Regression-based reference limits for urinary amino acids in a pediatric population.

Abstract

Age-related variation of the urinary excretion of amino acids has been widely reported. Instead of the conventional reference intervals, continuous reference limits might be a more practical approach during periods of rapid changes such as infancy or childhood. Regression-based reference limits have been produced for the urinary excretion of 23 amino acids analyzed in 148 healthy individuals from 0 to 12 years of age. Urinary excretion was modeled as a function of age by two parametric procedures: the polynomial and the piece-wise linear regression methods. Likewise, age-specific standard deviation was estimated by the regression of the absolute values of the residuals on age. Residual analysis was employed to select the best-fitting model and the 95% reference limits, and their 95% confidence intervals were calculated and plotted against the raw data. The urinary excretion of 19 amino acids decreased rapidly during the first year of life with a further slow decline thereafter. These amino acids fit better the piecewise model. The other four amino acids displayed a steady decrease in the urinary excretion from birth to puberty, and the excretion patterns fit better the quadratic or linear model. Fifteen amino acids showed a significant change in standard deviation with age. Regression-based reference limits differed consistently from conventional limits (<3 years) and narrower confidence intervals were obtained throughout the entire period studied. Avoiding partitioning gave rise to smoothly changing limits. Therefore, such alternative way of presenting amino acid reference limits may facilitate the follow-up of patients with inborn errors of amino acid metabolism.