In the present meta-analysis of the association between DPP4i and risk of gallbladder or biliary diseases, we investigated 75 RCTs involving 97,150 participants with type 2 diabetes. Our findings revealed that the use of DPP4i was associated with the risk of cholecystitis. In the subgroup analysis, the pooled RR of cholecystitis was influenced by the patient age and follow-up duration. That is, the risk of cholecystitis associated with DPP4i was concentrated in the group of patients of advanced age or in those who had been exposed to the drugs for a long period of time. However, the use of DPP4i did not increase the risks of cholelithiasis, cholangitis, bile duct stones, and biliary colic.

Current studies suggest that DPP4 inhibitor use increases the risk of cholecystitis, which is consistent with our present results. A previous European Medicines Agency Assessment Report also suggested a strong association between DPP4 inhibitor use and the risk of cholecystitis [8]. According to another meta-analysis, the use of DPP4i was associated with a 43% increased risk of cholecystitis [11]. Compared with this article, we have several differences and new ideas. First, there are several papers linking glucose control and insulin sensitivity to biliary diseases [12, 13]. It is theoretically possible that non-incretin therapies could reduce the risk of biliary diseases when compared with “placebo therapy”; therefore, the apparent “increased rate” associated with DPP4i therapies could be an artifact. To exclude this effect, a separate placebo-controlled group was tested. Second, we focused on the subgroup analyses of patient age and duration of drug use. We found that the cholecystitis risk associated with DPP4i was greater in patients of advanced age (≥ 60 years old) or in those who had experienced long-time drug exposure (≥ 52 weeks). Furthermore, we also performed a drug-based subgroup analysis and found no agent heterogeneity. These results provide a theoretical basis for the clinical identification of high-risk patient groups and extend upon the previous work.

In the aging process, the wall of the gallbladder gradually becomes atrophied, causing the contractile function of the organ to decrease. These degenerative changes can lead to cholestasis. At the same time, the end of the common bile duct and the sphincter of Oddi become relaxed, rendering them prone to retrograde infection. These pathophysiological changes lead to older people being more susceptible to cholecystitis. This may explain the higher risk of cholecystitis in the elderly who use DPP4i [14]. Additionally, GLP-1 has been shown to enhance the proliferative and functional activities of cholangiocytes and prevent their apoptosis. Preclinical evidence suggests that GLP-1 reduces the production of bile acids. Furthermore, animal studies have shown that incretin-based therapy induces the prolongation of gallbladder refilling [15]. Other drugs, such as naltrexone, bupropion, and octreotide, have been demonstrated to increase the risk of gallbladder disorders through this mechanism. [16, 17]

GLP-1RAs would cause a loss of appetite and a significant improvement in body weight [18]. Rapid weight loss mobilizes cholesterol in the liver and adipose tissue and may increase cholesterol saturation in bile, which has been recognized as one of the risk factors for gallbladder disease [19, 20]. By contrast, DPP4i have no direct effect on gastrointestinal motility and lack the weight-loss effect of GLP-1RAs [21]. However, the use of DPP4i also increased the risk of acute cholecystitis. Remarkably, the enzyme DPP4 is also expressed on immune-related cells, and it seems to influence T-cell growth, differentiation, and activation. DPP4i may influence potential anti-inflammatory effects in addition to increasing the risk of cholecystitis [22, 23]. Accordingly, some studies have shown that DPP4i are associated with an increased risk of infections, such as nasopharyngitis and urinary tract infection [24, 25]. Interestingly, the enzyme DPP4 is located throughout the body, and it influences many other peptides and hormones, such as glucagon-like peptide 2 (GLP-2), glucose-dependent insulinotropic peptide (GIP), and polypeptide-YY (PYY), which have all been proven to exert effects on gallbladder motility [26]. PYY is a strong inhibitor of gallbladder emptying [27], GLP-2 receptor activation greatly increased the gallbladder volume in mice [28], and GIP was associated with cholecystokinin secretion in mouse models [29]. These could be the potential mechanisms through which DPP4i induce cholecystitis.

Several strengths of this article are worth mentioning. We conducted a prespecified subgroup analysis and found that advanced age and long-term drug exposure significantly increased the incidence of cholecystitis associated with DPP4i, which suggests that clinicians should be more cautious when using these types of inhibitors in these populations. From the clinical point of view, several gallbladder diseases (e.g., cholecystitis, cholangitis) were selected to explore the relationship between DPP4i and gallbladder disease. The severity and clinical significance of these diseases are sufficient to arouse the attention of clinicians.

However, there are several limitations of our study that are worth considering: (1) Since the occurrence of gallbladder and biliary diseases was not a primary or secondary outcome of the RCTs included in this article, reporting bias is possible. (2) In these patients with type 2 diabetes included in RTCs, other risk factors such as weight change, dietary habits, drinking history, obesity and dyslipidemia might contribute to the occurrence of gallbladder and biliary diseases. However, the above information is not fully accessible, the bias caused by these factors cannot be corrected. (3) Information on serious gallbladder and biliary events, such as hospitalization, surgery, or even death, is not fully available. (4) We only included RCTs. However, recent recommendations on the methodological quality of systematic reviews (AMSTAR2) suggest that if observational studies based on large population databases are combined with RCTs, the meta-estimates will generate precise estimates of intervention effects. It may be necessary to conduct special reviews of observational studies in the future.