Abstract P3004: Plasma and Serum Water T ₂ are Strong Predictors of Cardiometabolic Health: Implications for Point-Of-Care Screening

Abstract

Introduction: Plasma and serum water T 2 are global biosensors of metabolism. They detect changes in the average rotational diffusion time of water induced by shifts in metabolic balance (Fig. 1). Water T 2 can be measured using a miniaturized magnetic resonance device (Fig. 2). However, the performance of T 2 parameters in combination with other readily available measures of cardiometabolic health is unknown. Hypothesis: The accuracy of water T 2 for predicting cardiometabolic health can be enhanced by including age, body mass index, and pulse rate. Specific Aim: Quantify the predictive accuracy of plasma or serum water T 2 with and without other measures. Methods: This is a secondary analysis of FWT2, an observational cross-sectional study of 72 disease-free adults. Other study exclusions were pregnancy and fasting <10 h. Hydrogen T 2 relaxation decay curves were recorded for plasma or serum at 37°C using a Bruker mq20 magnetic resonance relaxometer. Water T 2 was resolved using a discrete component analysis (XPFit, Alango Ltd.). Statistical analyses utilized JMP Pro v17.2 (SAS Institute): factor analysis, machine-learning predictor screening, and multi-variable linear regression. Goodness-of-fit and predictive accuracy were measured using adjusted R-squared (R 2 adj), root mean square error, corrected Akaike’s information criterion (AICc), Bayesian information criterion (BIC), and k-fold cross-validation R 2 (R 2 k-fold). Results: Table 1 shows the parameters for each regression model. Cardiometabolic health, the outcome or Y variable, was one of two latent variables generated by factor analysis: CMH1 for Phase 1&2 and CMH2 for Phase 2 only (Table 1). By itself, plasma water T 2 was a good predictor of cardiometabolic health (models 1A and 3A). The addition of age, BMI, and pulse rate improved the R 2 adj from 0.631 (3A) to 0.796 (3B). Improvements (decreases) were observed for RMS error, AICc, and BIC, with an improvement (increase) in R 2 k-fold. Models with serum water T 2 showed similar trends. Conclusions: Plasma and serum water T 2 are strong predictors of cardiometabolic health, especially when combined with age, BMI, and pulse rate. Water T 2, along with other predictors, explained up to 84% of the variance in cardiometabolic health. The predictive value of T 2 is independent of adiposity. Buoyed by predictive accuracy, measurement simplicity, and device portability, plasma and serum water T 2 show promise for metabolic health screening at the point of care.