PAGE. Abstracts of the Annual Meeting of the Population Approach Group in Europe.
PAGE 25 (2016) Abstr 5863 [www.page-meeting.org/?abstract=5863]
Click to open
Oral: Lewis Sheiner Student Session
Stein Schalkwijk (1,2), Rob ter Heine (1), Angela Colbers (1), Alwin Huitema (3), Paolo Denti (4), Kelly E. Dooley (5), Edmund Capparelli (6), Brookie Best (6), Tim Cressey (7,8), Rick Greupink (2), Frans GM Russel (2), Mark Mirochnick (9), and David Burger (1)
(1) Department of Pharmacy, Radboud University Medical Centre, Nijmegen, The Netherlands; (2) Department of Pharmacology & Toxicology, Radboud University Medical Centre, Nijmegen, The Netherlands; (3) Department of Pharmacy & Pharmacology, The Netherlands Cancer Institute, The Netherlands; (4) Division of Clinical Pharmacology, Department of Medicine, University of Cape Town, South Africa; (5) Johns Hopkins University School of Medicine, Baltimore, Maryland; (6) UC San Diego Skaggs School of Pharmacy and Pharmaceutical Sciences & School of Medicine, University of California San Diego; (7) Faculty of Associated Medical Sciences, Chiang Mai University, Chiang Mai, Thailand; (8) Harvard School of Public Health, Boston, MA, USA; (9) Boston University, Boston, MA, USA
Globally, efavirenz (EFV) is a cornerstone for the treatment of HIV infection. There is interest in exploring lower EFV doses, in part to avoid toxicities, but largely to reduce health program costs, with the goal of universal access to treatment. It has been shown that 400mg once-daily (QD) was non-inferior to the standard dose of 600mg QD in adults with regards to virologic response.(1) During pregnancy, the pharmacokinetics (PK) of antiretrovirals may be altered due to changes in protein binding, volume of distribution, and/or clearance, potentially leading to sub-therapeutic exposure and consequently a higher risk of treatment failure, drug resistance, and HIV transmission to the neonate.(2) Currently, it is unknown whether the 400mg dose is also appropriate for pregnant women. Pharmacokinetic studies have been performed to investigate the clinical relevance of pregnancy on the PK of EFV but no studies have investigated the lower 400mg dose. Moreover, large variability because of CYP2B6 polymorphisms influencing EFV clearance can make inference challenging.(3-7)
- To develop a physiologically-based (PB) population PK model to describe the PK of EFV in HIV-infected pregnant and non-pregnant women using the largest dataset yet available
- To simulate EFV exposure (C12) using 600mg and 400mg QD during third trimester of pregnancy.
PK data of pregnant and non-pregnant women taking EFV were collected from several clinical trials.(5, 7-13) Patients using potentially interacting concomitant medicines (e.g. rifampicin or isoniazid) were excluded. Population pharmacokinetic modeling was performed with NONMEM 7.3 with FOCE-I.(14)
To account for the relation between hepatic systemic and first-pass metabolism, a well-stirred liver model [eq.1] was implemented.(15)
Hepatic clearance (CLhep) is expressed as a function of hepatic plasma flow (Qhep,plasma), intrinsic hepatic clearance (CLint,hep), and fraction unbound (fu). A pregnancy-induced increase in Qhep,plasma [eq.2] and decrease in fu [eq.3] were included, a priori.
Hepatic blood flow (Qhep) and EFV protein (albumin)-binding dissociation constant (KD) were fixed to reported values.(16, 17) Polynomial relations (validated for use in PBPK models) between gestational age (GA) and albumin concentrations (P) [eq.4] and hematocrit values (Ht) [eq.5] were used to predict pregnancy-induced changes in fu and Qhep,plasma, respectively.(18, 19)
Flow parameters (^0.75) and volumes (^1) were allometrically scaled to a non-pregnant body weight of 70 kg. Additionally, pregnancy was tested as covariate (dichotomous) on all PK parameters. This effect was retained in the model when inclusion was statistically significant (ΔOFV≥3.84; p≤0.05), clinically relevant (>10% change) and physiologically plausible. Prediction-corrected visual predictive checks and routine goodness-of-fit plots were assessed throughout the model building process.(20)
Mid-dose concentrations (~C12) of <1.0 mg/L were previously related to treatment failure.(21) Therefore, the final model was used to simulate (1000x/condition) total and unbound C12 for 600mg and 400mg QD in pregnant (GA 38 weeks) and non-pregnant women.
PK profiles were available from 253 HIV-infected women (1699 samples). Paired observations during pregnancy (642) and non-pregnant (466) were available from 79 women. Median (IQR) non-pregnant weight was 59 (52-68) kg. Median (range) GA was 35 (25-39) weeks. A 2-compartment disposition model with first-order elimination and 3 absorption transit compartments best described the data. Data on CYP2B6 genotype (c.516G>T) in our population were limited (18%). A mixture model was implemented to account for the multi-modal distribution of CLint as a result of CYP2B6 polymorphisms by imputing the missing CYP2B6-related phenotypes; slow (SM), intermediate (IM) and fast (FM) metabolizers.(22) Stochastic simulation and estimation showed that the population frequencies of the mixture could not be identified; therefore population frequencies were fixed to 12, 36 and 52% for the SM, IM and FM, respectively, based on available data on race or region combined with known prevalence of the CYP2B6 genotypes.(23, 24) Final population estimates (reference 70kg) of EFV Ktr, CLint,SM/F, CLint,IM/F, CLint,FM/F, Vc/F, Q/F and Vp/F were 1.7 1/h, 1300 L/h, 3080 L/h, 4410 L/h, 116 L, 35 L/h, and 403 L, respectively. Inter-individual variability was 32% and 47% for CLhep and Ka, respectively. Inter-occasion variability was 26% for pre-hepatic bioavailability. The residual proportional error was 18%. CLhep was increased with 32%, 30%, and 28% at 38 weeks of gestation compared to non-pregnant for SM, IM, and FM. Pregnancy had no effect on CLint. Moreover, pregnancy was not a significant and relevant covariate for Ktr, Q, Vc and Vp.
For EFV 600mg QD, the simulated median (IQR) steady-state C12 in non-pregnant women were 6.59 (4.63-9.03), 2.53 (1.78-3.56), and 1.80 (1.28-2.54) mg/L for SM, IM, and FM, respectively. In pregnant women the predicted C12 were 4.92 (3.46-6.88), 1.83 (1.27-2.65), and 1.35 (0.90-1.96) mg/L. For 400mg QD, the predicted C12 in non-pregnant women were 4.37 (3.17-6.07), 1.74 (1.24-2.32), and 1.17 (0.84-1.64) mg/L for SM, IM and FM, respectively, as opposed to 3.22 (2.23-4.57), 1.26 (0.92-1.75), and 0.82 (0.58-1.20) mg/L during pregnancy. Although this apparent decrease in total concentration indicates sub-therapy, it should be noted that the predicted unbound concentrations were not altered (<2%) by pregnancy.
Pregnancy decreases total EFV C12 but unbound EFV concentrations are predicted to be unchanged. Although this finding warrants in-vivo confirmation, it indicates that a dose reduction to 400mg is feasible.
Discussion: As unbound EFV concentrations were not directly measured in the studies used for this analysis the absence of an effect of pregnancy on unbound concentrations must be confirmed. Nevertheless, the PB approach used in this study suggests that the lower EFV dose might be adequate during late pregnancy, a finding that may have been missed with empirical modeling of total concentrations. The current study uses the largest set of EFV PK data of pregnant and non-pregnant HIV-infected women compiled to date. We were able to differentiate between metabolic phenotypes and used a PB approach to account for pregnancy-induced alterations in fu and Qhep,plasma. Pregnancy was not identified as an additional covariate for CLint,hep. This indicates no or minor pregnancy-related induction of metabolic enzymes involved in EFV metabolism (e.g. CYP2B6). Of note, pregnancy-related induction of CYP2B6 has been suggested but was not confirmed in vivo.(25)