Factors to influence the accuracy of albumin adjusted free valproic acid concentration

Abstract

Therapeutic drug monitoring (TDM) is recommended during treatment with valproic acid (VPA), as is the measurement of free VPA concentration (MfVPA). However, MfVPA is unavailable in many institutions. Based on the highly protein-bound characteristics of VPA, an albumin-adjusted formula has been proposed to predict free VPA concentration (PfVPA). Nevertheless, the factors affecting the accuracy of this formula remain unknown, as does the concordance between MfVPA and PfVPA. Adult patients receiving VPA and undergoing TDM were enrolled. Free and total serum concentration (TVPA) were categorized as subtherapeutic, therapeutic, or supratherapeutic based on the reference range of 5-15 and 50-100μg/mL, respectively. Concordance was defined as MfVPA and PfVPA, or MfVPA and TVPA, falling within the same category. Multivariate logistic regression with generalized estimating equation was adopted to identify factors affecting concordance, and the receiver operating characteristic curve was applied to determine the cutoff values of predictors. A total of 98 data points from 51 participants were included for analysis. The concordance of MfVPA and PfVPA, and MfVPA and TVPA, was 72% and 44%, respectively. Blood urea nitrogen (BUN) (0.97 [0.95-0.99], P=0.01) and TVPA (0.97 [0.95-0.99], P=0.02) had a significant influence on the concordance of MfVPA and PfVPA. The cutoff values of TVPA and BUN for the accuracy of the albumin-adjusted formula were 56.4μg/mL and 51.05mg/dL, respectively. If MfVPA is not available, the albumin-adjusted formula should be applied before VPA dosage adjustment when TVPA is<56.4μg/mL and BUN is<51.05mg/dL.