Dosage adjustment of high-dose methotrexate using Bayesian estimation: a comparative study of two different...

Bayesian estimation (BE) of pharmacokinetic parameters enables the clinician to adjust the dosage of high-dose methotrexate (HDMTX) to correct the inter- and intraindividual variation of concentrations that are responsible for severe toxicity. In this study of 672 HDMTX infusions, we validated an approach that consisted of reaching as nearly as possible a theoretical concentration of 5.10(-4) M or 10(-3) M at the end of an 8-h infusion by adjusting, when necessary, the dosage at the 6th h. The BE of the clearance was compared with that obtained by maximum likelihood estimation (MLE), which was used as reference. BE performance was evaluated by calculating the bias and precision that indicated an overestimation of clearances obtained by BE compared with the higher clearance of the MLE in the group of patients receiving the higher dose (15 and 37.9%). Linear regression analysis of clearance obtained by BE and MLE showed a correlation (p < 0.0001) in both groups of patients with a closer link in those with the lower dose. However, in current clinical practice the important point is to obtain MTX concentration that is as close as possible to the desired concentration. Adjustments were evaluated by comparing the obtained concentrations with the desired theoretical concentration. There was no bias and precision was satisfactory in both groups of patients (15 and 12%, respectively, for 5.10(-4) M and 10(-3) M). This method makes it possible to limit the inter- and intraindividual variations of concentrations. As a result, severe complications were essentially nonexistent and were never life threatening.