Therapeutic drug monitoring databases for postmarketing surveillance of drug-drug interactions.

Drug-drug interactions can be associated with patient morbidity due to either increased toxicity or a potentially ineffective concentration. Because interactions cannot always be anticipated during drug development and actual patients receiving a drug for therapeutic use often differ from those included in clinical trials, postmarketing surveillance is essential. Therapeutic drug monitoring (TDM) databases offer a unique opportunity in this respect. Prerequisites for TDM databases to provide valid information in a pharmacoepidemiological perspective include the following: precise description of exposure to the potentially interacting drugs; measurement of parent compound and active metabolites through accurate and precise analytical techniques; documentation of relevant patient characteristics that may act as confounding factors (e.g. gender, age, smoking habits); repeated assessments over time if possible; and sound pharmacokinetic framework for data selection, analysis and interpretation. The contribution of TDM to the documentation of drug-drug interactions takes advantage of different possible study designs, discussed on the basis of recently published studies. The single case report plays an important role as an alert signal. It is illustrated for a patient on long-term treatment, who displayed an unexpectedly high clozapine concentration after the introduction of ciprofloxacin comedication. The prospective on and off comedication panel study shows advantages in terms of carefully selected inclusion criteria and control of treatment modalities. A study of the thioridazine-fluvoxamine interaction is presented, with patients followed on thioridazine monotherapy, after introduction of fluvoxamine and after its discontinuation. The main advantage of the retrospective large-scale TDM database screen is representativeness of patients actually treated, whereas drawbacks are related to quality of data and suitability for valid interpretation. Such an approach is illustrated by a review of data collected over 10 years of routine TDM that allowed documenting induction of nortriptyline metabolism by carbamazepine and inhibition by several phenothiazines. Finally, population pharmacokinetics is well suited to observational data collected for TDM purpose, provided quality is ascertained. Focus is placed on interindividual variability and relationship between pharmacokinetic parameters and patient characteristics, including comedication. The population approach is discussed with respect to a study that documented a 32% increase of haloperidol clearance associated with anticonvulsant comedication, in addition to effects of age and bodyweight. Among factors to consider for improved effectiveness in the use of TDM databases for postmarketing surveillance of drug-drug interactions, integration of efficacy and safety data in future studies and communication of expert recommendations to prescribing physicians are essential.