Refining oxygen consumption estimation using body composition and anthropometry for accurate cardiac output assessment in patients with heart failure
W Kawaharata, H Kouzu, S Katano, K Toda, N Nagano, K Ohori, N Yoshioka, T Yano, M FuruhashiAbstract
Background
Accurate measurement of cardiac output during right heart catheterization is crucial in patients with heart failure. However, direct measurement of oxygen consumption (VO2) in the Fick method is logistically challenging in routine practice. VO2 is therefore conventionally estimated using body surface area (BSA)-based equations, but their accuracy is limited. We hypothesized that body composition assessed by dual-energy X-ray absorptiometry (DEXA), a closer proxy for metabolic rate than BSA, could improve VO2 estimation accuracy. We further aimed to translate the DEXA-based body composition approach into an implementable method using simple limb anthropometry.
Methods and Results
We retrospectively analyzed 159 hospitalized patients with heart failure who underwent both cardiopulmonary exercise testing (CPX) and DEXA within 2 months. Median age was 57 years (interquartile range [IQR], 48–69), 68.6% were male, and median left ventricular ejection fraction was 37.5% (IQR, 28.1–51.3). DEXA quantified appendicular skeletal muscle mass (ASM), fat mass, and residual fat-free mass (rFFM; body weight excluding fat mass and ASM, comprising trunk and head lean soft tissue and bone mineral content). We evaluated the accuracy of the existing LaFarge equation against resting VO2 measured during CPX. The LaFarge equation systematically underestimated VO2 (median relative error = -15.2%, concordance correlation coefficient [CCC] = 0.505), yielding large errors (>25%) in 20.9% of patients. In multivariable analyses adjusted for age, sex, and heart rate, ASM, fat mass, and rFFM were each independently associated with VO2. A new DEXA-based model incorporating age, ASM, fat mass, and rFFM improved estimation (CCC = 0.825), reducing large errors (>25%) to 0.6%. To enhance practicality, we derived an anthropometry-based VO2 model by approximating ASM from body weight (BW), calf circumference (CC), mid-arm circumference (MAC), and sex using our previously published equation (Katano S, et al. Eur J Prev Cardiol 2021;28:e18) and estimating fat mass from body fat percentage using the Deurenberg adult equation, yielding a final model requiring only age, sex, BW, height, CC, and MAC. This simplified approach maintained robust predictive ability (CCC = 0.802) with a similarly low rate of large errors (>25%) at 0.7%. In a small external validation cohort (n = 27), the DEXA-based model showed a similar performance pattern, with better agreement and fewer large errors than the LaFarge equation (CCC 0.853 vs 0.457; >25% error 3.7% vs 44.4%).
Conclusions
Conventional BSA-based VO2 estimation was inaccurate and may underestimate Fick-derived cardiac index. A DEXA-derived model improved VO2 prediction, and a practical anthropometry-based model preserved performance with minimal large errors, enabling more accurate Fick-derived cardiac index assessment.