Potential of the Plasma Nutriproteome to Predict Micronutrient Distributions in Undernourished Settings
Keith West *
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Robert Cole
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Kerry Schulze
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Joshua Betz
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Ingo Ruczinski
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Sun Eun Lee
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Lee Wu
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
James Yager
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
John Groopman
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Sudeep Shrestha
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
Parul Christian
Johns Hopkins Medical Institutions, Baltimore, MD, USA.
*Author to whom correspondence should be addressed.
Abstract
Objectives: Micronutrient deficiencies are common but remain ‘hidden' due to difficulty of assessment. We explored the plasma proteome to identify nutrient-correlated biomarkers that may predict multiple micronutrient status and deficiencies, possibly on a single platform in the future.
Methods: We measured, in 500 6-8 year old Nepalese children, plasma concentrations of >20 vitamin/mineral indicators and acute phase proteins (APP) by conventional assays, and relative abundance of proteins by quantitative mass spectrometry, bioinformatics and linear mixed effects models (Herbrich S et al, 2012; Cole R et al, 2013). We identified ~980 proteins in >10% of subjects, and evaluated their strength of correlation with micronutrient and APP distributions. Comparisons were corrected for multiple comparisons, with a 10% threshold for false discoveries.
Results: 142 proteins were correlated with plasma retinol, 6 with 25(OH) vitamin D, 119 with α-tocopherol, 12 with γ-tocopherol, 6 with PIVKA-II (reflecting vitamin K status), 89 for vitamin B6, 35 for ferritin and 7 for transferrin receptor (reflecting iron status), 232 for copper, 3 for selenium and none for folate, thyroglobulin (reflecting iodine status) or vitamin B12 (q>0.1 for all comparisons). Initial models with up to 6 covariates suggest an ability to explain 60-80% of the variation (R2) in retinol, α-tocopherol, vitamin B6 and copper, ~50% of the variation in ferritin and, the carotenoid, β-cryptoxanthin and 80-85% of variation in CRP and AGP. Other nutrient-protein models will be presented.
Conclusions: Plasma nutrient-correlated proteomes exist that, with absolute quantification of candidate proteins, could provide a basis for multiple micronutrient status assessment of populations in the future.