VIENNA — Two experts from the UK who spoke at the European Association for the Study of Diabetes (EASD) 2025 Annual Meeting offered advice for avoiding misdiagnosis of adults who present with diabetes that doesn’t align with the classical characteristics of type 1 or type 2.

Angus Jones, PhD, MBBS, associate professor at the University of Exeter and honorary consultant physician at the Royal Devon and Exeter Hospitals in Exeter, England, spoke about distinguishing adult-onset type 1 diabetes (T1D) from type 2 diabetes (T2D).

The danger of misdiagnosing adult-onset T1D is in delaying the initiation of essential insulin, said Jones. Even if patients are eventually prescribed insulin, if they retain the T2D label, they may not be provided with access to standard of care education and technology. “This matters because it’s fundamental to treatment. We have completely different treatment guidance for type 1 and type 2,” he noted in his introduction.

Mark W.J. Strachan, MD, consultant physician in endocrinology, diabetes, and acute medicine at the Western General Hospital, and an honorary professor at the University of Edinburgh, Edinburgh, Scotland, discussed the role of routine biomarker assessment in flagging people with monogenic diabetes.

For those with monogenic diabetes who are misdiagnosed as having T1D, the mistake can mean a lifetime burden of unnecessary insulin treatment when simple, inexpensive sulfonylureas would have been the appropriate therapy, said Strachan, who presented data showing that “there’s a lot of undiagnosed monogenic diabetes out there.”

T1D or T2D? Age at Diagnosis Is Important

In a 2015 systematic review, Jones’ Exeter group showed that age at diagnosis and time to insulin were consistently the most discriminatory criteria for distinguishing T1D from T2D, while BMI didn’t add much more information. For predicting insulin deficiency, the best cutoffs — based on combined sensitivity and specificity — were age at diagnosis < 30 years (2 studies) or ≤ 40 years (4 studies) and time to requiring insulin, < 1 year (1 study) or ≤ 2 years (2 studies).

In a subsequent, unpublished study, Jones and his colleagues examined detailed data for 1800 adults newly diagnosed with diabetes and followed them for 4 years with annual C-peptide measurements. Features that added independently to age at diagnosis for differentiating T1D from T2D were BMI < 25 (sensitivity 47%; specificity 93% for T1D), unintentional weight loss (80% for T1D and 75% for T2D), and A1c > 115 mmol/mol (12.9%) at presentation (36% for T1D and 87% for T2D). The presence of diabetic ketoacidosis was very specific (96%), but not very sensitive (19%).

The reason age at onset is such a strong discriminator of diabetes type isn’t because T1D onset is less common in adults, but because “it is drowned out by the incredibly high rates of type 2 with older age. We have a needle in the haystack,” said Jones.

Indeed, he pointed to data showing that misclassification is very common. Between 8% and 20% of adults newly diagnosed with T1D actually have either T2D or monogenic diabetes. On the flip side, about 40% with definite T1D after age 30 are initially diagnosed and treated as T2D. In that study, nearly all of these patients rapidly progressed to insulin treatment within 3 years yet were still carrying the T2D diagnosis and weren’t receiving the technology or the education that is standard of care for T1D.

Biomarker testing is necessary to diagnose T1D in adults. Islet autoantibodies are most useful at or close to the time of diagnosis (ie, < 3 years), while C-peptide is most useful in longstanding diabetes, longer than 3-5 years. Autoantibodies predict progression to T1D, while C-peptide can still be high shortly after T1D onset. “What they tell us about is current treatment requirements,” Jones noted.

International guidelines now advise routine autoantibody testing at diagnosis of all adults with suspected T1D onset. Unpublished data from the Exeter group suggests that for an individual with suspected T1D — or given a T2D diagnosis but with rapid progression to insulin — a single positive islet autoantibody will usually confirm T1D, while multiple positive antibodies usually confirm the diagnosis regardless of clinical suspicion.

Negative autoantibodies don’t entirely exclude T1D, but “other diagnoses should strongly be considered,” Jones said.

Adding C-Peptide Testing to the Diagnostic Process

Strachan began his presentation by noting that measurement of C-peptide in blood is stable for at least 8 hours, doesn’t require special handling, and does not require the patient to fast or omit insulin. However, it is most discriminatory when blood glucose levels are above 4 mmol/L (72 mg/dL) because endogenous insulin production (if there is any) will be suppressed during hypoglycemia.

Urine C-peptide testing is less useful in a clinic setting, since it is best measured 2 hours after a main meal (usually evening), but it can be collected by the patient at home.

Abbott and Roche are the two main assay manufacturers of blood C-peptide tests. A study from Strachan’s group found that the Roche assay measures higher levels than Abbott’s. “You obviously have to be very careful about rigidly applying cutoffs…and use your clinical judgement,” he said.

In 2018, Strachan’s clinic in Edinburgh introduced a diagnostic algorithm for people diagnosed with T1D that added the C-peptide test to all the other routine lab tests in all patients with a clinician diagnosis of T1D for a minimum of 3 years. The move was triggered by a patient with a diagnosis of T1D who had been seen for over 20 years and who was identified through external family screening as actually having a form of sulfonylurea-sensitive monogenic diabetes.

The patient had been diagnosed with T1D at age 17. At age 36, her A1c with insulin treatment was 75-90 mmol/mol (9.0%-10.4%). She’d had two C-peptide tests, with values of 482 pmol/L in 2021 and 220 pmol/L in 2023, yet no action was taken at either time. Monogenic testing was finally done in late 2024, and she was found to have a variant. She’s now off insulin and takes gliclazide 20 mg twice daily. “Her glycemic control has been transformed. Her life has been transformed,” Strachan said.

“She had had insulin treatment for 25 years that she never needed to have. As a service, we were extremely upset about that…we wanted to do better,” he said.

Subsequently, several of her relatives who had also been diagnosed with T1D and had high C-peptides but with no action taken were reclassified as having the same monogenic diabetes type. One family member had been taking insulin unnecessarily for over 40 years, with suboptimal control. “So this family has nearly a hundred years of cumulative insulin exposure that they never needed to have. And they were all sitting there in a clinic, in plain sight,” Strachan said.

According to the diagnostic algorithm, a C-peptide level below 200 pmol/L confirms the T1D diagnosis. A C-peptide above 900 pmol/L with significant insulin resistance changes the person’s diagnosis to T2D.

For C-peptide levels of 200-900 pmol/L, indicating significant endogenous insulin production, T1D, T2D, and monogenic diabetes are all possible diagnoses. The algorithm then calls for measurement of autoantibodies GAD, IA2, and ZnT8 and goes through the various scenarios, with genetic risk scoring or genetic testing recommended when subsequent C-peptide testing is intermediate. 

In a study that used the algorithm for 859 people with a T1D diagnosis for at least 3 years, they found 114 with a C-peptide > 200 pmol/L. “That was a lot more than I had anticipated,” Strachan noted. Of those, 54 met criteria for T1D, while 58 (6.8% of the total 859) were reclassified: 44 with T2D and 14 with monogenic diabetes. “Again, that was a big surprise to me. I don’t think we’re bad diagnosticians. I think this reflects the diagnostic difficulty and challenge in making a diagnosis,” he commented.

Cost Effectiveness of the Algorithm

This algorithm turned out to be “astonishingly cost-effective,” said Strachan. The total cost of implementing the algorithm for the 859 people — including the C-peptide and antibody tests, the monogenic screen, and the Exeter T1D genetic risk score — is £23,262 (€26,750/$31,350), or £27 (€31/$36.4) per person. In contrast, a year’s worth of injectable insulin and consumables totals £2700 per person (€3105/$3639), not including continuous glucose monitoring.

“Insulin treatment is expensive…so you only need to identify one person with a sulfonylurea-sensitive form of monogenic diabetes in your clinic who was hitherto been unrecognized and treated with insulin to pay for the entirety of a testing regime. It’s beyond cost-effective,” he said. “And of course, it’s potentially life-transforming for that individual.”

The diagnostic biomarker testing program was rolled out across Scotland in late 2021. Scotland has a total population of 6 million people, of whom about 360,000, or 6%, have diabetes. Of those, 37,124 have a diagnosis of T1D. The uptake of C-peptide testing has varied widely across various regions in the country, with about 45% total receiving them. “That geographical variation is all clinician-dependent. There is no barrier to cost in Scotland for a C-peptide measurement. There is no barrier to access. It’s all about individual behaviors of healthcare professionals,” Strachan pointed out. 

However, the overall number of requests for monogenic diabetes tests and genetic risk scores across Scotland has risen since the COVID lockdown. “That to me says that the algorithm is being used and it’s working,” he said.

Currently in Scotland there are 554 people with a diagnosis of monogenic diabetes, or 1.5 of every 1000 diabetes cases nationwide. The population prevalence is about 1 in 10,000. But, without full uptake of the algorithm, about 50% are still estimated to be undiagnosed, he noted.

An abstract presented at EASD from the Exeter team showed that the population prevalence of pathogenic monogenic variants is about 1 in 1000, or between 0.63% and 0.0015% of people with diabetes. However, not everyone with those genetic variants actually will go on to develop diabetes, Strachan pointed out.

Strachan also often uses C-peptide in people diagnosed with T2D who are on insulin treatment and has identified many who actually have misdiagnosed T1D. 

Moreover, he observed that, despite the long-held belief that people with T2D eventually need insulin because their beta cells “burn out,” in his experience, “the vast majority of people with T2D who are on insulin who attend my clinic have got C-peptide levels well in excess of 1000 [pmol/L]. They don’t have beta cell failure at all, and actually, they’re individuals who almost certainly now would do very well with some of the super weight-loss drugs.” 

Neither Jones nor Strachan reported having disclosures.

Miriam E. Tucker is a freelance journalist based in the Washington, DC area. She is a regular contributor to Medscape Medical News, with other works appearing in the Washington Post, NPR’s Shots blog, and Diatribe. She is on X at @MiriamETucker and BlueSky at @miriametucker.bsky.social.