Challenges associated with test dose pharmacokinetic predictions of high dose melphalan exposure in patients with multiple myeloma

Aim To evaluate the accuracy of melphalan test dose pharmacokinetic (PK) predictions of the subsequent high dose (HDM) area under the concentration-versus-time curve (AUC) and to identify sources of prediction error (PE). Methods A prospective multicentre PK study was conducted in 40 myeloma patients of median age 60 (range:35–71) years using a 20 mg/m 2 test dose administered 1–3 days prior to HDM (predominantly 180 mg/m 2 ). PK data were collected post the test and high doses to compare predicted versus actual AUCs determined using the trapezoidal rule. Test and high dose infusion concentration, volume and duration and the time from preparation to infusion were compared using the paired Wilcoxin rank sign test. The impact of Melphalan administration parameters on PE was evaluated using the Mann–Whitney test. The predictive capacity of a previously published population PK (PopPK) model was also examined. Results Predicted HDM AUC was within 15% of the observed values in only 63% of patients when analysed using the trapezoidal rule and 70% of patients using PopPK. Test dose infusion concentration, volume, duration and time from preparation to infusion were significantly lower than for HDM ( p < 0.005). Test dose administration within 15 min of reconstitution ( n = 5) was associated with significantly lower PE than administration times of 16–60 min ( n = 22), p < 0.05. Test and HDM infusion concentrations were lower in patients with large PE (> ± 15%), but the differences were not significant ( p = 0.078, 0.228, respectively). Conclusion Test dose PK has the potential to predict subsequent HDM exposure to achieve a target AUC once melphalan administration parameters are optimised to account for stability issues in the formulation.