BACKGROUND

Muscle mass declines after menopause and is a key risk factor for frailty, falls and poor physical function as women age. The deuterated creatine (DCr) dilution method provides a direct assessment of muscle mass, but its accuracy in Asian women has not been evaluated. Our aim was to develop a new DCr algorithm incorporating anthropomorphic variables that can estimate fat-free muscle mass, using magnetic resonance imaging (MRI) as the reference standard.

METHODS

The Integrated Women's Health Programme (IWHP) enrolled 1201 healthy community-dwelling women, aged 45-69 years at baseline, who attended gynaecological clinics from 2014 to 2016. Between February 2021 and July 2023, 894 participants were recontacted, and 451 of the respondents agreed to ingest 30 mg of DCr and had available MRI measurements of fat-free thigh and erector spinae volumes. Urinary levels of creatine, creatinine and D-creatinine levels were measured by tandem mass spectrometry 4 days after ingestion of DCr. Muscle mass was estimated using the two DCr algorithms (DCr and DCr) in current use and a newly developed algorithm (DCr) incorporating anthropometric variables that estimate fat-free muscle volumes. Pearson's correlation analyses were used to compare the performances of the DCr algorithms with MRI. Bland-Altman analysis was used to ascertain agreement between DCr and MRI.

RESULTS

Participants (n = 451, mean age 62.6 ± 5.9) were randomly divided into training (n = 367) and validation (n = 84) cohorts. In the training cohort, stepwise multivariable regression modelling indicated that age (β = -0.011, p = 0.076) and ethnicity (β = 0.154, p = 0.317 [Indian]; β = -0.012, p = 0.942 [Malay] compared to Chinese) were not associated with fat-free muscle volumes. In the final model, DCr-determined creatine pool size (β = 0.032, p < 0.001), body weight (β = 0.030, p < 0.001) and height (β = 4.336, p < 0.001) were independently associated with fat-free muscle volumes and were incorporated into a new algorithm (DCr). In a separate validation cohort, muscle volumes estimated using the DCr algorithm (R = 0.813) had a higher correlation with MRI-measured fat-free muscle volumes than both DCr (R = 0.672) and DCr (R = 0.692) algorithms. Bland-Altman analysis indicated a low bias of 0.112 L and limits of agreement of -0.969 L to +1.190 L.

CONCLUSIONS

Muscle volumes estimated with the DCr algorithm had high correlation and agreement with MRI-measured fat-free muscle volumes. The convenience of the DCr method for participants suggests its potential to be a clinically relevant method for assessing fat-free muscle volumes in sarcopenia studies.