Therapeutic Drug Monitoring

Svein I. Johannessen

Philip N. Patsalos

Torbjörn Tomson

Emilio Perucca

Introduction

Therapeutic drug monitoring (TDM) can be defined as “the measurement and the clinical use of drug concentrations (levels) in body fluids (usually serum or plasma) to adjust each patient’s individual drug dosage and schedule to each patient’s individual therapeutic requirement.” In practice it is the patient who is treated and not the concentration, so that the dose of a drug is adjusted, using the drug concentration as a guide, to optimize its efficacy, avoid, minimize, or identify toxicity, and detect or confirm poor compliance. It is therefore important to appreciate that TDM begins before a drug concentration is measured and that the measurement itself is only part of the overall process of planning, monitoring, and optimally adjusting the dosage regimen.

Many pharmacologic properties of antiepileptic drugs (AEDs) support suitability for TDM, and during the last 40 years therapeutic monitoring of AEDs has had an undeniable impact on the management of patients with epilepsy.¹³ Major contributing factors are our enhanced understanding of AED pharmacokinetics and its relevance to drug therapeutics and the development of robust, reliable, and specific analytical methodologies underpinned by appropriate quality assurance schemes.¹⁰²^,¹⁰⁶ Assay results should be available quickly, preferably within 24 hours of sampling, because the most important use of blood concentrations are for making dosage adjustments and diagnosing toxicity when rapid decisions need to be made. Indeed, on-site, near-patient AED monitoring has been shown to have an immediate impact on clinical decision making and to enhance patient management.¹⁸^,⁴⁹^,⁶⁴

The criteria for valid TDM are the availability of accurate pharmacokinetic data; a poor correlation between dose and blood concentrations (i.e., substantial pharmacokinetic variability); good correlation between blood concentration and therapeutic effect, toxicity, or both, at least within individuals; a narrow therapeutic index (i.e., the therapeutic dose is close to the dose associated with toxicity); and the availability of simple, accurate, reproducible, and inexpensive analytical assays. However, some AEDs are more suitable candidates for monitoring than others. Phenytoin best fulfills these criteria because its saturable pharmacokinetics makes it very difficult to prescribe the optimum dose without measuring blood levels.

The goal of AED treatment is seizure freedom without side effects. However, even with the introduction of nine new AEDs, a significant number of people with epilepsy are still not achieving this goal. TDM can help to improve seizure control in numerous ways including the following:

Identification of therapeutic failure due to underdosage.
Detection of noncompliance with prescribed therapy, which may be responsible for avoidable therapeutic failure.
Identification of the uncommon situation in which overdosage causes increased seizures.
Detection of pharmacokinetic interactions that may compromise the adequacy of the therapy.

The following section provides a concise description of the current issues relating to the use of TDM for AEDs.

Principles of Therapeutic Drug Monitoring

Interest in the use of TDM as an aid to optimizing epilepsy treatment stems from the realization that serum AED concentrations may vary severalfold among patients who are prescribed the same dosage.²⁸ There are multiple reasons for such variability. To start with, the dose prescribed may differ from the dose actually taken, mainly due to variation in the degree of compliance. Even when drugs are taken as directed, serum concentrations may vary dramatically as a result of inter- and intraindividual differences in pharmacokinetics. Although most AEDs are absorbed efficiently from the gastrointestinal tract, the extent of absorption may be variably affected by factors such as age, interactions with foods or other medications, and characteristics of the pharmaceutical formulation used.²² More important, the metabolic and renal clearance of AEDs is subject to prominent variability, both within and between subjects, under the influence of the genetic background (e.g., variation in genes controlling drug metabolism and drug transport), physiologic factors (e.g., changes in drug-metabolizing-enzyme activity and renal function during development, pregnancy, and aging), drug interactions, and disease states, with special reference to disorders affecting the function of the liver and the kidney.

Because the concentration of AEDs in serum is in equilibrium with the drug concentration at the site(s) of action in the brain, it is not surprising that differences in serum drug concentrations represent an important source of variability in dose requirements. Based on this background, knowledge of serum drug concentrations provides a more direct estimate, compared with prescribed daily dose, of the amount of drug that is available to produce pharmacologic effects. In fact, it is reasonable to assume (and this assumption can be verified in the clinic under many circumstances) that therapeutic and toxic effects correlate better with the concentration of a drug in the serum than with the prescribed daily dose.²¹ In other words, under appropriate circumstances, the serum concentration of an AED can be used as a surrogate marker in assessing whether the patient is receiving an adequate amount of the prescribed drug.

Obviously, serum drug concentration measurements are superfluous when tools are available to measure drug effects directly. For example, we do not need to measure the serum concentrations of an antihypertensive drug or an oral anticoagulant because we can measure directly arterial blood pressure
or International Normalized Ratio (INR) ratio and individualize dosage directly based on therapeutic response. In the case of AEDs, however, direct measurement of pharmacologic response can be elusive. In a newly diagnosed patient with infrequent seizures, for example, it may take months to determine whether the dosage initially prescribed is sufficient to prevent seizure recurrence. Measuring the serum AED concentration in that patient and comparing it with the concentration range that in previous studies has been found to maximize the probability of seizure control helps in predicting the effectiveness of the administered treatment and in determining whether a dosage adjustment would be appropriate.

Factors Affecting Response to a Given Antiepileptic Drug Concentration

If all patients achieved the best response at the same AED concentration, optimizing treatment would be easy. We would simply need to administer an initial dose, measure the serum drug concentration, and adjust the dose accordingly until the “therapeutic” concentration is achieved. Unfortunately, in real life things are much more complicated because the serum concentration of any given AED required to produce an optimal response varies more than tenfold across individuals and can also change substantially over time within the same individual.²¹^,⁷⁵^,⁹⁴ This is due to a combination of factors, all of which may confound the relationship between serum drug concentration and pharmacologic effects, as follows.

Characteristics of the Seizure Disorder

There is evidence that the concentration of an AED required to produce seizure control differs in relation to the type and severity of the seizure disorder. Schmidt and Haenel⁸⁹ compared serum AED concentrations associated with complete seizure control in 84 patients stabilized on monotherapy with phenyt-oin, carbamazepine, or phenobarbital. In 26 of the 40 patients (65%) with generalized tonic–clonic seizures only, seizure control was achieved at concentrations of <60 μmol/L (<15 μg/mL) for phenytoin, <108 μmol/L (<25 μg/mL) for phenobarbital, and <25 μmol/L (<6 μg/mL) for carbamazepine. Conversely, only 3 of 32 well-controlled patients (9%) with complex partial seizures achieved remission at these relatively low drug concentrations. The AED concentration associated with seizure control also varied in relation to number of seizures in the first year of epilepsy. The 50 patients who were controlled at concentrations in the low range had suffered a median of 5 seizures (range 1 to 18) in the first year, whereas those who were controlled at higher concentrations had 61 seizures (range 0 to 300) over the same period. Recent studies also suggest that, within a given syndrome, responsiveness to AED treatment also depends on underlying etiology: For example, among patients with localization-related epilepsy, those showing underlying cortical atrophy or cerebrovascular disease may respond more favorably to AEDs than the remainder of the symptomatic group.⁵⁴

Age

Because the distribution of epilepsy syndromes and underlying etiologies vary across age groups, we would expect the optimal concentration of an AED to be influenced by age. Evidence has been provided, in particular, that many elderly patients achieve seizure control at lower serum AED concentrations than younger subjects, although they also tend to develop adverse effects such as impairment of gait and tremor at lower concentrations.²⁷^,³⁹

Duration of Treatment

For many AEDs, the response to a given drug concentration may change over time, most notably due to gradual development of tolerance to adverse effects. Patients with a prolonged exposure to phenobarbital or benzodiazepines, for example, might tolerate without major untoward effects serum drug concentrations that can cause extreme sedation and even coma in acutely exposed subjects.¹⁴^,³⁰ A gradual improvement in central nervous system (CNS) tolerability with slowly rising serum drug concentrations during the initial up-titration phase is also observed with topiramate, tiagabine, pregabalin, and many other AEDs.⁷³

Conditions Associated With Altered Drug Binding to Serum Proteins

Although routine methods for measuring AEDs in body fluids do not discriminate between drug molecules that are present in free (unbound) form and those that are bound to plasma proteins, only the free drug is available to move across the endothelium and to equilibrate with the concentration in the interstitial space in the brain. If the free fraction increases, the total drug concentration in serum will underestimate the amount of free, pharmacologically active drug, and under these circumstances therapeutic and toxic effects will be observed at total drug concentrations that are lower than usual.⁶⁸ Among AEDs that are commonly monitored, those that are most extensively bound to serum proteins are phenytoin and valproic acid. Impairment in the protein binding of these drugs may be caused by hypoalbuminemia (as observed during pregnancy, old age, liver disease, and many other pathologic conditions), accumulation of endogenous displacing agents (most notably in patients with renal insufficiency), and administration of other medications that compete for plasma protein–binding sites.²^,⁶⁸ In these conditions, interpretation of serum drug concentration requires special skills: For example, in a patient in late pregnancy, the change in unbound fraction of phenytoin can be predicted by measuring the serum albumin concentration.⁷⁵ Techniques are also available for the direct measurement of unbound drug concentrations. These techniques are not used routinely because measuring unbound drug concentrations is more expensive and more cumbersome than measuring total drug concentrations. Moreover, in most situations, the intersubject variability in unbound fraction is relatively small, and measurement of total concentrations is more than adequate for clinical purposes. However, if a major change in unbound fraction is expected (or suspected), measuring unbound drug concentrations might be justified.

Comorbidities

Several diseases, particularly those affecting the liver and the kidney, are associated with changes in the binding of AEDs to serum proteins, and by this mechanism they alter the relationship between total serum concentration of highly protein-bound drugs and clinical effects.⁷⁴ Associated disorders may also modify the response observed at any given serum AED concentration at the pharmacodynamic level. For example, it has been elegantly shown that the threshold serum carbamazepine concentration that induces ataxia is significantly lower in patients with magnetic resonance imaging (MRI) evidence
of preexisting cerebellar atrophy than in patients without atrophy.⁹³

Drug Interactions

As discussed previously, concomitant drugs may affect concentration–response relationships by displacing highly bound AEDs from plasma protein–binding sites. Valproic acid, in particular, causes a marked displacement of protein-bound phenytoin, and in valproic acid–treated patients, therapeutic and toxic effects of phenytoin are observed at total serum phenytoin concentrations lower than usual.⁴⁷^,⁶⁸ Concomitant drugs can also affect concentration–response relationships through pharmacodynamic interactions: For example, the tolerability associated with any given concentration of carbamazepine is reduced in the presence of adverse pharmacodynamic interactions caused by oxcarbazepine¹ or lamotrigine.¹²

Confounding Effects of Active Metabolites

At times, effects apparently unrelated to the drug concentration in serum can be explained by the presence of active metabolites. One example is the precipitation of carbamazepine toxicity following addition of valpromide, a valproic acid derivative: The toxic reaction is caused by a prominent increase in the concentration of the active metabolite carbamazepine-10,11-epoxide, whereas serum carbamazepine concentration is unaffected.⁷⁷

The Concept of Therapeutic Range

In the light of the variability discussed earlier, “therapeutic” ranges of serum AED concentrations reported in the literature (Table 1) must be interpreted flexibly. These ranges have merely a probabilistic value, that is, they represent the concentration interval at which the majority of patients are expected to show an optimal response. Indeed, some patients may do best at concentrations above and below the range, and there is no justification for modifying dosage in these patients if their seizures are well controlled and no adverse effects have emerged.³⁸^,¹⁰⁷

The validity of therapeutic ranges quoted in the literature depends on the population and methodology that led to their definition. In particular, the pivotal study that led to the widespread acceptance of a 40- to 80-μmol/L (10- to 20-μg/mL) therapeutic range for phenytoin⁴⁴ was conducted in a small and preselected population of patients with heterogeneous epileptic disorders who had not responded to phenyt-oin. Thus, it is likely that the study excluded patients who had become seizure free at concentrations <40 μmol/L (10 μg/mL). In fact, in different studies, the proportion of patients achieving complete seizure control at phenytoin concentration <40 μmol/L (10 μg/mL) has been on the order of 22%,⁸⁶ 35%,⁸⁸ and 73%.²³ Similar observations have been made with other AEDs, leading to the suggestion that the lower limit of the therapeutic range should be disregarded altogether, and that any measurable concentration up to the level at which toxicity is likely to occur should be regarded as potentially therapeutic.⁷⁵

In the light of these considerations, it is clear that therapeutic ranges must be interpreted flexibly, by taking into account all information that is available for the individual patient, and that no dosing changes should be made without careful evaluation of clinical response. Still, knowledge of the probability that a given concentration can be associated with seizure control or toxicity can be useful in specific situations. For example, in a patient with newly diagnosed epilepsy and infrequent seizures who is concerned about potential adverse drug effects and would not be excessively disturbed by the risk of having another seizure, it would make sense to adjust the dosage to achieve a relatively low target concentration. Conversely, a target concentration in the higher range would be appropriate whenever the patient’s social, professional, or psychological condition makes it imperative to reduce to the very minimum the risk of seizure recurrence.

The Individual Reference Concentration

The existence of a wide variability in the concentration at which different patients respond justifies an alternative (or complementary) approach to TDM, based on the identification of the so-called individual reference concentration.⁶⁹ In this approach, treatment is adjusted based on the individual’s clinical response (assisted, if appropriate, by monitoring serum concentrations) until a dose is identified that controls optimally the seizures without undue adverse effects. The serum concentration that is associated with that dose represents, for that patient, the individual reference concentration and can be helpful during subsequent management in ascertaining potential causes for a change in clinical status or in determining the need for dose adjustment. For example, if the patient will need an additional medication that is likely to increase or decrease the serum concentration of the AED, knowledge of the baseline (reference) AED concentration will allow adjusting the dose of the latter to compensate for the effect of the drug interaction. Likewise, knowledge of the individual reference concentration can be invaluable in clarifying the mechanisms responsible for the sudden onset of toxicity or for an unexpected recurrence of previously controlled seizures. An example is provided by a recent study that assessed serum AED concentrations in 52 patients shortly after the occurrence of a breakthrough seizure: In 44% of the cases, the serum concentration measured after the seizure was less than one-half the individual reference concentration that had been previously found to be associated with a good response in the same individual.⁹² In most cases, this was ascribed to poor compliance. Because the individual reference concentration is specific for each patient, it can be used in the absence of information on the serum concentration–response relationships in a population and when such a relationship is very variable across patients. There is, however, an important caveat: At times, concentration–response relationships can change within patients, as when drug responsiveness changes as a function of brain maturation in a child, or when a medication is added that causes displacement of the monitored drug from its serum protein–binding sites.

Assessment of the Impact of Therapeutic Drug Monitoring on Clinical Outcome

Although TDM has been established as a routine aid in the effort to individualize the dosing of AEDs since the 1960s, the impact of using TDM on the outcome of drug treatment of epilepsy has rarely been assessed in a systematic manner. Open uncontrolled studies have demonstrated that the introduction of a TDM service can result in a larger proportion of patients being treated with AED serum concentrations within the recommended ranges.³⁴^,⁴¹^,⁴⁹ However, studies on the effect of TDM on outcome in terms of seizure control are scarce. In fact, there are only two published randomized studies comparing the outcome of pharmacologic treatment with or

without the use of TDM.²⁵^,³⁵ In the first, 127 chronic epilepsy patients, most on polytherapy, were randomly assigned to treatment with or without the support of TDM.²⁵ Blood samples for determination of drug concentrations were drawn from all patients, but for one of the two groups the treating physician was not informed about the results. Of the randomized patients, 105 completed the 1-year follow-up. The therapeutic results of the two groups were not significantly different. However, a substantial proportion of patients, similar in both groups, had AED concentrations that fell outside of the therapeutic range. This observation suggests that the physicians responsible for the treatment did not use the information provided by the TDM service, which may have affected the negative outcome of the study. Such interpretation is in line with the observations made in a retrospective analysis of 164 patients with epilepsy.⁶ Seizure control one year before the introduction of TDM was compared to results one year after the service was made available. Seizure control was improved only when the physicians, according to the investigators, appropriately used information from the serum concentrations.

Table 1. Pharmacokinetic parameters of antiepileptic drugs in adults⁹^,⁷¹

Drug	Oral bioavailability	Time to peak concentration	Time to steady state	Half-life monotherapy	Half-life AED+ enzyme inducers	Serum protein binding	Main route of elimination	Tentative therapeutic range^a	Conversion factor F μmol/L = F × μg/mL)	Documentation of/ comment on usefulness of serum drug concentration monitoring
	(%)	(h)	(d)	(h)	(h)	(%)		(μmol/L)
Older AEDs
Carbamazepine	∼85	4–8^b	4–7	8–20	5–12	75	Oxidation	15–45	4.23	Useful
Clobazam	∼100	1–5	4–7	15–50	<30	85	Oxidation	0.1–1.0	3.32	Limited value
Clonazepam	∼100	1–3	5–10	20–60	12–46	82	Oxidation	60–220^c	3.17	Limited value
Ethosuximide	∼100	3–7	5–10	40–60	20–30	0	Oxidation	300–600	7.08	Useful
Phenobarbital	∼100	2–8	10–35	50–160	50–160	50	Oxidation, N-glucoside conjugation, and renal excretion	50–130	4.31	Useful
Phenytoin	∼100	2–8	4–8	7–60^d	7–60^d	90	Oxidation	40–80	3.96	Very useful
Primidone	∼100	2–5	1–3	4–12	4–12	0	Cleavage of the pyrimidine ring, oxidation to phenobarbital, and renal excretion	30–60	4.58	Monitor serum phenobarbital
Valproic acid	∼100	3–6^e	2–4	11–20	6–12	78–94^d	Oxidation and glucuronide conjugation	300–600	7.08	Useful
Newer AEDs
Felbamate	∼100	1–4	3–5	14–22	10–20	30	Oxidation and renal excretion	125–250	4.20	Potentially of value
Gabapentin	Up to 60, decreases with increasing dose	2–3	2	5–7	5–7	0	Renal excretion	70–120	5.83	Potentially of value
Lamotrigine	∼100	1–3	3–15	15–30 + VPA 30–90	8–20 + VPA 15–30	55	Glucuronide conjugation	10–60	3.90	Useful
Levetiracetam	∼100	1	2	6–8	5–8	0	Renal excretion and hydrolysis	35–120	5.88	Probably of limited value
Oxcarbazepine^f	Prodrug of MDH	4–6	2–3	8–15	7–12	40	Ketoreduction, then glucuronide conjugation of MHD	50–140	3.96	Serum MHD potentially of value
Pregabalin	∼100	1	2	6–7	5–7	0	Renal excretion	NE	6.33	Probably of limited value
Tiagabine	∼100	0.5–2	2	7–9	2–3	96	Oxidation	50–250^c	2.43	Probably of limited value
Topiramate	∼100	1–4	4–6	20–30	10–15	15	Renal excretion, oxidation	15–60	2.95	Potentially of value
Vigabatrin	60–70	1–2	1–2	5–8	5–8	0	Renal excretion	NA	7.74	Very limited value
Zonisamide	>65	4–7	5–12	50–70	25–35	60	Glucuronide conjugation, acetylation, oxidation, and renal excretion	45–180	4.71	Potentially of value
AED, antiepileptic drug; MDH, monohydroxy derivative; NA, not applicable; NE, not established; VPA, valproic acid. ^aThe lower limit of the therapeutic range has particularly limited value because many patients do well at serum concentrations below this limit. ^bConventional tablets. ^cμmol/L. ^dConcentration dependent. ^eEnteric-coated tablets. ^fMonohydroxy derivative.

Table 2 Some general indications for therapeutic drug monitoring

After initiation of treatment (to provide a baseline steady-state concentration)
After change in drug dose, in particular when nonlinear kinetics applies (to confirm new drug concentration)
At therapeutic failure (to confirm or exclude a pharmacokinetic explanation for uncontrolled seizures or adverse effects
To identify or control for drug–drug interactions
After a change in drug formulation
When pharmacokinetic alterations due to physiologic or pathologic changes are anticipated (e.g., pregnancy, hepatic disease, renal disease, gastrointestinal conditions potentially affecting drug absorption)
When poor compliance is suspected
Routine sampling annually of phenytoin because of saturation kinetics

The second and most recent randomized, controlled trial on the impact of TDM included 180 newly diagnosed, previously untreated patients with epilepsy who were about to start treatment with an older-generation AED.³⁵ The majority of the included patients were prescribed carbamazepine, and only a small fraction were on phenytoin. Patients were randomized to either treatment with dosage adjusted on clinical grounds alone or treatment with dosage adjusted to achieve serum concentrations within predefined target ranges. Both groups were followed for 24 months. There were no significant differences between the two groups with respect to patients achieving 12-month remission, patients remaining seizure free since initiation of treatment, time to first seizure or to 12-month remission, or frequency of adverse effects. Hence, this study could not demonstrate an effect of routine use of TDM on the clinical outcome of early treatment of epilepsy. In contrast, a small retrospective study from Mumbai, India, indicated that TDM can improve outcome in patients with established “generalized tonic–clonic epilepsy.”⁷⁹ The proportion of patients with seizure control was higher and the frequency of adverse reactions lower among 25 epilepsy patients who had undergone TDM at least twice a year compared with 25 matched epilepsy patients from the same epilepsy clinic who had not. Unfortunately, there is no information on the type of AEDs prescribed in this study. Although the retrospective design calls for caution in interpretation, the results suggest that TDM can have an impact on outcome in special situations and settings despite the negative findings from randomized studies.

The cost-effectiveness of TDM has recently been assessed in a systematic review.⁹⁶ Based on a low level of evidence (level 3), the review concluded that TDM of older-generation AEDs can lead to better control of patients with fewer side effects and be cost-effective. Furthermore, TDM of the new AEDs was considered useful in titrating patients whose epilepsy is difficult to control and in case of questionable compliance and drug–drug interactions.⁹⁷

Thus, there is an obvious need for studies assessing the impact of TDM on the outcome of treatment. Although the two available randomized studies did not provide evidence for the usefulness of routine monitoring of AEDs in general, this does not exclude the value of TDM in special situations.

Table 3 Enzyme-inducing properties of antiepileptic drugs (AEDs)

Non–enzyme-inducing AEDs	Broad-spectrum enzyme-inducing AEDs
Ethosuximide	Carbamaxepine, phenytoin, phenobarbital, primidone
Gabapentin
Levetiracetam	Narrow-spectrum enzyme-inducing AEDs,^a
Pregabalin	Felbamate, lamotrigine, oxcarbazepine, topiramate
Tiagabine
Valproic acid
Vigabatrin
Zonisamide
^aFelbamate, lamotrigine, topiramate (at doses >200 mg/d), and oxcarbazepine may reduce the serum concentrations of steroid oral contraceptives. Felbamate and oxcarbazepine may reduce the serum concentrations of some other cytochrome P450 substrates; both felbamate and oxcarbazepine may also act as enzyme inhibitors as well as enzyme inducers.