Prediabetes sits on the metabolic fault line between normal glucose metabolism and type 2 diabetes. It is defined by blood sugar levels that are elevated above the normal range but fall short of the diabetic threshold, and it affects hundreds of millions of adults worldwide. Insulin resistance, the progressive failure of tissues to respond efficiently to the hormone insulin, is considered a central engine driving the transition from prediabetes to diabetes. Yet measuring insulin sensitivity directly is impractical in routine care. The gold-standard technique, the hyperinsulinemic-euglycemic clamp, requires a prolonged intravenous infusion and repeated blood sampling, making it suitable only for specialized research settings. The eGDR was developed as a practical alternative: a formula that estimates how quickly the body disposes of glucose using variables that are already collected in almost every clinical encounter, including waist circumference as a proxy for abdominal adiposity, hypertension status, and levels of glycated hemoglobin, the long-term marker of blood glucose control.

In the new study, a team of researchers led by Ying Yang and Rong Li of Nanchang Medical College turned to the China Health and Retirement Longitudinal Study, a nationally representative survey of middle-aged and older Chinese adults, to ask whether baseline eGDR could forecast glycemic trajectories. The investigators focused on 2,630 adults who met criteria for prediabetes at the start of the study. Four years later, the participants’ metabolic fates had diverged in the way that prediabetes outcomes typically do: 1,631 remained in the prediabetic range, 382 had progressed to diabetes, and 617 had reverted to normal fasting glucose, the healthiest possible outcome. This three-way split allowed the researchers to model not only who got worse but also who got better, a dual perspective that most risk studies neglect.

Because the study’s central hypothesis concerned sex differences, the team analyzed men and women separately using multinomial logistic regression, a statistical framework suited to outcomes with more than two categories. The results were strikingly asymmetric. Among women, each unit increase in eGDR, reflecting better insulin sensitivity, was associated with a 16 percent lower relative risk of progressing to diabetes, with a relative risk ratio of 0.837 and a 95 percent confidence interval of 0.775 to 0.906. The same score was also linked to a higher likelihood of reverting to normal glucose, with a relative risk ratio of 1.099 and a confidence interval of 1.018 to 1.187. In other words, women whose bodies disposed of glucose more efficiently were both less likely to cross the diabetic threshold and more likely to climb back to metabolic health.

To probe the shape of these relationships, the researchers employed restricted cubic splines, a flexible modeling technique that allows the data to reveal whether an association is linear or follows a more complicated curve. The spline models indicated that the relationships between eGDR and both diabetes progression and reversion to normoglycemia were essentially linear in women, meaning that the prognostic benefit accrued steadily across the range of the score rather than kicking in only above some threshold. That linearity is clinically convenient, because it implies that even modest improvements in insulin sensitivity, as captured by the score, may shift a woman’s odds in the right direction. Among men, by contrast, neither the regression models nor the spline analyses uncovered any meaningful association between eGDR and glycemic outcomes, a null result that stood in sharp contrast to the findings in women.

Why would the same metabolic marker carry prognostic weight in one sex but not the other? The researchers offered a partial answer by examining whether blood lipids mediated the associations. Mediation analysis is a statistical method for decomposing an observed relationship into the portion that flows through an intermediate variable and the portion that operates through other pathways. In women, the lipid fractions explained a modest but measurable share of the link between eGDR and glycemic change. For progression to diabetes, high-density lipoprotein cholesterol and triglycerides accounted for statistically mediated proportions of 9.48 percent and 4.41 percent, respectively. For reversion to normal fasting glucose, low-density lipoprotein cholesterol and total cholesterol mediated 5.56 percent and 7.22 percent of the association. These figures suggest that insulin sensitivity and lipid metabolism travel together along at least part of the road from prediabetes to diabetes, particularly in women, although the bulk of the association remains unexplained by lipids alone.

The authors themselves urge caution in interpreting the sex difference. The formal statistical test for interaction between eGDR and sex yielded a P value of 0.071, just above the conventional threshold of 0.05, meaning the evidence for a true sex-specific effect is suggestive rather than definitive. Several other caveats apply. The cohort consisted of Chinese adults, and whether the findings generalize to other populations remains to be seen. The eGDR is an estimate, not a direct measurement, and it inherits the limitations of its component variables, including the crudeness of hypertension status as a stand-in for vascular effects. Residual confounding by diet, physical activity, medication use and other lifestyle factors cannot be excluded in an observational cohort, however carefully the models are adjusted.

Even with those qualifications, the study adds to a growing body of evidence that metabolic risk does not affect men and women identically, and that prognostic tools validated in one sex may not transfer cleanly to the other. Sex-specific differences in fat distribution, hormonal profiles, lipid metabolism and insulin action are well documented, and clinical guidelines increasingly recognize that cardiovascular risk calculators, for example, perform differently across the sexes. If the eGDR findings are replicated in independent cohorts, the score could offer an inexpensive way to flag women with prediabetes who warrant closer monitoring or earlier intervention, using nothing more than data already gathered in a routine checkup.

For now, the practical message is one of cautious promise rather than immediate practice change. The study demonstrates that a simple, zero-cost index of insulin sensitivity tracks the two most important glycemic outcomes in prediabetic women over a four-year horizon, and it hints that lipid fractions form one thread connecting insulin resistance to disease progression. Larger, longer and more diverse cohorts will be needed to confirm the sex-specific prognostic value of eGDR, to test whether the score improves care beyond existing risk factors, and to determine whether interventions that raise eGDR, such as weight loss and exercise, translate into more frequent reversion to normal glucose. In a condition as heterogeneous and reversible as prediabetes, tools that separate the trajectories of those who will progress from those who will recover are precisely what prevention efforts need most.

Subject of Research: Sex-specific prognostic value of the estimated glucose disposal rate for prediabetes outcomes

Article Title: The prognostic value of eGDR in prediabetes outcomes: a sex-stratified prospective cohort study

Article References: Yang, Y., Yang, J., Du, Y., Fan, Q., Fang, H., Zhang, H., You, J., Yu, Q., Lei, J., & Li, R. (2026). The prognostic value of eGDR in prediabetes outcomes: a sex-stratified prospective cohort study. BMC Endocrine Disorders. https://doi.org/10.1186/s12902-026-02560-z

Image Credits: AI Generated

DOI: 10.1186/s12902-026-02560-z

Keywords: prediabetes, estimated glucose disposal rate, insulin resistance, type 2 diabetes, sex differences, CHARLS, lipid mediation, normoglycemia, epidemiology, prospective cohort, endocrinology, risk prediction