Dipeptidyl peptidase-4 inhibitors and gallbladder or biliary disease in type 2 diabetes: systematic review and pairwise and network meta-analysis of randomised controlled trials.

OBJECTIVE: To examine the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases.
DESIGN: Systematic review and pairwise and network meta-analysis. DATA SOURCES: PubMed, EMBASE, Web of Science, and CENTRAL from inception until 31 July 2021. ELIGIBILITY CRITERIA: Randomised controlled trials of adult patients with type 2 diabetes who received dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, and sodium-glucose cotransporter-2 inhibitors compared with placebo or other antidiabetes drugs. MAIN OUTCOME MEASURES: Composite of gallbladder or biliary diseases, cholecystitis, cholelithiasis, and biliary diseases. DATA EXTRACTION AND DATA SYNTHESIS: Two reviewers independently extracted the data and assessed the quality of the studies. The quality of the evidence for each outcome was assessed using the Grading of Recommendations, Assessment, Development and Evaluations framework (GRADE) approach. The meta-analysis used pooled odds ratios and 95% confidence intervals.
RESULTS: A total of 82 randomised controlled trials with 104 833 participants were included in the pairwise meta-analysis. Compared with placebo or non-incretin drugs, dipeptidyl peptidase-4 inhibitors were significantly associated with an increased risk of the composite of gallbladder or biliary diseases (odds ratio 1.22 (95%confidence interval 1.04 to 1.43); risk difference 11 (2 to 21) more events per 10 000 person years) and cholecystitis (odds ratio 1.43 (1.14 to 1.79); risk difference 15 (5 to 27) more events per 10 000 person years) but not with the risk of cholelithiasis and biliary diseases. The associations tended to be observed in patients with a longer duration of dipeptidyl peptidase-4 inhibitor treatment. In the network meta-analysis of 184 trials, dipeptidyl peptidase-4 inhibitors increased the risk of the composite of gallbladder or biliary diseases and cholecystitis compared with sodium-glucose cotransporter-2 inhibitors but not compared with glucagon-like peptide-1 receptor agonists.
CONCLUSIONS: Dipeptidyl peptidase-4 inhibitors increased the risk of cholecystitis in randomised controlled trials, especially with a longer treatment duration, which requires more attention from physicians in clinical practice. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42021271647. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Introduction

Dipeptidyl peptidase-4 inhibitors are widely used in the treatment of type 2 diabetes.1 2 3 Given the large population of patients with type 2 diabetes receiving dipeptidyl peptidase-4 inhibitors worldwide,2 3 any safety concerns deserve attention.

Dipeptidyl peptidase-4 inhibitors enhance the bioavailability of endogenous glucagon-like peptide-1 and glucose dependent insulinotropic polypeptide,2 both of which might affect postprandial gallbladder motility,4 5 raising concerns about the risks of gallbladder or biliary diseases with dipeptidyl peptidase-4 inhibitors. Early studies have proposed associations between gallbladder related events and incretin based drugs, including dipeptidyl peptidase-4 inhibitors and glucagon-like peptide-1 receptor agonists.4 6 7 However, the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases remains unclear. Therefore, we aimed to do a systematic review and meta-analysis of randomised controlled trials to evaluate the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases in patients with type 2 diabetes.

Methods

The systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Statement for traditional pairwise meta-analysis and the PRISMA extension for network meta-analysis.8 9 The study protocol was prospectively registered (PROSPERO CRD 42021271647).

Study sources, searches, and identification

We searched PubMed, the Cochrane Library, EMBASE, and Web of Science for randomised controlled trials of dipeptidyl peptidase-4 inhibitors up to 31 July 2021. Three reviewers (LH, HZ, and JW) independently did literature searches with designed search strategies (supplementary table A). We also screening the reference lists of relevant systematic reviews and manually searched for grey literature in Google Scholar and public repositories.10 11 Four reviewers (LH, HZ, FP, and NY) independently identified the eligible studies according to the pre-formulated inclusion and exclusion criteria (supplementary table B). We included parallel group randomised controlled trials in adults with type 2 diabetes in the traditional pairwise meta-analysis if they compared dipeptidyl peptidase-4 inhibitors with placebo or non-incretin drugs, and we included them in the network meta-analysis if they compared dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, or sodium-glucose cotransporter-2 inhibitors with one another or with other antidiabetes drugs, placebo, or standard care. Disagreements or conflicts were resolved by discussion with other team members.

Data extraction and quality assessment

We extracted data from published articles and supplements, as well as public repositories. We assessed the risk of bias of eligible studies with the revised Cochrane risk of bias tool for randomised trials.12 We evaluated the certainty of the evidence from traditional pairwise meta-analysis and network meta-analysis by using the Grading of Recommendations, Assessment, Development and Evaluations framework (GRADE) for each outcome.13 14 15 Details about data extraction and quality assessment are in the supplementary methods.

Definition of outcomes

The outcomes were the composite of gallbladder or biliary diseases, cholecystitis, cholelithiasis, and biliary diseases. We identified the gallbladder or biliary diseases on the basis of the classifications in the Medical Dictionary for Regulatory Activities (MedDRA) version 22.0. The composite of gallbladder or biliary diseases incorporated cholelithiasis, cholecystitis, other gallbladder disorder, and biliary diseases. Cholecystitis, cholelithiasis, and biliary diseases were captured with the preferred terms, and we also included the lowest level terms that fell into the preferred terms listed above. Biliary diseases (preferred terms) included bile duct stone, bile duct obstruction, bile duct stenosis, biliary colic, biliary fistula, biliary cyst, and cholangitis.

Data synthesis and analysis

We did pairwise meta-analyses for dipeptidyl peptidase-4 inhibitors versus non-incretin based drugs or placebo. We assessed substantial heterogeneity by using χ2 tests (statistical heterogeneity: P<0.10) and τ2 (low: <0.04; low-moderate: 0.04-0.16; moderate-high: 0.16-0.36; high: >0.36).16 17 We calculated odds ratios and 95% confidence intervals by using fixed effect models with the Mantel-Haenszel method, and applied a correction proportional to the reciprocal of the size of the contrasting study arm to handle the zero events.18 19 20 We evaluated absolute risk differences on the basis of the calculated odds ratio and the mean event rate across the control groups for outcomes.18 21 We converted event rates into annual incidences for each outcome.

We assessed the effects of duration of treatment (≥26 weeks or <26 weeks), specific dipeptidyl peptidase-4 inhibitors, and types of control (pure or active controls) on gallbladder or biliary diseases. We evaluated publication bias by visual assessment of asymmetry of the funnel plots and the Egger's asymmetry test.22 23 We did sensitivity analyses by using different pooling methods,24 using random effect models, including double zero studies,25 and omitting each study.

We did network meta-analyses of randomised controlled trials with the frequentist method,26 comparing dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, and sodium-glucose cotransporter-2 inhibitors. We used fixed effect models to calculate the overall odds ratios and 95% confidence intervals and to evaluate the heterogeneities between studies for sparse network meta-analysis.27 We used forest plots and league tables of the effects to visualise comparisons of network estimations. We evaluated global inconsistency across different designs of treatment comparisons in the network by a design-by-treatment model with a generalised Q test.28 29 We used node splitting analysis to evaluate local inconsistency between direct and indirect results within each treatment comparison.30 31 32 We used comparison adjusted funnel plots to examine small study effects.33 We did sensitivity analysis by using random effects models to pool the results.

We used the meta, netmeta package in R (version 4.0.2) and Stata 15 for all analyses. Statistical significance was set at two tailed P<0.05.

Patient and public involvement

Patients and the public were not involved in the planning, design, and implementation of the study, as this study used secondary data. No patients were asked to advise on interpretation or writing up of the manuscript.

Results

Study selection and characteristics of studies included

We included 82 randomised controlled trials (fig 1) with 104 833 participants with type 2 diabetes in the traditional pairwise meta-analysis.34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 Characteristics of studies and participants included in the pairwise meta-analysis are shown in table 1 and supplementary table C. The average age of participants was 59.4 years, and 39.7% (n=41 618) were female. The mean body mass index was 29.7, and mean haemoglobin A1c was 8.1%. Almost all trials documented gallbladder or biliary diseases as serious adverse events.

Fig 1

PRISMA flow diagram. DPP-4=dipeptidyl peptidase-4; GDM=gestational diabetes mellitus; GLP-1=glucagon-like peptide-1; RCT=randomised controlled clinical trial; T1DM=type 1 diabetes mellitus

Table 1

Characteristics of studies and participants in pairwise meta-analysis

Trials	No of patients	Treatment duration	Mean age, years	No (%) female	Mean BMI	Mean HbA1c, %	DPP-4 inhibitors treatment	Comparators
Alogliptin
Nauck et al (2009)	524	26 weeks	55.0	262 (50)	32.0	7.9	Alogliptin 12.5/25 mg	Placebo
Pratley et al (2009)	500	26 weeks	56.7	239 (48)	30.1	NR	Alogliptin 12.5/25 mg	Placebo
Rosenstock et al (2009)	390	26 weeks	55.3	229 (59)	32.5	9.3	Alogliptin 12.5/25 mg	Placebo
Rosenstock et al (2010)	654	26 weeks	53.0	334 (51)	31.0	8.8	Alogliptin 12.5/25 mg	Placebo
DeFronzo et al (2012)	1553	26 weeks	54.4	857 (55)	31.2	8.5	Alogliptin 12.5/25 mg	Placebo; pioglitazone
EXAMINE (2013)	5380	18 months	61.0	1729 (32)	28.7	8.0	Alogliptin 25 mg	Placebo
Rosenstock et al (2013)	441	52 weeks	70.0	243 (55)	29.8	7.5	Alogliptin 25 mg	Glipizide
Pratley et al (2014)	768	26 weeks	53.1	376 (49)	30.4	NR	Alogliptin 12.5 mg	Placebo; metformin
ENDURE (2016)	2620	104 weeks	55.4	1327 (51)	31.2	7.6	Alogliptin 12.5/25 mg	Glipizide
SPEAD-A (2016)	322	12 months	64.6	134 (42)	24.7	7.3	Alogliptin 25 mg	Conventional treatment
Pan et al (2017)	505	16 weeks	52.3	206 (41)	25.8	8.0	Alogliptin 25 mg	Placebo; metformin; pioglitazone
Linagliptin
Gomis et al (2011)	389	24 weeks	57.4	152 (39)	29.2	8.6	Linagliptin 5 mg	Placebo
Owens et al (2011)	1055	24 weeks	57.9	557 (53)	28.3	8.1	Linagliptin 5 mg	Placebo
Gallwitz et al (2012)	1551	2 years	59.8	618 (40)	30.2	7.7	Linagliptin 5 mg	Glimepiride
Lewin et al (2012)	245	18 weeks	54.1	118 (48)	28.3	8.6	Linagliptin 5 mg	Placebo
Kawamori et al (2012)	481	52 weeks	60.1	143 (30)	25.1	8.1	Linagliptin 5/10 mg	Voglibose; placebo
Haak et al (2013)	566	54 weeks	55.6	256 (45)	29.0	7.6	Linagliptin 2.5 mg	Metformin
Yki-Järvinen et al (2013)	1261	52 weeks	60.0	603 (48)	31.0	8.3	Linagliptin 5 mg	Placebo
McGill et al (2014)	133	52 weeks	64.4	53 (40)	32.0	8.2	Linagliptin 5 mg	Placebo
Laakso et al (2015)	235	52 weeks	66.6	86 (37)	35.6	8.1	Linagliptin 5mg	Placebo; glimepiride
Ji et al (2015)	689	14 weeks	53.0	362 (53)	29.0	8.0	Linagliptin 5mg	Metformin
DeFronzo et al (2015)	818	52 weeks	55.8	380 (46)	31.1	8.0	Linagliptin 5mg	Empagliflozin
Wang et al (2016)	305	24 weeks	55.8	153 (50)	25.8	8.0	Linagliptin 5 mg	Placebo
CARMELINA (2018)	6979	2.2 years	65.8	2589 (37)	31.4	7.9	Linagliptin 5 mg	Placebo
Ledesma et al (2019)	302	24 weeks	72.4	119 (39)	28.1	8.1	Linagliptin 5 mg	Placebo
CAROLINA (2019)	6033	6.3 years	64.0	2414 (40)	30.1	7.2	Linagliptin 5 mg	Placebo
Yang et al (2020)	206	24 weeks	58.6	100 (49)	25.5	8.5	Linagliptin 5 mg	Placebo
TRUST2 (2021)	245	156 weeks	64.8	96 (39)	24.4	7.8	Linagliptin 5 mg	Placebo
Omarigliptin
Lee et al (2017)	306	24 weeks	57.8	160 (52)	31.2	8.5	Omarigliptin 25 mg	Placebo
Handelsman et al (2017)	750	54 weeks	58	336 (45)	31.5	7.5	Omarigliptin 25 mg	Glimepiride
MK-3102-018 (2017)	4192	96 weeks	63.7	1254 (30)	31.3	8.0	Omarigliptin 25 mg	Placebo
Shankar et al (2017)	402	104 weeks	57.1	199 (50)	32.5	8.1	Omarigliptin 25 mg	Placebo; glimepiride
Gantz et al (2017)	412	52 weeks	60.0	138 (34)	25.3	8.0	Omarigliptin 25 mg	Placebo
MK-3102-006 (2018)	485	78 weeks	55.5	114 (24)	29.6	8.1	Omarigliptin 25 mg	Placebo
Saxagliptin
Chacra et al (2009)	768	24 weeks	55.1	421 (55)	29.0	8.4	Saxagliptin 2.5/5 mg	Placebo
Jadzinsky et al (2009)	1306	24 weeks	52.0	663 (51)	30.1	9.5	Saxagliptin 5/10 mg	Metformin
Hollander et al (2009)	565	24 weeks	54.0	285 (50)	30.0	8.3	Saxagliptin 2.5/5 mg	Placebo
DeFronzo et al (2009)	743	24 weeks	54.6	366 (49)	31.4	8.1	Saxagliptin 2.5/5/10mg	Placebo
Yang et al (2011)	570	24 weeks	54.2	295 (52)	26.2	7.9	Saxagliptin 5 mg	Placebo
Frederich et al (2012)	365	76 weeks	54.9	197 (54)	30.5	8.0	Saxagliptin 2.5/5 mg	Placebo
PROMPT (2012)	286	24 weeks	58.7	122 (43)	31.7	7.8	Saxagliptin 5 mg	Metformin
SAVOR-TIMI 53 (2013)	16 492	2.1 years	65.1	5455 (33)	31.1	8.0	Saxagliptin 5 mg	Placebo
Barnett et al (2013)	455	52 weeks	57.2	267 (59)	32.7	8.7	Saxagliptin 5 mg	Placebo
Rosenstock et al (2013)	401	206 weeks	53.6	197 (49)	31.7	7.9	Saxagliptin 2.5/5/10mg	Placebo
Goke et al (2013)	858	52 weeks	57.5	414 (48)	31.4	7.7	Saxagliptin 5 mg	Glipizide
GENERATION (2015)	718	52 weeks	72.6	273 (38)	29.6	7.6	Saxagliptin 5 mg	Glimepiride
Matthaei et al (2015)	315	52 weeks	54.6	166 (53)	31.4	7.8	Saxagliptin 5 mg	Placebo
SMART (2016)	481	24 weeks	55.6	196 (41)	26.4	8.2	Saxagliptin 5 mg	Acarbose
START (2017)	639	24 weeks	50.1	211 (33)	26.6	9.4	Saxagliptin 5 mg	Placebo; metformin
SUPER (2017)	462	24 weeks	59.1	253 (55)	26.2	8.5	Saxagliptin 5 mg	Placebo
SPECIFY (2018)	374	48 weeks	53.5	145 (39)	25.5	8.0	Saxagliptin 5 mg	Glimepiride
DapaZu (2018)	939	52 weeks	58.4	339 (36)	32.9	8.3	Saxagliptin 5 mg	Dapagliflozin; glimepiride
DELIGHT (2019)	445	24 weeks	64.7	131 (29)	30.3	8.4	Saxagliptin 5 mg	Placebo; dapagliflozin
Rosenstock et al (2019)	588	24 weeks	56.7	283 (48)	31.9	8.2	Saxagliptin 5mg	Dapagliflozin
Sitagliptin
Rosenstock et al (2006)	353	24 weeks	56.2	157 (44)	31.5	8.1	Sitagliptin 100 mg	Placebo
Charbonnel et al (2006)	701	104 weeks	54.6	301 (43)	30.8	8.0	Sitagliptin 100 mg	Placebo
Nauck et al (2006)	1172	52 weeks	56.7	478 (41)	31.2	7.7	Sitagliptin 100 mg	Glipizide
MK0431-023 (2006)	521	54 weeks	55.1	238 (46)	32.1	8.1	Sitagliptin 100/200 mg	Placebo
Goldstein et al (2007)	1091	24 weeks	53.5	552 (51)	32.1	8.8	Sitagliptin 50/100 mg	Metformin; placebo
Hanefeld et al (2008)	555	12 weeks	55.5	267 (48)	31.8	7.7	Sitagliptin 25/50/100mg	Placebo
Mohan et al (2009)	530	18 weeks	50.9	224 (42)	25.0	8.7	Sitagliptin 100 mg	Placebo
Vilsboll et al (2010)	641	24 weeks	57.7	315 (49)	31.0	8.7	Sitagliptin 100 mg	Placebo
Barzilai et al (2011)	206	24 weeks	71.8	109 (53)	30.9	7.8	Sitagliptin 100 mg	Placebo
Ferreira et al (2013)	422	54 weeks	64.6	170 (40)	26.5	7.8	Sitagliptin 100 mg	Glipizide
CANTATA-D (2013)	1101	52 weeks	55.5	582 (53)	32.2	7.9	Sitagliptin 100 mg	Canagliflozin
CANTATA-D2 (2013)	755	52 weeks	56.7	333 (44)	31.5	8.1	Sitagliptin 100 mg	Canagliflozin
Dobs et al (2013)	262	54 weeks	55.2	111 (42)	30.4	8.8	Sitagliptin 100 mg	Placebo
Henry et al (2013)	1615	54 weeks	50.6	703 (44)	31.1	8.8	Sitagliptin 100 mg	Pioglitazone
Roden et al (2013)	899	24 weeks	55.4	348 (39)	28.4	7.9	Sitagliptin 100 mg	Placebo; empagliflozin
Ferrannini et al (2013)	444	78 weeks	60.0	220 (50)	30.1	7.9	Sitagliptin 100 mg	Empagliflozin
TECOS (2015)	14 540	3 years	65.4	4297 (30)	30.2	7.2	Sitagliptin 100 mg	Placebo
Roden et al (2015)	899	76 weeks	55.0	340 (38)	28.4	7.9	Sitagliptin 100mg	Placebo; empagliflozin
Moses et al (2016)	422	54 weeks	54.9	229 (54)	29.1	8.4	Sitagliptin 100 mg	Placebo; pioglitazone
SPIKE (2016)	274	104 weeks	63.7	105 (38)	25.0	8.1	Sitagliptin 100 mg	Conventional therapy
Wang et al (2017)	380	24 weeks	57.6	186 (49)	25.9	8.1	Sitagliptin 100 mg	Placebo
Pratley et al (2017)	1232	52 weeks	55.0	568 (46)	31.9	8.6	Sitagliptin 100 mg	Ertugliflozin
START-J (2017)	291	104 weeks	70.5	119 (41)	24.1	7.6	Sitagliptin 50 mg	Glimepiride
BEGIN (2013)	454	26 weeks	55.6	185 (41)	30.4	8.9	Sitagliptin 100 mg	Insulin degludec
CompoSIT-R (2018)	613	24 weeks	67.2	258 (42)	31.7	7.7	Sitagliptin 100 mg	Dapagliflozin
Vildagliptin
GALIANT (2009)	2627	12 weeks	55.7	1285 (49)	32.4	8.0	Vildagliptin 100 mg	Thiazolidinediones
Kanazawa et al (2017)	73	24 months	69.1	27 (37)	24.3	7.9	Vildagliptin 100 mg	Conventional therapy
VERIFY (2019)	1999	5 years	54.3	1060 (53)	31.1	6.7	Vildagliptin 100 mg	Placebo

BMI=body mass index; CVD=cardiovascular diseases; DPP-4=dipeptidyl-peptidase-4; OAD=other antidiabetes drugs; NR=not reported; SU=sulphonylureas; T2DM=type 2 diabetes mellitus.

Risk of bias and quality of evidence

The risk of bias of eligible studies is shown in supplementary table D and supplementary figure A. Sixty seven (82%) of 82 trials had low risks or some concerns of bias across all five domains evaluated. The quality of evidence for outcomes in pairwise meta-analysis was rated as moderate or low in the comparisons between dipeptidyl peptidase-4 inhibitors and placebo or non-incretin drugs (table 2; table 3).

Table 2

GRADE profile of DPP-4 inhibitors and risk of gallbladder or biliary diseases in patients with type 2 diabetes compared with placebo or non-incretin drugs in pairwise meta-analyses

No of participants (No of trials); mean follow-up	Quality assessment						Summary of findings					Overall quality of evidence
	Risk of bias	Inconsistency	Indirectness	Imprecision	Publication bias		Study event rates		Relative effect, OR (95% CI)	Anticipated absolute effect*
							With DPP-4 inhibitors	With control		Risk with control	Risk difference (95% CI) with DPP-4 inhibitors
Composite of gallbladder or biliary diseases
104 833 (82); 1.14 years	Some concerns†	Not serious	Not serious	Some concerns‡	None; symmetrical funnel plot; Egger’s test P=0.48		0.60% (338/56 005)	0.52% (256/48 828)	1.22 (1.04 to 1.43)	50	11 (2 to 21)	Low† ⨁⨁ ? ?
Cholecystitis
91 951 (60); 1.20 years	Some concerns†	Not serious	Not serious	Not serious	None; symmetrical funnel plot; Egger’s test P=0.49		0.36% (173/48 231)	0.25% (110/43 720)	1.43 (1.14 to 1.79)	35	15 (5 to 27)	Moderate† ⨁⨁⨁ ?
Cholelithiasis
88 423 (49); 1.36 years	Some concerns†	Not serious	Not serious	Some concerns‡	None; symmetrical funnel plot; Egger’s test P=0.51		0.23% (110/47 240)	0.23% (94/41 183)	1.08 (0.83 to 1.39)	26	2 (−4 to 10)	Low† ⨁⨁ ? ?
Biliary diseases
72 900 (28); 1.58 year	Some concerns†	Not serious	Not serious	Some concerns‡	None; symmetrical funnel plot; Egger’s test P=0.97		0.12% (46/37 591)	0.12% (42/35 309)	1.00 (0.68 to 1.47)	15	0 (−5 to 7)	Low† ⨁⨁ ? ?

CI=confidential interval; DPP-4=dipeptidyl-peptidase-4; GRADE=Grading of Recommendations Assessment, Development, and Evaluation; OR=odds ratio.

GRADE Working Group grades of evidence: high certainty=we are very confident that the true effect lies close to that of the estimate of the effect; moderate certainty=we are moderately confident in the effect estimate—the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different; low certainty=our confidence in the effect estimate is limited—the true effect may be substantially different from the estimate of the effect; very low certainty=we have very little confidence in the effect estimate—the true effect is likely to be substantially different from the estimate of effect.

Anticipated absolute effect is event rate per 10 000 person years.

All randomised controlled trials included in this analysis were designed to evaluate efficacy of DPP-4 inhibitors and did not report gallbladder or biliary diseases as safety endpoints of interest. Concerns might arise about selective reporting of results, considering that occurrence of gallbladder or biliary diseases might not be fully reported in trials.

Considering wide confidence intervals.

Table 3

Summary findings from pairwise and network meta-analysis

Analyses	Treatment comparisons	Composite of gallbladder or biliary diseases			Cholecystitis			Cholelithiasis			Biliary diseases
		Odds ratio (95% CI)	Quality of evidence*		Odds ratio (95% CI)	Quality of evidence*		Odds ratio (95% CI)	Quality of evidence		Odds ratio (95% CI)	Quality of evidence
Effects from traditional pairwise meta-analysis	DPP-4 inhibitors v placebo or non-incretin drugs	1.22 (1.04 to 1.43)	Low		1.43 (1.14, 1.79)	Moderate		1.08 (0.83 to 1.39)	Low		1.00 (0.68 to 1.47)	Low
Mixed effects from network meta-analysis (direct+indirect)	DPP-4 inhibitors v controls†	1.24 (1.06 to 1.46)	Low		1.45 (1.14 to 1.84)	Moderate		1.14 (0.87 to 1.49)	Low		0.95 (0.64 to 1.42)	Low
	GLP-1 receptor agonists v controls†	1.31 (1.17 to 1.47)	High		1.37 (1.13 to 1.65)	High		1.27 (1.08 to 1.51)	High		1.42 (0.97 to 2.08)	Moderate
	SGLT-2 inhibitors v controls†	0.94 (0.81 to 1.10)	Moderate		0.94 (0.76 to 1.16)	Moderate		0.87 (0.66 to 1.15)	Moderate		0.98 (0.67 to 1.45)	Low
	DPP-4 inhibitors v GLP-1 receptor agonists	0.95 (0.79 to 1.16)	Moderate		1.06 (0.79 to 1.43)	Low		0.89 (0.66 to 1.21)	Low		0.67 (0.39 to 1.15)	Low
	DPP-4 inhibitors v SGLT-2 inhibitors	1.32 (1.06 to 1.64)	Low		1.55 (1.13 to 2.12)	Moderate		1.3 1(0.90 to 1.91)	Low		0.97 (0.56 to 1.67)	Low
	GLP-1 receptor agonists v SGLT-2 inhibitors	1.39 (1.15 to 1.68)	Low		1.46 (1.09 to 1.94)	Low		1.46 (1.06 to 2.02)	Low		1.44 (0.84 to 2.46)	Low
Indirect effects from network meta-analysis	DPP-4 inhibitors v controls†	1.21 (0.70 to 2.11)	Low		1.00 (0.34 to 2.91)	Low		1.66 (0.75 to 3.71)	Low		1.56 (0.38 to 6.42)	Low
	GLP-1 receptor agonists v controls†	1.13 (0.65 to 1.95)	Moderate		2.18 (0.72 to 6.62)	Moderate		0.75 (0.35 to 1.60)	Moderate		0.69 (0.17 to 2.69)	Low
	SGLT-2 inhibitors v controls†	1.37 (0.62 to 3.02)	Low		0.85 (0.24 to 2.96)	Low		1.02 (0.31 to 3.33)	Low		1.29 (0.23 to 7.19)	Low
	DPP-4 inhibitors v GLP-1 receptor agonists	0.95 (0.77 to 1.16)	Moderate		1.12 (0.82 to 1.52)	Low		0.85 (0.61 to 1.18)	Low		0.61 (0.35 to 1.09)	Low
	DPP-4 inhibitors v SGLT-2 inhibitors	1.27 (1.02 to 1.59)	Low		1.44 (1.04 to 2.00)	Moderate		1.22 (0.82 to 1.80)	Low		0.97 (0.55 to 1.70)	Low
	GLP-1 receptor agonists v SGLT-2 inhibitors	1.41 (1.16 to 1.71)	Low		1.46 (1.09 to 1.94)	Low		1.50 (1.08 to 2.09)	Low		1.50 (0.87 to 2.60)	Low

CI=confidence interval ; DPP-4=dipeptidyl peptidase-4; GLP-1=glucagon-like peptide-1; SGLT-2=sodium-glucose cotransporter-2.

Quality of evidence was evaluated using non-contextualised GRADE approach.

Controls=placebo or other antidiabetes drugs (excluding DPP-4 inhibitors, GLP-1 receptor agonists, and SGLT-2 inhibitors).

Association of dipeptidyl peptidase-4 inhibitors with gallbladder or biliary diseases

We found a significant association between dipeptidyl peptidase-4 inhibitors and an increased risk of the composite of gallbladder or biliary diseases (odds ratio 1.22 (95% confidence interval 1.04 to 1.43); τ2=0.027; absolute risk difference 11 (2 to 21) more events per 10 000 person years) compared with placebo or non-incretin drugs (table 2; fig 2; supplementary figure B). Dipeptidyl peptidase-4 inhibitors significantly increased the risk of cholecystitis (odds ratio 1.43 (1.14 to 1.79); τ2=0; absolute risk difference 15 (5 to 27) more events per 10 000 person years) but not of cholelithiasis or biliary diseases (fig 2; supplementary figure B).

Fig 2

Risks of cholecystitis, cholelithiasis, and biliary disease in patients taking dipeptidyl peptidase-4 inhibitors (DPP-4i). Absolute risk difference (ARD) is number of events per 10 000 person years. Control groups=placebo or non-incretin drugs. CI=confidence interval

Effects of duration of treatment, specific dipeptidyl peptidase-4 inhibitors, and types of control

Dipeptidyl peptidase-4 inhibitors with a longer duration of treatment (≥26 weeks) were significantly associated with increased risks of the composite of gallbladder or biliary diseases (odds ratio 1.24, 1.04 to 1.48) and cholecystitis (1.51, 1.17 to 1.95), but those with a shorter duration (<26 weeks) were not (supplementary figure C). When we classified the duration of treatment into four groups, we observed a stepwise increase in the risk of outcomes as the duration of treatment increased (supplementary table E), although the group differences were not significant (P value for interaction >0.05).

The modified effects of specific dipeptidyl peptidase-4 inhibitors on the risks of cholecystitis were not significant (supplementary figure D). We observed no significant differences between pure and active controls for the outcomes (supplementary figure E).

Sensitive analyses

After use of different pooling methods and random effect models and inclusion of double arm, zero events studies (supplementary tables F-H), the results did not change for any outcomes. When we omitted each trial one by one (supplementary figure F) or removed the CAROLINA and CARMELINA trials simultaneously (supplementary table I), the results of outcomes remained stable, except for the composite of gallbladder or biliary diseases.

Publication bias

We observed no publication bias in the included studies for the composite of gallbladder or biliary diseases, cholelithiasis, cholecystitis, or biliary diseases when we used Egger’s test (table 2). The funnel plots were visually symmetrical (supplementary figure G) in traditional pairwise meta-analysis.

Comparisons between dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, and sodium-glucose cotransporter-2 inhibitors in network meta-analysis

We included 184 trials in a network of comparisons among the three classes of antidiabetes drugs (supplementary figure H; supplementary tables J and K). Network plots of treatment comparisons are shown in figure 3, and network estimates are shown in figure 4 and supplementary table L. Dipeptidyl peptidase-4 inhibitors increased the risks of the composite of gallbladder or biliary diseases (odds ratio 1.32, 1.06 to 1.64) and cholecystitis (1.55, 1.13 to 2.12) compared with sodium-glucose cotransporter-2 inhibitors, as well as compared with placebo or other antidiabetes drugs (fig 4). We found no significant differences between dipeptidyl peptidase-4 inhibitors and glucagon-like peptide-1 receptor agonists in any outcomes. The results from random effects models were similar to those from fixed effect models (supplementary tables L and M). Table 3 shows the summary of the results from traditional pairwise meta-analysis and the network meta-analysis.

Fig 3

Network plots of comparisons of dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and sodium-glucose cotransporter-2 (SGLT-2) inhibitors in network meta-analyses. Size of circle in each network is proportional to number of participants randomly assigned to treatment comparison. Width of each line is proportional to number of trials comparing two connected treatments. Number provided for each treatment class (in parentheses) indicates number of patients randomised to treatment in network. Controls are placebo or other antidiabetes drugs (excluding DPP-4 inhibitors, GLP-1 receptor agonists, and SGLT-2 inhibitors); k indicates number of comparisons; n indicates number of patients per comparison

Fig 4

Estimates in network meta-analysis comparing dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and sodium-glucose cotransporter-2 (SGLT-2) inhibitors. Controls are placebo or other antidiabetes drugs (excluding DPP-4 inhibitors, GLP-1 receptor agonists, and SGLT-2 inhibitors). No of studies indicates number of studies in direct comparison. CI=confidence interval

We observed no significant between study heterogeneities, between design inconsistencies, or inconsistencies between direct and indirect treatment comparisons for all outcomes in the network meta-analysis (supplementary tables N and O). We observed no small study effects in the studies included in the network meta-analysis (supplementary figure I). The quality of evidence for results from the network meta-analysis is shown in supplementary table P.

Discussion

To our knowledge, this is the first study to systematically assess the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases. Our findings showed that dipeptidyl peptidase-4 inhibitors increased the risk of cholecystitis, which tended to be observed in patients with a longer duration of treatment. The risk of cholecystitis was higher with dipeptidyl peptidase-4 inhibitors than with sodium-glucose cotransporter-2 inhibitors but similar to that with glucagon-like peptide-1 receptor agonists.

Comparison with other studies

Few studies have assessed the effects of dipeptidyl peptidase-4 inhibitors on gallbladder or biliary diseases.2 116 A retrospective study enrolled patients with type 2 diabetes and reported no significant associations between dipeptidyl peptidase-4 inhibitors and an increased risk of gallbladder or biliary diseases.7 However, the retrospective design could be associated with various biases and confounding, and the study did not analyse the specific risks of cholecystitis. We noted some discrepancies between the pooled results of our study and those of some large scale randomised controlled trials, such as SAVOR-TIMI trial, and the reasons for this are unknown. Possibly, differences in populations, periods of follow-up, specific dipeptidyl peptidase inhibitors, and reporting of outcomes may play a role. Intriguingly, although we found that glucagon-like peptide-1 receptor agonists increased the risks of cholecystitis compared with sodium-glucose cotransporter-2 inhibitors or controls (placebo or other antidiabetes drugs) in the network meta-analysis, consistent with previous studies,4 6 117 dipeptidyl peptidase-4 inhibitors were not associated with lower risks of cholecystitis than glucagon-like peptide-1 receptor agonists. Furthermore, the results from traditional pairwise meta-analysis were similar to those from the network meta-analysis that simultaneously combined direct and indirect effects.

The gallbladder or biliary effects of dipeptidyl peptidase-4 inhibitors might be attributed to the roles of glucagon-like peptide-1 and glucose dependent insulinotropic polypeptide.5 118 Glucagon-like peptide-1 is involved in impaired gallbladder mobility and contractility by inhibiting the secretion of cholecystokinin,118 which could contribute to the development of gallbladder or biliary diseases.4 Additionally, glucose dependent insulinotropic polypeptide was recently reported to play a role in gallbladder relaxation.5

Increased risks of gallbladder or biliary diseases tended to be apparent in patients receiving dipeptidyl peptidase-4 inhibitors for a longer duration, which suggested that a sufficiently long treatment duration was needed to allow the gallbladder or biliary effects of dipeptidyl peptidase-4 inhibitors to be explored. A previous study did not observe the effects of sitagliptin on gallbladder emptying,119 possibly owing to the short treatment duration (12 weeks). More importantly, as the duration of prescriptions for dipeptidyl peptidase-4 inhibitors is usually longer in routine practice than in clinical trials,120 awareness of the role of treatment duration might be of great clinical importance.

Strengths and limitations of study

This study has several strengths. Firstly, this is the first systematic review and meta-analysis pooling the results of randomised controlled trials to assess the association between dipeptidyl peptidase-4 inhibitors and gallbladder or biliary diseases. Secondly, our findings showed the possible effects of the duration of dipeptidyl peptidase-4 inhibitor treatment on gallbladder or biliary diseases. Thirdly, we did network meta-analysis to compare the three antidiabetes drugs.

Nevertheless, our study has some limitations. Firstly, the included studies were not specifically designed to evaluate the effects of dipeptidyl peptidase-4 inhibitors on gallbladder or biliary diseases. Secondly, gallbladder or biliary diseases were not predefined safety outcomes in the included studies and hence might be underreported. Although the data might not be fully collected, this is unlikely to bias the associations found. Thirdly, patient level data were inaccessible for the meta-analysis.

Policy implications

Dipeptidyl peptidase-4 inhibitors are considered to have a good safety profile,2 and concerns about potential adverse events have usually focused on heart failure,121 bullous pemphigoids,122 123 pancreatitis,124 and severe joint pain.125 The US Food and Drug Administration has released warnings about the risk of heart failure and severe joint pain with dipeptidyl peptidase-4 inhibitors.126 127 However, the potential risk of cholecystitis attributed to dipeptidyl peptidase-4 inhibitors has not yet been investigated. This study highlights the importance and raises awareness of the risks of cholecystitis with dipeptidyl peptidase-4 inhibitors for physicians and should encourage researchers to make them predefined safety endpoints. These findings may be also extended to double or triple agonists, including glucagon-like peptide-1, glucose dependent insulinotropic polypeptide, or both.

Although dipeptidyl peptidase-4 inhibitors increased relative risk for cholecystitis in patients with type 2 diabetes, the overall absolute risk increase was small (15 cases per 10 000 people per year). This absolute risk increase should be weighed against the benefits of dipeptidyl peptidase-4 inhibitor treatment.11 30

Conclusions

Dipeptidyl peptidase-4 inhibitors might increase the risk of cholecystitis in patients with type 2 diabetes. Our findings suggest that physicians should be concerned about increased risks of cholecystitis in patients with type 2 diabetes treated with dipeptidyl peptidase-4 inhibitors in clinical practice, especially with longer treatment durations. Future studies are encouraged to fully report gallbladder or biliary diseases.

Glucagon-like peptide-1 (GLP-1) is involved in the development of impaired gallbladder motility

Liraglutide, an important member of the GLP-1 receptor agonist class, has been reported to be associated with an increased risk of gallbladder or biliary diseases

Glucose dependent insulinotropic polypeptide has also been reported to affect gallbladder motility

Dipeptidyl peptidase-4 (DPP-4) inhibitors were significantly associated with an increased risk of the composite of gallbladder or biliary diseases and cholecystitis

DPP-4 inhibitors were not associated with increased risk of cholelithiasis and biliary diseases

Risk of cholecystitis was increased with DPP-4 inhibitors compared with sodium-glucose cotransporter-2 inhibitors but not GLP-1 receptor agonists