Cardiac Adipose Tissue and Pericoronary Inflammation
Summary¶
Two major cardiac adipose depots — epicardial/pericardial adipose tissue (EPAT) and pericoronary adipose tissue (PCAT) — represent distinct physiological, genetic, and diagnostic phenomena [1, 2, 3]. EPAT volume is a bulk ectopic fat depot whose genetic architecture maps directly to generic visceral adiposity; its polygenic risk score (PGS) disease associations fully attenuate after adjusting for abdominal visceral adipose tissue (VAT) [1]. In contrast, PCAT CT attenuation (measured within a $-190$ to $-30$ Hounsfield Unit window) and radiomic texture phenotyping directly reflect active vascular-wall inflammation and adipocyte lipolysis, showing complete physiological dissociation from PCAT volume and structural plaque burden [2, 3].
Biophysical & Anatomical Classification of Cardiac Fat Depots¶
| Depot | Anatomical Boundaries | Primary CT/MRI Metric | Primary Biological Process Reflected | Genomic Status |
|---|---|---|---|---|
| Epicardial / Pericardial Adipose Tissue (EPAT) | Enclosed within the pericardial sac surrounding the myocardium and coronary arteries [1]. | MRI cross-sectional area ($\text{cm}^2$) / volume ($\text{cm}^3$) [1]. | Bulk visceral adiposity & systemic metabolic syndrome [1]. | Well-mapped: 7 lead GWAS loci ($N=44,475$ MRI) [1]. |
| Pericoronary Adipose Tissue Volume (PCAT-v) | Fat within a 3 mm radial distance surrounding the vessel adventitia [2]. | Total voxel volume ($\text{cm}^3$) [2]. | Local ectopic fat accumulation [2]. | Correlates with age, BMI, and plaque burden [2]. |
| Pericoronary Fat Attenuation Index (PCAT-a / FAI) | Fat surrounding the proximal right coronary artery (RCA) or LAD [2, 3]. | Mean CT attenuation in Hounsfield Units ($-190$ to $-30$ HU) [2]. | Vascular wall inflammation & cytokine-driven lipolysis [2, 3]. | Unmapped genome-wide (Open Gap) [1, 2]. |
| PCAT Radiomic Phenotype | Proximal RCA and lesion-specific 3D spatial voxel heterogeneity [3]. | 1,103 texture, geometry, and wavelet features [3]. | Focal micro-inflammation, fibrosis, and adipocyte remodeling [3]. | Unmapped genome-wide [3]. |
Epicardial/Pericardial Adiposity Is Genetically Visceral Fat (Rämö et al. 2024)¶
Rämö et al. developed a deep-learning convolutional neural network (U-Net) to quantify EPAT area from four-chamber cardiac MRI in 44,475 UK Biobank participants, executed a genome-wide association study, and evaluated a genome-wide polygenic score (EPAT PGS) in 453,733 FinnGen participants [1].
Identified Loci & Visceral Adiposity Genes¶
GWAS identified 7 independent genome-wide significant loci ($p < 5 \times 10^{-8}$), all of which encode regulators of general adipogenesis, brown fat differentiation, or abdominal visceral fat distribution [1]: - EBF1 & *EBF2: Early B-cell factors regulating brown/beige adipocyte commitment and adipose morphology [1]. - CEBPA: CCAAT/enhancer-binding protein alpha, a master transcriptional regulator of adipocyte differentiation [1]. - WARS2 / *TRIB2: Regulators of mitochondrial translation in visceral adipose tissue depots [1]. - RERGL & *AHR*: Loci governing fat distribution and metabolic homeostasis [1].
┌────────────────────────────────────────────────────────┐
│ EPAT Deep-Learning MRI │
│ (N = 44,475 UKB; N = 453,733 FinnGen) │
└───────────────────────────┬────────────────────────────┘
│
┌──────────────┴──────────────┐
▼ ▼
┌─────────────────────────────┐ ┌─────────────────────────────┐
│ Unadjusted PGS Risk │ │ Multivariable VAT Model │
│ - CAD: OR 1.04 (p=4.1e-6) │ │ - CAD: HR 1.00 (p=0.98) │
│ - HF: OR 1.05 (p=3.8e-9) │ │ - HF: HR 1.02 (p=0.21) │
│ - T2D: OR 1.06 (p=1.2e-14) │ │ - T2D: HR 1.01 (p=0.45) │
└─────────────────────────────┘ └─────────────────────────────┘
│
▼
┌──────────────────────────────────┐
│ Attenuates to Non-Significance │
│ (EPAT is generic VAT, not CVD) │
└──────────────────────────────────┘
Complete Attenuation Upon Abdominal Visceral Fat (VAT) Adjustment¶
Unadjusted EPAT PGS predicted multiple cardiovascular and metabolic endpoints in FinnGen: T2D (OR $1.06$ per SD, $p = 1.2 \times 10^{-14}$), Heart Failure (OR $1.05$, $p = 3.8 \times 10^{-9}$), CAD (OR $1.04$, $p = 4.1 \times 10^{-6}$), Atrial Fibrillation (OR $1.04$, $p = 1.1 \times 10^{-5}$), and ischemic stroke (OR $1.02$, $p = 0.038$) [1].
However, in multivariable Cox models incorporating MRI-measured abdominal visceral adipose tissue (VAT) volume, all prospective disease associations attenuated to non-significance ($p > 0.05$) [1]. This proves that EPAT area does not represent a cardiac-specific pathogen, but rather serves as a surrogate marker of generic systemic visceral adiposity [1].
Methodological Rule for CVD Subtyping: EPAT PGS must be included as a visceral-adiposity covariate / negative control to confirm whether a proposed biomarker is independent of general ectopic fat, rather than as a primary coronary inflammation signal [1].
Dissociation of PCAT Attenuation (Inflammation) from PCAT Volume (Adiposity)¶
Yuvaraj et al. analyzed 108 matched patients undergoing coronary computed tomography angiography (CCTA) (54 Indigenous Australian, 54 non-Indigenous) to evaluate the spatial dissociation between PCAT attenuation (PCAT-a) and PCAT volume (PCAT-v) [2]:
- PCAT Attenuation (PCAT-a, Inflammation): Higher (less negative / increased attenuation) in males ($-86.7 \text{ HU}$ vs $-91.3 \text{ HU}, p = 0.003$) and in patients with obstructive CAD ($-86.5 \text{ HU}$ vs $-90.8 \text{ HU}, p = 0.030$) [2]. Crucially, PCAT-a exhibited zero correlation with structural plaque burden (Segment Involvement Score SIS $r = 0.14, p = 0.18$) [2].
- PCAT Volume (PCAT-v, Ectopic Fat): Strongly correlated with structural plaque burden (SIS $r = 0.29, p = 0.003$), age, hypertension, and body mass index ($p < 0.001$), with significantly higher total volume observed in Indigenous Australian participants [2].
This physiological separation confirms that pericoronary fat attenuation captures active vessel-wall cytokine release (IL-6, TNF-$\alpha$ inhibiting intracellular lipid accumulation and causing cell shrinkage), whereas PCAT volume reflects cumulative metabolic fat deposition [2].
Radiomic Texture Phenotyping & Acute MI Discrimination (Lin et al. 2020)¶
In a prospective matched case-control study, Lin et al. recruited 180 matched participants across three groups ($N=60$ acute myocardial infarction [MI] undergoing CCTA within 48h of admission before invasive angiography; $N=60$ stable CAD; $N=60$ no CAD controls) matched for age, sex, risk factors, medications, and CT tube voltage [3].
CCTA Scan (N = 180 Matched Subjects)
- 60 Acute MI (within 48h, pre-angiography)
- 60 Stable CAD
- 60 Controls (No CAD)
│
│ 3D PCAT Segmentation (Proximal RCA + Lesions)
▼
Radiomics Feature Extraction
(1,103 Quantitative Parameters)
│
┌──────────────────────┴──────────────────────┐
▼ ▼
Mean PCAT Attenuation (HU) Texture & Geometric Radiomics
- MI vs Controls: Differs - 20.3% Parameters Differ (p < 0.0006)
- MI vs Stable: Differs - 16.5% Parameters Differ (p < 0.0006)
- AUC = 0.77 - XGBoost AUC = 0.87 (p = 0.001) [3]
Quantitative Radiomic Extraction & Clustering¶
- Feature Matrix: Computed 1,103 radiomic parameters per PCAT volume, spanning GLCM, GLRLM, shape, geometry, and wavelet-filtered intensity distributions [3].
- Significant Features: 20.3% of radiomic parameters differed significantly between acute MI patients and controls ($p < 0.0006$), and 16.5% differed between acute MI and stable CAD [3]. In contrast, zero radiomic parameters differed between stable CAD and healthy controls ($p > 0.05$) [3].
Machine Learning Discriminative Models (XGBoost)¶
Using gradient boosted decision trees (XGBoost), Lin et al. evaluated multi-parametric diagnostic models for identifying acute MI [3]:
| Model Specification | Input Features Included | AUROC ($95\%\text{ CI}$) | Statistical Significance |
|---|---|---|---|
| Clinical Model | Age, sex, hypertension, dyslipidemia, smoking, lipids, hs-CRP | $0.76$ ($0.68\text{--}0.84$) | Baseline [3] |
| Clinical + PCAT Attenuation | Clinical features + mean HU attenuation | $0.77$ ($0.69\text{--}0.85$) | $p = 0.54$ vs Clinical alone [3] |
| Integrated Radiomics Model | Clinical features + mean HU attenuation + 3D Radiomic texture signature | $0.87$ ($0.81\text{--}0.93$) | $p = 0.001$ vs Attenuation model [3] |
6-Month Post-MI Temporal Stability & Persistent Pan-Coronary Inflammation¶
At 6-month CCTA follow-up post-MI, despite standard guideline-directed medical therapy (high-intensity statins, dual antiplatelet therapy), the PCAT radiomic phenotype around non-culprit lesions and the proximal RCA showed no significant change ($p > 0.05$) [3]. This demonstrates that acute MI triggers a persistent, pan-coronary vascular inflammatory state that remains detectable on radiomic CCTA long after clinical stabilization [3].
Open Gap in Cardiovascular Genomics¶
While coronary artery calcium (CAC) and CCTA plaque burden have large-scale published GWAS, pericoronary adipose tissue attenuation (PCAT-a) and PCAT radiomic phenotypes have no published genome-wide association studies to date [1, 2, 3]. Mapping the genetic architecture of PCAT-a and radiomic texture features represents an essential prerequisite for isolating non-lipid, vessel-specific inflammatory polygenic risk axes [1, 2, 3].
See Also¶
- Polygenic Subtyping of Cardiovascular Disease — integration of imaging endophenotypes into multi-axis risk models.
- Genetics of Subclinical Coronary Atherosclerosis Imaging — plaque burden and CAC GWAS resources.
Citations¶
[1] Rämö, J. T., Kany, S., Hou, C. R., Friedman, S. F., Roselli, C., Nauffal, V., Koyama, S., Karjalainen, J., FinnGen, Maddah, M., Palotie, A., Ellinor, P. T., & Pirruccello, J. P. (2024). Cardiovascular Significance and Genetics of Epicardial and Pericardial Adiposity. JAMA Cardiology, 9(5), 418--427. DOI: 10.1001/jamacardio.2024.0080. Source: jamacardiology_rm_2024_oi_240006_1714486932.70443.pdf. Supports: Deep-learning MRI sample size (N=44,475 UKB, N=453,733 FinnGen), 7 lead loci (EBF1, EBF2, CEBPA, WARS2, TRIB2), EPAT PGS disease ORs, and multivariable VAT adjustment attenuation. Location: Full text -- Abstract, Results, Figures 1-3. Verified 2026-07-30.
[2] Yuvaraj, J., Lim, E., Vo, T., Huynh, D., Rocco, C., Nerlekar, N., Cheng, K., Lin, A., Dey, D., Nicholls, S. J., Kangaharan, N., & Wong, D. T. L. (2023). Pericoronary adipose tissue attenuation on coronary computed tomography angiography associates with male sex and Indigenous Australian status. Scientific Reports, 13, 15509. DOI: 10.1038/s41598-023-41341-9. Source: 41598_2023_Article_41341.pdf. Supports: -190 to -30 HU attenuation window, PCAT-a vs PCAT-v dissociation, sex (-86.7 vs -91.3 HU), CAD status (-86.5 vs -90.8 HU), and SIS plaque correlation (r=0.29 vs r=0.14). Location: Full text -- Results, Discussion. Verified 2026-07-30.
[3] Lin, A., Kolossváry, M., Yuvaraj, J., Cadet, S., McElhinney, P. A., Jiang, C., Nerlekar, N., Nicholls, S. J., Slomka, P. J., Maurovich-Horvat, P., Wong, D. T. L., & Dey, D. (2020). Myocardial Infarction Associates With a Distinct Pericoronary Adipose Tissue Radiomic Phenotype. JACC: Cardiovascular Imaging, 13(11), 2371--2383. DOI: 10.1016/j.jcmg.2020.06.033. Source: lin-et-al-2020-myocardial-infarction-associates-with-a-distinct-pericoronary-adipose-tissue-radiomic-phenotype.pdf. Supports: Prospective case-control design (N=60 acute MI, N=60 stable CAD, N=60 controls), 1,103 radiomic parameters, XGBoost ML integration (AUC 0.87 vs 0.77 vs 0.76), 6-month post-MI persistence. Location: Full text -- Abstract, Methods, Results, Figures 1-5. Verified 2026-07-30.