TOPLINE

An observational imaging study using data from over 27,500 CT scans found that normalising epicardial adipose tissue volume (EATV) to total heart volume (EATVh), with a proposed cutoff of 0.1, effectively ruled out obstructive coronary artery disease (CAD) with a negative predictive value (NPV) of 97.1%.

METHODOLOGY

• Researchers analysed data of 25,155 participants (50.7% women) aged 50-64 years from a subset of the SCAPIS cohort and 2482 symptomatic patients (58.1% women) evaluated for suspected CAD at a single centre in Finland as a clinical cohort.
• They developed a fully automated, deep learning-based model to quantify EATV and epicardial adipose tissue attenuation (EATA) on non-contrast CT scans, with an expert radiologist manually reviewing and correcting any flagged cases.
• The total heart volume was measured from coronary CT angiography images using a deep learning model that segmented all cardiac structures; EATV was normalised to patient height, body surface area, total ventricular volume (right plus left ventricle), and total heart volume.
• The analysis included the Spearman rank correlation between EATA and normalised EATV, linear regression modelling, and the association of normalised EATV across quartiles with a high cardiovascular risk (Systematic Coronary Risk Evaluation 2 [SCORE2] ≥ 7.5%) and obstructive CAD (defined as stenosis of > 50% in any CA).
• A cutoff for normalised EATV was defined to include 95% of cases with obstructive CAD in the general population and validated in the clinical cohort.

TAKEAWAY

• Absolute EATV was significantly higher among men than among women across all age groups in both cohorts (median, 125.73 vs 95.49 mL in the population cohort), but EATVh effectively eliminated sex-based differences.
• EATVh improved the correlation between EATA and EATV compared with non-normalised values, with a correlation strengthening from -0.747 to -0.832 among men and from -0.721 to -0.796 among women in the population cohort (P < .0001 for both).
• In the population cohort, both the prevalence of obstructive CAD and a high cardiovascular risk by SCORE2 increased significantly across quartiles of EATVh in both sexes (P < .001 for trend). A similar CAD trend was observed in the clinical cohort (P < .05), for which SCORE2 data were unavailable.
• An EATVh cutoff of 0.1 had an NPV for obstructive CAD of 97.1% in the general population (95% in men and 99.2% in women) and 88.9% in the clinical cohort (82.6% in men and 93.5% in women).

IN PRACTICE

"Validation in a clinical cohort evaluated for chest pain and with different technical image data confirmed the utility of the concept, and a potential clinically useful cut-off for normalized EATV was proposed based on the occurrence of obstructive coronary artery disease," the authors wrote.

"This cut-off had an overall negative predictive value for obstructive CAD of 97.1% in the general population and of 88.9% in the population evaluated for chest pain," they added.

SOURCE

The study was led by David Molnar, Turku University Hospital, University of Turku, Turku, Finland. It was published online on September 04, 2026, in the European Journal of Radiology.

LIMITATIONS

The study was limited to a Northern European population, which could have restricted generalisability to other ethnic groups and broader age ranges. Automated image analysis had a risk for undetected segmentation errors. Furthermore, the clinical cohort had a higher rate of segmentation corrections due to inconsistent image quality and pacemaker artefacts.

DISCLOSURES

The SCAPIS was funded by the Swedish Heart Lung Foundation, the Knut and Alice Wallenberg Foundation, the Swedish Research Council, VINNOVA, and participating universities and university hospitals. The Finnish part of the study was funded by the Finnish Foundation for Cardiovascular Research, Finnish state research funding, and Research Council of Finland. One author reported receiving research funding from the European Union's Horizon Europe Framework programme, Swedish Medical Society, Swedish Heart Lung Foundation, Gothenburg Medical Society, and Swedish Heart Union. The authors reported no competing interests.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.