Lamivudine dosing for preterm infants exposed to HIV: a population pharmacokinetic modelling and simulation study

Abstract

Objectives

To develop a pragmatic twice daily lamivudine dosing strategy for preterm infants from 24 to 37 completed weeks of gestation.

Methods

Data were combined from eight pharmacokinetic studies in neonates and infants receiving lamivudine oral solution. A population pharmacokinetic model was developed using non-linear mixed effects regression. Different lamivudine dosing strategies, stratified by gestational age at birth (GA) bands, were simulated in a virtual population of preterm infants, aimed at maintaining lamivudine drug exposures (AUC_0-12) within a reference target range of 2.95 to 13.25 µg·h/mL, prior to switching to WHO-weight band doses when ≥4 weeks of age and weighing ≥3 kg.

Results

A total of 154 infants (59% female) contributed 858 lamivudine plasma concentrations. Median (range) GA at birth was 38 (27–41) weeks. At the time of first pharmacokinetic sampling infants were older with median postnatal age (PNA) of 6.3 (0.52–26.6) weeks. Lamivudine concentrations were described by a one-compartment model, with CL/F and V/F allometrically scaled to weight. Maturation of CL/F was described using an Emax model based on PNA. CL/F was also adjusted on GA to allow extrapolation for extreme prematurity. Simulations predicted an optimal lamivudine dosing for infants GA ≥24 to <30 weeks of 2 mg/kg twice daily from birth until weighing 3 kg; and for GA ≥30 to <37 weeks, 2 mg/kg twice daily for the first 4 weeks of life, followed by 4 mg/kg twice daily until weighing 3 kg.

Conclusions

Model-based predictions support twice daily pragmatic GA band dosing of lamivudine for preterm infants, but clinical validation is warranted.

Introduction

All neonates born to women with HIV, including preterm infants (<37 completed weeks of gestation), should receive antiretrovirals for prevention or early treatment.^1–3 Lamivudine is one of three antiretrovirals used off-label in preterm infants but with limited data to support current dosing guidance in very preterm (<32 weeks gestation) infants.⁴ Because lamivudine is eliminated unchanged by the kidneys,⁵ and preterm infants have underdeveloped renal function,⁶ renal maturation impacts lamivudine dosing in this population.

Lack of evidence to guide lamivudine preterm dosing at the lower limit of viability was identified as a critical knowledge gap at the most recent WHO-convened Pediatric Antiretroviral Drug Optimization (PADO-5) meeting.⁷ Subsequently, we pooled lamivudine pharmacokinetic (PK) data from existing neonatal and young infant trials to develop a population PK model to predict lamivudine exposures in preterm infants. Model simulations were performed to determine a pragmatic lamivudine twice daily dosing strategy, stratified by two gestational age (GA) bands:  ≥ 24 to <30 weeks; and ≥30 to <37 weeks.

Materials and methods

Study population

PK data were combined from eight completed studies in neonates and young infants receiving lamivudine oral solution (10 mg/mL). These comprised three HIV prevention studies (PACTG 353, 358 and 386)⁸ and five HIV treatment studies (PACTG 300,⁹ 356,⁸ IMPAACT P1069¹⁰ and P1106¹¹) and the Early Infant HIV Treatment in Botswana study.¹²

Lamivudine population PK analysis

A population PK model to describe lamivudine plasma concentrations was developed using non-linear mixed effects regression. Lamivudine concentration–time data were fit using first-order conditional estimation with interaction to estimate lamivudine population PK parameters (NONMEM v7.5). Individual models were run using Pirana (v2.9.6, Certara) and diagnostic graphs generated using R Studio (2023.06.1). PK structural and residual models were assessed using statistical and graphical methods. The influence of infant covariates was evaluated using stepwise forward inclusion and backward elimination procedures. Covariates included: birth weight, body weight (WT), GA, postnatal age (PNA), postmenstrual age (PMA; i.e. GA plus PNA) and plasma serum creatinine. A predicted-corrected visual predictive check and sampling importance resampling (SIR) analysis was performed to evaluate parameter uncertainty and model robustness.

Model simulations of lamivudine twice daily dosing in preterm infants

A virtual population of preterm infants was generated using postnatal growth reported in the INTERGROWTH-21^st Project.¹³ Data on postnatal growth for infants from GA 24 to 37 weeks, through to 12 weeks of age, were selected for boys and girls at the 10th, 50th and 90th centiles. Monte Carlo model simulations were performed with different dosing strategies to predict drug exposures (i.e. AUC_0-12) in preterm infants starting lamivudine at birth until 12 weeks of age, and prior to switching to the WHO weight-band dose of 30 mg twice daily when ≥4 weeks of age and weighing ≥3 kg. The lamivudine geometric mean (GM) AUC_0-12 target range reported in children was 2.95 to 13.25 µg·h/mL.¹⁴

Results

A total of 858 lamivudine plasma concentrations were available from 154 neonates and infants (59% were female). The median GA at birth was 38 (range, 27–41) weeks with 34/154 (22%) infants being preterm. However, at the time of the first PK sampling, infants were older: median PMA 42.8 (range, 35.7–64.6) weeks, PNA 6.3 (range, 0.52–26.6) weeks, and body weight 3.8 (range, 1.9–7.8) kg, respectively. The median serum creatinine was 0.4 (range, 0.1–1.2) mg/dL; available for 138/154 (89.6%) infants. Missing serum creatinine data from PACTG 356 were imputed using linear regression across all serum creatinine data.

Lamivudine plasma concentrations were best described by a one-compartment model. Lamivudine CL/F and V/F were allometrically scaled to body weight. Maturation of CL/F was described using an Emax model based on PNA, which also influenced V/F. Serum creatinine was a significant factor after inclusion of PNA. GA was highly correlated with body weight and serum creatinine but did not have an independent effect on lamivudine clearance. The final model lamivudine PK parameter estimates and model validation using the available lamivudine plasma concentration data are shown in Supplemental material (available as Supplementary data at JAC Online). However, the available lamivudine PK dataset lacked data in infants <34 weeks PMA; thus, the model had limitations for infants at lower GA in the first weeks of life. As lamivudine is primarily eliminated unchanged via the kidneys, it was assumed that the ontogeny of lamivudine clearance in very young preterm infants would mirror the physiological pattern described for serum creatinine and creatinine clearance. Consequently, the lamivudine PK model was adapted to include a maturation function based on early creatinine clearance changes to extrapolate lamivudine clearance for the simulations of extremely preterm infants down to 24 weeks GA. A model describing the dynamics of serum creatinine in neonates with birthweights <1.0 kg was applied to adjust lamivudine CL/F based on GA.¹⁵ The lamivudine PK parameters for this adapted lamivudine PK model are shown in Table 1. Using this GA-adjusted model, different lamivudine dose simulations were performed, stratified into two GA-bands, (1) >24 to <30 weeks and (2) ≥30 to <37 weeks, over the first 12 weeks of life. Model-predicted lamivudine GM exposures (AUC_0-12) remained within target for preterm infants GA >24 to <30 weeks using 2 mg/kg of lamivudine twice daily from birth until 3 kg; and for neonates GA ≥30 to <37 weeks using 2 mg/kg of lamivudine twice daily for the first 4 weeks, followed by 4 mg/kg twice daily until 3 kg (Figure 1). All infants can be switched to WHO weight-band doses when ≥4 weeks of age and weighing ≥3 kg.

Table 1.

Population pharmacokinetic parameters for the adjusted lamivudine model (based on gestational age) used for the dose simulations

	Final model		SIR
Lamivudine PK parameters	Estimate	RSE, %	Median	2.5th–97.5th percentile
K a, 1/h	1.33	11	1.33	1.03–1.66
CL/F, L/h
Maturation parameters
CLBL, L/h	0.335	8	0.334	0.278–0.385
Emax	1.12	6	1.12	0.99–1.27
TM50, d	82.7	5	82.8	73.3–90.6
Hill coefficient	1.31 (fixed)	—	—	—
θTM50-GA	−2.35 (fixed)	—	—	—
θCLBL-GA	1.6 (fixed)	—	—	—
V/F, L	2.84	5	2.84	2.53–3.14
θPNA	0.112	32	0.110	0.045–0.176
Interindividual variability (IIV)
ω²CL/F	0.096	20	0.095	0.058–0.134
ω²V/F	0.078	34	0.076	0.027–0.127
Residual variability
Proportional	0.202	7	0.203	0.174–0.232

CL_BL, clearance at baseline; CL/F, oral clearance; Emax, the maximum additional achieved clearance; GA, gestational age (weeks); Ka, absorption rate constant; PNA, postnatal age (days); RSE, relative standard error; SIR, sampling importance resampling; TM₅₀, the time point (days) where half of Emax is achieved; V/F, apparent volume of distribution.

PK model equations:

TM_50i = TM_50-pop × (GA/38)^θTm50-GA

CL/F_i (L/h) = [(CL_BL-pop × (GA/38) ^θCLBL-GA) + (Emax × PNA^Hill/TM₅₀ ^Hill + PNA ^Hill)] × WT^0.75

V/F_i (L) = V/F_pop × WT × (PNA + 7/90)^θPNA

Figure 1.

Predicted lamivudine (3TC) geometric mean AUC_0-12 through 12 weeks of life: GA 24 to <30 weeks: 2 mg/kg twice daily from birth; GA ≥30 to <37 weeks: 2 mg/kg twice daily for the first 4 weeks of life, then 4 mg/kg twice daily. Dotted lines represent lamivudine geometric mean AUC_0-12 target range.¹⁴ Infants reaching ≥3 kg and aged ≥4 weeks switch to follow WHO weight-band dosing recommendations and are not included in the figure. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Discussion

A pragmatic lamivudine twice daily dosing guidance for preterm infants from GA 24 to 37 weeks was developed based on PK modelling and simulation. Our proposed lamivudine dosing for very premature infants considers both expected renal developmental changes in preterm infants and weight gain over time. Given lamivudine is primarily eliminated by glomerular filtration and renal secretion, a good understanding of renal maturation is critical when developing an optimal lamivudine dosing strategy for neonates. Glomerular filtration rate, a proxy for renal drug clearance, is slow at birth, and even lower in preterm infants. Nephrogenesis continues until 34 to 36 weeks GA, but is altered with preterm delivery and impacts on clearance of renally excreted drugs.⁶ A PK study performed in 16 term neonates administered lamivudine at 2 mg/kg twice daily, showed that lamivudine apparent clearance was low at birth, then increased ∼1.6-fold (from 0.19 to 0.34 L/h/kg) within the first week of life.¹⁶ A similar rapid increase in drug clearance was observed in the PETITE study of term neonates who received 15 mg of lamivudine within a paediatric fixed-dose combination once daily for a month.¹⁷ Based on our model the typical lamivudine clearance was approximately 30% lower in infants born at GA 24 weeks compared with infants born at GA 32 weeks, and these immature infants also experienced a slower gain in lamivudine clearance during the first weeks of life.

To support the development of a robust PK model to inform lamivudine dosing recommendations in preterm infants, we combined data from multiple studies involving neonates and infants across a range of gestational and postnatal ages, including data from older preterm infants. By incorporating these PK datasets and considering our understanding of developmental pathways, the model was strengthened to enable extrapolation beyond the available PK data. This was necessary to provide dosing guidance for extreme preterm infants. Not surprisingly, we found that infant body weight and PNA were key factors influencing lamivudine PK parameters. To overcome the lack of lamivudine PK data in very immature infants, our model incorporated the normal physiological changes in serum creatinine and creatinine clearance related to GA (as a surrogate of lamivudine clearance ontogeny) observed in preterm infants with birthweights of <1.0 kg, to be able to extrapolate lamivudine dosing down to 24 weeks GA. As several assumptions were necessary, including the relative impact of gestational age on lamivudine CL/F at birth, there are potential limitations of the adapted PK model in very preterm neonates. Of note, any predicted lamivudine dosing strategy should result in exposures that fall within the range associated with proven efficacy and safety in children.

Current US paediatric lamivudine dosing recommendations include 2 mg/kg of lamivudine twice daily from birth to 4 weeks of age, followed by 4 mg/kg twice daily until 3 months of age, for infants ≥32 weeks GA.¹ This dosing is based on studies performed in term neonates and infants^8,18 but lamivudine dosing is unavailable for very preterm infants. The WHO recommends a simplified weight-band-dosing approach for neonates weighing ≥2 and <5 kg, independent of GA. Here, the birth weight determines the lamivudine starting dose, which is continued until 4 weeks of age, when it changes to 30 mg twice daily if the infant weighs ≥3 kg.³ Our proposed lamivudine dosing for preterm infants starts at 2 mg/kg twice daily, but to account for the effects of prematurity we recommend delaying the increase in lamivudine dose to 4 mg/kg twice daily until the preterm infant reaches a body weight of 3 kg, which can be up to 14 weeks for an average preterm infant of 24 weeks GA with a birth weight of 0.64 kg.¹³ All infants, independent of GA, can be transitioned to the WHO simplified weight-band dosing of 30 mg twice daily once they reach ≥3 kg and are aged ≥4 weeks.

This proposed lamivudine dosing guidance has been endorsed by the WHO Pediatric Antiretroviral Working Group, but should be clinically validated in a small number of preterm infants. Assessment of lamivudine drug concentrations in this population can provide vital information on renal maturation and a better understanding of the developmental pathways, supporting extrapolation of other neonatal-specific PK models in preterm infants where no or limited PK data exist.

Funding

Overall support for the International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) Network was provided by the National Institute of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health (NIH) under Award Numbers UM1AI068632 (IMPAACT LOC), UM1AI068616 (IMPAACT SDMC) and UM1AI106716 (IMPAACT LC), with co-funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) and the National Institute of Mental Health (NIMH). The World Health Organization provided support for the lamivudine PK modelling/simulation analysis.

Transparency declarations

None to declare.

Author contributions

A.B. and T.R.C. conceptualized and developed the first draft of this article. All authors contributed to drafts and revisions of this article and approved the final version. This article summarizes the work and perspective of the authors and does not reflect the official position of the World Health Organization.

Supplementary data

Supplementary material, Figures S1 and S2 and Table S1 are available as Supplementary data at JAC Online.