Sertraline is a commonly used antidepressant, including in pregnancy. Pharmacokinetic studies suggest that sertraline levels in the blood can decrease during pregnancy, particularly in the 2^nd and 3^rd trimester. However, the dose adjustment required remains uncertain. These changes in pharmacokinetics warrant further evaluation of the evidence on the pharmacokinetics, efficacy, and safety of sertraline during pregnancy to determine a suitable dosing strategy.

Physiologically-based pharmacokinetic (PBPK) computer model simulations can help determine an adequate sertraline dosing regimen during pregnancy by predicting maternal and fetal exposures (plasma concentrations of the drug over time). In our analysis, the PBPK modeling results of Almurjan and colleagues were used, alongside clinical evidence.

Antenatal sertraline dose recommendation in the Netherlands: link 
Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link

Almurjan and colleagues used the physiologically-based pharmacokinetic (PBPK) modeling software Simcyp. The software contains a population of ‘virtual’ non-pregnant adults, as well as a virtual pregnant women model. By also integrating a model representing the properties of sertraline, the software can predict sertraline concentrations after administration of various doses. First in the non-pregnant population and once accurate predictions have been achieved, in the pregnant population (Figure 1). Data from existing pharmacokinetic studies (measuring real concentration samples) in these populations are used to verify that the model predictions are correct.

Figure 1 (by Almurjan and colleagues): A workflow modeling approach for sertraline

Verification in non-pregnant adults
First, the model’s ability to predict sertraline pharmacokinetics accurately in non-pregnant adults was assessed by comparing the model predictions with data from four single-dose pharmacokinetic studies and one multiple-dose pharmacokinetics study. The model predicted sertraline plasma concentrations and other pharmacokinetic parameters in nonpregnant populations adequately.

Verification in pregnant individuals
Next, the model was verified against clinical data from 56 pregnant individuals (one study: Westin and colleagues). The range of predicted sertraline plasma concentrations was similar to the range of observed data reported in pregnancy. 

Conclusion: the PBPK model credibility is considered ‘satisfactory’ based on these verification steps.

Figure 2: Assessment of PBPK model credibility 

A more detailed description of the results of model verification results can be found in the full evidence & feasibility review.

Antenatal sertraline dose recommendation in the Netherlands: link 
Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link

After confirming that the model reliably predicts sertraline concentrations in pregnant women, various dosing regimens can be simulated to reach the intended plasma concentration (Figure 3). Although a clear target therapeutic range has not been identified for sertraline, Almurjan and colleagues used the target therapeutic range of 10 - 75 ng/mL, supported by clinical experience and data from a study by Braten and colleagues as a reference for their simulations. The optimal predicted doses were defined in such a way that less than 20% of the subjects had predicted trough plasma concentrations falling outside the therapeutic range.

Figure 3 (by Almurjan and colleagues): A workflow modeling approach for sertraline 

Impact of CYP2C19 polymorphism on sertraline plasma concentration during pregnancy
Sertraline is metabolized by the drug metabolizing enzyme CYP2C19. Genetic variation in this enzyme alters the plasma concentrations of sertraline (Dutch National Formulary, Dutch: 'Farmacotherapeutisch Kompas'). Therefore, Almurjan and colleagues first predicted the influence of this genetic variation on sertraline concentrations during pregnancy (Figure 4). They simulated a daily dose of 50 mg at different gestational ages. The predicted sertraline concentrations decreased most significantly in pregnant women with genetically ‘high’ CYP2C19 activity (UM, ultrarapid, green bars).

Figure 4 (by Almurjan and colleagues): Simulated sertraline trough plasma concentrations for various genetic CYP2C19 profiles

Dose simulations in pregnant women
Next, Almurjan and colleagues simulated sertraline concentrations during pregnancy for various CYP2C19 profiles (poor metabolizers (PMs), extensive metabolizers (EMs) and ultrarapid metabolizers (UMs, ‘high’) and performed dose-finding simulations for each profile. By investigating several doses, the optimal dose, ensuring that 80% of individuals would reach the therapeutic window, was determined. Figure 5 displays the proportion of patients with concentrations below the therapeutic target (higher chance of ineffective treatment) and above the therapeutic target (higher chance of toxicity) for various daily doses (see x-axis). The proportion of patients with concentrations below the therapeutic window was the highest in the 3^rd trimester.

 Figure 5 (by Almurjan and colleagues): Percentage of subjects with a sertraline concentration <10 ng/mL and >75 ng/mL after 25-300 mg in the 1^st (black), 2^nd (green) and 3^rd (red) trimester. 

Antenatal sertraline dose recommendation in the Netherlands: link 
Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link

Based on the simulations, Almurjan and colleagues propose the following dose regimen for pregnant women based on the CYP2C19 genotype (Table 1): 

Table 1: Dose regimens as suggested by PBPK modeling study of Almurjan and colleagues for various CYP2C19 phenotypes

Almurjan and colleagues recommend a daily dose of 100 mg during the second and third trimester of pregnancy for CYP2C19 poor metabolizers. However, based on their simulations, a daily dose of 150 mg would also results in concentrations within the therapeutic range for more than 80% of pregnant women. Hence, a maximum dose of 150 mg 2^nd and 3^rd trimester could also be considered for poor metabolizers (this is especially true given that Almurjan et al. used a conservative therapeutic range minimizing the chances of toxicity).

Antenatal sertraline dose recommendation in the Netherlands: link 
Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link

A model-informed dose as presented here is one of several pieces of information that need to be taken into account when defining pregnancy-adjusted doses for clinical use. The acceptability and feasibility of doses predicted by models is affected by multiple factors. This is also true for the model-informed doses proposed by Almurjan and colleagues:
-        Reference concentration range: Almurjan and colleagues used the target concentration range defined by Braten et al. (10-75 µg/mL) as a reference to test various dosing regimens. However, a clear relationship between concentration and effect of sertraline is lacking. In fact, sertraline dosing is typically titrated based on effect. This is true both in non-pregnant and pregnant individuals. Some individuals need low doses and low concentrations, others need very high concentrations for effective treatments.
-        20% outside therapeutic range: Almurjan and colleagues conducted various dosing simulations to determine optimal dose regimens for different sertraline CYP2C19 metabolizers. It was accepted that 20% of subjects may exhibit predicted concentrations below the reference concentration range. This suggests that a considerable portion of pregnant women may not achieve effective concentrations for the doses predicted by Almurjan and colleagues.
-        Differences in the relation between the concentration and the effect between the virtual population and real-world population of pregnant women and fetuses: Pregnancy is marked by a higher prevalence of depression and anxiety symptoms than outside of pregnancy. Changes in the clinical symptoms of pregnant individuals should also be taken into account when dosing sertraline.
-        Lack of model-informed dose for pregnant women with unknown CYP2C19-profile: Almurjan and colleagues propose a different model-informed dose for various metabolic profiles of CYP2C19. In the Netherlands, pregnancy is not a standard indication for CYP2C19 profile testing in sertraline users. Consequently, a dose recommendation for pregnant women without a known CYP2C19 profile is necessary.

A thorough review of the evidence on sertraline pharmacokinetics, efficacy and safety during pregnancy was conducted by the MADAM working committee to establish a sertraline dosing regimen ensuring effective and safe treatment of depression or anxiety symptoms during pregnancy.
 
On average, the plasma level of sertraline decrease by 10-20% during pregnancy. Most women have already initiated sertraline treatment before pregnancy and their dose has been titrated according to effect. Adjusting these women’s prescribed sertraline regimen once pregnant based on the regimen predicted by Almurjan may lead to underdosing or excessive doses in certain women. While it may be intuitive to increase doses in all women from the 2^nd trimester onwards, this may not be needed in all pregnant women. Therefore, clinical guidance for pregnant women, taking into account predictions from Almurjan and other pharmacokinetic studies, indicates the potential for decreased sertraline concentrations during pregnancy. This may require a dose increase in the event of worsening symptoms. 

Executive summary for healthcare practitioners: link
Executive summary for patients: link
Antenatal sertraline dose recommendation in the Netherlands: link
Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link

• Almurjan et al. 2021 (PBPK modeling study), PMID: 33851424, link
• Braten et al. 2020 (therapeutic range), PMID: 31649299, link
• Sertraline clinical dosing guidelines for non-pregnant patients: link
• Antenatal sertraline dose recommendation in the Netherlands: link
• Publication: Full evidence & feasibility review by Koldeweij et al. (2024): link