Population Pharmacokinetics of Bictegravir During Pregnancy and Postpartum: Role of Adherence in Maintaining Therapeutic Exposure

Abstract

Bictegravir is an integrase strand transfer inhibitor available in fixed-dose combination with emtricitabine and tenofovir alafenamide for HIV treatment. The objectives of this study were to develop a population pharmacokinetic model for bictegravir during pregnancy and postpartum, identify main drivers of between-subject variability, and evaluate the role of adherence patterns in maintaining therapeutic exposure. Intensive bictegravir concentration-time data were used from IMPAACT 2026, a pharmacokinetic study of selected antiretroviral drugs during pregnancy and postpartum. 508 bictegravir plasma concentrations from 27 participants during the second and third trimesters of pregnancy and postpartum were utilized for model development. A one-compartment structural model best described bictegravir PK. Pregnancy increased bictegravir apparent clearance (CL/F) by 61% compared to postpartum, while Black/African American race was associated with a 32% increase in apparent volume of distribution (Vd/F). Plasma albumin concentrations were associated with a 43% decrease in CL/F and body weight was associated with a 120% increase in Vd/F over the range of observed values. Monte Carlo simulations predicted median (90% prediction interval) pre-dose bictegravir concentrations of 920 ng/mL (265–2081) during the 3^rd trimester and 3,399 ng/mL (1423.2–6390.5) postpartum, exceeding the protein-adjusted 95% effective concentration (162 ng/mL). Adherence simulations predicted a single missed dose at steady-state during the third trimester results in 43.7% of virtual subjects with concentrations below pharmacodynamic target, while two consecutive missed doses result in 90.4% with concentrations below target. These results show that while standard bictegravir dosing is effective during pregnancy, consistent adherence is critical to maintain effective therapeutic exposures.

Introduction

Antiretroviral therapy (ART) is recommended for pregnant women with HIV for both primary treatment of HIV and prevention of perinatal HIV transmission.¹ The use of potent antiretroviral regimens during pregnancy can reduce the rate of perinatal HIV transmission to 1% or less.^2–4 While antiretroviral treatment options have increased for pregnant women with HIV, pharmacokinetic (PK) and safety data on newer agents during pregnancy remain limited.

Bictegravir is an HIV integrase strand transfer inhibitor (INSTI) with once-daily dosing, a high genetic barrier to resistance and a favorable safety profile.^5,6 It is approved as a fixed-dose-combination with emtricitabine and tenofovir alafenamide (B/F/TAF) for the treatment of HIV-1 infection in both pregnant and non-pregnant people with HIV.^7,8 After oral administration, bictegravir is rapidly absorbed and reaches peak plasma concentrations within 2–4 hours. Bictegravir is highly protein bound (>99%, primarily to albumin), has a plasma half-life of approximately 17.3 hours in non-pregnant adults, and is metabolized both by hepatic cytochrome P450 3A (CYP3A) and uridine diphosphate glucuronyltransferase type 1A1 (UGT1A1), with each contributing to ~50% of its metabolism. Bictegravir is excreted as parent drug and oxidative metabolites in feces (60%) and as glucuronide and other oxidative metabolites/conjugates in urine (35%).^5,7

Pregnancy introduces significant physiological changes – such as increased renal blood flow, altered hepatic enzyme activity, and expanded plasma volume – impacting drug disposition.^9–11 For many antiretroviral drugs, the physiologic changes of pregnancy reduce drug concentrations, potentially compromising maternal HIV viral suppression.¹² Unbound drug concentrations may also be affected, particularly for those with high protein binding.¹³ Inadequate HIV viral suppression during pregnancy may negatively impact maternal health, lead to emergence of maternal antiretroviral drug resistance, and increase the risk of perinatal transmission, potentially with drug-resistant HIV strains. For bictegravir, reduced exposure during pregnancy may be related to induction of hepatic metabolism via CYP3A and UGT1A1 enzymes, along with pregnancy-associated increases in plasma volume and reductions in plasma protein concentrations, which together enhance drug clearance and distribution.^9–11,14

We previously reported that plasma bictegravir drug exposure (AUC_0-tau) was 46% lower in the second trimester (n = 12), and 52% lower in the third trimester of pregnancy (n = 24), compared with paired postpartum data.¹⁴ Importantly, 88%–92% of participants in our study maintained viral suppression (e.g. <40 copies/mL) throughout pregnancy and postpartum and none of the 25 infants with follow-up tested positive for HIV. Median (Q1-Q3) bictegravir trough concentrations during the second and third trimesters were 1130 (590–1370) ng/mL and 950 (690–1210) ng/mL, respectively. These exposures are below the bictegravir trough concentrations reported in non-pregnant adults with HIV (2610 ng/mL) yet remain above the bictegravir protein-adjusted 95% effective concentration (paEC₉₅, 162 ng/mL).^14–17 Zhang et al. reported that the geometric least-squares mean ratios for bictegravir AUC_tau during pregnancy ranged from 41% to 45% compared to postpartum, while mean trough concentrations remained more than 6.5-fold above the paEC₉₅ in the second and third trimesters.¹⁸ However, the ratio of trough concentration to paEC₉₅ is relatively less during pregnancy, suggesting that treatment adherence to bictegravir will be critical in optimizing outcomes for pregnant women with HIV and their infants.¹⁴

Population pharmacokinetic (popPK) models of bictegravir have been developed for non-pregnant adults with HIV.¹⁹ However, to date, the popPK of bictegravir in pregnant women with HIV has not been reported. An understanding of the patterns of between-subject variability (BSV) in bictegravir exposure among pregnant and postpartum women is crucial for identifying the key factors that influence drug concentrations. Thus, the objectives of this analysis were to develop a popPK model to describe bictegravir concentrations during pregnancy and postpartum, identify the main drivers of BSV, and evaluate the role of adherence patterns in maintaining therapeutic exposure.

Methods

Study Population:

Pregnant women living with HIV of at least 20 weeks gestational age, receiving oral bictegravir as part of antiretroviral (ARV) therapy but not receiving concomitant tuberculosis (TB) medications were enrolled. Participants were enrolled in the International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) 2026 study (ClinicalTrials.gov NCT04518228) from the clinical sites in the US and Thailand.

All maternal participants provided written informed consent on their own behalf and the protocol was approved from all required institutional review boards, ethics committees, and applicable regulatory entities at all participating sites.

Drug Administration and Pharmacokinetic Sampling:

Participants received bictegravir 50 mg orally once daily as a component of B/F/TAF. Prior to PK sampling, participants were required to self-report adherence to bictegravir for the preceding two weeks and confirm consistent dosing times for the most recent three doses. Participants were enrolled between the 2^nd or 3^rd trimester. Intensive PK sampling was performed in the 2^nd trimester if enrolled before 27 weeks gestation, and for all participants during the 3^rd trimester and 6–12 weeks postpartum. On sample days, plasma samples were collected pre-dose, and at 1, 2, 4, 6, 8, 12, and 24 hours following an observed dose.

Bictegravir Plasma Concentration Measurements:

Quantitative determination of bictegravir in human plasma was accomplished by a validated high-performance liquid chromatography tandem mass spectrometry (LC-MS/MS) method as previously described.¹⁴

Population Pharmacokinetic Analysis:

Using the computer software NONMEM (version 7.5) with a GNU Fortran G77 compiler, bictegravir concentration-time data were fit using first-order conditional estimation (FOCE) method with interaction. A one-compartment PK structural model (ADVAN2, TRANS2 subroutine) with first-order absorption was initially used to describe the data prior to exploring alternative models, including a two-compartment model, zero-order absorption, and sequential zero- and first-order absorption models. The one-compartment model had the following parameters: apparent clearance (CL/F), apparent volume of distribution for the central compartment (Vd/F), and absorption rate constant (KA). BSV was described using an exponential-normal distribution error model, given by the equation:

$${\theta }_i={\theta }_{pop}\times e^{{\eta }_i}$$

where ${\theta }_i$ represents the individual parameter value, ${\theta }_{pop}$ represents the typical population value, and ${\eta }_i$ is a normally distributed random variable with a mean of 0 and variance ${\omega }^2$. The residual error model was evaluated using proportional, additive, and combined error models. The choice of the structural model, BSV, and residual error structure was guided by statistical criteria and a holistic evaluation of goodness-of-fit plots and parameter precision.

Age, time-varying body weight, plasma HIV-1 RNA viral load, hematocrit, CD4 cell count, aspartate transaminase (AST), alanine aminotransferase (ALT), total bilirubin, albumin, alpha-1-acid glycoprotein, serum creatinine, pregnancy status (2^nd and 3^rd trimester or postpartum), and race (Black/African American or Other) were evaluated as potential covariates for CL/F and Vd/F. Potential categorical covariates were included one at a time as a linear model and continuous covariates were included one at a time by a power model centered on the median value of the population in the analysis.

Categorical covariates (COV) were modeled using a power model approach:

$$COV=1;\left(groupofinterest\right)$$

$$COV=0;\left(referencegroup\right)$$

The structural parameter $\left(\theta \right)$ was adjusted as:

$$\theta ={\theta }_0·{{\theta }_{COV}}^{COV}$$

where ${\theta }_0$ represents the parameter estimate for the reference group and ${\theta }_{COV}$ represents the proportional change in the parameter associated with the covariate.

Continuous covariates ($\text{COV}$) were modeled using a power model:

$$COV={(x/x_{median})}^{{\theta }_x}$$

The structural parameter $\left(\theta \right)$ was adjusted as:

$$\theta ={\theta }_0·COV$$

where ${\theta }_0$ represents the parameter estimate when the covariate $x=x_{median}$. The exponent ${\theta }_x$ quantifies the extent and direction of the covariate’s effect on the parameter.

PK parameters were assumed to be log-normally distributed. An initial screening step was performed where covariates were retained if their inclusion in the model decreased the objective function value (OFV) by at least 3.84 (p ~ 0.05) points. Covariates that passed this screen were then evaluated in a full stepwise covariate modeling procedure, using a forward selection criterion of an OFV drop of ≥ 8.0 (p < ~0.005) followed by backwards elimination, where covariates were retained only if their removal increased the OFV by ≥ 10.83 (p < ~0.001). Empiric Bayesian estimates of individual PK parameters were generated from the final model using the POSTHOC routine. A non-parametric bootstrap (1000 replicates) assessments of the final model were performed using Wings for NONMEM (v. 7.4.1).

Monte Carlo simulations of the final popPK model were performed to evaluate differences in bictegravir exposures between pregnancy (3^rd trimester) and postpartum. The simulations accounted for variability between individuals and random unexplained variability in drug concentrations, reflecting the typical variability observed in the study population. Concentration-time profiles were generated for 2000 simulated subjects using Monte Carlo simulations based on four patient subgroup scenarios (500 subjects generated per subgroup): Black/African American and pregnant (3^rd trimester), Black/African American and 6 – 12 weeks postpartum, non-Black/African American and pregnant (3^rd trimester), and non-Black/African American and 6 – 12 weeks postpartum. Virtual subjects were generated by fixing key demographic and clinical covariates, including race, pregnancy status, and weight. Weight and albumin values were assigned using the median values observed between subgroups stratified by pregnancy status. These simulated profiles were used to compare predicted concentrations with observed data.

Simulations of unbound bictegravir concentrations were performed by scaling model-derived total bictegravir concentrations using previously reported percent bound estimates during pregnancy and postpartum. To capture variability, the interquartile range (IQR) of the previously reported percent bound estimates was applied to the predicted total concentrations, and the median and IQR of the resulting unbound concentrations were visualized.¹⁴

Simulations of adherence patterns were performed by evaluating three dosing scenarios at steady state: (i) no missed doses, (ii) one missed dose occurring at the scheduled dosing time 24 hours after a taken dose, and (iii) two consecutive missed doses occurring at 24 and 48 hours after a taken dose (500 replicate simulations for each dosing scenario for each group, for a total of 6000 simulations). For both simulations of unbound bictegravir concentrations and simulations of varying dose adherence scenarios, the probability of target attainment (PTA) was calculated by determining the proportion of simulated subjects with concentrations above the paEC₉₅ target for wild-type HIV-1 (162 ng/mL) or the unadjusted target concentration of 0.405 ng/mL (i.e. 162 ng/mL at 99.75% bound).^15–17,20 PTA values were used to assess the likelihood of maintaining therapeutic concentrations for unbound bictegravir concentrations and across different groups and dosing scenarios. Goodness-of-fit plots, observed and simulated concentration-time profiles were generated with the R programming language (version 4.3.1, R Foundation for Statistical Computing, Vienna, Austria).

Results

Participants:

A total of 508 plasma concentrations obtained from 27 participants were utilized for model development (96, 216, 196 and concentrations from the 2^nd trimester, 3^rd trimester, and postpartum, respectively). Demographics and participant characteristics are shown in Table 1. Fifteen participants (~56%) were Black/African American. Median body weight was 87 kg (2^nd trimester), 90 kg (3^rd trimester), and 80 kg (postpartum). Median albumin concentrations were 3.6 g/dL (2^nd trimester), 3.3 g/dL (3^rd trimester), and 4.5 g/dL (postpartum). Observed bictegravir concentrations are shown in Figure 1. Median concentrations in the 2^nd and 3^rd trimesters were largely overlapping, suggesting similar drug exposure across these pregnancy stages. Given the similarities in exposure between the 2^nd and 3^rd trimesters, these periods were grouped as a single “pregnant” category for the covariate analysis.

Table 1.

Participant Characteristics

Characteristic	Period
	Second Trimester, N = 121	Third Trimester, N = 271	Postpartum, N = 251
Age (years)	31.1 (22.7, 39.8)	31.8 (17.8, 44.1)	32.1 (18.0, 43.2)
Weight (kg)	87 (56, 122)	90 (46, 149)	80 (44, 141)
HIV-1 RNA (copies/mL)2	<20 (<20, 553)	<20 (<20, 429)	<20 (<20, 10300)
CD4 (cells/mm3)3	562 (27, 1526)	692 (204, 1814)	851 (817, 884)
AST (U/L)	17 (9, 27)	16 (8, 41)	23 (9, 116)
ALT (U/L)	9 (5, 23)	11 (6, 29)	18 (7, 187)
Albumin (g/dL)	3.60 (3.20, 4.00)	3.30 (2.90, 4.10)	4.50 (3.60, 4.80)
AAG (ng/mL)	40 (22, 91)	34 (19, 103)	80 (41, 264)
Total Bilirubin (mg/dL)	0.20 (0.15, 0.70)	0.20 (0.17, 0.90)	0.40 (0.16, 1.70)
Hematocrit (%)	33.5 (28.8, 35.9)	33.1 (25.8, 38.0)	37.8 (33.3, 42.9)
Serum Creatinine (mg/dL)	0.53 (0.40. 0.68)	0.53 (0.40, 0.70)	0.77 (0.56, 1.00)
Race4
Black or African American	7 (58%)	15 (56%)	14 (56%)
White	2 (17%)	4 (15%)	4 (16%)
Asian	1 (8.3%)	2 (7.4%)	1 (4.0%)
American Indian or Alaska Native	0 (0%)	1 (3.7%)	1 (4.0%)
Missing	2 (17%)	5 (19%)	5 (20%)

AAG, alpha-1-acid glycoprotein; ALT, alanine transaminase; AST, aspartate transaminase; CD4, clusters of differentiation 4; HIV-1, human immunodeficiency virus type 1; LOD, limit of detection; RNA, ribonucleic acid

¹Median (range); n / N (%)

²Many participant’s HIV-1 RNA counts were below the assay’s LOD (< 20 copies). 1 participant with HIV-1 RNA count of 10,300 copies was confirmed to be non-compliant with their treatment regimen.

³Missing 23 values from postpartum samples

⁴Most (96%) participants were enrolled in the United States; one was enrolled in Thailand.

Figure 1.

Observed steady-state total bictegravir concentrations in pregnant (2^nd and 3^rd trimester) and postpartum women with HIV receiving bictegravir 50 mg once daily as a component of B/F/TAF. Lines represent median concentrations.

Population Pharmacokinetic Analysis:

A one-compartment structural model with first-order absorption adequately described the data. In the univariate screen, weight, hematocrit, CD4 cell count, AST, ALT, total bilirubin, albumin, alpha-1-acid glycoprotein, serum creatinine, and pregnancy were identified as potential covariates for CL/F, while weight and race were identified as potential covariates for Vd/F.

Following multivariate analysis, albumin and pregnancy on CL/F and weight and race on Vd/F were significant covariates retained in the final model. The final popPK model described the data without bias (Figure 2). The final model parameter and variance estimates, along with the final model equations, are shown in Table 2. Bootstrap evaluations of the final model successfully converged 99.3% of the time. The final parameter estimates fell within the 95% confidence interval of the bootstrap analysis, supporting the reliability of parameter estimates. To further assess the predictive performance of the final model, observed median steady-state bictegravir `concentrations were overlaid with simulated concentrations from post-dose Monte Carlo simulations of 2000 virtual subjects. The close agreement between observed and simulated medians and interquartile ranges supports the adequacy of the model in capturing central tendency and variability in both pregnant and postpartum populations (Figure 3). Pregnancy was the most significant covariate in the model and was associated with a 61% increase in CL/F compared to being postpartum. Weight provided a 120% increase in Vd/F over the range of body weight (44 – 149 kg) present in the study population, while albumin provided a 43% decrease in CL over the range of albumin concentrations (2.9 – 4.8 g/dL). Black/African American race led to a 32% increase in Vd/F.

Figure 2.

Diagnostic plots for the final popPK model. (A) Individual predicted total bictegravir concentrations vs. observed concentrations for the 2^nd trimester, 3^rd trimester, and postpartum. Grey line represents line of unity. (B) Population predicted total bictegravir concentrations vs. observed concentrations for the 2^nd trimester, 3^rd trimester, and postpartum. Grey line represents line of unity. (C) Conditional weighted residuals vs. time after dose for the 2^nd trimester, 3^rd trimester, and postpartum. (D) Conditional weighted residuals vs. population predicted total bictegravir concentrations for the 2^nd trimester, 3^rd trimester, and postpartum.

Table 2.

Final PopPK Model Parameters and Bootstrap Estimates

Parameters	Final Parameter Estimates	Standard Error of Estimates	Bootstrap Estimatesa Median (95% CI)
θ1 (Vd/F; L)	12.6	0.897	12.5 (10.9 – 14.5)
θ2 (CL/F; L/h)	0.549	0.0457	0.551 (0.457 – 0.661)
θ3 (KA; h⁻¹)	1.88	0.444	1.88 (1.22 – 3.13)
θ4 (Pregnancy on CL)	1.61	0.166	1.62 (1.26 – 2.02)
θ5 (Albumin on CL; g/dL)	−1.13	0.461	−1.12 (−2.12 – −0.22)
θ6 (Weight on V; kg)	0.636	0.151	0.642 (0.297 – 0.940)
θ7 (Race on V)	1.32	0.0938	1.33 (1.12 – 1.53)
Variability (η)	CV (%)	Standard Error of Estimates	Bootstrap Estimatesa Median (95% CI)
Between-subject (Vd/F)	8.41	0.0502	6.00 (0.20 – 14.70)
Between-subject (CL/F)	28.5	0.0414	27.6 (19.2 – 35.2)
Error (ε)	CV (%)	Standard Error of Estimates	Bootstrap Estimatesa Median (95% CI)
Proportional	28.7	0.0201	28.2 (24.7 – 32.1)

CI, confidence interval; CL/F, apparent clearance; CV, coefficient of variation; KA, first-order absorption rate constant; Vd/F, apparent volume of distribution

^aBootstraps successfully converged 99.3% of the time.

$\left.Vd/F\left(L\right)=12.6\times {\left.\left(Weightinkg\right./70\right)}^{0.64}\times \left(1.32\right.ifRaceisBlackorAfricanAmerican\right)$

$\left.CL/F\left(L/hr\right)=0.549\times {\left.\left(Albumin\right./3.7\right)}^{-1.13}\times \left(1.61\right.ifpregnant\right)$

Figure 3.

Observed and simulated median steady-state bictegravir concentrations in pregnant and postpartum individuals. Post-dose Monte Carlo simulations (n = 2000) were conducted using the final popPK model for bictegravir in persons living with HIV. Simulated participants received the FDA-approved 50 mg once-daily oral dose. Shaded and open circles represent the observed median steady-state bictegravir concentrations during pregnancy and postpartum, respectively, with error bars indicating the interquartile range (25th–75th percentile). The solid and dashed lines represent simulated median concentrations. The horizontal dotted line at 162 ng/mL indicates the protein-adjusted EC95 for wild-type HIV-1.

Monte Carlo Simulations:

Monte Carlo simulations for virtual subjects (n = 2000) at steady-state receiving a 50 mg daily dose of bictegravir demonstrated a median (90% prediction interval) steady state concentration at time 0 (C0) of 920 ng/mL (265 – 2081) at the 3^rd trimester and 3399 ng/mL (1423 – 6391) postpartum. In simulated pregnant and postpartum subjects, over 99% of concentrations across the 24-hour dosing interval were above the minimum paEC₉₅ target therapeutic concentration of 162 ng/mL (Figure 4).^15–17

Figure 4.

Monte Carlo simulations of total steady-state bictegravir concentrations in pregnant and postpartum women with HIV receiving bictegravir 50 mg once daily. Solid lines display median concentrations and shaded area shows 90% (5–95) prediction intervals. Horizontal dotted lines at 162 ng/mL represent the pharmacodynamic target.

Simulations of unbound bictegravir concentrations are shown in Figure 5. The simulated median (IQR) C0 (ng/mL) of unbound bictegravir was 3.04 (2.39–4.23) at the 3^rd trimester, corresponding to a median free fraction of approximately 0.33% and 8.50 (7.14–9.52) at 6 – 12 weeks postpartum, corresponding to a median free fraction of approximately 0.25%. Over 99% of simulated subjects, regardless of antepartum or postpartum status, were above the unadjusted target of 0.405 ng/mL (162 ng/mL at 99.75% bound).²⁰

Figure 5.

Simulations of unbound steady-state bictegravir concentrations in pregnant and postpartum women with HIV receiving bictegravir 50 mg once daily. Solid lines display median concentrations and shaded area shows the interquartile range. Horizontal dashed lines at 0.405 ng/mL represents the unadjusted EC95 pharmacodynamic target.

Monte Carlo simulations for virtual subjects (n = 6000) assessing the impact of treatment adherence on total bictegravir concentrations are shown in Figure 6. With no missed doses at steady state, the median (90% prediction interval) trough concentration in the 3^rd trimester was 930.56 ng/mL (307.8 – 2228.5) and within 6–12 weeks postpartum was 3406.1 ng/mL (1499.2 – 6771.8). With one missed dose while at steady state, the median (90% prediction interval) trough concentration (just prior to resuming dosing) in the 3^rd trimester was 239.3 ng/mL (30.1 – 845.0) and postpartum was 1613.8 ng/mL (500.7 – 4261.3). With two consecutive missed doses while at steady state, the median (90% prediction interval) trough concentration (just prior to resuming dosing) in the 3^rd trimester was 41.5 ng/mL (1.56 – 305.3) and postpartum was 871.4 ng/mL (176.5 – 2695.5). In pregnant (3^rd trimester) virtual subjects who missed no doses, one dose, and two consecutive doses, 4.1%, 43.7%, and 90.4% of simulated subjects, respectively, had concentrations that fell below the target paEC₉₅ (162 ng/mL). Across all dosing scenarios in postpartum virtual subjects, over 96% of subjects maintained concentrations above the pharmacodynamic target (162 ng/mL).^15–17

Figure 6.

Monte Carlo simulations of total bictegravir concentrations assessing different adherence scenarios in pregnant and postpartum women with HIV. Data represent median concentrations and horizontal line at 162 ng/mL shows the pharmacodynamic target. Solid lines show steady-state (50 mg once daily) with no missed doses. Dashed lines show one missed dose. Dotted lines show two consecutive missed doses.

Discussion

This study aimed to develop a popPK model to characterize bictegravir disposition during pregnancy (2^nd and 3^rd trimesters) and postpartum, using intensive pharmacokinetic data from the IMPAACT 2026 study. Pregnancy was associated with significantly increased bictegravir CL/F compared to postpartum, while albumin concentrations were inversely correlated with bictegravir CL/F. Increasing weight and Black/African American race were associated with significantly increased Vd/F.

Bictegravir is a low hepatic extraction ratio drug, such that clearance is primarily governed by the product of the fraction unbound (f_u) and intrinsic clearance (CL_int) rather than hepatic blood flow.¹⁴ During pregnancy, reduced protein binding (increased f_u) can lead to an increase in CL/F, as observed in this study. Pregnancy also induces other physiological changes that may alter drug disposition. The increased activity of hepatic CYP3A4 and UGT1A1 during pregnancy may further contribute to an enhanced CL/F of bictegravir, which is primarily metabolized by these enzymes in the liver.^5,15 This increased clearance during pregnancy results in lower systemic drug exposure, which may have negative implications for maintaining therapeutic drug concentrations. Bictegravir metabolites were not measured in this study; however, quantifying metabolite concentrations could provide additional mechanistic insight into the contribution of enhanced clearance observed during pregnancy. The inverse relationship between albumin and CL/F suggests that as albumin concentrations decrease in pregnancy, protein binding decreases, effectively increasing f_u and thereby increasing CL/F. Black/African American race and higher weight were associated with increased Vd/F, potentially reflecting differences in body composition and tissue distribution. These pharmacokinetic changes contrast with the early postpartum period when the activity of the enzymes, clearance, and volume of distribution are expected to gradually return toward pre-pregnancy levels.²¹ In this study, postpartum sampling occurred between 6–12 weeks after delivery, a period commonly used to represent the early postpartum state. While this interval reflects the resolution of many pregnancy-related changes, it may not fully correspond to the non-pregnant steady-state condition.

Bictegravir is an important therapeutic option in the management of HIV during pregnancy due to its potency, favorable safety and tolerability profile, once daily dosing without pharmacokinetic boosting and high genetic barrier to resistance, making it preferred to first generation INSTIs such as raltegravir and elvitegravir. While elvitegravir requires pharmacokinetic boosting with the CYP3A inhibitor cobicistat, which increases the risk of drug-drug interactions (DDI), bictegravir does not require boosting, simplifying treatment. Additionally, bictegravir is co-formulated with tenofovir alafenamide (TAF) and emtricitabine (FTC), providing a more favorable renal and bone toxicity profile compared to formulations containing older nucleoside reverse transcriptase inhibitors such as tenofovir disoproxil fumarate. Dolutegravir, another potent INSTI, may be prescribed with abacavir and lamivudine. Use of abacavir delays immediate treatment initiation as HLA-B*5701 testing is recommended prior to starting therapy to avoid hypersensitivity reactions. In addition, abacavir carries potential cardiovascular risks.^5,6,22,23 Compared with dolutegravir-based triple-drug fixed-dose combinations, the B/F/TAF formulation also offers the advantage of a smaller single-tablet size, which may improve convenience and adherence during pregnancy.²⁴

The BSV observed in the model (8.41% for Vd/F and 28.5% for CL/F) suggests that while pregnancy plays a major role in influencing pharmacokinetic parameters, there are additional, unexplained factors contributing to variability in drug exposure. These factors may include biological and host genetics that differ between individuals, such as differences in body composition, metabolic enzyme activity, or transporter function. Notably, metabolic enzymes such as UGT1A1 and CYP3A, and transporters like P-gp and BCRP, could all contribute to the observed BSV in drug exposure.^25,26 Understanding the underlying mechanisms contributing to this variability could further refine population pharmacokinetic models and potentially enable more individualized treatment strategies. Simulations in four patient groups – pregnant Black/African American women, postpartum Black/African American women, pregnant non-Black/African American women, and postpartum non-Black/African American women living with HIV – receiving the standard adult dose 50 mg bictegravir orally once daily as a component of B/F/TAF resulted in the majority of simulated subjects (>99%) maintaining concentrations above the paEC₉₅ pharmacodynamic target for wild-type HIV-1.^15–17 Simulations of unbound bictegravir concentrations also demonstrated >99% of simulated subjects maintained concentrations above the unadjusted target of 0.405 ng/mL (162 ng/mL at 99.75% bound). Maintaining bictegravir concentrations above this target level is generally thought to be critical for achieving effective HIV-1 viral suppression, reducing the likelihood of viral rebound, development of resistance, and risk of vertical transmission with inherent risk of primary transmitted resistance to the INSTI class of antiretroviral drugs among infants who acquire HIV. However, the precise minimum effective trough concentration for bictegravir has not been formally established, and prior studies have used different targets (e.g. 2x or 4x the PA-IC95).

Black/African American race was identified as a significant predictor of Vd/F in the popPK model. This finding aligns with evidence from trials such as ADVANCE, which have shown that Black/African American women experience more pronounced weight gain compared to other groups during treatment with INSTIs, especially when combined with TAF. This increase in weight gain could potentially affect bictegravir pharmacokinetics by influencing body composition, which in turn alters distribution. However, stratified simulations indicated that racial differences in PK exposure were relatively small compared to the effect of pregnancy, with pregnancy remaining the primary driver of PK differences. As a result, the racial category was pooled in the simulations, focusing on the broader groups of pregnant and postpartum women living with HIV. Despite the lower concentrations observed in pregnancy compared to postpartum, the majority of subjects in the simulated pooled population maintained concentrations above the paEC₉₅.

Bictegravir is highly protein-bound (>99%) and pregnany-related changes in protein binding can influence the free drug concentration responsible for antiviral activity.^5,7 While total drug concentrations have historically been used for dosing decisions, our analysis also assessed unbound bictegravir concentrations and demonstrated that unbound bictegravir exposure remained above the pharmacodynamic target throughout the dosing interval, assuming no doses were missed. While there were not enough data to model unbound concentrations directly, future studies should aim to evaluate whether unbound bictegravir concentrations provide a more precise measure of pharmacologic activity in pregnancy, better guiding dose recommendations during pregnancy.

To further evaluate bictegravir exposure across pregnancy and postpartum, additional simulations explored the impact of missed doses on exposure. Monte Carlo simulations across the four participant groups modeled different adherence scenarios at steady state, including no missed doses, one missed dose, and two consecutive missed doses. Across all postpartum adherence scenarios, over 96% of simulated subjects maintained concentrations above the pharmacodynamic target even with two consecutive missed doses. However, in pregnant virtual subjects, the impact of missed doses was more pronounced. When no doses were missed, only 4.1% of virtual subjects experienced concentrations that fell below the pharmacodynamic target and in those who missed one dose, that proportion rises to 43.7%. When two consecutive doses were missed, this proportion further increases to 90.4%. These findings underscore that while the standard 50 mg daily dose of bictegravir maintains therapeutic plasma concentrations in pregnancy, consistent adherence is critical to maintain therapeutic exposure. Real-world adherence patterns, such as partial inconsistent adherence or make up doses, were not explored in these simulations.

Bictegravir has a long dissociation half-life from integrase-DNA complexes—163 hours compared to 96 hours, 10 hours, and 3.3 hours for other INSTIs like dolutegravir, raltegravir, and elvitegravir, respectively. This extended half-life provides an additional layer of protection against HIV replication, allowing for sustained viral suppression even when trough concentrations fall below the EC₉₅ threshold.²⁷ This characteristic, shared by other integrase inhibitors like dolutegravir, has been particularly relevant in rifamycin DDI studies, where individuals continued to experience viral suppression despite low plasma bictegravir and dolutegravir trough concentrations.^17,28 This extended protection could be particularly beneficial in the context of intermittent adherence, where trough concentrations may occasionally dip below therapeutic levels.

Models of bictegravir popPK in non-pregnant adults have been published. In a popPK model incorporating PK data from 8 clinical studies, data were fit to a one-compartment model with first-order absorption, lag time, and first-order elimination. While body weight, health status (healthy or living with HIV), and concomitant administration of proton pump inhibitors were statistically significant covariates retained in the final model, there was no clinically significant association between these covariates and bictegravir exposures (AUC_tau, C_max, and C_tau). The parameter estimates from the report are concordant with the estimates of our popPK model in the present analysis.¹⁹

The present study has a few limitations. The participants enrolled in the study were predominantly enrolled within the United States and may not be representative of the global population of pregnant women s living with HIV. Additionally, a majority of the participants self-identified as Black/African American race. While identification as Black/African American race was shown to be a significant covariate in the final model, other racial groups were not able to be assessed as covariates due to limited sample sizes.

Despite these limitations, this study has notable strengths. The use of intensive pharmacokinetic data from the IMPAACT 2026 cohort allowed for a comprehensive understanding of bictegravir disposition during pregnancy and postpartum, with very few missing data. The study also included data from the second and third trimesters of pregnancy and extended follow-up through 6 – 12 weeks postpartum. Our population pharmacokinetic model fit the data well and was consistent with previously published reports, strengthening the validity of the findings.¹⁹ Furthermore, simulations of unbound bictegravir concentrations, derived from the predicted total concentrations and using published protein binding estimates, led to similar conclusions regarding target attainment.

This study is the first to utilize a popPK model to simulate bictegravir adherence in pregnant women living with HIV, demonstrating the importance of adherence in maintaining therapeutic concentrations. The simulations showed that while the standard 50 mg daily dose of bictegravir as a component of B/F/TAF is effective in pregnancy, consistent adherence is essential to maintain concentrations above the pharmacodynamic target. The inclusion of adherence scenarios and their impact on drug exposure provides a novel contribution to understanding the importance of dosing consistency in pregnancy. Additionally, while the study predominantly focused on Black/African American women, the inclusion of both Black/African American and non-Black/African American women allows for important comparisons that can inform treatment strategies in different subgroups. The study’s ability to assess clinical factors such as weight, pregnancy, and race further enhances our understanding of how these variables impact drug pharmacokinetics and could help inform more personalized treatment strategies for pregnant women living with HIV.

Conclusions

In conclusion, the current study developed a popPK model for bictegravir in pregnant (2^nd and 3^rd trimester) and postpartum women living with HIV. Pregnancy was associated with an increase in CL/F while albumin concentration had an inverse relationship with CL/F. Weight and Black/African American race were independent predictors of Vd/F. Simulations demonstrate that the standard 50 mg daily bictegravir dose as a component of B/F/TAF is effective during pregnancy, though consistent adherence is critical to maintain effective therapeutic exposures.

Data Availability

The data that support the findings of this study may be available on request from the corresponding author.