A routine pair of blood tests—one measuring uric acid, the other a cholesterol ratio few patients have ever heard of—may tell very different stories about who is headed toward type 2 diabetes. That is the central finding of a large prospective cohort study published in BMC Endocrine Disorders, which followed more than 8,000 middle-aged and older adults in China for up to seven years and compared how two familiar biomarkers travel along distinct metabolic routes to the same disease endpoint. The results suggest that a simple, inexpensive metric known as the non-high-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio, or NHHR, carries an independent signal for incident diabetes that uric acid largely loses once body weight is taken into account.
The study, conducted by Weiqiang Qin, Bo Liu, Xiaoling Yi, Juan Lei and Lan Mai of Sun Yat-sen Memorial Hospital, Sun Yat-sen University in Guangzhou, drew on the China Health and Retirement Longitudinal Study (CHARLS), a nationally representative cohort of adults aged 45 and older. From the 2011 baseline wave through follow-up waves extending to 2018, the researchers identified 8,150 participants who were free of diabetes at enrollment. After accounting for missing data and attrition, 6,762 individuals remained in the fully adjusted analysis, providing a substantial sample with which to disentangle the effects of two biomarkers that often rise in parallel in metabolic disease.
Uric acid, the end product of purine metabolism, has long been associated with type 2 diabetes in observational research, and elevated serum levels are routinely noted in patients with obesity, hypertension and insulin resistance. NHHR, by contrast, is a derived lipid metric: it is calculated by dividing non-HDL cholesterol—which encompasses all atherogenic cholesterol fractions, including low-density lipoprotein and very-low-density lipoprotein particles—by HDL cholesterol, the fraction generally considered protective. Because it integrates both the “bad” and “good” sides of the cholesterol ledger in a single number, NHHR has attracted growing attention in recent years as a marker of atherogenic dyslipidemia, the lipid pattern closely tied to metabolic syndrome.
To compare the two markers fairly, the team employed Cox proportional hazards regression across four models of increasing adjustment, starting with crude associations and progressively controlling for age, sex, and an expanding battery of demographic, behavioral and clinical covariates. They additionally used restricted cubic splines to characterize dose-response relationships, allowing the data to reveal whether risk rises linearly or in threshold-like fashion with each biomarker. The analytical centerpiece, however, was a mediation analysis conducted under VanderWeele’s counterfactual framework, which partitioned each biomarker’s total association with incident diabetes into an indirect component channeled through body mass index and a direct component operating by other means.
The verdict was strikingly asymmetric. In the fully adjusted model, each 1-unit increase in NHHR was independently associated with incident type 2 diabetes, yielding a hazard ratio of 1.104 with a 95 percent confidence interval of 1.050 to 1.161, a result that held firm at P less than 0.001. Uric acid, by contrast, failed to maintain an independent association once covariates were fully accounted for: its hazard ratio of 1.048, with a confidence interval of 0.978 to 1.122, was statistically indistinguishable from no effect, with a P value of 0.184. In other words, when two people share the same age, sex, weight, blood pressure, lifestyle and other measured characteristics, the one with the higher cholesterol ratio still faces measurably greater diabetes risk, while the one with the higher uric acid level does not—at least not through any route the analysis could detect.
The mediation analysis went further, revealing why the two markers behave so differently. For NHHR, only 22.7 percent of the total association with incident diabetes was mediated through BMI, meaning that 77.3 percent of its effect operated independently of body weight. Uric acid displayed nearly the mirror image: 51.4 percent of its association was channeled through BMI, and whatever direct effect remained after accounting for that pathway was not statistically significant. The implication is that uric acid’s apparent link to diabetes may largely be a passenger of adiposity—rising and falling with body mass—whereas the cholesterol ratio tracks a more metabolically intrinsic pathway, one that persists even in individuals whose weight appears normal.
The researchers also examined joint exposure categories, and here again NHHR emerged as the more decisive signal. Participants with high uric acid but low NHHR—an intuitively worrisome combination—showed an incidence rate comparable to the low-risk reference group. Across the combined categories, elevated NHHR was the more prominent risk marker, suggesting that clinicians weighing the significance of an elevated uric acid reading might do better to check the patient’s cholesterol ratio before drawing alarmist conclusions.
Dyslipidemia emerged as a significant effect modifier in the relationship between NHHR and diabetes, with an interaction P value of 0.015. This indicates that the strength of the association between the cholesterol ratio and incident diabetes is not uniform across the population: in people with clinically diagnosed lipid abnormalities, the pathway linking NHHR to diabetes appears to operate differently, or with different intensity, than in those with normal lipid profiles. The finding adds nuance to how the metric should be interpreted and flags a subgroup—patients with dyslipidemia—in which the ratio may deserve heightened attention.
Yet the study’s authors are careful not to oversell NHHR as a clinical game-changer. When the team assessed the incremental predictive value of adding NHHR to an established base model, the improvement in discrimination was marginal: the concordance index, or C-statistic, rose by only 0.005, a change that fell short of conventional statistical significance at P equal to 0.091. Decision curve analysis, which evaluates the net benefit of acting on a model’s predictions across a range of risk thresholds, showed only limited added value, with a net benefit increment of 0.0049 at the prevalence threshold of 11.6 percent. Net reclassification improvement and integrated discrimination improvement metrics likewise painted a picture of modest, rather than transformative, gains.
The authors’ conclusion reflects this tension between biological interest and clinical pragmatism. NHHR may complement uric acid assessment in raising risk awareness among metabolically abnormal individuals with normal BMI—a population that conventional weight-based screening can miss—but its incremental predictive value over existing models is limited, and clinical implementation, they stress, requires validation in independent cohorts before any change in practice could be justified. This is a study about pathway profiles as much as prediction: its most durable contribution may be the demonstration that two biomarkers long treated as interchangeable components of “metabolic risk” in fact reach the same destination by different roads.
The findings arrive amid a broader wave of interest in NHHR. Recent analyses of the NHANES dataset in the United States have linked the ratio to prediabetes progression, to type 2 diabetes prevalence, and even to all-cause and cardiovascular mortality among adults with diabetes or prediabetes, with BMI frequently appearing as a mediating variable. The new CHARLS analysis distinguishes itself by placing NHHR and uric acid head to head within a single cohort and formally quantifying the BMI-mediated share of each pathway—an approach that allows the two markers’ biology to be compared directly rather than inferred from separate studies.
For the public, the practical takeaway is measured. High uric acid, long a source of anxiety for patients who associate it with gout and, increasingly, with diabetes warnings, may in many cases simply reflect body weight—and weight reduction would be expected to lower both the uric acid level and the associated risk. The cholesterol ratio, meanwhile, appears to flag a risk that travels independently of the scale, which means a person of normal weight with an unfavorable lipid profile is not necessarily in the clear. Both markers can be obtained from a standard lipid panel and a basic metabolic blood test, costing a fraction of what newer diabetes risk assays demand.
The study also carries implications for how researchers think about mediation in chronic disease epidemiology. By applying a counterfactual framework to a prospective cohort rather than a cross-sectional snapshot, the authors were able to estimate not merely whether BMI correlates with both biomarker and outcome, but how much of the biomarker’s temporal effect plausibly flows through it. The finding that roughly three-quarters of NHHR’s effect bypasses BMI invites mechanistic work into what those non-adiposity pathways might be—candidate mechanisms include the direct effects of atherogenic lipoprotein burden on pancreatic beta-cell function, hepatic insulin sensitivity, and chronic low-grade inflammation, all of which have been implicated in prior lipid-diabetes research.
Limitations remain, as they do in all observational work. CHARLS measures were largely self-reported or field-collected rather than drawn from standardized clinical examinations, incident diabetes was ascertained over waves rather than continuously, and residual confounding by unmeasured diet, physical activity and genetic factors cannot be excluded. The cohort is Chinese and middle-aged to elderly, and whether the same pathway partition holds in younger or non-Chinese populations is unknown. The authors note that the analysis is exploratory in its mediation component and that independent validation is essential.
Still, in an era when type 2 diabetes affects hundreds of millions of people worldwide and prevention depends on identifying risk early and cheaply, a study that clarifies which of two pennies-and-blood-work biomarkers carries genuinely independent information is far from trivial. The message of this cohort is ultimately one of refinement: uric acid’s warning is often obesity’s warning in disguise, while the ratio of non-HDL to HDL cholesterol speaks in its own metabolic voice—one that researchers, and perhaps eventually clinicians, would do well to keep listening to.
Subject of Research: Comparative prospective analysis of serum uric acid and the non-HDL cholesterol to HDL cholesterol ratio (NHHR) as predictors of incident type 2 diabetes, including BMI-mediated pathway profiles, in the CHARLS cohort
Subject of Research: Medicine
Article Title: Differential pathway profiles of uric acid and the non-high-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio (NHHR) in predicting incident type 2 diabetes: a prospective cohort study
Article References: Qin, W., Liu, B., Yi, X., Lei, J., & Mai, L. (2026). Differential pathway profiles of uric acid and the non-high-density lipoprotein cholesterol to high-density lipoprotein cholesterol ratio (NHHR) in predicting incident type 2 diabetes: a prospective cohort study. BMC Endocrine Disorders. https://doi.org/10.1186/s12902-026-02489-3
Image Credits: AI Generated
DOI: 10.1186/s12902-026-02489-3
Keywords: uric acid, NHHR, type 2 diabetes mellitus, mediation analysis, CHARLS, BMI, dyslipidemia, Cox regression, risk prediction, atherogenic dyslipidemia
Cite Scienmag News
APA
MLA
Chicago
Phoebe Ingram. (September 11, 2026). Uric acid and cholesterol ratio show different pathways to diabetes risk. Scienmag. https://scienmag.com/uric-acid-and-cholesterol-ratio-show-different-pathways-to-diabetes-risk/
Phoebe Ingram. “Uric acid and cholesterol ratio show different pathways to diabetes risk.” Scienmag, 11 September 2026, https://scienmag.com/uric-acid-and-cholesterol-ratio-show-different-pathways-to-diabetes-risk/. Accessed 11 September 2026.
Phoebe Ingram. “Uric acid and cholesterol ratio show different pathways to diabetes risk.” Scienmag. September 11, 2026. https://scienmag.com/uric-acid-and-cholesterol-ratio-show-different-pathways-to-diabetes-risk/
Copy citation
Download RIS
Tags: biomarkers for type 2 diabetes predictionbody weight and diabetes predictionChina Health and Retirement Longitudinal StudyChinese population study oncholesterol ratio as an independent diabetes markercholesterol ratios and metabolic healthdiabetes risk assessment in middle-aged adultsdistinct metabolic pathways to type 2 diabetesimpact of body weight on uric acid and cholesterol indicatorsinexpensive biomarkers for diabetes predictioninfluence of blood lipids on diabetes riskinfluence of uric acid on diabetes risklongitudinal analysis of middle-aged and older adultsmetabolic pathways to diabetes developmentNHHR as a diabetes biomarkernon-HDL to HDL cholesterol rationon-HDL to HDL cholesterol ratio significancepathways to diabetes riskprospective cohort study on diabetes biomarkersprospective cohort study on metabolic markersrole of uric acid in metabolic healthsignificance of inexpensive blood tests in diabetes screeningUric acid and cholesterol ratioUric acid and cholesterol ratio in diabetes risk
