Pharmacokinetics and Virological Efficacy after Switch to Once-Daily Lopinavir-Ritonavir in Treatment-Experienced HIV-1-Infected Children

Abstract

Lopinavir-ritonavir (LPV/r) is a protease inhibitor that is used twice daily (BID) in the treatment of HIV infection in children. In the context of a single-center observational study, a switch to a once-a-day (QD) LPV/r regimen was proposed for considerations of convenience and to support adherence. The aims of this study were to compare the pharmacokinetics, viral loads, percentages of CD4⁺ T cells, and lipid profiles after switching from a twice-daily to a once-daily regimen of LPV/r in experienced children. For this purpose, LPV concentrations, viral loads, CD4⁺ T cells, and biochemistry data were measured in routine therapeutic drug monitoring procedures in 45 children and adolescents. Thirty-six children were switched to the QD regimen. Nine children on the BID or QD regimen were added for pharmacokinetic-study purposes only. The QD trough concentrations (C_trough) of lopinavir in plasma were significantly lower than those observed with the BID regimen (P < 0.0001), but the 24-h exposure levels were not significantly lower with the QD than with the BID regimen (P = 0.09). Among 34 evaluable patients who switched from the BID to the QD regimen, the virological efficacy of LPV/r appeared to differ (P < 0.001), with 74% and 57% of viral loads, respectively, being <50 copies/ml (mean follow-up times, 33 and 20 months). Among 22 patients with stable virological control before the switch, 12 experienced either failure or blip (one observation of detectable viral load between two observations of undetectable viral load) after the switch. The change from the BID to the QD regimen did not result in significant differences in CD4⁺ T cell percentages or total cholesterol, high-density lipoprotein (HDL) cholesterol, or triglyceride levels. The switch from the BID to the QD LPV/r regimen led to equivalent exposure and lower C_trough values and resulted in lower levels of virological control in these antiretroviral-experienced children.

INTRODUCTION

Lopinavir-ritonavir (LPV/r) is the protease inhibitor that is most used in children today, administered twice daily (BID) and usually combined with two nucleoside analogues. The protease inhibitor LPV/r has been approved by the European Medicines Agency (EMEA) for children aged at least 2 years (8) and by the U.S. Food and Drug Administration (FDA) for children aged at least 14 days and older (1). LPV/r has demonstrated durable antiviral activity in antiretroviral (ARV)-naïve and protease inhibitor-experienced children, with viral suppression (HIV-1 RNA at <50 copies/ml) in more than 60% of the population treated (60 to 80% of adults demonstrated suppression in previous studies) (6, 20). Since 2006, LPV/r has been available as an oral solution or tablet, 200/50 mg or 100/25 mg, respectively.

There is a general trend toward the simplification of treatment with once-a-day (QD) administration to improve convenience and patient adherence (13). The simplification of LPV/r dosage from 400/100 mg BID to 800/200 mg QD was studied initially in ARV-naïve adults and recently in experienced adults and did not show any difference in therapeutic efficacy in randomized controlled studies (7, 23). The QD regimen option is now officially recommended in naïve adult patients (10). Data from adults often leads to case-by-case uncontrolled extrapolation to children. Physiological differences, mainly in terms of immaturity of enzyme systems and clearance mechanisms, justify additional studies in children to precisely evaluate the pharmacokinetics, efficacy, and acceptability of QD administration. The first observational study in ARV-experienced children showed that QD dosing provided undetectable viral loads (VLs) after 24 weeks of follow-up (21). A second study was then conducted on treatment-naive children infected by HIV and showed equivalence of pharmacokinetics between the two dosing schedules. High variability and low trough concentrations (C_trough) for the QD regimen suggested that therapeutic drug monitoring (TDM) is necessary to ensure adequate concentrations (19).

Lopinavir alone has a poor oral bioavailability and needs to be coadministered with subtherapeutic doses of ritonavir. The absolute bioavailability of LPV/r in humans remains unknown. The times to reach the maximum concentration following oral administration are, respectively, 4.4 h and 6.6 h for a twice-daily regimen or a once-a-day regimen. Lopinavir is highly bound to both albumin and α₁-acid glycoprotein (AAG), with a higher affinity for AAG. LPV undergoes extensive and rapid first-pass oxidative metabolism in the liver, mediated by the cytochrome P450 (CYP) 3A4 isoenzyme. The mean half-life and apparent oral clearance of lopinavir were 4.1 to 5.8 h and 6 to 7 liter/h, respectively, for an adult drug dosage of 400/100 mg twice daily (6).

This single-center observational pediatric retrospective study was based on clinical experience and routine follow-up, including TDM. For all children, the decision to switch from a BID regimen to QD was dictated only by considerations of convenience and support of adherence, outside the frame of pediatric marketing authorizations. It gave us an opportunity to (i) develop a population pharmacokinetic model to describe lopinavir concentrations over time and compare some pharmacokinetic efficiency parameters (exposure and trough concentrations) for the two dosing schedules and (ii) determine whether prescribing QD LPV/r in children/adolescents is equivalent to the BID regimen in terms of therapeutic efficacy and tolerance.

MATERIALS AND METHODS

Patients.

Data were collected from children followed in Hopital Necker—Enfants Malades, Paris, France, who received once-a-day LPV/r. Ethics committee approval and patient consent are not compulsory in France in order to retrospectively use therapeutic drug monitoring data, so no informed consent had to be collected. The ARV therapy was monitored on a routine basis for each sampling every 3 months. The drug, time after the last dosing, body weight, age, and combined treatments were recorded, as well as the VL, CD4⁺ T cell percentage, and total cholesterol (TC), HDL cholesterol (HDLc), and triglyceride (TG) levels obtained during the same day.

The assay for lopinavir was performed using high-performance liquid chromatography in combination with UV detection. The quantification limit of the method was 0.05 mg/liter according to a validated method (12). The determination of HIV RNA levels, or viral loads, in blood samples was performed with a Roche COBAS AmpliPrep/COBAS TaqMan HIV-1 test, with a limit of quantification (LOQ) of 50 copies/ml.

Treatments.

Children were administered LPV/r twice daily as an oral solution (80/20 mg per ml), as soft gelatin capsules (133.3/33.3 mg), or as tablets (100/25 mg or 200/50 mg) according to the dose recommended for their body weight (12/3 mg/kg of body weight from 7 to less than 15 kg, 10/2.5 mg/kg from 15 to 40 kg, and 400/100 mg above 40 kg). Children on the QD regimen received the same total daily dose as on the BID regimen. Children also received, as comedication, lamivudine (3TC; 54% of patients), abacavir (46%), didanosine (39%), zidovudine (19%), stavudine (18%), and tenofovir (11%).

Population pharmacokinetic modeling of LPV/r.

The data were analyzed using the nonlinear mixed-effects modeling program NONMEM (version VI, release 2; ICON Development Solutions, Ellicott City, MD) (4) driven by Wings for Nonmem (http://wfn.sourceforge.net). The first-order method with the INTERACTION option was used. Different structural models for LPV/r pharmacokinetics were investigated: one compartment with linear elimination and first-order absorption rate constant (K_a) or zero-order absorption, with or without a transit compartment for absorption. Because LPV/r was given exclusively by the oral route, clearance (CL) and volume of distribution (V) are apparent parameters, as V/F and CL/F, where F is the unknown bioavailability fraction.

The between-subject variabilities (BSVs) were assumed to be exponential. Proportional, additive, or mixed-error models were investigated to describe the residual variability.

The main covariates of interest in the population were age, sex, and body weight. Parameter estimates were standardized for a mean standard body weight using an allometric model as follows: P_i = P_STD(BW_i/BW_STD)^PWR, where P_STD is the standard value of the parameter for a patient with the standard body weight (BW) value and P_i and BW_i are the parameter value and body weight of the ith individual. The PWR (power) exponents may be estimated from the data. However, from allometric scaling theory, these are typically 0.75 for clearance and 1 for volume of distribution (2). All the covariates were tested via an upward model building. A covariate was selected if (i) its effect was biologically plausible, (ii) it produced a minimum decrease of 6.63 units (χ² with 1 df, 0.01) in the objective function value, and (iii) it produced a reduction in the variability of the pharmacokinetic parameter, assessed by the associated intersubject variability. All samples were included in the analysis except those with an undetectable LPV concentration (<0.05 mg/liter), representing less than 4% of the total observations. Graphical evaluation of the goodness-of-fit was mainly assessed by observed versus predicted concentrations (PRED-DV) and weighted residuals (WRES) versus time and/or weighted residuals versus PRED. The final population model was also evaluated by the normalized prediction distribution errors (npde) metrics (5) and a visual predictive check. The stability of the model and accuracy of the parameters were assessed by a bootstrap method implemented in Wings for Nonmem and diagnostic graphics, and distribution statistics were obtained using RfN (link on http://wfn.sourceforge.net) via the R program (17).

Empirical Bayesian estimates (EBE) of the pharmacokinetic parameters were then used to calculate the individual area under the concentration-time curve (drug exposure) from time zero to 24 h (AUC_0–24) and the trough concentration (C_trough). The AUC_0–24 values for the QD and BID doses, respectively, were calculated as dose₂₄/CL/F and (dose₁₂/CL/F) × 2. The effects of the switch from BID to QD treatment on lopinavir 24-h exposure and C_trough were examined and were assessed by a repeated-measure analysis of variance (ANOVA), performed using the nlme package (15) with the R program.

Effect of the switch on viral load.

For each patient with at least 2 VL observations before the switch and 3 VL observations after the switch, the VL profile before the switch was drawn, and patients were ranked among 3 categories: BID responder (fully controlled)—all the VLs were lower than 50 copies/ml (1.7 log₁₀); BID “blipper”—only one VL was above 50 copies/ml between 2 undetectable levels; and BID nonresponder (uncontrolled)—at least 2 successive VLs were >50 copies/ml.

Relationship between VL and genotypic resistance mutations.

The reverse transcriptase (RT) and protease gene sequences were determined from plasma samples with the Agence Nationale de Recherches sur le Sida (ANRS) consensus method, as previously described. Genotypic resistance was analyzed with the last genotypic profile available before the switch to the QD regimen. The LPV mutation score was determined according to the last tables of rules of the French ANRS AC11 Resistance study group (available at http://www.hivfrenchresistance.org/2009/tab2.html). This score indicated the number of the following mutations in the protease gene: L10F/I/R/V, K20M/N/R, L24I, L33F, M46I/L, I50V, F53L, I54L/T/V, L63P, A71I/L/V/T, V82A/F/S/T, I84V, and L90M (6, 14, 16, 20).

The score for genotypic susceptibility to drugs (or genotypic sensitivity score [GSS]) for the QD combination (including LPV/r) was interpreted using the French ANRS 2009 algorithm, as follows: 1, susceptible to drug; 0.5, possibly resistant to drug; and 0, resistant to drug.

Effects of the switch on the different markers of effects.

The effects of the switch from BID to QD treatment were examined for (i) viral loads, (ii) CD4⁺ T cell percentages, and (iii) lipid profiles, including HDLc, total cholesterol, and triglyceride levels. A repeated-measure ANOVA, performed using the nlme package (15) with the R program, was used to test possible differences between markers with QD and BID regimens. The distributions of patients with copies/ml lower than (success) or greater than 50 (failure) were also compared between BID and QD regimens with a repeated-measure analysis for binomial distribution using the lme4 package (3) with the R program.

RESULTS

Demographic data.

This single-center observational study enrolled 36 children (23 boys and 13 girls) treated with LPV/r and for whom a proposition was made to switch to QD treatment. Five patients were on first-line antiretroviral therapy, and 31 were ARV experienced. For the population pharmacokinetic study, additional data from 9 children receiving either BID only or first-line QD LPV/r were added. The median age was 9 years (range, 6 months to 19 years), and the median body weight was 31 kg (range, 7 to 68 kg). A total of 302 LPV concentrations (219 BID and 83 QD) were available for pharmacokinetic evaluation. The median number of samples per patient was 7 (minimum to maximum, 1 to 16), and the median range of blood samples after drug intake was 4.5 (0 to 37.4) hours. The mean follow-up times for the 45 children (27 boys and 18 girls) were, for BID treatment, 31 months (interquartile range, 12 to 43; median, 31), and for QD treatment, 11.5 months (interquartile range, 8 to 25; median, 12). The median BID and QD doses were 266 mg and 600 mg, respectively (Table 1). The children received LPV BID as capsules (59%), oral solution (28%), or tablets (13%) and LPV QD as tablets (92%), capsules (5%), or oral solution (3%).

Table 1.

Characteristics of the 45 children

Characteristic	Measure at baseline
Gender [no. (%)]
Male	27 (60)
Female	18 (40)
Age (yr) [median (range)]	8.8 (0.5–18.9)
Weight (kg) [median (range)]	31.3 (7–68)
ARV experienced [no. (%)]	40 (88.9)
Viral load
<1.7 log10 HIV RNA copies/ml [no. (%)]	11 (24.4)
>1.7 log10 HIV RNA copies/ml [no. (%)]	34 (75.6)
Median (range)	3.7 (1.8–5.9)
% CD4+ T cells [median (range)]	26 (1–48)
Dose (mg) [median (range)]
BID	266 (80–532)
QD	600 (266–800)
Length of follow-up (mo) [median (range)] of patients on:
BID regimen	31 (12–43)
QD regimen	12 (8–25)

Population pharmacokinetics.

The data were best described by a one-compartment model with first-order absorption and elimination. Between-subject variability could be precisely estimated only for apparent CL. Then, the BSVs for V and K_a were fixed to 30% and 50%, respectively. The combined proportional and additive model was used to describe the residual variability. No between-occasion variability parameter could be estimated. The parameter estimates of this basic model were as follows: CL/F, 2.6 liter/h (BSV 0.32); V, 42.4 liter; and K_a, 0.3 h⁻¹.

In a second step, CL and V were standardized to body weight using the fixed power exponents 0.75 and 1. This improved the predictive performance of the model and significantly decreased the variability of CL/F from 0.32 to 0.16. No other covariate effect could be identified (gender, combined use of protease inhibitors, or nucleoside or nonnucleoside analogues). Clearance and its associated BSV were accurately estimated, and the confidence intervals (CIs), derived from the bootstrap analysis, were reasonably narrow and did not include zero. The final pharmacokinetic parameters [mean (90% CI from bootstrap analysis)] were 4.5 liter · h⁻¹ 70 kg⁻¹ (4.23 to 4.79), 66.9 liter 70 kg⁻¹ (36 to 141), and 0.141 h⁻¹ (0.07 to 0.32) for CL/F, V/F, and K_a, respectively. The BSV estimate for CL/F was 0.16 (0.11 to 0.20). The constant and proportional components for the residual variability were 2.4 mg/liter (0.62 to 3.5) and 0.35 (0.24 to 0.43), respectively.

The visual predictive check performed on the final model showed that the average model prediction matched the observed concentration-time courses for the BID and QD regimens. Since patients received different drug dosages, the observed and predicted concentrations were normalized to the following mean dosages: 300 mg BID and 600 mg QD. Accordingly, 8% and 7% (exact binomial test 95% confidence intervals of 5 to 12% and 3 to 15%) of the BID and QD dosing observations were outside the 90% confidence limits (Fig. 1). The mean and variance of the npde metrics were not significantly different from 0 (P = 0.62) and 1 (P = 0.68), respectively, and their distribution was not different from a normal one (P = 0.18).

Fig. 1.

Visual predictive check standardized to 300-mg BID (left, THO 12) and 600-mg QD (right, THO 24) dosages. Solid lines and dashed lines are the median and prediction interval limits (5th and 95th percentiles) of 1,000 simulated predictions of the final model.

As expected, the C_trough values were significantly different with the BID and QD dosages (P < 0.0001). No significant or relevant differences were observed between QD and BID for the 24-h exposures (P = 0.09). The results are summarized in Table 2.

Table 2.

Effects of switch from BID to QD regimen on C_trough, AUC_0–24, viral load, CD4⁺ T cell percentage, and blood lipids

Parameter	BID regimen		QD regimen		P valuea
	Mean	Range	Mean	Range
Pharmacokinetic targets
Ctrough (mg/liter)	8.5	3.05–17.7	5.82	2.36–9.19	<0.0001
AUC0–24 (mg · h/liter)	232	83–520	212	90–318	0.09
Therapeutic markers
% of samples with VL of <50 copies/mlb	74	NA	57	NA	<0.001c
VL (log10 copies/ml)b	2.2	1.7–5.7	2.3	1.7–5.8	0.12
% CD4+ T cells	26.1	2–50	26.9	1–48	0.12
Tolerance markers
HDLc (mmol/liter)	1.5	0.7–2.8	1.4	0.5–2.3	0.01
Total cholesterol (mmol/liter)	5	2.5–9.2	4.8	1.1–9.3	0.42
Triglycerides (mmol/liter)	1.7	0.5–6.1	1.8	0.5–5.5	0.39

^aAssessed by repeated-measure ANOVA except where otherwise indicated.

^bLengths of follow-up for viral load, in months, for the BID and QD regimens, respectively, were as follows: means, 33.8 and 19.6, and ranges, 2 to 74 and 2 to 36.

^cAssessed by repeated-measure binomial distribution test.

Effect of the switch on viral load level profile.

In the 36 children who made the switch, 2 patients had too short a follow-up time or VL data missing before the switch and could not be included in the virological efficacy analysis. For the 34 evaluable patients, 294 viral load observations were available, with at least 2 VL observations before and 3 after the switch. Patients were sorted among the 3 categories described above according their BID VL profile (Fig. 2). Thereafter, their profiles after the switch to QD became as follows: for the 22 BID responders, 10, 5, and 7 patients became responders, blippers, and nonresponders, respectively; for the 5 BID blippers, 2, 1, and 2 patients became responders, blippers, and nonresponders, respectively; and the 7 BID nonresponders were unchanged.

Fig. 2.

Viral load profiles as the function of time of treatment for BID responders, BID blippers, and BID nonresponders. Time zero (symbolized by the vertical line) corresponds to the first evaluation after the switch from BID to QD; negative time corresponds to the delay before the switch and positive time to the delay after the switch. Each point corresponds to the viral load in one sample. The horizontal line represents the target viral load of 1.7 log₁₀ copies/ml.

Taking into account the LPV/r concentrations after the first 6 months of treatment, there was no significant overall difference in the mean log₁₀ copies/ml between the BID and QD treatments (Table 2). However, when expressed as repeated binary responses per patient, i.e., VLs of <50 copies/ml and >50 copies/ml, there was a significant difference between the two regimens: the proportion of samples with undetectable VL during the BID treatment was significantly greater than that observed during the QD treatment (Table 2): 74% versus 57% (P < 0.001). Figure 3 depicts the proportion of undetectable VLs as a function of treatment duration.

Fig. 3.

The proportions of patients with virologic success (VL < 50 copies/ml) at 4 observations each before and after the switch. Key: open circles, observed proportion; solid line, mean; dashed lines, 95% confidence interval.

Relationships between VLs, pharmacokinetic targets, compliance, and genotypic resistance mutations.

In the subgroup of 22 BID responders, 5 and 7 became blippers and nonresponders following the switch. In this subgroup, there were no significant differences between responders, blippers, and nonresponders for the mean values of AUC_0–24 and C_trough. Based on a subjective evaluation of compliance, the 10 children who maintained undetectable viral loads on the QD regimen were considered strictly compliant by the medical team, whereas 5 of the 7 QD nonresponders were less compliant. In contrast, the 5 children (blippers) for whom one detectable viral load was observed on the QD regimen appeared to be strictly compliant. For these 22 BID responders, the last genotypic profile of resistance available before the switch showed no difference between children who maintained a good virological response and all children with a lower level of control (blippers plus those with control failure) after the switch: the genotypic sensitivity scores—as a reflection of the drugs' combination potency—were 2.6 versus 2.4, respectively. As expressed by (International Aids Society) scoring, the number of children with 3 or more major mutations for resistance to LPV were 3/10 for the QD responders versus 3/12 for the QD blippers and nonresponders.

Effects of the switch on CD4⁺ percentage and lipid profile.

The mean CD4⁺ T cell percentages were compared for each patient before and after the switch and did not show clinically significant differences (Table 2).

No significant or no clinically relevant differences were observed between the total cholesterol, HDL cholesterol, and triglyceride levels following the change from the BID to the QD regimen (Table 2).

DISCUSSION

The pharmacokinetics of LPV/r were satisfactorily described by a one-compartment model with linear absorption and elimination, including body weight as a covariate. The apparent clearance estimate was 4.5 liters · h⁻¹ 70 kg⁻¹, similar to other estimates in adult patients, i.e., 4.57 liters · h⁻¹ (14) and 5.5 liters · h⁻¹ (11), supporting the use of the allometric scaling. The nonstandardized clearance, 2.6 liters · h⁻¹, was close to the value of 2.7 liters · h⁻¹ obtained in 19 children (1.4 to 12.9 years old) (16). As expected, the exposures (AUC_0–24) were similar between the BID and QD regimens, whereas the QD C_trough values were significantly lower than the BID C_trough values (7, 11).

The markers of ARV treatment studied here, CD4⁺ T cell percentage and HDLc, total cholesterol, and triglyceride levels, were not modified by the switch from the BID to the QD regimen, as observed in previous reports on pediatric patients (22) or on adult patients (7).

Expressed as binary observations (VL < 50 or VL > 50), there was a significant decrease in the proportion of undetectable VLs following the switch to QD regimen: 74% versus 57% VLs were <50 copies/ml. This decrease in the proportion of undetectable HIV RNA viral load after the switch could not be explained exclusively by pharmacokinetics; no relationship was found between virologic efficacy profiles and AUC_0–24 or C_trough. However, the QD regimen could provide a lower intrinsic efficacy than the BID regimen in these partially resistant viruses harboring several specific resistance mutations to LPV. Nevertheless, based on genotypic profiles obtained before LPV/r treatment, there was no difference in terms of number of mutations between children remaining in good virological control and those who failed. Also, for the 22 BID responders, the proportion of patients with a partially active nucleoside reverse transcriptase inhibitor (NRTI) backbone did not differ between those who maintained a good virological response and those with lower levels of control (blippers plus those with control failure), as assessed by their genotypic sensitivity scores estimated at the time of the switch (2.6 versus 2.4, respectively). Therefore, we can assume that the difference between the virological evolution of the different groups of patients is mainly due to the switch to once-daily LPV/r.

This decrease could be due to the LPV/r compliance behavior; indeed, all the patients who maintained virological control on the QD regimen were perfectly adherent to their treatment, whereas two-thirds of the nonresponders were less strictly compliant. Although sufficient in the BID regimen, this suboptimal compliance could have had a greater consequence in the QD regimen: each missed dose on the QD regimen induced a loss of antiviral protection for 24 h, versus 12 h on the BID regimen. The relationship between VL levels and compliance could not explain the overall loss of virological efficacy, since the children experiencing one blip after the switch were all highly compliant patients. Prior studies conducted on the BID regimen have shown that, in ARV-experienced children, the recommended dose of LPV/r could provide suboptimal exposure against virus partially resistant to LPV (16, 18). The once-a-day LPV/r regimen induced lower C_trough values than the BID regimen, which could also explain a loss of virological protection if the therapeutic target was not reached. This is in agreement with the FDA recommendation that the QD regimen be used in certain adult patients with fewer than 3 lopinavir resistance substitutions (9). This suggests a combined effect of a suboptimal LPV/r exposure for mutated viruses in the QD regimen together with suboptimal compliance to explain the clear loss of virological efficacy.

In conclusion, the switch from the BID to the QD LPV/r regimen led to equivalent exposure but resulted in lower levels of virological control in these ARV-experienced children. These assumptions should be prospectively confirmed by the PENTA Foundation KONCERT PENTA 18 trial, registered under number NCT01196195 in the ClinicalTrials.gov database (http://clinicaltrials.gov).