Can type 2 diabetes be cured? What remission trials show

The consensus group chose remission deliberately. Its report says glucose can return to the normal range in some people, spontaneously or after treatment, and can stay there after drugs are withdrawn, but that the long-term risks and benefits of that state are not well understood ^[1]. It recommended continued observation of people in remission rather than discharge from care ^[1]. The same report was published at the same time in Diabetes Care, in Diabetologia, the journal of the European Association for the Study of Diabetes, and in Diabetic Medicine, the journal of Diabetes UK ^[1][2][3], which is why the definition is often described as the ADA, EASD and Diabetes UK definition.

Three details of the definition matter when reading any claim of "reversal":

• The threshold is HbA1c below 6.5%, the diagnostic line for diabetes, not a fully normal value ^[1].
• The drugs must be stopped, and the measurement taken at least 3 months later ^[1]. Normal glucose while still on a drug is good control, not remission.
• It can relapse. The trials below show remission rates falling year by year, which is the practical reason the word cure is not used.

DiRECT randomised 49 primary care practices in Scotland and north-east England, with 149 participants in each arm ^[4]. It enrolled adults aged 20 to 65, diagnosed within the previous 6 years, with a BMI of 27 to 45 and not on insulin ^[4]. The intervention withdrew diabetes and blood pressure drugs, replaced food with an 825 to 853 kcal per day formula diet for 12 to 20 weeks, reintroduced food over 2 to 8 weeks, and then gave structured support to maintain weight loss ^[5]. At 12 months, 46% of the intervention group (68 of 149) were in remission ^[5]. At 24 months the figure was 36%, against 3% in the control group ^[5]. Weight was the driver: among participants who kept off at least 10 kg, 64% were in remission at 2 years ^[5]. At 5 years, 11 of 85 participants who continued in an extension programme of low-intensity support, 13%, were in remission ^[6].

DIADEM-I, in Qatar, tested a similar total diet replacement programme with physical activity support in younger adults, aged 18 to 50, diagnosed within the previous 3 years ^[7]. After 12 months, mean weight loss was 11.98 kg against 3.98 kg with usual care, and the authors report remission in over 60% of the intervention group ^[7]. Its participants were younger and earlier in the disease than DiRECT's.

Both programmes were run by trained staff, began by stopping glucose-lowering drugs under supervision, and selected people early in the disease. A very-low-calorie diet while still taking insulin or a sulfonylurea carries a risk of hypoglycemia, which is why these trials managed medicines alongside the diet; our guide to glucose pills and lows explains that risk.

A 2025 meta-analysis of 18 non-surgical randomised trials found remission more likely with the interventions than with control care, but noted small samples, strict protocols, narrow selection of participants and inconsistent definitions of remission ^[10]. The rates in the table come from selected groups under trial conditions and are not a forecast for any individual.

The longest randomised data come from a single-centre trial in Rome of 60 adults with a BMI of at least 35, diabetes for more than 5 years and HbA1c above 7% ^[8]. Ten years after surgery, 37.5% of surgically treated participants had stayed in remission throughout ^[8]. Rates differed by operation: 50% after biliopancreatic diversion, 25% after Roux-en-Y gastric bypass, and 5.5% with medical therapy ^[8]. Relapse was common: 58.8% of those in remission at 2 years saw hyperglycemia return, although their mean HbA1c at 10 years was 6.7% ^[8]. Surgical participants had fewer diabetes complications, while serious adverse events were more frequent after biliopancreatic diversion ^[8].

The ADA Standards of Care 2026 include metabolic surgery among the treatments for type 2 diabetes in people with obesity, alongside lifestyle and drug therapy for weight management ^[11]. Surgery is a clinical decision with its own risks, made with a specialist team.

Under the consensus definition, glucose that is normal only while a drug is being taken does not count as remission ^[1]. That is the central limit on GLP-1 data. An Italian observational study of 14,141 people starting a GLP-1 receptor agonist, followed for about 4 years, found remission in 5.8% under the strictest definition (no glucose-lowering drugs) and in up to 18.3% when ongoing treatment was ignored ^[12]. Remission was more likely with shorter diabetes duration and higher BMI ^[12].

For tirzepatide, a meta-analysis found it raised the odds of reaching an HbA1c below 5.7%, the normal range, more than 16-fold compared with control treatments ^[13]. Those participants were still on the drug, so this is normoglycemia on treatment rather than remission. The tirzepatide record and semaglutide record hold the trial details.

For the amylin combination CagriSema, the type 2 diabetes trial in people early in the disease, REIMAGINE 1, reported HbA1c falling by 1.8 percentage points over 40 weeks against 0.1 on placebo, from a baseline of 7.8% ^[14]. Its primary endpoint was HbA1c change while on treatment, so remission off treatment was not the question it asked ^[14]. A Europe PMC search in October 2026 found no published trial of cagrilintide or CagriSema reporting remission off treatment. CagriSema is not approved anywhere; see new weight-loss drugs for its status.

What happens when these drugs stop is the open question. A 2026 trial protocol summarises the published evidence that most people regain weight after abruptly stopping semaglutide, with its cardiometabolic benefits reversing ^[15]. Since weight drives remission in the diet and surgery trials ^[5], that regain is the main reason drug-induced normal glucose is not described as a cure.

People searching for an alternative treatment for type 2 diabetes usually mean one of three things: a non-drug route (diet, weight loss, surgery), a different drug, or a supplement. The first two are covered above and in our overview of type 2 diabetes treatments. For supplements, the evidence is thinner.

A 2025 umbrella review of herbal medicines for type 2 diabetes found that ginger, turmeric and the Chinese preparation Jinlida lowered HbA1c compared with placebo by about 0.3 to 0.5 percentage points, with moderate to high certainty for those outcomes ^[16]. It also judged the underlying reviews to be of low to very low quality, said herbal medicines should be used only as an adjunct to conventional treatment, and warned of hypoglycemia when they are combined with diabetes drugs because herb-drug interaction data are lacking ^[16]. The outcome these reviews measured was HbA1c change, not remission ^[16].

For comparison, the diet-based trials above produced remission in a third to over half of participants at 1 to 2 years ^[5][7]. For people with prediabetes rather than diabetes, prevention evidence is on our prediabetes medication page.