The researchers analyzed patients admitted to the Department of Endocrinology at the First Affiliated Hospital of Harbin Medical University in Harbin, China, from January through December 2024. Hyperuricemia was defined as a serum uric acid concentration of at least 360 μmol/L in women or 420 μmol/L in men. Of the 959 participants, 259, or 27.0%, met those criteria. Patients with hyperuricemia had higher GLM7 values than those without it, along with higher triglycerides, fasting insulin, body mass index, total cholesterol and serum creatinine; they also had lower estimated glomerular filtration rate. The groups did not differ significantly in fasting blood glucose or glycated hemoglobin, making the overall association more informative than a simple comparison of glucose control.

GLM7 is calculated as the base-10 logarithm of the product of age, body mass index, fasting blood glucose, insulin, triglycerides, low-density lipoprotein cholesterol and the inverse of high-density lipoprotein cholesterol, using specified units. Its construction means that a rising value can reflect several metabolic characteristics simultaneously rather than one isolated measurement. When the researchers divided the index into quartiles, the third and fourth groups had higher adjusted odds of hyperuricemia than the lowest group. The highest quartile had 2.04 times the odds of the lowest quartile (95% confidence interval, 1.28–3.26; P = 0.003), while the trend across quartiles was statistically significant. A one-standard-deviation increase in GLM7 was associated with 1.38 times the odds of hyperuricemia (95% confidence interval, 1.17–1.64; P < 0.001).

The same pattern appeared when the outcome was measured as a continuous serum uric acid concentration. In the fully adjusted linear model, a one-unit increase in GLM7 was associated with serum uric acid being 37.88 μmol/L higher (95% confidence interval, 28.22–47.55; P < 0.001). The estimated increase associated with one standard deviation of GLM7 was 22.60 μmol/L (95% confidence interval, 16.83–28.36; P < 0.001). Compared with the lowest GLM7 quartile, the highest quartile had serum uric acid concentrations averaging 57.07 μmol/L higher (95% confidence interval, 40.74–73.40; P < 0.001). Restricted cubic spline analysis found a significant overall relationship but no evidence of a nonlinear association across the observed range (P for nonlinearity = 0.656), supporting an approximately linear pattern in this dataset rather than proving a biological dose-response relationship.

Subgroup analyses suggested that the association was directionally consistent among strata defined by age, sex, smoking, alcohol consumption, hypertension and body mass index, with no statistically significant interaction detected. Results also remained similar in sensitivity models that included either serum creatinine or estimated glomerular filtration rate separately, rather than together. In those analyses, the odds ratio per one-unit increase in GLM7 was 1.760 when creatinine was included and 1.813 when estimated glomerular filtration rate was included, both with P < 0.001. These checks suggest that the observed association was not substantially explained by the two correlated measures of kidney function, although the study did not directly measure tubular urate transport, urinary urate excretion, intestinal excretion or hepatic urate production.

Despite the strength of the statistical association, GLM7 performed weakly on its own as an identifier of hyperuricemia. Its area under the receiver operating characteristic curve was 0.599 (95% confidence interval, 0.559–0.640), only slightly above the value of 0.50 that corresponds to chance-level discrimination. The reported optimal cutoff was 7.959, with 63.3% sensitivity and 55.0% specificity, meaning a positive result would miss a substantial share of cases and incorrectly flag a similar share of people without hyperuricemia. GLM7 did not significantly outperform the triglyceride-glucose index, triglycerides or body mass index in these comparisons, so its larger collection of inputs did not provide superior stand-alone classification in this hospitalized population.

Adding GLM7 to a conventional clinical model also failed to significantly improve discrimination. The clinical model included age, sex, body mass index, smoking, alcohol consumption, hypertension, glycated hemoglobin, serum creatinine, liver enzymes and estimated glomerular filtration rate; its area under the curve was 0.754 (95% confidence interval, 0.719–0.789). Adding GLM7 raised the value only to 0.757 (95% confidence interval, 0.723–0.792), a difference that was not statistically significant by the DeLong test (P = 0.438). A likelihood-ratio test did indicate better model fit after GLM7 was included (χ² = 5.70; P = 0.017), but that improvement did not translate into a statistically significant increase in discriminatory performance. The findings therefore do not validate GLM7 for diagnosis, screening or risk stratification of hyperuricemia in patients with type 2 diabetes.

Several design features limit how far the findings can be generalized. The study was retrospective and cross-sectional, so it could not establish whether metabolic abnormalities preceded high uric acid or whether uric acid and related conditions contributed to the metabolic profile captured by GLM7; reverse causation therefore remains possible. All participants came from one tertiary hospital and had established type 2 diabetes, making them more metabolically complex than many people with diabetes managed in community settings. The cohort was also predominantly male, although sex was included in the adjusted analysis. Detailed information on purine intake, fructose consumption, physical activity and actual alcohol intake was unavailable, leaving possible residual confounding. Moreover, fasting glucose is itself part of GLM7, so values and thresholds derived in general populations may not transfer directly to people with diabetes.

For now, the most defensible interpretation is that GLM7 is a composite marker of the overall glucolipid metabolic profile among hospitalized patients with type 2 diabetes, and higher values are associated with both elevated serum uric acid and hyperuricemia. That association survived extensive statistical adjustment and subgroup testing, yet the index lacked enough accuracy to replace direct measurement of serum uric acid or add clearly useful discriminatory information to an existing clinical model. Prospective, multicenter studies involving broader diabetic populations, external validation of any proposed thresholds, and direct measurements of urate production and excretion would be needed before the marker’s clinical value or the biological processes behind the association could be established.

Subject of Research: Association of the glucose-lipid metabolism 7-factor index with hyperuricemia in hospitalized patients with type 2 diabetes

Article Title: Association between the glucose-lipid metabolism 7-factor index and hyperuricemia in patients with type 2 diabetes: a retrospective study

Article References: Zheng, Z., Li, Q., & Zhang, H. (2026). Association between the glucose-lipid metabolism 7-factor index and hyperuricemia in patients with type 2 diabetes: a retrospective study. Frontiers in Endocrinology, 17, Article 1961640. https://doi.org/10.3389/fendo.2026.1961640

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1961640

Keywords: GLM7, hyperuricemia, type 2 diabetes, serum uric acid, metabolic index, insulin resistance, dyslipidemia, retrospective study, diabetes, Association, between, glucose-lipid