Bayesian therapeutic drug monitoring software: past, present and future

摘要

Therapeutic drug monitoring (TDM) is a powerful tool to individualize drug therapy. TDM is justified when inter-patient pharmacokinetic (PK) variability exceeds the limits of safe and effective variability of the drug, and where downstream markers either provide a poor gauge of drug effect, or they are clinically devastating (e.g., seizures) [1]. These criteria are fulfilled by many drugs, including selected antimicrobials, immunosuppressants, psychotropics and anti-epileptics [2-6]. Bayesian TDM methods formally incorporate information about a drug's population PK, individual patient variables and the measured concentrations. From these, individual PK parameters are estimated to optimize dose-regimens to attain a chosen target. The advantages of the Bayesian approach over alternative methods have been demonstrated in numerous studies, mostly with respect to reduced toxicity or cost [2,7-9]. Yet, Bayesian TDM has not been widely adopted in clinical practice [10]. This reflects the challenges of TDM in general: lack of well-defined exposure targets, resource limitations and lack of required expertise. Furthermore, there remains a dearth of randomized controlled trials comparing TDM based dose-optimization to alternative methods of dosing. Bayesian TDM is perceived as complicated as it uses specialized software and PK concepts, which may be unfamiliar to many clinicians.