Therapeutic drug monitoring: applying the ‘Goldilocks Principle’ to clinical pharmacology

An old joke quips that the definition of a toxicologist is a pharmacologist who does not know the dose. Beneath the internecine rivalry, and academic humor lies an important truth: pharmacology is the study of the interactions between ligands and receptors – and the extent of that interaction (and therefore the biological response) depends upon the concentration of the ligand. Too low a concentration results in therapeutic failure, too high a concentration results in toxicity – either by having too large an effect on the intended target of the drug, or due to off-target effects. The process of drug discovery and development takes this important consideration into account and aims to design drugs that have a wide therapeutic window and predictable pharmacokinetics. When used as medicines, these drugs can usually be administered according to fixed dosing schedules and require little to no monitoring beyond routine follow-up of the patient. Nevertheless, many important drugs have narrow therapeutic windows, meaning that accurate dosing is essential to ensure that treatment is effective, and to avoid toxicity. Furthermore, some classes of patients (particularly children, older adults, and those with co-morbidities) are often more susceptible to variability in the response to medicines. Therefore, particular care is needed in these situations to ensure treatment doses are appropriate to the patient and the situation. This can be achieved through therapeutic drug monitoring (TDM) ‘a multidisciplinary clinical specialty aimed at improving patient care by individually adjusting the dose of drugs for which clinical experience or clinical trials have shown it improved outcome in the general or special populations’ [2]. Whilst the concept of TDM is not new, the field is rapidly evolving owing to developments in technology and clinical practice, and thus it is fitting that this special issue of Expert Review of Clinical Pharmacology is focused on the topic. Recent developments in TDM include new ways of measuring drug concentrations in the body. Laboratoryanalysis of blood samples taken via phlebotomy can give very precise and accurate determinations of plasma concentrations of drugs. However, the process is invasive for the patient (not ideal for routine care), and inevitably involves a delay between taking the sample, and getting the result (not practical in emergency situations). Therefore, advancements in techniques to measure the concentrations of drugs in the body are welcome. In this issue, Touw provides a comprehensive update as to the status of saliva sampling methods, as an alternative to blood in TDM [3]. Owing to its noninvasive nature, saliva sampling is likely to be much more acceptable to patients and may increase the range of situations in which TDM is feasible. Challenges to saliva sampling include problems of sample treatment and contamination (saliva may be viscous and contains micro-organisms), and the fact that whilst blood contains both protein-bound and free drug, only the free drug can be detected in saliva. As such saliva monitoring is unlikely to replace blood monitoring in all TDM situations, but it represents an important area of development in the field [3]. Where it is necessary to use blood in drug monitoring, the process can be made less invasive by using smaller volumes of blood, and by employing capillary sampling rather than venipuncture to obtain the sample. In a comprehensive review, Muller et al. describe the benefits and challenges associated with the use of dried blood-spot sampling in TDM [4]. TDM has aways drawn on models, such as physiologicallybased-pharmacokinetic (PBPK) techniques, to understand the relationship between an administered dose of drug, and the concentration achieved at the site of action. Machine learning, artificial intelligence, and improved modeling techniques allow for a more sophisticated use of a greater number of variables to make better predictions about the effects of specific doses of drugs on individual patients. In this issue, Yang et al. describe a machine-learning model that can be used to predict plasma concentrations of quetiapine (an antipsychotic drug also used in depression) [5]. Quetiapine therapy is associated with adverse effects, such as drowsiness, hypotension, and anticholinergic adverse effects, through its action on histaminergic, adrenergic, and muscarinic receptors [5]. By preventing supra-therapeutic concentrations of drugs, TDM could reduce the incidence and severity of adverse effects, and therefore improve adherence to therapy and patient outcomes.