The prediction of total skeletal muscle mass in a Caucasian population – comparison of Magnetic resonance imaging (MRI) and Dual‐energy X‐ray absorptiometry (DXA)

• Published in: Clinical physiology and functional imaging Vol. 37; no. 2; pp. 168 - 172
• Main Authors: Geisler, Corinna, Pourhassan, Maryam, Braun, Wiebke, Schweitzer, Lisa, Müller, Manfred J.
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.03.2017
• Subjects: Absorptiometry, Photon; Adiposity - ethnology; Adolescent; Adult; Aged; Algorithms; body composition; Body Mass Index; ethnicity; European Continental Ancestry Group; Female; Humans; lean body mass; Magnetic Resonance Imaging; Male; method bias; Middle Aged; Models, Biological; Muscle, Skeletal - diagnostic imaging; Plethysmography; Predictive Value of Tests; Reproducibility of Results; Young Adult; lean body mass; body composition; magnetic resonance imaging; ethnicity; method bias
• ISSN: 1475-0961; 1475-097X; 1475-097X
• DOI: 10.1111/cpf.12282

Summary Dual‐energy X‐ray (DXA) is an alternative to magnetic resonance imaging (MRI) to measure skeletal muscle mass. DXA assesses lean body mass (LBM), and MRI measures skeletal muscle mass (SMM). Kim et al. (Am J Clin Nutr 2002; 76: 378; J Appl Physiol (1985) 2004; 97: 655) developed MRI‐based algorithms to estimate whole‐body SMM by DXA. These algorithms were based on an ethnically mixed study population (Kim et al., Am J Clin Nutr 2002; 76: 378; J Appl Physiol (1985) 2004; 97: 655). It is unclear whether Kim's algorithms are accurate in an exclusive Caucasian population. The aim of our study was to validate Kim's equation in a Caucasian population of 346 subjects. SMMMRI was assessed using MRI, and LBM and BMCDXA were measured by DXA and fat mass (FMADP) by air‐displacement plethysmographie (ADP). SMMMRI and predicted SMM were highly correlated (r = 0·944; P<0·05). The standard error of estimate of the regression equation was 2·4 kg. However, Bland–Altman plots showed a significant (P<0·001) systematic bias between SMMMRI (median 25·1 kg; IQ 20·2–31·1 kg) and predicted SMM (median 26·3 kg; IQ 22·6–33·0 kg), overestimating SMM by 9·8%. Multiple regression analyses showed that weight explained 4·4% of the variance in the differences between SMMMRI and predicted SMM with the major part unexplained. Kim's algorithm has a systematic unexplained bias and is not recommended in Caucasians.