Opportunistic CT-based assessment of pectoralis major sarcopenia and risk of incident atrial fibrillation
G Cobb, G Boswell, L De Sousa Domingues Da Silva, F Amin, E Hu, S Newlander, G Wesbey, J StojanovskaAbstract
Background
Atrial fibrillation (AF) is the most common sustained arrhythmia and affects 59 million people worldwide [1]. Although coronary CT angiography (CCTA) and coronary artery calcium scoring (CACS) are widely used for cardiovascular risk stratification, their value for identifying nonvascular predictors of AF is less studied. Sarcopenia is an emerging marker of adverse cardiometabolic outcomes [2]. Pectoralis major muscle size on CT correlates with whole-body muscle mass and predicts mortality and cardiovascular events [3], but links to incident AF remain unclear.
Purpose
To use cardiac CT–based opportunistic screening to measure pectoralis major muscle size and quality and test for an independent association with new-onset AF.
Methods
We conducted a longitudinal study of adults undergoing clinically indicated CCTA (EHR; Apr 1, 2017–Nov 30, 2020). AF status was adjudicated through Nov 1, 2020 using EHR review; AF was confirmed by ECG/expert review. Patients with AF prior to index CCTA or prior mitral valve surgery were excluded. Using nearest-neighbor propensity matching (age, sex, smoking, HTN, HLD, diabetes), 47 participants with incident AF after CCTA but occurring by Nov 1, 2020 were matched 1:1 to 47 without AF as of Nov 1, 2020 (n=94). Participants were then followed from index CCTA through Nov 1, 2025 for incident AF events for time-to-event analyses. On non-contrast calcium-score scans at 120 kVp, the left pectoralis major was manually segmented at the level of maximal axial thickness by a single blinded radiologist. Metrics included cross-sectional area, maximal thickness, mean attenuation (HU), and HU SD. Exclusions included three participants (and their matched counterparts; n=6), yielding n=88. During extended follow-up to Nov 1, 2025, five additional incident AF events occurred in controls and were included in time-to-event analyses (total events=49).
Results
Eighty-eight patients were included (49 AF events, 39 censored). Sex distribution was similar (p=0.33), and body surface area (BSA) was not associated with AF (p>0.05). Age showed borderline association (p=0.053). Mean pectoralis attenuation was lower in those who developed AF (Wilcoxon p=0.0048; Kruskal–Wallis p=0.0047). Mean time to AF onset was 446±610 days (median 185). In Cox models adjusted for sex, age, and BSA, mean HU was independently associated with time to AF (β=−0.0176, SE=0.00813, p=0.0305). Hazard ratio was 0.983 per 1-HU increase (95% CI 0.967–0.998) and 0.84 per 10-HU increase (95% CI 0.72–0.98).
Conclusion
CT-derived muscle quality (attenuation), a surrogate for myosteatosis and metabolic health, is independently associated with incident AF, whereas muscle quantity is not. Lower attenuation may reflect lipid infiltration and impaired cardiometabolic reserve, potentially contributing to atrial remodeling and arrhythmic risk. Larger studies should validate muscle quality as an imaging biomarker for frailty and cardiometabolic risk.