Dipeptidyl peptidase–4 inhibitors may be associated with higher risks of certain autoimmune conditions compared with other commonly used diabetes drug classes among patients with type 2 diabetes.
Investigators conducted 3 emulated target trials using de-identified electronic health record data from 152 health care organizations in the TriNetX Global Collaborative Network. They included adult patients newly diagnosed with type 2 diabetes between 2016 and 2023 who initiated monotherapy with a dipeptidyl peptidase (DPP)-4 inhibitor, glucagon-like peptide (GLP)-1 receptor agonist, or sodium-glucose cotransporter-2 (SGLT2) inhibitor within 1 year of diagnosis. Patients with prior autoimmune disease or immunosuppressant exposure were excluded. Following 1:1 propensity score matching, the analyses included 118,419 matched DPP-4i inhibitor vs. GLP-1 receptor agonist pairs, 102,810 DPP-4 inhibitor vs. SGLT2 inhibitor pairs, and 105,869 GLP-1 receptor agonist vs. SGLT2 inhibitor pairs.
The investigators assessed 3-year incidence of autoimmune diseases, including rheumatoid arthritis, psoriasis, psoriatic arthritis, multiple sclerosis, inflammatory bowel disease, giant cell arteritis, dermatomyositis, systemic sclerosis, autoimmune thyroiditis, and bullous pemphigoid. Follow-up began at a 6-month landmark following treatment initiation to reduce immortal time and reverse causation biases. Propensity score matching accounted for demographics, comorbidities, smoking status, medications, hemoglobin A1c, estimated glomerular filtration rate, and other measured characteristics. The investigators also evaluated 3 negative control outcomes and changes in body mass index and hemoglobin A1c as positive controls.
Compared with GLP-1 receptor agonists, DPP-4 inhibitors were associated with a lower likelihood of psoriasis, psoriatic arthritis, and autoimmune thyroiditis but 2.18 times the likelihood of dermatomyositis and 1.78 times the likelihood of bullous pemphigoid. No statistically significant differences were observed for the prespecified negative control outcomes.
The DPP-4 inhibitor vs. SGLT2 inhibitor analysis showed a similar pattern. DPP-4 inhibitor treatment was associated with a lower likelihood of psoriasis and autoimmune thyroiditis, but 2.03 times the likelihood of dermatomyositis and 2.07 times the likelihood of bullous pemphigoid. DPP-4 inhibitors were also associated with 1.83 times the likelihood of giant cell arteritis. No statistically significant differences emerged for the negative control outcomes.
In contrast, the investigators found no statistically significant differences in autoimmune disease risk between GLP-1 receptor agonists and SGLT2 inhibitors after adjustment for multiple comparisons. Although point estimates suggested possible associations for Crohn's disease and giant cell arteritis, neither remained statistically significant after adjustment.
The increased bullous pemphigoid risk and decreased psoriasis risk associated with DPP-4 inhibitors were consistent with emerging pharmacovigilance findings. However, they cautioned that residual confounding related to obesity or diabetes severity could partly explain some associations. The dermatomyositis association represented a potentially novel safety signal because it had not previously been reported in large-scale epidemiologic studies and appeared consistently across the 2 DPP-4 inhibitor comparator analyses.
The study was limited by its observational design, including potential selection bias, exposure and outcome misclassification, confounding by indication, and residual confounding. The electronic health record data lacked socioeconomic and insurance information and incompletely captured some risk factors, including smoking. Unmeasured differences in diabetes severity and duration may have influenced the associations. The absence of an untreated comparator meant the findings reflected relative differences between the drug classes rather than absolute effects on autoimmune disease risk. Limited follow-up and the rarity of some autoimmune diseases may have reduced the ability to detect differences.
“These findings provide novel safety signals and may inform antidiabetic drug selection while guiding future mechanistic and prospective research,” wrote lead study author Arjun Mahajan, MS, of Harvard Medical School and the Division of Rheumatology, Inflammation, and Immunity at Brigham and Women’s Hospital, and colleagues.
Full disclosures of the study authors can be found in the study.
Source: ACR Open Rheumatology
