Longitudinal association between muscle and bone loss: Results of US and Japanese cohort studies

Muscle and bone are physiologically interconnected, but joint changes of muscle and bone with aging, and whether the muscle-bone changes are different by sex and by country has been little studied. We examined longitudinal associations of bone mineral density (BMD) and muscle mass or muscle strength in community-dwelling 65 years or older in the United States and Japan.

The present analytic sample included 1129 women and men from the Baltimore Longitudinal Study of Aging (BLSA) (mean age, 74.5 ± 7.5 years; women, 49.8%) and 1998 women and men from the National Institute for Longevity Sciences-Longitudinal Study of Aging (NILS-LSA) (mean age, 70.0 ± 4.5 years; women, 51.4%). Median follow-up was 4.6 (min-max, 0-15.4) years in the BLSA and 4.0 (min-max, 0-13.4) years in the NILS-LSA. We selected visits at which participants had BMD (whole body, pelvic, femoral neck, trochanter, and Ward's triangle BMDs) and muscle mass [appendicular lean mass, (ALM)] measured by DXA scan. In each bone site, we ran cohort-specific bivariate linear mixed-effects models adjusted for baseline age, sex, body height, body weight, fat mass, education year, and smoking status. Race was an additional adjustment in the BLSA. Additionally, we performed sex-specific analyses.

In the BLSA, the rate of change in ALM positively correlated with the rate of change in the whole body (rho = 0.30, P < 0.0001) and pelvic BMD (rho = 0.24, P < 0.0001), but not in trochanter, femoral neck, or Ward's triangle BMD (P > 0.05). In the NILS-LSA, ALM positively correlated with the rate of change in all bone sites (rho ranged from 0.20 to 0.71, P < 0.01). In women, ALM positively correlated with the rate of change in all bone sites in both cohorts (in the NILS-LSA, rho ranged from 0.35 to 0.91, P < 0.01; in the BLSA, rho ranged from 0.26 to 0.56, P < 0.05) except for femoral neck BMD in the BLSA. In men, ALM positively correlated with pelvic, trochanter, and Ward's triangle BMD in the NILS-LSA (rho ranged from 0.45 to 0.68, P < 0.0001), and whole body and trochanter BMD in the BLSA (both, rho = 0.20, P < 0.05).

Muscle loss co-occurred with bone loss in both cohorts, but the association in the NILS-LSA tended to be stronger than in the BLSA, and the association was higher in women than in men, implying that the association may differ by sex and country.