, and isolated Two patients in the placebo group had the risk of emergent acute pancreatitis. liraglutide-associated elevations in lipase chronic pancreatitis events confirmed by and amylase were not predictive of subse- adjudication (Supplementary Table 7). Predictive Value of Lipase and Amylase quent acute pancreatitis. Liraglutide was Figure 3 shows the time course of Table 1 and Supplementary Tables 10 and also associated with increases in the pro- acute pancreatitis events, which was sim- 11 show the predictive values of measur- portion of patients with elevated enzyme ilar in both groups. No patients in the ing lipase and amylase in predicting acute levels over the course of therapy, including liraglutide group developed acute pancre- pancreatitis. Whether the data were an- some with threefold or more increases in atitis within the first 6 months of therapy. alyzed using the .ULN or $33 ULN, lipase (albeit none of these had acute pancreatitis), findings consistent with prior studies of liraglutide in patients with normoglycemia, prediabetes, and diabetes (16). The current study did not evaluate pancreatic enzymes after liraglutide was discontinued, but prior work has shown that serum pancreatic enzymes return to baseline when liraglutide is stopped (17). The cause of serum pancreatic enzyme elevations, especially lipase, induced by liraglutide is unknown. Indeed, the mech- anisms by which pancreatic enzymes enter and exit the blood compartment have not been elucidated. One hypothesis is that enzymes enter the blood by passing Figure 3—Time to confirmed acute pancreatitis. Kaplan-Meier plot of confirmed acute pancreatitis through the basolateral membrane of the first index events (full analysis set). Values underneath the graph are number of subjects at risk. HR, acinar cell allowing access into the blood hazard ratio; Lira, liraglutide. (“spillover” of a minor percentage of the care.diabetesjournals.org Steinberg and Associates 5

fi Table 1—Predictive value of increased lipase or amylase levels for confirmed acute agonist lixisenatide reported ve and eight pancreatitis patients with acute pancreatitis in the lix- Liraglutide Placebo isenatide and placebo groups, respectively (30). Additional cardiovascular outcomes N % N % trials for other GLP-1 receptor agonists Lipase .ULN during trial 2,604 1,682 areexpectedtoreportinthecoming Subsequent acute pancreatitis 7 0.27 11 0.65 years and may provide further data on $ 3 Lipase 3 ULN during trial 339 216 this topic. Subsequent acute pancreatitis 0 0.00 2 0.93 The time course of acute pancreatitis in . Amylase ULN during trial 1,382 1,084 both liraglutide and placebo groups was Subsequent acute pancreatitis 3 0.22 5 0.46 similar. The majority of patients in both Amylase $33 ULN during trial 38 35 Subsequent acute pancreatitis 0 0.00 0 0.00 groups had acute pancreatitis attacks after participating in the trial for .12 months. fi Data are for con rmed acute pancreatitis (full analysis set). Subjects are included in the elevated Most therapies associated with drug- lipase or amylase category from the date of first elevation within the stated categories (.ULN and $33 ULN lipase or amylase). Hazard ratios and 95% CIs derived from data in this table are induced pancreatitis are thought to be shown in Supplementary Fig. 2. %, percentage of subjects; N, number of subjects. due to a hypersensitivity reaction (31). Hypersensitivity reactions tend to have much shorter latencies (the time interval between starting the drug and develop- enzymes present in acinar cells). The elim- on serum lipase (and, less so, amylase) has ment of acute pancreatitis) of up to ination of lipase from the blood is thought not been sufficiently delineated. 12 weeks (32,33). Long latencies as seen in part to be due to renal degradation of Despite this background of increased li- in the present population suggest a differ- the enzyme (18). Although previous stud- pase and amylase associated with liraglutide, ent mechanism than hypersensitivity. An- ies reported controversial views on the rate of acute pancreatitis was quite low other possibility is the development of whether human pancreatic acinar cells overall and 22% lower in the liraglutide gallstones leading to gallstone pancreati- express GLP-1 receptors (19,20), recent than in the placebo group, albeit not statis- tis. Patients with type 2 diabetes (not on animal studies indicate that pancreatic tically significant. The liraglutide group GLP-1 receptor agonists) have an in- acinar cells have GLP-1 receptors that, also included six patients who were off creased risk of developing gallstones when stimulated, increase the production liraglutide for 28–637 days when acute and acute pancreatitis (6), and liraglutide and secretion of pancreatic enzymes (20). pancreatitis developed. The relationship therapy in patients without diabetes in- However, increases in pancreatic secre- between acute pancreatitis and drugs creases the risk of gallstones and acute tion into the pancreatic ductular system that have been stopped for long periods cholecystitis compared with placebo and intestine alone do not explain higher of time remains obscure. In pooled data (34). Gallstones may take an extended levels in the blood. Along these lines, se- from prior studies of the GLP-1 receptor period of time to form, which may explain rum pancreatic enzymes do not rise agonists exenatide, liraglutide, and lixise- the latency between drug initiation and after a meal despite an increase in pan- natide, an excess of acute pancreatitis was acute pancreatitis in some patients. In creatic secretion (21). It appears possible found compared with comparator treat- LEADER, an increased risk of develop- that liraglutide somehow enhances the ments (2.1/1,000 PYO and 1.5/1,000 ment of acute gallstone disease was basolateral secretion of pancreatic en- PYO, respectively) (8). The lower event also observed in the liraglutide group zymes, especially lipase, into the blood rate for acute pancreatitis in the current compared with placebo (3.1% vs. 1.9%, or may affect the degradation of lipase study (1.1/1,000 PYO) compared with the P , 0.001) (13). Nevertheless, the ob- in the kidney. Alternative explanations rate reported in the analysis by Meier and served imbalance in acute gallstone dis- are an increase in pancreatic exocrine Nauck (8) may be the result of adjudica- ease did not translate into an increased weight due to stimulation of protein syn- tion in the current study; adjudication was risk of gallstone-associated acute pancre- thesis, as shown in rodents treated with not performed in all studies contributing atitis, since a similar proportion of pancre- liraglutide, with a likely stimulation of to the pooled analysis. atitis events (38.9% vs. 43.5%) was found pancreatic enzyme synthesis as part of The current study population consisted to be associated with signs of gallstone this general enhancement in protein syn- of patients with type 2 diabetes at high disease being present at the time of thesis (22), although similar effects have risk for cardiovascular events, which is event in the liraglutide group compared not been obvious in studying nonhuman different from populations previously with placebo. The protocol required all primates (23). The changes in lipase and studied. Interestingly, another recent adverse events of acute gallstone disease amylase may also be the consequence of cardiovascular outcomes trial for the to be reported; however, asymptom- changes in food preferences (as part of once-weekly GLP-1 analog semaglutide atic gallstones were not required to be well-established reductions in appetite reported a lower number of patients reported, and furthermore, systematic and caloric intake induced by liraglutide) with acute pancreatitis events in the sem- clinical evaluation for gallstone disease (24,25) leading to altered expression of aglutide versus placebo group (9 and at baseline in the LEADER trial was not pancreatic enzymes specifically favoring 12 for semaglutide and placebo, respec- part of the protocol. the digestion of preferred substrates tively), as confirmed by adjudication (29). It has been suggested that liraglutide (26–28). Altogether, the mechanism of Similarly, the cardiovascular outcomes tri- may induce pancreatitis in patients with GLP-1 receptor stimulation with liraglutide al for the short-acting GLP-1 receptor prior history of pancreatitis (35). In fact, 6 Amylase, Lipase, and Pancreatitis in LEADER Diabetes Care

prescribing information for liraglutide ad- Finally, to ensure objective assessment received consulting fees from Novo Nordisk and vises exercising caution in patients with a of events of suspected acute pancreatitis, owns stock in Novo Nordisk. J.B.B. reports receiv- history of acute pancreatitis and that a committee blinded to treatment evalu- ing contracted consulting fees paid to his institu- tion and travel support from Novo Nordisk, Eli other therapies should be considered. In ated all cases and used the Atlanta criteria Lilly, Bristol-Myers Squibb, GI Dynamics, Elcelyx, the current study, pancreatitis was not an to confirm a diagnosis of acute pancrea- Merck, Metavention, vTv Therapeutics, PhaseBio, exclusion criterion for study entry, and in titis. A potential weakness of this study is AstraZeneca, Dance Biopharm, Quest Diagnos- patients with prior history of pancreatitis, that the trial population only included pa- tics, Sanofi, Lexicon Pharmaceuticals, Orexigen Therapeutics, Takeda Pharmaceuticals, Adocia, and only 2 of 147 (1.4%) in the liraglutide tients with type 2 diabetes, the majority Roche; receiving grant support from Eli Lilly, Bristol- group and 6 of 120 (5.0%) in the placebo being men, with a higher mean age and Myers Squibb, GI Dynamics, Merck, PhaseBio, group developed acute pancreatitis. Our longer diabetes duration than the wider AstraZeneca, Medtronic, Sanofi, Tolerex, Osiris results do not support the exclusion of population of patients with type 2 diabe- Therapeutics, Halozyme Therapeutics, Johnson patients with previous pancreatitis from tes. Thus, our findings may not be gener- & Johnson, Andromeda, Boehringer Ingelheim, GlaxoSmithKline, Astellas Pharma, MacroGenics, treatment with GLP-1 receptor agonists in alizable. Another potential limitation is Intarcia Therapeutics, Lexicon, Scion NeuroStim, general terms and with liraglutide in par- the low number of acute pancreatitis Orexigen Therapeutics, Takeda Pharmaceuticals, ticular. Nevertheless, the study protocol events, which leads to wide CIs, meaning Theracos, and Roche; receiving fees from and did not call for an active investigation to that we cannot completely rule out a po- holding stock options in PhaseBio; and serving characterize previous pancreatic or biliary tential increased risk (Fig. 3). Finally, on the boards of the AstraZeneca Healthcare Foundation and the Bristol-Myers Squibb To- structural or functional damage. because patients could discontinue treat- gether on Diabetes Foundation. M.L.M.G. and The U.S. Food and Drug Administration ment or not reach the 36-month visit (if D.D.Ø. are employees and shareholders of Novo and other regulatory authorities have they discontinued the trial or died), the re- Nordisk. M.A.N. reports receiving fees for serving mandated that studies involving GLP-1 relationship between lipase levels and pan- on advisory boards from Berlin-Chemie, Boehringer ceptor agonists monitor lipase and amy- creatitis events could be diluted and/or Ingelheim, Eli Lilly, GlaxoSmithKline, Merck Sharp & Dohme, Novo Nordisk, Sanofi,Versartis,Intarcia lase levels under the assumption that this the association between liraglutide treat- Therapeutics, AstraZeneca, Roche, Novartis, and might predict patients’ development of ment and risk of acute pancreatitis could Janssen; lecture fees from Berlin-Chemie, Boehringer acute pancreatitis. This study and others be affected. However, analyses that in- Ingelheim, Eli Lilly, GlaxoSmithKline, Merck Sharp & (16) have shown that the positive predic- cluded only the on-treatment population Dohme, Novo Nordisk, Sanofi,AstraZeneca, Novartis, Janssen, and Medscape; travel support tive value of an isolated elevation of these showed similar results to those of the in- from Berlin-Chemie, Boehringer Ingelheim, Eli enzymes for acute pancreatitis is very low tention-to-treat analyses, potentially due to Lilly, GlaxoSmithKline, Merck Sharp & Dohme, (,1.0%). Indeed, in the population of all the high degree of exposure in the trial. Novo Nordisk, Sanofi, Versartis, Intarcia Thera- patients treated with liraglutide in the In summary, this study confirmed and peutics, AstraZeneca, Roche, Novartis, Janssen, LEADER trial who developed substantially extended observations concerning the ef- and Medscape; and grant support from Eli Lilly, $ 3 GlaxoSmithKline, Merck Sharp & Dohme, Novo elevated serum lipase ( 3 ULN) while fect of liraglutide on the pancreas. In Nordisk, AstraZeneca, and Novartis. No other po- being asymptomatic, none developed acute LEADER, numerically fewer events of tential conflicts of interest relevant to this article pancreatitis later in the trial. In fact, among acute pancreatitis were observed in were reported. patients with categorical enzyme elevations liraglutide-treated patients compared Author Contributions. W.M.S., J.B.B., and there were relatively more in the placebo with the placebo group. Liraglutide in- M.A.N. participated in the conduct and oversight of the protocol. W.M.S., J.B.B., M.L.M.G., D.D.Ø., group who subsequently developed acute creased serum amylase, and especially li- and M.A.N. drafted the manuscript. J.B.B. partic- pancreatitis (based on small numbers) (Sup- pase, and this increase plateaued and ipated in developing the protocol. All authors par- plementary Fig. 2). The reasons for this remained stable for the duration of the ticipated in analysis and interpretation of data, remain obscure. study. Liraglutide-induced elevations edited the manuscript for important intellectual fi The present analysis provides details of lipase and amylase in asymptomatic content, and approved the nal version. W.M.S. is the guarantor of this work and, as such, had full on the blinded adjudication process patients were not predictive of the de- access to all the data in the study and takes re- (Supplementary Tables 3–6) that demon- velopment of acute pancreatitis. sponsibility for the integrity of the data and the strate sufficient data were available for ei- accuracy of the data analysis. ther confirming or refuting the diagnosis of Prior Presentation. Parts of this study were presented in abstract form at the 76th Scientific acute pancreatitis for those events that Sessions of the American Diabetes Association, Acknowledgments. The authors thank Søren were adjudicated. These data also show New Orleans, LA, 10–14 June 2016; presented in Rasmussen (Novo Nordisk) for statistical support; that despite more suspected events sent abstract form at the 52nd Annual Meeting of the Florian Baeres and Anders Jespersen (Novo European Association for the Study of Diabetes, for adjudication with liraglutide treatment Nordisk) for scientific input; Joseph Murphy (in- Munich, Germany, 12–16 September 2016; and (most likely due to lipase and amylase independent medical writer, Kansas City, MO) and published by Marso et al. (13). The full trial pro- Watermeadow Medical for editorial support creases as well as abdominal pain) versus tocol was previously published by Marso et al. (funded by the sponsor); the participants, investi- placebo, among those events not confirmed (13). by adjudication, the extent of workup was gators (full list of investigators available in the Supplementary Data online), and trial site staff; similar between the two groups. and the leadership, employees, and contractors of References Strengths of the current study include the sponsor who were involved in the conduct of 1. Drucker DJ, Nauck MA. The incretin system: its large population, placebo-controlled the trial. glucagon-like peptide-1 receptor agonists and di- design, and relatively long-term observa- Duality of Interest. This study was supported by peptidyl peptidase-4 inhibitors in type 2 diabetes. Novo Nordisk. W.M.S. has served as a legal con- Lancet 2006;368:1696–1705 tion time (for a minimum of 3.5 and up to sultant for Eli Lilly and Company, Amylin, and 2. Singh S, Chang HY, Richards TM, Weiner JP, 5.0 years). Serum samples were collected Novo Nordisk; has served on the steering commit- Clark JM, Segal JB. Glucagonlike peptide 1-based prospectively at specified intervals. tee of the Novo Nordisk LEADER study; and has therapies and risk of hospitalization for acute care.diabetesjournals.org Steinberg and Associates 7

pancreatitis in type 2 diabetes mellitus: a 14. Steinberg WM, Nauck MA, Zinman B, et al.; 24. Secher A, Jelsing J, Baquero AF, et al. The population-based matched case-control study. LEADER Trial investigators. LEADER 3--lipase and arcuate nucleus mediates GLP-1 receptor agonist JAMA Intern Med 2013;173:534–539 amylase activity in subjects with type 2 diabetes: liraglutide-dependent weight loss. J Clin Invest 3. Girman CJ, Kou TD, Cai B, et al. Patients with baseline data from over 9000 subjects in the 2014;124:4473–4488 type 2 diabetes mellitus have higher risk for acute LEADER Trial. Pancreas 2014;43:1223–1231 25. Vilsbøll T, Christensen M, Junker AE, Knop FK, pancreatitis compared with those without diabe- 15. Banks PA, Bollen TL, Dervenis C, et al.; Acute Gluud LL. Effects of glucagon-like peptide-1 recep- tes. Diabetes Obes Metab 2010;12:766–771 Pancreatitis Classification Working Group. Classi- tor agonists on weight loss: systematic review and 4. Lando HM, Alattar M, Dua AP. Elevated fication of acute pancreatitis–2012: revision of the meta-analyses of randomised controlled trials. amylase and lipase levels in patients using gluca- Atlanta classification and definitions by interna- BMJ 2012;344:d7771 gonlike peptide-1 receptor agonists or dipeptidyl- tional consensus. Gut 2013;62:102–111 26. Wicker C, Puigserver A. Changes in mRNA peptidase-4 inhibitors in the outpatient setting. 16. Steinberg WM, Rosenstock J, Wadden TA, levels of rat pancreatic lipase in the early days Endocr Pract 2012;18:472–477 Donsmark M, Jensen CB, DeVries JH. Impact of of consumption of a high-lipid diet. Eur J Biochem 5. Malloy J, Gurney K, Shan K, Yan P, Chen S. In- liraglutide on amylase, lipase and acute pancrea- 1989;180:563–567 creasedvariability and abnormalitiesin pancreatic titis in participants with overweight/obesity and 27. Wicker C, Puigserver A. Expression of rat pan- enzyme concentrations in otherwise asymptom- normoglycemia, prediabetes or type 2 diabetes: creatic lipase gene is modulated by a lipid-rich diet atic subjects with type 2 diabetes. Diabetes Metab secondary analysesof pooleddata from the SCALE at a transcriptional level. Biochem Biophys Res Syndr Obes 2012;5:419–424 clinical development program. Diabetes Care Commun 1990;166:358–364 6. Noel RA, Braun DK, Patterson RE, Bloomgren 2017;40:XXX–XXX 28. Tsai A, Cowan MR, Johnson DG, Brannon PM. GL. Increased risk of acute pancreatitis and biliary 17. Steinberg WM, DeVries JH, Wadden TA, Regulation of pancreatic amylase and lipase gene disease observed in patients with type 2 diabetes: Jensen CB, Svendsen CB, Rosenstock J. Elevated expression by diet and insulin in diabetic rats. Am a retrospective cohort study. Diabetes Care 2009; serum lipase activity in adults with type 2 diabetes J Physiol 1994;267:G 575–G583 32:834–838 and no gastrointestinal symptoms (Abstract). Gas- 29. Marso SP, Bain SC, Consoli A, et al.; SUSTAIN- 7. Egan AG, Blind E, Dunder K, et al. Pancreatic troenterology 2012;142(Suppl. 1):A422 6 Investigators. Semaglutide and cardiovascular safety of incretin-based drugs–FDAandEMAas- 18. Junge W, Malyusz´ M, Ehrens HJ. The role of outcomes in patients with type 2 diabetes. N sessment. N Engl J Med 2014;370:794–797 the kidney in the elimination of pancreatic lipase Engl J Med 2016;375:1834–1844 8. Meier JJ, Nauck MA. Risk of pancreatitis in pa- and amylase from blood. J Clin Chem Clin Biochem 30. Pfeffer MA, Claggett B, Diaz R, et al.; ELIXA tients treated with incretin-based therapies. Dia- 1985;23:387–392 Investigators. Lixisenatide in patients with type 2 betologia 2014;57:1320–1324 19. Waser B, Blank A, Karamitopoulou E, Perren diabetes and acute coronary syndrome. N Engl J 9. Buse JB, Bethel MA, Green JB, et al.; TECOS A, Reubi JC. Glucagon-like-peptide-1 receptor ex- Med 2015;373:2247–2257 Study Group. Pancreatic Safety of Sitagliptin in pression in normal and diseased human thyroid 31. Badalov N, Baradarian R, Iswara K, Li J, the TECOS Study. Diabetes Care 2017;40:164– and pancreas. Mod Pathol 2015;28:391–402 Steinberg W, Tenner S. Drug-induced acute pan- 170 20. Pyke C, Heller RS, Kirk RK, et al. GLP-1 recep- creatitis: an evidence-based review. Clin Gastro- 10. Abbas AS, Dehbi HM, Ray KK. Cardiovascular tor localization in monkey and human tissue: enterol Hepatol 2007;5:648–661; quiz 644 and non-cardiovascular safety of dipeptidyl novel distribution revealed with extensively vali- 32. Andersen V, Sonne J, Andersen M. Spontane- peptidase-4 inhibition: a meta-analysis of ran- dated monoclonal antibody. Endocrinology 2014; ous reports on drug-induced pancreatitis in Den- domized controlled cardiovascular outcome 155:1280–1290 mark from 1968 to 1999. Eur J Clin Pharmacol trials. Diabetes Obes Metab 2016;18:295–299 21. Sonne DP, Vilsbøll T, Knop FK. Pancreatic am- 2001;57:517–521 11. Tenner S, Dubner H, Steinberg W. Predicting ylase and lipase plasma concentrations are unaf- 33. Eland IA, van Puijenbroek EP, Sturkenboom gallstone pancreatitis with laboratory parame- fected by increments in endogenous GLP-1 levels MJ, Wilson JH, Stricker BH. Drug-associated acute ters: a meta-analysis. Am J Gastroenterol 1994; following liquid meal tests. Diabetes Care 2015; pancreatitis: twenty-one years of spontaneous re- 89:1863–1866 38:e71–e72 porting in The Netherlands. Am J Gastroenterol 12. Marso SP, Poulter NR, Nissen SE, et al. Design 22. Koehler JA, Baggio LL, Cao X, et al. Glucagon- 1999;94:2417–2422 of the liraglutide effect and action in diabetes: like peptide-1 receptor agonists increase pancreatic 34. Pi-Sunyer X, Astrup A, Fujioka K, et al.; SCALE evaluation of cardiovascular outcome results mass by induction of protein synthesis. Diabetes Obesity and Prediabetes NN8022-1839 Study (LEADER) trial. Am Heart J 2013;166:823–830.e5 2015;64:1046–1056 Group. A randomized, controlled trial of 3.0 mg 13. Marso SP, Daniels GH, Brown-Frandsen K, 23. Nyborg NC, Mølck AM, Madsen LW, Knudsen of liraglutide in weight management. N Engl J Med et al.; LEADER Steering Committee; LEADER Trial LB. The human GLP-1 analog liraglutide and the 2015;373:11–22 Investigators. Liraglutide and cardiovascular out- pancreas: evidence for the absence of structural 35. Meier JJ, Rosenstock J. Therapy: gastrointes- comes in type 2 diabetes. N Engl J Med 2016;375: pancreatic changes in three species. Diabetes tinal safety of incretin therapies: are we there yet? 311–322 2012;61:1243–1249 Nat Rev Gastroenterol Hepatol 2016;13:630–632