Soft tissue radiodensity parameters mediate the relationship between self-reported physical activity and lower extremity function in AGES-Reykjavík participants

• Published in: Izvestiya TINRO
• Main Authors: Edmunds, Kyle J., Okonkwo, Ozioma C., Sigurdsson, Sigurdur, Lose, Sarah R., Gudnason, Vilmundur, Carraro, Ugo, Gargiulo, Paolo
• Format: Text
• Language: English
• Published: Nature Publishing Group UK 2021
• Subjects: Article; Reykjavík
• Online Access: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC8505499/; http://www.ncbi.nlm.nih.gov/pubmed/34635746; https://doi.org/10.1038/s41598-021-99699-7

Although previous studies have highlighted the association between physical activity and lower extremity function (LEF) in elderly individuals, the mechanisms underlying this relationship remain debated. Our recent work has recognized the utility of nonlinear trimodal regression analysis (NTRA) parameters in characterizing changes in soft tissue radiodensity as a quantitative construct for sarcopenia in the longitudinal, population-based cohort of the AGES-Reykjavík study. For the present work, we assembled a series of prospective multivariate regression models to interrogate whether NTRA parameters mediate the 5-year longitudinal relationship between physical activity and LEF in AGES-Reykjavík participants. Healthy elderly volunteers from the AGES-Reykjavík cohort underwent mid-thigh X-ray CT scans along with a four-part battery of LEF tasks: normal gait speed, fastest-comfortable gait speed, isometric leg strength, and timed up-and-go. These data were recorded at two study timepoints which were separated by approximately 5 years: AGES-I (n = 3157) and AGES-II (n = 3098). Participants in AGES-I were likewise administered a survey to approximate their weekly frequency of engaging in moderate-to-vigorous physical activity (PA(AGES-I)). Using a multivariate mediation analysis framework, linear regression models were assembled to test whether NTRA parameters mediated the longitudinal relationship between PA(AGES-I) and LEF(AGES-II); all models were covariate-adjusted for age, sex, BMI, and baseline LEF, and results were corrected for multiple statistical comparisons. Our first series of models confirmed that all four LEF tasks were significantly related to PA(AGES-I); next, modelling the relationship between PA(AGES-I) and NTRA(AGES-II) identified muscle amplitude (N(m)) and location (μ(m)) as potential mediators of LEF to test. Finally, adding these two parameters into our PA(AGES-I) → LEF(AGES-II) models attenuated the prior effect of PA(AGES-I); bootstrapping confirmed N(m) and μ(m) as significant partial ...