Prediabetes Remission Without Weight Loss: What the Data Show

Normal glucose levels may tell you more than the scale

For people with prediabetes, returning to normal glucose regulation predicted a much lower later risk of type 2 diabetes—even among people who did not lose weight or gained weight. That result does not mean weight loss has stopped mattering. It means the scale can miss a clinically relevant change.

The evidence comes from a 2025 Nature Medicine analysis by Sandforth et al. that examined lifestyle-intervention data from the Prediabetes Lifestyle Intervention Study (PLIS) and replicated the finding in the U.S. Diabetes Prevention Program (DPP) repository. The analysis was post hoc: researchers did not randomly assign participants to pursue glucose remission rather than a weight-loss goal. Still, the association was large enough to justify tracking glucose normalization as an outcome in its own right.

There is a practical lesson here. A person can improve fasting glucose, oral-glucose-tolerance results, and HbA1c without seeing a satisfying number on the bathroom scale. Metabolism has several moving parts; body weight is one of them, not the whole dashboard.

What the PLIS and DPP analyses found

Sandforth and colleagues defined prediabetes remission at 12 months as a full return to normal glycemia: fasting plasma glucose below 5.6 mmol/L, 2-hour glucose below 7.8 mmol/L, and HbA1c below 39 mmol/mol. This was not a partial reduction in glucose or a single favorable lab value.

In PLIS, 234 of 1,105 participants had failed to lose weight or had gained weight after 12 months. Yet 51 of those 234 participants achieved remission. During follow-up of up to 10 years, remission in that subgroup was associated with a 71% lower relative risk of incident type 2 diabetes, with a risk ratio of 0.29 (95% CI, 0.09 to 0.91). The confidence interval excludes 1.0, so this association met the conventional threshold for statistical significance. The study did not report an absolute risk difference in the information available here, so the relative-risk reduction should not be mistaken for a guaranteed personal benefit of 71%.

The DPP repository analysis pointed the same way. Among 494 participants who also did not lose weight or gained weight at one year, those who achieved remission had a 73% lower future diabetes risk over a median 4.11 years: risk ratio 0.27 (95% CI, 0.11 to 0.66). Replication in a second dataset makes the signal more persuasive than a single subgroup result, but it still does not establish that targeting remission causes the lower risk.

The paper also compared remission with weight-loss-goal attainment in earlier PLIS- and DPP-derived analyses. Weight-loss-induced remission was associated with a 73% lower future diabetes risk than meeting a weight-loss goal alone. Achieving both remission and the 7% weight-loss target was associated with a 76% lower risk than reaching the weight-loss goal alone. These comparisons support a narrower, useful conclusion: in these datasets, glucose remission contained prognostic information beyond whether someone reached a weight target.

Weight loss remains a useful tool for diabetes prevention. But a weight-loss target alone does not fully describe metabolic improvement or later diabetes risk.

Why glucose can improve while weight does not

The PLIS analysis found metabolic differences between people who achieved remission and those who did not despite similar weight trajectories. Remitters had better insulin sensitivity, early insulin secretion, beta-cell function, beta-cell sensitivity to GLP-1, and glucagon suppression. They also had a more favorable fat-distribution pattern: visceral adipose tissue remained stable, while the ratio of subcutaneous to visceral fat was higher.

These findings fit basic physiology. Glucose regulation depends on how effectively tissues respond to insulin, how well beta cells release insulin, how much glucose the liver produces, and where fat is stored. Scale weight does not directly measure any of those processes. It can be useful, but it cannot inspect the pancreas.

Other remission research supports that physiological framing. A post hoc analysis of 301 participants in the IDEATE trial found that people with higher baseline disposition or adaptation indices were more likely to achieve type 2 diabetes remission. In the CORDIOPREV trial analysis, better beta-cell function and lower hepatic insulin resistance predicted remission and a lower need to begin glucose-lowering medication among people with newly diagnosed type 2 diabetes. These studies concern diabetes rather than prediabetes, so they cannot prove the PLIS result. They do show why remission is likely to reflect metabolic reserve and organ-specific dysfunction, not sheer willpower or a single behavior.

Omics adds a biological clue, not a consumer test

A separate 2025 Diabetes Care cohort study by Barovic et al. offers a reason remission and progression may be more than opposite labels. Investigators studied up to 134 people with new-onset diabetes, prediabetes, or normoglycemia. The cohort included 108 participants who had prediabetes one year earlier. They measured 1,389 proteins and 152 metabolites using fasting samples and oral-glucose-challenge data.

People who progressed to diabetes differed from people who reverted to normoglycemia before their outcomes fully separated. The study reported 14 proteins elevated in progressors versus reverters, and some proteins were associated with insulin-sensitivity indices. People with new-onset diabetes also had higher oral-glucose-test area-under-the-curve measures for IDL-related lipoprotein features and glutamate.

This is biologically interesting, but it is not a biomarker shopping list. The study was observational, followed participants for one year, involved a modest sample, and tested many analytes. It did not test an intervention, establish that any molecule caused progression, or provide enough detail to select a clinically useful consumer panel. A glutamate result on a direct-to-consumer report is not a treatment plan wearing a lab coat.

Prediabetes is not one condition with one response

People classified as having prediabetes do not all have the same mix of impaired fasting glucose, impaired glucose tolerance, insulin resistance, beta-cell dysfunction, or ectopic fat. That heterogeneity limits any universal claim about remission targets.

A systematic review and meta-analysis by Rong et al. found that lifestyle-intervention responses differed across prediabetes phenotypes. A separate study by Wang and colleagues reported that lifestyle intervention was more effective among people with high 1-hour post-load glucose than among people with conventional prediabetes for restoring beta-cell function, reducing ectopic fat, and preventing type 2 diabetes. These results do not contradict the PLIS/DPP finding. They explain why its size and practical meaning may vary by baseline phenotype.

They also argue against treating remission as a detached behavioral target. Glucose normalization may reflect real improvement, but the route to it can differ substantially between people. Diet, physical activity, medication when indicated, sleep, and weight reduction can all matter; the right mix depends on the person's physiology and clinical context.

What this evidence can—and cannot—change

The strongest claim supported by the evidence is straightforward: clinicians and patients should track remission to normoglycemia alongside weight change in prediabetes. The PLIS and DPP analyses found lower later diabetes risk among remitters, including people whose weight did not decline. Fasting glucose, 2-hour oral-glucose-tolerance-test glucose, and HbA1c provide a more direct view of that outcome than body weight alone.

The evidence does not support replacing weight-loss targets with a remission-first doctrine. The anchor analyses were post hoc, the key PLIS subgroup included only 51 remitters among 234 participants without weight loss, and the DPP result came from a repository analysis limited to participants with the required follow-up data. In PLIS, remitters were somewhat younger and numerically more often female, leaving uncertainty about how consistently the finding applies across age, sex, adiposity patterns, and baseline metabolic profiles.

A prespecified randomized trial that compares a remission-targeted strategy with a conventional weight-loss-targeted strategy would answer the central causal question. Until then, the sensible approach is less dramatic and more useful: keep weight loss in the toolkit, measure glucose outcomes directly, and avoid mistaking early omics signals for personalized treatment instructions.