Level-Headed: Correlating Antiseizure Medication Drug Concentrations With Adverse Effects and Seizure Response

Abstract

Value of drug level concentrations of brivaracetam, lacosamide, and perampanel in care of people with epilepsy

Hentschel M, Stoffel-Wagner B, Surges R, Surges R, Dolscheid-Pommerich RC. Epilepsia. 2024 Mar;65(3):620–629. doi:10.1111/epi.17873. Epub 2024 Jan 19

Objective: The aim of this study was to determine whether clinical efficacy and reported adverse effects (AEs) of the newer antiseizure medications (ASMs) brivaracetam (BRV), lacosamide (LCM), and perampanel (PER) have been associated with plasma levels of these ASMs. We also investigated whether plasma levels outside the reference range have led to dose adjustments. Methods: Plasma levels of 300 people with epilepsy (PWE) seen at our tertiary epilepsy center were determined by liquid chromatography-tandem mass spectrometry. PWE received BRV (n = 100), LCM (n = 100), or PER (n = 100), in most cases in polytherapy. Demographic and clinical data were retrospectively analyzed and related to plasma levels. The clinical efficacy of BRV, LCM, or PER was assessed retrospectively by comparing seizure frequency at the time of the current blood draw with seizure frequency at the time of first administration. AEs were also recorded and, if reported, compared retrospectively with the time of first administration. Results: No significant associations were found between plasma levels of BRV, LCM, or PER and seizure freedom (BRV, P = 1.000; LCM, P = .243; PER, P = .113) or responder status (BRV, P = .118; LCM, P = .478; PER, P = .069) at presentation. There was also no pattern between plasma levels and the occurrence of AEs. In the majority of cases, drug levels outside the reference ranges have not led to adjustments in the daily doses of BRV (93.5%), LCM (93.9%), or PER (89.1%). Significance: Plasma levels at a given time point did not allow conclusions to be drawn about seizure control or the occurrence of AEs. Our findings indicate that efficacy and tolerability cannot be predicted based on averaged data from a single plasma measurement due to high interindividual variability. Instead, individual reference values should be established when sufficient clinical data are available, in line with the 2008 International League Against Epilepsy position paper on therapeutic drug monitoring.

Commentary

While the history and physical provide a foundation to guide care, interview questions and neurological examination have limits. Accordingly, we supplement data collected during a visit with objective tests, to see what we could otherwise not see with our own eyes.

Checking antiseizure medication (ASM) levels provides an apt example. While it is easy enough to ask the patient or caregiver regarding their perception of ASM adverse effects, seizure response, or medication adherence, unstructured interview questions are imperfect probes. Structured questionnaires demonstrate that adverse effects go undetected ¹ and nonspecific symptoms can be confused for adverse effects, patients may under- or overreport their true seizure burden, ² and many factors beyond adherence influence a patient's blood concentration level (eg, pregnancy, nutritional status, drug interactions, and hepatorenal function). Therefore, checking ASM levels has intuitive appeal to supplement our monitoring toxicity and response, testing clinical hypotheses regarding whether the problem might be “too much” or “too little” medication during periods of worsening.

Despite this appeal, evidence is lacking regarding how ASM levels predict outcomes to guide management. Suggested lab ranges can be arbitrary in the absence of evidence-based reference controls or individualized numerical targets, inter- or even intra-lab variability can make values difficult to interpret across time, and the correlation between levels and physiological response may not be linear or simple. Therefore, the leap between drug concentration and its impact on the patient can be complex.

Hentschel et al ³ recently investigated whether ASM response and adverse effects correlated with plasma levels of 3 ASMs—brivaracetam, lacosamide, and perampanel. They included 100 patients currently taking each ASM and assessed current adverse effects and seizure frequency compared with initial seizure frequency at the time of ASM initiation, per electronic medical records.

They found no such correlations. For example, patients deemed to be currently seizure-free had a median lacosamide plasma level of 9.37 μg/mL, compared to a median level of 10.04 μg/mL among those not currently seizure-free (P = .24). Similar results were obtained when stratifying according to 50% responder rates contrasting current versus prelacosamide seizure frequencies. Median plasma levels were 9.37 μg/mL versus 9.95 μg/mL in patients with versus without suspected lacosamide-related adverse effects without clear differences between groups, albeit the sample size with adverse effects (N = 9) was small for any convincing comparison. Not only did plasma levels fail to distinguish groups, but also even among the 49 patients in whom plasma levels were outside of reference ranges, in only 3 (6%) cases were ASM doses modified.

So, should we ever monitor levels? One pessimistic interpretation of these data would be that levels did not predict anything important and rarely changed management, so why bother? It is true that more information is not always better, even if obtaining such information has face validity that it should be useful. Refusing to act upon supra- or subreference levels may be reasonable (even if casting doubt upon the utility of having checked in the first place), given we treat the patient rather than just numbers. Also, linear pharmacokinetics relating dose to blood concentrations does not mean that physiological/psychological responses have any sort of such linear relationship, so the clinical interpretation or implication of a “high” or “low” number is not straightforward.

However, never say never. Checking levels likely have value in ways not investigated by the above study. The study by Hentschel et al focused only on between-patient differences. In contrast, common situations in which levels might still have face validity include following within-patient changes over time. For example, one could obtain a level during a period of good control, which would enable a comparison in case a patient transitions into a future period of worsened seizures or increased possible adverse effects. Likewise, one may check baseline levels before pregnancy to guide future dose escalations as pregnancy progresses, targeting stability knowing that certain drug levels tend to drop as metabolism and blood volume increase into later trimesters. Levels may also inform adherence (“pseudorefractoriness”), for example, to increase confidence that the patient has maximized medical therapy before advancing toward surgical evaluation.

Moreover, key study limitations temper conclusions.

This cross-sectional study does not rule out a true causal relationship between levels and seizure response or levels and adverse effects. Trials typically find dose-response relationships (eg, brivaracetam going from 50 mg to 100 mg per day ⁴ ), making level-response or level-side effect relationships likely. However, a cross-sectional study cannot distinguish causation (high levels lead to adverse effects) from reverse causation (adverse effects lead to decreasing levels), thus true and spurious effects might cancel out.

Between-subject comparisons are assuredly confounded. Many factors may vary between patients that may explain both the studied outcomes and levels of adherence such as behaviors, polypharmacy, depression/anxiety, and comorbidities. One cannot draw between-subject conclusions without having equalized such confounders between groups.

The reasons for checking ASM levels at that particular moment are nonrandom, as clinical deterioration may lead to checking levels at that moment affecting apparent correlations.

This population may not generalize well to other patients, given a relatively high seizure frequency among patients that have tried a mean of 4 to 5 ASMs already, some of whom were inpatients at the time of the blood draw further raising issues of cause and effect.

The electronic medical record may not be the best “gold standard” to assess seizure frequency or adverse effect burden, without structured validated questionnaires measuring these constructs.

This study contained a small sample size of a limited number of ASMs, thus with potentially limited power to detect differences.

Future studies addressing this question could therefore focus on longitudinal assessments within individuals over time, capture variables that might independently influence the perception of adverse effects or seizures, and apply more rigorous outcome assessments. It would, therefore, be inappropriate to conclude from these data that there is no correlation between levels and outcomes.

As with anything, whether and when to obtain levels is case-dependent. At least, it may “rule in” clinical hypotheses, such as an undetectable level in a seizing patient or a clearly high level with a clinical constellation consistent with that medication's toxic profile. Though even then, the true test of such hypotheses is whether outcomes change in response to “correcting” such levels. Levels in the proper context may provide one important clue to the clinician that can be easily tracked over time, and blood levels remain one tool at the clinician's disposal when seeking to track adherence (“trust but verify”) or assess whether too much or too little drug might underly apparent nonresponse or symptoms that might suggest toxicity. However, buyers beware that it may not be straightforward to apply the number to your patient with so many potentially confounding or competing factors at play, particularly in the absence of well-validated individualized lab reference ranges.