(OR06-01-26) Multi-metabolite Scores Predictive of Sodium and Potassium Recovery Biomarkers

Objectives: Dietary sodium and potassium have been linked to human health but are poorly measured by dietary assessment tools such as food frequency questionnaires (FFQs), with correlations between FFQs and recovery biomarkers of < 0.2 for sodium and < 0.4 for potassium. Recovery biomarkers reflect absolute dietary intake but require urine collection over a 24-hr period, which is not feasible in large epidemiologic studies. Our objective was to develop serum multi-metabolite scores (MMS) that are predictive of the recovery biomarkers for sodium, potassium, and the sodium:potassium ratio.

Methods: The Interactive Diet and Activity Tracking in AARP Study included 663 adults who provided two serum and 24-hr urine samples, collected 6 months apart. Ion-selective electrode potentiometry was used to measure urinary concentrations of sodium and potassium. Ultra-high performance liquid chromatography coupled with tandem mass spectrometry was used to measure 1464 metabolites in serum. Recovery biomarkers and metabolites were averaged across sample collections to reduce within-person variation. The dataset was randomly split into training and test sets. MMS were derived in the training set using elastic net regression with repeated resampling to identify consistently selected metabolites. Predictive performance was evaluated using Pearson correlations between predicted and observed biomarkers.

Results: The mean (IQR) age of participants was 63 (59-68) years, 93% were non-Hispanic White, and 51% were female. From stability-selection, 37 metabolites were selected for the sodium MMS (primarily lipids [41%] and xenobiotics [19%]), 52 metabolites for the potassium MMS (29% lipids, 15% amino acids, 15% cofactors and vitamins), and 27 metabolites for the sodium:potassium ratio MMS (33% lipids, 26% unknown compounds). Correlations between the MMS and biomarkers were moderate to strong: 0.65 for sodium, 0.70 for potassium, and 0.63 for the ratio in the training set, and 0.41 for sodium, 0.51 for potassium, and 0.57 for the ratio in the test set.

Conclusions: Blood-based MMS have the potential to advance research on sodium, potassium, and chronic disease risk in large epidemiologic studies.

Funding Sources: This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH).