Population Pharmacokinetics of Total and Free Erdafitinib in Adult Healthy Volunteers and Cancer Patients: Analysis of Phase 1 and Phase 2 Studies

The Journal of Clinical Pharmacology 2019, 0(0) 1–13
ⓍC 2019, The American College of
Clinical Pharmacology DOI: 10.1002/jcph.1547
Anne-Gaelle Dosne, PhD1, Elodie Valade, PhD1, Kim Stuyckens, MSc1, Lilian Y. Li, PhD2, Daniele Ouellet, PhD2, and Juan Jose Perez-Ruixo, PhD1

Abstract
A population pharmacokinetic (PK) model was developed using data pooled from 6 clinical studies (3 in healthy volunteers and 3 in cancer patients) to characterize total and free plasma concentrations of erdafitinib following single- and multiple-dose administration, to understand clinically relevant covariates, and to quantify the inter- and intraindividual variability in erdafitinib PK. An open, linear, 3-compartment disposition model with first-order absorption and a lag time was used to describe the PK profile of total and free erdafitinib plasma concentrations. The PK of erdafitinib were linear and time independent. After oral administration, erdafitinib was rapidly absorbed, with a time to maximum concentration between 2 and 4 hours. In patients, erdafitinib total apparent oral clearance was 0.200 L/h (median free fraction 0.24%), and the effective terminal half-life of total drug was 76.4 hours. Interindividual variability in PK parameters was moderate for oral clearance and central volume of distribution, and large for absorption rate and peripheral volume of distribution. Sex and renal function were significant covariates on free oral clearance, while weight, sex, and α1-acid-glycoprotein were significant on oral central volume of distribution. Age, race, and mild hepatic impairment were not significant covariates of erdafitinib exposure. Given that the magnitude of the covariate effects were within 25% of reference values and that the recommended dosing regimen of erdafitinib comprises individual dose up-titrations and reductions based on presence or absence of toxicities, the clinical relevance of the investigated covariates is expected to be limited, and no dose adjustments are warranted.

Keywords
α1-glycoprotein, covariate analysis, erdafitinib, fibroblast growth factor receptor, free fraction, plasma protein binding, population pharmacokinetics, tyrosine kinase inhibitor

Pharmacometrics

Erdafitinib, a potent oral selective pan–fibroblast growth factor receptor (FGFR) tyrosine kinase inhibitor,1 has demonstrated target inhibition and pathway modulation in FGFR pathway–activated cellular models of various cancers, including bladder cancer.1 In the first-in-human phase 1 study (EDI1001- [NCT01703481]), erdafitinib 9 mg once-daily and erdafitinib 10 mg on a 7-days-on/7-days-off schedule demonstrated a manageable safety profile and achieved exposures resulting in clinical response, as measured by relative change from baseline in target lesion size.2 Recently, in a multicenter, open-label, phase 2 study (BLC2001 [NCT02365597]), erdafitinib demonstrated efficacy and safety in adult patients with locally advanced or metastatic urothelial cancer, whose tumors harbor certain FGFR genetic alterations.3,4 Based on this evidence, erdafitinib was granted accelerated approval by the US Food and Drug Administration for the treatment of adult patients with locally advanced or metastatic urothelial carcinoma that has susceptible FGFR3 or FGFR2 genetic alterations and progressed during or following at least 1 line of prior platinum-containing chemotherapy including within 12 months of neoadjuvant or adjuvant platinum-containing chemotherapy.5 Erdafitinib dose is individually titrated based on serum phosphate concentrations, a target engagement biomarker that reflects the extent of erdafitinib FGFR inhibition, and is expected to be linked to efficacy and safety outcomes. Due to its mechanism of action, administration of erdafitinib results in an increase in serum phosphate concentrations.6 Specifically, the impairment of renal fibroblast growth factor 23/Klotho signaling through 1Global Clinical Pharmacology, Janssen Research & Development, Beerse, Belgium 2Global Clinical Pharmacology, Janssen Research & Development, Spring- house, Pennsylvania, USA Submitted for publication 2 July 2019; accepted 14 October 2019.

Corresponding Author:
Anne Gaelle Dosne, PhD, Janssen Research & Development, a Division of Janssen Pharmaceutica N.V. Turnhoutseweg 30, B-2340 Beerse, Belgium. Email: [email protected]

FGFR inhibition induces hyperphosphatemia via the deregulation of cytochrome P450 (CYP) 27B1 and CYP24A1.7 Hyperphosphatemia has been seen in nonclinical studies of selective FGFR small-molecule kinase inhibitors,7 as well as in clinical studies of erdafitinib,2 where phosphate levels typically increased following erdafitinib dosing and returned to baseline levels after drug interruption. Elevated phosphate levels also correlate with improved response rates.2,6 Erdafitinib dosing starts at 8 mg daily, with the possibility of up-titration to 9 mg daily if serum phosphate concentrations are <5.5 mg/dL 2 weeks after treatment initiation and there are no safety concerns.
Clinical pharmacokinetics (PK) of erdafitinib have been assessed previously in healthy volunteers and in patients with cancer. In the combined single-/multiple-ascending-dose study,8 the systemic exposure (maximum concentration [Cmax] and area under the concentration-time curve) increased with increasing dose with no consistent deviations from dose proportionality.9 Apparent terminal half - life for erdafitinib in cancer patients ranged from 42 to 74 hours. Based on total plasma concentration of erdafitinib, the apparent volume of distribution (26 L) and plasma clearance (0.26 L/h) across the doses were relatively low.2 Erdafitinib binds to α1- acid glycoprotein (AGP), with free fraction (fu) in the range of 0.250% to 0.506%.10 Erdafitinib fu varies between patients with cancer and healthy volunteers, and a complete understanding of erdafitinib exposure requires the development of a PK model that describes both free and total plasma concentrations.

Erdafitinib is cleared via multiple clearance path- ways. Based on totality of available in vitro and clinical PK data, a physiologically based PK model estimated the clearance of erdafitinib to be composed of metabolism by CYP2C9 (39%), CYP3A4 (20%), renal clearance (13%), intestinal secretion (21%), and additional hepatic clearance (8%).5 Bioavailability of erdafitinib was comparable under a fed or fasted state and among various oral formulations used throughout clinical development, including solution, capsule, and 2 oral tablet formulations.
A population PK analysis was conducted by pooling data from healthy volunteers and cancer patients to characterize total and free erdafitinib PK following single- and multiple-dose administration, to under- stand clinically relevant covariates, such as age, sex, race, weight, AGP, and renal and hepatic function, and to quantify the inter- and intraindividual variability in erdafitinib PK.

Methods
Patients and Study Design
All studies were conducted in accordance with the Declaration of Helsinki and in adherence with the Good Clinical Practice guidelines and were approved by the institutional review board for each study. Informed consents were obtained from all healthy volunteers and patients with cancer before the enrollment. A total of 6 clinical studies (3 in healthy volunteers and 3 in cancer patients) were included in the pooled data set for population PK. Details of study design, sample size, formulation, dosing regimen, and PK sam- pling schedule of these studies are described in Table 1. The 3 studies conducted in healthy volunteers were (1): EDI1002 (NCT02218073), a PK study to assess the relative bioavailability between the tablet and so- lution formulations; (2) EDI1003 (NCT02231489), a PK study to assess the relative bioavailability be- tween the tablet and capsule formulations; and (3) EDI1005 (NCT02692677), a mass balance study. In these healthy-volunteer studies, adult men or women (if women, they must be postmenopausal [no spontaneous menses for ?2 years] or surgically sterile) with body mass index between 18 and 30 kg/m2 who were non- smokers for ?6 months before screening were included.

Participants were excluded if they had a history of or current clinically significant medical illness, including but not limited to cardiac disease, hematologic disease, pulmonary disease, any clinically relevant laboratory abnormalities, and/or clinically significant abnormal physical examination, vital signs, or 12-lead electrocar- diogram at screening. The 3 studies conducted in cancer patients were (1) EDI1001 (NCT01703481), a first-in-human multiple- dose study conducted in 4 parts to evaluate the PK, pharmacodynamics, and safety of erdafitinib; iden- tify recommended phase 2 doses; and explore po- tential clinical activity of these doses; (2) GAC1001 (NCT01962532), a multiple-dose study to evaluate the safety, PK, pharmacodynamics, and clinical activity of erdafitinib in Asian patients9; and (3) BLC2001 (NCT02365597), a multiple-dose, open-label study con- ducted to determine the efficacy and the safety of 2 different dose regimens of erdafitinib in patients with metastatic or surgically unresectable urothelial cancers that harbor selected FGFR genomic aberrations.3 In the phase 2 study, erdafitinib dose was administered either daily or 7 days on/7 days off in 28-day cycles, with a pharmacodynamically guided up-titration after 2 to 4 weeks of erdafitinib treatment. The dose was increased if serum phosphate concentrations were below a given threshold. During study conduct, erdafitinib dose could also be reduced at any point based on toxicities. Overall,

Data and Bioanalytical Methods
Measurements included in the PK analysis included total concentrations; free concentrations; and AGP, albumin, and total plasma protein levels. Total concen- trations were measured at all PK time points specified in the study protocols. Free concentrations, AGP, al- bumin, and total plasma protein levels were measured only at time points where plasma protein binding (PPB) assessment was planned, typically 1 to 2 times per subject. Free fraction was derived from total and free concentration measurements, which coincides with PPB assessment time points. PK data for 1 subject then consisted of many total concentrations and few free concentrations. For each subject, fu, AGP, albumin, and total plasma protein levels were carried forward at each PK time point from their value at the last available PPB assessment time point.

All venous blood samples were collected in hep- arinized tubes and centrifuged, and separated plasma was stored at 20°C for analysis. Plasma concentra- tions of erdafitinib were quantified using nonchiral liquid chromatography–mass spectrometry assays de- veloped and validated in the bioanalytic laboratory of Janssen R&D, a division of Janssen Pharmaceutica in Beerse, Belgium (concentration range, 0.5-500 ng/mL). This method consisted of protein precipitation with acetonitrile after addition of stable isotope labeled internal standard (erdafitinib 13CD3). The resulting supernatant was diluted with water and then injected on a reversed phase high-density liquid chromatography column using a gradient method. Detection was done by tandem mass spectrometry in the multiple reaction monitoring mode with TurboIonSpray (Sciex, Fram- ingham, Massachusetts) ionization in the positive ion mode. A similar method was validated at PRA Health Sciences in Assen, The Netherlands (concentration
range, 1-2000 ng/mL). Both assays produced equivalent results as demonstrated by cross-validation using the same lot of quality control samples. The lower limit of quantification ranged from 0.5 to 1 ng/mL.

Aliquots from the PPB sample were used to deter- mine the fu via equilibrium dialysis. Predose (baseline, before any dose) PPB samples from the clinical studies were fortified with an erdafitinib stock solution before equilibrium dialysis, while postdose PPB samples were subject to equilibrium dialysis as is. The concentration of erdafitinib in buffer and plasma was determined using a qualified liquid chromatography–mass spec- trometry assay. All analytical batches were accepted based on the calibration curve and bioanalytical quality control acceptance criteria in line with current US Food and Drug Administration guidelines. Using a separate aliquot from the same sample, AGP, albumin, and total plasma protein levels were measured via a clinical chemistry analyzer.

Pharmacokinetic Analysis
Software. The population PK analysis was per- formed in accordance with appropriate guidelines,11–13 using nonlinear mixed-effects modeling with the first- order conditional estimation method as implemented in the NONMEM version 7.3.0 (ICON plc, Dublin, Ire- land). Compilations were achieved using the Fortran 64 Compiler Professional, version 11.1 (Intel, Santa Clara, California). Exploratory analysis, diagnostic plots, and postprocessing of NONMEM analysis results were carried out in R version 3.4.1.

Base Structural Model
Based on preliminary data exploration, the starting PK model describing the time profiles of erdafitinib total plasma concentrations (Ctot) consisted of an open, linear, 3-compartment mammillary model with first- order absorption and elimination. The population PK model was parameterized in terms of apparent clear- ance (CL/F [L/h]), apparent volumes of distribution (central V2/F and 2 peripheral volumes V3/F and V4/F [L]), apparent intercompartmental clearances (Q3/F and Q4/F [L/h]), lag time (tlag [hours]), and apparent first-order absorption rate constant (ka [h−1]). Both Ctot and free concentrations (Cfree) were simultaneously modeled using a single PK model, where Cfree were predicted from the modeled Ctot. That is, the model was describing both Ctot and Cfree at all PK time points, even if actual Cfree data were available only at PPB assessment time points (see Data and Bioanalytical Methods section for rationale). In the base model, Cfree was predicted based on modeled Ctot (equation 1) and the predicted fu. Specifically, fu was predicted based on measured AGP and the estimated binding constant (dissociation constant [Kd]) and was referred to as

Final population PK model for erdafitinib. Cfree, free concentrations; CLfree/F, apparent clearance of free erdafitinib; Ctot, total concentrations; fu, free fraction; ka, apparent first-order absorption rate constant; Q3,free/F and Q4,free/F, apparent intercompartmental clearances of free erdafitinib; V2,tot, apparent volume of distribution of the central compartment of the total drug; V3/F and V4/F, apparent peripheral volumes of distribution. “estimated” fu (f u,est) (equation 2). The actual measured fu from the PPB assessment was not used in the base model. C free = fu,est · Ctot (1) f u,est and fu was moderate to high (r 0.73). These improvements led to the reference structural model as presented in Figure 1 and equations 3 to 5. Compared to the base model, only free drug amounts (ie, total drug amounts multiplied by fu) could leave the central The basic assumptions of this model were (1) the erdafitinib binding to AGP is a nonlinear process with regard to AGP and a linear process with regards to erdafitinib concentration, and (2) there is only one binding site of erdafitinib to AGP.

Reference Structural Model
The base structural model implied that changes in erdafitinib binding to AGP, triggered by changes in
conduct, absorption parameters (ka and tlag) could vary among oral formulations (solution, capsule, and tablet) despite the fact that comparable bioavailability among them was demonstrated through noncompartmental analysis. The effect of formulation was included in the reference structural model as a covariate on the absorption parameters: tlag and first-order absorption rate.

In the selection of a preferable model, models that converged successfully, had a successful estimation of the standard errors, produced reasonable parame- ter estimates, and had low interindividual variability and correlations among random effects were preferred over others. The improvement in the fit obtained was assessed by examination of several diagnostics. The change in the minimum value of objective function (MVOF) was examined, with the addition of fixed effects being considered at an α confidence level of
.001. Goodness-of-fit plots displaying observations vs model predictions and model residuals needed to show the absence of trends. Prediction-corrected visual pre- dictive check14 was also used for model evaluation to provide a visual comparison between the distributions of simulated and observed erdafitinib Ctot and Cfree using the final population PK model and the observed data. Sampling importance resampling was used in
addition to the asymptotic variance-covariance matrix to estimate parameter uncertainty and compute the 95%CI of model parameters.14,15

Results
The final PK analysis data set included data from
373 participants receiving oral single doses (range, 10-12 mg), continuous once-daily doses (range, 0.5-12 mg), or intermittent (7 days on, 7 days off) once-daily doses (range, 10-12 mg) of erdafitinib. In total, 5060 total erdafitinib plasma concentrations and 650 free erdafitinib plasma concentrations were available after 41 total concentrations were excluded due to data below the limit of quantification (n 20), incomplete dosing/sample times (n 19), and graphical inspection (n 2 considered data errors: one 1000-fold higher than other concentrations at similar time and one not below the limit of quantification 3.5 months after the last dose, which is more than 3 times the maximum postdose follow-up time in the rest of the data set). Four free con- centrations were excluded due to data below the limit of quantification (n 2) and incomplete dosing/sample times (n 2). Total erdafitinib PK profiles after a single dose and over the 24-hour dosing interval at steady state are shown in Figure S1 after dose normalization to erdafitinib 8 mg, the recommended starting dose. Baseline characteristics of the study population are summarized in Table 2. Free fraction was missing for 13 participants. For these participants, fu was imputed using individual AGP for 7 participants and using AGP study median for 6 participants. Hepatic and renal impairment status was missing for 3 and 1 participants, respectively. These participants’ status was imputed to the most common status in the data set, that is, normal hepatic function and mild renal impairment, respectively. Mean (standard deviation) weight in the 3 weight categories were 52.2 (5.45) kg for <60 kg (n 93), 69.9 (5.78) kg for 60 to 80 kg (n 179) and 93.0 (11.0) kg for >80 kg (n 101). For age, it was 52.8 (9.18) years for <65 years (n 232), 69.1 (3.09) years for 65 to 75 years (n 105), and 80.2 (3.61) years for >75 years (n 36).

Conclusions
The plasma PK of total and free erdafitinib concentra- tion and its variability were adequately described using a population approach across different dose levels, regimens, and formulations administered to healthy volunteers and cancer patients. There was no evidence of concentration- and time-dependent PK of erdafi- tinib within the range of oral doses evaluated. Age, race, and mild hepatic impairment were not significant covariates of erdafitinib exposure. The effect of sex and renal function on CLfree/F, as well as the effect of weight, sex, and AGP on oral volume of distribution were considered to be of limited clinical relevance. In addition, since erdafitinib dose is adjusted on the basis of serum phosphate, a mechanism of action biomarker reflective of the target engagement, no additional dose adjustments based on the covariates evaluated in this analysis are warranted.

Acknowledgments
Erdafitinib (JNJ-42756493) was discovered in collaboration with Astex Pharmaceuticals. The authors thank all the pa- tients for their participation in this study and acknowledge the collaboration and commitment of all investigators and their staff. The authors also thank Marc de Meulder for the bioanalysis. Writing assistance was provided by Ramji Narayanan, MPharm, ISMPP CMPP (SIRO Clinpharm Pvt. Ltd.) funded by Janssen Global Services and additional editorial support for this manuscript was provided by Harry Ma, PhD (Janssen Global Services, LLC).

Conflicts of Interest
All authors were employees of Janssen Research & Develop- ment.

Funding
This study was supported by Janssen Research & De- velopment LLC, USA.

Author Contributions
Conception and design: JJPR, AGD, EV, and KS; collection of data: AGD, EV, KS; data analysis and interpretation: all authors; manuscript writing and edit- ing: all authors. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors meet International Committee of Medical Journal Editors criteria and all those who fulfilled those criteria are listed as authors. All authors provided direction and comments on the manuscript, made the final decision about where to publish these data, and approved submission to this journal.

Data Sharing
The data sharing policy of the study sponsor, Janssen Pharmaceutical Companies of Johnson & Johnson, is available at https://www.janssen.com/clinical-trials/ transparency. As noted on this site, requests for access to the study data can be submitted through Yale Open Data Access Project site at http://yoda.yale.edu.

References
1. Perera TPS, Jovcheva E, Mevellec L, et al. Discovery and pharmacological characterization of JNJ-42756493 (erdafitinib), a functionally selective small-molecule FGFR family inhibitor. Mol Cancer Ther. 2017;16(6):1010-1020.
2. Tabernero J, Bahleda R, Dienstmann R, et al. Phase I dose- escalation study of JNJ-42756493, an oral pan-fibroblast growth factor receptor inhibitor, in patients with advanced solid tumors. J Clin Oncol. 2015;33(30):3401-3408.
3. Loriot Y, Necchi A, Park SH, et al. Erdafitinib (ERDA; JNJ- 42756493), a pan-fibroblast growth factor receptor (FGFR) in- hibitor, in patients (pts) with metastatic or unresectable urothelial carcinoma (mUC) and FGFR alterations (FGFRa): phase 2 con- tinuous versus intermittent dosing. J Clin Oncol. 2018;36(suppl 6S):abstr 411.
4. Siefker-Radtke A, Necchi A, Park S, et al. First results from the primary analysis population of the phase 2 study of erdafitinib (ERDA; JNJ-42756493) in patients (pts) with metastatic or unresectable urothelial carcinoma (mUC) and FGFR alterations (FGFRalt). J Clin Oncol. 2018;36(suppl):abstract 4503.
5. BALVERSA (erdafitinib) [prescribing information]. Janssen Pharmaceuticals; 2019.
6. Loriot Y, Necchi A, Park SH, et al. Erdafitinib in locally advanced or metastatic urothelial carcinoma. N Engl J Med. 2019;381(4):338-348.
7. Wohrle S, Bonny O, Beluch N, et al. FGF receptors control vitamin D and phosphate homeostasis by mediating renal FGF- 23 signaling and regulating FGF-23 expression in bone. J Bone Miner Res. 2011;26(10):2486-2497.
8. Bahleda R, Italiano A, Hierro C, et al. Multicenter phase I study of erdafitinib (JNJ-42756493), oral pan-fibroblast growth factor receptor inhibitor, in patients with advanced or refractory solid tumors (published online ahead of print May 14, 2019. Clin Cancer Res. 10.1158/1078-0432.CCR-18-3334.
9. Nishina T, Takahashi S, Iwasawa R, Noguchi H, Aoki M, Doi T. Safety, pharmacokinetic, and pharmacodynamics of erdafitinib, a pan-fibroblast growth factor receptor (FGFR) tyrosine kinase inhibitor, in patients with advanced or refractory solid tumors. Invest New Drugs. 2018;36(3):424-434.
10. Lilian Y Li YG, Gonzalez M, Ouellet D. Plasma protein binding of erdafitinib across clinical studies. Presented at American Society for Clinical Pharmacology and Therapeutics (ASCPT) 2019 Annual Meeting, Washington, DC. March 13-16, 2019. Abstract PII068.
11. Committee for Medicinal Products for Human Use. Guideline on reporting the results of population pharmacokinetic analyses. http://www.ema.europa.eu/docs/en_GB/document_library/ Scientific_guideline/2009/09/WC500003067.pdf. Accessed September 5, 2018.
12. Menon-Andersen D, Yu B, Madabushi R, et al. Essential phar- macokinetic information for drug dosage decisions: a concise visual presentation in the drug label. Clin Pharmacol Ther. 2011;90(3):471-474.
13. US Food and Drug Administration. Guidance for industry: population pharmacokinetics. http://www.fda.gov/downloads/ Drugs/GuidanceComplianceRegulatoryInformation/Guidances
/UCM072137.pdf. Published 1999. Accessed November 19,
2018.
14. Dosne AG, Bergstrand M, Karlsson MO. An automated sam- pling importance resampling procedure for estimating parameter uncertainty. J Pharmacokinet Pharmacodyn. 2017;44(6):509-520.
15. Dosne AG, Bergstrand M, Harling K, Karlsson MO. Improving the estimation of parameter uncertainty distributions in nonlin- ear mixed effects models using sampling importance resampling. J Pharmacokinet Pharmacodyn. 2016;43(6):583-596.
16. Schmidt S, Gonzalez D, Derendorf H. Significance of protein binding in pharmacokinetics and pharmacodynamics. J Pharm Sci. 2010;99(3):1107-1122.
17. Dosne A-G, Valade E, Stuyckens K, et al. Optimizing biomarker-based dosing algorithm for JNJ-42756493 erdafitinib through PK- PD modeling and simulations. Annual Meeting of the Popula- tion Approach Group in Europe (PAGE); 2018. Abstract 8584.

Supplemental Information
Additional supplemental information can be found by clicking the Supplements link in the PDF toolbar or the Supplemental Information section at the end of web- based version of this article.