Bites of Bio

Why insulin resistance can develop before blood sugar rises

Share



Hospital corpsman prepares clinical blood sample collection equipment in a medical setting.
Credit: Credit: U.S. Marine Corps / Lance Cpl. Afton Smiley, via Unsplash

Insulin resistance can emerge while blood glucose remains within the normal range, and a study published November 16, 2026, in Nature Metabolism identified circulating protein patterns that could help characterize this earlier metabolic dysfunction. Researchers combined laboratory measurements of insulin sensitivity in 161 people with blood-protein profiling, then examined whether a 13-protein prediction model tracked future disease risk in 20,612 UK Biobank participants. The findings point toward more detailed metabolic risk assessment, not a new test ready for routine clinical use.

Normal glucose readings can coexist with reduced insulin sensitivity

Insulin allows tissues to respond to circulating glucose after a meal. Skeletal muscle normally takes up much of that glucose, while insulin suppresses glucose production by the liver and influences how fat cells store and release energy. In insulin resistance, these actions become less effective. The pancreas can initially compensate by secreting more insulin, maintaining glucose concentrations that appear normal on routine tests.

As compensation becomes insufficient, fasting glucose or longer-term glycemic exposure may rise into the prediabetes range and eventually meet criteria for type 2 diabetes. The timing varies across individuals. Insulin resistance is not synonymous with prediabetes, and a normal hemoglobin A1c (HbA1c) result does not directly establish normal insulin sensitivity.

These differences reflect dysfunction across several tissues, not one pathway. Recent research on Miro1 and muscle insulin resistance examines one potential molecular contributor, while studies of visceral fat and metabolic risk illustrate why body-fat distribution also matters.

The Nature Metabolism researchers measured insulin sensitivity directly

Jeppe Kjærgaard, Nigel Kurgan and colleagues examined 161 participants across three research cohorts. Insulin sensitivity was assessed using the hyperinsulinemic-euglycemic clamp, a research procedure in which insulin is infused and glucose is adjusted to maintain a stable blood glucose concentration. The amount of glucose needed provides a direct measure of whole-body insulin-stimulated glucose disposal.

The investigators combined several mass spectrometry methods with affinity-based protein measurements, identifying more than 8,000 distinct circulating proteins. They found 488 proteins associated with clamp-measured insulin sensitivity in analyses adjusted for sex. After additional adjustment for age and body-fat percentage, 62 proteins remained significantly associated.

Signals were linked to liver, fat tissue, muscle and immune biology. This reinforces the concept that insulin resistance is a multiorgan process. However, assigning a circulating protein to a tissue using reference expression data does not establish that the organ caused the observed change.

A 13-protein model predicted insulin sensitivity better than clinical measures alone

The researchers developed a statistical model using 13 proteins, including insulin-like growth factor binding protein 1 (IGFBP1), leptin (LEP), growth differentiation factor 15 (GDF15), paraoxonase 3 (PON3) and the low-density lipoprotein receptor (LDLR). They combined the protein measurements with sex, HbA1c and the ratio of triglycerides to high-density lipoprotein cholesterol.

Across the clamp cohorts, the combined model explained 67% of the variation in measured insulin sensitivity during cross-validation (R2 = 0.67). In a separate validation cohort, it explained 73% (R2 = 0.73), compared with 49% for a model using clinical measurements alone.

Applied to 20,612 UK Biobank participants, the model also showed stronger discrimination for future type 2 diabetes than the triglyceride-to-high-density lipoprotein cholesterol ratio alone, with concordance indices of 0.86 and 0.74, respectively. A concordance index measures how well a model ranks people by subsequent risk; it does not mean the test predicts diabetes correctly in 86% of patients, nor does it establish benefit from using the model in clinical care.

Exercise shifted some protein markers toward a more insulin-sensitive pattern

One research cohort included 43 men who completed an eight-week high-intensity interval training program. The study reported a 30–40% improvement in clamp-measured insulin sensitivity following training. Twenty-two circulating proteins changed significantly, and a broader set of protein signals shifted in the direction associated with greater insulin sensitivity.

These observations suggest the protein signatures are not simply fixed markers of inherited risk. Still, the exercise analysis was relatively small and included men only. It did not demonstrate that monitoring this 13-protein panel improves exercise prescriptions or prevents diabetes.

Today’s glucose tests assess glycemia, not insulin sensitivity itself

The American Diabetes Association (ADA) recommends established blood tests to screen for prediabetes and diabetes. For nonpregnant adults, prediabetes is defined by an HbA1c of 5.7–6.4%, a fasting plasma glucose of 100–125 mg/dL, or a two-hour oral glucose tolerance test result of 140–199 mg/dL. These tests measure glucose exposure or glucose handling rather than directly quantifying how strongly tissues respond to insulin.

HbA1c reflects average glucose exposure over recent months. Fasting glucose measures glucose at a single point in time. An oral glucose tolerance test can reveal impaired glucose handling that a fasting result does not show. The tests are clinically useful, but their normal ranges cannot exclude early insulin resistance. The ADA also states that there is insufficient evidence to use continuous glucose monitoring to screen for or diagnose prediabetes or diabetes.

Routine fasting-insulin calculations, including the homeostatic model assessment of insulin resistance (HOMA-IR), are used in research and some clinical contexts, but are not interchangeable with clamp-derived whole-body insulin sensitivity. The new study compared the protein-based approach with several such surrogate indices; it did not replace current diagnostic criteria.

Clinical risk assessment still depends on established factors

Factors associated with elevated diabetes risk include excess abdominal fat, physical inactivity, a family history of type 2 diabetes, previous gestational diabetes, high triglycerides, low high-density lipoprotein cholesterol, hypertension and polycystic ovary syndrome. Some people have insulin resistance without obvious symptoms or excess body weight. None of these features, alone, confirms insulin resistance.

ADA guidance recommends diabetes screening for most adults beginning at age 35, with earlier testing when risk factors are present. For people at higher risk, structured lifestyle programs emphasize an individualized eating pattern, regular physical activity and weight management where appropriate. The 2026 recommendations include at least 150 minutes of moderate-intensity physical activity per week and, for adults with overweight or obesity at high risk, a sustained 5–7% reduction in starting weight.

The new molecular findings do not change those recommendations. A person concerned about early metabolic risk can discuss their family history, blood pressure, cholesterol, weight pattern and appropriate glucose testing with a clinician rather than seek an unvalidated protein panel.

What the evidence cannot yet answer

The clamp cohorts were modest in size, and the study populations were primarily of European ancestry. Some tissue-source assignments were inferred from gene-expression reference databases rather than measured directly in each organ. Medication effects were not systematically investigated. The exercise analyses were exploratory, and the protein panel was constrained by the assays available in the large validation datasets.

Most importantly, the study tested associations and prediction performance, not a clinical screening strategy. It did not randomize patients to receive protein-guided care or show that adding the panel to existing practice reduces diabetes, heart attacks or other complications. The authors declared no competing interests and reported support from the Novo Nordisk Foundation, Lundbeck Foundation and other research funders; they stated that funders did not direct the study.

The researchers conclude that their work demonstrates the feasibility of capturing the molecular architecture of insulin resistance from blood samples. Whether such information improves clinical decisions before glucose abnormalities become apparent remains a question for larger, more diverse prospective studies.

References

  1. Kjærgaard J, Kurgan N, Gräf JF, et al. Plasma proteomics identifies insulin resistance signatures predictive of cardiometabolic disease risk. Nature Metabolism. Published November 16, 2026. DOI: 10.1038/s42255-026-01662-1.
  2. American Diabetes Association Professional Practice Committee. 2. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes 2026. Diabetes Care. 2026;49(Suppl 1):S27. ADA guideline.
  3. American Diabetes Association Professional Practice Committee. 3. Prevention or Delay of Diabetes and Associated Comorbidities: Standards of Care in Diabetes 2026. Diabetes Care. 2026;49(Suppl 1):S50. DOI: 10.2337/dc26-S003.

Leave a Comment

Your email address will not be published. Required fields are marked *