Predicting Unbound Phenytoin Concentrations: Effects of Albumin Concentration and Kidney Dysfunction

Several methods are available to predict unbound (free) phenytoin concentrations in patients with hypoalbuminemia; however, predictive methods have not been evaluated in patients with concurrent hypoalbuminemia and kidney dysfunction or in patients with mild to moderate (estimated glomerular filtration rate [eGFR] 30-90 ml/min/1.73 m² ) kidney dysfunction alone. Thus the objective was to evaluate the accuracy and precision of predictive methods to estimate free phenytoin concentrations in patients with varying albumin concentrations and/or kidney dysfunction.

Retrospective chart review.

Large academic medical center.

A total of 344 patients with free and total phenytoin, albumin, and serum creatinine concentrations obtained between November 2012 and May 2017.

Free phenytoin concentrations were estimated in patients without kidney dysfunction using the Winter-Tozer, Anderson, Kane, and Cheng equations. For the analysis in patients with eGFR lower than 90 ml/min/1.73 m² , free phenytoin concentrations were estimated using the Shiner-Tozer derivation with adjusted affinity coefficients (C = 0.15, 0.20, 0.25, and 0.30). For both analyses, accuracy of predictive methods was evaluated by P20, the proportion of estimations within 20% of the measured free phenytoin concentration. In 158 patients with normal kidney function/normal albumin concentrations, 73 with normal kidney function/hypoalbuminemia, or 47 with mild kidney dysfunction/normal albumin concentrations, the Anderson method had the highest accuracy (86%, 82%, and 92%, respectively) and highest precision compared with the other methods. In 47 patients with normal albumin concentrations and mild kidney dysfunction or 13 with moderate kidney dysfunction, the free fraction was unchanged, and total phenytoin concentrations accurately reflected free concentrations. In 17 patients with hypoalbuminemia and mild or 17 with moderate kidney dysfunction, the Winter-Tozer (67% and 50%, respectively) and the Anderson (56% and 67%, respectively) methods had the highest accuracy compared with other methods with significantly lower accuracy compared with patients with normal kidney function. In the 14 patients with severe kidney dysfunction and hypoalbuminemia, none of the coefficients had a P20 accuracy greater than 45%.

In patients with normal albumin concentrations, with or without mild or moderate kidney dysfunction and not receiving a protein-binding displacer, the free fraction of phenytoin is unchanged, and it is not necessary to measure a free phenytoin concentration. Free phenytoin concentrations should be measured directly in patients with hypoalbuminemia and kidney dysfunction.