EchoMRI is an emerging quantitative magnetic resonance technique for measuring body composition. It uses a static magnetic field to detect the hydrogen atoms of fat, lean tissue, and water by their specific spin characteristics. While EchoMRI has been validated in adults, only one prior study has examined the precision and accuracy of EchoMRI in infants, which concluded that EchoMRI was as or more reliable than existing techniques. In the current analysis, we examined the associations between anthropometry and fat mass (by EchoMRI) in our Hispanic Cohort of infants aged 1-12 months.
EchoMRI and skinfold data was available for 68 infants. Multiple linear regression was used to measure associations and predictions with fat mass, as measured by EchoMRI.
Infant weight showed the strongest correlation with fat mass (r = 0.93), followed by age (r = 0.78), midthigh skinfold (r = 0.77), suprascapular skinfold (r = 0.70), tricep skinfold (r = 0.62), and subscapular skinfold (r = 0.54; all p < 0.0001). Infant growth z-scores, calculated using WHO standards, showed lesser correlations (r = 0.24 – 0.53, p < 0.0001). In regression models, we found that including infant weight (kg), age (days) and sex resulted in a model with an R2 of 0.951 (p < 0.0001). The addition of skinfold measurements (tricep, midthigh, subscapular, suprailiac), resulted in an R2 change of 0.035 for a final model R2 of 0.951 (p < 0.0001). Therefore, skinfold measurements only slightly improved the predictive model. Replacing infant weight with any of the WHO z-scores resulted in a weaker model (R2 = 0.50 – 0.69, p < 0.0001).
Infant weight, sex, and age were the most important predictors of fat mass. Skinfold measurements only explained an additional 3.5% of the variance in fat mass. WHO z-scores were not significant predictors of infant fat mass beyond infant weight. This information could be useful for predicting fat mass in Hispanic infants that do not have EchoMRI data available.