Red Flags for Misdiagnosed Type 2, When to Order Antibodies and C‑Peptide

Experts describe clinical warning signs that should prompt reconsideration of a type 2 diabetes label and outline how autoantibody and C‑peptide testing can help differentiate type 1 and autoimmune diabetes in adults.

In this segment, Samson shifts the focus from baseline phenotype to dynamic clinical clues that may signal a misdiagnosis of type 2 diabetes. She discusses scenarios in which patients with presumed type 2 diabetes exhibit poor response to noninsulin therapies despite strong adherence, rapidly progress to insulin dependence within a few years of diagnosis, or require very high insulin doses (eg, >2 units/kg) suggestive of extreme insulin resistance or another underlying process.

Additional red flags include the development of ketosis, particularly in the setting of sodium-glucose cotransporter 2 (SGLT2) inhibitor use, and lack of expected response to oral hypoglycemic agents such as metformin or insulin secretagogues. Samson notes that these patterns should trigger clinicians to question whether the original diagnosis of type 2 diabetes is correct.

Umpierrez then details how autoantibody testing and C‑peptide measurement are integrated into the 2026 AACE classification algorithm when type 1 diabetes or latent autoimmune diabetes in adults (LADA) is suspected. He explains that glutamic acid decarboxylase 65 (GAD65) antibodies are the most commonly used and often remain detectable in adults, including those with LADA. Additional autoantibodies include those against IA‑2, zinc transporter 8, and insulin. He cautions that up to 10% of patients with type 1 diabetes may be antibody-negative and that immunosuppressive therapies, including corticosteroids, can mask antibody positivity. When testing for insulin autoantibodies, he emphasizes that patients must be insulin-naive; exogenous insulin exposure can itself induce autoantibody formation.

Regarding C‑peptide, Samson and Umpierrez explain that its interpretation must be contextualized with simultaneous glucose levels and recent glycemic control. Umpierrez notes that severe hyperglycemia or recent diabetic ketoacidosis can suppress endogenous insulin secretion—a phenomenon of glucose toxicity—leading to transiently low C‑peptide levels.

The 2026 AACE algorithm, aligned with recommendations from major diabetes societies, considers a low C‑peptide (eg, <200 pmol/L) coupled with at least 1 positive autoantibody as supportive of type 1 diabetes in adults, particularly at very low C‑peptide levels (<35 pmol/L) after age 35 years. By contrast, a C‑peptide concentration greater than approximately 600 pmol/L suggests type 2 diabetes, with intermediate levels requiring integration of clinical features, family history, and autoantibody results. Through this framework, Samson and Umpierrez illustrate how the algorithm operationalizes laboratory tools to refine diabetes classification beyond phenotype alone.