Leveraging Physiologically-Based Pharmacokinetic (PBPK) Modeling to Optimize Dosing for Lopinavir/Ritonavir with Rifampin in Pediatric Patients

Abstract

Study Objective:

Treatment of HIV and tuberculosis co-infection leads to significant mortality in pediatric patients, and treatment can be challenging due to the clinically significant drug-drug interaction (DDI) between lopinavir/ritonavir (LPV/RTV) and rifampin. Doubling LPV/RTV results in insufficient lopinavir trough concentrations in pediatric patients. The objective of this study is to leverage physiologically-based pharmacokinetic (PBPK) modeling to optimize the adjusted doses of LPV/RTV in children receiving the WHO-revised doses of rifampin (15 mg/kg daily).

Design:

Adult and pediatric PBPK models for LPV/RTV with rifampin were developed, including CYP3A and P-glycoprotein inhibition and induction.

Setting (or Data Source):

Data for LPV/RTV model development and evaluation was available from the pediatric AIDS Clinical Trials Group.

Patients:

Dosing simulations were next performed to optimize dosing in children (2 months to 8 years of age).

Intervention:

Exposure following super-boosted LPV/RTV with 10 mg/kg and 15 mg/kg PO daily rifampin was simulated.

Measurements and Main Results:

Simulated parameters were within two-fold of observations for LPV, RTV, and rifampin in adults and children ≥ 2 weeks old. The model predicted that, in healthy adults receiving 400/100 mg oral LPV/RTV twice daily (BID), co-treatment with 600 mg oral rifampin daily decreased the steady-state area under the concentration vs. time curve of LPV by 79%, in line with the observed change of 75%. Simulated and observed concentration profiles were comparable for LPV/RTV (230/57.5 mg/m²) PO BID without rifampin and 230/230 mg/m² LPV/RTV PO BID with 10 mg/kg PO daily rifampin in pediatric patients. 16 mg/kg of super-boosted LPV (LPV/RTV 1:1) PO BID with 15 mg/kg PO daily rifampin achieved simulated LPV troughs >1 mg/L in ≥93% of virtual children weighing 3.0-24.9 kg, which was comparable with 10 mg/kg PO daily rifampin.

Conclusions:

Super-boosted LPV/RTV with 15 mg/kg rifampin achieves therapeutic LPV troughs in HIV/TB infected simulated children.

Introduction

Human immunodeficiency virus (HIV) and tuberculosis (TB) co-infection is a serious problem worldwide and is particularly problematic in sub-Saharan Africa. HIV prevalence among children with TB ranges from 10 to 60% in countries with moderate to high TB prevalence. TB is the leading cause of death among HIV-infected children resulting in 52,000 TB-related deaths globally among HIV-positive children in 2016.¹ Lopinavir (LPV) is a protease inhibitor that is co-administered with low-dose ritonavir (RTV) in the fixed-dose combination (Kaletra®) for enhanced bioavailability and prolonged half-life, through P-glycoprotein and cytochrome P450 (CYP) 3A inhibition. LPV/RTV in combination with two nucleoside reverse transcriptase inhibitors (NRTIs) is the first-line regimen of choice in treatment-naïve children from 2 weeks to <3 years of age, and it is also commonly used as second-line therapy in children failing a non-nucleoside reverse transcriptase inhibitor (NNRTI) based initial regimen.² Rifampin is a key medication used for TB treatment in both adults and children. LPV/RTV with rifampin is often unavoidable for the management of HIV/TB treatment in pediatric patients despite a clinically significant interaction.

LPV is primarily metabolized by CYP3A with 2.2% and 19.8% of drug excreted unchanged in urine and feces, respectively.^3,4 LPV is highly protein bound with a concentration dependent decrease in binding ranging from 97.4% at 100 μg/mL LPV to 99.7% at 0.1 μg/mL LPV in human plasma.⁴ Although LPV binds to both human serum albumin and α1-acid-glycoprotein (AAG), it has a higher affinity for AAG.^4,5 RTV is predominantly metabolized by CYP3A with a minor pathway through CYP2D6.⁶ RTV is cleared primarily through hepatobiliary elimination with 3.5% and 33.8% of drug excreted unchanged in urine and feces, respectively.⁷ The protein binding of RTV in human serum ranged from 99.3 to 99.5% at RTV concentrations ranging from 0.01 to 30 μg/mL, and both human serum albumin and AAG contribute to protein binding.^7,8 LPV and RTV are substrates, inhibitors, and inducers of P-glycoprotein.^9–12 Additionally, LPV and RTV are time-dependent inhibitors of CYP3A but RTV is a more potent time-dependent inhibitor.¹³ RTV is also a mixed competitive CYP3A inhibitor, and RTV can induce CYP3A through the pregnane X receptor (PXR).^12,14

Rifampin can weakly inhibit but is primarily a strong inducer of CYP3A and P-glycoprotein through upregulation of PXR.^15–18 Co-administration of rifampin 600 mg oral (PO) daily with LPV/RTV [400/100 mg PO twice daily (BID) for 1 week] in HIV-infected adults resulted in a significant decrease in the median (interquartile range) LPV trough concentrations from 0.13 (0.10–0.18) mg/L without rifampin to 0.03 (0.01–0.05) mg/L a week after commending rifampin.¹⁹ In healthy and HIV-infected adults, this interaction can be overcome by super-boosting the RTV dose (LPV/RTV 400/400 mg PO BID) or doubling the dose (LPV/RTV 800/200 mg PO BID).^19–21 However in young children, doubling the dose of LPV/RTV (460/115 mg/m² by body-surface area PO BID) in combination with rifampin 10 mg/kg PO once daily resulted in 60% of children [median (interquartile range) age of 1.25 (0.98–1.93) years] achieving sub-therapeutic LPV trough concentrations (<1 mg/L) compared to 8% of children [median (interquartile range) age of 1.59 (1.15–2.23) years] receiving LPV/RTV (230/57.5 mg/m² PO BID) without rifampin.²² Super-boosting LPV/RTV (230/230 mg/m² PO BID) was able to overcome the induction effects of rifampin (10 mg/kg PO daily) in children [median (interquartile range) age of 16 (14–24) months], where 87% of these children achieved therapeutic LPV trough values (>1 mg/L).²³ Furthermore, a study in 96 HIV-infected children with TB with a median (interquartile range) age of 18.2 (9.6, 26.8) months reported a non-inferior difference in the morning minimal concentration (C_min) of LPV for children receiving super-boosted LPV/RTV with median (interquartile range) 12.4 (11.2, 13.4) mg/kg of rifampin compared to children receiving standard LPV/RTV without rifampin.²⁴

However, the drug-drug interaction (DDI) between LPV/RTV with the revised World-Health Organization (WHO) rifampin dosing of approximately 15 mg/kg using the fixed-dose combination (75 mg rifampin with 50 mg isoniazid and 150 mg pyrazinamide) has not been evaluated.²⁵ Optimal dosing of LPV/RTV with rifampin in pediatric patients requires further research since this interaction can result in therapeutic failure and antiretroviral drug resistance. Therefore, we developed a PBPK model for LPV/RTV with rifampin in order to optimize super-boosted LPV/RTV dosing in HIV/TB-infected pediatric patients (between 3.0 and 24.9 kg body weight) concurrently receiving 15 mg/kg PO daily rifampin.

Methods

Adult RTV PBPK Model Development

A 29-year-old European male with expressions of CYP3A4, CYP3A5, and P-glycoprotein using the Array Database was used for LPV/RTV model development and evaluation using PK-Sim®/MoBi® version 9.0 (www.open-systems-pharmacology.org). We first developed and evaluated an adult PBPK model for RTV administered as a boosting agent with elvitegravir for 10 days in healthy adults, and RTV administered from 200–500 mg PO every 12 hours for 2 weeks in HIV-infected males (Supplementary Table 1).^26,27 The model included CYP3A4, CYP3A5, and CYP2D6 metabolism, P-glycoprotein transport, renal clearance via glomerular filtration, time-dependent inhibition and competitive inhibition and induction for CYP3A4 and CYP3A5, as well as competitive inhibition for P-glycoprotein (Table 1). The aqueous solubility, transcellular intestinal permeability, and P-glycoprotein maximal velocity (V_max) were optimized using parameter identification with the Monte Carlo algorithm. To best characterize the maximum concentration (C_max) and area under the plasma concentration-time curve (AUC) data for doses ranging from 20 mg to 500 mg, one transcellular intestinal permeability value was optimized for doses <200 mg and another value was optimized for doses ≥ 200 mg (Table 1). Cellular permeability was calculated using the default PK-Sim® algorithm, and the tissue-to-plasma partition coefficients were calculated using the Rodgers & Rowland method.^28,29 The LPV and RTV PO formulation absorption characteristics were optimized using the Weibull formulation with a 50% dissolution time of 30 minutes, a lag-time of 30 minutes, and a dissolution shape of 0.92. Evaluation of RTV CYP3A and P-glycoprotein mediated DDI potential is described in the Supplementary Methods.

Table 1.

Final physiologically-based pharmacokinetic (PBPK) model parameters for lopinavir, ritonavir, and rifampin.

Parameter	Lopinavir	Ritonavir	Rifampin
Physiochemical Properties
Lipophilicity	3.32a	3.83a	2.91a
Molecular Weight	629 g/mol [45]	721 g/mol [45]	822.94 g/mol [45]
Fraction Unbound	0.01 [45]	0.01 [45]	0.11 [45]
Protein Binding	AAG	AAG	Albumin
Solubility	0.00192 mg/mL [45]	100 μg/mLa	2.8 mg/mL [36]
pKa (Acid or Base)	Neutral [45]	2.84 (Base) [45]	7.90 (Base)/1.70 (Acid) [36]
Transcellular Intestinal Permeability	9.91·10−4 cm/mina	2.33 * 10−5 cm/min (<200 mg)a 8.91 * 10−5 cm/min (≥ 200 mg)a	3.00 *10−5 cm/min (adults)a 7.50 *10−6 cm/min (pediatric)a
Metabolism and Transport
CYP3A4 KM	6.8 μM [3]	0.068 μM [6]	-
CYP3A4 Vmax	9.4 nmol/mg/min [3]	1.37 pmol/min/pmol [6]	-
CYP3A5 KM	-	0.047 μM [6]	-
CYP3A5 Vmax	-	1.00 pmol/min/pmol [6]	-
CYP2D6 KM	-	1.0 μM [6]	-
CYP2D6 Vmax	-	0.93 pmol/min/pmol [6]	-
P-glycoprotein KM	0.13 μM [9]	0.13 μM [9]	-
P-glycoprotein Vmax	1.28 pmol/min/pmola	1.68 pmol/min/pmola	-
P-glycoprotein Kcat	-	-	0.61 1/min [36]
AADAC KM	-	-	162.9 μM [46]
AADAC Vmax	-	-	47.4 pmol/min/pmola
GFR fraction b	1.00	1.00	1.80a
OATP1B1 Vmax	-	-	8.37*10−4 pmol/min/pmol [47]
OATP1B1 KM	-	-	1.50 μM [47]
Interactions
CYP3A4 Kinact,half	0.41 μM [13]	0.10 μM [13]	-
CYP3A4 Kinact	0.10 1/min [13]	0.32 1/min [13]	-
CYP3A4 KI	-	0.03 μM [14]	18.5 μM [18]
CYP3A4 EC50	-	1.0 μM [48]	0.34 μM [15]
CYP3A4 Emax	-	68.5 [48]	9.0 [15]
CYP3A5 EC50	-	1.0 μM [48]	-
CYP3A5 Emax	-	68.5 [48]	-
CYP3A5 Kinact,half	1.00 μM [13]	0.12 μM [13]	-
CYP3A5 Kinact	0.05 1/min [13]	0.08 1/min [13]	-
CYP3A5 KI	-	0.03 μM [14]	-
P-glycoprotein KI	-	0.2 μM [49]	169 μM [17]
P-glycoprotein EC50	-	-	0.34 μM [15,36]
P-glycoprotein Emax	-	-	2.5 [16]
OATP1B1 EC50	-	-	10 μM [50]
OATP1B1 Emax	-	-	5.0 [50]
AADAC EC50	-	-	0.9 μM [39]
AADAC Emax	-	-	5.0a

^aOptimized value.

^bAdult renal clearance via the GFR fraction was predicted using a GFR fraction to account for tubular reabsorption or secretion as the empiric renal clearance (literature renal clearance value in adults) divided by the expected renal clearance as a result of GFR if there was no tubular reabsorption (fraction unbound × normal GFR).

Abbreviations: pKa: negative base-10 logarithm of the acid dissociation constant; CYP3A4: cytochrome P450 3A4; CYP3A5: cytochrome P450 3A5; CYP2D6: cytochrome P450 2D6; K_M: concentration of half-maximum metabolism or transport; V_max: maximum rate of metabolism or transport; AADAC: human arylacetamide deacetylase; OATP1B1: organic-anion-transporting polypeptide (OATP) 1B1; EC₅₀: concentration of the drug that gives half-maximum induction; E_max: maximum induction; K_inact,half: concentration of half-maximum inactivation; K_inact: maximum rate of inactivation; K_I: concentration of half-maximum inhibition; AAG: α1-acid glycoprotein; glomerular filtration rate.

Adult LPV PBPK Model Development

The LPV PBPK model was developed and evaluated using LPV pharmacokinetic (PK) data administered concurrently with RTV in heathy adults (Supplementary Table 1).^30–32 The model included Michaelis-Menten CYP3A4 metabolism and P-glycoprotein transport, renal clearance via glomerular filtration, and CYP3A4 and CYP3A5 time-dependent inhibition (Table 1). We assumed that LPV inhibition of P-glycoprotein was minimum relative to RTV. Transcellular intestinal permeability and P-glycoprotein V_max were optimized using parameter identification with the Monte Carlo algorithm. Cellular permeability was calculated using the default PK-Sim® algorithm, and the tissue-to-plasma partition coefficients were calculated using the Rodgers & Rowland method.^28,29

Adult LPV/RTV PBPK Model Evaluation

Model evaluation for literature data was determined by comparing the ratio of simulated and observed PK parameters (C_max, AUC) for LPV and RTV (Supplementary Table 1). Additionally, we used a model evaluation dataset consisting of LPV/RTV (400/100 mg) (Kaletra®) administered PO BID to 12 healthy adults (84 plasma samples) and 12 HIV-infected adults (94 plasma samples).^33–35 Population simulations were performed based on 100 virtual white American adults (50% female) from 25 to 47 years of age. The simulated AUC within a dosing interval at steady-state (AUC_0-τ,ss) (corresponding to 0 to 12 hours) as well as the maximum concentration at steady-state (C_max,ss) for LPV and RTV were compared to observed parameters in healthy adults as well as HIV-infected subjects. Additionally, the average fold error (AFE) was calculated comparing the simulated geometric mean with observed concentration using Equation 1. A two-fold error (0.5 to 2.0) for AFE and the simulated to observed PK parameters was considered acceptable for all adult and pediatric model evaluations.

$$\text{AFE}={10}^{\left(1/n\right)\left(\sum \text{log}\left(\text{simulated}/\text{observed}\right)\right)}$$ (1)

DDI Evaluation for Adults receiving LPV/RTV with Rifampin

The adult rifampin model was based upon a slightly revised published PBPK model, and then evaluated using PK data in adults with TB (Supplementary Methods).³⁶ Population simulations were performed in 100 virtual white American adults (60% female) from 22 to 70 years of age receiving standard dosing (400/100 mg LPV/RTV PO BID), double dosing (800/200 mg LPV/RTV PO BID), or super-boosting dosing (400/400 mg LPV/RTV PO BID) with 600 mg PO daily rifampin. Simulated geometric mean fold ratios (90% confidence intervals) for LPV C_max,ss, AUC_0-τ,ss, and the minimum steady-state concentration (C_min,ss) were compared with observed data in healthy and HIV-infected subjects (Supplementary Methods).^5,19,21

LPV/RTV Pediatric PBPK Model Development and Evaluation

Pediatric data for LPV/RTV model development and evaluation was available from the pediatric AIDS Clinical Trials Group (Supplementary Materials; Supplementary Table 1).³⁷ There were 160 pediatric HIV-1 infected subjects (1203 LPV samples) from 0.115 to <18 years of age available for model evaluation. A virtual 30 year-old white American was scaled to pediatric virtual subjects between 2 weeks and 18 years of age based on pre-established age-dependent algorithms in PK-Sim® to generate anatomical and physiological parameters (e.g., body weight, height, organ weights), as well as to account for the maturation of CYP3A4, P-glycoprotein, and AAG (Supplementary Methods). Based on an adult bioequivalence study using the soft-gel capsules as reference, liquid Kaletra® was adjusted to be a fraction of the administered dose, which was 0.8 for LPV and 0.7 for RTV in the fasting-state and 0.9 for both LPV/RTV in the fed-state (Supplementary Methods).³

Population simulations were performed with 100 virtual pediatric subjects simulated in each age category (2 to <6 months, 0.5 to <2 years, 2 to < 6 years, 6 to <12 years, and 12 to <18 years of age) and then compared with the observed data for LPV and RTV. Since infants do not typically begin eating solid foods until at least 6 months of age, the liquid formulation of Kaletra® in the fasted-state was used for the < 2 years of age population simulation. The majority of children ≥ 6 years of age received either the soft-gel capsules or the melt-extrusion tablet so the previously described Weibull function for LPV/RTV was implemented. For the 2 to < 6 years of age population simulation, only subjects receiving the tablet formulation were included. The plasma concentration data was dose-normalized linearly to the most frequent administration per age group: 120/30, 160/40, 200/50, and 400/100 mg of LPV/RTV for the 0 to < 2 years, 2 to <6 years, 6 to <12 years, and 12 to <18 years of age groups, respectively. The AFE for concentrations were calculated using the simulated geometric mean.

DDI Simulations between LPV/RTV with Rifampin in Pediatric Patients

The adult rifampin PBPK model was scaled to pediatric patients and evaluated using PK data collected from 76 South African children with TB (2 months to 11.3 years of age) receiving 10 or 15 mg/kg of rifampin (Supplementary Methods). We simulated standard dosing (LPV/RTV 230/57.5 mg/m² PO BID) with and without rifampin 10 mg/kg PO daily and super-boosted LPV/RTV dosing (230/230 mg/m²) PO BID with rifampin 10 mg/kg daily PO. The population simulations were performed in 100 virtual children between 0.5 and 4.5 years of age for 2 weeks using the liquid formulation of Kaletra® in the fed-state. The simulated PK parameters were compared with observed data in children.²³

Dosing simulations were next performed to optimize dosing in children (2 months to 8 years of age) receiving super-boosted LPV/RTV with 10 mg/kg and 15 mg/kg PO daily rifampin. We simulated 100 virtual subjects per WHO simplified weight bands (3–5.9, 6–9.9, 10–13.9, 14 to 19.9, and 20–24.9 kg), and targeted weight based dosing for super-boosted LPV/RTV (1:1 ratio) that would achieve ≥ 90% of virtual subjects obtaining LPV trough values > 1 mg/L.²⁵ The PBPK-simulated results were also compared with suggested dosing obtained from a population based PK (PopPK) study developed in HIV-infected children with and without TB receiving super-boosted LPV/RTV with and without 10 mg/kg PO daily rifampin.³⁸

Results

Adult LPV/RTV PBPK Model Evaluation

The LPV/RTV PBPK model included CYP3A4, CYP3A5, and CYP2D6 metabolism; P-glycoprotein transport; glomerular filtration; CYP3A4 and CYP3A5 time-dependent and competitive inhibition and induction; and P-glycoprotein competitive inhibition (Table 1). The simulated to observed AUC_0-τ,ss, AUC_0–24,ss and C_max,ss mean ratios were between 0.7 and 2.0 (Table 2; Figure 1; Supplementary Table 2, Supplementary Table 3; Supplementary Table 4).

Table 2.

Comparison between the observed and simulated lopinavir and ritonavir area under the concentration vs. time curve and maximum concentration following the Kaletra® fixed-dose combination (400/100 mg lopinavir/ritonavir) in healthy and HIV-infected adults.

	Lopinavir AUC0-τ,ss (μg·h/mL)	Ritonavir AUC0-τ,ss (μg·h/mL)	Lopinavir Cmax,ss (μg/mL)	Ritonavir Cmax,ss (μg/mL)
Virtual adults
Simulateda	93.7	4.5	8.9	0.53
Healthy adults
Observedb	96.8 ± 21.8	4.4 ± 1.8	11.2 ± 2.9	0.85 ± 0.45
Ratiob	1.0	1.0	0.8	0.6
Observedc	70.9 ± 37.0	3.08 ± 2.79	7.67 ± 2.93	0.42 ± 0.36
Ratioc	1.3	1.5	1.2	1.3
HIV-infected adults
Observedd	92.6 ± 36.7	N/A	9.8 ± 3.7	N/A
Ratiod	1.0	N/A	0.9	N/A
Observede	95.3 (60.3–119.3)	4.97 (2.82–8.74)	N/A	N/A
Ratioe	1.0	0.9	N/A	N/A

^aSimulated data presented as the geometric mean in a virtual population of 100 white Americans (50% female) from 25 to 47 years of age. Ratio calculated as the ratio of the simulated to observed mean value.

^bObserved data presented as the mean ± standard deviation for 16 healthy volunteers receiving Kaletra (400 mg lopinavir with 100 mg ritonavir) oral (PO) BID for 16 days administered as the fixed-dose melt extrusion tablet formulation (Kaletra®)³².

^cObserved data presented as the mean ± standard deviation for 24 healthy adults receiving Kaletra (400 mg lopinavir with 100 mg ritonavir) PO BID for 11 days.³³

^dObserved data presented as the steady-state mean ± standard deviation for 19 HIV-infected subjects receiving Kaletra (400 mg lopinavir with 100 mg ritonavir) PO BID as the fixed-dose melt extrusion tablet formulation⁵.

^eObserved data presented as the mean (min-max) obtained from eight HIV-infected adults receiving Kaletra 400 mg lopinavir with 100 mg ritonavir) PO BID with tenofovir with or two other nucleotide reverse transcriptase inhibitor drugs for 2 weeks³⁵.

Abbreviations: HIV: human immunodeficiency virus; AUC_0-τ,ss: area under the concentration vs. time curve from 0 to tau (0 to 12 hours) at steady-state; C_max,ss: maximum concentration at steady-state, N/A: not-available.

Figure 1.

Population simulations of lopinavir and ritonavir (400/100 mg) concentration vs. time profiles administered PO BID for two weeks BID HIV-infected and healthy adults.

Population simulations for lopinavir (A) and (B) and ritonavir (C) and (D) based on 100 virtual white American subjects (50% female) from 25 to 47 years of age receiving 400/100 mg lopinavir/ritonavir orally BID for two weeks. The 90% prediction intervals is shown as a solid grey region and the geometric mean is shown as a solid black line. Observed data for lopinavir and ritonavir in HIV infected subjects with plasma samples collected over 12 hours between study weeks 2 and 4^34,35 (B and D) and in healthy adults with plasma samples collected on day 11³³ (A and C) are presented as dots with color stratified by individual. The average fold error, calculated using the equation with the simulated geometric mean ${10}^{\left(1/\text{n}\right)\left(\sum \text{log}\left(\frac{\text{simulated}}{\text{observed}}\right)\right)}$, was 0.90 for lopinavir and 1.42 for ritonavir for both healthy and HIV-infected subjects.

Adult LPV/RTV with Rifampin DDI Evaluation

A previously published adult PBPK model for rifampin was modified using PK data in adult patients with newly diagnosed pulmonary TB receiving rifampin at 450 mg PO daily (body weight < 50 kg) or 600 mg PO daily (body weight ≥ 50 kg) in combination with isoniazid, pyrazinamide, and ethambutol as fixed-dose combination tablets (Supplementary Results; Table 1; Supplementary Figure 1).³⁹ The simulated to observed geometric mean fold ratios for C_max,ss, AUC_0-τ,ss, and C_min,ss LPV were all within 2-fold except for the C_min,ss at standard dosing (LPV/RTV 400/100 mg) with rifampin 600 mg PO daily that was ~9-fold higher (Table 3).

Table 3.

Comparison of observed and simulated lopinavir geometric mean fold ratios (associated 90% prediction intervals) for the area under the concentration vs. time curve, maximum concentration, and minimum concentration at steady-state following co-administration with ritonavir and rifampin in healthy and HIV/TB co-infected adults.

LPV/RTV Dose (mg) + Rifampin	Data Source	Lopinavir Cmax,ss ratio	Lopinavir AUC0-τ,ss ratio	Lopinavir Cmin,ss ratio
400/100 mg LPV/RTV	Observeda	0.45 (0.40, 0.51)	0.25 (0.21, 0.29)	0.01 (0.01, 0.02)
	Simulatedb	0.29 (0.28, 0.31)	0.21 (0.21, 0.22)	0.09 (0.09, 0.10)
	Observedc	0.46 (0.42–0.50)	0.32 (0.28–0.36)	0.05 (0.03, 0.06)
400/400 mg LPV/RTV	Observedd	0.93 (0.81, 1.07)	0.98 (0.81, 1.17)	1.03 (0.68, 1.56)
	Simulatedb	0.96 (0.94, 0.99)	0.94 (0.94, 0.99)	0.85 (0.80, 0.90)
800/200 mg LPV/RTV	Observedd	1.02 (0.85, 1.23)	0.84 (0.64, 1.10)	0.43 (0.19, 0.96)
	Simulatedb	0.82 (0.79, 0.85)	0.86 (0.83, 0.89)	0.73 (0.59, 0.91)
	Observedc	1.13 (0.97, 1.31)	1.02 (0.90, 1.15)	N/A

Data is presented as the geometric mean ratio and the associated 90% confidence interval, except for the observed lopinavir C_min,ss which is presented as the ratio of the median (inter-quartile range).

^aObserved data is presented as the geometric mean (90% confidence interval (CI)) ratio at steady-state based on 22 healthy adults receiving 400 mg lopinavir with 100 mg ritonavir soft-gel capsules BID in combination with 600 mg oral (PO) once daily rifampin relative to 400 mg lopinavir with 100 mg ritonavir soft-gel capsules BID for 10 days⁵.

^bSimulated data based on 100 virtual white American adults (60% female) from 22 to 70 years of age.

^cObserved data presented as the geometric mean ratio (90% CI) in 21 (86% female) HIV-infected adults receiving LPV/RTV with dual nucleoside reverse transcriptase inhibitors at 400/100 mg PO BID, with 600 mg PO daily rifampin, and at 800/200 mg PO BID with 600 mg PO daily rifampin¹⁹.

^dObserved data presented as the geometric mean ratio (90% CI) in healthy adults receiving 800 mg lopinavir with 200 mg ritonavir or 400 mg lopinavir with 400 mg ritonavir soft-gel capsules BID in combination with 600 mg PO daily rifampin for 7 days relative to receiving 400 mg lopinavir with 100 mg ritonavir soft-gel capsules BID for 10 days²¹.

Abbreviations: HIV: human immunodeficiency virus; TB: tuberculosis; N/A: not available; LPV/RTV: lopinavir/ritonavir; CI: confidence interval; C_max,ss: maximum concentration at steady-state; C_min,ss: minimum concentration at steady-state; AUC_0-τ,ss: area under the concentration vs. time curve from 0 to tau (0 to 12 hours) at steady-state.

Pediatric LPV/RTV PBPK Model Evaluation

Population simulations were performed and compared with observed data in 160 HIV-1 infected pediatric subjects (1203 LPV samples) who received LPV/RTV.³⁷ The simulated and observed plasma concentrations for LPV and RTV were all within 0.5 to 2-fold except for infants <6 months of age, and the simulated versus observed median AUC_0–12, C_min, and C_max were comparable (Table 4; Supplementary Figures 2–6).

Table 4.

Comparison of observed and simulated lopinavir area under the concentration vs. time curve, maximum concentration, and minimum concentration at steady-state in infants and children from 0.5 to 4.5 years of age.

Dosing Regimen	LPV/RTV 230/57.5 mg/m2 PO BID alone		LPV/RTV 230/230 mg/m2 PO BID + Rifampin 10 mg/kg PO Daily
PK Parameter	Observeda	Simulated	Observeda	Simulated
AUC0–12,ss (mg*h/L)	117.8 (80.4, 176.1)23	105.9 (74.5, 148.2)	80.9 (50.9, 121.7)23	78.5 (57.3, 111.4)
AUC0–8,ss (mg*h/L)	49.2 (40.7, 86.6)22	78.8 (55.3, 109.1)	N/A	59.5 (44.2, 83.6)
Cmin,ss (mg/L)	4.6 (2.3, 10.4)23	5.5 (3.3, 9.0)	3.9 (2.3, 7.7)23	3.3 (2.3, 5.75)
Cmax,ss (mg/L)	14.2 (11.9, 23.5)23	11.9 (9.2, 17.1)	10.5 (7.1, 14.3)23	9.9 (7.4, 13.6)

Data are presented as the median (interquartile range).

^aObserved data were reported in a population of 15 children (7 months to 3.9 years) co-infected with HIV/TB receiving 230/230 mg/m² (super-boosted dose) of liquid LPV/RTV oral (PO) twice daily (BID) with 10 mg/kg PO daily rifampin²³. Simulations in a population of 100 virtual children from 6 months to 4.5 years of age.

Abbreviations: LPV: lopinavir; RTV: ritonavir; C_min,ss: minimum concentration at steady-state; C_max,ss: maximum concentration at steady-state; AUC_0–12,ss: area under the concentration vs. time curve from 0 to 12 hours at steady-state; AUC_0–8,ss: area under the concentration vs. time curve from 0 to 8 hours at steady-state; PO: PO administration.

Pediatric LPV/RTV with Rifampin DDI Model Evaluation

The AFE comparing simulated and observed rifampin concentrations ranged from 0.82 to 1.25 in TB positive children from 2 months to 11.3 years of age (Supplementary Results, Supplementary Figure 7; Supplementary Table 5).⁴⁰ The simulated and observed PK parameters were comparable in children between 0.5 and 4.5 years receiving the 230/57.5 mg/m² PO BID LPV/RTV and 230/230 mg/m² PO BID LPV/RTV with 10 mg/kg PO daily rifampin regimens.²³ The simulated RTV AUC_0-τ,ss was lower in children than adults possibly due to lower expression of P-glycoprotein and CYP3A4 in pediatric patients (Figure 2).

Figure 2.

Simulated lopinavir and ritonavir area under the concentration vs. time curve (AUC) from 0 to 12 hours for 2 weeks in the presence and absence of rifampin in virtual adults, infants, and children.

Population simulations were performed in 100 virtual infants and children (0.5 to 4.5 years of age) and 100 virtual adults (22 to 70 years of age) for lopinavir/ritonavir (LPV/RTV) with and without rifampin. (A) Boxplots for lopinavir area under the concentration vs. time curve (AUC) from 0 to 12 hours at steady-state. (B) Boxplots for ritonavir AUC at steady-state from 0 to 12 hours. Data was stratified by adults (orange) and children (blue). The simulated dosing regimens in adults were: 400/100 mg LPV/RTV orally (PO) BID (standard dose with no rifampin (RIF)); 400/100 mg LPV/RTV PO BID with 600 mg PO daily rifampin (standard dose with RIF); 400/400 mg LPV/RTV PO BID with 600 mg PO daily rifampin (super-boosted dose with RIF. The simulated dosing regimens in children were: 230/57.5 mg/m² LPV/RTV PO BID (standard dose no RIF); 230/57.5 mg/m² LPV/RTV PO BID with 10 mg/kg PO daily rifampin (standard dose with RIF); 230/230 mg/m² LPV/RTV PO BID with 10 mg/kg PO daily rifampin (super-boosted dose with RIF). The boxplots displays the median (inter-quartile range), the upper whiskers is the 75^th percentile to 1.5 times the inter-quartile range, the lower whisker is the 25^th percentile subtract 1.5 times the inter-quartile range, and observations outside the whiskers are represented as black dots.

Dosing Simulations to Optimize Dosing for LPV/RTV in Pediatric Patients

The simplified weight-based dosing for super-boosted LPV/RTV in children (2 months to 8 years of age) was optimized when administered concurrently with rifampin 10 mg/kg and 15 mg/kg (Table 5). 14–16 mg/kg of super-boosted LPV/RTV (1:1 ratio administered PO BID) in combination with 10 mg/kg PO daily rifampin resulted in 94% to 100% of simulated subjects achieving LPV C_min,ss values > 1 mg/L (Table 5). Increasing the rifampin dosing to 15 mg/kg PO daily with 16 mg/kg PO BID of super-boosted LPV/RTV resulted in 93% to 98% of simulated subjects achieving LPV C_min,ss values > 1 mg/L (Table 5).

Table 5.

Comparison of observed and simulated lopinavir area under the concentration vs. time curve, maximum concentration, and minimum concentration at steady-state following the simulated dosing recommendations for pediatric patients receiving lopinavir/ritonavir with rifampin.

Age Range (years)	Weight (kg)	Rifampin daily dose PO (mg/kg)	PBPK LPV/RTV (1:1) BID dosing PO (mg/kg)	PopPK LPV/RTV (1:1) BID dosing PO (mg/kg)	LPVAUCτ (mg*h/L)a	LPV Cmax,ss (mg/L)a	LPV Cmin,ss (mg/L)a	LPV Cmin,ss > 1 mg/L (%)
0.02 – 0.5	3.0 – 5.9	10	16	22	161	16.9	9.6	100
0.5 – 1.8	6.0 – 9.9	10	16	16	142	14.9	8.2	95
0.6 – 4.7	10.0 – 13.9	10	16	14	123	8.5	5.7	95
1.5 – 6	14.0 – 19.9	10	14	12	107	11.9	5.5	94
1.5 – 7.9	20 – 24.9	10	14	N/A	96	11.1	4.6	96
0.02 – 0.5	3.0 – 5.9	15	16	N/A	139	15.0	7.9	98
0.5 – 1.8	6.0 – 9.9	15	16	N/A	130	13.9	7.3	95
0.6 – 4.7	10.0 – 13.9	15	16	N/A	113	12.7	4.9	94
1 – 7.5	14.0 – 19.9	15	16	N/A	104	11.6	5.8	94
1.5 – 7.9	20 – 24.9	15	16	N/A	94	10.9	4.5	93

^aSimulated data presented as the geometric mean. Dosing simulations were performed to optimize dosing in infants and children (2 months to 8 years of age) receiving the super-boosted dose of lopinavir/ritonavir (LPV/RTV) in combination with 10 mg/kg daily rifampin and the World Health Organization (WHO) revised rifampin dosing recommendations of 15 mg/kg daily rifampin.¹ We simulated 100 virtual subjects per WHO simplified weight bands (3–5.9 kg, 6–9.9 kg, 10–13.9 kg, 14–19.9 kg, and 20–24.9 kg), and targeted weight-based dosing for super-boosted LPV/RTV (1:1 ratio) that would achieve ≥ 85% of virtual infants and children achieving LPV trough values greater than 1 mg/L. The physiologically-based pharmacokinetic (PBPK) model simulated results were also compared with suggested dosing obtained from a population PK (PopPK) model developed using data from HIV-infected adults and children with and without tuberculosis receiving LPV/RTV with or without rifampin.³⁸

Abbreviations: LPV: lopinavir; RTV: ritonavir: BID: BID: PO: PO administration; AUC_0-τ: area under the concentration vs. time curve from 0 to tau (0 to 12 hours) at steady-state: C_max,ss: maximum concentration at steady-state: C_min,ss: minimum concentration at steady-state; N/A: not available; PBPK: physiologically-based pharmacokinetic; PopPK: population pharmacokinetic.

Discussion

We developed a PBPK model for LPV/RTV with rifampin in order to optimize LPV/RTV super-boosted dosing with rifampin in pediatric patients co-infected with HIV and TB. Adult simulations for LPV/RTV with rifampin were performed using a published rifampin model (with some revised model assumptions), and then simulations were compared with observed data reported in healthy and HIV/TB co-infected adults.³⁶ The LPV/RTV and rifampin PBPK models were scaled to pediatric patients and evaluated using pediatric PK data with and without rifampicin. Finally, dosing recommendations of super-boosted LPV/RTV (1:1 ratio) were simulated in pediatric patients from 2 months to 8 years of age using the WHO revised dosing of rifampin (15 mg/kg) stratified by WHO simplified weight bands.²⁵

The RTV PBPK model leveraged literature data and then the solubility, lipophilicity, transcellular intestinal permeability, and P-glycoprotein V_max were optimized (Table 1). RTV is essentially insoluble in water but the soft-gel capsule and the melt-extrusion tablet formulations of Kaletra® have higher RTV solubility.⁴¹ Since the solubility of RTV is unknown in these formulations, the aqueous solubility of RTV was optimized. The transcellular intestinal permeability was also optimized using a higher value in simulations with doses ≥ 200 mg. Another published PBPK model developed for RTV also optimized the fraction absorbed from 0.08 to 1.00 for doses ranging from 20 mg to 200 mg.⁴² This suggests that there may be some model misspecification regarding CYP3A4 and/or P-glycoprotein intestinal inhibition in the model, however, RTV is typically administered at doses under 200 mg in adults and children. However, RTV-mediated DDI potential for CYP3A4/5 and P-glycoprotein were evaluated based on comparing simulations with clinical DDI data for RTV plus midazolam and sildenafil (CYP3A probes) and digoxin (P-glycoprotein probe) (Supplementary Methods).

The LPV PBPK model was developed and evaluated as the LPV/RTV co-formulation since clinical data for LPV alone is limited due to its poor bioavailability (Table 1; Supplementary Table 1). RTV significantly increases LPV bioavailability, presumably through P-glycoprotein and CYP3A inhibition. LPV-mediated P-glycoprotein inhibition was not included in the model although both induction and inhibition of P-glycoprotein have been reported experimentally for LPV.¹⁰ In healthy human volunteers, co-administration of a single 400 mg dose of LPV with 50 mg RTV enhanced the AUC_0−∞ of LPV in plasma by 77-fold relative to dosing of LPV alone.⁸ Additionally, LPV doses up to 600 mg did not affect the PK profile of 50 mg or 100 mg RTV after multiple dosing of LPV/RTV.³ This suggests that P-glycoprotein and/or CYP3A inhibition by LPV is negligible relative to RTV when administered concurrently.

A previously developed adult PBPK model for rifampin was modified using data from adult TB patients, and then the DDI between LPV/RTV with rifampin was evaluated in adults.³⁶ We simulated standard dosing (400/100 mg), double dosing (800/200 mg), and super-boosted dosing (400/400 mg) of LPV/RTV PO BID with rifampin 600 mg PO daily, and compared these results with observed values reported in healthy and HIV-infected adults. The simulated LPV AUC_0-τ,ss for standard dosing was decreased by 79% with rifampin, compared to 75% reported in healthy adults and 68% reported in HIV-infected adults.^5,19 However, the simulated LPV C_min,ss geometric mean fold ratio for standard dosing with rifampin was 9-fold higher than reported in healthy adults. The actual timing of the trough measurements may differ since the study details were not well described in the package insert.⁵ Additionally, a diurnal pattern for RTV has been reported clinically with lower exposures and higher troughs following the evening dose which was not accounted for in the PBPK model.²⁷ Consistent with clinical studies in HIV-infected and healthy adults, the simulations for double and super-boosted dosing of LPV/RTV overcame rifampin induction and resulted in similar exposures as the standard dosing without rifampin (Table 3; Figure 2).

We focused our simulations on super-boosted LPV/RTV with rifampin in pediatric patients since prior studies reported that doubling the dose of LPV/RTV with rifampin 10 mg/kg resulted in sub-therapeutic LPV trough concentrations in pediatric patients.²² Therefore, we simulated weight-based dosing for super-boosted LPV/RTV with rifampin in virtual pediatric subjects from 2 months to 8 years of age between 3.0–24.9 kg body weights. A prospective study in 96 HIV-infected children with TB (median age of 18.2 months) found that there was a non-inferior difference, measured by LPV troughs > 1 mg/L, of −1.1% (95% confidence interval of −6.9 to 3.2) in children receiving super-boosted LPV/RTV with 10 mg/kg rifampin relative to standard LPV/RTV without rifampin.⁴³ However, studies have not yet evaluated super-boosted dosing of LPV/RTV in combination with the WHO revised rifampin dosing of 15 mg/kg in the fixed-dose combination. The clinical end point was to achieve LPV troughs >1 mg/L based on the concentration of half maximum inhibition of 0.1 μM for wild-type HIV in 50% serum.³ 16 mg/kg of super-boosted (LPV/RTV 1:1 ratio) PO BID with rifampin 15 mg/kg PO daily in children weighing between 3.0 kg to 24.9 kg resulted in >90% of virtual subjects with LPV C_min,ss > 1 mg/L. In contrast, the WHO recommends standard LPV/RTV (4:1) weight-based without rifampin of approximately 20–40 mg/kg, 12–20 mg/kg, 12–16 mg/kg, 10–14 mg/kg, and 10–12 mg/kg for the 3–5.9 kg, 6–9.9 kg, 10–13.9 kg, 14–19.9 kg, and 20–24.9 kg weight-bands, respectively. Thus, we are predicting a less-pronounced age effect on LPV/RTV disposition than the WHO dosing.⁴⁴ Interestingly, we identified minor differences in our simulated dosing recommendations for virtual infants and children receiving rifampin 10 mg/kg or 15 mg/kg of PO daily. This is likely because the mean rifampin average concentration during the dosing interval in children receiving the 10-mg/kg daily dose was 0.67 mg/L, which exceeds the CYP3A4 and P-glycoprotein EC₅₀ value (0.28 mg/L) thus induction is already functioning close to the maximal induction effect (E_max).

We leveraged PBPK modeling to optimize dosing for HIV/TB co-infected pediatric patients receiving LPV/RTV plus rifampin. There are some notable limitations and key assumptions that require further discussion. We performed the DDI simulations assuming that LPV/RTV had negligible impact on rifampin exposures. Both LPV and RTV potently inhibited OATP1B1, however, we did not observe significant differences in rifampin exposure with and without inclusion of these in-vitro parameters (Supplementary Results). There may also be physiological differences in HIV and TB co-infected children, such as malnourishment as well as differences between the White or Black American virtual population and South Africans, which are not incorporated in the pediatric virtual population. Future studies could evaluate these simulated results in pediatric subjects receiving super-boosted LPV/RTV with rifampin. In conclusion, our model simulations suggest that administering super-boosted LPV (LPV/RTV 1:1 ratio) 16 mg/kg PO BID should be sufficient to overcome the inductive effects of rifampin at the revised WHO recommended doses of 15 mg/kg PO daily in children.