Empiric dosing strategies to predict lamotrigine concentrations during pregnancy

Abstract

Introduction:

Maintaining seizure control with lamotrigine is complicated by altered pharmacokinetics and existence of subpopulations in whom clearance increases or remains constant during pregnancy.

Objective:

Our objective was to characterize the potential for particular dosing scenarios to lead to increased seizure risk or toxicity.

Methods:

Lamotrigine pharmacokinetic parameters obtained from our previous study were applied to a one-compartment model structure with subpopulations (75:25%) exhibiting different clearance changes. A single-patient simulation was conducted with typical pharmacokinetic parameter values from each subpopulation. Population level simulations (N=48,000) included six dosing scenarios and considered four preconception doses using the R package mrgsolve (Metrum Research Group). Thresholds for efficacy and toxicity were selected as drug concentration that are 65% lower than preconception concentrations and doubling of preconception concentrations, respectively.

Results:

Individual simulation results demonstrated that without dose increases, concentrations fell below 0.65 at 6–8 weeks in the high clearance subpopulation depending on preconception clearance. While no simulated dosing regimen allowed all women in both subpopulations to maintain preconception concentrations, some regimens provided a more balanced risk profile than others. Predicted concentrations suggested potential increased seizure risk for 7 to 100% of women in the high clearance change group depending on preconception dose and subpopulation. Additionally, in 63% of dosing scenarios for women with low clearance change, there was an increased risk of toxicity (34 to 100% of women).

Significance:

A substantial percentage of simulated individuals had concentrations low enough to potentially increase seizure risk or high enough to create toxicity. Early clearance changes indicate possible subpopulation categorization if therapeutic drug monitoring is conducted in the first trimester. An arbitrary “one-size fits all” philosophy may not work well for lamotrigine dosing adjustments during pregnancy and reinforces the need for therapeutic drug monitoring until a patient is determined to be in a low or high clearance change group.

Introduction

Managing epilepsy during pregnancy is crucial to the health of both the mother and developing fetus. Antiseizure medication (ASM) initiated prior to conception is typically continued throughout pregnancy for stable seizure control. Lamotrigine is the most commonly used ASM for epilepsy during pregnancy due to its relatively low risk for structural and neurodevelopmental teratogenicity compared with most other treatments.^1–4 Maintaining seizure control with lamotrigine is complicated due to substantially altered pharmacokinetics during pregnancy. The most notable of these changes is the significant increase in clearance with wide individual variability. When ASM concentrations, particularly lamotrigine, fall below 65% of their preconception baseline values, there is a risk for seizures worsening.^{5, 6} Thus, an individualized 0.65 ratio to target concentration (RTC), the ratio of total concentration over the target concentration, is a threshold for potential increased seizure risk.^{5, 6} Similarly, when exposure rises significantly over baseline, excess fetal exposure or maternal toxicity can occur.³

To further complicate dosing, two subpopulations of pregnant women have been identified: 1) pregnant women with clearance increases of over 200% on average by the end of pregnancy (high clearance (HC)), and 2) pregnant women with little to no marked change in clearance (low clearance (LC)).^{7, 8} Approximately 77% of pregnant women with epilepsy are HC, while 23% of women are LC.⁷ There are a number of physiological changes that occur during pregnancy that can affect drug pharmacokinetics including metabolic and renal function changes.¹ However, the specific etiology to determine whether a woman is HC or LC is not currently known. Although it is not presently possible to identify which group a woman belongs to prior to pregnancy, predicting concentrations with different dosing adjustment strategies can reveal which strategies are most likely to result in loss of efficacy or toxicity.

There are currently no consistent clinical guidelines for when or how to change doses when women become pregnant. Literature regarding lamotrigine concentrations during pregnancy partly inform a pre-emptive dosing strategy.⁹ Even when therapeutic drug monitoring is available, there is no consensus on optimal timing for obtaining concentrations.¹⁰ Variation in recommendations include the obtainment of drug concentrations every 4 weeks after conception or arbitrary changes in the lamotrigine dose by 20–25% of preconception dose at every instance where the pregnancy lamotrigine concentrations fall below a reference preconception lamotrigine plasma concentration established for that individual.^{9, 11} Even when therapeutic drug monitoring is employed, there is no consensus on optimal timing for obtaining samples for concentration measurement.¹⁰ A recent study exploring dose changes during pregnancy used dose increases of 25 mg, 50 mg, and 25 mg for each subsequent trimester, assuming a preconception dose of 150 mg.¹² However, the study did not address the potential clinical impact of two subpopulations of pregnant women with differing clearance changes and the effects of differing preconception doses and dose change regimens,

Caring for a pregnant patient with epilepsy without therapeutic drug monitoring or comprehensive dosing guidelines may result in increased seizure frequency or unnecessary concentration overshoot, potentially affecting mother and baby. Although therapeutic drug monitoring is currently an optimal strategy,⁵ it is not always available in every clinical situation. Enrolling pregnant women on particular dosing regimens is not practical. Moreover, dosing recommendations that can sustain seizure control in both the HC and LC subpopulations throughout the duration of pregnancy are needed. Simulation strategies support research in special populations where clinical trials would be challenging or unethical. Since a clinical trial of dosing strategies in pregnant patients would be very difficult, simulations based on known pharmacokinetic data are a viable option to predict concentrations in patients, presenting the best- and worst-case scenarios resulting from various dosing regimens. In this study, we aimed to predict lamotrigine concentrations and potential development of loss of seizure control or potential adverse effects from various dose adjustment regimens in pregnant women with epilepsy, based on published pharmacokinetic data. In particular, we evaluated the proposed clinically-relevant dosing strategies with consideration of how therapy differs between the two pregnancy subpopulations and with different preconception starting doses.

Methods

Model Development

The lamotrigine pharmacokinetic parameters used in the model were previously estimated from published data in nonpregnant and pregnant women with epilepsy (Table 1).^{7, 13} Preconception parameter values including absorption rate constant (3.17 hr⁻¹), volume of distribution (90/61 kg), and clearance (2.15 L/hr/61kg) were included in the model, and pregnancy-related changes in clearance were included as a slope for HC (0.12 week⁻¹) and LC (0.0115 week⁻¹).⁷ Based on previous studies, a one-compartment model was used with gestation-related changes modeled as a linear function of gestational age for clearance and allometric weight function for volume.^{7, 13} Between subject variability for all parameters except absorption rate constant were set to 30–46.7% with a residual unexplained variability of 30% based on a previously published model.⁷ In addition, an LC and HC group were simulated to reflect both pregnancy subpopulations.⁷ Women were assumed to be on lamotrigine monotherapy with no interacting concomitant medications, a common scenario during pregnancy. Concentration values below 0.65 RTC were used to represent increased risk of seizure.⁵ As toxicity thresholds are not possible to determine, we used a doubling of baseline concentrations as a measure of “overshoot” to indicate unnecessary fetal exposure and/or potential toxicity to the mother. The model was developed and simulated in the R package mrgsolve (Metrum Research Group v 0.10.0).¹⁴

Table 1.

Final parameters used in the individual and population level simulations

PARAMETER (unit)	VALUE	BETWEEN-SUBJECT VARIABILITY (%)
Preconception Parameters
Absorption Rate Constant (hr−1)	3.17	-
Apparent Volume of Distribution (L/61kg)	90	30
Apparent Clearance (L/hr/61kg)	2.15	40.6
Pregnancy-Related Changes
Clearance: Slope for HCa (week−1)	0.12	46.7
Clearance: Slope for LCb (week−1)	0.0115	46.7
Residual Unexplained Variability (%)	30

^aHC: Pregnant Woman with High Clearance Change

^bLC: Pregnant Woman with Low Clearance Change

Individual Level Simulations

Individual simulations provided predictions for two hypothetical, representative patients (HC and LC) under various dosing conditions. Typical values (population averages) for lamotrigine pharmacokinetic parameters were used for individual simulations.^{7, 12} The scenarios simulated for these individuals were that her physician made: a) no change in lamotrigine dose during pregnancy and b) increased lamotrigine dose once by either 25 mg or doubling the preconception dose at first instance of 0.65 RTC. Additional simulations were conducted to evaluate the effects of different preconception clearances, resulting in different preconception baseline concentrations, have on a single regimen change for a typical HC and LC patient, where the lamotrigine dose was doubled twice during pregnancy when serum concentrations fell below 0.65 RTC. The effects of dose increases were simulated in both populations.

Population Level Simulations

A second set of simulations were performed at the population level. Lamotrigine clearance values, between subject variability, and distribution of populations were assumed as described in our previous study to reflect real world variability in patients.⁷ To characterize the effect of clinically plausible dose change scenarios on lamotrigine concentrations and evaluate their potential clinical consequences, we identified six dose change scenarios: no dose change, double the dose at weeks 8 and 25, dose increase of 25 mg every 15 days, dose increase of 25 mg every 4 weeks, dose increase of 50 mg every 4 weeks, and dose increase of 100 mg every 4 weeks. These scenarios were identified as pragmatic options given the understanding of the magnitude of change in clearance and clinical experience, as there are no current recommendations of empiric dose changes during pregnancy. For each dose change scenario, we categorized women into four preconception target concentration groups based on different preconception total daily doses ranging from 100–400 mg. Simulations were conducted for both HC and LC subpopulations. For each group, the six dose change scenarios were considered at each of the four preconception doses. One thousand individuals from each of the clearance change groups were simulated for each dose scenario and preconception concentration, resulting in a total of 48 simulation sets and 48,000 unique individuals. Population simulation results were assembled into a heatmap to illustrate higher and lower percentages of the population in a particular category.

Simulation Analyses

Metrics were calculated for each simulation dataset to characterize dose change scenarios in the two clearance subpopulations. The percentage of the simulated population with at least one incidence of <0.65 RTC and at least one incidence of double (2x) the preconception concentration was calculated to quantify the frequency of potential subtherapeutic or overshoot concentrations, respectively, predicted in each simulation group. Finally, the average of time to the first incidence of <0.65 RTC and time to the first incidence of 2x preconception concentration were calculated to show the time during which concentrations remained in the therapeutic window across the population for each simulation group.

Results

Individual Level Simulations

Simulation of a typical HC individual patient showed that lamotrigine concentrations fell below 0.65 RTC as early as 8 weeks gestational age if doses were not adjusted (Fig 1), whereas, lamotrigine concentrations in a LC never fell below 0.65 RTC at any dose change. Increasing the preconception dose by 25 mg and doubling of the dose in HC maintained lamotrigine concentrations above 0.65 RTC for an additional 5 and 17 weeks, respectively (Fig 1A). Although the same dose changes in a LC patient resulted in higher concentrations, the concentrations never crossed the pre-defined thresholds (0.65 RTC or 2x the preconception concentration) (Fig 1B). The predicted concentrations for women at two different preconception concentration baselines are also shown, demonstrating the effect of doubling the dose over time (Fig 2). Women who have a higher preconception baseline (Fig 2A) experience their first and second instances of 0.65 RTC at 6 and 20 weeks, whereas women with a lower preconception baseline do not reach these thresholds until 8 and 25 weeks (Fig 2B). HC women with a higher baseline also experience an additional subtherapeutic event near the end of pregnancy, potentially requiring a third dose adjustment (Fig 2A). LC, regardless of their preconception baseline, will experience significantly increased concentrations (i.e., higher than 2x preconception concentration) upon the second dose adjustment, whereas HC receiving the same dose adjustments experience similar concentrations throughout pregnancy.

Figure 1:

Single dose change with 3 dosing strategies (No dose change, 25 mg dose increases, and double dose change) for a typical HC (A) and LC (B). The dashed line refers to the start of gestation. The shaded area refers to the region between 0.65 RTC and 2x preconception concentration.

Figure 2:

Scenarios for a double dose change at 0.65 RTC for a typical individual with higher baseline concentration (A) and an individual with lower baseline concentration (B). The blue lines refer to LC and red lines indicate HC. The dashed line refers to start of gestation. The shaded area refers to the region between 0.65 RTC and 2x preconception concentration.

Population Level Simulations

Women in the LC group have very low risk of falling below 0.65 RTC even without changes in lamotrigine dose (data not shown), however, there is a high chance with most preconception doses and regimen changes for overshooting of preconception concentrations (modeled here as concentrations above the 2x preconception concentration) (Fig 3), with 63% of scenarios indicating risk in 34–100% of women. Doubling the dose results in 100% of LC being at risk of overshoot, with 100 mg every 4 weeks increase regimen change achieving similar risk. The optimal regimen for LC is no change as it limits the risk of overshoot.

Figure 3:

Clinically relevant risk profiles for LC and HC are presented, with the percent of LC experiencing concentrations above double their preconception baseline at least once during pregnancy (left) and the percent of the HC experiencing concentrations below 0.65 RTC at least once during pregnancy (right), stratified by preconception dose and dose change strategy.

Alternatively compared to LC, women in the HC group have almost no risk of reaching the toxicity threshold (data not shown). Instead, they are most at risk for lamotrigine concentrations decreasing below 0.65 RTC, though the extent can vary depending on the preconception dose and dose change regimen (Fig 3). All scenarios demonstrated some measure of risk, ranging from 7–100% of women. The scenario 100 mg every 4 weeks showed the least risk when compared with other regimen options for all preconception dose groups of HC, with 7.1 to 52.4% of women having concentrations that fell below 0.65. The percentage of the population experiencing 0.65 RTC concentrations for the 25 mg every 15 days scenario was similar, but only for low preconception doses (100 mg and 200 mg). The no change scenario resulted in all of the population being at risk of concentrations falling below 0.65 RTC. As opposed to LC and toxicity concerns, the primary concern for HC women is falling below 0.65 RTC indicating potential subtherapeutic treatment.

Time to Incidence Analysis

The average time that women are predicted to have concentrations below the 0.65 RTC threshold in either the HC or LC is shown in Figure 4. For HC all scenarios resulted in concentrations below 0.65 RTC, as early as 4 weeks of gestation for those taking 100 mg dose (100 mg every 4 weeks increase regimen) (Fig 4A). All dose change scenarios in HC predicted a decrease in concentration below the efficacy threshold between 3 to 26 weeks. Alternatively, for LC, only the no change regimen resulted in concentrations predicted to fall below 0.65 RTC during late pregnancy. Further, at all initial doses for LC in the no change regimen, simulations predicted a concentration below 0.65 RTC by 33 weeks of pregnancy (Fig 4B).

Figure 4:

Average time and standard deviation of first incidence of 0.65 RTC for HC (A) and LC (B). Regimens with no bar had no incidences of 0.65 RTC.

For each regimen, lower preconception target concentrations resulted in a shorter average time to 2x preconception concentration (i.e., potential toxicity). Individuals in the LC group take longer to reach 2x preconception concentration during pregnancy, typically later than 20 weeks. However, regimens at lower preconception doses lead to concentrations above the 2x preconception concentration threshold around 10 weeks’ gestation, particularly with 100 mg and 200 mg preconception dose and the 25 mg increase every 15 days regimen (Fig 5).

Figure 5:

Average time and standard deviation of first incidence of 2x preconception concentration for HC (A) and LC (B). Regimens with no bar had no incidences of 2x preconception concentration.

Discussion

Clinicians face a conundrum since lamotrigine concentrations are known to change dramatically during pregnancy and there are two subpopulations of women that exhibit different changes in lamotrigine clearance when taking lamotrigine monotherapy during pregnancy. Therefore, dose increases that are needed in some individuals to maintain their established target concentration will result in overshoot and possible toxicity in other women.

Individual simulations demonstrated the overall changes in and time course of predicted lamotrigine concentrations for a typical HC and LC woman. Results from our population-level simulations suggest that in the absence of a mechanism to identify HC and LC women, a pre-determined dose adjustment schedule will result in either undesirably low or high concentrations. While a recent simulation study identified a dosing regimen of 175 mg, 225 mg, and 250 mg twice daily for respective trimesters for a typical pregnant woman taking 150 mg twice daily preconception as sufficient for maintaining preconception concentrations, little patient variability was included with no adjustments for subpopulations being considered.¹² As most women are maintained to an individual optimal dose prior to pregnancy it is more relevant to consider changes in preconception dose instead of absolute doses. Arbitrary dose changes with limited consideration of patient-specific characteristics, including subpopulation status and preconception dose, could result in toxicity issues or loss of seizure control. Our results provide guidance to clinicians for balancing risk at various preconception doses for both subpopulations with clinically relevant dose change regimens. Using Figure 3, for example, we can consider a woman with brittle epilepsy, meaning her body reacts dramatically to small changes in medication, who is on a preconception dose of 200 mg/day. For her, falling below 0.65 RTC would put her at risk of a breakthrough seizure. The best regimen for dose changes during pregnancy for this brittle epilepsy patient might be 25 mg increase every 15 days, as this regimen minimizes the risk of overly low drug concentrations in both HC and LC groups. Additionally, the patient should be watched closely for signs of toxicity symptoms in the case she is LC.

Results from the individual simulations and time-to-event analysis show that approximately 8 and 25-weeks gestational age, dependent on preconception baseline clearance, are indicated as critical times for potential increased seizures and overshoot. Simulations of 48,000 patients showed that none of the empiric dosing simulation scenarios were able to maintain concentrations within the desired window for both groups (LC and HC) of women. For example, the same a priori empiric dosing strategy that reduces the risk of undertreatment more than any other regimen (100 mg increase every 4 weeks) leads to a high percentage of double preconception baseline concentrations at every preconception dose. Choosing a dosing strategy must balance the dual goals of reducing potential maternal and fetal toxicity and maintaining adequate seizure control for the woman throughout pregnancy while adjusting for the changes in lamotrigine pharmacokinetics. Our results indicate that knowing which group a patient falls into would allow selection of a safer dosing strategy. Clinical concern differs depending on if a woman is LC or HC. For example, HC women are unlikely to reach toxicity thresholds but much more likely to experience 0.65 RTC events during pregnancy. Elucidation of the timing of potential clearance changes provides guidance for important time considerations when monitoring patients. In our study the time during pregnancy when concentrations decreased below desired levels is consistent with earlier reports.¹⁵ If a concentration can be obtained early in pregnancy at approximately 8 weeks, it may be possible for the clinician to determine, based on the change or lack of change in concentration compared to dose change, whether the patient is in the HC or LC group. This categorization can then be used to guide both timing and amount of dose adjustments later in pregnancy.

If therapeutic drug monitoring is not available, individual risk could also be considered when choosing an empiric regimen. The clinician could consider two issues. The first is whether the woman has brittle epilepsy that represents a high risk to her and her baby. If so, it may be prudent to opt for possible over-dosing vs under-dosing. The second is whether the patient has any current symptoms that suggest she is at the threshold of toxicity, or if she has been on higher doses in the past without toxicity. This would provide information that might determine if very higher lamotrigine concentrations could be tolerated. Different regimens are more or less problematic, depending on starting dose. Our simulations show the timing of critical changes in pharmacokinetics, thus indicate when changes in clinical symptoms may be observed. A woman in the HC group would have lamotrigine concentrations that fall below 0.65 RTC as early as 8 weeks gestational age if doses are not increased prior to 8 weeks. On average, a single dose change will not be sufficient to maintain lamotrigine concentrations at preconception concentrations throughout pregnancy. Therefore, the amount of dose increase determines the time above the 0.65 RTC during pregnancy assuming a constant change in clearance.

Pharmacokinetic modeling is a critical mechanism to understand drug exposure changes during pregnancy, since pharmacokinetic changes in pregnant women rely on observational studies as clinical trials investigating dose changes in this population is not feasible. Currently, regular and repeated therapeutic drug monitoring during pregnancy is the only way to maintain lamotrigine concentrations in the individualized target range. However, serial monitoring is clinically challenging due to patient time, costs, or lack of availability of technology. As with all models, certain assumptions were made, including independence of lamotrigine clearance and lamotrigine dose. Clinically, women on higher lamotrigine doses may have a higher baseline clearance. Since associations between clearance and dose were not available, lamotrigine clearances for these simulations for all women were drawn from the same distribution regardless of preconception dose. Our findings may not be directly relatable to women receiving polytherapy; however, our results can potentially be used to help inform additional monitoring for these women as it provides a timeline for a medication with known pharmacokinetic changes during pregnancy.

Based on our simulations, a “one-size fits all” philosophy does not work well for empirical lamotrigine dosing in pregnant women with epilepsy based on current knowledge and reinforces the need for therapeutic drug monitoring. The simulations offer some practical guidance for clinicians without access to monitoring and emphasize that no change at all is a poor option for most women. Although some dosage regimens are shown to be beneficial, there is a substantial percentage of the subpopulations that will ultimately have subtherapeutic concentrations or concentrations that overshoot the ideally established target. Currently, prospective identification of clearance change group is not possible as the etiology of the two groups has not been identified. However, simulation presents a means to provide the extent of concentrations that would be experienced in each group of women and allow consideration of possible outcomes. For instance, it could be possible that therapeutic drug monitoring has identified a woman in the high clearance group in week 9, allowing for identification of important times to monitor for symptoms, week 8 and 25, especially when monitoring is not available. While some work has been done to explore the mechanism of clearance changes and genetic polymorphisms in the enzymes that metabolize lamotrigine,^{16, 17} further studies are needed to identify the etiology of the two clearance change groups to prospectively determine which clearance change group the woman falls in and to more accurately predict serum lamotrigine concentrations with different dosing strategies.

Choice of dosing regimen unfortunately depends on both group (unknown) and preconception dose (established). Quantifying the extent of predicted drug exposure in each group of women, leading to changes in efficacy or toxicity, can be helpful to support clinician assessment of risk to the patient. Characterization of pharmacokinetic changes can aid in filling knowledge gaps by presenting the possible clinical sequelae (i.e., seizures or adverse events) that may be experienced and elucidate the optimal time windows to monitor patients.