Longitudinal validation of an original bias-reduced fat-free mass metabolic coefficient in inpatients with obesity.

BACKGROUND AND AIMS: In obese patients subjects, conventional resting energy expenditure (REE) normalization using the REE/fat-free mass (FFM) ratio is confounded by the metabolic contribution of adipose tissue, introducing an anthropometric bias across varying levels of adiposity. This study validates a new corrected metabolic coefficient (KFFM) designed to neutralize this bias and evaluates its longitudinal responsiveness and clinical reliability during a 8-week multidisciplinary rehabilitation intervention. METHOD AND RESULTS: Longitudinal study of 167 patients subjects with Class I-III obesity undergoing an 8-week residential program. REE and body composition were assessed by indirect calorimetry and DXA. KFFM was calculated by adjusting REE for fat mass. Bias was evaluated by Pearson correlation; longitudinal changes by Linear Mixed Models; metabolic convergence by Brown-Forsythe test and bootstrapping (5000 iterations). MDC was set at 2.30 kJ/kg/day. Allometric modeling confirmed a non-linear REE-FFM relationship with an exponent of β = 0.82 (p = 0.0007). The KFFM achieved complete bias neutralization (r = 0.098,p = 0.21), whereas the traditional REE/FFM ratio remained significantly correlated with adiposity (r = 0.25,p = 0.001). Longitudinal analysis revealed a significant overall metabolic shift (p = 0.0009), with a highly significant reduction in KFFM specifically for Class III obesity (from 118.11 to 109.95 kJ/kg FFM/day; p = 0.0004). Furthermore, 91.02% of patients subjects exhibited changes exceeding the minimum detectable change (MDC = 2.30 kJ/kg/day), confirming true physiological signals. Bootstrap analysis confirmed the stability of inter-individual variance (95% CI: -14.74 to 7.08). CONCLUSIONS: KFFM could be a bias-reduced, robust metric for REE normalization in obesity, showing stable inter-individual variability and sensitivity to clinically meaningful bioenergetic changes beyond traditional ratio-based methods.