Statistical Analysis

The primary outcome measures of each MRI segmentation were the TAC volume (mm³), temporalis volume (muscle volume, mm³), and the TAC volume to muscle volume ratio (TAC:muscleR, %).

ICCs were computed to measure intra-rater reliability for muscle and TAC volumes. ICCs less than 0.4, between 0.4 and 0.75, and greater than 0.75 were considered poor, fair to good, and excellent, respectively.

A multivariate analysis of covariance (MANCOVA) was used to test the effect of sex, age, and the interaction sex-by-age on muscle and TAC volumes. MANCOVA was used as muscle, and TAC volumes (dependent variables) were positively correlated (Spearman's ρ = 0.852, p < 0.001). Sex was included in the model as a fixed factor (categorical variable), and age as a covariate (continuous variable). The inclusion of the interaction sex-by-age in the statistical model allowed to account for a potential effect of different age entry in puberty between male and female individuals.

A univariate analysis of covariance (ANCOVA) was used to test the effect of sex, age, and the interaction sex-by-age on TAC:muscleR.

The normality of standardized residuals was tested using the Kolgomorov-Smirnov test. Since residuals of muscle and TAC volumes were not normally distributed, these measures were logarithmically transformed.

A minimum sample size of 72 MRI scans was required to obtain 80% power with a small effect size (f² = 0.10, α = 0.05) considering the abovementioned factor (sex) and covariate (age), and their interaction.

All statistical analyses were performed in SPSS Statistics for Windows (Version 24.0., IBM, Armonk, NY) by a single operator (I.C.) who was blinded to age and sex of MRI studies (dataset masking). The level of statistically significance was set at p < 0.05.