Simple Blood Ratio Flags Hidden Metabolic Risk in Adults with Obesity

A single calculation performed on two routine blood tests may help clinicians spot which patients with obesity are quietly developing the cluster of metabolic disturbances that precedes type 2 diabetes and cardiovascular disease. In a large cross-sectional study conducted at Kayseri City Hospital in Turkey, researchers found that the uric acid-to-HDL cholesterol ratio, known by the abbreviation UHR, was strongly associated with a study-defined dysmetabolic phenotype in adults with a body mass index of 30 kg/m² or higher. The work, published in BMC Endocrine Disorders, adds to a growing body of literature exploring whether cheap, widely available laboratory markers can be combined into composite indices that outperform their individual components. Yet the study also delivers a sobering caveat: while the ratio performed respectably on its own, it added only a sliver of diagnostic power beyond HDL cholesterol itself, a finding that tempers enthusiasm for adopting it as a standalone screening tool.

The research team, led by İfakat İrem Bicer and colleagues in the hospital’s Department of Internal Medicine, analyzed records from 2,453 adults with obesity who presented to the hospital between March 2025 and March 2026. The cohort was overwhelmingly female, at 81.5 percent, reflecting both the demographics of obesity care in the region and the referral patterns of the endocrine and internal medicine services. Rather than relying on a generic definition of metabolic syndrome, the investigators constructed a deliberately stringent phenotype, labeled Group 2, which required central obesity together with reduced HDL cholesterol accompanied by either elevated triglycerides or impaired glucose metabolism, including diabetes. Everyone who did not meet those criteria was placed in Group 1. Crucially, the authors stipulated that reduced HDL cholesterol alone was insufficient for classification, a design choice intended to prevent the analysis from becoming circular when the marker under investigation contains HDL cholesterol in its own formula.

That methodological caution matters because UHR is calculated as uric acid concentration in milligrams per deciliter divided by HDL cholesterol in milligrams per deciliter, multiplied by one hundred. Both ingredients are measured in virtually every basic metabolic panel and lipid profile ordered worldwide, which is precisely what makes the ratio attractive. Uric acid, the end product of purine metabolism, has long been linked to insulin resistance, hypertension, and fatty liver disease, while HDL cholesterol is the well-known protective lipoprotein whose levels fall as metabolic health deteriorates. Dividing one by the other therefore creates an index that rises when the harmful component climbs and the protective component drops, theoretically capturing two dimensions of metabolic dysregulation in a single number.

The results showed a striking separation between the groups. Among the 2,453 participants, 878, or 35.8 percent, met the criteria for the dysmetabolic phenotype. Their median UHR was 13.95, with an interquartile range spanning 11.40 to 17.75, compared with 10.19 and an interquartile range of 7.82 to 13.09 in the remaining participants, a difference that was highly statistically significant. Perhaps more revealing, however, was what the researchers found when they looked at reduced HDL cholesterol in isolation: 39.4 percent of participants in Group 1, the metabolically healthier group by the study’s definition, also had low HDL cholesterol. This observation confirmed that low HDL on its own does not determine whether a patient displays the full dysmetabolic picture, and it underscored why the phenotype definition required additional abnormalities to be present.

To quantify how well UHR distinguished between the two groups, the team performed receiver operating characteristic analysis, a standard technique that plots sensitivity against specificity across all possible cut-off values and summarizes the result as the area under the curve, or AUC. An AUC of 0.5 indicates performance no better than a coin flip, while 1.0 represents perfect discrimination. UHR achieved an AUC of 0.745, which is conventionally described as moderate, with an optimal cut-off of 11.70 yielding a sensitivity of 70.8 percent and a specificity of 66.8 percent. In practical terms, roughly seven in ten patients with the dysmetabolic phenotype would be flagged by the ratio, while about two-thirds of those without it would be correctly reassured. For a marker that costs nothing beyond tests already ordered, those figures are respectable, though far from the precision clinicians would want for high-stakes decisions.

The comparative analysis proved more humbling. When UHR was pitted against its individual components, it significantly outperformed uric acid alone, which managed an AUC of just 0.580, barely better than chance. But HDL cholesterol by itself was substantially stronger, achieving an AUC of 0.825, and both comparisons were statistically significant according to DeLong’s test, a method designed to compare correlated ROC curves. This asymmetry reveals that most of the discriminatory signal in the ratio comes from the HDL cholesterol denominator rather than the uric acid numerator. In other words, the composite index largely repackages information that HDL cholesterol already provides, and the uric acid contribution, while statistically detectable, does not transform the marker’s performance.

Multivariable logistic regression reinforced the association while adding an important nuance about its shape. After adjustment for relevant covariates, each unit increase in UHR was associated with odds of the dysmetabolic phenotype rising by a factor of 1.356, with a 95 percent confidence interval of 1.316 to 1.397 and a p value below 0.001. However, the model also included a quadratic term for UHR that was itself highly significant, indicating that the relationship is not a straight line across the range of observed values. Nonlinear associations of this kind can arise when the risk conferred by a biomarker plateaus or accelerates at higher concentrations, and they complicate the interpretation of a single odds ratio, reminding readers that summary statistics can obscure the true dose-response pattern.

The most consequential test was whether UHR adds anything to a clinical model that already contains HDL cholesterol. The answer was technically yes, but only barely. Adding the ratio to an HDL-containing model improved the AUC from 0.871 to 0.873, an increment of two thousandths that, while statistically evaluated, is unlikely to alter any individual patient’s care. This is a familiar pattern in biomarker research, where novel composite indices demonstrate impressive univariate associations but deliver marginal incremental value once established markers are accounted for. The authors themselves concluded that UHR should be regarded as a complementary biochemical index rather than a replacement for HDL cholesterol or for direct assessment of metabolic abnormalities, a measured interpretation that stands in contrast to the breathless enthusiasm such ratios sometimes attract on social media and in commercial health testing.

Several limitations frame the findings. The study is retrospective and cross-sectional, meaning it captures a single moment in time and cannot establish whether an elevated UHR precedes the development of the dysmetabolic phenotype or merely accompanies it. The phenotype itself was defined by the investigators rather than by an external gold standard such as the International Diabetes Federation criteria, which, although the authors referenced IDF and WHO frameworks in their methodology, means the results speak most directly to the study’s own classification scheme. The cohort’s composition, dominated by women and drawn from a single hospital in Kayseri, also raises questions about generalizability to other populations, and the shared dependence of both the marker and the phenotype on HDL cholesterol required careful statistical handling, including variance inflation factor checks for collinearity.

Even with those caveats, the study offers a useful lesson in biomarker epidemiology. It demonstrates that a transparent, reproducible calculation from routine labs can stratify metabolic risk in a large obesity cohort, and it does so with the kind of head-to-head benchmarking against component markers that many published ratios never bother to include. For clinicians, the practical takeaway is that HDL cholesterol remains the dominant laboratory signal, and that patients with obesity and low HDL warrant a fuller metabolic workup regardless of what any derived ratio suggests. For researchers, the modest incremental AUC of 0.002 poses a pointed question about where the field should go next: toward ever more elaborate composite indices, or toward longitudinal studies that test whether these markers genuinely predict incident disease. As uric acid metabolism, HDL biology, and insulin resistance continue to be untangled at the molecular level, ratios like UHR may yet find a more definitive role, but this study suggests that role will be supporting, not starring.

Subject of Research: Association between the uric acid-to-HDL cholesterol ratio and a dysmetabolic phenotype in adults with obesity

Article Title: Association of the uric acid-to-HDL cholesterol ratio with a study-defined dysmetabolic phenotype in adults with obesity: a cross-sectional study

Article References: Association of the uric acid-to-HDL cholesterol ratio with a study-defined dysmetabolic phenotype in adults with obesity: a cross-sectional study. (n.d.). https://doi.org/10.1186/s12902-026-02557-8

Image Credits: AI Generated

DOI: 10.1186/s12902-026-02557-8

Keywords: uric acid-to-HDL cholesterol ratio, obesity, metabolic syndrome, dyslipidemia, HDL cholesterol, insulin resistance, cross-sectional study, biomarkers, metabolic risk, triglycerides, type 2 diabetes, ROC analysis