Characterization of the metabolic profile associated with serum 25-hydroxyvitamin D: a cross-sectional analysis in population-based data

Abstract

Background: Numerous observational studies have observed associations between vitamin D deficiency and cardiometabolic diseases, but these findings might be confounded by obesity. A characterization of the metabolic profile associated with serum 25-hydroxyvitamin D [25(OH)D] levels, in general and stratified by abdominal obesity, may help to untangle the relationship between vitamin D, obesity and cardiometabolic health.

Methods: Serum metabolomics measurements were obtained from a nuclear magnetic resonance spectroscopy (NMR)- and a mass spectrometry (MS)-based platform. The discovery was conducted in 1726 participants of the population-based KORA-F4 study, in which the associations of the concentrations of 415 metabolites with 25(OH)D levels were assessed in linear models. The results were replicated in 6759 participants (NMR) and 609 (MS) participants, respectively, of the population-based FINRISK 1997 study.

Results: Mean [standard deviation (SD)] 25(OH)D levels were 15.2 (7.5) ng/ml in KORA F4 and 13.8 (5.9) ng/ml in FINRISK 1997; 37 metabolites were associated with 25(OH)D in KORA F4 at P < 0.05/415. Of these, 30 associations were replicated in FINRISK 1997 at P < 0.05/37. Among these were constituents of (very) large very-low-density lipoprotein and small low-density lipoprotein subclasses and related measures like serum triglycerides as well as fatty acids and measures reflecting the degree of fatty acid saturation. The observed associations were independent of waist circumference and generally similar in abdominally obese and non-obese participants.

Conclusions: Independently of abdominal obesity, higher 25(OH)D levels were associated with a metabolite profile characterized by lower concentrations of atherogenic lipids and a higher degree of fatty acid polyunsaturation. These results indicate that the relationship between vitamin D deficiency and cardiometabolic diseases is unlikely to merely reflect obesity-related pathomechanisms.

Keywords: 25(OH)D; metabolomics; molecular epidemiology; obesity; vitamin D.

Figure 1.

Individual metabolites associated with 25(OH)D in KORA F4 and FINRISK 1997. β-coefficients (standardized), 95% confidence intervals and P-values of the 37 metabolites associated with 25(OH)D (P-value threshold: 1.20 x 10⁻⁴) in KORA F4 (square). Of these, 30 associations were replicated in FINRISK 1997 (diamond) at P < 1.35 x 10⁻³. Main models: adjusted for age, sex, season (KORA F4 only), WC (continuous), physical activity, time outdoors (KORA F4 only), smoking status and alcohol consumption.

Figure 2.

Individual metabolite associations in abdominally obese and non-obese participants in KORA F4. β-coefficients (standardized), 95% confidence intervals and P-values of the 37 metabolites associated with 25(OH)D from the main models for abdominally obese (circle) and non-obese (triangle) participants. In addition, 25(OH)D was associated with glucose (N) (β = −0.0204, P-value = 0.00004) in abdominally obese participants and with 3-methyl-2-oxovalerate (M) (β = −0.0162, P-value = 0.00005) in non-obese participants (P-value threshold was set at 6.02 x 10⁻⁰⁵).