Li Zhihong, Gao Weiqi, Liu Guifen, Chen Weihong
Department of Pharmacy, Shanxi Bethune Hospital (Shanxi Academy of Medical Sciences); and.
Department of Health Statistics, School of Public Health, Shanxi Medical University, Taiyuan, China.
Ther Drug Monit. 2020 Aug;42(4):610-616. doi: 10.1097/FTD.0000000000000749.
In patients with hypoalbuminemia after craniotomy, total serum concentrations of valproic acid (VPA) may provide poor clinical insights, owing to saturated protein binding and increased unbound fractions. However, very few clinical laboratories routinely analyze free concentrations of the drug. The aim of this study was to develop a model to predict serum-free and cerebrospinal fluid (CSF) levels of VPA based on its total concentration and to investigate the model's applicability.
Total serum and CSF concentrations of VPA in 79 patients were measured using a validated immunoassay between January 2015 and December 2015. The demographic, clinical, and laboratory information of patients were retrieved from medical records. A multiple linear regression analysis was adopted to determine the potential variations and establish the functional relationship between CSF concentration and significant clinical factors.
Based on the stepwise multiple linear regression analysis performed using the natural logarithm of the concentration of VPA in the CSF as the dependent variable, serum concentrations of VPA (X1, β' = 0.844), serum albumin concentration (X2, β' = -0.393), and CSF protein concentration (X3, β' = 0.098) were identified as the 3 variables that significantly predicted the dependent variable: (Equation is included in full-text article.), with a coefficient of determination (R) of 0.874. As the CSF protein level is often unavailable, the model was redefined to include 2 variables-serum concentrations of VPA (X1, β' = 0.840) and serum albumin concentration (X2, β' = -0.359): (Equation is included in full-text article.), with R = 0.813.
Based on total VPA and serum albumin concentrations, we developed a model to predict serum-free and CSF levels of VPA. This model is useful for correcting dose adjustment in patients with hypoalbuminemia after craniotomy.
开颅术后出现低蛋白血症的患者，由于丙戊酸（VPA）的蛋白结合饱和及游离部分增加，血清总浓度可能无法提供可靠的临床信息。然而，很少有临床实验室常规分析该药物的游离浓度。本研究旨在建立一个基于VPA总浓度预测其血清游离浓度和脑脊液（CSF）浓度的模型，并探讨该模型的适用性。
2015年1月至2015年12月期间，采用经过验证的免疫分析法测定了79例患者的血清和脑脊液中VPA的总浓度。从病历中获取患者的人口统计学、临床和实验室信息。采用多元线性回归分析确定潜在变量，并建立脑脊液浓度与重要临床因素之间的函数关系。
以脑脊液中VPA浓度的自然对数作为因变量进行逐步多元线性回归分析，结果显示VPA血清浓度（X1，β' = 0.844）、血清白蛋白浓度（X2，β' = -0.393）和脑脊液蛋白浓度（X3，β' = 0.098）是显著预测因变量的3个变量：（方程包含在全文中），决定系数（R）为0.874。由于脑脊液蛋白水平通常难以获得，该模型重新定义为包含2个变量——VPA血清浓度（X1，β' = 0.840）和血清白蛋白浓度（X2，β' = -0.359）：（方程包含在全文中），R = 0.813。
基于VPA总浓度和血清白蛋白浓度，我们建立了一个预测VPA血清游离浓度和脑脊液浓度的模型。该模型有助于校正开颅术后低蛋白血症患者的剂量调整。
