Impact of adherence and anthropometric characteristics on nevirapine pharmacokinetics and exposure among HIV-infected Kenyan children.

CLINICAL SCIENCE

Impact of Adherence and Anthropometric Characteristics on Nevirapine Pharmacokinetics and Exposure Among HIV-Infected Kenyan Children Rachel C. Vreeman, MD, MS,*†‡ Winstone M. Nyandiko, MBChB, MMED, MPH,‡§ Edward A. Liechty, MD,* Naftali Busakhala, MBChB, MMED,‡k Imke H. Bartelink, PhD,¶ Rada M. Savic, PhD,¶ Michael L. Scanlon, MPH,*‡ Samual O. Ayaya, MBChB, MMED,‡§ and Terry F. Blaschke, MD¶#

Background: There are insufficient data on pediatric antiretroviral therapy (ART) pharmacokinetics (PK), particularly for children in low- and middle-income countries.

Methods: We conducted a prospective nevirapine (NVP) PK study among HIV-infected Kenyan children aged 3–13 years initiating an NVP-based ART regimen. NVP dose timing was measured through medication event monitors. Participants underwent 2 inpatient assessments: 1 at 4–8 weeks after ART initiation and 1 at 3–4 months after ART initiation. Allometric scaling of oral clearance (CL)/bioavailability (F) and volume of distribution (Vd)/F values were computed. Nonlinear mixed-effects modeling using the first-order conditional estimation with interaction method was performed with covariates. The impact of adherence on time below minimum effective concentration was assessed in the final PK model using medication event monitors data and model-estimated individual parameters.

impact on CL/F (P , 0.05), with an estimated decrease in CL of 6.2% for each 1-year increase in age. Total body water percentage was significantly associated with Vd/F (P , 0.001). No children had .10% of time below minimum effective concentration when the PK model assumed perfect adherence compared with 10 children when adherence data were used.

Conclusions: Age and body composition were significantly associated with children’s NVP PK parameters. ART adherence significantly impacted drug exposure over time, revealing subtherapeutic windows that may lead to viral resistance. Key Words: pharmacokinetics, adherence, HIV-infected children, resource-limited settings (J Acquir Immune Defic Syndr 2014;67:277–286)

Results: Among 21 children enrolled, mean age was 5.4 years

INTRODUCTION

and 57% were female. CL/F was 1.67 L/h and Vd/F was 3.8 L for a median child weighing 15 kg. Participants’ age had a significant

Combination antiretroviral therapy (ART) that includes a nonnucleoside reverse transcriptase inhibitor has been shown to effectively suppress the HIV virus and significantly reduce the risk of opportunistic infections and HIV-related morbidity and mortality in HIV-infected adults and children.1–7 The success of ART, however, requires appropriate dosing for optimal disease treatment. Exposure to supratherapeutic levels of antiretroviral drugs increases the risk for drug toxicity, whereas exposure to subtherapeutic levels increases the risk for viral resistance.8,9 Currently, there are insufficient pharmacokinetic (PK) data to guide ART dosing for children.10 Current pediatric dosing recommendations are often extrapolated from ART PK data obtained in adults, and studies suggest that commonly used fixed-dose combinations of ART may result in inappropriate dosing in children,11–14 particularly dosing based on the age of the child, with younger children often clearing the drug more rapidly.15,16 Other studies show that even international recommendations for ART dosing based on children’s weights can yield suboptimal drug concentrations.17–21 In resource-limited settings, the prevalence of malnutrition may also impact the metabolism of drugs and thus the adequacy of dosing guidelines.22,23 Optimal HIV therapy for children is limited by both the paucity of pediatric ART PK data and the lack of pediatric-specific ART formulations.

Received for publication February 27, 2014; accepted July 11, 2014. From the *Department of Pediatrics, School of Medicine, Indiana University, Indianapolis, IN; †The Regenstrief Institute, Inc, Indianapolis, IN; ‡Academic Model Providing Access to Healthcare (AMPATH), Eldoret, Kenya; Departments of §Child Health and Paediatrics; kMedicine, School of Medicine, College of Health Sciences, Moi University, Eldoret, Kenya; ¶Department of Bioengineering and Therapeutic Sciences, University of California, San Francisco, San Francisco, CA; and #Department of Medicine (Clinical Pharmacology) and Molecular Pharmacology, School of Medicine, Stanford University, Stanford, CA. Supported in part by a career development award to R.C.V. from the Indiana CTSI (KL2RR025760-01) and by the President’s Emergency Plan for AIDS Relief (PEPFAR) through USAID under the terms of Cooperative Agreement No. AID-623-A-12-0001. Grant (1K23MH087225). The primary author had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. The authors have no conflicts of interest to disclose. The views expressed in this article are those of the authors and do not necessarily represent the view of the Indiana University School of Medicine or the Moi University School of Medicine. Correspondence to: Rachel C. Vreeman, MD, MS, Children’s Health Services Research, Department of Pediatrics, School of Medicine, Indiana University, 410 W, 10th Street, HITS Suite 1000, Indianapolis, IN 46202 (e-mail: [email protected]). Copyright © 2014 by Lippincott Williams & Wilkins

J Acquir Immune Defic Syndr Volume 67, Number 3, November 1, 2014

www.jaids.com |

277


Vreeman et al

Pediatric ART PK and drug exposure data are particularly inadequate for pediatric populations in sub-Saharan Africa, where 90% of the world’s 3.4 million HIV-infected children live.24 Nevirapine (NVP) is among the most common agents used as part of combination ART for children in resource-limited settings. The objectives of this study were to assess NVP PK parameters in HIV-infected Kenyan children and to use mixed-effects modeling to assess sources of variation in NVP PK parameters and drug exposure, focusing on body composition and adherence to ART.

METHODS Setting This PK study was conducted in western Kenya within the Academic Model Providing Access to Healthcare (AMPATH) program.25–27 As of September 2013, AMPATH has enrolled more than 130,000 patients in western Kenya and currently follows more than 63,000 patients (including more than 14,000 children) at 58 urban and rural clinic and satellite clinic locations. AMPATH provides free ART to all patients who qualify for therapy, as well as comprehensive nutrition services, psychosocial support, and economic development training. This PK study took place at AMPATH’s largest clinic at the Moi Teaching and Referral Hospital in Eldoret, Kenya, between May 2008 and May 2009.

Study Population Treatment-naive HIV-infected children aged 3–13 years receiving care and treatment in the Moi Teaching and Referral Hospital AMPATH clinic, who met clinical or immunological criteria for initiating ART that included NVP, were eligible for enrollment. Because previous studies suggested that NVP PK may vary in children younger than 3 years and because smaller children were typically started on a different ART regimen (though with an NVP backbone), we focused this initial evaluation on children of 3 years and older. Caregivers of all children meeting the inclusion criteria and initiating ART at the study clinic were approached consecutively with information about the study until a sample of 21 patients was reached. We had no refusals to participate in the consecutive sample and no withdrawals, so the sample reflected the 21 children initiating ART containing NVP after the study start date. HIV infection was documented by DNA-PCR (Amplicor; Roche, Basel, Switzerland) for children younger than 18 months at the time of diagnosis and by 2 parallel HIV rapid ELISA tests using Determine and Uni-Gold for children older than 18 months. At the time of the study, AMPATH’s pediatric HIV care protocols recommended ART to be initiated for any child younger than 6 years with a CD4 cell percentage of ,15%, any child older than 6 years with a CD4 count ,200 cells per cubic millimeter, and for any child with WHO clinical stages 3 or 4 or CDC stage C. The standard initial ART regimens were zidovudine/lamivudine/NVP for children weighing ,10 kg or stavudine/lamivudine/NVP for children weighing .10 kg. The dosage of NVP was determined by weight- and age-based dosage recommendations in accordance with Kenya National Guidelines

278

| www.jaids.com

and WHO recommendations. The dosage of NVP was 4 mg$kg21$d21 by mouth once a day for the first 2 weeks, then 7 mg$kg21$dose21 by mouth twice a day for those younger than 8 years, or 4 mg$kg21$dose21 by mouth twice a day for those older than 8 years. Current AMPATH protocols still recommend NVP in the standard first-line pediatric ART regimens. All children enrolled in the study completed a 2-week “lead-in” phase of NVP dosing before starting their twice-a-day NVP regimen of either liquid or pill formulations. After the 2-week lead in phase, patients attended regularly scheduled monthly visits throughout the course of the study, and any changes to the patient’s regimen and dosage (mostly by weight-based criteria) were made by the patient’s physician at AMPATH. Children who had entered puberty (as determined by physician Tanner staging with .Tanner stage 1) and children on any medications known to induce, inhibit, or act as a substrate for hepatic metabolism of drugs, including but not limited to medications for the treatment of tuberculosis, anticonvulsant medications, oral antifungal medications, and macrolide antibiotics, were excluded. Because the primary objectives of this study were establishing the feasibility of these procedures in this setting and to generate exploratory data for pharmacokinetic parameters for NVP in this specific population and setting, precise sample size estimates were not possible. By examining only HIV-infected children initiating therapy with NVP, by optimizing the number of plasma drug levels drawn for each child and by drawing these levels on 2 separate occasions for each child, we optimized the PK data available from even this small sample. Software was used to generate the optimal times and number of samples for this sample size.28

Study Design Children were enrolled in the PK study within 2 weeks of initiating ART. At study start, participants received NVP in bottles with electronic monitoring caps called Medication Event Monitoring System (MEMS caps, AARDEX Group Ltd, Sion, Switzerland). The children were scheduled for 2 overnight inpatient hospital admissions during which the study PK assessments were carried out. The first PK assessment was conducted 4–8 weeks after initiating NVP, when NVP concentration should have reached steady state, and a minimum of 2 weeks after twice-a-day dosing of NVP was initiated. The second assessment was conducted 3–4 months after the initiation of ART. All changes in a patient’s regimen and dosage were recorded through chart review by the study staff. To evaluate the PK of NVP, blood samples at steady state were collected predose, and at 1, 3, 8, and 12 hours after an observed dose of NVP. The optimization of the sampling times was determined using PFIM 1.2 software (S. Retout and F. Mentré, ISERM EMI 0357/DEBRC, Bichat Claude Bernard University Hospital, Paris, France).28 At the time of the observed dose of NVP, the participants also received an oral dose of deuterium-labeled water, which allowed for total body water (TBW) measurements, as determined by a dilution method from the same timed plasma samples. Nutritional status of the participants was also assessed through anthropometric measurements taken at the time of each visit, which 2014 Lippincott Williams & Wilkins


included height, weight, mid-arm circumference, and serum albumin measurements. The assessment of PK and nutritional status was carried out in conjunction with adherence assessments. The primary method of adherence assessment was through electronic monitoring of the timing of medication bottle opening using MEMS to estimate dose intake. Electronic dose monitoring of NVP began at enrollment and continued for the duration of the study. The dosing history data from the MEMS were collected at the completion of the study. Adherence was defined by prescribed doses taken by MEMS, and patients were categorized into 3 adherence levels: good adherence ($90% doses taken), average adherence (,90% and $75% doses taken), and poor adherence (,75% doses taken). Children and their caregivers were informed that they were receiving their medicines in bottles that help the clinic to know how they are using the medicines. Participants also received standard adherence counseling in line with standard AMPATH protocol on ART initiation, as well as additional counseling by study staff at the initiation of the study.

Drug and Body Water Analysis Plasma NVP concentrations were measured using a rapid automated enzyme immunoassay developed by ARK Diagnostics in Sunnyvale, CA.29 The ARK NVP assay is based on competitive binding to antibody between drug in the sample and drug-labeled enzymes. Drug concentration was measured spectrophotometrically in terms of enzyme activity with a Roche/Hitachi 902 chemistry analyzer (Roche Diagnostics Corporation, Indianapolis, IN). Each test required 4 mL of plasma that was obtained from a 0.5 mL whole-blood specimen drawn at each time point after centrifugation and separation of plasma. Assay sensitivity was 0.5 mg/mL. Validation data for controls (0.5–8 mg/mL) show interassay precision of ,8.5% coefficient of variation. Accuracy was 216.4% deviation at 0.5 mg/mL and within 6.5% for remaining controls. No interference was noted from other antiretroviral drugs or blank plasma samples.30 The enzyme immunoassay shows good correlation with high-performance liquid chromatography and provides a cost-effective way to determine NVP concentrations in areas with high HIV prevalence and limited testing resources.31 For TBW measurement, the subject was given an oral dose of nonradioactive deuterated water (D2O) at 0.50 g/kg body weight. Two hours was allowed, during which the D2O distributes uniformly in the intra- and extracellular body water, and then blood samples obtained were used to determine the dilution of D2O in the body water. The enrichment of the predose blood sample, postdose blood sample, and the dose given was measured using isotope-ratio mass spectrometry. The deuterium oxide dilution space is divided by a factor of 1.04 to calculate TBW.32

Pharmacokinetic Modeling All data were analyzed using nonlinear mixed-effects modeling software (NONMEM 7 Software, ICON Development Solutions, San Antonio, TX) with the first-order conditional estimation with interaction method. A basic model structure was established using the full PK profile data from the first visit. To account for developmental 2014 Lippincott Williams & Wilkins

Pharmacokinetics in HIV-Infected Kenyan Children

changes, PK parameters were allometrically scaled and centered at the observed median of 15 kg. In the final model, flip-flop kinetics was accounted for to avoid the estimation of a slower rate of absorption than the rate of elimination.33 The individual parameters were assumed to be log-normally distributed, and proportional error was used for the description of residual variability. The model-building procedure was guided by the likelihood ratio test, diagnostic plots, and internal model validation techniques, including visual and numerical predictive checks. The covariate search was performed using a stepwise covariate model-building procedure described elsewhere34 using the full data set (ie, used data from both visits for all subjects). The procedure included a forward inclusion step, where parameter–covariate relationships are added to the model in a stepwise manner until no further relationship is statistically significant (P , 0.05). Backward elimination steps followed, where the identified relationships were excluded from the model if they failed to achieve stricter statistical significance (P , 0.01), to account for the multiple testing. Linear and nonlinear relationships were investigated. The covariates tested for inclusion were age, sex, total body water percentage (TBW%), dilution space, TBW/weight (%H2O), and the following anthropometric measurements: weight, height, mid-upper arm circumference, and derived weight-for-age Z (WAZ) score. The final PK model for NVP, using the MEMS dosing history and the model-estimated individual parameters, was used to assess the impact of adherence and the individual’s NVP PK on NVP exposure, including projected NVP plasma concentration profiles over the entire study period. For each child, we estimated the following over the 16 weeks of the trial: (1) the number of doses taken within a 4-hour interval of the specified dosing time according to the MEMS-recorded dosing history and (2) the cumulative time with concentration below the minimum effective concentration (MEC), defined as ,3 mg/L that is consistent with previous PK studies.35,36 We compared the exposure outcomes of the data assuming full adherence to the exposure outcomes using MEMSrecorded dosing history. To further enhance our understanding of the PK of NVP in these Kenyan children, the concentration–time profiles were compared with a PK model derived from children in the United States, Thailand, and Zambia.37 We conducted 1000 simulations based on dosing regimen, body weight, and age of the patients in our sample, using the model equations (including equations for body weight and age, variability, and error estimates) obtained from the population PK model presented by Nikanjam et al from US, Thai, and Zambian children. The presence of cytochrome P450 (CYP)2B6 516 GT single nucleotide polymorphism (causing decreased NVP metabolism) was assessed in 26% of children, where they found a mutation frequency of 11% homozygous poor metabolizers (CYP2B6 516 TT). A similar mutation frequency has previously been reported in Kenyan adults (12-16% homozygous CYP2B6 516 TT).38 Therefore, the assumption was made that the Kenyan children and children from the study of Nikanjam et al had the same mutation frequency. Thus, for the simulations in the Kenyan children, the missing genetic www.jaids.com |

279


Vreeman et al

TABLE 1. Study Sample and Patient Characteristics Total Number Patients Samples Females, n (%) Adherence category, n (%) Good Average Poor

21 206 12 (57) 11 (52) 7 (33) 3 (15) Median (range) Initial

Age (yrs) Weight (kg) WAZ score Mid-upper arm circumference (cm) Total body water (%) Dilution space (mL$kg21$L21)

Ethics Committee in Eldoret, Kenya, and by the Indiana University School of Medicine Institutional Review Board in Indianapolis, IN.

4.4 15.5 21.3 14.3

Follow-up

(3 to 13) (11 to 43) (24.7 to 0.9) (12 to 21)

4.5 15.25 21.4 14.4

74.6 (45.8 to 133.9) 11.9 (7.02 to 41.0)

(3.3 to 14) (11 to 44) (24.7 to 0.9) (11.5 to 28)

68.3 (33.0 to 161.7) 11.7 (5.5 to 33.9)

covariate was imputed to the mean allele frequency. We did not evaluate different solid formulations as opposed to some patients described by Nikanjam et al, and our patients did not use ritonavir. The effect of the estimation of poor metabolism did not affect the estimated theta clearance (CL) in our sample (0.198 L/h versus 0.200 L/h), and the adjusted CL value of 0.198 L/h was used to estimate CL.

Regulatory Approvals All parents and guardians gave witnessed verbal informed consent based on a written informed consent document before participating in the study. Children of 10 years and older also provided verbal assent. Participants received a modest honorarium to defray their time and transportation costs. Ethics approval was granted by the Moi University School of Medicine Institutional Research

RESULTS Among 21 HIV-infected Kenyan children who completed the study, 12 (57%) were female, and the mean age and WAZ score were 5.4 years (range: 3–13 years) and 21.5 (range: 24.7–0.9), respectively (Table 1). Most patients had good adherence (52%), 33% had average adherence, and 15% had poor adherence during the study period. At 12-hour postdose (C12h), mean NVP concentration for all participants across assessments was 5.25 mg/L (range: 0.58–16.12). Mean NVP concentration at C12h was slightly lower at the initial assessment (5.05 mg/L) compared with follow-up assessment (5.45 mg/L). At the initial assessment, 6 participants had subtherapeutic Ctrough NVP concentrations (C0h or C12h , 3.0 mg/L). At repeated follow-up assessment, 6 had subtherapeutic Ctrough NVP concentrations, with 3 of these patients having subtherapeutic Ctrough levels at both occasions. Of the 9 patients with subtherapeutic Ctrough NVP concentrations at the initial or follow-up assessment, 4 were classified as having good adherence, 4 as having average adherence, and 1 as having poor adherence. A 1-compartment model described NVP disposition well. In this model, the rate of absorption was slower than the rate of elimination. Evidence from adult populations suggests that NVP shows rapid absorption,39 and flip-flop kinetics were avoided by the estimation of the absorption constant as KA = Kel + uKa. Final population PK parameters are shown in Table 2. Allometric scaling based on body weight was found to be the best size-related predictor of PK parameters (P , 0.001). Additionally, age was found to be significantly associated with CL/bioavailability (F) (P , 0.05). For each 1-year increase in age, CL/F was found to significantly decrease by 6.4%. Body composition parameters were strong predictors of volume of distribution (Vd)/ F. In addition to allometric scaling with body weight, TBW % was significantly associated with drug distribution (P ,

TABLE 2. Population PK Parameter Estimates of the Final PK Model in an Average Patient of 4 Years and 15 kg Structural Model

Estimate

RSE, %

95% CI

CL15kg (L/h)* V15kg (L)† KA (h21)‡ Error uAge* u%BWT† Random variability IOV on CL IIV on V Additive error (mg/L)

1.81 29.5 0.24 1.23 20.0052 20.0134

5.7 11.8 21.8 7.1 12.3 65.7

1.61 to 2.01 22.68 to 36.32 0.14 to 0.34 1.06 to 1.40 20.01 to 0.00 20.03 to 0.00

42.50% 38.20% 1.23

27.0 27.6 7.1

20% to 65% 18% to 59% 1.06 to 1.40

Shrinkage, %

27 34

Bootstrap

RSE, %

Interquartile Range

1.789 31.6 0.25 1.197 20.005 20.013

7 43 59 8 27 66

1.71 to 1.87 28.0 to 39.3 0.16 to 0.27 1.13 to 1.27 20.01 to 0.00 20.02 to 20.01

35% 37% 1.52

63 110 8.3

27% to 43% 29% to 48% 1.27 to 1.77

*CL/F = CL15kg · (WT/15)0.75 · [1 + uAge · (age: 53 months)]. †Vd=F ¼ Vd15kg · ðWT=15Þ1 · e½u%BWT · ð%BWT274:6%Þ : ‡KA = Kel + uKa, where Kel = CL/V. CL/F, oral clearance; Vd/F, volume of distribution; KA, absorption constant; Kel, elimination constant; %BWT, percentage body water; WT, weight.

280

| www.jaids.com

2014 Lippincott Williams & Wilkins



FIGURE 1. PK profiles by adherence level for 3 patients. PK profiles were derived from individual patient PK parameters and dosing history (gray lines) over 3.5 months for 3 patients with different adherence levels (patient 8—good adherence; patient 16—average adherence; patient 3—poor adherence). Observed concentrations are shown (black dots). Cumulative time with concentration values below MEC is indicated by the staircase line. The target MEC between 3 and 8 mg/L is indicated by the broken line. The adherence pattern of the patient is shown along the x-axis. Green fields indicate correct dose intake, orange bars represent 1 missed dose, and red bars indicate 2 or more missed doses. Blue bars indicate overdosing (,7 hours between 2 doses).

0.001). As a child’s weight increased, the TBW% per kilogram decreased. The full drug concentration profile over 16 weeks and the accumulated duration of time below the MEC were 2014 Lippincott Williams & Wilkins

calculated for all patients. Drug concentration profiles for 3 typical patients with good, average, and poor adherence are shown in Figure 1. Simulation of perfect adherence (ie, all doses taken at correct times) using the final PK parameter model for www.jaids.com |

281


Vreeman et al

TABLE 3. Percentage of Time With Concentrations Below Minimum Effective Concentration Patient ID 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Simulated Full Adherence (%)

Including Adherence Data (%)

Adherence Category

0.9 1.1 0.5 1.3 0.5 2.3 0.7 0.5 0.6 0.8 1.2 0.5 0.5 0.7 0.6 1.0 0.6 0.6 0.8 1.0 2.0

9.7 0.3 41.8 1.1 3.5 11.2 15.1 1.4 39.6 6.0 15.9 3.3 5.8 11.9 2.8 15.7 23.2 22.1 43.0 0.0 6.1

Good Good Poor Good Good Average Average Good Poor Good Average Good Good Average Good Average Average Average Poor Good Good

Derived from 21 patients’ average individual parameter estimates using dosing history (ie, MEMS data) versus PK profiles assuming full adherence. Time below MEC was relative to the total time of inclusion in the study (%).

each individual patient showed that with perfect adherence, no patients would have had NVP concentrations with ,10% of the time below MEC (Table 3). When the MEMS dosing histories were used, 10 patients (45%) experienced more than 10% of the time below MEC during the study period. Most children in this cohort had evidence of malnutrition, with a median WAZ score of 21.3 (range: 24.7 to 0.9), but WAZ score was not associated with drug distribution. There were no consistent significant differences in anthropometric measures reflecting children’s nutritional status from the first evaluation to the second among this small sample of children (Fig. 2). In addition, there were no significant differences in PK from the first evaluation to the second. When a subset of individuals who improved with respect to 1 of 4 nutritional status parameters (weight, upper arm circumference, TBW, and TBW%) were examined, there was evidence of changes in PK parameters, but the changes were not consistent (Fig. 2). Although TBW% was significantly associated with drug distribution in this malnourished cohort, the concentration–time profiles from this sample were still consistent with a PK model derived from wellnourished children in the United States and Thailand also using standard WHO dosing guidelines37 (Fig. 3).

DISCUSSION In this study, we estimated population PK parameters for NVP among a cohort of HIV-infected Kenyan children in which malnutrition was prevalent. Despite the prevalence of

282

| www.jaids.com

malnutrition as measured by WAZ scores, the PK parameters of this cohort of children in Kenya fit relatively well within PK models established from children in the United States, Thailand, and Zambia.37 This finding is encouraging in regard to the adequacy of using ART dosing recommendations developed in mostly resource-rich settings for children in resource-limited settings. Variations in PK and drug exposure among children in sub-Saharan Africa have rarely been studied. Pediatric populations in various resource-limited settings may have different characteristics than children in resource-rich countries that may influence ART PK and drug exposure, such as more malnutrition,40,41 different levels of adherence to ART,42 and different prevalence of the particular cytochrome P450 2B6 polymorphisms that influence drug plasma concentrations.43,44 Children in resource-limited settings often have limited access to second- and third-line regimens if they fail their first-line therapy.45,46 Understanding factors influencing ART PK and exposure in children, thus, becomes crucial to improving pediatric ART dosing and long-term outcomes in resource-limited settings. As in our setting, children’s NVP dosing is usually based on weight bands,47 but variations in age,48,49 nutritional status,22,23 and other factors such as genetic differences and concomitant infections10 may influence appropriate dosing. We found that 6 children (29%) had subtherapeutic NVP concentrations based on current dosing guidelines. Although our sample was small, this was a relatively high proportion of patients with subtherapeutic drug levels. Recent studies assessing the PK of NVP in similarly aged children in India,50 Thailand,11,51 Malawi,14,47,52 and Zambia12,13 vary significantly in terms of the proportion of children with subtherapeutic NVP concentrations. For example, 3 studies among 4 groups of children from Malawi and Zambia (2 groups from each country) found subtherapeutic drug concentrations (also defined as ,3 mg/L) between 6% and 27% of children, although these studies were all evaluating NVP within fixed-dose combinations.13,14,47 In this study, age and TBW% were the most significant predictors of PK parameters after allometric scaling based on body weight. The effect of age on PK parameters is unclear, but our results are relatively consistent to those reported by Nikanjam et al37 that modeled age as a nonlinear maturation function. In their study of children from the United States, Thailand, and Zambia, the effect of age on CL was “modest” and was most significant at very young ages. Younger children generally have increased NVP CL compared with adults,53 but there are still few PK data across the pediatric age range. A recent study of children in Malawi found that for every 1-year increase in age, there was an 88% increase in the odds of a therapeutic NVP concentration.14 More data are needed for dosing guidelines, particularly for very young children. We had hypothesized that the children would have significant improvements in their nutritional status during their first months on ART. In this care system, families also had access to nutritional support, including 6 months of food supplements for those with food insecurity at ART initiation. We did not find consistent positive changes in children’s body 2014 Lippincott Williams & Wilkins



FIGURE 2. Oral CL by nutritional parameters. The upper panels show the measurements of nutritional statue at both occasions for all 21 children. Nutritional parameters are WT, weight (kg); PERC, total body water percentage; MUAC, mid-upper arm circumference (cm). The lower panels show the CL values for patients who improved in a specific parameter of nutrition. The values per individual and the median (bold) are shown.

weight over the course of the evaluation, nor did the PK parameters differ significantly from the first evaluation to the second. However, this was a small sample of children followed over a relatively short time period. When we examined a subset of children who improved on at least 1 of 4 nutritional parameters, we did not find clear associations between improved nutritional status and CL. Whether nutritional marker improvements affect NVP PK is inconclusive with these data, and further research using larger samples over longer time periods are needed. Other studies also show conflicting results. In a study of HIV-infected children in 2014 Lippincott Williams & Wilkins

Malawi and Zambia, a higher body mass index for age (ie, lack of wasting) was associated with lower NVP concentrations12; however, another study in Malawi found no association between malnutrition (defined by weight-for-height) and NVP concentration.14 By combining detailed adherence data from electronic monitoring with an individual’s PK model, we identified many more children at risk for viral resistance because of inadequate drug exposure. Examining this combination, a significant proportion of children in our sample spent more than 10% of time below the MEC. Studies have www.jaids.com |

283

Vreeman et al


FIGURE 3. VPC-based simulations (left) and prediction-corrected VPC (right) using an established model based on children from resource-rich countries (the established PK model was taken from Nikanjam et al37). The solid gray lines represent the 2.5th percentile, median, and 95th percentile of the (prediction corrected) observed plasma concentrations. The semitransparent dark gray field represents a simulation-based 95% confidence interval for the median and the semitransparent light gray fields show the 95% confidence intervals of the simulated data. Adaptations are themselves works protected by copyright. So in order to publish this adaptation, authorization must be obtained both from the owner of the copyright in the original work and from the owner of copyright in the translation or adaptation.

284

| www.jaids.com



reported on the use of adherence data alongside PK and pharmacodynamic data in HIV viral models for antiviral response and long-term HIV outcomes54–57; however, most PK models assume static steady-state drug concentrations. Inconsistent drug exposure due to variable patterns of adherence can lead to biased PK parameter estimates and residual variability.58–60 Our results support the importance of using richly sampled adherence data in models of drug profiles and drug exposure, yielding a more accurate assessment of the total time spent below the MEC. Moreover, the results suggest that supporting medication adherence for children in resource-limited settings may be a more important target than dose adjustment based on malnutrition. Almost half of the children had poor or average adherence to ART as measured by MEMS in this study (ie, ,90% doses taken). Although MEMS is often used as the reference standard for adherence to ART,61 routine use of MEMS is usually not feasible because of cost and few validated adherence measurement strategies exist, particularly for children in low-income settings like Kenya.42 There is a critical need to develop and implement accurate, low-cost, and routine adherence monitoring for HIVinfected children. This study has several limitations that merit consideration. The PK model included only a small sample of patients; however, similar sample sizes are frequently used in modeling PK parameters.62,63 The population was also limited to children from a particular geographical region in western Kenya. Nonetheless, because the current body of literature lacks detailed data on NVP PK for children in sub-Saharan Africa, findings from this population in Kenya provide an important addition. Furthermore, we found that our PK model and data were relatively consistent with PK models derived in other settings. We did not have data on virologic outcomes for this sample, and so we could not assess the relationship between time below MEC and virologic failure or resistance. Finally, although electronic dose monitoring is commonly considered an acceptable gold standard for adherence to ART, there are limitations to using MEMS that have not been adequately explored in this setting.64

CONCLUSIONS Evaluating PK parameters for ART for HIV-infected children in Kenya is feasible and necessary for guiding pediatric dosing in resource-limited settings. We found good consistency between our PK model from a sub-Saharan African population and models from other settings. Adherence data were critical to understanding drug exposure in this cohort and revealed a significant proportion of children who spent more than 10% of their treatment time below the MEC. The importance of improving children’s nutritional status merits further attention.

ACKNOWLEDGMENTS Kenneth C. Kasper, ARK Diagnostics, Inc (www.arktdm.com/), provided the assay kits that were used in Kenya. The authors thank Liné Labbe for modeling support, Eunice 2014 Lippincott Williams & Wilkins


Gift Walumbe, Josephine Aluoch Okoyo, and the participating patients and their families. REFERENCES 1. Doerholt K, Duong T, Tookey P, et al. Outcomes for human immunodeficiency virus-1-infected infants in the United Kingdom and Republic of Ireland in the era of effective antiretroviral therapy. Pediatr Infect Dis J. 2006;25:420–426. 2. Gibb DM, Goodall RL, Giacomet V, et al. Adherence to prescribed antiretroviral therapy in human immunodeficiency virus-infected children in the PENTA 5 trial. Pediatr Infect Dis J. 2003;22:56–62. 3. Hogg RS, Yip B, Kully C, et al. Improved survival among HIV-infected patients after initiation of triple-drug antiretroviral regimens. CMAJ. 1999;160:659–665. 4. Needham DM, Hogg RS, Yip B, et al. The impact of antiretroviral therapy on AIDS survival observed in a province-wide drug treatment programme. Int J STD AIDS. 1998;9:370–372. 5. Nyandiko WM, Ayaya S, Nabakwe E, et al. Outcomes of HIV-infected orphaned and non-orphaned children on antiretroviral therapy in western Kenya. J Acquir Immune Defic Syndr. 2006;43:418–425. 6. Peacock-Villada E, Richardson BA, John-Stewart GC. Post-HAART outcomes in pediatric populations: comparison of resource-limited and developed countries. Pediatrics. 2011;127:e423–e441. 7. Wamalwa DC, Farquhar C, Obimbo EM, et al. Early response to highly active antiretroviral therapy in HIV-1-infected Kenyan children. J Acquir Immune Defic Syndr. 2007;45:311–317. 8. Mullen J, Leech S, O’Shea S, et al. Antiretroviral drug resistance among HIV-1 infected children failing treatment. J Med Virol. 2002;68: 299–304. 9. Penazzato M, Dona D, Wool PS, et al. Update on antiretroviral therapy in paediatrics. Antivir Res. 2010;85:266–275. 10. Neely MN, Rakhmanina NY. Pharmacokinetic optimization of antiretroviral therapy in children and adolescents. Clin Pharmacokinet. 2011;50: 143–189. 11. Chokephaibulkit K, Plipat N, Cressey TR, et al. Pharmacokinetics of nevirapine in HIV-infected children receiving an adult fixed-dose combination of stavudine, lamivudine and nevirapine. AIDS. 2005;19:1495– 1499. 12. Ellis JC, L’Homme RF, Ewings FM, et al. Nevirapine concentrations in HIV-infected children treated with divided fixed-dose combination antiretroviral tablets in Malawi and Zambia. Antivir Ther. 2007;12:253–260. 13. L’homme RF, Kabamba D, Ewings FM, et al. Nevirapine, stavudine, and lamivudine pharmacokinetics in African children on paediatric fixeddose combination tablets. AIDS. 2008;22:557–565. 14. Pollock L, Else L, Poerksen G, et al. Pharmacokinetics of nevirapine in HIV-infected children with and without malnutrition receiving divided adult fixed-dose combination tablets. J Antimicrob Chemother. 2009;64: 1251–1259. 15. Burger DM, Verweel G, Rakhmanina N, et al. Age-dependent pharmacokinetics of lamivudine in HIV-infected children. Clin Pharmacol Ther. 2007;81:517–520. 16. Tremoulet AH, Capparelli EV, Patel P, et al. Population pharmacokinetics of lamivudine in human immunodeficiency virus-exposed and -infected infants. Antimicrob Agents Chemother. 2007;51:4297–4302. 17. von Hentig N, Koenigs C, Elanjikal S, et al. Need for therapeutic drug monitoring in HIV-1 infected children receiving efavirenz doses according to international guidelines. Eur J Med Res. 2006;11:377–380. 18. Ren Y, Nuttall JJ, Egbers C, et al. High prevalence of subtherapeutic plasma concentrations of efavirenz in children. J Acquir Immune Defic Syndr. 2007;45:133–136. 19. ter Heine R, Scherpbier HJ, Crommentuyn KM, et al. A pharmacokinetic and pharmacogenetic study of efavirenz in children: dosing guidelines can result in subtherapeutic concentrations. Antivir Ther. 2008;13:779–787. 20. Verweel G, Burger DM, Sheehan NL, et al. Plasma concentrations of the HIV-protease inhibitor lopinavir are suboptimal in children aged 2 years and below. Antivir Ther. 2007;12:453–458. 21. Wintergerst U, Hoffmann F, Jansson A, et al. Antiviral efficacy, tolerability and pharmacokinetics of efavirenz in an unselected cohort of HIV-infected children. J Antimicrob Chemother. 2008;61: 1336–1339.

www.jaids.com |

285

Vreeman et al


22. Krishnaswamy K. Drug metabolism and pharmacokinetics in malnourished children. Clin Pharmacokinet. 1989;17(suppl 1):68–88. 23. Jorquera F, Culebras JM, Gonzalez-Gallego J. Influence of nutrition on liver oxidative metabolism. Nutrition. 1996;12:442–447. 24. World Health Organization. Global HIV/AIDS Response: Epidemic Update and Health Sector Progress Towards Universal Access—Progress Report 2011. Geneva, Switzerland: World Health Organization; 2011. 25. Quigley F. Walking Together, Walking Far. Indianapolis, IN: Indiana University Press; 2009. 26. Einterz RM, Kimaiyo S, Mengech HN, et al. Responding to the HIV pandemic: the power of an academic medical partnership. Acad Med. 2007;82:812–818. 27. Einterz RM, Kelley CR, Mamlin JJ, et al. Partnerships in international health. The Indiana University-Moi University experience. Infect Dis Clin North Am. 1995;9:453–455. 28. Retout S, Mentre F. Optimization of individual and population designs using Splus. J Pharmacokinet Pharmacodyn. 2003;30:417–443. 29. Kasper KC, Moon B, Nguyen J, et al. Homogeneous Enzyme Immunoassay for HIV-1 Antiretroviral Drugs: Darunavir and Maraviroc. Stuttgart, Germany: International Association of Therapeutic Drug Monitoring & Clinical Toxicology Kultur- und Kongresszentrum Liederhalle; 2011. 30. Holland D, Espina-Quinto R, Stefanski E, et al. Rapid, automated immunoassay for therapeutic drug monitoring of nevirapine using ARK NVP-test: method validation, application and comparison with HPLC plasma method. Paper presented at: 45th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC), 2005; Washington, DC. 31. Cusato M, Villani P, Bartoli A, et al. Comparison of an HPLC-UV method with a rapid automated immunoassay (EIA) to measure nevirapine concentration in HIV+ patients. Paper presented at: 8th International Workshop on Clinical Pharmacology of HIV Therapy, 2007; Budapest, Hungary. 32. Schoeller DA, van Santen E, Peterson DW, et al. Total body water measurement in humans with 18O and 2H labeled water. Am J Clin Nutr. 1980;33:2686–2693. 33. Yanez JA, Remsberg CM, Sayre CL, et al. Flip-flop pharmacokinetics– delivering a reversal of disposition: challenges and opportunities during drug development. Ther Delivery. 2011;2:643–672. 34. Wahlby U, Jonsson EN, Karlsson MO. Comparison of stepwise covariate model building strategies in population pharmacokineticpharmacodynamic analysis. AAPS PharmSci. 2002;4:E27. 35. de Vries-Sluijs TE, Dieleman JP, Arts D, et al. Low nevirapine plasma concentrations predict virological failure in an unselected HIV-1-infected population. Clin Pharmacokinet. 2003;42:599–605. 36. Duong M, Buisson M, Peytavin G, et al. Low trough plasma concentrations of nevirapine associated with virologic rebounds in HIV-infected patients who switched from protease inhibitors. Ann Pharmacotherapy. 2005;39:603–609. 37. Nikanjam M, Kabamba D, Cressey TR, et al. Nevirapine exposure with WHO pediatric weight band dosing: enhanced therapeutic concentrations predicted based on extensive international pharmacokinetic experience. Antimicrob Agents Chemother. 2012;56:5374–5380. 38. Oluka MN. Pharmacogenetics of Drug Metabolizing Enzymes and Clinical Implications in Selected Kenyan Populations. Nairobi, Kenya: College of Humanities and Social Sciences, University of Nairobi; 2012. 39. Mirochnick M, Clarke DF, Dorenbaum A. Nevirapine: pharmacokinetic considerations in children and pregnant women. Clin Pharmacokinet. 2000;39:281–293. 40. UNICEF. The State of the World’s Children 2007. UNICEF, Geneva, Switzerland; 2007. 41. Padmapriyadarsini C, Pooranagangadevi N, Chandrasekaran K, et al. Prevalence of underweight, stunting, and wasting among children infected with human immunodeficiency virus in South India. Int J Pediatr. 2009;2009:837627. 42. Vreeman RC, Wiehe SE, Pearce EC, et al. A systematic review of pediatric adherence to antiretroviral therapy in low- and middle-income countries. Pediatr Infect Dis J. 2008;27:686–691.

286

| www.jaids.com

43. Penzak SR, Kabuye G, Mugyenyi P, et al. Cytochrome P450 2B6 (CYP2B6) G516T influences nevirapine plasma concentrations in HIVinfected patients in Uganda. HIV Med. 2007;8:86–91. 44. Saitoh A, Sarles E, Capparelli E, et al. CYP2B6 genetic variants are associated with nevirapine pharmacokinetics and clinical response in HIV-1-infected children. AIDS. 2007;21:2191–2199. 45. Zyl GU, Rabie H, Nuttall JJ, et al. It is time to consider third-line options in antiretroviral-experienced paediatric patients? J Int AIDS Soc. 2011;14:55. 46. Bartlett JA, Shao JF. Successes, challenges, and limitations of current antiretroviral therapy in low-income and middle-income countries. Lancet Infect Dis. 2009;9:637–649. 47. Poerksen G, Pollock L, Moons P, et al. Steady-state nevirapine, lamivudine and stavudine levels in Malawian HIV-infected children on antiretroviral therapy using split Triomune 30 tablets. Antivir Ther. 2010;15:343–350. 48. Foissac F, Bouazza N, Frange P, et al. Evaluation of nevirapine dosing recommendations in HIV-infected children. Br J Clin Pharmacol. 2012; 76:137–144. 49. Fillekes Q, Mulenga V, Kabamba D, et al. Pharmacokinetics of nevirapine in HIV-infected infants weighing 3 kg to less than 6 kg taking paediatric fixed dose combination tablets. AIDS. 2012;26:1795–1800. 50. Swaminathan S, Ramachandran G, Agibothu Kupparam HK, et al. Factors influencing plasma nevirapine levels: a study in HIV-infected children on generic antiretroviral treatment in India. J Antimicrob Chemother. 2011;66:1354–1359. 51. Chokephaibulkit K, Cressey TR, Capparelli E, et al. Pharmacokinetics and safety of a new paediatric fixed-dose combination of zidovudine/lamivudine/nevirapine in HIV-infected children. Antivir Ther. 2011;16: 1287–1295. 52. Corbett AH, Hosseinipour MC, Nyirenda J, et al. Pharmacokinetics of generic and trade formulations of lamivudine, stavudine and nevirapine in HIV-infected Malawian children. Antivir Ther. 2010;15:83–90. 53. Luzuriaga K, Bryson Y, McSherry G, et al. Pharmacokinetics, safety, and activity of nevirapine in human immunodeficiency virus type 1-infected children. J Infect Dis. 1996;174:713–721. 54. Huang Y, Rosenkranz SL, Wu H. Modeling HIV dynamics and antiviral response with consideration of time-varying drug exposures, adherence and phenotypic sensitivity. Math Biosci. 2003;184:165–186. 55. Huang Y. Long-term HIV dynamic models incorporating drug adherence and resistance to treatment for prediction of virological responses. Comput Stat Data Anal. 2008;52:3765–3778. 56. Wu H, Huang Y, Acosta EP, et al. Pharmacodynamics of antiretroviral agents in HIV-1 infected patients: using viral dynamic models that incorporate drug susceptibility and adherence. J Pharmacokinet Pharmacodyn. 2006;33:399–419. 57. Labbe L, Verotta D. A non-linear mixed effect dynamic model incorporating prior exposure and adherence to treatment to describe long-term therapy outcome in HIV-patients. J Pharmacokinet Pharmacodyn. 2006; 33:519–542. 58. Girard P, Sheiner LB, Kastrissios H, et al. Do we need full compliance data for population pharmacokinetic analysis? J Pharmacokinet Biopharm. 1996;24:265–282. 59. Vrijens B, Goetghebeur E. The impact of compliance in pharmacokinetic studies. Stat Methods Med Res. 1999;8:247–262. 60. Vrijens B, Goetghebeur E. Electronic monitoring of variation in drug intakes can reduce bias and improve precision in pharmacokinetic/pharmacodynamic population studies. Stat Med. 2004;23:531–544. 61. Deschamps AE, De Geest S, Vandamme AM, et al. Diagnostic value of different adherence measures using electronic monitoring and virologic failure as reference standards. AIDS Patient Care STD. 2008;22:735–743. 62. De Cock RF, Piana C, Krekels EH, et al. The role of population PK-PD modelling in paediatric clinical research. Eur J Clin Pharmacol. 2011;67 (suppl 1):5–16. 63. Kile DA, MaWhinney S, Aquilante CL, et al. A population pharmacokinetic-pharmacogenetic analysis of atazanavir. AIDS Res Hum Retroviruses. 2012;28:1227–1234. 64. Berg KM, Arnsten JH. Practical and conceptual challenges in measuring antiretroviral adherence. J Acquir Immune Defic Syndr. 2006;43(suppl 1): S79–S87.


Anthropometric, gait and strength characteristics of kenyan distance runners.

Plasma nevirapine concentrations predict virological and adherence failure in Kenyan HIV-1 infected patients with extensive antiretroviral treatment exposure.

Pharmacokinetics-based adherence measures for antiretroviral therapy in HIV-infected Kenyan children.

Effect of diurnal variation, CYP2B6 genotype and age on the pharmacokinetics of nevirapine in African children.

Relationships among age, gender, anthropometric characteristics, and dynamic balance in children 5 to 12 years old.

Nevirapine-related adverse events among children switched from efavirenz to nevirapine as compared to children who were started on nevirapine-based antiretroviral therapy directly.

Nevirapine Resistance in Previously Nevirapine-Unexposed HIV-1-Infected Kenyan Infants Initiating Early Antiretroviral Therapy.

Implementation and Operational Research: Correlates of Adherence and Treatment Failure Among Kenyan Patients on Long-term Highly Active Antiretroviral Therapy.

Clinical Characteristics of Celiac Disease and Dietary Adherence to Gluten-Free Diet among Saudi Children.

An assessment of the impact of host polymorphisms on Plasmodium falciparum var gene expression patterns among Kenyan children.

The effects of prenatal HIV exposure on language functioning in Kenyan children: establishing an evaluative framework.

Impact of Threat Level, Task Instruction, and Individual Characteristics on Cold Pressor Pain and Fear among Children and Their Parents.

Anthropometric characteristics and sex influence magnitude of skin cooling following exposure to whole body cryotherapy.

Impact of Integrated Services on HIV Testing: A Nonrandomized Trial among Kenyan Family Planning Clients.

Effect of blood transfusion on survival among children in a Kenyan hospital.

In utero exposure to iodine-131 from Chernobyl fallout and anthropometric characteristics in adolescence.

Physical activity and anthropometric and functional characteristics of mildly malnourished Senegalese children.

Physiological, Behavioral, and Dietary Characteristics Associated with Hypertension among Kenyan Defence Forces.

Impact of 20 Week Lifestyle Intervention Package on Anthropometric Biochemical and Behavioral Characteristics of Schoolchildren in North India.

Physical fitness and anthropometric characteristics among adolescents living in urban or rural areas of Kosovo.

Impact of psychosocial status and disease knowledge on deferoxamine adherence among thalassaemia major adolescents.

Characteristics of a Kenyan camelpox virus.

Sociocultural and Victimization Factors That Impact Attitudes Toward Intimate Partner Violence Among Kenyan Women.

Comparison of HIV-1 nef and gag Variations and Host HLA Characteristics as Determinants of Disease Progression among HIV-1 Vertically Infected Kenyan Children.