For the millions of people with type 2 diabetes who take daily basal insulin injections, the prospect of eventually discontinuing insulin is a meaningful quality-of-life goal. GLP-1 receptor agonists have generated enormous enthusiasm as potential insulin-sparing agents, but whether their metabolic benefits actually translate into freedom from insulin — compared with other add-on therapies — has not been rigorously tested until now.
This target trial emulation used U.S. Veterans Health Administration electronic health records to construct nearly 8,900 matched sets of patients with type 2 diabetes already on basal insulin who then initiated either a GLP-1 receptor agonist (predominantly semaglutide at 76.6%), an SGLT-2 inhibitor (nearly all empagliflozin), or a DPP-4 inhibitor (predominantly alogliptin). Over three years of follow-up, insulin discontinuation — defined as a gap of at least 12 months in insulin fills — occurred in 16.7% of GLP-1RA initiators, 17.9% of SGLT-2i initiators, and 17.1% of DPP-4i initiators. The risk ratio for GLP-1RA versus SGLT-2i was 0.93 (95% CI, 0.86–1.01), and versus DPP-4i was 0.98 (CI, 0.87–1.09) — neither reaching statistical significance.
These findings challenge a widely held clinical assumption. Despite GLP-1 receptor agonists' superior HbA1c-lowering potency and weight-loss effects relative to DPP-4 inhibitors — and their presumed advantage over SGLT-2 inhibitors in beta-cell function restoration — the real-world rate of achieving durable insulin discontinuation was essentially identical across all three drug classes. This is a notable null result. The study population was predominantly older, male, and White veterans, limiting generalizability to women and younger or more diverse populations. Additionally, the 12-month gap definition is a pragmatic proxy for true discontinuation and may not capture intentional deprescribing. The target trial emulation design, while a methodological strength, cannot fully eliminate confounding by indication. Clinically, the data suggest that for most insulin-dependent patients with type 2 diabetes, drug class selection alone is unlikely to determine whether insulin cessation is achievable — patient-level factors and structured deprescribing protocols may matter more.