TyG-WHtR Predicts Incident Aortic Stenosis in UK Biobank

Key Takeaways

• Adults in UK Biobank with CKM stages 0–3 were followed prospectively, and higher TyG and obesity-derived TyG indices were associated with later aortic stenosis.
• Participants in the highest tertile of each TyG-related index had greater aortic stenosis risk than those in the lowest tertile after multivariable adjustment.
• All six TyG-related indices showed linear dose-response relationships with aortic stenosis, and systolic blood pressure statistically accounted for part of the association.
• TyG-waist-to-height ratio (TyG-WHtR) showed the strongest incremental predictive improvement beyond the baseline model.

Aortic stenosis is often recognized only after a long subclinical period, when valvular calcification is already advanced and earlier risk identification is difficult. In a UK Biobank prospective cohort study of TyG-related indices and incident aortic stenosis in CKM stages 0-3, adults aged 40 to 69 years recruited from 2006 to 2010 were narrowed from 408,984 screened participants to 363,907 after exclusions. The exposure measures were triglyceride-glucose (TyG), TyG-body mass index (BMI), TyG-waist, TyG-waist-to-hip ratio (WHR), TyG-waist-to-height ratio (WHtR), and TyG-a body shape index (ABSI). Incident aortic stenosis (AS) was identified from hospital records with ICD-10 codes I35.0 and I35.2, with follow-up until AS, death, loss to follow-up, or the end of hospital records. Analyses used Cox models, Fine-Gray competing risk models with all-cause mortality as the competing event, restricted cubic splines, mediation analysis for systolic blood pressure (SBP), and prediction modeling, with adjustment that included estimated glomerular filtration rate (eGFR).

Over a median 15.25 years, 3,326 incident AS cases occurred. After multivariable adjustment, all six indices were positively associated with incident AS; in the detailed competing-risk results, per-SD estimates ranged from 3% higher risk for TyG-ABSI to 24% for TyG-BMI. Across tertiles, the highest-versus-lowest association ranged from hazard ratio 1.13 for TyG-ABSI to 1.77 for TyG-BMI, and spline analyses supported linear dose-response relationships across all six measures.

SBP statistically accounted for 12.1% to 18.1% of the observed associations. Adding TyG-WHtR to the baseline model increased the Harrell C-index from 0.796 to 0.806, with a net reclassification improvement (NRI) of 0.308, and TyG-WHtR performed best among the six indices. Subgroup analyses suggested stronger associations in earlier CKM stages and several lower-risk subgroups, while sensitivity analyses remained directionally consistent after early-event exclusion, medication adjustment, and additional biomarker, physical activity, or diet adjustment.

AS ascertainment relied on inpatient ICD-10 coding and could have missed milder or asymptomatic disease, and the cohort did not include systematic echocardiographic severity data. TyG-related indices were measured only at baseline, so the analysis does not address trajectories or cumulative exposure over time. Because the SBP mediation analysis used baseline cross-sectional measurements, it supports statistical accounting rather than proof of a causal pathway. These findings tracked future AS risk in UK Biobank participants from England, Scotland, and Wales, but they do not directly extend to CKM stage 4 populations and may not generalize cleanly beyond a predominantly White volunteer cohort.

The authors reported that six TyG-related indices were independently associated with long-term incident AS in CKM stages 0–3, and that TyG-WHtR showed the largest incremental predictive gain within the models tested. In this observational UK cohort, the findings link baseline metabolic and anthropometric markers with later AS rather than establishing causation.

Clinician Questions

How was incident aortic stenosis defined in the UK Biobank CKM stage 0–3 analysis?

Incident aortic stenosis among UK Biobank participants with CKM stages 0–3 was identified from hospital diagnosis records using ICD-10 codes I35.0 and I35.2, with follow-up continuing until AS, death, loss to follow-up, or the end of hospital episode records.

Which patients did these TyG-related aortic stenosis findings not directly apply to?

These findings did not directly apply to people with CKM stage 4, who were excluded by design, and they are not generalizable to people with established cardiovascular disease or to more diverse populations because the cohort was predominantly White and reflected UK Biobank volunteer selection.

What did the subgroup analyses suggest about TyG-related indices across CKM stages?

Subgroup analyses suggested that associations were generally stronger in earlier CKM stages, with elevated TyG linked to higher AS risk mainly in CKM stages 0–1, while obesity-derived indices remained significantly associated across CKM stages 0–3; effect modification by CKM stage was reported for most indices except TyG-BMI.

Why should the SBP mediation result be interpreted cautiously in TyG-related aortic stenosis risk?

The SBP mediation analysis used baseline cross-sectional SBP and TyG-related measurements, so the reported mediated proportion reflects statistical accounting within shared pathways rather than establishing temporal or causal mediation for AS risk.