Population pharmacokinetics of paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin in hepatitis C virus genotype 1 infection: analysis of six phase III trials

Abstract

Aim

The aim of the current study was to characterize the population pharmacokinetics of a triple direct‐acting antiviral (DAA) regimen (3D) (ombitasvir, paritaprevir–ritonavir and dasabuvir) and adjunctive ribavirin, and estimate covariate effects in a broad spectrum of subjects with hepatitis C virus (HCV) genotype 1 infection.

Methods

Pharmacokinetic data from six phase III studies and one phase II study in subjects receiving the currently approved doses of the 3D ± ribavirin regimen for treating HCV genotype 1 infection for 12 weeks or 24 weeks were characterized using separate population pharmacokinetic models, built using each component of the regimen from nonlinear mixed‐effects methodology in NONMEM 7.3. In the models, demographic and clinical covariates were tested. Models were assessed via goodness‐of‐fit plots, visual predictive checks and bootstrap evaluations.

Results

The population pharmacokinetic models for each component of the 3D ± ribavirin regimen (DAAs and ritonavir, n = 2348) and ribavirin (n = 1841) adequately described their respective plasma concentration–time data. Model parameter estimates were precise and robust, and all models showed good predictive ability. Significant covariate effects associated with apparent clearance and volume of distribution included age, body weight, gender, cirrhosis, HCV subtype, opioid or antidiabetic agent use, and creatinine clearance.

Conclusion

The population pharmacokinetics of the 3D ± ribavirin regimen components in HCV‐infected patients were characterized using phase II and III HCV clinical trial data. Although several statistically significant covariates were identified, their effects were modest and not clinically meaningful to necessitate dose adjustments for any component of the 3D regimen.

What is Already Known about this Subject

• Triple drug (3D) therapy with ombitasvir, paritaprevir–ritonavir and dasabuvir, with or without ribavirin, is highly effective in patients with hepatitis C virus (HCV) infection.

• Population pharmacokinetic models developed using phase Ib and II data in noncirrhotic HCV patients showed demographic and clinical parameters to have significant effects on clearance and volume of distribution.

What this Study Adds

• Population pharmacokinetic models for the components of the 3D ± ribavirin regimen were developed using data from the pivotal phase III studies at the currently approved doses and formulations across a more diverse patient population.

• Despite significant covariate effects, the magnitudes of changes in exposure were not clinically meaningful to necessitate dose adjustments for any component of the 3D regimen.

Tables of Links

TARGETS
Transporters 2	Enzymes 3
OATP1B1/B3	CYP3A
P‐gp	CYP2C8
BCRP

LIGANDS
Ritonavir
Ribavirin

These Tables list key protein targets and ligands in this article that are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 1, and are permanently archived in the Concise Guide to PHARMACOLOGY 2015/16 2, 3.

Introduction

Hepatitis C virus (HCV) infection presents a global healthcare burden, with 130 to 150 million people infected chronically worldwide 4 and an estimated 2.7 million in the United States 5. Persistent infection is associated with a 20‐year liver cirrhosis risk of 15–30%, and approximately 500 000 people die each year from HCV‐related liver diseases 4. Of the seven recognized HCV genotypes, genotype 1 is the most prevalent worldwide, with subtypes 1a and 1b responsible for the vast majority of infections 6. Genotype 1 infection represents a treatment challenge. Pegylated interferon‐based treatment is associated with clinically significant systemic adverse reactions, and response is influenced by variables such as age, race, HCV genotype, interleukin (IL) 28B genotype, baseline viral load and degree of fibrosis 7, 8, 9, 10, 11. However, with the availability of direct‐acting antiviral agents (DAAs), effective, well‐tolerated, all‐oral, interferon‐free regimens are now the mainstay of treatment for a broad range of HCV patient types 12.

An oral triple DAA drug (3D) regimen, recently approved by numerous regulatory agencies globally, including the United States Food and Drug Administration (ViekiraPak) 13 and the European Medicines Agency (Viekirax and Exviera) 14, 15, has achieved high rates of sustained virological response and is well tolerated in clinical trials of subjects with HCV genotype 1 infection 16, 17, 18, 19, 20, 21, 22. The 3D regimen consists of paritaprevir, a potent nonstructural protein 3/4A protease inhibitor; ombitasvir, a novel nonstructural protein 5A inhibitor; and dasabuvir, a non‐nucleoside, nonstructural protein 5B polymerase inhibitor. Paritaprevir is metabolized primarily by cytochrome P450 (CYP) 3A and is administered with a low dose of ritonavir (the combination is denoted as paritaprevir/r), which enhances its exposure and enables once‐daily administration 23. Dasabuvir is predominantly metabolized by CYP2C8, and ombitasvir is metabolized by amide hydrolysis 13, 24. Other than ritonavir, a potent CYP3A4 inhibitor, the DAAs are not inhibitors or inducers of CYP enzymes 24. At clinically relevant concentrations, paritaprevir is an inhibitor of organic anion transporting polypeptide (OATP) 1B1/B3, and paritaprevir, ritonavir and dasabuvir are inhibitors of P‐glycoprotein (P‐gp) and breast cancer resistance protein. The DAAs and ritonavir are in vitro substrates of P‐gp and breast cancer resistance protein, and paritaprevir is also a substrate of OATP1B1/B3 13, 24. The daily dosage of the 3D regimen consists of two coformulated tablets of ombitasvir/paritaprevir/ritonavir 12.5 mg/75 mg/50 mg and one tablet of dasabuvir 250 mg twice daily 13. Ribavirin is added to the regimen for patients with HCV genotype 1a infection with or without cirrhosis 13.

The clinical development programme for the 3D regimen has yielded a robust dataset, encompassing various dosing schemes and patient populations reflective of those encountered in clinical practice. The population diversity of the phase III studies allows for robust assessments of potential patient variables that could influence drug exposure, which in turn could affect the efficacy and safety of the 3D regimen. Population pharmacokinetic models for paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin based on data from phase Ib and IIa/b studies have shown that age, gender, body weight, non‐Hispanic ethnicity, CYP2C8 inhibitor use and creatinine clearance are significant covariates with apparent clearance or apparent volume parameters; however, based on the magnitude of the effect, no dose adjustments for the 3D regimen are needed based on any patient demographic or clinical characteristic 25.

Comprehensive phase III clinical trials have investigated the efficacy and safety of the 3D regimen with or without ribavirin, across a broad spectrum of more than 2300 genotype 1‐infected patients, including treatment‐naive patients, nonresponders to previous pegylated interferon‐based therapy and patients with cirrhosis 16, 18, 19, 21, 22. All phase III studies evaluated the currently approved 3D regimen. The population pharmacokinetics of paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin in HCV genotype1‐infected subjects were characterized using combined data from six phase III studies and one phase II study, all of which used the currently approved 3D treatment regimen. The objective of the present study was to develop population pharmacokinetic models to identify and evaluate demographic, pathophysiological and treatment factors that influence the pharmacokinetics of the components of the 3D regimen and ribavirin, and their exposures. These data can help to inform clinicians of the need to make dose adjustments in special patient populations.

Methods

Study population

This analysis included adult subjects with HCV genotype 1 infection who were enrolled in one phase II and six phase III studies of paritaprevir/r, ombitasvir and dasabuvir, with or without ribavirin 16, 18, 19, 21, 22, 26. The subjects ranged in age from 18 years to 70 years and had a plasma HCV RNA level exceeding 10 000 IU ml⁻¹. Subjects who tested positive for hepatitis B surface antigen or anti‐HIV antibody were excluded. Of the seven studies, three enrolled treatment‐naive subjects only 18, 19, two enrolled treatment‐experienced subjects 16, 22 and two enrolled both treatment‐naive and treatment‐experienced subjects 21, 26. Two studies focused only on subjects with HCV genotype 1b infection 16, 19 and one study included those subjects with HCV genotype 1a infection 19. Other than one study which enrolled subjects with compensated cirrhosis 21, subjects with cirrhosis were excluded. One study (phase II) enrolled subjects who were on methadone or buprenorphine, with or without naloxone 26.

All studies were conducted in accordance with the Good Clinical Practice Guideline (US Code of Federal Regulations (CFR), 21 CFR parts, 50, 56 and 312) as defined by the International Conference on Harmonization, the Declaration of Helsinki, and/or all applicable federal and local regulations, and Institutional Review Boards, as appropriate. All study participants provided written informed consent.

Study design and treatment

Plasma concentration data for the five drugs of the 3D with or without ribavirin regimens from one phase II study 26 and six phase III studies 16, 18, 19, 21, 22 were used for the analyses. Table 1 summarizes the study design and treatment regimens used in each study. Subjects received coformulated paritaprevir/r–ombitasvir (at a once‐daily dose of paritaprevir 150 mg, ritonavir 100 mg, and ombitasvir 25 mg), dasabuvir (250 mg twice daily), with or without weight‐based ribavirin administered twice daily (1000 mg daily for body weight <75 kg and 1200 mg daily for body weight ≥75 kg), or matching placebo for 12 weeks or 24 weeks (only in patients with compensated cirrhosis).

Table 1.

Summary of clinical trials

Study ( ClinicalTrials.gov identifier)	Treatment regimen (duration) a	N b	Study description	Pharmacokinetic sampling
Phase II 26 (NCT01911845)	Paritaprevir/r/ombitasvir +dasabuvir + RBV (12 weeks)	38	Phase II, single‐arm, open‐label, multicentre study in HCV GT1‐infected adults who were peg‐IFN/RBV treatment‐naive or treatment‐experienced, noncirrhotic, and who were on stable opioid replacement therapyc	At each study visit or on patient discontinuation and intensive sampling of 22 patients (predose and at 2, 4, 6 and 24 h postdose) at a single visit on or after the week 2 study visit
PEARL‐II 16 (NCT01674725)	Paritaprevir/r/ombitasvir +dasabuvir ± RBV (12 weeks)	186	Phase III, open‐label, randomized study in peg‐IFN/RBV treatment‐experienced, noncirrhotic, HCV GT1b–infected adults	At each study visit (weeks 1–12) or on patient discontinuation
PEARL‐III 19 (NCT01767116)	Paritaprevir/r/ombitasvir +dasabuvir ± RBV (12 weeks)	419	Phase III, randomized, double‐blind, controlled, multicentre study in treatment‐naive, HCV GT1b–infected, noncirrhotic adults	Predose on day 1 and at each study visit up to week 12 or on patient discontinuation
PEARL‐IV 19 (NCT01833533)	Paritaprevir/r/ombitasvir +dasabuvir ± RBV (12 weeks)	305	Phase III, randomized, double‐blind, controlled, multicentre study in treatment‐naive, HCV GT1a‐infected, noncirrhotic adults	Predose on day 1 and at each study visit up to week 12 or on patient discontinuation
TURQUOISE‐II 21 (NCT01704755)	Paritaprevir/r/ombitasvir +dasabuvir + RBV (12 or 24 weeks)	380	Phase III, randomized, open‐label, multicentre study in HCV GT1‐infected treatment‐naive and previous peg‐IFN/RBV treatment‐experienced adults with compensated cirrhosis	Predose and 2 h postdose on day 1, and single samples at each study visit up to week 12 or 24, or on patient discontinuation
SAPPHIRE‐I 18 (NCT01716585)	Paritaprevir/r/ombitasvir +dasabuvir + RBV (12 weeks)	628	Phase III, randomized, double‐blind, placebo‐controlled, multicentre study in treatment‐naive, noncirrhotic, HCV GT1‐infected adults	Predose on day 1 and at each study visit up to week 12 or 24, or on patient discontinuation
SAPPHIRE‐II 22 (NCT01715415)	Paritaprevir/r/ombitasvir +dasabuvir + RBV (12 weeks)	392	Phase III, randomized, double‐blind, placebo‐controlled, multicentre study in treatment‐experienced, noncirrhotic, HCV GT1‐infected adults	Predose on day 1 and at each study visit up to week 12 or 24, or on patient discontinuation

GT1, genotype 1; HCV, hepatitis C virus; peg‐IFN, pegylated interferon alpha; RBV, ribavirin

^aCoformulated paritaprevir/r–ombitasvir was dosed at 150 mg paritaprevir, 100 mg ritonavir, and 25 mg ombitasvir once daily and dasabuvir at 250 mg twice daily. Ribavirin was dosed at 1000 mg or 1200 mg daily, divided into two doses in accordance with local labelling

^b N represents the number of subjects with observed or available pharmacokinetic data

^cStable opioid replacement therapy was defined as receipt of methadone or buprenorphine with or without naloxone for at least 6 months before screening

Assessments

Samples were collected before the first dose and without regard to dosing time at each study visit up to week 12 or 24 of active treatment or on subject discontinuation; the daily dosing histories were obtained using the medication event monitoring system (MEMS™, AARDEX Group Ltd., Sion, Switzerland) for all DAAs and ribavirin in two phase III studies (TURQUOISE‐II 21, PEARL‐II 16). In the other four phase III studies (PEARL‐IV 19, SAPPHIRE‐1 18, SAPPHIRE‐II 22, PEARL III 19), where MEMS caps were placed only on ribavirin bottles, dosing times for ombitasvir/paritaprevir/ritonavir and dasabuvir were assumed to be the same time as for ribavirin (a.m. dosing time for ombitasvir/paritaprevir/ritonavir once‐daily tablets; a.m. and p.m. dosing times for dasabuvir twice‐daily tablets). In the phase II study, daily dosing histories were obtained using MEMS for all DAAs and ribavirin. In this phase II study, in addition to the above‐described samples from all subjects at each visit, intensive pharmacokinetic samples were collected at the week 2 or at a later visit from a subset of subjects (n = 22) 26. The actual blood sample collection times were determined based on electronically recorded last dose time by MEMS and used for population pharmacokinetics analyses.

Blood samples were collected according to the protocol specifications for each study and were processed, and concentrations of DAAs and ritonavir were determined according to a previously described validated liquid chromatography tandem mass spectrometric detection method 27. Lower limits of quantification ranges were 0.573–0.601 ng ml⁻¹ for paritaprevir, 0.442–0.462 ng ml⁻¹ for ombitasvir, 4.53–4.58 ng ml⁻¹ for dasabuvir, 4.73–4.93 ng ml⁻¹ for ritonavir and 98.1 ng ml⁻¹ for ribavirin.

Pharmacokinetic model development and statistical analysis

Data from all subjects who received paritaprevir/r, ombitasvir and dasabuvir, with or without ribavirin, and who had pharmacokinetic assessments were included in the analyses.

For each subject, the first below the limit of quantification (BLQ) observation after the last non‐BLQ observation, if any, was substituted with a value equal to the lower limit of quantification divided by two (LLOQ/2). All subsequent BLQs were excluded from the analyses. This approach, commonly referred to as the M6 method 28, has been shown to provide less‐biased parameter estimates 29.

Population pharmacokinetic models for each compound were built using a nonlinear mixed‐effects modelling approach (NONMEM 7.3, ICON Development Solutions, Hanover, MD, USA). Exact pharmacokinetic sampling times were calculated based on the electronically recorded last dose time from MEMS. The first‐order conditional estimation method with ƞ‐ε interaction (FOCE‐INT) was used for model building. Only linear pharmacokinetic models were considered. A base model that defined the structural pharmacokinetic model and the models for interindividual and residual variabilities was developed first. Both one‐ and two‐compartment models were fitted to the data and in case of ribavirin, a three‐compartment model was also tested. The development of the structural model involved a comparison of all potential model representations based on available data and parameter identifiability and plausibility of parameter estimates.

Interindividual variability (IIV) in pharmacokinetic parameters was modelled using the following exponential error equation:

$${\mathit{\theta }}_{\mathrm{i}}={\mathit{\theta }}_{\mathrm{TV}}\times exp\left({\mathrm{n}}^{\mathrm{\theta i}}\right)$$

where θi is the parameter estimate for the ith individual, θ _TV is the typical value of the parameter in the population, η ^θi are individual‐specific random effects for the ith individual and assumed to be normally distributed with mean zero and variance ω²: η ~ N (0, ω²). This assumption imparts a log‐normal distribution on the parameter of interest and expresses the variability as an approximate percentage coefficient of variation (%CV).

Residual unexplained variability (RUV) was modelled using a combined error model, except for paritaprevir, which was modelled using an additive error model on log‐transformed data as follows:

$$C_{\mathit{ij}}={Ĉ}_{\mathit{ij}}\times \left(\mathit{1}+{\epsilon }_{\mathit{1}\mathit{ij}}\right)+{\epsilon }_{\mathit{2}\mathit{ij}}$$

$$\mathit{log}\left(C_{\mathit{ij}}\right)=\mathit{log}\left({Ĉ}_{\mathit{ij}}\right)+{\epsilon }_{\mathit{ij}}$$

where C _ij is the jth measured plasma concentration in individual i, Ĉ _ij is the jth model‐predicted value in individual i and ε _ij is the residual random error for individual i and measurement j. ε _1ij is the proportional component and ε _2ij (or ε _ij for the RUV model representing an additive error on log‐transformed data) is the additive component of the residual random error. The ε values were assumed to be independently and identically distributed with a mean of zero and variance of σ ²: ε ~ N (0, σ ²). A RUV model was selected based on Bayesian information criterion values, model stability and diagnostic plots.

Evaluation of covariate effects

For apparent clearance (CL/F), the tested covariates included body weight, body mass index, body surface area, gender, age, baseline creatinine clearance, status of treatment experience (treatment experienced vs. naive), HCV subgenotype (1a or 1b), race (black vs. nonblack), ethnicity (Hispanic Latino vs. non‐Hispanic Latino and Asian vs. non‐Asian), status of cirrhosis (no/yes), methadone/buprenorphine use, use of ribavirin and use of comedications. For apparent volume parameters (apparent volume of central compartment [V_c/F] and peripheral compartment [V_p/F]), the covariates tested were age, gender, body weight, body surface area and body mass index.

More than 1200 different comedications were concomitantly used by the phase III subjects 30. The effects of these medications on CL/F were evaluated by categorizing the medications into drug classes (those used commonly with the 3D regimen) and into inducer/inhibitor categories. The methodology of the comedication analysis has been described previously 30. Briefly, the potential for a drug class or category to influence the pharmacokinetics of DAAs, ritonavir or ribavirin was determined by calculating steady‐state area under the plasma concentration–time curve (AUC) values from the base models for subjects receiving each drug class or category and comparing these values with control groups (no comedication). A comedication was included in the covariate model evaluation procedure if the minimum initial screening criteria were met. The minimum initial screening criteria, aimed at minimizing selection of random nonmeaningful covariates, were defined as a geometric mean AUC ratio (comedication/control) of ≤0.5 or ≥2.0 for drug classes, ≤0.5 for inducer categories or ≥2.0 for inhibitor categories, in a minimum of 15 subjects receiving the comedication.

Covariate effects were added to the model in a multiplicative manner. Continuous covariates were normalized to a reference value and were included in the model with a power function. Power models were used to describe the relationship between continuous covariates and the typical value of pharmacokinetic parameters:

$${\theta }_{\mathit{TV}}={\theta }_{\mathit{REF}}\text{·}{\left(\frac{X_i}{X_{\mathit{REF}}}\right)}^{{\theta }_x}$$

where θ _REF and θ _x are the fixed‐effect parameters and X _REF is a reference value of the covariate X _i. Reference values represent the median of the covariate values in the population pharmacokinetic dataset.

Categorical covariates were tested with a multiplicative model to obtain the fractional difference of pharmacokinetic parameters between the tested categorical groups. The relationship between binary categorical covariates (X _i) and the typical value of pharmacokinetic parameters was modelled as:

$${\mathrm{\theta }}_{\mathit{TV}}={\mathrm{\theta }}_{\mathit{REF}}*{\left({\mathrm{\theta }}_{\mathrm{x}}\right)}^{\mathrm{X}i}$$

where θ _REF and θ _x are fixed‐effect parameters and X _i is the indicator variable, which is equal to 1 or zero, dependent on the category of the covariates.

The influence of covariates was examined with the stepwise forward inclusion (P < 0.01) and backward elimination (P < 0.001) procedures, and the final model was selected using several predefined criteria. The model selection criteria included the likelihood ratio test (for nested models only), the Bayesian information criterion, improved goodness of fit and residual plots, physiological relevance of estimated parameters, increased precision of parameter estimates, and reduced variance of intersubject and residual errors. The likelihood ratio test was used to discriminate among alternative nested models. When comparing nested models, one additional model parameter, corresponding to one degree of freedom in the higher‐order model, was considered significant if it lowered the objective function value (OFV) by more than 6.63, corresponding to P < 0.01. For two degrees of freedom, the required reduction in OFV was 9.21. For the backward elimination steps of the covariate selection procedure (P < 0.001), the required changes in OFV were 10.83 and 13.82 for one and two degrees of freedom, respectively.

Model evaluation

The models were evaluated with goodness‐of‐fit plots, visual predictive checks and bootstrap methods. For visual predictive checks, 250 model‐simulated replicates of the observed data were generated using NONMEM. The simulated values were compared with the observed data by superimposing the median, 5th and 95th percentiles of the observed data on the median, 5th and 95th percentiles (along with their 90% confidence intervals) of the simulated data for each unique observation time bin. Additionally, 500 bootstrap replicates were constructed by randomly sampling (with replacement) N subjects from the original dataset, where N is the number of subjects in the original dataset. The parameter estimates of the final model were compared against the medians and 2.5th and 97.5th percentiles from the bootstrap results.

Inferences about the clinical importance of covariate effects were made based on the magnitude and precision of covariate parameter estimates. The magnitude of covariate effects was derived from the simulated exposures [steady‐state maximum plasma concentration (C_max,ss) and AUC at 24 hours at steady state (AUC_24,ss)], using the final population model parameters, and is presented in Forest plots. For all the DAAs, a change in exposure of 0.5–2.0‐fold from the overall population estimates was not anticipated to alter the efficacy or safety profile to an extent that would require dosage adjustments 31.

Results

Data sources

Paritaprevir, ombitasvir, dasabuvir and ritonavir assessments were based on about 17 000 plasma concentrations for each compound from 2348 subjects, yielding approximately seven concentrations per subject. The analysis of ribavirin population pharmacokinetics was based on about 13 500 plasma concentrations from 1841 subjects, yielding approximately seven concentrations per subject. The percentage of BLQ observations for any agent ranged from 0.5% to <2.5%; these observations were excluded from the analysis as they were expected to have minimal to no impact on the overall modelling results.

Subject demographic and baseline characteristics

Demographic characteristics are presented in Table 2 for subjects who received the 3D regimen with or without ribavirin, and 3D regimen plus ribavirin. The median age of subjects was 54 years; 42% were female, 2% were Asian, 7% were black and 6% were Hispanic/Latino. Slightly more than one‐third of subjects had previously received interferon‐based therapy. Compensated cirrhosis was present in 16% of subjects who received the 3D regimen with or without ribavirin, and 21% of subjects who received the 3D regimen plus ribavirin. HCV genotype 1a was present in 53% of subjects who received the 3D regimen with or without ribavirin, and 57% of subjects who received the 3D regimen plus ribavirin.

Table 2.

Patient demographic and clinical characteristicsa

Characteristic	DAA pharmacokinetic data (from 3D ± ribavirin regimens) (n = 2348)	Ribavirin pharmacokinetic data (from 3D + ribavirin regimens) (n = 1841)
Gender, n (%)
Male	1353 (58)	1081 (59)
Female	995 (42)	760 (41)
Grouped race, n (%)
Asian	38 (2)	32 (2)
Non‐Asian	2310 (98)	1809 (98)
Grouped race, n (%)
Black	153 (7)	109 (6)
Nonblack	2194 (93)	1732 (94)
Unknown	1 (0)	0
Ethnicity, n (%)
Hispanic/Latino	144 (6)	119 (6)
Other	2204 (94)	1722 (94)
Age, year	54.0 (18.0, 71.0)	54.0 (18.0, 71.0)
Weight, kg	76.0 (42.0, 129.0)	77.0 (42.0, 129.0)
BSA, kg m −2	1.9 (1.3, 2.6)	1.9 (1.3, 2.6)
BMI, kg m −2	26.0 (18.0, 39.4)	26.1 (18.0, 39.4)
Baseline CrCL, mg dl –1	104.0 (37.0, 281.4)	111.2 (37.0, 241.0)
Presence of compensated cirrhosis, n (%)
Yes	380 (16)	380 (21)
No	1968 (84)	1461 (79)
HCV treatment experience, n (%)
Naive	1548 (66)	1136 (62)
Experienced	800 (34)	705 (38)
HCV genotype, n (%)
1a	1253 (53)	1049 (57)
1b	1094 (47)	791 (43)
Unknown	1 (0)	1 (0)
Methadone/buprenorphine use, n (%)
Yes	38 (2)	38 (2)
No	2310 (98)	1803 (98)

3D, triple DAA drug; BMI, body mass index; BSA, body surface area; CrCL, creatinine clearance; DAA, direct‐acting antiviral agent; HCV, hepatitis C virus

Data are presented as median (range) unless otherwise indicated

Comedication analysis

Based on the potential comedication drug class or category selection analysis, only five drug classes (opioids, antipsychotic agents, antiepileptic drugs/anticonvulsants, antidiabetic agents and hormone replacement therapies) met the criteria (N ≥ 15, AUC_24,ss ratio: ≤0.5 or ≥2.0) for the paritaprevir step‐wise covariate model building. None of the drug categories (enzyme/transporter inducer/inhibitors) met the selection criteria for paritaprevir. None of the drug classes or categories met the criteria for ombitasvir, dasabuvir, ritonavir and ribavirin 30.

Ombitasvir population pharmacokinetics

A one‐compartment model with first‐order absorption and elimination and a combined residual error model characterized ombitasvir plasma concentration–time data, with good predictive performance (Figure 1). Cirrhosis, gender, age and body weight had significant covariate effects on CL/F, and age and body weight were significant covariates for V_c/F (Table 3). The mean ombitasvir CL/F was 453 l day⁻¹ (18.9 l h⁻¹), and mean V_c/F was 50.1 l (Table 3). Gender was the only covariate to affect ombitasvir exposures (C_max,ss and AUC_24,ss), which were 46–54% higher in female subjects than in male subjects. Changes in exposure for each 10‐year change in age, 10 kg change in body weight, or due to cirrhosis were ≤11% (Figure 2). The steps in developing the final population model for ombitasvir are shown in the Supporting Information; changes in OFV and covariate effects are detailed in Table S1.

Figure 1.

Visual predictive checks for population pharmacokinetic models. Filled circles represent observed median values (5th and 95th percentile error bars). The simulated median is represented by a solid line and the associated 90% confidence interval of the simulated median by blue shading. The simulated 5th and 95th percentiles are represented by dashed lines and the associated 90% confidence interval of the simulated 5th and 95th percentile by purple shading. The profile for ribavirin starts 2 weeks into the treatment period

Table 3.

Population pharmacokinetic parameter estimates and significant covariatesa

Model characteristic	Ombitasvir	Paritaprevir	Dasabuvir	Ritonavir	Ribavirin
Pharmacokinetic model	One compartment	One compartment with absorption lag time	Two compartment	One compartment	Two compartment
k a , day −1	1.08 (1.01, 1.14)	1.74 (fixed)	4.61 (3.99, 5.45)	2.32 (1.47, 2.77)	21.3 (18.7, 24.1)
ALAG, day	—	0.0400 (fixed)	—	—	—
CL/F, l day −1	453 (441, 467)	1580 (1450, 1710)	1150 (1100, 1200)	439 (369, 554)	427 (419, 436)
V c /F, l	50.1 (44.9, 55.8)	16.7 (11.8, 22.6)	110 (93.3, 133)	21.5 (6.85, 43.9)	1100 (983, 1230)
Q/F, l day −1	—	—	182 (111, 295)	—	877 (791, 977)
V p /F, l	—	—	286 (190, 408)	—	3230 (3070, 3380)
IIV of CL/F	0.143 (0.131, 0.155)	1.18 (1.10, 1.26)	0.263 (0.213, 0.299)	0.810 (0.679, 1.11)	0.062 (0.057, 0.067)
IIV of V c /F, V p /F	—	—	—	—	0.197 (0.171, 0.222)
RUV	0.107 (0.096, 0.122)b 2.4 × 10−5 (6 × 10−6, 4.1 × 10−5)c	1.14 (1.09, 1.20)a	0.260 (0.242, 0.283)b 4.00 × 10−3 ( 1.00 × 10− 3,7.00 × 10−3)c	0.533(0.465, 0.558)b 4 × 10−6 (2 × 10−7, 8 × 10−6)c	0.0170 (0.0140, 0.0190)b 0.0390 (0.0300, 0.0480)c
Significant covariates on CL/F	Cirrhosis, gender, age, body weight	Cirrhosis, gender, age, opioid use, antidiabetic agent use	Cirrhosis, gender, body weight, CrCL	Gender, CrCL, genotype	Cirrhosis, gender, CrCL
Significant covariates on Vc/F	Age, body weight	Age, body weight	Age, body weight	None	Gender

ALAG, absorption lag time; CL/F, apparent clearance; F, bioavailability; IIV, interindividual variability; k_a, first‐order absorption rate constant; Q/F, apparent intercompartmental clearance between the central and peripheral compartment; RUV, residual unexplained variability; V_c/F, apparent volume of central compartment; V_p/F, apparent volume of peripheral compartment

Data are presented as population estimate (bootstrap evaluation 95% confidence interval)

^aAdditive error on log‐transformed data RUV

^bProportional RUV

^cAdditive RUV

Figure 2.

Population pharmacokinetic evaluations. Model‐derived estimates of steady‐state maximum plasma concentration (C_max,ss) and area under the plasma concentration–time curve (AUC_24,ss) for covariates identified as significantly associated with apparent clearance or apparent volume parameters for paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin (600 mg twice daily). A ratio of 1.0 indicates similar values between the subgroup analysed and the reference subject characteristics of the respective population. Error bars represent the 95% confidence intervals. Body weight (BWT) groups were stratified as 66 kg and 86 kg (and compared with the median population value of 76 kg), and age was plotted for 44 years and 64 years (and compared with the median population value of 54 years). Creatinine clearance (CrCL) was plotted for 75 ml min⁻¹ and 115 ml min⁻¹ (and compared with the median value of 104 ml min⁻¹ from direct‐acting antiviral agent and ritonavir pharmacokinetic datasets, and 105 ml min⁻¹ from the ribavirin pharmacokinetic dataset). GT, genotype

Paritaprevir population pharmacokinetics

A one‐compartment model with linear absorption and elimination, a lag time in absorption, a model representing additive residual error on log‐transformed data, and IIV in CL/F adequately described paritaprevir plasma concentration–time data, with good predictive performance (Figure 1). Model‐derived population pharmacokinetic parameter estimates and significant covariates are shown in Table 3. Cirrhosis, gender, age, opioid use and antidiabetic agent use were significant covariates for CL/F, and body weight and age were significant covariates for V_c/F; use of antiepileptic drugs/anticonvulsants and hormone replacement therapies were not significant. The mean paritaprevir CL/F in the final model was 1580 l day⁻¹ (65.9 l h⁻¹), and the population mean V_c/F was 16.7 l. Changes in predicted paritaprevir steady‐state exposures (C_max,ss and AUC_24,ss) were 92–96% higher for female subjects and 122–140% higher for subjects with cirrhosis; use of opioid medications and antidiabetic agents was associated with more modest increases of 48–56% and 40–46%, respectively, whereas a 10‐year change in age resulted in a minimal effect of <20%, and a 10 kg change in body weight resulted in no changes in exposure (Figure 2). The steps in developing the final population model for paritaprevir are shown in the Supporting Information; changes in OFV and covariate effects are detailed in Table S2.

Ritonavir population pharmacokinetics

A one‐compartment model with linear absorption and elimination and a combined residual error model adequately described ritonavir plasma concentration–time data, with good predictive performance (Figure 1). Gender, creatinine clearance and HCV genotype were significant covariates of CL/F (Table 3). The population mean ritonavir CL/F was 439 l day⁻¹ (18.3 l h⁻¹), and the mean V_c/F was 21.5 l (Table 3). Ritonavir exposures (C_max,ss and AUC_24,ss) were predicted to be 28–33% higher in subjects with HCV genotype 1a infection compared with genotype 1b infection and ≤15% higher in female subjects and subjects with mild renal impairment (creatinine clearance of 75 ml min⁻¹) compared to those with normal renal function (creatinine clearance of 104 ml min⁻¹) (Figure 2). The steps in developing the final population model for ritonavir are shown in the Supporting Information; changes in OFV and covariate effects are detailed in Table S3.

Dasabuvir population pharmacokinetics

A two‐compartment model with linear absorption and elimination and a combined residual error model characterized the dasabuvir plasma concentration–time data optimally, with good predictive performance (Figure 1). Significant covariates were cirrhosis, gender, creatinine clearance and body weight on CL/F, and age and body weight on V_c/F and V_p/F (Table 3). The population mean dasabuvir CL/F was 1150 l day⁻¹ (47.9 l h⁻¹), and the population means of V_c/F and V_p/F were 110 l and 286 l, respectively (Table 3). Based on covariate analysis, the exposures (C_max,ss and AUC_24,ss) were 29–39% higher in subjects with cirrhosis compared with non‐cirrhotic subjects, and 16–21% higher in female subjects than in male subjects. A 10‐year change in age, 10 kg change in body weight or the presence of mild renal impairment (creatinine clearance of 75 ml min⁻¹) was associated with <10% change in exposure to dasabuvir (Figure 2). The steps in developing the final population model for dasabuvir are shown in the Supporting Information; changes in OFV and covariate effects are detailed in Table S4.

Ribavirin population pharmacokinetics

A two‐compartment model with linear absorption and elimination, combined residual error and correlated IIV on CL/F and V_c/V_p adequately described ribavirin plasma concentration–time data, with good predictive performance (Figure 1). Significant covariate effects were observed for cirrhosis, gender and creatinine clearance on CL/F, and gender on V_c/F and V_p/F (Table 3). The population mean ribavirin CL/F was 427 l day⁻¹ (17.8 l h⁻¹), and mean V_c/F and V_p/F values were 1100 l and 3230 l, respectively (Table 3). Ribavirin exposures (C_max,ss and AUC_24,ss) were 31–29% higher in female subjects than in male subjects, <10% higher in subjects with mild renal impairment (creatinine clearance of 75 ml min⁻¹) compared to those with normal renal function (creatinine clearance of 105 ml min⁻¹) and similar in subjects with or without compensated cirrhosis. The steps in developing the final population model for ribavirin are shown in the Supporting Information; changes in OFV and covariate effects are detailed in Table S5.

For all agents, goodness‐of‐fit plots (not shown), visual predictive checks (Figure 1), and bootstrap analyses (Table 3) indicated that the data were well characterized by the pharmacokinetic models.

Discussion

The combination of paritaprevir/r, ombitasvir and dasabuvir with or without ribavirin is an established treatment option for patients with HCV genotype 1 infection. Phase III clinical trials showed that this 3D regimen resulted in high rates of sustained virological response across a wide spectrum of patients who are likely to be encountered in clinical practice, including those with challenging treatment characteristics. With such potentially broad clinical use, it is important to establish whether patient‐related variables influence the pharmacokinetics of the 3D regimen or adjunctive ribavirin, necessitating dose adjustments to maintain efficacy and safety. To address this issue, we used the large pharmacokinetic dataset from the broad phase III development programme, to generate population pharmacokinetic models for paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin. Our analysis showed that the population pharmacokinetic models for each agent adequately described the observed plasma concentration–time data, with precise model parameter estimates and good predictive performance.

Although female gender was associated with about 100% and 50% higher paritaprevir and ombitasvir exposures, respectively, these increases were not considered clinically meaningful. The higher exposures of paritaprevir and ombitasvir in women could be due to inherent differences in the pharmacokinetics of DAAs between men and women, potentially arising from hormonal and other physiological influences on complex metabolic transporter pathways. The phase II and phase III studies overall had 42% representation of female subjects and thus the safety and efficacy were well characterized in female subjects at the higher exposures.

The presence of compensated cirrhosis increased paritaprevir exposures by up to 140%, and dasabuvir exposures by up to 39%. These changes in exposure do not adversely affect the clinical efficacy and tolerability of the 3D regimen in patients with compensated cirrhosis, as shown by high sustained virological response rates and low discontinuation rates in the TURQUOISE‐II study, which enrolled treatment‐naive and treatment‐experienced patients with compensated cirrhosis (Child Pugh stage A) 21. The clinical trials used in the present analysis enrolled subjects with no cirrhosis or with compensated cirrhosis (Child Pugh stage A) only; therefore, the influence of more severe hepatic impairment on 3D pharmacokinetics could not be assessed. The present analysis indicated that the 3D regimen can be administered without dosage adjustment in patients with mild hepatic impairment. The regimen is contraindicated in patients with moderate or severe hepatic impairment (Child Pugh stages B or C) 13.

HCV genotype appeared to influence the pharmacokinetics of ritonavir, resulting in up to 33% higher exposure in subjects with HCV genotype 1a infection than in those with HCV genotype 1b infection. As there is no clear physiological basis for any influence of HCV subtype on ritonavir pharmacokinetics, this is likely to be a chance finding based on random variation and is not considered clinically meaningful.

Potential comedication drug classes did not appreciably alter the pharmacokinetics of the 3D regimen. Paritaprevir was the only medication whose pharmacokinetics were influenced by medications, showing an approximately 50% increase in exposure with opioids or antidiabetic agents, which does not necessitate paritaprevir dose adjustments. A specific mechanism for the interaction is difficult to ascertain because of the variety of opioids and antidiabetic agents with different mechanisms of metabolism/disposition that were included in the analysis datasets. However, given the complexity of the metabolic pathways (enzymes and transporters), inhibition of CYP3A or OATP1B1/1B3 by opioids or antidiabetic agents could be a possible reason for the interaction. No inducer or inhibitors of metabolic enzymes or transporters exerted any significant effect on the pharmacokinetics of the DAAs 30.

Although a number of covariates had significant effects on clearance and volume parameters for ombitasvir, dasabuvir and ritonavir, changes in exposures for these agents in the various subpopulations were modest, and hence no dose modifications are required in these groups.

Ribavirin enhances the antiviral response in patients with HCV genotype 1a infection when added to the 3D regimen 19, and is currently included in the recommended treatment regimen for HCV genotype 1a infection 13. This analysis showed that ribavirin coadministration did not affect the pharmacokinetics of any of the DAAs in HCV genotype 1‐infected subjects, as ribavirin was not identified as a covariate in any of the analyses. Ribavirin values of clearance and volume parameters estimated from the population pharmacokinetic analysis were in close agreement with those reported in the population pharmacokinetic analysis of phase Ib and II studies of the 3D regimen 25 and in the literature 32, 33, suggesting that the pharmacokinetics of ribavirin were not altered when coadministered with the 3D regimen. The weight‐based ribavirin dosing used in the clinical trials that contributed to the present analysis provided consistent exposures across the various covariates. Although gender was a covariate for both ribavirin clearance and volume parameters in the current model, steady‐state exposure in female subjects was only about 30% higher than in male subjects. The influence of gender on ribavirin plasma parameters was reported in the phase Ib and II population pharmacokinetic analysis of the 3D regimen 25, and in population pharmacokinetic studies of peginterferon‐based regimens 33, 34 but not in other studies 32, 35. Ribavirin dose adjustment does not appear to be necessary when given with the 3D regimen; however, certain severe adverse reactions or laboratory abnormalities may require dose modifications, and dose reduction is advised for patients with renal impairment 36.

Most patient variables that emerged as significant covariates for clearance or volume parameters showed relatively modest changes in steady‐state exposure. In general, a change in exposure of 0.5–2.0‐fold is not expected to alter the efficacy or safety of 3D therapy 31, 37. The majority of covariates increased predicted drug exposures by 30% or less and were not expected to affect the safety profile or efficacy to an extent that would require an adjustment in doses of the 3D regimen or ribavirin.

A limitation of the present study was that the patient population was determined by the inclusion and exclusion criteria as delineated by trial protocols, hence any additional factors influencing 3D pharmacokinetics were not explored. Also, the comedication analysis is subject to limitations, related particularly to the timing and duration of use. Subjects were taking multiple comedications that could have had opposite effects, thereby masking the true interaction of a certain class of medication or pathway of disposition. Formal drug interaction studies have provided more robust information on the influence of comedications on the pharmacokinetics of the 3D regimen 24, 37, 38, 39, 40, 41, 42, 43.

In conclusion, population pharmacokinetic models were developed successfully for paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin based on phase II and III data from subjects with HCV genotype 1 infection treated with the approved 3D regimen. Each drug model described the plasma concentration–time data accurately, with precise and robust model parameter estimates and good predictive performance. Based on model‐predicted steady‐state drug exposures, no clinical or demographic patient characteristics were identified that would require a dose adjustment for the 3D regimen.

Competing Interests

All authors have completed the Unified Competing Interest form at http://www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare S. M., D. E., S. S., A. R. P., A. K., T. J. P., W. M. A., R. M. M. and S. D., are (or were) employees of AbbVie Inc. at the time the study was performed and hold stock/shares with AbbVie Inc. SS is now employed by Gilead Sciences, Inc., Foster City, CA, USA, and may hold stock/shares with Gilead Sciences, Inc.

The authors thank AbbVie employees Zhongqing He for building the datasets for the analyses and Natalie Hycner for writing support. The authors also thank Lamara D. Shrode PhD, CMPP, of The JB Ashtin Group, Inc. and Meher M. Dustoor PhD, under contract with JB Ashtin, for assistance (writing, technical editing and proofreading) in preparing this manuscript for publication on behalf of AbbVie Inc.

Contributors

This work was supported by AbbVie Inc. AbbVie contributed to the study design, research and interpretation of data, and the writing, review and approval of the manuscript for publication. The manuscript was prepared according to the International Society for Medical Publication Professionals' ‘Good Publication Practice for Communicating Company‐Sponsored Medical Research (GPP3)’. The authors have met all the requirements for full authorship and have reviewed and modified the manuscript in detail at every stage of development. The authors received no compensation for this work. All authors agree to be accountable for all aspects of the work, ensuring the accuracy and integrity of the publication. SM and DE contributed equally to this work. A.K., T.J.P., W.M.A., R.M.M. and S.D. were responsible for original project conception and study design. S.S., A.R.P., A.K. and R.M.M. were responsible for data acquisition. S.M., D.E., S.S., A.R.P., A.K., T.J.P., W.M.A., R.M.M. and S.D. were responsible for statistical analysis, data interpretation, review and critique of the manuscript throughout the editorial process, and approval of the final manuscript draft submitted for publication.

Supporting information

Table S1 P values for each covariate for ombitasvir

Table S2 P values for each covariate for paritaprevir

Table S3 P values for each covariate for ritonavir

Table S4 P values for each covariate for dasabuvir

Table S5 P values for each covariate for ribavirin

Supporting info item

Mensing, S. , Eckert, D. , Sharma, S. , Polepally, A. R. , Khatri, A. , Podsadecki, T. J. , Awni, W. M. , Menon, R. M. , and Dutta, S. (2017) Population pharmacokinetics of paritaprevir, ombitasvir, dasabuvir, ritonavir and ribavirin in hepatitis C virus genotype 1 infection: analysis of six phase III trials. Br J Clin Pharmacol, 83: 527–539. doi: 10.1111/bcp.13138.