CT Myosteatosis Across Cardiovascular Settings

Evidence

Two independent observational studies examined myosteatosis—fat infiltration that lowers skeletal-muscle attenuation on CT—as a cardiovascular risk signal in sharply different settings. The newer study analyzed people free of recognized cardiovascular disease at baseline. The second studied patients after an acute ST-elevation myocardial infarction (STEMI). Their results point in the same direction, but they do not test the same threshold, muscle region, or outcome.

The fresh study retrospectively analyzed 5,739 participants in the Multi-Ethnic Study of Atherosclerosis who had undergone coronary artery calcium (CAC) scanning. Mean age was 62.1 years, 52.3% were female, and participants had no cardiovascular disease at baseline. An artificial-intelligence system measured mean attenuation across visible thoracic skeletal muscle. Myosteatosis was defined using sex-specific cutoffs, and the investigators compared the lowest and highest attenuation quartiles with Cox models.

Over 19 years, the cohort recorded 1,826 cardiovascular events, including 1,139 atrial-fibrillation events and 359 heart-failure events. After adjustment for cardiovascular risk factors, inflammatory markers, insulin resistance, CAC burden, muscle volume, and social determinants of health, myosteatosis was associated with higher rates of total cardiovascular disease (hazard ratio 1.48, 95% confidence interval 1.25–1.75), atrial fibrillation (1.68, 1.37–2.07), and heart failure (1.61, 1.18–2.19). These are relative associations between quartiles, not probabilities for an individual.

The imaging measurement added statistical discrimination to the Agatston CAC score in this dataset. The time-dependent area under the receiver-operating-characteristic curve increased from 0.74 to 0.80 for total cardiovascular disease, from 0.68 to 0.76 for atrial fibrillation, and from 0.73 to 0.78 for heart failure. Because the source was available to ingestion only as a PubMed abstract, those reported results can be described, but the model's full validation procedures and sensitivity analyses cannot be independently assessed here.

The independent study used a hospital cohort of 324 patients who underwent emergency percutaneous coronary intervention for STEMI at one center in China between 2017 and 2020. Chest CT was performed within 72 hours of admission. Instead of automated whole-thorax segmentation, two observers manually outlined the pectoralis muscles at a specified thoracic level and calculated mean muscle attenuation. The cohort was 83.3% male, and 35 patients died during follow-up.

The researchers derived 32.5 Hounsfield units as the cutoff that best separated all-cause mortality in their own data. This classified 116 patients into a lower-attenuation group and 208 into a higher-attenuation group. Twenty-four of the 35 deaths occurred in the lower-attenuation group. In the reported multivariable analysis, higher attenuation versus lower attenuation was associated with a lower mortality rate (hazard ratio 0.30, 95% confidence interval 0.13–0.70). The direction therefore matches the community cohort: lower muscle attenuation tracked worse outcomes.

The STEMI team also combined muscle attenuation with clinical variables in a survival nomogram. Its reported concordance index was 0.86, and calibration plots compared predicted with observed survival at one, three, and five years. Those are internal performance results from the same small cohort used to choose the cutoff and construct the model. They do not establish performance in another hospital or population.

Analysis — Muscle Quality as a Cross-Setting Signal

The cross-study inference is that CT muscle attenuation may capture a broad vulnerability signal that is visible before recognized cardiovascular disease and remains associated with outcome after an acute coronary event. This is analysis, not a demonstrated mechanism or a validated clinical rule. The community study found associations with later atrial fibrillation and heart failure even after accounting for coronary calcium, muscle volume, metabolic factors, inflammation, and social variables. The STEMI study found the same adverse direction in a much sicker, post-infarction cohort using a different CT method. That convergence makes simple muscle quantity an incomplete explanation and supports further study of muscle quality as a systemic phenotype. However, the signal could still reflect age, frailty, physical inactivity, metabolic disease, or other residual confounding rather than a causal process. A stronger test would use a prespecified attenuation measure and threshold, modern standardized CT, blinded analysis, and external validation across both community and acute-care cohorts. It would also test whether attenuation adds reproducible information beyond established risk models without assuming that changing the measurement would change outcomes.

Limitations

Both studies were observational, so neither shows that myosteatosis causes atrial fibrillation, heart failure, or death. The fresh MESA paper was abstract-only in the source pack; details about missing data, model calibration, threshold derivation, and all sensitivity analyses were unavailable. Its baseline scans preceded the outcomes by many years, and a single muscle measurement cannot show whether muscle quality changed before an event.

The STEMI study was retrospective, single-center, and small, with only 35 deaths and relatively few women. Requiring an admission chest CT may have introduced selection bias. Its manual pectoralis measurement differs from the automated whole-thorax method, and its 32.5-unit cutoff was selected in the same dataset used for modeling. The nomogram had no external validation, while calibration in development data can look optimistic. The cohorts, CT protocols, muscle regions, thresholds, and endpoints cannot be pooled. These population-level associations do not predict any individual's course and do not establish a screening or treatment strategy.