Losing weight is the central recommendation for anyone told they have prediabetes. A German analysis reports that for one identifiable subgroup, it did not work.
People classified in what researchers call Tübingen type 2 diabetes risk cluster 5 lost roughly 8 percent of their body weight and kept it off for years. Over that period, their blood glucose still deteriorated, their insulin secretion fell more steeply than any other cluster, and their risk of developing type 2 diabetes stayed high.
The work comes from the German Center for Diabetes Research, University Hospital Tübingen, and Helmholtz Munich, and was published in the journal Diabetes. It draws on the Tübingen Lifestyle Intervention Program.
One timing note before anything else: this research was published and announced in April. It resurfaced this week through an aggregator summary. The findings are not new as of this week, and treating them as breaking news would misrepresent them.
What the Clusters Are
The framework matters, because "cluster 5" means nothing without it.
In 2021, researchers at these institutions reported in Nature Medicine that prediabetes is not one condition. Using detailed metabolic phenotyping, they identified six distinguishable subtypes that differ in how often diabetes develops and how complications progress.
Clusters 1, 2, and 4 carry low risk. Clusters 3, 5, and 6 carry elevated risk, with clusters 3 and 5 the highest for progression to type 2 diabetes. Cluster 5 in particular has been associated with higher risk not only of diabetes but of atherosclerotic cardiovascular disease and mortality.
Assignment to a cluster is not exotic laboratory work. According to Medscape's report of the study, classification uses sex, body mass index, waist and hip circumference, fasting glucose, two-hour oral glucose tolerance test values, triglycerides, and HDL cholesterol. The criteria were derived from the Tübingen cohorts and later replicated in the London-based Whitehall II study.
The mechanisms behind cluster 5 have been described in earlier work as fatty liver disease and insulin resistance, which the researchers say makes this group particularly susceptible to both diabetes and cardiovascular disease.
What This Analysis Did and Did Not Do
This is a reanalysis of participants in an existing lifestyle intervention program, grouped retrospectively by cluster. It is not a new randomized trial comparing interventions in cluster 5 against anything else.
That distinction sets the ceiling on what can be concluded. Subgroup analyses of existing cohorts generate hypotheses. They are vulnerable to small sample sizes within each subgroup, to differences between groups that were never randomized away, and to the possibility that an apparent pattern reflects chance across multiple comparisons.
The researchers are explicit about this themselves. Their stated position is that if the findings are confirmed in a prospective study, a more tailored approach to diabetes prevention would be required, with high-risk phenotypes, such as cluster 5, potentially requiring more intensive or targeted interventions.
One figure that should accompany any responsible account of this study is the number of participants who fell into cluster 5. That number is not stated in the institutional announcement or the secondary coverage reviewed for this article, and it determines how much confidence the subgroup finding can carry. It should be taken from the paper before publication.
Lead author Norbert Stefan, professor of clinical and experimental diabetology at the University of Tübingen and head of the department of pathophysiology of prediabetes at Helmholtz Munich, told Medscape that "alternative or intensified interventions should be considered for people in Tübingen T2D risk cluster 5."
He also framed the practical implication as knowing a person's risk profile before starting a lifestyle intervention, rather than after it has failed.
What This Does Not Mean for Someone with Prediabetes
Nothing here is a reason to stop losing weight, and it should not be read that way.
Weight loss and increased physical activity remain effective in preventing type 2 diabetes across the population, and the same analysis found that other clusters also benefited. The finding is that benefit is not uniform, not that it is absent.
Weight loss also produces benefits that this analysis did not measure, including effects on blood pressure, lipids, joint symptoms, sleep apnea, and cardiovascular risk. A person in cluster 5 who loses weight is not wasting the effort even if their glucose trajectory does not change.
Cluster assignment is also not something available at a routine appointment in the United States. This is a research classification, not a clinical test you can request. Some of the inputs, such as a two-hour oral glucose tolerance test, are not part of standard prediabetes screening in US practice.
What a person with prediabetes can reasonably do is ask their clinician whether their blood sugar is actually improving with the changes they have made, rather than assuming it must be. If HbA1c or fasting glucose is rising despite sustained weight loss and activity, that is worth naming as a specific question rather than being told to keep going. Whether medication should be added is a decision for a clinician, and no one should start, stop, or change a medication based on a news article.
What Would Confirm It
The researchers have named the next step themselves: a prospective study that assigns cluster membership in advance and then tests whether tailored or intensified interventions produce different outcomes in cluster 5 than standard lifestyle advice.
Until that exists, this retrospective subgroup analysis, published in April, suggests that a widely recommended approach may not work equally well for everyone who receives it.
The confirmed finding is that in this analysis, one high-risk prediabetes subgroup showed worsening glucose and the steepest decline in insulin secretion despite sustained weight loss of about 8 percent. The people this may eventually matter to are those with prediabetes whose numbers are not improving despite doing what they were told. The most reasonable action is to ask whether your own glucose measures are actually responding. The central uncertainty is whether the pattern holds prospectively, which no one yet knows.
Frequently Asked Questions
What did the researchers find? People in Tübingen type 2 diabetes risk cluster 5 had worsening blood glucose, the steepest decline in insulin secretion among the clusters, and continued high diabetes risk despite sustained weight loss of about 8 percent.
What are the prediabetes clusters? Six subtypes of prediabetes were identified in 2021 that differ in diabetes risk and complication patterns. Clusters 1, 2, and 4 are lower risk; clusters 3, 5, and 6 are higher.
How is cluster 5 identified? Through sex, body mass index, waist and hip circumference, fasting glucose, two-hour oral glucose tolerance test results, triglycerides, and HDL cholesterol.
Can I find out my cluster? Not through routine care in the United States. This is a research classification, and some inputs such as the two-hour glucose tolerance test are not standard in US prediabetes screening.
Is this a new clinical trial? No. It is a reanalysis of participants in an existing lifestyle intervention program, grouped retrospectively by cluster. The researchers say prospective confirmation is needed.
Should I stop trying to lose weight? No. Other clusters did benefit, and weight loss carries benefits beyond glucose. The finding is that benefit is uneven, not absent.
What should I ask my doctor? Whether your blood sugar measures are actually improving with the changes you have made, and what the plan is if they are not.
