LAUSR

In a study on a subgroup of the UK Biobank population in this issue of Obesity , the authors addressed the association of 180 blood metabolites (by nuclear magnetic resonance [NMR] spectrometry) with liver fat (by magnetic resonance imaging) assessed 5 years later [1]. Liver fat was associated with obesity, diabetes, and higher levels of apolipoprotein B, very low-density lipoprotein triglycerides, and fatty acids (including saturated fatty acids and monounsaturated fatty acids). Age, sex, smoking, drinking, and body mass index (BMI) (model 2) explained 13.5% of the variance in liver fat. Adding high-density lipoprotein cholesterol, very low-density lipoprotein triglycerides, diabetes, and liver enzymes increased R 2 to 23.6% (model 4). Adding new biomarkers, as calculated from the first 10 principal components of the NMR metabolites, showed a 15% significant improvement in risk prediction relative to BMI alone ( R 2 of 27.8% compared to 13.5%), whereas this was nonsignificant compared to the predictive value of model 4 (29.6% vs. 23.6%). However, all these risk scores did not include strong determinants of liver fat (i.e., plasma insulin levels and insulin resistance [2, 3]). Although these observational findings are informative, they should be considered preliminary.