Model-based assessment of VRC07-523LS dosing in infants through population pharmacokinetic -pharmacodynamic modelling in adults and infants

Abstract

Objectives

To develop a population pharmacokinetic (popPK) model to understand factors affecting the broadly neutralizing HIV antibody (bNAb) VR07-523LS disposition and infant predicted serum neutralization 80% inhibitory dilution titre ratio (PT₈₀) and its potential relationship to antiviral effects.

Methods

PK data from two studies were used: infants exposed to HIV initiating therapy ≤5 days of age (IMPAACT P1112, n = 21) and healthy adults (VRC605, n = 25). The infant study implemented fixed subcutaneous (s.c.) dosing whereas the adult study implemented weight-based dosing, via both intravenous (i.v.) and s.c. routes. Monte Carlo simulation assessed two doses (80 mg followed by 100 mg at 12 weeks) to determine the maintenance of levels >10 µg/mL in virtual infants. The broadly neutralizing antibodies (bNAbs) concentration/IC₈₀ ratio (PT₈₀) was calculated on the basis of simulated concentrations from the popPK model and VRC07-523LS HIV-1 sensitivity (IC80) distribution.

Results

This model predicts typical infant CL/F and t_1/2 of 159 mL/day/70 kg and 39 days. Comparatively, predicted adult typical CL/F and t_1/2 were 269 mL/day/70 kg and 31 days. Concentrations were >10 µg/mL in >87% of virtual infants at 12 weeks following one dose and >98% at 24 weeks following two doses. Median (90% predictive interval) PT₈₀ was 163 (71–383), 116 (51–274) and 61 (26–144) at 6, 8 and 12 weeks, respectively. Significant model covariates included age (infant versus adult) and multiple dosing.

Conclusions

Age and repeat dosing influence VRC07-523LS PK. An 80-mg dose followed by 100 mg at 12 weeks rapidly achieves and maintains concentrations >10 µg/mL for >24 weeks. This dosing strategy is expected to result in antiviral effects in patients with sensitive viruses and supports further evaluation of VRC07-523LS.

Introduction

Dramatic reduction in infant HIV-1 transmission has been achieved with antiretroviral therapy (ART) in pregnant and breastfeeding individuals living with HIV-1, along with antiretroviral (ARV) prophylaxis in infants. Despite interventions, an estimated 120 000 children (age 0–14 years) were newly diagnosed with HIV worldwide in 2023.¹ Late diagnosis of maternal HIV-1, incomplete adherence to ART, and HIV-1 acquisition during pregnancy and breastfeeding account for most perinatal transmissions.² While improved maternal diagnosis and adherence to treatment can reduce transmission in those with known HIV-1, better infant prophylaxis strategies are needed, especially interventions which impact HIV-1 acquisition during breastfeeding. Broadly neutralizing antibodies (bNAbs) are a novel class of agents characterized by long half-lives and infrequent dosing, resulting in high adherence, thus having the potential to prevent infant HIV-1 acquisition.³

BNAbs specific for HIV-1 have been developed for several different HIV-1 envelope protein target sites. VRC01, a prototype bNAb that binds to the CD4 binding site of gp120 on HIV-1 viruses,⁴ was tested in the Antibody Mediated Prevention (AMP) studies⁵ where it demonstrated efficacy in reducing sexual transmission of HIV-1 for viral isolates sensitive to VRC01. VRC07-523LS is a more potent bNAb that targets the CD4 binding site with activity against >95% of Clade B and C isolates.⁶ The LS mutation alters antibody metabolism, reducing clearance (CL) and prolonging serum t_1/2 compared with VRC01,⁷ thereby maintaining target serum concentration levels for >2 months after a single dose in adults. On the basis of desired pharmacokinetics (PK), safety, and tolerability, VRC07-523LS is under development as a novel agent for HIV prophylaxis and treatment in infants and adults.^6,8–11

In the AMP studies, the IC₈₀, against an HIV-1 pseudovirus in vitro alone was an incomplete predictor of prevention. Gilbert et al. suggested that PT₈₀, the ratio of the serum bNAb concentration and the IC₈₀, may be a metric to assess potential prophylaxis efficacy¹² as it incorporates both bNAb concentrations and virus sensitivity. Linking these two measurements, PT₈₀ was devised to predict prevention.

The current study developed a combined infant and adult population pharmacokinetic (popPK) model to characterize VRC07-523LS disposition and evaluate differences between the two populations. The model was used in simulations to derive PT₈₀, which can provide a rationale for infant dosing recommendations for VRC07-523LS.

Methods

Patient population

VRC07-523LS PK data from two clinical trials were combined for this analysis. The infant study was conducted by the International Maternal Pediatric Adolescent AIDS Clinical Trials (IMPAACT) Network (Study IMPAACT P1112, NCT02256631) at 14 domestic and international sites.¹¹ Infants born to a birth parent living with HIV-1, ≥36 weeks gestational age, and ≥2 kg birth weight were enrolled from July 2015 to August 2020 and received VRC07-523LS in the first 5 days of life. The study in healthy adults was conducted by the Vaccine Research Center (VRC), National Institute of Allergy and Infectious Diseases, National Institutes of Health (NIH) (Study VRC605, NCT03015181) with enrolment from February to September 2017.⁶ The studies were approved by the institutional review board for each participating site, and written informed consent was obtained from participating adults and parents/guardians of infants. The US Department of Health and Human Services guidelines governing protection of human participants involved in research, were followed.¹³

Drug administration and PK sampling

In the VRC605 study, four groups of adult participants received a single intravenous (i.v.) dose of 1, 5, 20 or 40 mg/kg of VRC07-523LS, and one group received a single 5 mg/kg subcutaneous (s.c.) dose. In addition, two groups received three doses of either 20 mg/kg i.v. VRC07-523LS or 5 mg/kg s.c. VRC07-523LS at 12-week intervals. PK samples were collected from participants up to 24 weeks after the last dose (Table 1).

Table 1.

Baseline demographics and study designs

	IMPAACT	VRC
Study	P1112, Arm 5	Study 605
	(N = 22)	(N = 25)
Study population	Infants (HIV-exposed)	Adult
i.v. administration dose (N)	N/A	Single dose
		1, 5, 20, 40 mg/kg × 1 (12)
		Multiple dose
		20 mg/kg q 12 weeks × 3 (5)
s.c. administration dose (N)	Single dose (non-breast-fed)	Single dose
	80 mg at birth (12)	5 mg/kg × 1 (3)
	Multiple dose (breast-fed)a	Multiple dose
	80 mg at birth and 100 mg at week 12 (9)	5 mg/kg q 12 weeks × 3 (5)
Doses received (N = 1 | 2| 3)	13 | 8 | 0	17 | 8 | 8
PK sampling	Hour: 0, 24, 72	Hour: 0, 1, 3, 6, 24, 48, 72
	Week: 1, 2, 4, 8, 24, 36, 48, 60, 72, 84	Week: 1–4, 8, 12, 16, 20, 24
Total PK samples	211	480
Median gestational age (range)	38.6 weeks (37–42)	NA
Median post-natal age (range)	2.5 days (1–5)	29.4 years (22–48)
Median weight (kg; range)	2.8 (2.2–4.3)	71.1 (45–97)
Median dose (mg/kg, range)	28.6 (19.2–40.0)	5 (1–40)
Sex (N = male | female)	12 | 9	11 | 14

I.V., intravenous; S.C., subcutaneous; Q12W, every 12 weeks

^aInfants were administered a second dose of 100 mg at 12 weeks if still breastfeeding.

In the IMPAACT P1112 study, Cohort 1 included formula-fed infants receiving a single 80 mg dose of s.c. VRC07-523LS within 3 days of birth. Cohort 2 included breast-fed infants receiving an initial 80 mg dose within 5 days of birth and a second dose of 100 mg at Week 12 if complete cessation of breastfeeding was not achieved. PK sampling for Cohort 1 included up to 12 samples between the completion of infusion and 96 weeks post-dose. Cohort 2 had eight PK samples drawn between the completion of the first infusion and 12 weeks, followed by an additional five PK samples at weeks 14 to 96, following the Week 12 dose (Table 1).

Pharmacokinetic analysis

Quantification of VRC07-523LS concentrations in participant serum involved using an anti-idiotype antibody capture ELISA method described elsewhere.^6,11

PK data were analysed using the computer program NONMEM (version 7.5) with a GNU Fortran G77 compiler and first-order conditional estimation with interaction. A two-compartment PK structural model (ADVAN4, TRANS4 subroutine) with zero-order input was used to describe the data. The model had the following parameters: clearance (CL), intercompartmental clearance (Q), steady state volume of distribution (V_ss), zero-order absorption duration (D) and bioavailability (F).

The popPK analysis was performed in two phases. The route, range of dosing and frequency of sampling were more frequent in the adult study; thus, an initial analysis was performed on the adult data alone followed by a combined analysis in infants and adults. Since there was no i.v. administration in the infants, the denser PK information from adults helped establish estimates for VRC07-523LS absorption and distribution parameters that could be extrapolated to infants. PK parameters were scaled allometrically before evaluation of other potential covariates, CL and Q by weight (WT^0.85) and V_ss by weight (WT^1.0).¹⁴ Weight was updated at each PK study visit for both the modelling and simulation. Age (adult versus infant), dose (mg/kg as a linear function), dose number (initial versus subsequent dose), and sex were evaluated as potential covariates for CL, V_ss, D and F. Between-subject variability (BSV) was estimated for CL, V_ss and D. A combined proportional and additive residual error model. The first PK sample within each PK profile with concentrations less than the quantitative limit (QL) were set to half the QL (<1 mcg/mL) for modelling. Subsequent PK samples (after the first) less than the QL were excluded from PK analysis.

Potential covariates were added to the model one at a time, as a linear or categorical function, with covariates that improved the model fitting at a statistically significant level (P < ∼0.05; change in objective function of 3.84 for loss of one degree of freedom for each covariate added) being selected for assessment in the subsequent multivariate analysis. During the multivariate assessment, using a forward-addition approach, covariates found to improve the model fitting by a reduction in the objective function of at least 7.88 (P < ∼0.005) were retained in the final model.

Empirical Bayesian estimates of the individual PK parameters were generated from the model using the POSTHOC routine. A 1000-sample bootstrap assessment of parameter precision for the model was performed using Wings for NONMEM.¹⁵

Monte Carlo simulations of 1990 virtual infants were conducted to assess the current proposed infant dose (80 mg s.c. within the first week of life, followed by a 100 mg s.c. dose at Week 12). The simulations used median infant body weights incorporating changes in weight with post-natal age.¹⁶ Each virtual infant was randomly assigned an IC₈₀ value derived from a NIH VRC HIV-1 viral panel of viruses (199 HIV-1 viral isolates; n = 10 infants per unique IC₈₀ value), where the median IC₈₀ was 0.238 µg/mL (interquartile range 0.103–0.545). Five percent of viral isolates were considered completely resistant to VRC07-523LS (≥50 µg/mL) and were excluded from simulations.

The frequency of simulated concentrations above the pre-defined target of 10 µg/mL was determined. The PT₈₀ value was calculated as the ratio of serum bNAb concentration and the assigned IC₈₀ value of the reference virus across the bNAb concentration versus time profile.

Results

Participants

A total of 638 VRC07-523LS serum concentrations collected over a span of 84 weeks were used in the two-compartment popPK model. PK sampling was obtained from 21 infants exposed to HIV and 25 healthy adults. Table 1 summarizes participant characteristics at the first PK visit and study designs. Dose-normalized concentration profiles of VRC07-523LS after the first s.c. dose for infants and i.v. dose for adults, respectively, are shown in Figure 1.

Figure 1.

Observed infant s.c. VRC07-523LS concentrations dose-normalized to 25 mg/kg after first administration in comparison to observed adult i.v. VRC07-523LS concentration from 20 mg/kg doses. Dashed (triangles) and solid (circles) lines represent median concentrations for infants and adults, respectively. Data suggest infants absorb s.c. VRC07-523LS in a similar manner to i.v. absorption in adults.

Population pharmacokinetic analysis

A two-compartment PK model with zero-order input described the infant and adult data well without significant bias (Figures S1A–C and S2A–C, available as Supplementary data at JAC Online). The univariate screen found population (infant versus adult) as a potential predictor of CL, V_ss and D. Multiple dosing was also a potential predictor of V_ss and F, while dose was a potential predictor of CL and V_ss. Repeat dosing on V_ss, in addition to age (infant versus. adult) on CL and D, were significant covariates retained in the final model (Table 2). Although the effect of dose (if i.v.; mg/kg as a linear function) was seen in the adult model and trended in the combined population model, it was found to not significantly impact our model, probably since infants received a single dose amount and a different s.c. dose than adults. BSV was assessed for CL, V_ss and D. A combined, additive and proportional error was used to characterize residual error.

Table 2.

Final combined population pharmacokinetic model parameters and bootstrap estimates

	Final parameter estimates	Relative standard error	Bootstrap estimates median (95% CI)
Θ1 (Vc; L/70)	1.48	0.57	1.47 (1.23–1.71)
Θ2 (Vp; L/70)	2.28	0.68	2.21 (1.68–2.78)
Θ3 (CL; mL/d/70 kg)	47.76	17.40	47.76 (36.72–58.32)
Θ4 (Q; L/h/70 kg)	0.0243	0.0040	0.0251 (0.0172–0.0354)
Θ6 (F)	0.30	0.092	0.30 (0.22–0.38)
Θ7 (D1; hours)	36	1.9	35.7 (30.4–40.1)
Θ8 (repeat dose factor Vc  +  Vp)	1.49	0.10	1.48 (1.25–1.65)
Θ9 (adult factor CL)	1.69	0.21	1.69 (1.48–1.93)
Θ10 (adult factor D1)	2.79	0.24	2.72 (1.71–3.44)
BSV (η)
IIV on Vc + Vp (%)	35.7%	20.4%	28.9% (20.6–37.1%)
IIV on CL (%)	29.3%	20.6%	9.4% (0.5–27.1%)
IIV on D1 (%)	10.0%	37.2%	84.8% (61.3–95.4%)
Residual error (ε)
Proportional (%)	22.6%	5.5%	22.0% (19.5%–24.6%)
Additive (μg/mL)	0.31	1.5	0.31 (0.005–1.63)

V _ss/F (L) = 12.53 × (WTKG/70)^1.0 × 1.49 (if repeat dose).

CL/F (mL/d) = 158.4 × (WTKG/70)^0.85 × 1.69 (if adult) D1 (hours) = 36  ×  2.79 (if adult).

IIV, interindividual variability; KA, first-order absorption; V_c, central volume of distribution; V_p, peripheral volume of distribution; V_ss/F, apparent steady state volume of distribution; WTKG, weight (kg).

The visual predictive check demonstrated that the model was able to adequately simulate infant exposure (Figure S3). Shrinkage estimates for between- subject variability were low for V_ss (2.46%) and CL (2.75%), but higher for D (61.9%). Final model parameters and variance estimates are shown in Table 2. Bootstrap evaluation of the final model successfully converged 89% of the time. The parameter estimates fell within the 95% confidence interval and deviated minimally from the median estimates, suggesting that the model represents the populations well. Between-subject variabilities for V_ss, CL, and D were 35.7%, 29.3% and 10.0%, respectively. Weight-normalized CL in adults was 69% higher than in infants. In addition, D was 179% higher in adults than in infants. With respect to the effect of repeat dosing on V_ss, V_ss increased by 49% with the second and subsequent doses in comparison to the first dose. This increase was observed both in adult-only and combined analysis (Table S1).

Monte Carlo simulations and pharmacodynamic analysis

Monte Carlo simulations of 1990 virtual infants exposed to HIV were conducted to assess the P1112 dose regimen (80 mg at birth followed by 100 mg at 12 weeks). These simulations indicate that >87% of infants would maintain trough concentrations >10 μg/mL at 12 weeks, and >98% at 24 weeks (Figure 2 and Figure S4).

Figure 2.

Monte Carlo simulations of VRC07-523LS pharmacokinetic in HIV-exposed infants using the final model and 80 mg s.c followed by 100 mg s.c. at Week 12. Infant weights are based on median infant weight for postnatal growth preterm infants from Villar et al. Simulated VRC07-523LS concentration versus time profile assessing maintenance of target concentrations >10 μg/mL based on the P1112 protocol dosing. Shaded area represents the 90% (5–95) prediction interval. Solid black curve represents the median simulated concentrations. The horizontal solid and dotted lines represent target maintenance VRC07-523LS concentrations.

The expected PT₈₀ based on the infant popPK model and expected distribution of viral sensitivity (Figure S5) were also evaluated. By combining the simulated IC₈₀s and time-varying concentrations, PT₈₀ values were generated throughout the concentration–time profile and are presented in the simulation output (Figure 3). Simulations provided median (90% PI) PT₈₀ values of 163 (71–383), 116 (51–274) and 61 (26–144) at 6, 8 and 12 weeks, respectively (Table S2). Detailed VRC07-523LS concentrations, PT₈₀ over the concentration–time profile, and predicted prevention efficacy from weeks 4 to 24 are shown in Table S2. Most (>88%) infants achieved and sustained a PT₈₀ of >32 for 12 weeks, however, a PT₈₀ > 200 was only sustained for about 4 weeks (Figure 3). A PT₈₀ > 200 was correlated with 90% of maximal antiviral efficacy from the AMP study (Figure S6).

Figure 3.

Calculated PT₈₀ (predicted serum neutralization 80% inhibitory dilution titre), defined as an individual’s VRC07-523LS serum concentration at each time point divided by the IC₈₀ (in vitro 80% inhibitory dilution) from a distribution of 199 isolates from Vaccine Research Center internal data. There were originally 210 isolates from Clades B/C, however ∼5% (11) of isolates that were fully resistant to VRC07-523LS (IC₈₀ > 50 ug/mL) were excluded from this analysis. PT₈₀ is the proposed bNAb titre biomarker integrating these two measurements, which was found to be a correlate of HIV-1 prevention efficacy with data from the AMP trials (Gilbert et al.). It is suggested that a sustained PT₈₀ of >200 over the 12-week dosing period is necessary to achieve 90% prevention efficacy against acquisition of circulating HIV-1 viruses. The dashed horizontal line represents a PT₈₀ of 200 (90% prevention efficacy) whereas the solid horizontal line represents a PT₈₀ of 32 (50% prevention efficacy). Shaded area represents the 90% (5–95) prediction interval. Solid black curve represents the median simulated PT₈₀.

Further simulations were conducted to assess alternative VRC07-523LS dosing, with a larger initial dose and repeat dosing at 6 and 8 weeks in an attempt to sustain VRC07-523LS PT₈₀ ratios above 200 throughout the dosing interval (Figures S7 and S8, Tables S2 and S3). For the first scenario, 80 mg within the first week after birth and 100 mg at Week 6, the median VRC07-523LS concentration level and PT₈₀ just before the second dose (trough level; expected to have least bNAb protection) were 39.74 μg/mL and 164, respectively. For the second scenario, 80 mg within the first week after birth and 100 mg at Week 8, median concentration level and PT₈₀ were 27.81 μg/mL and 117, respectively. For the third scenario, 100 mg within the first week after birth and 100 mg at Week 6, median concentration level and PT₈₀ were 49.67 μg/mL and 205, respectively. For the fourth scenario, 100 mg within the first week after birth and 100 mg at Week 8, median concentration level and PT₈₀ were 34.89 μg/mL and 146, respectively. For the last scenario, 100 mg within the first week after birth and 100 mg at Week 12, median concentration level and PT₈₀ were 18.34 μg/mL and 76, respectively. Of the five additional analyses conducted, dosing of 100 mg within the first week after birth and 100 mg at Week 6 which achieved and sustained a PT₈₀ > 200 throughout the dosing interval. Monte Carlo simulations provided median (90% PI) PT₈₀ values of 295 (127–689), 205 (89–480), 306 (131–710) and 213 (91–495) at 4, 6, 8 and 12 weeks, respectively (Table S3).

Discussion

Antibodies from the VRC01 class, which target the HIV-1 envelope gp120 CD4-binding site, are advancing in clinical development for adults and infants. However, little is known about their pharmacokinetics and optimal dosing in infants.¹¹ This is the first VRC07-523LS popPK analysis in infants and represents one of the few popPK models combining adult and infant monoclonal antibody PK data in a single analysis. Leveraging sparse infant PK with intensive adult PK data improved infant simulations. The adult popPK of VRC07-523LS has been described by Chawana et al. in an analysis of 124 participants across 3 HVTN/HPTN studies.¹⁷ Their estimated CL/F was nearly identical to our estimated value in adults (255 mL/d versus 269 mL/d). Our model also generated values similar to previous non-compartmental analyses from infants for VRC07-523LS CL/F, V_ss/F and t_1/2.¹¹ The estimated median adult CL and t_1/2 of VRC07-523LS values were within 15% of those from previous non-compartmental analyses.⁶

We assessed factors influencing VRC07-523LS PK. In our final popPK model, CL (allometrically scaled) was 69% higher in adults than in infants and absorption duration in adults was 2.79 times longer than in infants. Repeated dosing also led to a 49% increase in V_ss compared with the initial dose. This phenomena could involve target-mediated disposition or changes in FcRn binding but requires additional study for mechanistic understanding. The rapid infant s.c. absorption indicates high VRC07-523LS concentrations can be rapidly achieved with s.c. administration shortly after birth, which is important for prevention of vertical transmission since HIV-1 exposure is greatest contemporaneously with the initial antibody dose.

Furthermore, our model suggests VRC07-523LS metabolism is slower in infants than adults. While higher FcRn expression in infants could help protect VRC07-523LS from metabolism¹⁸ this difference is countered by the rapid infant growth during the first few months of life. There was nearly a doubling in infant weight for birth to 12 weeks of age, which resulted in a significant dilutional effect on observed bNAb concentrations. Thus, the observed infant t_1/2 represents both growth and metabolism. While growth needs to be considered in characterizing infant VRC07-523LS PK, it is difficult to include in non-compartmental analyses. This population PK analysis incorporated weight as a dynamic variable, updating the values at each infant visit to account for growth. Thus, our model can be used to generate precise, unbiased estimates of CL/F and V_ss/F at any time during the first months of life.

Other bNAbs, including VRC01 and VRC01LS, have demonstrated safety with favourable PK profiles in infants but with some PK differences compared with adults.^16,19,20 Similar to PK in adults, VRC07-523LS has lower CL/F and a longer t_1/2 than VRC01. The estimated VRC07-523LS infant CL was 45.6 mL/d/70 kg compared with 63.9 mL/d/70 kg for VRC01, whereas the infant VRC07-523LS CL was greater and t_1/2 shorter than for VRC01LS, estimated to be 35.5 mL/d/70 kg and 39 days, respectively. While differences in CL/F between the adults and infants were seen in VRC01, VRC01LS and VRC07-523LS, the best description of these differences has varied. Without infant i.v. administration, age differences in bNAb metabolism have been confounded by potential age differences in bioavailability. Age group was a potential covariate for CL and F in the univariate screen but its impact on CL was greater and age group was not found to be an independent factor on F during the multivariate assessment. The lower adult s.c. doses (5 mg/kg in adults versus ∼25 mg/kg in infants) as well as the rapid infant growth add to the complexity of determining the separate impact of age on CL versus F.

Absorption in our popPK models was rapid, with concentrations exceeding the target levels by Day 1. This is consistent with rapid absorption seen in VRC01 and VRC01LS.^21,22 The absorption was faster in infants than adults, resulting in high bNAb concentrations by 24 hours in all infants. Despite limited early PK sampling, which prevented precise characterization of infant C_max, infant concentrations at 24 hours were higher than the next sample at Week 2, whereas adult C_max did not occur until Day 3.

Although simulations of the P1112 VRC07-523LS regimen predicted that large numbers of virtual infants exposed to HIV rapidly achieve and maintain target serum concentration levels of bNAbs (above 10 µg/mL) for >24 weeks with two doses, this may not translate into optimal efficacy. In addition to using protocol-specified target bNAb concentrations, we also evaluated expected PT₈₀ based on the infant popPK model and expected distribution of viral sensitivity. The PT₈₀ following the dosing used in P1112 did not translate to maintenance levels >200 (the value correlated with 90% of maximal antiviral efficacy in the AMP study) at 12 weeks. The simulation of P1112 dosing resulted in a Week 12 median PT₈₀ of 61 (90% PI 26–144), suggesting that some effect would be seen for 12 weeks in nearly all infants,¹² but would be less than optimal.

Further Monte Carlo simulations were conducted to assess alternative VRC07-523LS dosing for the prevention of HIV-1 transmission to infants and in an attempt to sustain VRC07-523LS PT₈₀ ratios >200 throughout the dosing interval, predicted achievable trough concentrations above the target for 100% of virtual infants exposed to HIV, but also that serum concentrations would be maintained at high levels throughout the first 6 and 8 weeks of life. While most of these virtual infants maintained the target bNAb levels and PT₈₀s with all dosing strategies, the PT₈₀ calculations provide additional insight into how efficacy may differ dramatically based on the distribution of IC₈₀. Thus, while most infants achieved some predicted prophylactic activity with the IMPAACT P1112 dosing regimen, for a higher maximal antiviral effi9cacy threshold (i.e. 90%), when more resistant viruses are expected, then more frequent dosing or combination with other bNAbs or ART may be necessary to overcome less sensitive viruses and to prevent HIV-1 transmission.

Our study had several limitations. First, the VRC07-523LS PK sampling schedule in the infant study did not allow detailed characterization of the s.c. absorption rate but did indicate faster absorption in infants. Importantly, observed and modelled concentrations consistently achieved target levels at 1 day after administration in the infants. Second, because infants only received VRC07-523LS via s.c. administration, the bioavailability was assumed to be similar between infants and adults but differences may exist due to administration technique and microvascular architecture. The potential age differences in CL and V_ss were confounded by differences in F. In general, the covariates and parameter estimates in the popPK models for the adult-only and combined infant and adult data were similar. The one exception was F, which was lower in the combined data. This suggests the age differences in CL may be confounded with some undescribed age differences in F. While the typical parameters appeared to be well estimated, the small numbers limited the precision of the BSV of CL and D1. Last, while PT₈₀ has been found to correlate with bNAb preexposure prophylaxis efficacy with VRC01 against horizontal HIV-1 transmission, extrapolating these findings to VRC07-523LS and vertical transmission is not straightforward. HIV-1 acquisition is exclusively via the oral route in infants while multiple routes are involved in adult transmission. VRC07-523LS and VRC01 are different bNAbs with different variable and fixed regions, which may affect their distribution PK and in vitro versus in vivo PD relationship.

In addition, the target PT₈₀ levels that correlated with viral effect proposed by Gilbert et al. were generated by midpoint concentrations at steady state,¹² and in adults. Infants in the current study only received one or two doses, with little opportunity for accumulation. Using a conservative approach, we focused presenting the entire time profile with emphasis on the point of least bNAb protection, just prior to a second dose of 12 weeks after the first dose. While the overall shape of pharmacodynamic relationship is probably similar throughout the dosing interval, estimated PT_80wk12 values would be lower than PT_80mid from the AMP study analysis. Additional studies are needed to determine the appropriate target VRC07-523LS PT₈₀ values to prevent vertical transmission. If the antiviral pharmacodynamic effects of VRC07-523LS are similar to VRC01 from the AMP trials, then the expected prevention efficacy that we expect in an infant population can be simulated from the popPK model and expected viral sensitivity (Figures S7 and S8).

In conclusion, we developed a popPK model for VRC07-523LS using infant and adult data to better understand the factors affecting VRC07-523LS disposition, predict infant exposures for specific dosing regimens that could be administered during breastfeeding, and to help guide future trial dosing regimens. Simulations of infant exposure linked to the distribution of HIV neutralization sensitivity were used to predict the prevention efficacy of VRC07-523LS, allowing for a rational and mechanistic approach to optimize dosing. The assessed dose (80 mg s.c. followed by 100 mg at 12 weeks) rapidly achieves and maintains concentrations >10 µg/mL for more than 24 weeks. This is expected to result in at least modest (∼50%) prevention efficacy in all infants, but high PT_80s and predicted prevention efficacy (∼90%) is only maintained for about 4 weeks post-dose. This suggests that a shorter dosing interval and/or combination with additional bNAb or ART may be necessary for maximal prevention of HIV acquisition and to overcome resistant viruses. Further evaluation of VRC07-523LS in breastfeeding infants is needed.

Funding

Overall support for the International Maternal Pediatric Adolescent AIDS Clinical Trials Network (IMPAACT) was provided by the National Institute of Allergy and Infectious Diseases (NIAID) 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), all components of the National Institutes of Health (NIH), under Award Numbers UM1AI068632 (IMPAACT LOC), UM1AI068616 (IMPAACT SDMC) and UM1AI106716 (IMPAACT LC), and by NICHD contract number HSN275201800001I. Funding for this study was also provided by the Intramural Research Program of the VRC at NIAID. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Transparency declaration

None to declare.

Supplementary data

Figures S1 to S8 and Tables S1 to S3 are available as Supplementary data at JAC Online.