Establishing age-stratified red blood cell fatty acid reference ranges using model-based clustering and iterative...

The current assessment of nutritional status and diagnosis of essential fatty acids deficiency (EFAD) utilizes the analysis of long-chain fatty acids (LCFAs) in serum or plasma; however, these concentrations do not represent habitual LCFA intake. LCFAs in red blood cells (RBCs) are less prone to intra-individual variability and exclude the need for fasting, which is unrealistic in pediatric populations. Our study objective was to characterize the RBC LCFA profiles in pediatric and adult reference populations and establish age-specific reference intervals (RIs).

Twenty-one LCFAs in RBCs were measured in 523 pediatric and adult controls by gas chromatography-mass spectrometry. Model-based clustering was used to identify possible age subgroups. After removing outliers by the Tukey method, initial age subgroups were then compared using the Harris-Boyd method in an iterative manner. RIs (95%), with confidence intervals (90%), in the final age groups were established using parametric or non-parametric statistics.

Our data showed heterogeneous changes in the concentrations of most LCFAs and the EFAD biomarkers (mead acid, Triene/Tetraene ratio) during infancy. Model-based clustering identified six initial age subgroups per fatty acid, on average. Our application of the iterative Harris-Boyd method decreased the average number of age groups to three per fatty acid, with 13 total unique age cut-offs. Finally, using these age groups, we established age-specific RIs for 21 fatty acids, six group totals, and the Triene/Tetraene ratio.

Our study revealed significant age-dependent changes in RBC fatty acid profiles warranting separate pediatric and adults RIs. Model-based clustering and the iterative application of the Harris-Boyd method were successfully used to establish RBC fatty acid RIs for an objective assessment of long-term nutritional status in pediatric and adult populations.