People whose bodies show both poor insulin sensitivity and high levels of an inherited cholesterol particle carry the greatest cardiovascular risk among the markers studied, according to an analysis of more than 286,000 UK Biobank participants. The study, published in Cardiovascular Diabetology, examined two very different measures: estimated glucose disposal rate, or eGDR, a composite that summarizes waist circumference, hypertension and glycated hemoglobin, and lipoprotein(a), often abbreviated Lp(a), which reflects a largely inherited atherogenic pathway. Over a median follow-up of 14.1 years, 34,921 participants developed cardiovascular disease, giving the researchers enough events to test how the two markers behave alone and together.
The team defined low eGDR as 6.28 mg/kg/min or less and high Lp(a) as 125 nmol/L or more. In multivariable Cox models, each one-standard-deviation lower eGDR was associated with a 37 percent higher hazard of cardiovascular disease, with a hazard ratio of 1.37 and a 95 percent confidence interval of 1.35 to 1.39. Each one-standard-deviation higher log-transformed Lp(a) carried a smaller but still significant association, with a hazard ratio of 1.07 and a 95 percent confidence interval of 1.06 to 1.08. Compared with participants who had both high eGDR and low Lp(a), those with low eGDR and high Lp(a) had a hazard ratio of 1.51, with a 95 percent confidence interval of 1.43 to 1.59.
Absolute risk told a similar story. Standardized 10-year cumulative incidence, calculated while accounting for competing death, was 7.15 percent in the high eGDR, low Lp(a) group and 10.45 percent in the low eGDR, high Lp(a) group. That three-percentage-point gap means the combined classification does more than rank people on a relative scale; it identifies a group in which roughly one in ten develops cardiovascular disease within a decade.
What the combination did not show was interaction. Neither the multiplicative interaction test, which returned a P value of 0.207, nor the 10-year absolute-risk interaction contrast, estimated at minus 0.08 percentage points with a 95 percent confidence interval of minus 0.58 to 0.47, reached statistical significance. In practical terms, the two markers appear to act as independent, additive signals rather than amplifying each other. The researchers also noted that the two measures were minimally correlated, which supports treating them as complementary rather than redundant, since each captures a distinct risk pathway: one cardiometabolic, the other lipoprotein-related.
The predictive-value analysis was deliberately stringent. The investigators built a comprehensive conventional risk-factor model and then tested whether adding either marker improved discrimination, measured by the C-index, using fivefold cross-validation. Adding eGDR raised the C-index from 0.71048 to 0.71174. Adding Lp(a) on top of eGDR nudged it to 0.71208. The study also found no clear additional benefit on decision-curve analysis, which evaluates net benefit across a range of risk thresholds. The authors conclude that incremental predictive value beyond comprehensive conventional risk assessment was small, particularly after adding Lp(a) to a model already containing eGDR.
That pattern is worth understanding precisely. A hazard ratio of 1.51 describes association, not causation, and the C-index gains, while statistically real, translate into little change in how well a model separates people who will and will not have an event. Decision curves similarly showed that reclassifying patients with these markers would not clearly improve clinical net benefit. The researchers therefore frame the joint classification as a way to summarize distinct risk characteristics, not as a ready-made screening upgrade, and they note that additional clinical net benefit was not demonstrated.
The study design has specific strengths and limits. Following 286,025 participants without baseline cardiovascular disease for a median of 14.1 years, with competing-risk-adjusted absolute-risk estimates and cross-validated prediction metrics, gives unusually robust evidence for the associations reported. At the same time, eGDR is an estimated composite rather than a direct measure of insulin sensitivity, Lp(a) thresholds were fixed at 125 nmol/L, and the analysis population is UK Biobank participants, so generalization to other populations and to routine clinical practice requires further study. The published version is an early-release, peer-reviewed accepted manuscript that remains citable with a permanent DOI and is subject to further edits before the final Version of Record.
For readers tracking cardiometabolic risk, the practical takeaway is nuanced. Lower eGDR and higher Lp(a) were independently associated with incident cardiovascular disease, and their combination identified the participants at greatest risk in this cohort, without evidence of statistical interaction on either the relative-hazard or the absolute-risk scale. The markers summarize different biology, one reflecting waist circumference, hypertension and glycated hemoglobin together, the other a largely inherited atherogenic lipoprotein. But the study also shows that knowing both does not substantially improve prediction beyond a comprehensive conventional risk-factor model. In other words, the combination is informative for describing who is at highest risk, yet its incremental value for changing risk prediction in practice remains small.
Subject of Research: Joint associations of estimated glucose disposal rate and lipoprotein(a) with incident cardiovascular disease in the UK Biobank
Article Title: Estimated glucose disposal rate and lipoprotein(a) as complementary markers of cardiovascular risk in the UK Biobank
Article References: Wen, S., Huang, Z., Wang, J., Chen, Z., Meng, L., Huang, Y., & Chen, Z. (2026). Estimated glucose disposal rate and lipoprotein(a) as complementary markers of cardiovascular risk in the UK Biobank. Cardiovascular Diabetology. https://doi.org/10.1186/s12933-026-03430-7
Image Credits: AI Generated
DOI: 10.1186/s12933-026-03430-7
Keywords: eGDR, lipoprotein(a), cardiovascular disease, UK Biobank, risk prediction, Cox model, C-index, absolute risk, insulin sensitivity, atherogenic lipoprotein, competing risk, cross-validation
News Source: Ophelia Keating. (November 4, 2026). Two blood markers together flag highest heart risk in UK Biobank study. Scienmag.
