Utility, promise, and limitations of liquid chromatography‐mass spectrometry‐based therapeutic drug monitoring in precision medicine

Abstract

Therapeutic drug monitoring (TDM) is typically referred to as the measurement of the concentration of drugs in patient blood. Although in the past, TDM was restricted to drugs with a narrow therapeutic range in order to avoid drug toxicity, TDM has recently become a major tool for precision medicine being applied to many more drugs. Through compensating for interindividual differences in a drug's pharmacokinetics, improved dosing of individual patients based on TDM ensures maximum drug effectiveness while minimizing side effects. This is especially relevant for individuals that present a particularly high intervariability in pharmacokinetics, such as newborns, or for critically/severely ill patients. In this article, we will review the applications for and limitations of TDM, discuss for which patients TDM is most beneficial and why, examine which techniques are being used for TDM, and demonstrate how mass spectrometry is increasingly becoming a reliable and convenient alternative for the TDM of different classes of drugs. We will also highlight the advances, challenges, and limitations of the existing repertoire of TDM methods and discuss future opportunities for TDM‐based precision medicine.

1. INTRODUCTION

Drug pharmacodynamics is, whenever possible, monitored by directly measuring physiological indices of therapeutic response. For example, blood glucose levels are measured to adjust insulin doses for diabetic patients. For many drugs, however, methods for the direct measurement of a drug's effect are either not available or not sufficiently sensitive. In these cases, the percentage of a drug, or its active moiety that actually reaches the systemic circulation and becomes available to its target, known as the drug bioavailability, has become the most accepted indicator of efficacy. This explains the importance of therapeutic drug monitoring (TDM), which is the analysis, assessment, and evaluation of the circulating concentrations of drugs in serum, plasma, or whole blood in order to ensure that the patient's drug concentrations are within the established therapeutic range for the respective indication. ¹ , ²

The therapeutic reference range (or therapeutic window) of a drug in blood is defined as the concentration range below which the therapeutic response is unlikely to occur and above which drug tolerability decreases without improving therapeutic response, with excessive dosing eventually leading to toxic effects (Figure 1A). ³ , ⁴ It is, therefore, important to ensure—at the time of drug administration—that the patients' blood levels are still within the therapeutic window, that they will remain within it, and that a next dosage will be administered once the drug reaches its “trough level = predose sample C₀” (the lowest level where a drug should be, before the next dose is administrated). There are different methods for determining the dose according to the safety and efficacy studies conducted for each drug. Many drugs are dosed based on body weight where there is an acceptable relationship between drug response and drug dose, although others are given in fixed doses, with the dosage recommendations typically being determined by dose–response studies performed during Phase II of drug development ⁵ along with a concentration range accounting for interindividual variability. ⁶ For adults, fixed‐dosing regimens can be used for some drugs (such as cyclosporine microemulsion, recombinant activated Factor VII, and epoetin α), whereas weight‐based dosing is commonly used in clinical practice. ⁷ In pediatrics, weight‐based dosing or body surface‐area‐based dosing are commonly used (e.g., for methotrexate, prednisone, growth hormone, and 13‐cis‐retinoic acid). In some cases, however, dosing based on age is more appropriate (e.g., for intravenous busulfan and dalteparin). Although these dosing recommendations work well for most drugs and in most patients, it has been reported that certain drugs produce highly variable plasma levels in individual patients (Figure 1B) and that certain diseases such as renal and liver disease greatly affect drug metabolism and thus require TDM even for drugs that do not have a narrow therapeutic range. Clozapine, an anti‐psychotic drug used to treat schizophrenia, has shown to have an up‐to‐45‐fold interindividual variability in serum levels in patients fixed dose treatment ⁸ , ⁹ for many reasons such as age, gender, smoking habits, and pharmacological interactions with other drugs. ¹⁰ Moreover, even drugs that can be safely dosed based on age and/or weight may need close TDM in patients with renal and liver disorders to avoid adverse effects.

FIGURE 1.

The therapeutic reference range of a drug within which a drug response without major toxic side effects can be expected. (A) Above the maximum therapeutic concentration (MTC), there is an increased risk of toxicity, and below the minimum effective concentration (MEC), there is likely to be a lack of response. The peak concentration C _max is reached a certain time after the dose has been administered, and the trough level C ₀ is when the next dose should be administered. A representative AUC for the first 12 h after dosing (AUC_0–12h) is indicated. (B) Interpatient variability: An example of an interference with drug absorption, leading to a low C _max and low C ₀ (orange curve) compared with the ideal case (blue curve). An example of an interference with drug elimination (red curve), which is commonly caused by either kidney dysfunction or liver dysfunction. C _max is normal, and C ₀ is high after the first dose, and both do gradually increase with subsequent doses

Drug pharmacokinetics is therefore affected by many patient‐specific factors other than weight and age that are idiosyncratic (e.g., sex, diet, genetics, hydration state, body fat percentage, and renal and liver function), and this can lead to unpredictable effects. For example, one of the most important pathways involved in drug detoxification acts through the Cytochrome P450 (CYPs) enzyme superfamily that catalyzes oxidation, the most common phase I reaction involved in drug metabolism. Interindividual variation in the expression levels and/or activities of CYPs can translate into significant variations in CYP‐mediated drug metabolism. ¹¹ Moreover, it has been shown that CYPs can be affected by diet, such as the inhibition of CYP3A by grapefruit juice, leading to an increased bioavailability of some orally administered drugs. ¹² Despite the availability of methods for the genetic and phenotypic assessment of variation in the activity of such enzymes, ¹¹ a more practical, comprehensive, and straightforward approach is needed to adjust the dose of a drug for each patient. Ideally, this method would consider the activity of enzymes involved in the drug's metabolism, as well as all other factors that could affect the drug pharmacokinetics, whether these factors are known or not.

Drug dosage adjustment can be guided by TDM. This has been shown to accelerate the recovery of many patients, while also minimizing side effects and reducing healthcare costs because unnecessary and/or excessive dosing can be avoided. ³ Immunoassays, such as radioimmunoassays (RIAs), chemiluminescence immunoassays (CLIAs), or enzyme‐linked immunosorbent assays (ELISAs), are techniques that are commonly used for TDM. ¹³ , ¹⁴ However, significant imprecision has been reported for such antibody‐based methods, ¹³ , ¹⁵ , ¹⁶ , ¹⁷ because they can suffer from low specificity ¹⁸ , ¹⁹ which can lead to interference and false‐positive results. ²⁰ , ²¹ Compared with immunoassays, drug monitoring using mass spectrometry (MS) is a more flexible and more specific technology and enables the multiplexed quantitation of several targets simultaneously. Multiplexed assays are extremely relevant, because of the increasing treatment not only for patients with multiple drugs, including polypharmacy (i.e., the regular use of at least five drugs) but also for patients with unclear medical history. Patients who are typically treated with multiple drugs include hepatitis C virus patients, patients with autoimmune disorders, psychiatric patients (including drugs‐of‐abuse), and cancer patients.

MS has been used for a broad range of applications, from fundamental research to clinical assays. ¹⁷ , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ Moreover, a wide range of TDM assays have been developed as laboratory‐developed tests (LDTs) and are currently being used by major clinical diagnostic laboratories in the United States (see Figures 2 and 3 and Table S1).

FIGURE 2.

Therapeutic drug monitoring (TDM) assays currently being used by major clinical laboratories in the United States. (A) Methods used. GC, gas chromatography. (B) Drug classes monitored ²⁷ , ²⁸

FIGURE 3.

Principles of some of the most‐widely used methods for therapeutic drug monitoring (TDM). (A) Liquid chromatography–mass spectrometry (LC–MS). The target drug is directly measured in plasma and quantified using internal standards. Protein drugs are usually proteolytically digested prior to their quantitation using “proteotypic” peptides. (B) Enzyme‐linked immunosorbent assay (ELISA). In a competitive ELISA, a small molecule drug and an enzyme‐conjugated form of the drug are captured through an immobilized antibody. After washing, the enzyme reaction is used as colorimetric readout: The signal is inversely proportional to the original concentration of the analyte. (C) Bioassay (diffusion). A bacterial strain that is susceptible to the antibiotic to be measured is grown on defined wells in an agar plate. Equal volumes for samples and standards are then added to these wells, and after ca. 24 h of incubation, the diameters of the “inhibition zones” are measured as an indirect readout of the concentration of the antibiotic. (D) Drug concentrations are determined using external or internal calibration curves

2. WHEN IS TDM MOST BENEFICIAL?

Personalized therapies can be adjusted based on the concentration of a therapeutic drug in the patient's blood. ³ , ²⁹ , ³⁰ , ³¹ Thus, the use of TDM has increased greatly in this current era of precision medicine, despite having previously been limited to certain drug groups and certain patients. Initially, TDM was used in cases where (і) the therapeutic dose yielded responses that were either absent or poor or where side effects indicated that a patient could not tolerate the dose; (іі) out‐patient compliance with the prescribed dosage was uncertain, (most often relevant for psychiatric patients); (ііі) combination therapies may lead to drug–drug interactions (DDIs) affecting the pharmacokinetics of both drugs; (iv) drugs with a narrow therapeutic range were used (such as lithium, a common medication for treating mental health diseases, that is normally used at serum concentrations of 0.6–1.2 mEq/L but becomes toxic at 1.5 mEq/L). ³²

To date, there is much evidence that supports the benefits of TDM on drug effectiveness and the minimization of side effects, especially for individuals who present with unusual pharmacokinetics (see Figure 1b), suffer from diseases that require constant medication (autoimmune disorders, epilepsy, mental‐health, cardiac conditions), or are in an especially vulnerable state (newborn, preterm infants, elderly, transplanted). Below, we discuss several patient groups for which TDM has been most beneficial.

2.1. Patients with organ transplants: Immunosuppressants

TDM of immunosuppressants is a fundamental part of transplant protocols, because immunosuppressants play a critical role in the acceptance of a newly transplanted organ by the recipient's immune system. ³³ The immunosuppressant's blood level should be precisely maintained to prevent organ rejection while at the same time avoiding oversuppression of the immune system, because this can lead to life threatening infections, ³⁴ or to carcinogenesis induced by viral infections that target tumor suppressive pathways in the absence of a fully functional immune system. ³⁵

The optimal blood levels of immunosuppressants for patients with transplants differ according to (і) the type of organ transplanted; (іі) the type of the therapy (i.e., monotherapy or combination therapy with other immunosuppressants); and (ііі) the aim of the treatment (induction or maintenance). Moreover, the transition from innovator to generic medications for cost‐saving purposes caused serious side effects, especially in pediatric solid organ transplant recipients due to significant variations in the total integrated area under the plasma drug concentration‐time curve (AUC), for example, the AUC_0–12h for the first 12 h after injection, making close TDM essential, ³⁶ Thus, for one of the most widely used immunosuppressants for organ transplantation, cyclosporine, a study reported a 16.7% drop of the AUC and a 13.1% drop of the 2‐h concentration after oral administration when switching from innovator drug to generic cyclosporine, ³⁷ indicating a reduced bioavailability of the generic drug, which could be associated with treatment failure (= organ rejection) or the need for more frequent dosing. For tacrolimus, another widely used calcineurin inhibitor (i.e., a drug that inhibits the T‐cells of the immune system), it was found that patients with identical C ₀ may have very different AUC_0–12h, which explains differences in treatment efficacy. Therefore, instead of using C ₀ alone to guide tacrolimus therapy, it is now recommended to use both the C ₀ and AUC_0–12 for recipient management, at least once within the early period after transplantation and at another time within the stable phase. ³⁸

2.2. Patients with autoimmune disorders

Autoimmune disorders cause the body's immune system to fight its own cells and/or tissues under largely unknown triggers. There are currently no cures available for autoimmune disorders, and life‐long treatment is often required to ease the symptoms. Some examples of autoimmune disorders include rheumatoid arthritis, inflammatory bowel disease, multiple sclerosis, type I diabetes mellitus, and psoriasis. Type I diabetes patients do not produce their own insulin and depend on a life‐long therapy—an example of a therapy that has already been personalized for the past 35 years. ⁶ The tumor necrosis factor (TNF) family of pro‐inflammatory cytokines plays a central role in the pathogenesis of several autoimmune disorders. ³⁹ , ⁴⁰ , ⁴¹ TNF‐alpha inhibitors are used to treat chronic immune disorders and cancer related inflammations ⁴⁰ , ⁴² and are generally well‐tolerated, but some patients show loss of response over time, due to multiple clinical, genetic, and immunopharmacologial factors. ⁴³ Moreover, serious adverse effects such as infections, lymphomas, congestive heart failure, or the generation of autoantibodies have been reported. ⁴⁴ Because these complications ultimately lead to treatment failure, ⁴³ TDM of patients at risk for non‐response is imperative, as it allows physicians to determine whether there is a need for drug optimization (dose escalation), concomitant immunosuppression (to prevent production of anti‐drug antibody [ADA]), or a change of medication. ⁴⁵ , ⁴⁶

Adalimumab and infliximab are therapeutic monoclonal antibodies (mAbs) that target TNF and are prescribed for the treatment of rheumatoid arthritis. Compared with small molecule drugs, the pharmacokinetics and pharmacodynamics of mAbs are more complex. mAb drugs have to be administered intravenously, subcutaneously, or intramuscularly, rather than orally, due to their large size (~150 kDa) and the resulting low cell permeability. ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ In the body, their distribution is limited to the vascular and interstitial fluids, and the clearance of mAbs is distinctly different from that of small‐molecule drugs. Although small‐molecule drugs undergo renal and/or hepatic clearance, mAbs are too large to be cleared from the body through these elimination routes; thus, they are eliminated from the body by either biliary excretion or intracellular catabolism by lysosomal degradation. ⁴⁹ , ⁵¹ Most mAbs are administrated either in fixed doses and time intervals, or based on body size. ⁵¹ Such strict dosing schedules, however, do not account for the interpatient variability of mAb pharmacokinetics. ⁵² , ⁵³ , ⁵⁴ This variability can derive from many factors, including sex, body weight, the expression level of the mAb target, the levels of covariates of mAb clearance in blood (such as, for some mAbs, the levels of albumin or C‐reactive protein), and immunogenicity leading to the formation of ADAs. ⁴⁵ , ⁵¹ , ⁵⁵ Although adalimumab is administered at 40 mg every second week, infliximab is administered at 5 mg/kg of body weight at Weeks 0, 2, and 6 and then every 8 weeks.

Indeed, studies have shown that higher plasma concentrations of infliximab correlate with better disease control in inflammatory bowel disease patients in maintenance stage. ⁵⁶ , ⁵⁷ , ⁵⁸ Moreover, patients who have high plasma concentrations of adalimumab or infliximab but show no response or poor response to the treatment can benefit from a switch to a different TNF inhibitor. ⁵⁹ Thus, TDM can help to better guide therapeutic decisions and prevent treatment failure.

2.3. Patients receiving neuropsychiatric drugs

TDM for neuropsychiatric drugs has been well‐studied and continuously evaluated. This is particularly true because these drugs change the brain's biochemistry and affect the central nervous system and autonomic functions that are necessary for living. Accordingly, the consensus guidelines for TDM in neuropsychopharmacology classify neuropsychiatric drugs based on the necessity for TDM to assess toxicity into four categories from “highly recommended or obligatory” for many drugs, to “recommended,” “useful,” or “probably useful” for another 139 agents. The drug group with the highest level of recommendation for TDM (level 1) includes (і) the mood stabilizers carbamazepine and valproate with the exception of lithium where TDM is obligatory throughout the treatment; (іі) most tricyclic antidepressants and their metabolites such as amitriptyline and imipramine; (ііі) antipsychotics such as haloperidol and thioridazine; (іv) the anticonvulsants phenobarbital and phenytoin; (v) the selective serotonin reuptake inhibitor (SSRI) citalopram. Because of interpatient variability observed for all the level 1 drugs, TDM is obligatory both during initial dosage optimization and after any change in dosage. ³

Besides avoiding toxicity or side effects, TDM is also used in neuropsychiatric pharmacotherapy to meet the growing demand for personalized therapies. Individual patients need different therapeutic concentrations to achieve an optimal response, and this concentration can deviate from the therapeutic reference range. The use of TDM as part of a clinical routine to monitor the response to neuropsychiatric therapeutics is very complex. It involves important measurements of each drug in addition to the therapeutic reference range. These include the drug half‐life (t _1/2), laboratory alert levels, and the metabolite‐to‐parent ratios (MPRs). Laboratory alert levels are used as an alarm signal, and the prescribing doctor has to be informed immediately about the need to consider a dose reduction when clinical signs of adverse drug reactions (ADRs) are present or suspected. MPRs are calculated by dividing the concentration of a major metabolite by the concentration of the parent compound. They provide valuable information regarding the enzymatic activities involved in the metabolism of a patient (e.g., an ultrarapid or poor metabolizer), patient compliance, and DDIs on a pharmacokinetic level.

This is, for example, important for patients with seizure disorders who are dependent on antiepileptic drugs for prevention of seizures and for minimization of the negative effects on general well‐being. ⁶⁰ For these patients, the type and severity of epilepsy, pathological and psychological states, as well as DDIs should be considered, and dosage regiments should be individualized accordingly. ⁶⁰ Valproic acid (VPA) is a small‐molecule drug that has been approved for the treatment of several types of epilepsy, bipolar disorder, and migraine prophylaxis. VPA plasma concentrations do not increase linearly with VPA dosing, because of plasma‐protein binding. ⁶¹ At high VPA doses, plasma proteins are saturated by VPA and the concentration of free (unbound) VPA concentration increases. Because VPA can induce its own metabolism, the higher the share of free VPA is, the faster VPA is metabolized and the lower its plasma concentration as well as its therapeutic efficacy become. Any co‐administrated drug that competes for binding with plasma proteins can therefore affect the free VPA concentration. In this case, TDM is utilized as a tool for efficacy assessment and dose adjustment, ⁶² and in cases of unpredictable relationship between dose and serum concentration, it is important to individualize and maintain therapies using TDM. ⁶³

2.4. Patients with CVD

Drugs to treat cardiovascular disease (CVD) are among the most commonly prescribed medications. For CVD, TDM was initially restricted to some antiarrhythmic medications with narrow therapeutic ranges, such as digoxin and digitoxin, where selecting the dose based on existing blood levels greatly decreases the incidence of digitalis intoxication. For class I antiarrhythmic drugs and amiodarone, TDM prevents many ADRs, especially the extracardiac ADRs such as hypoglycemia due to cibenzoline. ⁶⁴

For some other medications, TDM of the concentrations of both the drug and its active metabolite is highly recommended. For example, procainamide, a classic antiarrhythmic drug, and its metabolite N‐acetylprocainamide (NAPA) are both monitored. NAPA, which has antiarrhythmic effects, is eliminated mainly by the kidney, and its level varies according to the activity of the enzyme that converts procainamide to NAPA, N‐acetyltransferase (NAT). Therefore, patients who are not monitored for NAPA may suffer from ADRs despite having low procainamide concentrations, as NAPA may increase to a level that is higher than that of procainamide in patients with high NAT activity and renal dysfunction. ⁶⁵

Regarding antimicrobial medications used for the treatment of infective endocarditis, such as vancomycin, aminoglycosides, and teicoplanin, there are strict guidelines that recommend TDM with certain trough concentrations for each drug at specific time points from the beginning of the treatment. Moreover, for vancomycin and aminoglycosides, the AUC/minimum inhibitory concentration (MIC) ratio is found to be more useful to ensure the clinical and bacteriological efficacy, which is extremely important in such a serious infection to prevent serious cardiac and systemic complications. In addition, this “careful TDM” (compared with the solely C ₀‐based TDM) is used to avoid serious side effects known for these antibiotics such as nephrotoxicity. ⁶⁶

More recently, TDM has gained more applications in cardiovascular therapy, as most patients require more than just one medication. This combinatory treatment carries more risk of ADRs as a consequence of more DDIs. Therefore, many commercial laboratories use high‐performance liquid chromatography (HPLC) and LC–MS/MS assays for simultaneous detection and monitoring of a broad range of cardiovascular drug classes. For example, Dias et al. have developed and evaluated an LC–MS/MS method for 34 commonly prescribed cardiovascular drugs or drug metabolites using 100 μl of serum or plasma. ⁶⁷ Drug classes monitored by this assay include: anticoagulants, angiotensin converting enzyme inhibitors (ACEIs), angiotensin II receptor blockers (ARBs), beta blockers, calcium channel blockers, diuretics, statins, and vasodilators, as well as digoxin, fenofibrate, and niacin.

2.5. Patients with cancer

Cancer is one of the main causes of death worldwide, with 14 billion new cases per year. ⁶⁸ The medication given to cancer patients to combat tumors is typically cytotoxic, thus causing a large number of severe side effects. ⁶⁹ Commonly used drugs in cancer treatments are small‐molecule inhibitors (SMIs) and mAbs, while newly designed interfering RNA molecules, and microRNAs are very promising treatments. ⁷⁰ Considerable effort has been made to better predict which drugs would cause better responses and also to define concentration‐effect relationships, as large interindividual pharmacokinetic variability is influenced by the pharmacogenetic background of the patient (e.g., cytochrome P450 and ATP‐binding cassette transporters polymorphisms) as well as by the genetic heterogeneity of the patient's drug targets. ⁷¹ , ⁷² Thus, for some cancers, oncologists are now able to predict which drugs to use for a more personalized treatment based on the tumor's genome. However, factors such as age, gender, smoking habits, and pharmacological interaction with other drugs can also affect drug metabolism. Sunitinib, nilotinib, dasatinib, sorafenib, and lapatinib are some examples of SMIs that have been used for cancer treatment for over a decade and which are characterized by interindividual PK variability, risk of DDI, leading to suboptimal responses and sometimes failure of treatment. ⁷³ , ⁷⁴ Therapeutic mAbs have the ability to kill tumor cells, and also engage the immune system to develop response against the tumor. ⁷⁵ TDM methods have been developed for many cancer drugs. ⁷⁶ , ⁷⁷ , ⁷⁸ , ⁷⁹ , ⁸⁰ , ⁸¹ , ⁸² and an effort to better define concentration‐effect relationships would certainly be beneficial for patient treatment. ⁷⁰ , ⁷¹ , ⁷² , ⁸³ Another important reason for performing cancer medication TDM is to ensure patient adherence, which is crucial for a successful outcome and preventing recurrence, minimal toxicity, and reduced healthcare costs when patients are taking oral medications. ⁸⁴ Nonadherence has been frequently reported, for a variety of reasons. ⁸⁵ , ⁸⁶ , ⁸⁷ , ⁸⁸

2.6. Infants and preterm neonates

Hospitalized infants and preterm neonates face specific issues regarding prescribed drugs due to the lack of studies in this population; therefore, they are exposed to drugs that are prescribed off‐label. ⁸⁹ , ⁹⁰ , ⁹¹ Moreover, medication dosages for infants differ greatly from those for older children. For example, for the commonly used antibiotic vancomycin, the dose for preterm neonates (<1.2 kg) is 15 mg kg⁻¹ day⁻¹, where for infants (>1 month of age) it is 40–45 mg kg⁻¹ day⁻¹. ⁹¹ Reasons for the complicated dosing guidelines and high pharmacokinetic variability of infants are (i) constantly changing proportions of body water, fat, and protein that affect drug distribution and (ii) the maturation of major organ systems affecting drug elimination, during infancy and childhood, for example, those affecting renal elimination mechanisms. ⁸⁹ , ⁹² Moreover, the pathophysiology of some diseases and pharmacologic receptor functions change during infancy and childhood and differ from those of adults.

TDM may ensure the safety of infants and preterm neonates who need medication at this vulnerable phase of their lives, especially when they are being treated in Neonatal Intensive Care Units (NICUs). ⁸⁹ , ⁹³ , ⁹⁴ Some of the common medications for this group of patients in NICUs are ampicillin, gentamicin, caffeine citrate, vancomycin, beractant, furosemide, fentanyl, dopamine, midazolam, and calfactan. ⁹⁰ Among these, TDM is only routinely used for gentamicin and vancomycin. ⁸⁹ Caffeine, for example, has become one of the most prescribed medications in NICUs and the preferred first‐line treatment for apnea of prematurity in preterm infants. ⁹⁵ , ⁹⁶ It has been shown to reduce the rate of bronchopulmonary dysplasia in newborns and to increase extubation success, ⁹⁵ , ⁹⁷ , ⁹⁸ , ⁹⁹ , ¹⁰⁰ but adverse side effects such as tachycardia, irritability, and diminished weight gain may occur. Moreover, accidental administration of high doses has led to temporary toxicity. ¹⁰¹ , ¹⁰² Because of the wide therapeutic range, and relative safety of this drug, TDM is not routinely used—it is, however, advisable as an optimal therapeutic strategy for newborns. ⁹⁵ , ⁹⁶ , ⁹⁸ , ¹⁰⁰

The major limitation for performing TDM in neonates and pediatric populations is their small circulating blood volume. Thus, repeated blood sampling can be risky in neonatal care, and the maximum amount of blood that can be collected is currently a point of discussion. ⁸⁹ Consequently, dried blood spot (DBS) collection, a minimally invasive technique for microvolume sampling, has become well‐established to collect a few μl of a newborn's blood on a paper card after a heel prick and is widely used for neonatal screening for specific analytes. ¹⁰³ DBS screening is typically combined with LC–MS/MS analysis. ¹⁰⁴ , ¹⁰⁵ DBS analysis has the potential to facilitate caffeine‐level determination in neonates, ¹⁰⁶ , ¹⁰⁷ , ¹⁰⁸ allowing optimal strategies for precision medication. Thus, the neonatal population would definitely benefit from studies that enable individualized treatment, because most drugs used in neonatal medicine are prescribed off‐label.

2.7. ICUs patients

Effective treatment of intensive care unit (ICU) patients while avoiding predictable and unpredictable adverse effects is highly challenging. In the ICU, 16% of all adverse drug events are caused by DDIs. ¹⁰⁹ The rate that ICU patients are exposed to potential DDI is almost twice as high as patients on general wards in some health institutions. ¹¹⁰ This is mainly because ICU patients commonly receive a large number of drugs, some of which can inhibit or induce the metabolism of others. Examples of inhibitors that are widely administered in the ICU setting include macrolide antibiotics (erythromycin, clarithromycin), azole antifungals (ketoconazole), protease inhibitors (ritonavir, saquinavir), SSRIs (paroxetine, fluoxetine), cimetidine, and amiodarone. Moreover, many patients in the ICU may develop stage 1 acute kidney injury (AKI), ¹¹¹ acquired liver injury and hepatotoxicity, intestinal dysfunction, or hypovolemia at some point during their ICU stay. ¹¹² These conditions greatly affect drug absorption, distribution, and elimination. Therfore, many drugs require more continous and careful TDM and dosage adjustment when administered to ICU patients. For example, for out‐patients who are administered digioxin orally and for whom the steady state with the indicated therapeutic concentration has been reached, TDM of digoxin concentrations is only recommended in certain cases, such as electrocardiogram (ECG) changes, when ADRs are suspected, or when changes in the patient's condition may affect the pharmacokinetics or efficacy of digoxin. ⁶⁴ In contrast, for ICU patients who are administered digioxin intravenously, the trough level at 12–24 h has to be determined after every dose, even after reaching the steady state. ¹¹³

2.8. TDM for antimicrobial therapeutics

Antimicrobial agents and particularly antibiotics should be administered with very accurate dosing schemes to reach the appropriate therapeutic target and prevent therapeutic failure, toxicity, and antimicrobial resistance. ¹¹⁴ Antimicrobials were classified by Roberts et al. according to the pharmacokinetic and pharmacodynamic indices that describe their efficacy into (i) time‐dependent antimicrobials such as β‐Lactams, whose efficacy is related to the time for which the antimicrobial concentration is maintained above a certain threshold, typically the MIC; (ii) concentration‐dependent antimicrobials such as metronidazole, whose efficacy is related to the ratio of the peak concentration during a dosing interval and the MIC (C _max/MIC); and (iii) concentration‐dependent antimicrobials with time dependence such as tetracyclines, whose efficacy is related to the ratio of AUC_0–24/MIC. ¹¹⁵ Therefore, although most antibiotics (β‐lactams, macrolides, quinolones) have a wide therapeutic index, numerous studies showed a better outcome when pharmacodynamic targets of these antibiotics are maintained whith the help of TDM, especially in cases of profound PK variability. ¹¹⁶ Some other antibiotics such as aminoglycosides (gentamicin, tobramycin, amikacin) and vancomycin have narrow therapeutic indexes and toxicity may be severe or irreversible (e.g., nephrotoxicity). Therefore, these have been monitored for a long time using many different assays, and their TDM guidelines have been stablished for specific patient conditions. For example, aminoglycosides antibiotics that are concentration dependent are normaly monitored by the C _max/MIC ratio which should be maintained between 8 and 10 when analyzing a blood sample that was taken 30 min after the end of the intravenous infusion, in order to determine the C _max. In contrast, for patients with impaired renal function, burns, or sepsis, two samples (peak and trough) are needed for an improved adjustment of the dose and/or the dosing frequency. ¹¹⁵ , ¹¹⁷ Therefore, it is crucial to develop multiplexed methods to accuratly measure antibiotics levels in plasma for the precise individualization of antimicrobial therapies. Barco et al. have introduced a clinicaly validated LC–MS/MS multiplexed method for rapid quantitation of 14 antibiotics (amikacin, amoxicillin, ceftazidime, ciprofloxacin, colistin, daptomycin, gentamicin, linezolid, meropenem, piperacillin, teicoplanin, tigecycline, tobramycin, and vancomycin) and a beta‐lactamase inhibitor (tazobactam). ¹¹⁸ The assay has an overall turnaround time of 20 min using a small volume of plasma and is suitable for real‐time TDM‐guided personalization of antimicrobial treatment in critically ill patients.

There is also increasing interest in TDM of many antifungal agents (such as flucytosine, itraconazole, voriconazole, and posaconazole ¹¹⁹ ) and antiviral agents (such as anti‐hepatitis C virus drugs and antiretroviral [anti‐HIV] drugs ¹²⁰ ).

3. TECHNIQUES AND CHALLENGES FOR TDM

A variety of techniques and methods have been used for TDM (Table 1).

TABLE 1.

Methods that have been used for therapeutic drug monitoring (TDM)

Immunoassays	Mass spectrometry
Radioimmunoassays (RIAs)	LC–MS/MS
Enzyme‐linked immunosorbent assay (EIA)	MALDI‐MS
Enzyme‐multiplied immunoassay technique (EMIT)	GC–MS
Heterogeneous enzyme‐linked immunosorbent assay (ELISA)	Other techniques
Fluorescence polarization immunoassay (FPIA)	High‐performance liquid chromatography (HPLC)
Competitive fluorescent microsphere immunoassay (CFIA)	Atomic absorption spectrometry
Chemiluminescence immunoassay (CLIA)	Flame photometry
Immunochromatography	Electrode technique
Latex immunoagglutination inhibition method (PENTINIA)	Chromogenic technique

In the late 1950s, Berson and Yalow developed the first immunoassay for insulin, ¹²¹ which paved the way for the development of RIAs for thousands of analytes. ¹²² RIAs have been the most commonly used technique for TDM, for example, for digoxin ¹²³ and digitoxin. ¹²⁴ The major advantage of RIA for measuring compounds in biological fluids is the high precision and extreme sensitivity, which cannot be achieved by other immunoassays. ¹²⁵ Major disadvantages of RIAs, however, are the use of radiation and the associated environmental issues, as well as the cost of waste disposal. Homogenous enzyme immunoassays (EIAs) were a major breakthrough in immunoassay technology and have been used for the TDM of some pharmaceutical compounds, for example, aminoglycoside antibiotics. ¹²⁵

Then immunoassays were converted into high‐throughput assays such as enzyme multiplied immunoassay (EMIA), fluorescence polarization fluoroimmunoassay (FPFIA), and CLIA. Automated analyzers were developed using these techniques and became commercially available. ¹²⁶ , ¹²⁷ , ¹²⁸

Immunoassays are still widely used for TDM (see Figure 2), because they can be automated and are easy to perform, but the quality, purity, selectivity, and specificity of antibodies used are a critical part of the assay development process. ¹²⁹ Being antibody based, immunoassays can suffer from interferences, lack of sensitivity, and are not readily available for every drug. ¹³ , ²⁰ , ¹³⁰ , ¹³¹ Significant interference has, for example, been reported for digoxin immunoassays. ¹³²

A technology that is increasingly being used for TDM is MS, which allows determination of the mass‐to‐charge ratio (m/z) of ions in the gas phase. MS can be used not only to identify but, more importantly, to quantify biomolecules with high specificity, precision, and accuracy. Quantitation by MS is usually performed using reference standards that can be either stable isotope‐labeled synthetic versions of the analyte of interest that have the same physicochemical properties as the endogenous analytes but a different mass, which allows them to be differentiated by MS, or standards that are chemically highly similar to the analyte of interest. Known amounts of these reference standards are then spiked into each sample as early as possible during the sample preparation workflow in order to compensate for potential losses during sample preparation. These internal standards are then used for determining the concentration of the endogenous analyte, either through internal or external calibration.

Gas chromatography (GC) coupled to MS has been used in clinical laboratories for several decades for the quantitative analysis of small molecules, and, in recent years, the coupling of HPLC with MS (LC–MS) has become increasingly important in clinical laboratories. Although HPLC combined with ultraviolet detection (HPLC‐UV) has been used in clinical laboratories for the detection of some analytes, coupling HPLC to MS provides more accurate results. ¹³³ , ¹³⁴ As the number of LC–MS instruments in clinical laboratories around the world continues to grow, LC–MS has become the method of choice for the analysis of hormones and proteins, ¹³⁵ especially the ones that are present at low levels, such as testosterone and estrogen hormones, where sensitivity and specificity are required. ¹³⁶ , ¹³⁷

LC–MS‐based quantitation in clinical settings usually involves multiple reaction monitoring (MRM) on triple quadruple mass spectrometers. MRM allows the fast and specific quantitation of analytes over a large dynamic range with high precision and accuracy. ¹³⁸ In MRM, the first quadruple of the MS is used as a filter that allows only molecule ions having a predefined m/z range (the precursor ion) to enter the second quadrupole, where the collision with inert gas molecules leads to a fragmentation of this precursor ion into product ions. The third quadruple is then again used as a filter that allows only product ions of a predefined m/z range to pass to the detector. Thus, only if the correct predefined combination of precursor ion m/z (Q1) and product ion m/z (Q3) appear, a signal will be acquired. Although the LC immediately in front of the MS is typically used to reduce sample complexity and thus improve the limits of detection (LOD) and quantitation (LOQ), the analyte‐specific elution profile and retention time add additional confidence, particularly in reference to the internal standard. If stable isotope‐labeled standards are used, these typically co‐elute exactly with the endogenous molecule (with the exception of deuterium‐labeled standards). Thus, MS‐based quantitation is inherently linked with a confirmatory identification.

Although small molecules and metabolites are typically measured as intact molecules or after a specific derivatization to increase their stability or detectability, protein drugs are usually measured on the peptide level, after controlled proteolytic digestion. ¹³⁹ The digestion considerably reduces the large heterogeneity of proteins with regard to size, hydrophobicity, and structure, which facilitates the separation by, the detection by MS, but also the generation of appropriate reference standards (typically stable‐isotope labeled peptides). Notably, for protein assays, it is important that the “quantifier peptide” is unique for the protein of interest, shows a good and reproducible digestion efficiency, ¹⁴⁰ is known to be not modified (e.g., by glycosylation), and ideally does not contain amino acids or amino acid sequences that are easily modified or degraded during sample processing (such as methionine that is easily oxidized, or asparagine‐glycine that tends to deamidate). ⁷⁶

LC–MS is a versatile technology that allows the development of assays by clinical laboratories themselves, that is, “LDTs”. ¹⁴¹ It has enabled the TDM of many drugs that could otherwise not be quantified because of the lack of available tests, because LC–MS can be applied to a much wider range of biological molecules than GC–MS that is restricted to volatile compounds. ¹³⁵ , ¹⁴² Some challenges in implementing MS in clinical laboratories are related to specific infrastructure needs, such as the high initial cost of the equipment, the need for a stable and reliable supply of electricity, ventilation, and high purity gases, as well as the requirement for staff members with a higher level of technical expertise than, for example, for performing ELISAs. Although clinical MS is not without pitfalls, the appropriate choice of sample preparation and MS methods, combined with the use of internal standards and full validation of assays, enables accurate quantitation of drugs in biological fluids. ¹⁴³ Theoretically, hundreds of drugs and toxins could be screened in a single analytical run, ¹³⁵ including low molecular weight compounds. LC–MS methods usually have easier sample‐preparation workflows than conventional HPLC or GC–MS that often requires derivatization for detection or volatilization. Moreover, once an LC–MS method has been validated, the actual cost per sample is significantly lower than the alternative techniques. ¹⁴⁴

There has been a combined effort by industry and clinical laboratories towards maximizing automation and facilitating assay implementation for routine work in clinical laboratories. Examples of automated systems that are currently available are the Shimadzu CLAM System ¹⁴⁵ and ThermoFisher Scientific's Cascadion™ SM Clinical Analyzer. ¹⁴⁶ The latter has been tested and compared with routine tests for Vitamin D and immunosuppressive drugs (cyclosporine A, sirolimus, tacrolimus, and everolimus) and showed excellent analytical performance. ¹⁴⁷ , ¹⁴⁸ For Vitamin D, all samples were within 10% of the reference concentrations, and for the immunosuppressive drugs, the mean deviation from the immunoassays was 17%–19%, ¹⁴⁷ , ¹⁴⁸ suggesting that automated LC–MS provides robust clinical data.

Another advantage of LC–MS is that low amounts of material are required for the analysis, which makes the use of less invasive sampling techniques appealing. The collection of DBSs, which can be combined with MS analysis, ¹⁰⁵ , ¹⁴⁹ also facilitates longitudinal and/or remote testing, which is preferable for some patients in order to spare them from having to visit frequently crowded test centers in hospitals, a topic that has become more important during the COVID‐19 pandemic. ¹⁰⁵ Saliva is becoming a promising alternative sample type, ⁹⁴ , ¹⁵⁰ but more correlation studies with plasma are required to establish saliva as a useful alternative, which would be particularly relevant for vulnerable patients where blood sampling can be risky, for instance for neonates.

4. CONCLUSION

There is much evidence that supports the benefits of TDM on drug effectiveness and the minimization of side effects, especially for individuals, that present a high intervariability, such as neonatal, or critically/severely ill patients. TDM is a valuable tool to personalize drug administration and to ensure that patients receive the next drug dosage when, or just before, the trough level is reached. Personalized drug administration is the next step towards precision medicine, which—in contrast to a “one‐size‐fits‐all” approach—takes into account an individual's variability in genes, environment, and lifestyle, leading to the selection of drugs tailored to a specific patient and his/her disease. ¹⁴¹ The goal of personalized medicine is to optimize the benefit and minimize the harm of an intervention, by using genomics and further omics data to support the personalization of treatments. ⁴² With targeted agents, and with technology that makes it possible to monitor a drug's concentration in an individual, TDM should be used as a precision approach that is part of personalized medicine for patient care to help ensure that patients receive the most appropriate drugs and treatment schedules. There are, however, still some challenges to the routine application of TDM. These include a lack of available tests, sampling logistics, and even the interpretation of the results. MS will help overcome these limitations by expanding the number of drug assays (MS‐based LDTs (e.g., Table S1 lists 93 drugs that are currently being measured by LC–MS and seven that are currently being measured by GC–MS), and by simplifying the sampling, as it requires minimal amounts of material. The wide availability of standardized MS‐based assays will help to establish reliable reference ranges for specific drugs and indications, and the increasing availability of more‐and‐more complex data on individual patients in combination with cutting‐edge computational approaches such as artificial intelligence will enable the generation of more robust prediction models. As mass spectrometers become more robust and easy‐to‐use and become more common in clinical laboratories, we envision that MS‐based TDM will successfully answer additional clinical and analytical questions and will become an even more widespread approach in the era of precision medicine.

CONFLICT OF INTEREST

CHB is the Chief Scientific Officer of MRM Proteomics, Inc. RPZ is the Chief Executive Officer of MRM Proteomics Inc. The other authors declare no competing financial interests.

Supporting information

Table S1: TDM assays being used in diagnostic laboratories

Sources: (A) ARUP Consult. https://arupconsult.com/content/therapeutic-drug-monitoring. (Accessed Mar. 22, 2021). (B) Mayo Clinic Laboratories. https://www.mayocliniclabs.com/test-info/drug-book/therapeutic.html. (Accessed Mar. 22, 2021).

Gaspar VP, Ibrahim S, Zahedi RP, Borchers CH. Utility, promise, and limitations of liquid chromatography‐mass spectrometry‐based therapeutic drug monitoring in precision medicine. J Mass Spectrom. 2021;56(11):e4788. doi: 10.1002/jms.4788

DATA AVAILABILITY STATEMENT

n/a