Clin Pharmacokinet https://doi.org/10.1007/s40262-018-0629-6 ORIGINAL RESEARCH ARTICLE Population Pharmacokinetics of Elagolix in Healthy Women and Women with Endometriosis Insa Winzenborg1 • Ahmed Nader1 • Akshanth R. Polepally1 • Mohan Liu1 • Jacob Degner1 • Cheri E. Klein1 • Nael M. Mostafa1 • Peter Noertersheuser1 • Juki Ng1  Springer International Publishing AG, part of Springer Nature 2018 Abstract Introduction Elagolix is a novel, orally active, non-peptide, competitive gonadotropin-releasing hormone (GnRH) receptor antagonist in development for the management of endometriosis with associated pain and heavy menstrual bleeding due to uterine fibroids. The pharmacokinetics of elagolix have been well-characterized in phase I studies; however, elagolix population pharmacokinetics have not been previously reported. Therefore, a robust model was developed to describe elagolix population pharmacokinetics and to evaluate factors affecting elagolix pharmacokinetic parameters. Methods The data from nine clinical studies (a total of 1624 women) were included in the analysis: five phase I studies in healthy, premenopausal women and four phase III studies in premenopausal women with endometriosis. Results Elagolix population pharmacokinetics were best described by a two-compartment model with a lag time in absorption. Of the 15 covariates tested for effect on elagolix apparent clearance (CL/F) and/or volume of distribution only one covariate, organic anion transporting polypeptide (OATP) 1B1 genotype status, had a Insa Winzenborg and Ahmed Nader contributed equally to the pharmacokinetic model design and execution as well as to the writing of this manuscript. Electronic supplementary material The online version of this article (https://doi.org/10.1007/s40262-018-0629-6) contains supplementary material, which is available to authorized users. & Ahmed Nader ahmed.nader@abbvie.com 1 Clinical Pharmacology and Pharmacometrics, Department R4PK, AbbVie Inc., Building AP31-3, 1 North Waukegan Road, North Chicago, IL 60064, USA statistically significant, but not clinically meaningful, effect on elagolix CL/F. Conclusion Elagolix pharmacokinetics were not affected by patient demographics and were similar between healthy women and women with endometriosis. Clinical Trial Registration Numbers NCT01403038, NCT01620528, NCT01760954, NCT01931670, NCT02143713. Key Points Elagolix population pharmacokinetics in 313 healthy women and 1311 women with endometriosis were best described by a two-compartment model with a lag time in absorption. Organic anion transporting polypeptide (OATP) 1B1 genotype status was the only covariate affecting elagolix apparent clearance, but did not result in clinically meaningful changes in elagolix plasma exposure. Elagolix pharmacokinetics were similar between healthy women and women with endometriosis. 1 Introduction Endometriosis, which affects approximately 6–10% of women of reproductive age [2], is characterized by the presence of endometrial-like tissue outside the uterus, which provokes a chronic, inflammatory condition [1]. I. Winzenborg et al. Endometriosis primarily affects women of childbearing age and is frequently associated with a range of symptoms including dysmenorrhea, non-menstrual pelvic pain, as well as dyspareunia [3]. Current first-line therapies for endometriosis-related pain include pain relievers (nonsteroidal anti-inflammatory drugs [NSAIDs]) and oral contraceptives. Second-line therapies involve injectable depot formulations of gonadotropin-releasing hormone (GnRH) agonists, such as leuprolide acetate. Medical treatment options remain limited and there is a significant unmet medical need for non-surgical treatments. Elagolix is a novel, orally active, non-peptide, competitive GnRH receptor antagonist in development for the management of endometriosis with associated pain and heavy menstrual bleeding due to uterine fibroids. Elagolix offers patients many benefits as it has a rapid onset of action, is orally bioavailable, does not produce a flare effect, and can be readily discontinued if necessary [4, 5]. In addition, elagolix has the ability to reduce gonadotropins (luteinizing hormone [LH], and follicle-stimulating hormone [FSH]) and ovarian (estradiol) hormones in a dosedependent manner, raising the possibility of tailoring the dose to achieve an acceptable balance between therapeutic efficacy and unwanted adverse effects [6]. Data from phase II and phase III trials have demonstrated efficacy and safety of elagolix as a therapy for endometriosis with associated pain [7–9]. The pharmacokinetic profile of elagolix has been wellcharacterized in a number of phase I studies. Following oral dosing, elagolix is rapidly absorbed reaching maximum concentrations at 1.0–1.5 h with a half-life of 4–6 h [6]. Elagolix doses of 150 mg once daily or 200 mg twice daily were tested in phase III studies for the treatment of endometriosis with associated pain. Elagolix population pharmacokinetics have not been previously reported. The elagolix clinical development program has yielded the largest dataset in women with endometriosis, allowing for development of a robust population pharmacokinetic model. The aim of this analysis was to develop a model to describe elagolix population pharmacokinetics and to evaluate factors affecting elagolix exposures in order to better assess key elagolix efficacy and safety outcomes measures. 2 Methods 2.1 Clinical Studies Included in the Analysis The data from nine clinical studies (five phase I and four phase III studies) were included in this population pharmacokinetic analysis. The phase I studies enrolled healthy, premenopausal women and the phase III studies enrolled premenopausal women with moderate-to-severe endometriosis-associated pain. Study design information for the phase I studies and treatment regimens for all nine studies are described in Table 1. Subjects in the phase I studies had a history of regular menstrual cycles, with a duration of 24–32 days with at least 3 and no more than 7 days of bleeding per month for a minimum of 3 months prior to study drug administration. Study design information for the phase 3 studies, including inclusion and exclusion criteria, was previously reported [8]. Elagolix dosing regimens for the studies in this analysis included single doses of 150 or 200 mg and multiple doses ranging from 100 mg once daily to 400 mg twice daily; dosing duration for multiple doses ranged from 21 days to 12 months [6, 8, 10–12]. All study participants provided written informed consent. Study protocols were approved by the institutional review boards or ethics committees, and study procedures were conducted in accordance with International Council for Harmonization Good Clinical Practice guidelines and ethical principles that have their origin in the Declaration of Helsinki. 2.2 Study Drug Dosing and Pharmacokinetic Sampling For Studies 1, 2, 3 and 4 (Table 1), doses were administered on-site; therefore, dosing times were based on observed study drug administration. Doses for Study 5 were self-administered by the study subjects and dosing times for that study were based on daily diaries obtained from individual subjects. The phase III studies (women with endometriosis) employed electronic compliance packaging to dispense study drug to study patients. Each patient received a monthly kit which contained weekly blister packs for dosing. Patients returned the compliance kits during their monthly visits; the kits were sent to Information Mediary Corporation (IMC, Ottawa, ON, Canada) where they were scanned to generate study- and patient-specific compliance reports. This method provided a detailed profile of patients’ dosing behavior and exact dosing times for the majority of doses in the population pharmacokinetic analysis. Blood sample collection times varied according to the protocol specifications for each study; sampling times are detailed in Table 1. The actual blood sample collection times were used for the population pharmacokinetic analysis. 2.3 Bioanalysis Blood samples for assay of elagolix were collected in K2ethylenediaminetetraacetic acid (EDTA)-containing Study (Clinicaltrials.gov identifier) Study site N IRB Study design Elagolix treatment Pharmacokinetic sampling times Bioanalysis informationa Duration Regimen 150 mg qd, 100 mg bid, 200 mg bid, 300 mg bid, or 400 mg bid Day 1: 0, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, and 24 h after dosing; prior to dosing on Days 5, 7, 9, 15, 17, and 19; day 21: 0, 0.5, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, 24, 36 and 48 h after dosing Precision: B 11.4% 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, and 24 h after dosing in each period Precision: B 2.3% Phase I studies (healthy, premenopausal women) Study 1 [6] Study 2 Study 3 Study 4 Study 5 (NCT01403038) [10–12] 34 23 23 54 179 ACPRU, Waukegan, IL, USA Vista Health System, Waukegan, IL, USA Randomized, placebo-controlled, MAD, double-blind study to evaluate safety and PKs/PDs in healthy women 21 days ACPRU, Waukegan, IL, USA Vista Health System, Waukegan, IL, USA Randomized, open-label, singledose, crossover study to evaluate safety and PKs in healthy women Single doses 150 mg ACPRU, Grayslake, IL, USA Vista Health System, Waukegan, IL, USA Randomized, open-label, singledose, crossover study to evaluate safety and PKs in healthy women Single doses 200 mg PPD Development, LP, Austin, TX, USA Salus IRB, Austin, TX, USA Randomized, open-label, singledose, crossover study to evaluate safety and PKs in healthy women Single doses 20 sites in the USA 7 IRBs provided protocol review and approval Open-label, multiple-dose study to evaluate safety and PDs in healthy women 3 cycles (84 days) Accuracy: - 6.9 to 7.2% LLOQ: 0.126 ng/ mL Accuracy: - 3.8 to 2% LLOQ: 1.57 ng/mL 2 9 100 mg 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, and 24 h after dosing in each period Precision: B 11.3% Accuracy: - 1.6 to 8.5% LLOQ: 0.115 ng/ mL 2 9 100 mg 0, 0.25, 0.5, 0.75, 1, 1.5, 2, 3, 4, 6, 8, 10, 12, 16, 24, 30, and 36 h after dosing in each period Precision: B 8.1% Accuracy: - 3.6 to 4.7% LLOQ: 0.102 ng/ mL 100 mg qd, 150 mg qd, 200 mg qd, 100 mg bid, 200 mg bid, 300 mg bid One sample on Weeks 2 and 4 of Cycles 1, 2, and 3 during clinic visit Precision: B 12.3% Accuracy: - 2.4 to 7.4% LLOQ: 0.126 ng/ mL Elagolix Population Pharmacokinetics Table 1 Studies included in the population pharmacokinetic model Table 1 continued Study (Clinicaltrials.gov identifier) N Study site IRB Study design Elagolix treatment Duration Regimen 6 months 150 mg qd or 200 mg bid Pharmacokinetic sampling times Bioanalysis informationa Day 1 (approximately 1 h after dosing, for pivotal studies only) and at Months 1, 2, 3, 4, 5, and 6 (or D/C) clinic visits Precision: B 374% Phase III studies (premenopausal women with moderate-to-severe endometriosis-associated pain) Elaris EM-I (NCT01620528) [8] Elaris EM-I-ext (NCT01760954) [8] 1311 Pivotal studies with extensions Multicenter, double-blind, randomized, placebo-controlled, phase III trials of 6-month treatment with elagolix in women with moderate or severe endometriosis-associated pain Accuracy: - 7.3 to 5.8% LLOQ: 0.102 ng/ mL Precision: B 124% Accuracy: - 3.7 to 25% LLOQ: 0.102 ng/ mL Elaris EM-II (NCT01931670) [8] Elaris EM-II-ext (NCT02143713) [8] Precision: B 8.3% Accuracy: - 3.1 to 4.9% LLOQ: 0.102 ng/ mL Precision: B 9.0% Accuracy: 0.6–2.6% LLOQ: 0.995 ng/ mL ACPRU AbbVie Clinical Pharmacology Research Unit, bid twice daily, D/C premature discontinuation, IRB institutional review board, LLOQ lower limit of quantification, MAD multipleascending dose, N number of volunteers included in the population pharmacokinetic analysis, PD pharmacodynamic, PK pharmacokinetic, qd once daily a Precision is expressed as % coefficient of variation (%CV); accuracy is expressed as percentage bias I. Winzenborg et al. Elagolix Population Pharmacokinetics collection tubes. Sufficient blood was collected to provide 1.5–3 mL plasma from each sample. Immediately after collection, the blood samples were inverted several times to ensure good mixing of the blood and anticoagulant; samples were then placed in an ice bath. The blood samples for elagolix assay were centrifuged (1100–1600 rpm for 10 min) using a refrigerated centrifuge to separate the plasma within 2 h of collection. The plasma samples were transferred using plastic pipettes into labeled, screw-capped polypropylene tubes. The plasma samples were placed in the freezer within 2 h after collection and maintained at - 20 C or colder until shipped to the sponsor. The frozen plasma samples for elagolix assay were packed in dry ice sufficient to last during transport to the sponsor for analysis. The samples were processed and assayed using a validated salt-assisted protein precipitation extraction, high-performance liquid chromatography–tandem mass spectrometric (LC–MS/MS) method similar to the methods previously published (AbbVie Bioanalysis Lab, North Chicago, IL, USA) [6, 13]. Samples were analyzed by subject. The precision (coefficient of variation) for elagolix plasma concentration determination was B 11.4% and accuracy (expressed as percentage bias) ranged from - 6.6 to 7.2%. The lower limits of quantitation were between 0.1 and 1.0 ng/mL for the phase I studies and were approximately 0.1 ng/mL for the phase III studies. Precision, accuracy, and the lower limit of quantification (LLOQ) for each study are provided in Table 1. 2.4 Pharmacogenetic Testing Deoxyribonucleic acid was isolated from whole blood samples using Qiagen reagent kits (Qiagen Inc., Valencia, CA, USA) applied to an AutoGenprep 3000 automated DNA extraction instrument. Genotypes were determined for the following nucleotide polymorphisms in OATP1B1 for rs4149056 [14]. All genotypes were determined using the pyrosequencing detection method (Qiagen Inc., Valencia, CA, USA). SLCO1B1 genetic variant 521T[C(S5) genotype was assayed to classify subjects into one of three different OATP1B1 transporter status categories, extensive transporter (ET); intermediate transporter (IT); or poor transporter (PT), based on the following criteria: • • • Homozygous variant 521T[C(s5) ? PT Heterozygous for 521T[C(s5) ? IT Homozygous wild-type 521T[C(s5) ? ET. 2.5 Population Pharmacokinetic Model Development Population pharmacokinetic models were built using nonlinear mixed-effects modeling in NONMEM 7.3 (ICON Development Solutions, Hanover, MD, USA). Model building was conducted using the first-order conditional estimation method with g–e interaction (FOCEI) within NONMEM and included development of the appropriate structural model, models for inter-individual variability (IIV) and residual error, and the testing of potential covariate–parameter relationships. One-, two-, and threecompartment pharmacokinetic models with first-order absorption (with or without lag time for absorption) and elimination were evaluated and the model that best described the observed concentration–time data was selected. Covariate selection was conducted using forward inclusion and backward elimination procedures to identify statistically significant covariate–parameter relationships. Several criteria were used to evaluate improvement in model performance and to select the final model. The likelihood ratio test was used to discriminate among alternative nested models using the difference in objective function values (OFV) provided by NONMEM (p\0.01 for forward inclusion and p\0.001 for backward elimination in covariate testing). Model development was guided by goodness-of-fit plots, likelihood ratio tests, physiologic plausibility, and precision of parameter estimates, as well as previous knowledge of elagolix pharmacokinetics. The covariates investigated for influence on elagolix apparent clearance (CL/F) and apparent volume of central compartment distribution (Vc/F) included age, body weight, body mass index (BMI), race, ethnicity, albumin, bilirubin, aspartate aminotransferase (AST), alanine aminotransferase (ALT), tobacco use, alcohol use, organic anion transporting polypeptide (OATP) 1B1 genotype status, and study region (subjects in the USA versus those outside the USA). Serum creatinine and creatinine clearance (estimated using Cockcroft–Gault equation) were also tested for effect on CL/F. OATP1B1 genotype status was determined since elagolix is a substrate of OATP1B1 and, thus, its elimination through hepatic metabolism may be dependent on OATP1B1-mediated hepatic uptake [15]. Continuous covariates were entered in the model using either a power (Eq. 1) or a multiplicative function (Eq. 2), normalized or centered around the median value as shown in the following example equations: I. Winzenborg et al. h2 WTKGi ; TVCLi ¼ h1
median WTKG ð1Þ TVCLi ¼ h1  ð1 þ h2  ðALBi ð2Þ  median ALBÞÞ; where TVCLi is the typical value of elagolix CL/F for individual patient i, h1 is the typical value of CL/F for an individual with a median weight (WTKG) in Eq. 1, and an individual with a median baseline albumin (ALB) level in Eq. 2, and h2 is the covariate effect constant relating the fixed effect (e.g., WTKGi and ALBi) to the elagolix CL/F. Dichotomous covariates were entered in the model via a binary indicator variable in a multiplicative way. For example (Eq. 3): TVCLi ¼ h1
ð1 þ h2 ÞETHi ; ð3Þ where TVCLi is the typical value of elagolix CL/F for individual i, and ETHi is the individual’s covariate value (coded as either 0 or 1), such that if ETHi = 0 then TVCLi = h1 and if ETHi = 1 then TVCLi = h1
(1 ? h2). IIV was modeled using an exponential error model and residual variability was modeled using a combined additive and proportional error model. The final model individual parameter estimates and daily dosing records were used to estimate daily concentration–time profiles and to calculate individual elagolix pharmacokinetic exposures. Steady-state functionality was used for modeling in the case of missing dosing times, i.e., if kits were not returned, could not be read, or if patients extracted multiple doses from the kit at the same time. Missing dosing times for Study 5 were imputed based on the last valid dosing information. Since intensive pharmacokinetic sampling and closely monitored dosing times were not feasible in phase I Study 5 and the phase III studies, rules for outlier exclusion for the sparse data were created for the pharmacokinetic analysis based on concentrations observed in the phase I studies with intensive sampling. A data exclusion rule was developed using the distribution of intensive pharmacokinetic concentration–time profiles from 134 women in Studies 1–4 (phase I) and was used to remove highly unlikely measurements collected from studies with sparse sampling (Study 5 and the phase III studies). An analysis of variance (ANOVA) was used on data from the phase I studies with intensive sampling to identify variance that can be explained by time after last dose (TALD) and administered dose. The residual error of the ANOVA model was used to describe the deviations from the mean for each TALD and dose. The standard deviation of the residual error was then quantified by the root mean squared error (RMSE). This analysis used the log10-transformed observed concentrations, hence the residuals were approximately normally distributed (99.7% of data expected to be within a three-fold RMSE around the mean). A more conservative exclusion range of four-fold RMSE was considered as the plausible interval for determination of outlier data for Study 5 as well as phase III studies. Furthermore, a seven-fold RMSE around the mean was used for the initial absorption phase (until 0.75 h after last dose) in order to avoid exclusion of too many samples in the steep incline phase, where values outside the four-fold RMSE are more likely to occur if small deviations in individual subject absorption rates exist. Elagolix concentrations outside these intervals were classified as outlying measurements. Additional information regarding the data exclusion rule is provided in the Electronic Supplementary Material. 2.6 Model Evaluation Model evaluation included diagnostic goodness-of-fit plots, visual predictive checks (VPCs), and bootstrap evaluation. For VPCs, final model parameters were used to simulate 500 replicates of the predictions using the NONMEM $SIMULATION option. VPCs of the concentration–time profile were then created by superimposing the individual observed concentrations over relevant percentiles (median, 5th, and 95th) of the simulations. 95% prediction intervals around each of the simulation percentiles were also included to demonstrate variability in simulated concentrations. For phase I VPCs, a prediction-correction was performed since several dosing regimens are presented, as described by Bergstrand et al. [16]. Furthermore, in order to estimate confidence intervals of the model parameters, 1000 bootstrap replicates were constructed by randomly sampling N patients with replacement from the original dataset, where N is the number of patients in the original dataset. Model parameters from the bootstrap replicates that had successfully converged were summarized to yield medians and 2.5th and 97.5th percentiles. Final model parameters based on the original dataset were compared against the bootstrap results. 3 Results The population pharmacokinetic analysis was based on concentrations derived from 1624 women enrolled in five phase I and four phase III studies. A summary of the demographic characteristics for volunteers included in the analysis is presented in Table 2. Elagolix Population Pharmacokinetics Table 2 Baseline demographics and characteristics All study volunteers (N = 1624)a Demographic characteristic (units) Age (years) 32 (18–49) Body weight (kg) 74 (40–148) Body mass index (kg/m2) 27 (16–56) Serum creatinine (mg/dL) 0.7 (0.4–1.3) Creatinine clearance (mL/min) 133.3 (53.6–346.8) Aspartate aminotransferase (IU/L) 18.8 (9.0–275.0) Alanine aminotransferase (IU/L) 15.9 (4.0–367.0) Albumin (g/dL) 4.5 (3.6–5.4) Total bilirubin (mg/dL) 0.5 (0.1–1.9) Race Black 194 (12) White and others 1430 (88) Geographic region USA 1287 (79) Outside the USA OATP1B1 genotype status 337 (21) Extensive transporter (ET) 858 (53) Intermediate transporter (IT) 279 (17) Poor transporter (PT) 30 (2) Unknowna 457 (28) Values are given as mean (range) or n (%) OATP1B1 organic anion-transporting polypeptide 1B1 a For some volunteers, no pharmacogenetic sample was available for analysis 3.1 Data Disposition and Exclusions The population pharmacokinetic analysis included a total of 3655 plasma concentrations from 313 healthy, premenopausal women in phase I studies and a total of 8637 plasma concentrations from 1311 premenopausal women with endometriosis in phase III studies (including women who switched from placebo in the pivotal studies to elagolix treatment in the extension studies). For Studies 1–4 (phase I studies with intensive sampling), all available pharmacokinetic data were used in the calculation of the pharmacokinetic parameters; therefore, all subjects were included in the pharmacokinetic analysis. For Study 5 (phase I study with sparse sampling), seven subjects were excluded from the pharmacokinetic analysis because they had no measurable elagolix plasma concentrations above the LLOQ at any visit. For the phase III studies (sparse sampling), 127 patients were excluded due to missing sample times or incorrect sample identifiers and 72 patients were excluded because they did not have any measurable elagolix plasma concentrations above the LLOQ. Data below the LLOQ were imputed for the population pharmacokinetic analysis, as follows: the first individual plasma concentration below the LLOQ following the study drug administration was set to one-half the LLOQ (LLOQ/2); all subsequent observations below the LLOQ for that subject after the same dose were excluded from the analyses. A data exclusion rule was developed and was used to remove implausible measurements collected from studies with sparse sampling (Study 5 and the phase III studies) from further analysis. All available elagolix concentrations from the studies with sparse sampling were included in the population pharmacokinetic analysis, with the exception of samples that fulfilled at least one of the five criteria from the data exclusion rule (criteria are presented in the Electronic Supplementary Material). No phase I intensive pharmacokinetic data were excluded based on this data exclusion rule. In total, pharmacokinetic concentrations from 24 patients from Elaris Endometriosis (EM)-I and 41 patients from Elaris EM-II were completely excluded from the pharmacokinetic model development because all their concentration measurements were below the limit of quantification (BLQ). Two patients in Elaris EM-I and one patient in Elaris EM-II that had at least two non-BLQ pharmacokinetic samples less than 48 h after last dose were completely excluded from the pharmacokinetic analysis due to the data exclusion rule. Additional information regarding the data exclusion rule is provided in the Electronic Supplementary Material. I. Winzenborg et al. Table 3 Parameter estimates and covariate effects for elagolix based on final population pharmacokinetic model and bootstrap evaluation Parameter Final pharmacokinetic model Population estimate (SEE) Bootstrap evaluation % RSE a Median 2.5th and 97.5th percentiles Pharmacokinetic parameters CL/F (L/h) 118 (1.54) 1.30 118 (114, 121) Vc/F (L) 257 (4.46) 1.74 257 (246, 270) ka (1/h) 2.49 (0.076) 3.04 2.50 (2.21, 2.83) Q/F (L/h) 5.83 (0.247) 4.23 5.83 (4.46, 7.00) Vp/F (L) Lag-time (h) OATP1B1 = IT/PT on CL/F 59.6 (1.56) 2.62 0.202 (0.002) - 0.142 (0.020) 0.925 - 13.7 59.8 (53.5, 68.5) 0.203 - 0.142 (0.192, 0.215) (- 0.182, - 0.102) IIV and residual variability IIV on CL/F (%CV)b 0.166 (42.5) 5.12 0.167 (0.147, 0.191) IIV on Vc/F (%CV)b 0.231 (51.0) 6.36 0.233 (0.194, 0.279) Covariance of IIV on CL/F and Vc/F 0.133 (37.7) 7.18 IIV on residual error (% CV)b 0.131 (37.4) 6.46 0.128 (0.084, 0.166) Proportional error 0.228 (0.006) Additive error 0.149 (0.016) 2.79 10.6 0.228 (0.215, 0.240) 0.143 (0.072, 0.251) CL/F apparent clearance, IIV inter-individual variability, IT/PT intermediate transporter/poor transporter, ka first-order absorption rate constant, OATP1B1 organic anion-transporting polypeptide 1B1, Q/F apparent inter-compartmental clearance, SEE standard error of estimate, Vc/F apparent volume of central compartment distribution, Vp/F apparent volume of distribution in the peripheral compartment, % CV percentage coefficient of variation, % RSE percentage relative standard error a % RSE was estimated as: SEE/population estimate 9 100 b % CV = SQRT(exp(ETA) - 1) 9 100 3.2 Population Pharmacokinetic Model Results A two-compartment model with a lag-time in absorption, IIV on CL/F and Vc/F, and a combined residual error best described elagolix pharmacokinetics (Table 3). The IIV on residual error [ETA(3)] was introduced in order to account for individual differences in the accuracy of dosing or pharmacokinetic sampling times (Eq. 4): Y ¼ IPRED  ð1 þ EPSð1Þ  EXPðETAð3ÞÞÞ þ EPSð2Þ  EXPðETAð3ÞÞ; ð4Þ where IPRED is the individual-predicted concentration, EPS(1) is the proportional error term, and EPS(2) is the additive error term. The only covariate lowering the OFV significantly (44 units) was OATP1B1 genotype as a covariate on CL/ F (Fig. 1). Women with genotype status of IT or PT had 14% lower CL/F than those with ET (extended transporter) genotype status for OATP1B1. The addition of OATP1B1 as a covariate on elagolix CL/F resulted in a minor decrease in IIV on elagolix CL/F from 43.9 to 42.5%. None of the other covariates, including body weight and BMI, were significant in the final population pharmacokinetic model (Table 3). Evaluation of the relationship of post hoc CL/F and Vc/F estimates against healthy subjects or patients with endometriosis showed no clear trend for differences in pharmacokinetic parameters between the two populations (Fig. 2). Accordingly, elagolix pharmacokinetics was considered similar between healthy subjects and patients with endometriosis. Predicted elagolix exposures using the final pharmacokinetic model were determined for the 150 mg oncedaily and 200 mg twice-daily dosing regimens. The median (5–95% percentiles) predicted average plasma concentration (Cavg) and area under the plasma concentration–time curve from time zero to 24 h (AUC24) values were 43.6 (14.2–99.9) ng/mL and 1050 (340–2400) ng
h/mL, respectively, for the 150 mg once-daily dosing regimen and were 120 (38.2–262) ng/mL and 2890 (916–6300) ng
h/mL, respectively, for the 200 mg twice-daily dosing regimen. 3.3 Final Model Evaluation Results The goodness-of-fit for the final model was evaluated graphically and is presented in Fig. 3. Overall, the plot of predicted and observed concentrations indicated that the model adequately described the observations over the entire range of elagolix plasma concentrations. The conditional weighted residuals did not show any major trends Elagolix Population Pharmacokinetics Fig. 1 Organic anion-transporting polypeptide (OATP)–clearance covariate relationships. The box shows the IQR with a median line. Lower/upper whiskers extend to the lowest/highest value within 1.5 9 IQR. Data beyond the end of the whiskers are shown as filled circles. For the unknown transporter group, a single outlier (with clearance\ 20 L/h) was excluded from this figure in order to better represent the covariate relationship. ET extensive transporter, IQR interquartile range, IT/PT intermediate transporter/poor transporter (IT and PT groups were combined for this analysis), OATP1B1 organic anion-transporting polypeptide 1B1 genotype status Fig. 2 Covariate relationships in healthy subjects and endometriosis patients: a elagolix clearance; and b elagolix central volume of distribution. In the figure, the box shows the IQR with a median line. Lower/upper whiskers extend to the lowest/ highest value within 1.5 9 IQR. Data beyond the end of the whiskers are shown as filled circles. IQR interquartile range when plotted against sampling times or population predictions, indicating that the model was appropriately unbiased. Based on 500 simulations, the VPCs for the elagolix plasma concentration–time profiles showed that the model accurately describe the central tendency and variability of I. Winzenborg et al. Fig. 3 GOF plots for the final population pharmacokinetic model: a population-predicted concentrations versus observed concentrations; b individual-predicted concentrations versus observed concentrations; c CWRES versus time; and d CWRES versus population predictions. CWRES conditional weighted residuals, GOF goodness-of-fit the data for all studies as well as for phase I and III studies separately. For the 150 mg once-daily dose, some deviations occurred in describing the 5th percentile, possibly due to outlying measurements not captured by the implemented data exclusion rules (Electronic Supplementary Material). The VPCs for the final model are presented in Fig. 4. The final population pharmacokinetic model was used to estimate confidence intervals of the model parameters. A total of 967 of 1000 bootstrap replicates plus the original dataset converged successfully. The estimated pharmacokinetic parameter values based on the original dataset were in good agreement with the medians of the parameter values estimated from the bootstrap replicates and none of the bootstrap confidence intervals included zero (Table 3). women with endometriosis employed compliance packaging technology to monitor patient compliance to study drug administration. Such technology enabled a more accurate record of dosing times for inclusion in the population pharmacokinetic analysis. The inclusion of data from several phase I and III studies enabled development of a robust population pharmacokinetic model, with all structural pharmacokinetic parameters estimated with good precision (RSE: 1–4%). Comparison of post hoc CL/F and Vc/F estimates between healthy subjects and women with endometriosis demonstrated similarity between these two populations. Although the comparison was not formally tested during the covariate selection process (as it would be confounded by the differences in sampling frequency between phase I and III studies), the observed similarity in pharmacokinetic parameter estimates indicate that elagolix pharmacokinetics and exposures are comparable in healthy women and women with endometriosis. Of the 15 covariates investigated for influence on CL/ F or Vc/F, OATP1B1 genotype status was the only covariate with a statistically significant effect on elagolix CL/F. Women with OATP1B1 genotype status of PT or IT 4 Discussion A population pharmacokinetic model was developed based on the largest pharmacokinetic dataset for elagolix, to date, and included substantial representation of healthy premenopausal women and premenopausal women with endometriosis. The phase III clinical trials for elagolix in Elagolix Population Pharmacokinetics b Fig. 4 Visual predictive checks for the final elagolix population pharmacokinetic model: a phase III, 150 mg qd; b phase III, 200 mg bid; c phase I, prediction-corrected. bid twice daily, qd once daily, TSLD time since last dose, VPC visual predictive checks had an elagolix CL/F 14% lower than that for women with OATP1B1 genotype status of ET. This effect is expected given that elagolix elimination through hepatic metabolism may be dependent on OATP1B1 transporter-mediated hepatic uptake [15]. However, the minor difference in elagolix CL/F shown in the model is not considered to be clinically meaningful. None of the other tested covariates (including body size, race/ethnicity, hepatic function, or renal function) were significantly associated with elagolix pharmacokinetic parameters and, hence, would not result in different elagolix dosing recommendations for patients with varying demographic characteristics. A challenge that occurs when modeling phase III pharmacokinetic data for drugs with a short half-life and frequent dosing is that small deviations in dosing or sampling times can lead to concentration measurements that are beyond the expected concentration range for the TALD. Attempts to model population pharmacokinetics while including all pharmacokinetic data resulted in minimization termination during the first NONMEM iteration. Therefore, a data exclusion rule was applied in order to remove concentration data that are not physiologically plausible based on the observed phase I data in 134 women. This approach needs to be implemented with care, as it is not suited for drugs where differences in pharmacokinetic exposures between healthy subjects and patients are expected. In the case of elagolix, the percentages of sparse samples removed by this rule for healthy women in Study 5 were in the same range as those for phase III studies in women with endometriosis. Limitations of this analysis included the constraints associated with utilizing sparse sampling data and outpatient dosing to develop population pharmacokinetic models. These challenges were addressed by using a substantial intensive pharmacokinetic dataset along with the data exclusion rule. 5 Conclusions Overall, elagolix population pharmacokinetics are adequately described by a two-compartment model with a lag time in absorption. The model demonstrated that elagolix pharmacokinetic parameters are similar between healthy women and women with endometriosis. OATP1B1 genotype status was the only statistically significant covariate on elagolix CL/F. However, OATP1B1 genotype status did I. Winzenborg et al. not have a clinically meaningful impact on elagolix exposure. None of the other tested covariates, including body weight and BMI, were significantly associated with elagolix pharmacokinetic parameters. The final population pharmacokinetic model will be utilized for exposure–response models for key elagolix efficacy and safety outcome measures. Acknowledgements We thank AbbVie employee Sonja Kemmis Causemaker for medical writing support for this manuscript. We also thank AbbVie employee Xiaohua Du for dataset programming efforts. 6. 7. 8. 9. Author contributions All authors contributed to the study design and analysis and interpretation of the data. All authors participated in the drafting and revising of the manuscript. 10. Compliance with Ethical Standards Funding AbbVie provided financial support for the study and participated in the design, study conduct, analysis, and interpretation of data as well as the writing, review, and approval of the manuscript. 11. Conflict of interest Insa Winzenborg, Ahmed Nader, Akshanth R. Polepally, Mohan Liu, Jacob Degner, Cheri E. Klein, Nael M. Mostafa, Peter Noertersheuser, and Juki Ng are employees of AbbVie, Inc. and may hold stock or stock options. 12. References 1. Kennedy S, Bergqvist A, Chapron C, et al. ESHRE guideline for the diagnosis and treatment of endometriosis. Hum Reprod. 2005;20(10):2698–704. 2. Viganò P, Parazzini F, Somigliana E, Vercellini P. Endometriosis: epidemiology and aetiological factors. Best Pract Res Clin Obstet Gynaecol. 2004;18(2):177–200. 3. Facchin F, Barbara G, Saita E, et al. Impact of endometriosis on quality of life and mental health: pelvic pain makes the difference. J Psychosom Obstet Gynaecol. 2015;36(4):135–41. 4. Struthers RS, Chen T, Campbell B, et al. Suppression of serum luteinizing hormone in postmenopausal women by an orally administered nonpeptide antagonist of the gonadotropin-releasing hormone receptor (NBI-42902). J Clin Endocrinol Metab. 2006;91(10):3903–7. 5. Struthers RS, Nicholls AJ, Grundy J, et al. Suppression of gonadotropins and estradiol in premenopausal women by oral 13. 14. 15. 16. administration of the nonpeptide gonadotropin-releasing hormone antagonist elagolix. J Clin Endocrinol Metab. 2009;94(2):545–51. Ng J, Chwalisz K, Carter DC, Klein CE. Dose-dependent suppression of gonadotropins and ovarian hormones by elagolix in healthy premenopausal women. J Clin Endocrinol Metab. 2017;102(5):1683–91. Carr B, Dmowski WP, O’Brien C, et al. Elagolix, an oral GnRH antagonist, versus subcutaneous depot medroxyprogesterone acetate for the treatment of endometriosis: effects on bone mineral density. Reprod Sci. 2014;21(11):1341–51. Taylor HS, Giudice LC, Lessey BA, et al. Treatment of endometriosis-associated pain with elagolix, an oral GnRH antagonist. N Engl J Med. 2017;377(1):28–40. Surrey E, Taylor HS, Giudice LC, et al. Long-term safety and efficacy of elagolix treatment in women with endometriosis-associated pain: primary results from two phase 3 extension studies [abstract no. O-233]. Fertil Steril. 2017;108(3 Suppl):e95–6. Archer D, Ng J, Chiu Y-L, Klein C, Chwalisz K. Dose-dependent suppression of ovulation and ovarian activity by elagolix in healthy premenopausal women. Society for Reproductive Investigation (SRI) 64th annual meeting, 15–18 Mar 2017, Orlando. Ng J, Chiu Y-L, Chwalisz K, Klein C. Effects of elagolix, an oral gonadotropin-releasing hormone (GnRH), on gonadotropins and ovarian sex hormones in healthy premenopausal women [poster no. PS02-003]. World Congress on Endometriosis (WCE), 17–20 May 2017, Vancouver. Ng J, Archer D, Chiu Y-L, Klein C, Chwalisz K. Effects of elagolix, an oral gonadotropin-releasing hormone (GnRH), on ovulation, ovarian activity, and ovarian reserve in healthy premenopausal women [poster no. PS02-004]. World Congress on Endometriosis (WCE), 17–20 May 2017, Vancouver. Chen C, Wu D, Guo Z, et al. Discovery of sodium r-(?)-4-{2-[5(2-fluoro-3-methoxyphenyl)-3-(2-fluoro-6-[trifluoromethyl]benzyl)-4-methyl-2,6-dioxo-3,6-dihydro-2h-pyrimidin-1-yl]-1phenylethylamino}butyrate (elagolix), a potent and orally available nonpeptide antagonist of the human gonadotropin-releasing hormone receptor. J Med Chem. 2008;51(23):7478–85. Hagenbuch B, Meier PJ. Organic anion transporting polypeptides of the OATP/SLC21 family: Phylogenetic classification as OATP/SLCO superfamily, new nomenclature and molecular/functional properties. Pflugers Arch. 2004;447(5):653–65. Ng J, Salem AH, Carter D, Williams LA, Klein CE. Effects of the coadministration of single and multiple doses of rifampin on the pharmacokinetics and safety of elagolix in healthy premenopausal females [poster no. PI-089]. American Society for Clinical Pharmacology and Therapeutics (ASCPT), 8–12 Mar 2016, San Diego. Bergstrand M, Hooker AC, Wallin JE, Karlsson MO. Predictioncorrected visual predictive checks for diagnosing nonlinear mixed-effects models. AAPS J. 2011;13(2):143–51.