Quantitative MRI burden score and risk of first-ever stroke: A nested case-control study in the UK Biobank

Composite scores of cerebral small vessel disease (CSVD) burden are used to estimate stroke risk, but most rely on semi-quantitative visual ratings and may miss diffuse microstructural injury. We developed a quantitative magnetic resonance imaging (MRI) Burden Score integrating volumetric and diffusion tensor imaging (DTI) metrics and evaluated its association with first-ever stroke in the general population.

This nested case-control study was conducted within the UK Biobank imaging cohort (N = 83,951). We identified 78 participants with incident first-ever stroke and matched them to 780 controls using incidence-density sampling. A principal component analysis-derived MRI Burden Score was constructed from 12 automated MRI markers, including white matter hyperintensity volume, total brain volume, and tract-specific fractional anisotropy and mean diffusivity. Associations with incident stroke were assessed using multivariable Cox models, restricted cubic splines, and reclassification analyses.

The MRI Burden Score was higher in stroke cases than in controls (P = 0.036). After multivariable adjustment, each 1-SD increase in the score was associated with a 15% higher hazard of first-ever stroke (hazard ratio 1.15, 95% CI 1.04-1.28). Stroke risk increased nonlinearly, rising steeply above a score of approximately 0.45. Adding the MRI Burden Score to a clinical model improved risk stratification (continuous net reclassification improvement 10.8%; integrated discrimination improvement 1.9%; both P < 0.001).

A data-driven MRI Burden Score capturing cumulative macro- and microstructural brain injury independently predicts first-ever stroke and provides incremental value beyond conventional clinical risk factors.