Even though therapeutic drug monitoring of fluorouracil was proven by many studies to have advantages over the commonly used body surface area-based dosing, it still struggles to find its place in clinical routine.
The aim of the current study was to reassess the application of a previously developed population pharmacokinetic model for 5-FU dose individualization and to further analyze pharmacokinetically-guided dose adjustment of fluorouracil during continuous intravenous infusion in colorectal cancer patients. First, an external validation of the population pharmacokinetic model was performed. Then, potential covariates affecting the interindividual variability of fluorouracil were analyzed. Finally, the agreement between two different AUC estimation methods (the “rectangle method” and the population pharmacokinetic model), being used for pharmacokinetically-guided 5-FU dose adjustment, was assessed. 
The external evaluation of the population pharmacokinetic model was performed using a validation dataset of 62 patients with single sampling and multiple 5-FU therapy cycles. Some minor model adjustments were necessary to enable a proper parameter estimation. After these adjustments, the model described the validation population reasonably well. Predictive performance was characterized by a reasonably low bias but also a low precision. During the stepwise covariate modeling process, both BSA and duration of infusion were included as covariates on clearance, resulting in a reduction of the IIV in CL from 20.2% to 14.8%. 
AUC estimation using the “rectangle method” and the PK model showed a lack of agreement dependent on the AUC value. The more the average AUC value differed from the population mean, the larger was the difference between the two estimates. However, application of the population pharmacokinetic model showed advantages over the “rectangle method” in dealing with outliers and for therapy regimens with additional 5-FU bolus. Therefore, it may be beneficial for TDM compared to the “rectangle method”. 
In conclusion, the adapted population PK model can facilitate individual dose adaptation of 5-FU in colorectal cancer patients. Future studies should aim at improving the precision of the model and developing easy-to-use software for wider clinical application.