Glucagon-like peptide-1 (GLP-1) receptor agonists are a class of hypodermic hypoglycemic drugs that are effective for the treatment of type 2 diabetes mellitus. A number of GLP-1 receptor agonists have already been approved by the Food and Drug Administration (FDA) for type 2 diabetes treatment, including exenatide, liraglutide, semaglutide, lixisenatide, dulaglutide, albiglutide, and others. Based on the findings of several cardiovascular outcome trials (CVOTs) [
However, some studies have indicated that GLP-1 receptor agonists may be associated with an increased risk of malignant neoplasia. In animal models, GLP-1 receptor agonist treatment was linked to an increased risk of pancreatic cancer and thyroid C-cell cancer [
This systematic review was registered in PROSPERO (registration number
In this systematic review and meta-analysis, we regarded studies as eligible for the inclusion criteria if they were randomized controlled trials (RCTs) that included adult patients with type 2 diabetes, compared a GLP-1 receptor agonist to another treatment strategy with a minimum treatment duration of 24 weeks, and reported the number of participants who developed neoplasms during follow-up. We retained all potentially eligible studies for review, independent of the primary outcome of each study. We searched the MEDLINE, Embase, and Cochrane databases for eligible trials, with a language restriction of English. The search strategy was based on “subject terms+free terms.” Subject terms used in the searches were “glucagon-like peptide-1 receptor agonist,” “exenatide,” “liraglutide,” “semaglutide,” “lixisenatide,” “dulaglutide,” “albiglutide,” “neoplasms,” and “diabetes mellitus.” With regard to neoplasms, we included studies of all types of malignant tumors and excluded those evaluating benign tumor formation. For the search for RCTs, we used available filters to search only for RCTs from the Harvard Library. Searches were done through October 1, 2018.
Two independent investigators reviewed study titles and abstracts, and studies that satisfied the inclusion criteria were assessed by screening of the full text. Trials selected for detailed analysis and data extraction were analyzed by two investigators, and disagreements were resolved by a third investigator. For quality assessment, Cochrane Collaboration’s Tool for Assessing Risk of Bias in RCTs was used.
When specific data were not available, requests for the information were sent to the corresponding authors of the trial articles. We calculated ORs and 95% confidence intervals (CIs) for the numbers of neoplasia events by the treatment group. We used a fixed effects model meta-analysis if the between-trial statistical heterogeneity was low. We used a random effects model meta-analysis if the between-trial statistical heterogeneity was high. The possibility of publication bias was assessed by constructing a funnel plot of each trial’s effect size against the standard error. The heterogeneity of treatment effects between trials was assessed by the
Our database searches identified 209 studies, of which 34 (presenting data for 50452 participants) were included in our analysis (Figure
Study selection.
Trial design features and results.
Author/year | Trial/program | Interventions | Trial duration (weeks) | Experimental | Control | |||
---|---|---|---|---|---|---|---|---|
Experimental | Control | Number of events | Number of events | |||||
Gallwitz et al. 2012 [ |
EUREXA | Exenatide | Glimepiride | 102 | 3 | 515 | 1 | 514 |
Russel-Jones et al. 2012 [ |
DURATION-4 | Exenatide | Metformin | 26 | 0 | 248 | 1 | 246 |
Kadowaki et al. 2010 [ | Exenatide | Placebo | 24 | 1 | 144 | 0 | 35 | |
Xu et al. 2014 [ |
CONFIDENCE | Exenatide | Premixed insulin | 48 | 1 | 110 | 0 | 114 |
Jaiswal et al. 2015 [ | Exenatide | Glargine | 77 | 1 | 22 | 0 | 24 | |
Diamant et al. 2014 [ |
DURATION-3 | Exenatide | Glargine | 156 | 3 | 140 | 2 | 147 |
Diamant et al. 2012 [ | Exenatide | Glargine | 84 | 2 | 173 | 1 | 173 | |
Bergenstal et al. 2010 [ |
DURATION-2 | Exenatide | Sitagliptin | 26 | 0 | 160 | 1 | 166 |
Holman et al. 2017 [ |
EXSCEL | Exenatide | Placebo | 167 | 355 | 7356 | 361 | 7396 |
Gadde et al. 2017 [ |
DURATION-NEO-2 | Exenatide | Placebo | 28 | 0 | 181 | 0 | 61 |
Weinstock et al. 2015 [ |
AWARD-5 | Dulaglutide | Sitagliptin | 26 | 8 | 606 | 5 | 315 |
Araki et al. 2015 [ | Dulaglutide | Glargine | 26 | 0 | 181 | 0 | 180 | |
Blonde et al. 2015 [ |
AWARD-4 | Dulaglutide | Glargine | 52 | 0 | 588 | 0 | 296 |
Umpierrez et al. 2014 [ |
AWARD-3 | Dulaglutide | Metformin | 52 | 0 | 539 | 0 | 268 |
Pozzilli et al. 2017 [ |
AWARD-9 | Dulaglutide | Glargine | 28 | 0 | 150 | 1 | 150 |
Miyagawa et al. 2015 [ |
Dulaglutide | Placebo | 52 | 0 | 281 | 0 | 70 | |
Liraglutide | 1 | 141 | ||||||
Bailey et al. 2016 [ |
LIRA-SWITCH | Liraglutide | Sitagliptin | 26 | 0 | 202 | 2 | 204 |
Zang et al. 2016 [ | Liraglutide | Sitagliptin | 26 | 0 | 183 | 2 | 184 | |
Marso et al. 2016 (Liraglutide) [ |
LEADER | Liraglutide | Placebo | 198 | 296 | 4668 | 279 | 4672 |
Pratley et al. 2011 [ | Liraglutide | Sitagliptin | 52 | 4 | 446 | 1 | 219 | |
le et al. 2017 [ |
SCALE | Liraglutide | Placebo | 160 | 11 | 1501 | 1 | 747 |
Marso et al. 2016 (Semaglutide) [ |
SUSTAIN-6 | Semaglutide | Placebo | 104 | 1 | 1648 | 4 | 1649 |
Ahrén et al. 2017 [ |
SUSTAIN 2 | Semaglutide | Sitagliptin | 52 | 2 | 818 | 2 | 407 |
Sorli et al. 2017 [ |
SUSTAIN 1 | Semaglutide | Placebo | 30 | 4 | 258 | 0 | 129 |
Davies et al. 2017 [ | Semaglutide | Placebo | 26 | 1 | 419 | 0 | 71 | |
Aroda et al. 2017 [ |
SUSTAIN 4 | Semaglutide | Glargine | 30 | 4 | 722 | 1 | 360 |
Reusch et al. 2014 [ |
HARMONY 1 | Albiglutide | Placebo | 52 | 0 | 150 | 0 | 151 |
Home et al. 2015 [ |
HARMONY 5 | Albiglutide | Placebo | 156 | 0 | 271 | 1 | 115 |
Weissman et al. 2014 [ |
HARMONY 4 | Albiglutide | Glargine | 52 | 0 | 504 | 0 | 241 |
Nauck et al. 2015 [ |
HARMONY 2 | Albiglutide | Placebo | 52 | 2 | 204 | 0 | 105 |
Leiter et al. 2014 [ | Albiglutide | Sitagliptin | 52 | 0 | 249 | 0 | 246 | |
Leiter et al. 2017 [ | Albiglutide | Lispro | 52 | 1 | 285 | 0 | 281 | |
Yu et al. 2014 [ |
GetGoal-M-Asia | Lixisenatide | Placebo | 24 | 0 | 196 | 0 | 194 |
Pfeffer et al. 2015 [ |
ELIXA | Lixisenatide | Placebo | 108 | 72 | 3031 | 61 | 3032 |
Fourteen trials compared treatment outcomes achieved with GLP-1 receptor agonists versus placebo with or without oral antidiabetic drugs. The other 20 included trials compared outcomes achieved with GLP-1 receptor agonists to those obtained with metformin (2 trials), sitagliptin (7 trials), glimepiride (1 trial), premixed insulin (1 trial), glargine (8 trials), or lispro (1 trial). Many types of neoplasms occurred in these trials, including pancreatic cancer, thyroid cancer, breast cancer, gastrointestinal cancer, skin cancer, leukemia, lymphoma, prostate cancer, and others. For trials belonging to the same program, such as the DURATION-2, DURATION-3, and DURATION-4 studies, detailed assessment was performed to exclude duplicate data.
Compared with placebo or other interventions, GLP-1 receptor agonist use showed no association with an increased risk of neoplasm development (OR 1.04, 95% CI 0.94–1.15;
Meta-analysis of the incidence of neoplasms with the use of all tested GLP-1 receptor agonists versus placebo or other antidiabetic treatments.
Funnel plot for the comparison of the incidence of neoplasia with the use of GLP-1 receptor agonists versus placebo or other antidiabetic treatments.
Among all 34 included trials, 6 trials (with data for 13237 patients) employed liraglutide as the experimental agent. Compared with placebo or other interventions, liraglutide use was not associated with an increased incidence of neoplasms (OR 1.08, 95% CI 0.91–1.27;
Meta-analysis of the incidence of neoplasms with the use of specific GLP-1 receptor agonists versus placebo or other antidiabetic treatments.
Among all 34 included trials, 10 trials (with data for 17925 patients) employed exenatide as the experimental agent. Compared with placebo or other interventions, exenatide use was not associated with an increased incidence of neoplasia (OR 1.00, 95% CI 0.86–1.16;
Among the 34 included trials, 5 trials (with data for 6481 patients) employed semaglutide as the experimental agent. Compared with placebo or other interventions, semaglutide use was not associated with an increase in neoplasm formation (OR 0.89, 95% CI 0.35–2.22;
Among the 34 included trials, 6 trials (with data for 2802 patients) employed albiglutide as the experimental agent. Compared with placebo or other interventions, albiglutide use was not associated with an increased incidence of neoplasia (OR 1.07, 95% CI 0.23–4.88;
Among the 34 included trials, 14 trials (with data for 38876 patients) chose placebo as the only control treatment. Compared with placebo only, GLP-1 receptor agonist use was not associated with an increased incidence of neoplasia (OR 1.04, 95% CI 0.94–1.16;
Meta-analysis of the incidence of neoplasia with placebo versus GLP-1 receptor agonists used.
Among the 34 included trials, 5 trials (with data for 1309 patients) had a study duration of at least 3 years. The subanalysis of only these 5 trials showed that, compared with placebo or other antidiabetic treatments, GLP-1 receptor agonist use was not associated with an increased incidence of neoplasia (OR 1.03, 95% CI 0.92–1.15;
Meta-analysis of the incidence of neoplasia with the use of a GLP-1 receptor agonist versus placebo or other antidiabetic treatments, based on studies with a minimum duration of 3 years.
Previous studies have reported that GLP-1 receptor agonist use correlated with an increased risk of pancreatic cancer [
GLP-1 receptor agonists promote cell proliferation and survival by activating signaling pathways in human islet cells, such as those involving phosphate idylinositol 3 kinase (PI3K) and extracellular regulated kinases 1 and 2 (ERK1/2), which are also frequently activated in human colon cancer cells. ERK1 and ERK2 act on transcription factors such as E1k-1, c-myc, c-fos, c-jun, activating transcription factor (ATF), nuclear factor- (NF-) kB, and activator protein- (AP-) 1, to promote the expression of genes closely related to cell proliferation and differentiation [
Research about dipeptidyl peptidase-4 (DPP-4) inhibitors, which function via a similar mechanism as GLP-1 receptor agonists, failed to verify an association between the use of these drugs and an increased risk of site-specific cancer, and this was attributed to the small number of studies for each cancer type and their relatively short duration [
The main strength of this review is that all of the included studies were RCTs. Among all included trials, two RCTs contributed considerable weight to the pooled analysis [
We excluded studies with an intervention duration less than 24 weeks to prevent detection bias or even reverse causality. According to the incidence rates of malignant tumors in humans [
A limitation of the present meta-analysis was the lack of trials with long-term duration, given that neoplasm formation may occur over an extended period. Additionally, patients included in RCTs are generally healthier than real-world patients and are therefore less likely to develop neoplasms than the general patient population or patients in observational studies of new drugs. Moreover, as the occurrence of neoplasia was not the primary or secondary outcome in these RCTs, reporting bias is possible. Although most of the included studies were published in high-impact journals, potential risks of bias such as an open-label design and funding from pharmaceutical companies are still possible, as outlined in the supplementary tables (supplementary material (available
GLP-1 receptor agonists can be used without safety concerns related to the risk of malignant neoplasia in patients with type 2 diabetes.
All authors have no conflicts of interest.
All authors read the manuscript, contributed to its revision, and approved the final submitted version. Yufang Liu and Xiaomei Zhang contributed equally to this work.
The risk of bias graph review of authors’ judgements about each risk of bias item presented as percentages across all included studies.