Assessment of the relative dose-response test based on serum retinol-binding protein instead of serum retinol in determining low hepatic vitamin A stores

Background: The relative dose-response (RDR) test, which measures the percentage of change in serum retinol concentration in response to an oral vitamin A (VA) dose, is a functional reference method to assess low hepatic VA stores. However, problems due to the relative instability of retinol, which is measured in the traditional RDR test, could be circumvented if retinol-binding protein (RBP), a more stable marker of VA, could be measured instead of retinol to provide the RDR value. Objective: The objective was to compare classification of VA status assessed by retinol-RDR with that assessed by using RBP-RDR. Design: With the use of serum samples from 57 lactating women in northern Kenya collected in August-September 2006, we assessed the accuracy of RBP-RDR in predicting low hepatic VA stores through receiver operator characteristic (ROC) analysis using retinol-RDR values as the gold standard. By using regression analysis, we explored the effects of 1) body mass index (BMI) on RBP-RDR performance and 2) the oral VA dose on the retinol-RBP molar ratio. Results: The classificatory accuracy of RBP-RDR was low to moderate (n = 50; ROC area: 0.56-0.72) depending on the cutoffs used. RBP-RDR systematically overestimated VA deficiency with higher BMI, although it was superior to a single measurement of serum retinol. The discrepancy between RBP-RDR and retinol-RDR appears to originate in a retinol concentration-dependent alteration of the retinol-RBP molar ratio triggered by the oral dose. Conclusions: RBP-RDR has the potential to serve as a moderately accurate surrogate measure of retinol-RDR if the variation associated with BMI is understood and adjusted. Further studies should clarify the dynamics of the retinol-RBP molar ratio and its link to RBP-RDR performance. Am J Clin Nutr 2009; 90: 217-24.