Risk of pancreatic adverse events associated with the use of glucagon-like peptide-1 receptor agonist and dipeptidyl peptidase-4 inhibitor drugs: A systematic review and meta-analysis of randomized trials

Abstract

AIM: To systematically assess risk of pancreatic adverse events with glucagon-like peptide-1 (GLP-1) receptor agonist and dipeptidyl peptidase-4 (DPP-4) inhibitor drugs.

METHODS: We searched PubMed, Embase, CINAHL, Cochrane review of clinical trials, pharmaceutical company clinical trials register, United States Food and Drug Administration website, European Medicines Agency website and ClinicalTrials.gov for randomized controlled trials from inception to October 2013. Randomized control trial studies were selected for inclusion if they reported on pancreatic complication events and/or changes in pancreatic enzyme levels (serum amylase and serum lipase) as adverse events or as serious adverse events for patients who were on GLP-1 receptor agonist and DPP-4 inhibitor drugs. Two independent reviewers extracted data directly. We performed Peto odds ratio (OR) fixed effect meta-analysis of pancreatic adverse events a, and assessed heterogeneity with the I² statistic.

RESULTS: Sixty-eight randomized controlled trials were eligible. A total of 60720 patients were included in our analysis of the association of risk of pancreatic complication events with GLP-1 agents. A total of 89 pancreatic related adverse events occurred among the GLP-1 agents compared to 74 events among the controls. There was a statistically significant increased risk of elevation of pancreatic enzymes associated with GLP-1 agents compared with control (Peto OR = 3.15, 95%CI: 1.56-6.39, P = 0.001, I² = 0%). There was no statistically significant difference in the risk of pancreatic adverse event associated with GLP-1 agent compared with controls (Peto OR = 1.00, 95%CI: 0.73-1.37, P = 1.00, I² = 0%). There were a total of 71 pancreatitis events in patients on GLP-1 agents and 56 pancreatitis events occurred in the control patients. There were 36 reports of pancreatic cancer in these studies. Of these cases, 2 used linagliptin, 2 used alogliptin, 1 used vildagliptin, 7 used saxagliptin while 6 used sitagliptin. The remaining 18 cases occurred among controls.

CONCLUSION: Although GLP-1 based agents are associated with pancreatic enzyme elevation, we were unable to confirm a significant risk of pancreatitis or pancreatic cancer.

Key Words: Diabetes mellitus; Pancreatitis; Glucagon-like peptide-1 agonists; Dipeptidyl peptidase-4 inhibitors; Meta-analysis

Core tip: There is conflicting data on the risk of pancreatitis with glucagon-like peptide-1 (GLP-1) receptor agonists and dipeptidyl peptidase-4 (DPP-4) inhibitors. We performed a meta-analysis of 68 randomized controlled trials of 11 different GLP-1 or DPP-4 targeted drugs. The incidence of pancreatic adverse events in the trials was generally low and we did not find any definitive evidence for pancreatitis or pancreatic cancer amongst the trials. However, we found a significantly raised risk of elevated pancreatic enzymes in a small number of trials that reported such enzyme elevations.

INTRODUCTION

Glucagon-like peptide-1 (GLP-1) is a naturally occurring gut hormone that is mainly secreted by the intestinal L cell. It is a potent antihyperglycemic hormone, inducing glucose-dependent stimulation of insulin secretion while suppressing glucagon secretion. Once in the circulation, GLP-1 has a half-life of less than 2 min, due to the rapid degradation by the enzyme dipeptidyl peptidase-4 (DPP-4). The GLP-1 based therapies include GLP-1 receptor agonists and DPP-4 inhibitors. As GLP-1 is a gut hormone, it is possible that patients may experience adverse effects on the gastrointestinal system such as nausea or abdominal pain.

There are already several GLP-1 receptor agonists and DPP-4 inhibitor drugs approved by the Food and Drug Administration (FDA) or the European Medicines Agency (EMA), and we are aware of additional agents in development. However, 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[1-4]. A recent case-control study showed a significant increased risk of hospitalization for acute pancreatitis associated with the use of sitagliptin or exenatide among adult patients with type-2 diabetes mellitus[5]. A meta-analysis of clinical trials reported no difference for sitagliptin use compared with placebo or other oral hypoglycemic in the incidence rates of pancreatitis[6]. Although complications involving the pancreas (acute pancreatitis, chronic pancreatitis and pancreatic cancer) are potentially serious adverse effects of GLP-1 receptor agonist drugs, there is a paucity of data available to clinicians regarding these effects of GLP-1 receptor agonist drugs. A recent meta-analysis[7] suggested that neither exenatide nor liraglutide increases the risk for acute pancreatitis when used in the treatment of type-2 diabetes mellitus. This analysis, however, was based on small studies, non-clinical evaluation of pancreatitis in the included RCTs and residual confounding in the observational studies that were included. None of the previous studies have adequately evaluated the role of pancreatic enzyme elevations. These studies have not evaluated the occurrence of reports of pancreatic cancer in these trials. Finally, the risk of pancreatic complication associated with individual therapies has not been evaluated.

Our objective was to conduct a systematic review to ascertain the risk of pancreatic complications (acute and chronic pancreatitis and pancreatic cancer) and pancreatic enzyme elevations associated with GLP-1 based therapies, as compared to placebo or other oral hypoglycemic drugs in randomized controlled trials of GLP-1 based therapies.

MATERIALS AND METHODS

Methods

We defined study aims and procedures in the study protocol registered with PROSPERO register of systematic reviews[8].

Data sources and searches

We searched MEDLINE, EMBASE, CINAHL and the Cochrane database from inception to October 2013 using the search terms: (drug name OR chemical compound OR drug class) AND [“Pancreatic Neoplasms”(Mesh) OR “Pancreatitis”(Mesh) OR “pancreas”(tiab) OR “pancreatitis”(tiab) OR “pancreatic”(tiab) OR “pancreatic cancer”(tiab) OR “serum amylase”(tiab) OR “serum lipase”(tiab) OR “Islet Cell Adenoma”(tiab) OR “Insulinoma”(tiab) OR “Islet Cell Carcinoma”(tiab) OR “Gastrinoma”(tiab) OR “Glucagonoma”(tiab) OR “Somatostatinoma”(tiab) OR “Vipoma”(tiab) OR “Pancreatic Ductal Carcinoma”(tiab) OR “Islet Cell Adenomas”(tiab) OR “Insulinomas”(tiab) OR “Islet Cell Carcinomas”(tiab) OR “Gastrinomas”(tiab) OR “Glucagonomas”(tiab) OR “Somatostatinomas”(tiab) OR “Vipomas”(tiab) OR “Pancreatic Ductal Carcinomas”(tiab)] AND English(lang) NOT [“Animals”(Mesh)] NOT [“Animals”(Mesh) AND “Humans”(Mesh)].

We did not specify any language or population restrictions. To identify any unpublished studies, we keyed in the names of specific drug compounds into the search boxes of all GLP-1 agent pharmaceutical companies, three of which had publicly available clinical trials, these were Boehringer Ingelheim clinical trials register, Novartis clinical trials register and Takeda Pharmaceuticals register. We also searched the FDA, the EMA and ClinicalTrials.gov up to August 2013. Bibliographies of included studies and recent review articles were checked for additional relevant studies.

Study selection

We selected randomized controlled trials that enrolled participants using GLP-1 agonist and DPP-4 inhibitor drugs and reported on the risk of pancreatic complications either as adverse events or as serious adverse events. We included studies that reported on the use of FDA approved GLP-1 receptor agonists such as Exenatide (Byetta, Bydureon), Liraglutide (Victoza) and Albiglutide (Tanzeum). Other GLP-1 receptor agonists that were studied but have not yet been approved by FDA included Taspoglutide, Lixisenatide (Lyxumia), Dulaglutide and Semaglutide were included. Studies that also used FDA approved DPP-4 inhibitors such as Vidagliptin (Eucreas, Galvus, Icandra, Jalra, Xiliarx, Zomarist), Sitagliptin (Efficib, Januvia, Janumet, Ristaben, Ristfor, Tesavel, Velmetia, Xelevia), Saxagliptin (Komboglyze, Onglyza), Linagliptin (Jentadueto, Trajenta) and Alogliptin (Nesina) were included. Other DPP-4 inhibitors in development were included in our search. These include Septagliptin, Anagliptin, Bisegliptin, Carmegliptin, Denagliptin, Dutogliptin, Gosogliptin, Isoleucine Thiazolidide, Valine pyrrolidide, Evogliptin, Gemigliptin, Melogliptin, Omarigliptin, Tenegliptin and Trelagliptin. We did not restrict studies by healthcare settings, methods of diagnosing pancreatitis or by indication for the drug.

Data extraction and quality assessment

Two reviewers (HMS and TA) evaluated all titles and abstracts for studies that met the inclusion criteria, and excluded any articles that clearly did not meet the selection criteria. Full reports of potentially relevant studies were retrieved and independently checked for eligibility. Data from the included studies were then extracted independently by two reviewers (HMS and TA) who collected information on study design, study location, study population description, drug exposure, pancreatic complication (acute pancreatitis, chronic pancreatitis, pancreatic cancer) events, pancreatic enzyme derangement (elevated serum pancreatic amylase and/or pancreatic serum lipase) data, mortality from pancreatic events, how the pancreatic events were defined and monitored, confounders for pancreatic events and characteristics of participants onto a pre-formatted spreadsheet. Another reviewer (YKL or SS) then checked the data. Any uncertainties or discrepancies were resolved through rechecking against the source papers, and through discussion with a third reviewer.

We used a pre-specified spreadsheet to record the location and duration of the randomized controlled trials (in years), dose and frequency of GLP-1 agonist drug and DPP-4 inhibitor drug and placebo or alternative hypoglycemic agent, mean age and sex of participants, number of pancreatic complication events and confounders.

The Cochrane toolkit was used for the assessment of bias in evaluating each trial for the reporting of randomization, allocation concealment, the use of blinding of participants and staff, and information on loss to follow-up or withdrawal rates[9]. In accordance with the Cochrane handbook of systematic reviews, we assessed the quality of data on adverse events by recording how they were monitored and recorded by the investigators[10]. We aimed to generate funnel plots to assess the possibility of publication bias, provided that there were > 10 studies available in the meta-analysis, with no evidence of substantial statistical heterogeneity[11].

Statistical analysis

We used RevMan[12] 5.3 to conduct meta-analysis based on the summary statistic of Peto Odds Ratios, which is the recommended approach for rare events[9]. We assumed similarity between the risk ratio and OR because the incidence of adverse outcomes was low[13]. We evaluated both adjusted and unadjusted data from primary studies, although we preferentially used adjusted data where available.

Statistical heterogeneity was assessed using I² statistic[14], with I² values of 30%-60% representing a moderate level of heterogeneity. Pre-specified subgroup analysis was performed by evaluating the effect of study design, study setting and outcome ascertainment.

The statistical methods of this study were reviewed by Yoon K Loke, convenor of the Cochrane Adverse Effects Methods Group.

RESULTS

After a review of 3583 citations, we identified 68 randomized controlled trials (Figure 1) with a total of 60811 patients for inclusion in our analysis of the association of risk of pancreatic complication events with the use of GLP-1 agonist and DPP-4 inhibitor drugs.

Figure 1 Flow diagram of studies identified and selected.

Description of studies

The study characteristics are listed in Table 1 and quality assessment of the trials in Table 2.

Table 1 Characteristics of glucagon-like peptide-1 based agents in randomized controlled trials included in analysis of pancreatic events.

Ref.	Location (No. of centers)	Year of study completion	Total duration (wk)	Duration of GLP-1 exposure (wk)	Participant disease	Arms	No. of participants	Mean age, yr (SD)	Female, n (%)
Ross et al[21]	Multi-national (84 centers in 9 countries)	2010	43	12	Type 2 diabetes	Linagliptin 2.5 mg bid	223	58.7 (9.9)	85 (38.1)
						Linagliptin 5 mg qd	224	58.4 (10.6)	103 (46.0)
						Placebo	44	59.9 (10.7)	23 (52.3)
Haak et al[22]	Multi-national (133 clinics in 14 countries)	2010	73	24	Type 2 diabetes	Linagliptin 5 mg qd	142	56.2 (10.8)	62 (43.7)
						Metformin 500 mg bid	144	52.9 (10.4)	62 (43.1)
						Metformin 1000 mg bid	147	55.2 (10.6)	69 (46.9)
						Linagliptin 2.5 mg qd + Metformin 500 mg bid	143	55.6 (11.2)	73 (49.0)
						Linagliptin 2.5 mg qd + Metformin 1000 mg bid	143	56.4 (10.7)	66 (46.2)
						Placebo	72	55.7 (11.0)	36 (50)
NCT00328172[23]	Multi-national (71 sites in 6 countries)	2007	65	12	Type 2 diabetes	Linagliptin 0.5 mg	58	58.0 (9.4)	13 (22.4)
						Linagliptin 2.5 mg	57	59.8 (10.3)	30 (52.6)
						Linagliptin 5.0 mg	55	56.6 (9.6)	24 (43.6)
						Metformin	65	53.7 (10.7)	26 (40.0)
						Placebo	67	58.6 (8.9)	34 (50.7)
Yki-Jarvinen et al[24,25]	Multi-national (169 sites in 19 countries)	2011	108	52	Type 2 diabetes	Linagliptin 5.0 mg	631	59.7 (9.9)	302 (47.9)
						Placebo	630	60.4 (10.0)	301 (47.8)
NCT00654381[26]	Japan	2010	91	12	Type 2 diabetes	Linagliptin 5.0 mg	159	60.3 (9.4)	48 (30.2)
						Linagliptin 10.0 mg	160	61.3 (10.0)	48 (30.0)
						Voglibose	162	58.5 (9.9)	47 (29.0)
						Placebo	80	59.7 (8.9)	23 (28.7)
NCT00622284[27]	Multi-national (221 sites in 16 countries)	2010	146	104	Type 2 diabetes	Linagliptin	776	59.8 (9.4)	314 (40.5)
						Glimepiride	775	59.8 (9.4)	304 (39.2)
BI Trial No: 1218.15/ U09-2519-01[28]	Multi-national (43 sites in 7 countries)	2009	61	24	Type 2 diabetes	Linagliptin 5 mg + Pioglitazone 30 mg	259	NR	NR
						Pioglitazone 30 mg + Placebo	130	NR	NR
BI Trial No: 1218.52/U11-1782-01[29]	Multi-national (101 sites in 14 countries)	2011	102	54	Type 2 diabetes	Linagliptin 2.5 mg + Metformin (500 mg and 1000 mg bid)	396	NR	NR
						Metformin 1000 mg bid	170	NR	NR
BI Trial No: 1218.63/U11-1781-02[30]	Multi-national (33 sites in 5 countries)	2011	67	24	Type 2 diabetes	Linagliptin 5 mg	162	NR	46 (28.4)
						Placebo	79	NR	30 (38.0)
BI Trial No: 1218.75/U12-3204-01[31]	Multi-center study (Black/African American patients only)	2011	55	24	Type 2 diabetes	Linagliptin 5 mg	106	NR	NR
						Placebo	120	NR	NR
BI Trial No: 1218.61/U13-3124-01[32]	Multi-national study (4 countries)	2012	123	24	Type 2 diabetes	Linagliptin 5 mg	183	NR	NR
						Placebo	89	NR	NR
BI Trial No: 1218.65/U12-2143-01[33]	Multi-national study (19 sites in 3 countries)	2012	74	24	Type 2 diabetes	Linagliptin 5 mg	205	82%(< 65 yr)	NR
						Placebo	100	83% (< 65 yr)	NR
BI Trial No: 1218.64/U13-1283-01[34]	Multi-national study (52 sites in 9 countries)	2012	117	52	Type 2 diabetes	Linagliptin 5 mg	113	NR	43 (38.1)
						Placebo (first 12 wk)/ Glimepiride (next 40 wk)	122	NR	43 (35.2)
BI Trial No: 1218.66/U12-2076-01[35]	Multi-national study (19 sites in 3 countries	2012	80	24	Type 2 diabetes	Linagliptin 5 mg	200	84.0% (< 65 yr)	NR
						Placebo	99	89.9% (< 65 yr)	NR
Rosenstock et al[36]	Multi-national study (110 sites in 13 countries)	2007	65	26	Type 2 diabetes	Alogliptin 12.5 mg	131	55.4 (9.8)	76 (58)
						Alogliptin 25 mg	129	55.9 (10.2)	85 (66)
						Placebo	130	55.0 (10.6)	68 (52)
White et al[37]	Multi-national study (898 centers in 49 countries)	2013	193	76 (median)	Type 2 diabetes	Alogliptin	2701	36.0% (≥ 65 yr)	873 (32.3)
						Placebo	2679	34.9% (≥ 65 yr)	856 (32.0)
NCT01318135[38]	Japan (58 sites)	2010	52	52	Type 2 diabetes	Alogliptin 12.5 mg qd + Glimepiride 1-6 mg qd or bid	150	38.0% (≥ 65 yr)	53 (35.3)
						Alogliptin 25 mg qd + Glimepiride 1-6 mg qd or bid	152	30.9% (≥ 65 yr)	52 (34.2)
NCT01289119[39]	Multi-national study (21 sites in 3 countries)	2011	52	16	Type 2 diabetes	Alogliptin monotherapy	92	51.6 (10.41)	37 (40.2)
						Metformin	98	53.2 (9.46)	50 (51.0)
						Metformin + Alogliptin Add-on Therapy	99	53.0 (9.88)	48 (48.5)
						Pioglitazone	63	51.8 (10.37)	24 (38.1)
						Pioglitazone + Alogliptin Add-on Therapy	61	52.6 (9.44)	33 (54.1)
						Placebo	93	53.1 (8.88)	39 (41.9)
NCT01263496[40]	Japan (58 sites)	2008	72	52	Type 2 diabetes	Alogliptin 6.25 mg qd	96	28.1 (≥ 65 yr)	26 (27.1)
						Alogliptin 12.5 mg qd	101	33.7 (≥ 6 5 yr)	29 (28.7)
						Alogliptin 25 mg qd	97	34.0 (≥ 65 yr)	22 (22.7)
						Alogliptin 50 mg qd	97	32.9 (≥ 65 yr)	29 (29.9)
						Voglibose 0.2 mg tid	83	38.6 (≥ 65 yr)	27 (32.5)
NCT00328627[41]	Multi-national study (90 sites in 19 countries)	2008	93	26	Type 2 diabetes	Alogliptin 12.5 mg + Placebo	128	53.1 (9.59)	61 (47.6)
						Alogliptin 25 mg + Placebo	129	53.7 (9.31)	79 (61.2)
						Placebo	129	55.2 (9.89)	68 (52.7)
NCT00395512[42]	Multi-national study (268 sites in 23 countries)	2008	67	26	Type 2 diabetes	Alogliptin 25 mg + Pioglitazone 30 mg	164	52.8 (11.01)	91 (55.5)
						Alogliptin 12.5 mg + Pioglitazone 30 mg	164	53.5 (11.37)	83 (50.6)
						Pioglitazone 30 mg	163	51.5 (10.72)	73 (44.8)
Kikuchi et al[43]	Japan (26 sites)	2007	52	12	Type 2 diabetes	Vildagliptin 50 mg bid + glimepiride	102	59.2 (9.8)	27 (26.5)
						Placebo + glimepiride	100	60.3 (10.1)	31 (31.0)
Lukashevich et al[44]	Multi-national study (12 countries)	2010	291	24	Type 2 diabetes	Vildagliptin 50 mg qd (moderate RI)	165	67.7 (8.8)	69 (41.8)
						Placebo (moderate RI)	129	69.7 (7.3)	49 (38.0)
						Vildagliptin 50 mg qd (severe RI)	124	64.1 (9.2)	59 (47.6)
						Placebo (severe RI)	97	64.5 (10.8)	44 (45.4)
Strain et al[45]	Multi-national study (45 centers in 7 countries)	2012	64	24	Type 2 diabetes	Vildagliptin	139	75.1 (4.3)	66 (47.5)
						Placebo	139	74.4 (4.0)	86 (61.9)
NCT00106340[46] (CLAF237A2308)	Multi-national study (402 centers in 25 countries)	2008	166	104	Type 2 diabetes	Vildagliptin 50 mg bid + Metformin	1562	57.5 (9.07)	733 (46.9)
						Glimepiride up to 6 mg qd + Metformin	1556	57.5 (9.19)	718 (46.1)
NCT00300287[47] (CLAF237A2307)	Multi-national study (69 centers in 6 countries)	2006	85	52	Type 2 diabetes	Vildagliptin 50 mg qd	156	63.27 (10.18)	63 (40.4)
						Placebo	150	62.84 (11.03)	61 (40.7)
CLAF237A1301[48]	Japan (51 centers)	2007	44	12	Type 2 diabetes	Vildagliptin 50 mg bid	188	60.3 (10.48)	67 (35.6)
						Voglibose 0.2 mg tid	192	58.0 (9.32)	62 (32.3)
CLAF237A23119[49]	United States (796 centers)	2007	53	12	Type 2 diabetes	Vildagliptin 100 mg + Metformin	1776	55.3	864 (48.6)
						Thiazolinedione + Metformin	888	56.2	467 (52.6)
NCT00110240[50] (CLAF237A2323)	Multi-national study (31 centers in 3 countries)	2006	87	24	Type 2 diabetes	Vildagliptin 50 mg bid	441	51.79 (10.13)	176 (40.0)
						Acarbose up to 100 mg tid	220	51.93 (10.34)	81 (37.0)
NCT00327015[51]	Multi-national study (211 sites in 13 countries)	2007	78	24	Type 2 diabetes	Saxagliptin 5 mg + Metformin 500 mg	320	51.95 (10.43)	155 (48.4)
						Saxagliptin 10 mg + Metformin 500 mg	323	52.08 (11.59)	177 (54.8)
						Metformin 500 mg + Placebo	328	51.83 (10.74)	165 (50.3)
Hollander et al[52]	Multi-national study (133 sites in 7 countries)	2007	82	24	Type 2 diabetes	Saxagliptin 2.5 mg + TZD	195	54.9 (9.7)	89 (45.6)
(NCT00295633)						Saxagliptin 5 mg + TZD	186	53.2 (10.6)	97 (52.2)
						Placebo + TZD	184	54.0 (10.1)	99 (53.8)
NCT00757588[53]	Multi-national study (80 sites in 10 countries)	2010	73	24	Type 2 diabetes	Saxagliptin 5 mg + Insulin	304	57.2 (9.4)	184 (60.5)
						Placebo + Insulin	151	57.3 (9.3)	83 (54.9)
Scirica et al[54]	Multi-national study (788 sites in 26 countries)	2013	156	109	Type 2 diabetes	Saxagliptin	8280	65.1 (8.5)	2768 (33.4)
						Placebo	8212	65 (8.6)	2687 (32.7)
Göke et al[55]	Multi-national study (130 sites in 11 countries)	2010	139	104	Type 2 diabetes	Saxagliptin + Metformin	428	57.5 (10.26)	216 (50.5)
						Glipizide + Metformin	430	57.59 (10.37)	198 (46.1)
NCT00316082[56]	Multi-national study (74 sites in 4 countries)	2007	74	24	Type 2 diabetes	Saxagliptin 2.5/5 mg QAM	71	54.28 (10.93)	34 (47.9)
						Saxagliptin 2.5 mg QAM	74	55.24 (10.44)	49 (66.2)
						Saxagliptin 5 mg QAM	74	54.66 (9.71)	36 (48.6)
						Saxagliptin 5 mg QPM	72	55.11 (10.35)	39 (54.2)
						Placebo	74	55.57 (10.32)	39 (52.7)
NCT00614939[57]	Multi-national study (74 sites in 14 countries)	2009	74	12	Type 2 diabetes	Saxagliptin	85	66.8 (8.3)	53 (62.4)
						Placebo	85	66.2 (9.1)	44 (51.8)
Chan et al[58,59]	Multi-national study (30 sites in 13 countries)	2006	NR	54	Type 2 diabetes	Sitagliptin 50 mg or 25 mg once daily	65	68.9 (9.8)	34 (52.3)
						Glipizide	26	65.3 (9.7)	10 (38.5)
Kojima et al[60]	Japan (Japanese Red Cross Medical Center)	2011	65	12	Type 2 diabetes	Sitagliptin	20	63.85 (12.92)	5 (0.25)
						Nateglinide	16	66.44 (9.02)	4 (0.25)
NCT00509262	Multi-national study	2011	178	54	Type 2 diabetes	Sitagliptin	211	64.2 (10.7)	80 (37.9)
(Arjona Ferreira et al[61,62])
						Glipizide	212	64.2 (9.4)	90 (42.5)
Henry et al[63,64]	Multi-national study	2010	108	54	Type 2 diabetes	Sitagliptin 100 mg/Pioglitazone 15 mg	230	NR	112 (48.7)
						Sitagliptin 100 mg/Pioglitazone 30 mg	231	NR	96 (41.6)
						Sitagliptin 100 mg/Pioglitazone 45 mg	230	NR
									95 (41.3)
						Pioglitazone 15 mg	230	NR	82 (35.7)
						Pioglitazone 30 mg	233	NR	106 (45.5)
						Pioglitazone 45 mg	230	NR	117 (50.9)
Raz et al[65,66]	Multi-national study (30 sites in 13 countries)	2007	47	30	Type 2 diabetes	Sitagliptin 100 mg	96	53.6 (9.5)	47 (48.9)
						Placebo	94	56.1 (9.5)	55 (58.5)
NCT01131182[67]	NR	2010	22	4	Type 2 diabetes	Sitagliptin	507	55.0 (11.0)	238 (46.9)
						Sulfonylurea	514	55.0 (11.0)	259 (50.4)
Goldstein et al[68,69]	Multi-national study	2006	69	54	Type 2 diabetes	Sitagliptin 50 mg bid + Metformin 500 mg bid	190	54.1 (10.0)	85 (44.7)
						Sitagliptin 50 mg bid + Metformin 1000 mg bid	182	53.3 (9.6)	105 (57.7)
						Sitagliptin 50 mg bid + Metformin 1000 mg bid (Open Label Cohort)	117	52.6 (10.0)	50 (42.7)
						Metformin 500 mg bid	182	53.4 (10.2)	93 (51.1)
						Metformin 1000 mg bid	182	53.2 (9.6)	100 (54.9)
						Placebo/Metformin 1000 mg bid	176	53.6 (10.0)	83 (47.2)
Arechavaleta et al[70,71]	Multi-national study	2009	74	30	Type 2 diabetes	Sitagliptin	516	56.3 (9.7)	232 (44.9)
						Glimepiride	519	56.2 (10.1)	240 (46.2)
NCT00086515 et al[72,73]	Multi-national study	2007	135	24	Type 2 diabetes	Sitagliptin 100 mg	464	54.4 (10.4)	205 (44.2)
						Placebo/Glipizide 5 mg	237	54.7 (9.7)	96 (40.5)
Bergenstal et al[74,75]	Multi-national study (62 sites in 3 countries)	2009	56	26	Type 2 diabetes	Exenatide once weekly	160	52.4 (10.41)	71 (44.4)
						Sitagliptin	166	52.2 (10.54)	80 (48.2)
						Pioglitazone	165	53.0 (9.92)	86 (52.1)
NCT00094757[76]	Multi-national study	2006	78	54	Type 2 diabetes	Sitagliptin 100 mg	205	54.5 (10.0)	95 (46.3)
						Sitagliptin 200 mg	206	55.4 (9.2)	102 (49.5)
						Placebo/Pioglitazone	110	55.5 (10.1)	41 (37.3)
NCT00094770[77]	Multi-national study (173 sites in 27 countries)	2006	139	104	Type 2 diabetes	Sitagliptin 100 mg	588	56.8 (9.3)	252 (42.8)
						Glipizide	584	56.6 (9.8)	226 (38.7)
NCT01137812[78,79]	Multi-national study (182 sites in 17 countries)	2012	87	52	Type 2 diabetes	Sitagliptin 100 mg	378	56.6 (9.33)	163 (43.1)
						Canagliflozin 300 mg	377	56.5 (9.62)	170 (45.1)
NCT00482729[80]	Multi-national study (209 sites in United States)	2008	74	44	Type 2 diabetes	Sitagliptin/Meformin-Fixed Dose Combination	625	49.5 (10.5)	272 (43.5)
						Metformin	621	50.0 (10.5)	266 (42.8)
Bunck et al[81]	Multi-national study (3 sites in 3 countries)	2007	154	52	Type 2 diabetes	Exenatide	36	58.4 (1.4)	13 (36.1)
						Insulin glargine	33	58.3 (1.3)	11 (33.3)
Diamant et al[82]	Multi-national study (72 sites in 7 countries)	2009	53	26	Type 2 diabetes	Exenatide	233	58.0 (10.0)	113 (48.0)
						Insulin glargine	223	58 (9.0)	100 (45.0)
Inagaki et al[83]	Japan (22 sites)	2010	61	26	Type 2 diabetes	Exenatide once weekly	215	57.07 (10.44)	73 (34.0)
						Insulin glargine once daily	212	56.44 (11.16)	64 (30.2)
Russell-Jones et al[84]	Multi-national study (106 sites in 22 countries)	2010	82	26	Type 2 diabetes	Exenatide 2 mg once weekly + Oral placebo	248	53.7 (10.91)	109 (43.9)
						Sitagliptin 100 mg/d + SC placebo	163	52.3 (11.05)	69 (42.3)
						Metformin starting at 1000 mg/d + SC placebo	246	53.7 (11.08)	92 (37.4)
						Pioglitazone starting at 30 mg/d+ SC placebo	163	55.3 (10.96)	66 (40.5)
NCT01003184[85]	34 sites in Ireland and United Kingdom	2011	91	26	Type 2 diabetes	Exenatide once weekly	111	59.2 (9.86)	40 (36.04)
						Insulin Detemir twice daily	105	57.8 (9.48)	33 (31.4)
Astrup et al[86]	Multi-national study (19 sites in 8 European countries)	2009	117	104	Type 2 diabetes	Liraglutide 1.2 mg	95	47.18 (9.72)	73 (76.8)
						Liraglutide 1.8 mg	90	45.53 (10.9)	68 (75.6)
						Liraglutide 2.4 mg	93	45.01 (11.09)	71 (76.3)
						Liraglutide 3.0 mg	93	45.91 (10.71)	70 (75.3)
						Placebo	98	45.86 (10.28)	74 (75.5)
Garber et al[87]	126 sites in United States and 12 sites in Mexico	2007	91	52	Type 2 diabetes	Liraglutide 1.2 mg	251	53.7 (11.0)	134 (53.4)
						Liraglutide 1.8 mg	247	52.0 (10.8)	126 (51.0)
						Glimepiride 8 mg	248	53.4 (10.9)	115 (46.4)
Nauck et al[88]	Multi-national study (170 sites in 21 countries)	2007	52	26	Type 2 diabetes	Once daily Liraglutide (0.6 mg)	242	56.0 (11.0)	91 (37.6)
						Once daily Liraglutide (1.2 mg)	241	57 (9.0)	111 (46.1)
						Once daily Liraglutide (1.8 mg)	242	57 (9.0)	100 (41.3)
						Once daily Glimepiride (4 mg)	244	57 (9.0)	103 (42.2)
						Placebo	122	56 (9.0)	49 (40.2)
Marre et al[89]	Multi-national study (116 sites in 21 countries)	NR	NR	26	Type 2 diabetes	Liraglutide 0.6 mg	233	55.7 (9.9)	107 (46.0)
						Liraglutide 1.2 mg	228	57.7 (9.0)	125 (55.0)
						Liraglutide 1.8 mg	234	55.6 (10.0)	110 (47.0)
						Placebo	114	54.7 (10.0)	60 (53.0)
Zinman et al[90]	90 sites in United States and Canada	2007	65	26	Type 2 diabetes	Liraglutide 1.2 mg	178	55.0 (10.0)	77 (43.0)
						Liraglutide 1.8 mg	178	55.0 (11.0)	87 (49.0)
						Placebo	177	55.0 (10.0)	67 (38.0)
Raz et al[91]	Multi-national study (53 centers in 11 countries)	2011	134	24	Type 2 diabetes	Taspoglutide 10 mg	116	NR	NR
						Taspoglutide 20 mg	129	NR	NR
						Placebo	123	NR	NR
Rosenstock et al[92]	Multi-national study (118 sites in 4 countries)	2008	56	16	Type 2 diabetes	Albiglutide 4 mg weekly	35	50.4 (10.3)	20 (57.1)
						Albiglutide 15 mg weekly	35	55.5 (10.5)	17 (48.6)
						Albiglutide 30 mg weekly	31	54.2 (9.7)	23 (74.2)
						Albiglutide 15 mg biweekly	33	52.5 (9.6)	19 (57.6)
						Albiglutide 30 mg biweekly	32	55.5 (9.9)	16 (50.0)
						Albiglutide 50 mg biweekly	35	51.1 (10.3)	16 (45.7)
						Albiglutide 50 mg monthly	35	54.1 (11.3)	18 (51.4)
						Albiglutide 100 mg monthly	34	54.4 (9.9)	15 (44.1)
						Placebo	51	54.0 (10.6)	23 (45.1)
Seino et al[93]	Multi-national study (57 centers in 4 Asian countries)	NR	NR	24	Type 2 diabetes	Lixisenatide (10 ug for 1 wk, 15 mg for 1 wk, then 20 mg-maintenance dose)	154	58.7 (10.2)	85 (55.2)
						Placebo	157	58.0 (10.1)	77 (49.0)
Umpierrez et al[94]	36 sites in United States and 3 in Puerto Rico	2008	39	16	Type 2 diabetes	LY2189265 (LY 0.5/1.0)	66	59.0 (12.0)	31 (47.0)
						LY2189265 (LY 1.0/1.0)	65	57.0 (12.0)	30 (46.0)
						LY2189265 (LY 1.0/2.0)	65	54.0 (11.0)	31 (48.0)
						Placebo	66	56.0 (12.0)	37 (56.0)

Table 2 Quality assessment of glucagon-like peptide-1 based agents in randomized controlled trials included in analysis of pancreatic events.

Ref.	Sequence generation	Blinding	Allocation concealment	Was Pancreatitis an AE or SAE?	Adverse event monitoring	Arms	Withdrawal rate (%)	Loss to follow- up (%)
Ross et al[21]	Central computer based; randomization: block in a 5:5:1 ratio	Double blind	Adequate	AE	Safety and tolerability end-points were the incidence of adverse events (including adverse changes observed during physical examinations or ECGs), protocol-specified significant AEs, hypoglycemia and changes from baseline in vital signs, clinical laboratory parameters and body weight	Linagliptin 2.5 mg bid	7.2	0
						Linagliptin 5 mg qd	4.5	0
						Placebo	2.3	0
Haak et al[22]	NR	Double blind	Adequate	AE	Incidence of AEs, serious AEs, discontinuation due to AEs,12-lead ECGs, vital signs and clinical laboratory parameters. The causal relationships between study medications and AEs were evaluated by the investigators at the site	Linagliptin 5 mg qd	14.8	2.1
						Metformin 500 mg bid	11.8	2.1
						Metformin 1000 mg bid	14.3	2.7
						Linagliptin 2.5 mg qd + Metformin 500 mg bid	11.2	2.8
						Linagliptin 2.5 mg qd + Metformin 1000 mg bid	7.7	0
						Placebo	25.0	1.4
NCT00328172[23]	NR	Double blind	NR	SAE	NR	Linagliptin 0.5 mg	24.1	1.7
						Linagliptin 2.5 mg	17.5	3.5
						Linagliptin 5.0 mg	23.6	1.8
						Metformin	7.7	1.5
						Placebo	32.8	1.5
Yki-Jarvinen et al[24,25]	NR	Double blind	NR	SAE	NR	Linagliptin 5.0 mg	13.9	2.2
						Placebo	17.5	1.3
NCT00654381[26]	NR	Double blind	NR	SAE	NR	Linagliptin 5.0 mg	1.89	0
						Linagliptin 10.0 mg	3.13	0
						Voglibose	2.5	0
						Placebo	7.5	0
NCT00622284[27]	NR	Double blind	NR	SAE	NR	Linagliptin	24.4	1.4
						Glimepiride	22.1	1.7
BI Trial No: 1218.15/ U09-2519-01[28]	Randomized into 1;2 ratio to receive either placebo or linagliptin	Double blind	Adequate	SAE	Incidence and intensity of AEs, withdrawals due to AEs, physical examination, 12-lead ECG, vital signs, clinical laboratory parameters	Linagliptin 5 mg + Pioglitazone 30 mg	5.8	NR
						Pioglitazone 30 mg + Placebo	14.6	NR
BI Trial No: 1218.52/U11-1782-01[29]	NR	Double blind	NR	SAE	Safety endpoints were the incidence and intensity of AEs, withdrawals due to AEs, clinically relevant new or worsening findings in physical examination, 12-lead ECG, vital signs and clinical laboratory parameters	Linagliptin 2.5 mg + Metformin (500 mg and 1000 mg bid)	0.0	NR
						Metformin 1000 mg bid	0.6	NR
BI Trial No: 1218.63/U11-1781-02[30]	NR	Double blind	NR	SAE	Incidence and intensity of AEs, withdrawals due to AEs, physical examination, 12-lead ECG, vital signs, clinical laboratory parameters	Linagliptin 5 mg	1.23	NR
						Placebo	1.26	NR
BI Trial No: 1218.75/U12-3204-01[31]	NR	Double blind	NR	AE	Incidence and intensity of AEs, withdrawals due to AEs, clinically relevant changes from baseline in vital signs (blood pressure and pulse rate), clinically relevant new or worsening findings in 12-lead ECG as reported as AEs, clinically relevant changes from baseline in clinical laboratory assessments, cardiac and cerebrovascular events adjudicated CEC	Linagliptin 5 mg	12.3	NR
						Placebo	12.5	NR
BI Trial No: 1218.61/U13-3124-01[32]	NR	Double blind	NR	AE	Incidence and intensity of AEs, primarily based on spontaneous AEs; withdrawal due to AEs; clinically relevant new or worsening findings in physical examination reported as AEs; changes from baseline in vital signs (BP and pulse); clinically relevant new or worsening findings in 12 lead ECG reported as AEs; changes from baseline in clinical lab assessments; and hypoglycemic events	Linagliptin 5 mg	2.2	NR
						Placebo	0.0	NR
BI Trial No: 1218.65/U12-2143-01[33]	NR	Double blind	NR	SAE	Incidence and intensity of adverse events, withdrawals due to AEs, physical examination, ECGs, change from baseline in clinical lab parameters and cardiovascular events (Clinical Event Committee adjudication results)	Linagliptin 5 mg	0.98	NR
						Placebo	3.0	NR
BI Trial No: 1218.64/U13-1283-01[34]	NR	Double blind	NR	AE	Incidence and intensity of adverse events (AEs), withdrawals due to AEs, physical examination, vital signs, 12 lead ECG, change from baseline in clinical lab parameters	Linagliptin 5 mg	0.0	NR
						Placebo (first 12 wk)/ Glimepiride (next 40 wk)	1.64	NR
BI Trial No: 1218.66/U12-2076-01[35]	NR	Double blind	NR	SAE	Incidence and intensity of adverse events, withdrawals due to AEs, physical examination and vital signs, 12-lead ECG, clinical laboratory assessments	Linagliptin 5 mg	5.1	NR
						Placebo	2.0	NR
Rosenstock et al[36]	Automated interactive voice response system using a randomization schedule	Double blind	NR	SAE	During the treatment period, patients were reviewed for adverse event evaluations. Further safety assessments included clinical examination of skin and digits. Hematology, serum chemistry, vital signs, physical exam and ECG parameters were done	Alogliptin 12.5 mg	36.6	3.05
						Alogliptin 25 mg	40.3	2.33
						Placebo	57.7	1.54
White et al[37]	NR	Double blind	NR	SAE	The principal secondary safety end point was the primary composite end point with the addition of urgent revascularization due to unstable angina within 24 h after hospital admission. Additional safety end points included angioedema, hypoglycemia, pancreatitis, cancer, and the results of laboratory testing	Alogliptin	NR	NR
						Placebo	NR	NR
NCT01318135[38]	NR	Open Label	Inadequate	SAE (Pancreatic cancer only)		Alogliptin 12.5 mg qd + Glimepiride 1-6 mg qd or bid	NR	NR
						Alogliptin 25 mg qd + Glimepiride 1-6 mg qd or bid	NR	NR
NCT01289119[39]	NR	Double blind	NR	SAE	TEAE were defined as any adverse events that started on or after the date of the first dose of double-blind study drug and within 14 d after the date of the last dose of double-blind study drug	Alogliptin monotherapy	9.78	3.26
						Metformin	9.18	0
						Metformin + Alogliptin Add-on Therapy	6.06	0
						Pioglitazone	7.94	0
						Pioglitazone + Alogliptin Add-on Therapy	6.56	1.64
						Placebo	9.78	0
NCT01263496[40]	NR	Open Label	Inadequate	SAE	A TEAE is defined as an adverse event with an onset that occurs after receiving study drug and within 30 d after receiving the last dose of study drug	Alogliptin 6.25 mg qd	NR	NR
						Alogliptin 12.5 mg qd	NR	NR
						Alogliptin 25 mg qd	NR	NR
						Alogliptin 50 mg qd	NR	NR
						Voglibose 0.2 mg tid	NR	NR
NCT00328627[41]	NR	Double blind	NR	SAE	NR	Alogliptin 12.5 mg + Placebo	24.2	1.56
						Alogliptin 25 mg + Placebo	21.7	1.55
						Placebo	45.7	3.1
NCT00395512[42]	NR	Double blind	Adequate	SAE	NR	Alogliptin 25 mg + Pioglitazone 30 mg	17.1	3.05
						Alogliptin 12.5 mg + Pioglitazone 30 mg	23.2	3.05
						Pioglitazone 30 mg	22.7	3.68
Kikuchi et al[43]	Dynamic randomization	Double blind	NR	SAE	Adverse events were recorded at each visit, and these AEs were assessed for severity and suspected relationship to the study drug. Hematology, biochemistry and urinalysis were performed at each scheduled visit. All laboratory assessments were processed at a central testing to ensure consistency	Vildagliptin 50 mg bid + glimepiride	2.9	NR
						Placebo + glimepiride	4	NR
Lukashevich et al[44]	NR	Double blind	NR	SAE	All treatment emergent AEs were recorded and assessed by the investigator as to severity and potential relationship to study drug. Particular attention was paid to hepatic, infections, skin, pancreatitis as well as edema and cardiovascular safety	Vildagliptin 50 mg qd (moderate RI)	10.3	2.4
						Placebo (moderate RI)	10.9	1.6
						Vildagliptin 50 mg qd (severe RI)	13.7	1.6
						Placebo (severe RI)	13.4	2.1
Strain et al[45]	Validated automated system	Double blind	Adequate	AE	All AEs and their severity, serious AEs, and their presumed relation with the study drug were monitored and recorded at each study visit	Vildagliptin	5.8	0.72
						Placebo	5.8	0
NCT00106340[46] (CLAF237A2308)	NR	Double blind	NR	SAE	Safety assessments included monitoring and recording all AEs, SAEs and pregnancies; regular monitoring of hematology, blood chemistry, and urine (performed at a central lab); and regular assessments of vital signs, ECG, physical condition and body weight. Severity and relationship to study drug were recorded for all AEs and SAEs	Vildagliptin 50 mg bid + Metformin	36.4	0
						Glimepiride up to 6 mg qd + Metformin	38.8	0
NCT00300287[47]	NR	Double blind	NR	SAE	Safety assessments included monitoring and recording all AEs, SAEs with their severity and presumed relationship to study drug and pregnancies, recording of hypoglycemic events, the regular monitoring of hematology, blood chemistry and urine, and regular assessments of vital signs, physical condition, body weight, and ECGs	Vildagliptin 50 mg qd	14.7	0.6
						Placebo	12.7	0.7
(CLAF237A2307) CLAF237A1301[48]	NR	Double blind	NR	AE (elevated pancreatic enzymes)	Safety assessments included monitoring and recording all AEs, SAEs with their severity and presumed relationship to study drug and pregnancies, recording of hypoglycemic events, the regular monitoring of hematology, blood chemistry and urine, and regular assessments of vital signs, physical condition, body weight, and ECGs	Vildagliptin	4.8	NR
						50 mg bid Voglibose 0.2 mg tid	5.2	NR
CLAF237A23119[49]	NR	Open Label	NA	SAE	Safety assessments included monitoring and recording all AEs, SAEs with their severity and presumed relationship to study drug and pregnancies, recording of hypoglycemic events, the regular monitoring of hematology, blood chemistry and urine, and regular assessments of vital signs, physical condition, body weight, and ECGs	Vildagliptin 100 mg + Metformin	10.4	2.5
						Thiazolinedione + Metformin	11.8	2.1
NCT00110240[50] (CLAF237A2323)	NR	Double Blind	NR	SAE	Safety assessments included adverse events, hypoglycemic events and serious adverse events, physical examination, vital signs, laboratory evaluations, and ECGs	Vildagliptin 50 mg bid	9.5	1.6
						Acarbose up to 100 mg tid	12.7	1.4
NCT00327015[51]	NR	Double Blind	NR	SAE	Safety and tolerability end-points included incidence of AEs, SAEs, discontinuation due to AEs, physical and ECG examinations, vital signs and results of clinical laboratory tests	Saxagliptin 5 mg + Metformin 500 mg	28.4	6.9
						Saxagliptin 10 mg + Metformin 500 mg	28.5	7.1
						Metformin 500 mg + Placebo	33.2	6.7
Hollander et al[52] (NCT00295633)	NR	Double Blind	NR	SAE	Safety assessments included incidence of AEs, SAEs and discontinuation due to AEs, changes from baseline lab parameters; changes from baseline vital signs; and incidence of marked clinical laboratory abnormalities	Saxagliptin 2.5 mg + TZD	31.8	NR
						Saxagliptin 5 mg + TZD	36	NR
						Placebo + TZD	41.3	NR
NCT00757588[53]	Interactive voice response system	Double Blind	NR	SAE	Safety end points included AEs, hypoglycemia and weight gain	Saxagliptin 5 mg + Insulin	11.8	0.98
						Placebo + Insulin	11.3	3.31
Scirica et al[54]	Central computerized telephone or web based system	Double Blind	NR	NR (Safety End Point)	A clinical events committee comprising specialists in cardiovascular and pancreatic medicine, all of whom were unaware of the study group assignments, adjudicated	Saxagliptin	NR	NR
						Placebo	NR	NR
Goke et al[55]	NR	Double Blind	NR	SAE	Safety and tolerability assessments included AEs and SAEs, lab measurements, vital signs, physical examination and ECG testing	Saxagliptin + Metformin	61.4	0.23
						Glipizide + Metformin	65.8	0.69
NCT00316082[56]	NR	Double Blind	NR	SAE	NR	Saxagliptin 2.5/5 mg QAM	38.0	9.9
						Saxagliptin 2.5 mg QAM	44.6	9.5
						Saxagliptin 5 mg QAM	29.7	8.1
						Saxagliptin 5 mg QPM	36.1	11.1
						Placebo	35.1	8.1
NCT00614939[57]	Interactive voice response system	Double Blind	NR	SAE	Safety and tolerability assessments included AEs, SAEs, treatment-related AEs, discontinuations of randomized study medication because of AEs, deaths, AEs of special interest and hypoglycemic events	Saxagliptin	71.8	NR
						Placebo	80.0	NR
Chan et al[58,59]	Computer generated randomization schedule	Double Blind	Adequate	SAE	Assessment of safety and tolerability included evaluation of the data from physical examinations, vital signs and ECGs collected at specified study visits. All adverse experiences were rated by the investigators for intensity and relationship to study drug	Sitagliptin 50 mg or 25 mg once daily	29.2	NR
						Placebo/Glipizide	23.1	NR
Kojima et al[60]	Random allocation sequence performed centrally	Open label	NA	AE	NR	Sitagliptin	NR	NR
						Nateglinide	NR	NR
NCT00509262 (Arjona Ferreira JC et al[61,62])	Computer generated randomization schedule	Double	NR	SAE	Safety measurements included evaluation of AEs, physical exam and vital signs, and ECG. Lab safety studies included serum chemistry, hematology and urinalysis. All AEs were rated by the investigator for intensity and relationship to study drug	Sitagliptin	210
						Glipizide	212
Henry RR et al[63,64]	NR	Blind Double blind	NR	SAE	Safety and tolerability were evaluated throughout the study by physical examination, monitoring of vital signs and safety lab measurements that included serum chemistry, hematology and urinalysis. AEs were monitored and evaluated by the investigators for intensity (severity), duration, outcome and relationship to study drug	Sitagliptin 100 mg/Pioglitazone 15 mg	20.9	3.5
						Sitagliptin 100 mg/Pioglitazone 30 mg	22.9	6.9
						Sitagliptin 100 mg/Pioglitazone 45 mg	22.2	5.7
						Pioglitazone 15 mg	31.3	6.1
						Pioglitazone 30 mg	27.9	9
						Pioglitazone 45 mg	27.4	5.7
Raz I et al[65,66]	Computer generated schedule	Double blind	NR	SAE	Safety and tolerability were evaluated by physical examination, vital signs and lab measurements that included routine serum chemistry, hematology, urinalysis and pregnancy testing. AEs were monitored through the study for intensity, duration, outcome, relationship to study drug and level of severity	Sitagliptin 100 mg	17.7
								3.13
						Placebo	14.9	3.19
NCT01131182[67]	NR	Open label	NA	SAE	NR	Sitagliptin	NR	NR
						Sulfonylurea	NR	NR
Goldstein et al[68,69]	NR	Double blind	NR	SAE	Data were collected regarding AEs, physical exam, vital signs, ECGs and body weight throughout the study. All AEs were rated by investigators for intensity and relationship to study drug	Sitagliptin 50 mg bid + Metformin 500 mg bid	22.1	2.6
						Sitagliptin 50 mg bid + Metformin 1000 mg bid	22.5	5.5
						Sitagliptin 50 mg bid + Metformin 1000 mg bid (OLC)	32.5	2.6
						Metformin 500 mg bid	30.8	2.2
						Metformin 1000 mg bid	25.8	3.8
						Placebo/Metformin 1000 mg bid	34.7	5.1
Arechavaleta et al[70,71]	Concealed computer-generated allocation schedule	Double blind	Adequate	SAE	Safety and tolerability were assessed by a review of all safety parameters including adverse experiences, laboratory safety parameters, body weight and vital signs	Sitagliptin	9.3	1.7
						Glimepiride	9.8	1.7
NCT00086515 et al[72,73]	NR	Double blind	NR	SAE	Safety and tolerability were assessed throughout the study. Monitoring for adverse experiences, physical examinations, vital signs, body weight, 12-lead ECGs (read at a central reading laboratory), and safety laboratory measurements comprising routine hematology, serum chemistry, and urinalysis were performed	Sitagliptin 100 mg	10.6	0.86
						Placebo/Glipizide 5 mg	18.9	2.11
Bergenstal et al[74,75]	Interactive voice response system	Double blind	Adequate	SAE	NR	Exenatide once weekly	26.9	5
						Sitagliptin	16.9	5.4
						Pioglitazone	24.8	7.8
NCT00094757[76]	NR	Double blind	NR	SAE	Data for adverse experiences, physical examinations, vital signs, ECGs, and body weight were collected throughout the study	Sitagliptin 100 mg	25.8	1.5
						Sitagliptin 200 mg	30.1	2.4
						Placebo/Pioglitazone	27.3	5.4
NCT00094770[77]	NR	Double blind	NR	SAE	Data on adverse experiences, physical examinations, vital signs, ECGs and body weight were collected throughout the study. All adverse experiences were rated by the study site investigators for intensity and relationship to study drug. Laboratory safety evaluations included blood chemistry, haematology and urinalysis	Sitagliptin 100 mg	34.4	3.2
						Glipizide	29.5	1.7
NCT01137812[78,79]	Interactive Voice Response System/Interactive Web Response System	Double blind	Adequate	SAE	Safety evaluations included AEs, clinical laboratory tests, vital sign measurements, physical examinations, self-monitored blood glucose, 12-lead electrocardiograms, and documentation of hypoglycemic episodes	Sitagliptin 100 mg	44.4	2.1
						Canagliflozin 300 mg	32.6	1.6
NCT00482729[80]	NR	Double blind	NR	SAE	NR	Sitagliptin/Meformin-Fixed Dose Combination	34.7 (217/626)	13.7 (86/626)
						Metformin	34.9 (218/624)	10.6 (66/624)
Bunck et al[81]	NR	Open label	NA	SAE	NR	Exenatide	16.7	0
						Insulin glargine	9.1	3.03
Diamant et al[82]	Computer generated randomization sequence	Open label	NA	SAE	Safety endpoints were adverse events, clinical lab assessments, vital signs, and hypoglycemia. We defined adverse events as those occurring at or after randomization or worsening during the study	Exenatide	10.3	0.86
						Insulin glargine	6.3	0.45
Inagaki et al[83]	Computer generated randomization sequence	Open label	NA	AE	Safety profile end points included AEs and hypoglycemia	Exenatide once weekly	10.2	0.47
						Insulin glargine once daily	5.2	0
Russell-Jones et al[84]	Computer generated randomization sequence	Double blind	Adequate	SAE	Safety end points were adverse events, clinical lab assessments, vital signs, hypoglycemia and antibodies to exenatide. Treatment emergent adverse events were defined as those occurring or worsening after the first dose of study drug	Exenatide 2 mg once weekly + Oral placebo	15.3	1.6
						Sitagliptin 100 mg/d + SC placebo	14.1	2.4
						Metformin starting at 1000 mg/d + SC placebo	13.4	0.4
						Pioglitazone starting at 30 mg/d + SC placebo	1.8	1.8
NCT01003184[85]	NR	Open label	NR	SAE	NR	Exenatide once weekly	17.1	0.9
						Insulin Detemir twice daily	11.4	0
Astrup et al[86]	NR	Double blind (first 20 wk) Weeks 20-104: Open label	NR	SAE	Safety assessments included adverse events, recorded at every visit, standard lab tests and serum liraglutide antibodies. A safety committee for data surveillance was established	Liraglutide 1.2 mg	10.5	0
						Liraglutide 1.8 mg	17.8	0
						Liraglutide 2.4 mg	21.5	0
						Liraglutide 3.0 mg	11.8	0
						Placebo	19.4	0
Garber et al[87]	Telephone based or web-based systems	Double blind	Adequate	SAE	Key safety assessments were tolerability (including nausea and other gastrointestinal adverse events), serum calcitonin and hypoglycemic episodes	Liraglutide 1.2 mg	35.5	NR
						Liraglutide 1.8 mg	29.7	NR
						Glimepiride 8 mg	38.7	NR
Nauck et al[88]	Telephone based or web-based randomization systems	Double blind	Adequate	SAE	Safety variables included adverse events, vital signs, ECG, biochemical and hematology measures and subject reported hypoglycemic episodes	Once daily Liraglutide (0.6 mg)	14.0	0
						Once daily Liraglutide (1.2 mg)	18.0	0.4
						Once daily Liraglutide (1.8 mg)	21.0	0
						Once daily Glimepiride (4 mg)	14.0	0
						Placebo	39.0	0
Marre et al[89]	NR	Double blind	NR	SAE	Safety variables included hypoglycemic episodes, liraglutide antibodies, tolerability (gastrointestinal complaints) and pulse. AEs, vital signs, ECG, biochemical and hematology measures including calcitonin were also monitored	Liraglutide 0.6 mg	10.7	NR
						Liraglutide 1.2 mg	14.0	NR
						Liraglutide 1.8 mg	8.9	NR
						Placebo	27.2	NR
Zinman et al[90]	Telephone based or web-based randomization systems	Double blind	Adequate	SAE	Safety variables included AEs, vital signs, ECG, biochemical and hematology measures and subject reported hypoglycemic episodes	Liraglutide 1.2 mg	14.0	NR
						Liraglutide 1.8 mg	25.0	NR
						Placebo	32.0	NR
Raz et al[91]	NR	Double blind	NR	SAE	Safety assessments included AEs, vital signs, physical examinations, clinical lab tests, ECG and hypoglycemia	Taspoglutide 10 mg	11.2	NR
						Taspoglutide 20 mg	13.2	NR
						Placebo	3.3	NR
Rosenstock et al[92]	NR	Double blind	NR	SAE	Adverse event assessments and safety analyses were conducted throughout the study	Albiglutide 4 mg weekly	48.6	5.7
						Albiglutide 15 mg weekly	31.4	8.6
						Albiglutide 30 mg weekly	32.3	3.2
						Albiglutide 15 mg biweekly	45.5	9.1
						Albiglutide 30 mg biweekly	24.2	0
						Albiglutide 50 mg biweekly	42.9	2.9
						Albiglutide 50 mg monthly	14.3	2.9
						Albiglutide 100 mg monthly	44.1	2.9
						Placebo	23.5	0
Seino et al[93]	Interactive voice response system	Double blind	Adequate	SAE	Safety and tolerability included reported AEs and other safety information such as symptomatic hypoglycemia	Lixisenatide (10 ug for 1 wk, 15 ug for 1 wk, then 20 ug-maintenance dose)	NR	NR
						Placebo	NR	NR
Umpierrez et al[94]	Computer generated random sequence	Double blind	Adequate	SAE	Safety measures included AEs, vital signs, hypoglycemia events and lab tests	LY2189265 (LY 0.5/1.0)	12.1	1.5
						LY2189265 (LY 1.0/1.0)	10.8	1.5
						LY2189265 (LY 1.0/2.0)	13.8	1.5
						Placebo	9.1	1.5

NR: Not reported; AE: Adverse event; SAE: Serious adverse event; ECGs: Electrocardiograms; TEAE: Treatment-emergent adverse events; CEC: Clinical events committee.

Of these 69 studies, data was abstracted from 28 published reports, 32 studies from clinicaltrials.gov and 9 studies were abstracted from pharmaceutical company databases (Boehringer Ingelheim, Novartis and Takeda Pharmaceutical Company). Almost all the trials were multicenter or multinational studies in patients with type II diabetes mellitus.

The majority of studies did not report on the method of generating the random sequence, or on the means of concealing allocation. However, most of the trials (n = 55) were double-blinded, thus reducing risk of bias in the diagnosis of pancreatic adverse events. We found that most of the trials, except for two, did not specific pancreatitis as a part of their safety monitoring protocol. As such, there is a strong possibility that pancreatic adverse events may have been missed or wrongly diagnosed. Moreover, the included studies did not specify whether they applied similar criteria in defining cases of pancreatitis.

Overall

Total pancreatic related adverse events: With a total of 89 pancreatic related adverse events among the 34340 number of patients receiving GLP-1 agents and 74 events among 26380 patients receiving the control agents, there was no statistically significant difference in the risk of pancreatic adverse event associated with GLP-1 agent compared with controls (Peto OR = 0.99, 95%CI: 0.72-1.36, P = 0.96; I² = 0%) (Figure 2).

Figure 2 Risk of pancreatic adverse events in patients treated with glucagon-like peptide-1 based therapies (A and B).

Pancreatitis: There were a total of 71 pancreatitis events in patients on GLP-1 agents and 56 pancreatitis events occurred in the control patients. There was no statistically significant difference in the risk of pancreatic adverse event associated with GLP-1 agent compared with controls (Peto OR = 1.07, 95%CI: 0.75-1.52, P = 0.72, I² = 0%) (Figure 3).

Figure 3 Risk of pancreatitis in patients treated with glucagon-like peptide-1 based therapies (A and B).

Elevated pancreatic enzymes: Eight studies reported on elevated pancreatic enzymes. There was a statistically significant increased risk of elevation of pancreatic enzymes associated with GLP-1 agents compared with control (Peto OR = 3.15, 95%CI: 1.56-6.39, P = 0.001, I² = 0%) (Figure 4).

Figure 4 Risk of elevated pancreatic enzymes for glucagon-like peptide-1 based agents.

Pancreatic cancer: Eighteen studies reported on pancreatic cancer (Table 3). There were a total of 35 cases of pancreatic cancer reported from studies that used GLP-1 agents. Seventeen cases of pancreatic occurred among 18259 patients taking GLP-1 agents compared to 18 cases among 15785 controls. Of these cases, 2 used linagliptin, 2 used alogliptin, 1 used vildagliptin, 7 used saxagliptin while 5 used sitagliptin. The remaining 18 cases occurred among controls.

Table 3 Pancreatic cancer events in randomized controlled trials of glucagon-like peptide-1 agents.

Ref.	Duration of GLP-1 exposure (wk)	Arms	No. of participants	No. of cases
NCT00654381[26]	52	Linagliptin 5 mg	159	0
		Linagliptin 10 mg	160	1
		Voglibose	162	0
		Placebo	80	0
NCT00622284[27]	104	Linagliptin	776	1
		Glimepiride	775	2
BI Trial No: 1218.15/U09-2519-01[28]	24	Linagliptin 5 mg + Pioglitazone 30 mg	130	0
		Pioglitazone 30 mg + Placebo	259	1
White et al[37]	76	Alogliptin	2701	0
		Placebo	2679	0
NCT01318135[38]	52	Alogliptin 12.5 mg qd + Metformin 500 mg bid or 750 mg tid	142	1
		Metformin 500 mg bid or 750 mg tid	145	0
NCT01263496[40]	52	Alogliptin 6.25 mg qd	96	0
		Alogliptin 12.5 mg qd	101	0
		Alogliptin 25 mg qd	97	1
		Alogliptin 50 mg qd	97	0
		Voglibose 0.2 mg tid	83	0
CLAF237A23119[49]	12	Vildagliptin 100 mg + Metformin	1756	1
		Thiazolinedione + Metformin	871	NR
NCT00757588[53]	52	Saxagliptin 5 mg + Insulin	304	1
		Placebo + Insulin	151	0
Scirica et al[54]	109	Saxagliptin	8280	5
		Placebo	8212	12
NCT00316082[56]	24	Saxagliptin 2.5/5 mg QAM	71	1
		Saxagliptin 2.5 mg QAM	74	0
		Saxagliptin 5 mg QAM	74	0
		Saxagliptin 5 mg QPM	72	0
		Placebo	74	0
Chan et al[58,59]	54	Sitagliptin 50 mg or 25 mg once daily	65	1
		Placebo/Glipizide	26	0
Ferreira et al[61,62]	54	Sitagliptin	210	1
		Glipizide	212	0
Henry et al[63,64]	54	Pioglitazone 15 mg	230	0
		Pioglitazone 30 mg	233	0
		Pioglitazone 45 mg	230	0
		Sitagliptin 100 mg/Pioglitazone 15 mg	230	0
		Sitagliptin 100 mg/Pioglitazone 30 mg	231	1
		Sitagliptin 100 mg/Pioglitazone 45 mg	230	0
Raz et al[65,66]	30	Sitagliptin 100 mg	96	0
		Placebo	94	1
Goldstein et al[68,69]	104	Metformin 500 mg bid	182	0
		Metformin 1000 mg bid	182	0
		Sitagliptin 50 mg bid + Metformin 500 mg bid	190	0
		Sitagliptin 50 mg bid + Metformin 1000 mg bid	182	0
		Sitagliptin 50 mg bid + Metformin 1000 mg bid	117	0
		Placebo/Metformin 1000 mg bid	176	1
Arechavaleta et al[70,71]	30	Sitagliptin	516	1
		Glimepiride	518	0
Charbonnel et al[72,73]	104	Sitagliptin 100 mg	464	1
		Placebo/Glipizide 5 mg	237	0
NCT00094757[76]	54	Sitagliptin 100 mg	205	0
		Sitagliptin 200 mg	206	0
		Placebo/Pioglitazone	110	1

GLP-1: Glucagon-like peptide-1.

Individual GLP-1 agents

DPP-4 inhibitors: (1) Linagliptin: Fifteen studies that used Linagliptin had a total of 7263 patients. There was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 1.14, 95%CI: 0.32-4.13) or pancreatitis (Peto OR = 2.90, 95%CI: 0.49-17.36) associated with linagliptin compared with controls; (2) Alogliptin: Nine studies that used Alogliptin had a total of 7914 patients. In comparison with control, there was no increased risk of having a pancreatic adverse event (Peto OR = 1.59, 95%CI: 0.82-3.07) or pancreatitis (Peto OR = 1.50, 95%CI: 0.77-2.94) with alogiptin; (3) Vildagliptin: Seven studies that used Vildagliptin had a total of 7687 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 0.87, 95%CI: 0.26-2.94) or pancreatitis (Peto OR = 0.75, 95%CI: 0.21-2.67) with vildagliptin; (4) Saxagliptin: Seven studies that used Saxagliptin had a total of 19876 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 0.79, 95%CI: 0.49-1.25) or pancreatitis (Peto OR 0.91, 95%CI: 0.53-1.56) with saxaglipitin; and (5) sitagliptin: Sixteen studies that used Sitagliptin had a total of 10360 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 0.66, 95%CI: 0.27-1.63) or pancreatitis (Peto OR = 0.45, 95%CI: 0.14-1.43) with sitagliptin.

GLP-1 receptor agonists

Exenatide: Five studies that used Exenatide had a total of 1690 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 1.53, 95%CI: 0.15-15.29) or pancreatitis (Peto OR = 1.53, 95%CI: 0.15-15.29) with exenatide.

Liraglutide: Six studies that used Liraglutide had a total of 4373 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 1.71, 95%CI: 0.29-10.04) or pancreatitis (Peto OR = 1.71, 95%CI: 0.29-10.04 with liraglutide.

Dulaglutide: One study that used Dulaglutide had 262 patients. In comparison with control, there was no statistically significant difference in the risk of pancreatic adverse event (Peto OR = 3.83, 95%CI: 0.16-93.74) or pancreatitis (Peto OR = 3.83, 95%CI: 0.16-93.74) with dulaglutide.

Taspoglutide, albiglutide and lixisenatide: Taspoglutide, Albiglutide and Lixisenatide all had 1 study each with 368, 321 and 311 patients each. The effect estimates were not estimable due to the small number of events.

In a post-hoc analysis, we examined whether there was any difference between DPP-4 inhibitors and GLP-1 based therapies. The results showed that neither the DPP-4 inhibitors nor the GLP- 1 based therapies were associated with a risk of pancreatic complications (Figure 5).

Figure 5 Risk of pancreatic events for glucagon-like peptide-1 receptor agonist drugs only.

Publication bias

We did not detect any publication bias in the funnel plot (Figure 6).

Figure 6 Funnel plot for risk of pancreatic adverse events.

DISCUSSION

Summary of results

Our study showed a significantly increased risk of pancreatic enzyme elevation with GLP-1 based therapies. However, the use of GLP-1 based therapies was not associated with a statistically significant increased risk of pancreatic complication events in patients with type 2 diabetes in randomized controlled trials. Additionally, when we examined individual agents, none of the DPP-4 inhibitors or GLP-1 agonists was associated with a statistically significant increased risk of pancreatitis (Figure 3). Despite the lack of statistical significance the upper bounds of the CI in several analyses, particularly for the GLP-1 receptor agonists (exenatide, liraglutide and albiglutide) exceeded 1 and could not rule out a clinically significant hazard. There were an insufficient number of cases of pancreatic cancer to allow for the estimation of meaningful differences between GLP-1 based agents and controls.

Explanations

These discordant results-no significant effect on the outcome of acute pancreatitis but significant increase in the risk of pancreatic enzyme elevation associated with GLP-1 based therapies in a small number of studies may have two alternative explanations.

These could indicate that injury with GLP-1 based therapies is sub-threshold and result in pancreatic inflammation that may not reach the level of acute pancreatitis. Alternatively, the ascertainment of pancreatic adverse events/complications may have been more complete in this subset of studies showing an elevation in pancreatic enzymes. It was not clear whether pancreatitis adverse events were rigorously defined or captured in an objective rather than subjective manner across the trials, potentially biasing towards the null due to misclassification. In contrast, measurement of elevated pancreatic enzymes is a more objective measure, serial enzyme measurements should be regularly checked in trial participants on GLP-1 agents who present with gastrointestinal symptoms. Lack of awareness for the need to assess pancreatic enzymes could lead to under-ascertainment of pancreatic adverse events in patients presenting with upper abdominal symptoms. Among patients with type 2 diabetes, one previous study reported an increase in enzyme associated with DPP-4 inhibitors compared to controls (36% vs 18%), suggesting that this adverse reaction deserves further investigation[15].

Our meta-analysis should be seen in the light of other recent studies. A recent review reported a slightly increased trend for reporting of acute pancreatitis associated with GLP-1 receptor agonists but not with DPP-IV inhibitors[16]. Two other systematic reviews reported no increased risk of acute pancreatitis, but with very wide confidence intervals that could not rule out a significant increase[6,17]. However, one such meta-analysis included observational studies, which may be prone to confounding[17]. The difference in meta-analysis should reflect differences in inclusion of trials and ascertainment of events. Importantly none of the previous meta-analysis have reported on elevations in pancreatic enyzmes associated with GLP-1 based therapies. However, the CIs were wide in all meta-analyses and we could not rule out a significant increase in the risk of pancreatitis with GLP-1 based therapies. The lack of statistical significance may reflect incomplete ascertainment of pancreatic adverse events in clinical trials of GLP-1 based therapies or inadequate statistical power to detect rare but serious complication such as pancreatitis. Observational studies have also shown inconsistent results between GLP-1 based therapies and acute pancreatitis due to incomplete ascertainment of covariates, or poor performance of the diagnostic codes for acute pancreatitis[5,18-20]. It is also unclear whether the inflammatory process from recurrent or chronic pancreatitis is a predisposing factor to subsequent development of pancreatic cancer.

Limitations

Our study has some limitations. We limited our analysis to published RCTs. However; there may be unpublished studies that report on this outcome. We did not have access to data to conduct individual patient data meta-analysis and ascertain time to the occurrence of pancreatic enzyme elevations. Importantly, clinical trials may not have ascertained the occurrence of pancreatitis on participants who withdrew from the trial (as a result of the complication). This may bias our estimates towards the null. The availability of sponsors of individual patient data to independent investigators may allow for further analyses.

Our meta-analysis shows a three-fold increased risk of pancreatic enzyme elevation with GLP-1 based agents compared to controls, without an a significant increased risk of pancreatitis or pancreatic cancer due to small number of cases. Future adequately powered observational studies with well validated codes for pancreatitis and pancreatic cancer and careful control of confounding are needed to evaluate the risk of pancreatic enzyme elevation, pancreatitis and pancreatic cancer with GLP-1 based therapies.

COMMENTS

Background

Recent developments have led to an increasingly wide range of glucose lowering drugs being trialed for treatment of type II diabetes mellitus. However, a variety of concerns have been raised regarding the safety of these new agents for long-term chronic use. This has led to tightening of the regulatory landscape and closer scrutiny of data regarding serious rare adverse events.

Research frontiers

Many trials have been conducted to demonstrate the efficacy of glucagon-like-peptide-1 (GLP-1) agonists and dipeptidyl peptidase-4 (DPP-4) inhibitors in reducing blood glucose levels. However, there have been suggestions of a potential increase in risk of pancreatic adverse events with these drugs due to a postulated proliferative effect on pancreatic cells. The existing evidence base is conflicting, and difficult to interpret due to the very low incidence of pancreatic adverse events.

Innovations and breakthroughs

The findings of this meta-analysis are that risks of pancreatitis or pancreatic cancer have not been definitively established with any of the GLP-1 agonists or DPP-4 inhibitors. However, there is a signal suggesting increased risk of elevated pancreatic enzymes, which has not previously been described in other systematic reviews.

Applications

GLP-1 agonists or DPP-4 inhibitors may have some relationship with elevations in the pancreatic enzyme levels. Further large scale studies are needed to determine if these elevations may or may not be associated with adverse clinical outcomes.

Terminology

GLP-1 belongs to the incretin group of hormones which act to stimulate insulin secretion dependent on glucose levels. GLP-1 receptor agonists are drugs developed as incretin-mimetics. DPP-4 is an enzyme that breaks down GLP-1, thus causing GLP-1 to have a short half-life. Drugs that inhibit DPP-4 would be expected to increase the availability of endogenous GLP-1.

Peer-review

This manuscript has a great collecting data about this topic.