How to define low muscle mass: A critical exploration of current definitions supports a reference equation-based...

Current definitions of low muscle mass rely on ratios in combination with gender-specific cut-offs. While these approaches aim to adjust for differences in body metrics, their mathematical validity has not been systematically examined and may introduce bias-not only in relation to muscle mass itself, but also in relation to adiposity.

In the National Health and Nutrition Examination Survey (NHANES) cohort 1999-2006 (n = 8325) DXA-based definitions for low muscle mass of the European Working Group on Sarcopenia in Older People (ALST/height²), of the Foundation for the National Institutes of Health sarcopenia project (ALST/BMI) and of the ESPEN and EASO consensus for sarcopenic obesity (ALST/weight) were analyzed for proportional (isometric) scaling by regression analysis, calculation of scaling exponents of the denominators and graphical illustrations. The complex relationship between muscle mass, fat mass and BMI was analyzed and the adiposity related bias, cut-offs entail, explored. To illustrate the clinical relevance, we examined the relationship between low muscle mass and fatty liver disease in a second NHANES cohort (2017-2018, n = 673).

While ALST/height² approximated isometric scaling, ALST/BMI and ALST/weight showed a major mathematical bias. Instead of the assumed β = 1, BMI scaled to ALST with β = 0.6 and weight with β = 0.7. In addition to the increase with BMI, variation of muscle mass at a given BMI was tightly linked to adiposity: age and BMI adjusted standard deviation scores for muscle and fat mass correlated with r = 0.833 and r = 0.799 in men and women, respectively (p < 0.001). As none of the ratio-based definitions correctly adjusted for these associations, cut-offs unmasked a major, adiposity-related bias. Using ALST/height² the group with low muscle mass had a ∼7 kg/m² lower BMI, opposed by ALST/BMI and ALST/weight with a ∼6 kg/m² and ∼7 kg/m² higher BMI. In line, fatty liver disease was less prevalent in subjects with low muscle mass following the ALSTI/height² definition (PR = 0.30; CI 0.09-0.97), but more prevalent following the ALST/BMI and ALST/weight definitions (PR = 1.92; CI 1.38-2.67 and PR = 2.20; CI 1.58-3.06). Latter associations were no longer statistically significant when BMI and weight were corrected for β values (BMI^0.6 and weight^0.7).

In contrast to ALST/height², ALST/BMI and ALST/weight were not mathematically valid. Normalization of muscle mass by these ratios cannot correctly account for the complex association of muscle mass with fat mass, BMI and age and therefore potentially confounds results when cut-offs are applied. Reference equations that adjust for age and BMI and yield standard deviation scores may offer a more valid and clinically meaningful alternative.