For people with type 2 diabetes who already have nerve damage in their feet, a matched comparison of 39,540 US patients found fewer foot ulcers on GLP-1 receptor agonists than on DPP-4 inhibitors — and nearly twice the rate of Charcot neuroarthropathy, a joint-destroying complication of the foot. Amputation and bone-infection rates were the same in both groups. The practical conclusion the authors draw is not to avoid the drugs but to watch the feet more closely. It was published on 7 September in the Journal of Neurology.
The numbers
Using the TriNetX US Collaborative Network, the team identified adults with type 2 diabetes and diabetic neuropathy (ICD-10-CM E11.40) starting either drug class, then matched them 1:1 on demographics, comorbidities and medications, leaving 19,770 patients per arm. Outcomes were followed for one and two years using Kaplan-Meier estimates and Cox proportional hazards models, with a Bonferroni-corrected threshold of p < 0.01.
At one year, diabetic foot ulcers occurred in 2.2% of the GLP-1 group against 2.7% on DPP-4 inhibitors (HR 0.813, 95% CI 0.716–0.922). That half-point difference works out to roughly 435 patients against 534, given the rounding — about one ulcer avoided for every 200 people treated for a year. Amputation was unchanged at 0.5% in both arms (HR 1.073, 95% CI 0.810–1.421), as was foot and ankle osteomyelitis at 0.4% (HR 0.978, 95% CI 0.708–1.349). Both of those intervals straddle 1, meaning the data are equally consistent with a small benefit and a small harm.
Charcot neuroarthropathy went the other way: 0.3% on GLP-1 drugs against 0.2% on DPP-4 inhibitors, HR 1.993 (95% CI 1.297–3.064). In absolute terms that is roughly 59 cases against 40 — about 19 extra, or one per thousand people treated. All-cause mortality was lower with GLP-1 drugs, a finding the authors themselves flag as hypothesis-generating rather than a result to act on.
Why the Charcot signal is plausible but unsettled
The authors list treatment-induced neuropathy of diabetes among their keywords, and it is the obvious candidate mechanism. It is a recognised, acute worsening of nerve function that can follow a rapid improvement in glucose control, and GLP-1 drugs lower glucose quickly. That makes the association biologically coherent. It does not make it causal, and this study was not built to settle that.
What it does change is monitoring. Anyone with established diabetic neuropathy starting one of these drugs has a reason to keep foot checks on the schedule rather than letting them lapse, and to treat a hot, swollen, painless foot as urgent rather than as a sprain. Charcot is most treatable when it is caught before the joint collapses.
What is not yet known. This is a retrospective analysis of de-identified electronic health records, not a randomised trial. Propensity matching balanced the recorded variables, but people prescribed a GLP-1 drug differ from people prescribed a DPP-4 inhibitor in ways the records do not fully capture — weight, kidney function, how advanced the neuropathy already was, and how often they see a specialist. That last one matters twice over, because Charcot is under-diagnosed and frequently coded late, so both arms almost certainly undercount it, and the arm under closer surveillance will undercount it less. The confidence interval on the Charcot hazard ratio, 1.30 to 3.06, is consistent with anything from a third more cases to triple. The two-year results are not broken out in the abstract, and no funding was declared.
Sources: Tawfik F, Yalley E, Sienkaniec J, et al. “GLP-1 receptor agonists vs DPP-4 inhibitors and neuropathic complications in type 2 diabetes.” Journal of Neurology 2026;273(10), published online 7 September 2026. PubMed 42704495
