Assessment of limb muscle and adipose tissue by dual-energy X-ray absorptiometry using magnetic resonance imaging for comparison

OBJECTIVE: To use magnetic resonance imaging (MRI) to validate estimates of muscle and adipose tissue (AT) in lower limb sections obtained by dual-ene rgy X-ray absorptiometry (DXA) modelling. DESIGN: MRI measurements were used as reference for validating limb muscle and AT estimates obtained by DXA models that assume fat-free soft tissue (F FST) comprised mainly muscle: model A accounted for bone hydration only; mo del B also applied constants for FFST in bone and skin and fat in muscle an d AT; model C was as model B but allowing for variable fat in muscle and AT . SUBJECTS: Healthy men (n = 8) and women (n = 8), ages 41 - 62 y; mean (s.d. ) body mass indices (BMIs) of 28.6 (5.4) kg/m(2) and 25.1 (5.4) kg/m2, resp ectively. MEASUREMENTS: MRI scans of the legs and whole body DXA scans were analysed for muscle and AT content of thigh (20 cm) and lower leg (10 cm) sections; 24 h creatinine excretion was measured. RESULTS: Model A overestimated thigh muscle volume (MRI mean, 2.3 l) substa ntially (bias 0.36 l), whereas model B underestimated it by only 2% (bias 0 .045 l). Lower leg muscle (MRI mean, 0.6 l) was better predicted using mode l A (bias 0.04 l, 7% overestimate) than model B (bias 0.1 l, 17% underestim ate). The 95% limits of agreement were high for these models (thigh,+/- 20% ; lower leg,+/- 47%). Model C predictions were more discrepant than those o f model B. There was generally less agreement between MRI and all DXA model s for AT. Measurement variability was generally less for DXA measurements o f FFST (coefficient of variation 0.7 - 1.8%) and fat (0.8 - 3.3%) than mode l B estimates of muscle (0.5-2.6%) and AT (3.3 - 6.8%), respectively. Despi te strong relationships between them, muscle mass was overestimated by crea tinine excretion with highly variable predictability. CONCLUSION: This study has shown the value of DXA models for assessment of muscle and AT in leg sections, but suggests the need to re-evaluate some of the assumptions upon which they are based.