Ponesimod

Effect of Ponesimod Exposure on Total Lymphocyte Dynamics in Patients with Multiple Sclerosis

Belén Valenzuela1 · Juan‑José Pérez‑Ruixo1 · Quentin Leirens2,3 · Sivi Ouwerkerk‑Mahadevan4 · Italo Poggesi5

Accepted: 25 March 2021
© The Author(s), under exclusive licence to Springer Nature Switzerland AG 2021

Abstract

Objective

The aim of this study was to characterize the relationship between ponesimod plasma concentrations and the temporal evolution of lymphocyte counts in multiple sclerosis (MS) patients.

Methods

Population pharmacokinetic (PK) and PK/pharmacodynamic (PD) models were developed using data from phase I, II, and III trials, and the impact of clinically relevant covariates on PK and PD parameters was assessed. Simulations were conducted to evaluate the maximal lymphocyte count reduction after ponesimod treatment, and the time required for total lymphocyte counts to return to normal values after treatment interruption.

Results

In MS patients, ponesimod PK were characterized by a low mean apparent plasma clearance (5.52 L/h) and a moder-
ate mean apparent volume of distribution at steady state (239 L). The model developed indicated that none of the evaluated covariates (age, sex, formulation, food, body weight, clinical condition, and renal impairment) had a clinically relevant impact on the PK/PD parameters. In MS patients, total lymphocyte counts were characterized by a maximum reduction of 88.0% and a half maximal inhibitory concentration (IC50) of 54.9 ng/mL. Simulations indicated that in patients with normal hepatic function treated with ponesimod 20 mg daily, total lymphocyte counts were reduced to 41% of baseline at trough. After stopping treatment, lymphocyte counts were restored to normal levels within one week.

Conclusions

The population PK/PD model well-characterized the PK of ponesimod and the time course of total lymphocyte counts in MS patients. Additionally, none of the evaluated covariates had a clinically relevant impact. This should be taken into considera- tion when assessing the risk of infection, administration of live-attenuated vaccines, and concomitant use of immunosuppressants.

1 Introduction

Multiple sclerosis (MS) is the most common inflammatory neurological disease in young adults, with a mean age of diagnosis of approximately 30 years [1]. Global MS preva- lence is rising, with over 2.2 million reported cases in 2016, a 10.4% increase from 1990 [2]. Although current medica- tions have improved the management of MS, these thera- pies have limitations. Specifically, many need to be injected (interferon [IFN]-β1-a, IFN-β1-b, glatiramer acetate, pegylated IFN-β-1a) or infused (alemtuzumab, mitoxantrone hydrochloride, ocrelizumab, natalizumab) [3], have a long half-life (fingolimod half-life 6–9 days; teriflunomide half- life approximately 19 days) [4, 5], and can lose efficacy or tolerability due to their potential immunogenicity [6]. Phar- maceutical research has made advancements in oral targeted therapies and has addressed some of these limitations.

Ponesimod (JNJ-67896153/ACT-128800) is orally active and is a selective sphingosine-1-phosphate 1 (S1P1) receptor modulator [7, 8]. S1P1 is widely expressed in many tissues [9] and its expression on lymphocytes controls lymphocyte egress from lymphoid organs [10]. Through modulation of S1P1, ponesimod causes a rapid, dose-dependent, reversible reduction in peripheral blood lymphocyte counts [11]. In blocking the movement of lymphocytes out of lymphoid organs and into lymphatic and vascular circulation, ponesimod prevents lymphocyte recruitment to sites of inflammation [11].

Ponesimod is an effective MS therapy [12, 13]. In the phase II AC-058B201 study (NCT01006265; abbreviated B201 from here), once-daily treatment with ponesimod at a 10, 20, or 40 mg daily dose significantly reduced the number of new T1 Gd+ lesions and showed a beneficial effect on clinical endpoints: the mean cumulative number of new T1 Gd+ lesions at weeks 12–24 was significantly reduced in the ponesimod 10 mg (43.4 %; p = 0.0318), 20 mg (83.0 %, p < 0.0001), and 40 mg (77.4%; p < 0.0001) groups compared with placebo. Similarly, the mean annu- alized relapse rate (ARR) was reduced by 36.8% with ponesimod 10 mg (p = 0.162), by 20.7% with ponesimod 20 mg (p = 0.442), and by 52.2% (p = 0.0363) with pone- simod 40 mg compared with placebo [12]. Recently, the phase I I I OPTIMUM s tudy (NCT02425644; abbreviated B301 from here) compared the efficacy and safety of ponesimod with teriflunomide in patients with relapsing MS [13, 14]. Treatment with ponesimod (20 mg) was found to significantly reduce the ARR, the study’s primary endpoint, by 30.5% (95% con- fidence interval [CI] 15.2–43.0%; p = 0.0003) compared with teriflunomide (14 mg) [14]. Ponesimod pharmacokinetics (PK) and pharmacody- namics (PD) have been investigated in several phase I and II clinical trials [15]. The PK of ponesimod was dose pro- portional, having a high absolute bioavailability (83.3%) and rapid absorption, achieving a peak plasma concentra- tion within 2–4 h after dosing. In addition, ponesimod is characterized by a moderate volume distribution and low clearance, which translates to a terminal half-life of 33 h, and a mild to moderate interindividual variability (IIV) [16]. A two-compartment model with sequential zero/first- order absorption, including lag time, intercompartmental drug flow, and first-order clearance, was previously used and adequately described the time course of ponesimod plasma concentrations after oral administration using data from healthy volunteers and MS patients [17]. In healthy volunteers, single daily doses of ponesimod ≥ 5 mg caused a reduction in peripheral blood lymphocyte counts, with the maximum reduction observed 6 h postdose. This is caused by reversible sequestration of lymphocytes in lym- phoid tissues. An indirect response PK/PD model with circadian rhythm was previously used to describe the time course of total lymphocyte counts following ponesimod therapy in healthy volunteers [18]. This model considered that the appearance of lymphocytes in blood followed a zero-order input rate, which is reduced by ponesimod plasma concentrations, thereby mimicking the mechanism of action of ponesimod, which reversibly blocks the egress of lymphocytes from lymphoid organs to blood. The current analysis was aimed at characterizing the PK of ponesimod in MS patients in the B201 and B301 studies and to explore the relationship between ponesimod plasma concentrations and the temporal evolution of lymphocyte counts. In this study, the previous population PK and PK/ PD models using the phase II and III studies have been expanded to better understand dose, exposure, and lympho- cyte response relationships following ponesimod treatment in patients with MS. These data are critical in assessing ben- efit/risk profiles and appropriate dosing regimens of ponesi- mod in MS patients. 2 Materials and Methods 2.1 Clinical Studies, Sample Collection, and Bioanalytical Methods The ponesimod population PK analysis using a previous PK dataset, which included data from 11 phase I studies and two phase II studies [17], was expanded to include PK data from the B301 study. This pooled PK dataset, including a total of 14 clinical studies, was used to reassess the per- formance of the previous population PK model describing the PK in the MS population. For the population PK/PD analysis, the previous PK/PD dataset, which included data from 81 healthy volunteers pooled from three phase I studies (studies AC-058-101, AC-058-102, and AC-058-109) [18], was expanded to include data from MS patients from the B201and B301 studies. Details regarding the phase I studies, patient characteristics, and sample collection related to the PK and PK/PD analyses have been previously published [17, 18], and the characteristics from the B201 and B301 studies are described in Table 1. Ponesimod plasma concentrations were analyzed using validated liquid chromatography coupled with tandem mass spectrometry [19]. The lower limit of quantification was 1 ng/mL. The method accuracy was higher than 87.5% and the within- and between-run precision values were satisfactory, with coefficient of variation (CV) values of 0.600–13.7% and 1.30–11.9%, respectively. Total lympho- cyte counts were measured by hematocytometry as part of the standard hematology assessments. 2.2 Pharmacokinetic/Pharmacodynamic (PK/PD) Model Development 2.2.1 Structural and Statistical PK/PD Model The previously available structural PK model [17] was an open, linear, two-compartment model with sequential zero/ first-order absorption after a lag time, parameterized in terms of clearance and volume of distribution, and was refined to characterize the time course of ponesimod in MS patients. In this PK model, IIV was estimated for all PK parameters. The previous structural PD model was an indirect response model including a first-order rate (Rin) describing the appearance of lymphocytes in peripheral blood and its circadian variation, ponesimod concentration-dependent reduction in Rin, and a first-order rate constant (kout) charac- terizing the disappearance of lymphocytes from the blood stream [18]. A schematic of the structural PK/PD model is shown in Fig. 1. The circadian variation was modeled with a cosine function over 24 h, and parameterized with an amplitude (amp) and a time shift (shift). The effect of ponesimod on Rin was described by an Imax model, with Imax denoting the maximum possible fractional lymphocyte count reduction, and IC50, denoting the ponesimod plasma concen- tration required to reach half the maximum reduction. This PK/PD model was refined to characterize the time course of lymphocyte count following ponesimod treatment in MS patients. For the PK/PD development, a sequential modeling approach was followed, and the resulting individual esti- mates of PK parameters were used to obtain the time course of ponesimod plasma concentration, which was used as an input function of the PK/PD model [20]. The IIV in the PK/PD model parameters was assumed to follow a log-normal distribution and, consequently, an exponential error model was used, with the exception of shift and Imax because they were logistically transformed to only allow values between 0–24 and 0–1 for time shift and Imax, respectively. Consequently, the IIV was modeled as an addi- tive random effect in the logit domain. IIV was included in all the PK/PD parameters except for Rin. Residual variability was evaluated using a combined additive and proportional error model for PK and proportional error model for PD measurements. 2.2.2 Covariate Analysis Body weight, race, disease status, hepatic impairment, drug formulation, and food effect were previously included in the population PK model [17] and were evaluated with the updated dataset in order to understand similarities or differ- ences between the MS population and healthy volunteers. Table 2 shows the summary of the main characteristics of the MS participants from B201 and B301 at baseline, strati- fied by study. Characteristics of the healthy volunteer popu- lation have been previously published [17, 18]. The previously available population PK/PD model explored the effect of age, sex, and body weight [18]. Only age was reported to have a significant, although not clini- cally relevant, effect on the disappearance rate of lympho- cytes in healthy volunteers and was also explored in the MS population. The covariate analysis was performed using the forward selection and backward elimination methodology [21]. Dur- ing forward selection, a decrease of at least 6.63 points in the minimum objective function value (OFV; Chi-square test, degrees of freedom [df] = 1; p < 0.01) was selected as a statis- tically significant criterion to retain a covariate in the model. During backward elimination, a covariate was considered significant if it contributed to a decrease of at least 10.8 points in the OFV (Chi-square test, df = 1; p < 0.001) when removed from the model. In addition, statistically significant covariate effects were considered not clinically relevant if the 90% CI of the geometric mean ratio was fully included in the 80–125% interval. With this methodology, only covariates showing sig- nificant and clinically relevant contributions were conserved in the population PK or PK/PD models. Fig. 1 Schematic representation of the population PK/PD model of lymphocytes. Modified from Lott et al. [17]. Circles denote compart- ments, solid lines denote drug flow with associated PK parameters, dotted lines indicate relationships. amp amplitude of the circadian rhythm, CL/F apparent clearance, Doral oral dose, Fr fraction of the drug absorbed via zero order, IC50 concentration required to reach half-maximum ponesimod effect, Imax maximum ponesimod effect, ka first-order absorption rate constant, kout first-order output rate con- stant, shift time shift of the circadian rhythm, PD pharmacodynamics, PK pharmacokinetics, Q/F apparent intercompartmental drug flow, Rin zero-order input rate, Tk0 duration of the zero-order absorption, Tlag absorption lag time, Vc/F apparent central volume of distribution, Vp/F apparent peripheral volume of distribution. To fully understand the impact of the covariates on ponesimod exposure in MS subjects, exploratory forest plots were created using the post hoc estimates of the individual exposure obtained from the updated popu- lation PK model. The dose-normalized area under the plasma concentration-time curve (AUC) for ponesimod was calculated for each subject. The forest plot represents the geometric mean ratio and its 90% CI for the AUC of ponesimod in a specific subpopulation relative to the value in the reference subpopulation, after adjusting other covariates. A covariate effect can be considered not clini- cally relevant if the observed differences fall within the 80–125% interval. Similarly, an additional forest plot with the final PK/PD model was also created to understand the impact of the selected covariates over the absolute change in the lymphocyte counts at steady-state. 2.2.3 Model Evaluation Goodness-of-fit (GOF) plots, reduction on the OFV, and acceptable precision on the parameter estimates were used as complementary tools for the evaluation of the model fits. Moreover, a prediction-corrected visual predictive check (pcVPC; stratified by dose for the B201 study) was used as an internal validation method for the population’s models [22]. The final parameter estimates were used to simulate 1000 replicates of the population models, stratified by study and dose levels. The prediction-corrected median and 95% prediction intervals were computed and summarized across replicates as the mean and 95% CI across the 2.5th, 50th and 97.5th percentiles of the observed data. 2.2.4 Model‑Based Simulations To calculate the required time for total lymphocyte counts to return to normal values after ponesimod treatment interrup- tion in individuals with normal liver function and with mild hepatic impairment, the final PK/PD model was used for simulating the lymphocyte counts after multiple ponesimod doses of 20 mg once daily. The sampling scheme used to simulate lymphocyte dynamics included assessment every 12 h for 30 days. It was assumed that the steady-state dose was reached after 10 days of treatment, considering ponesi- mod terminal half-life (33 h) and recovery for the remainder of 30 days. Recovery was assessed considering 50%, 90%, and 99% of patients back-to-normal values of lymphocyte counts. The absolute change from baseline and the percent- age of change from baseline were also derived from total lymphocyte counts. In addition, the percentage of patients returning to normal lymphocyte counts within 2, 4, and 8 days after stopping treatment was also computed. 2.2.5 Computer Software Data analysis was performed using NONMEM version 7.3.0 (NONMEM 7.3; ICON Development Solutions, Hanover, MD, USA) [23]. The Fortran compiler was Intel® Fortran 64 Compiler Professional, version 11.1. The NONMEM analy- ses were performed in a validated environment, High Perfor- mance Pharmacometrics Platform (HP3) System (Rudraya Sonic Version 4), based on Good Automated Manufacturing Practice and in accordance with 21 CFR Part 11 and Good Clinical Practice regulations. Small modifications to the analysis dataset, exploratory analysis, diagnostic graphics, and post-processing of NONMEM analysis results were car- ried out using the R Project for Statistical Computing, ver- sion 3.4.1 for Windows (Comprehensive R Network, http:// cran.r-project.org) [24] in the validated HP3 environment. The first-order conditional estimation (FOCE) method with INTERACTION and Monte Carlo importance sampling hemi-hydrated crystalline form, while Form C is an anhy- drous polymorphic form and is the most stable form at room temperature. Form A was used in the initial phase I studies, and Form C was used in the phase II and III studies. Ponesimod film-coated tablets contain Form C with the following inactive ingredients: lactose monohydrate, micro- crystalline cellulose, povidone, sodium lauryl sulfate, cro- scarmellose sodium, silica colloidal anhydrous, and mag- nesium stearate for the core tablet and OPADRY® II for the coating.CrCl creatinine clearance, EDSS expanded disability status score, LC lymphocyte count, N number of participants with data, n number of participants in the specified category, NCIc National Cancer Institute classification system Continuous variables are expressed as median (range) and categorical variables are expressed as n (%) Renal function was defined, taking into account the CrCl values (cal- culated using the Cockcroft–Gault formula), as follows: normal: CrCl ≥90 mL/min; mild: CrCl from 60 to 89 mL/min; moderate: CrCl from 30 to 59 mL/min; severe: CrCl<30 mL/min. Hepatic impairment was defined and computed using the NCI criteria [28, 29] (IMP) were sequentially used for the population PK and PK/ PD models. The IMP method was used to evaluate the accu- racy and precision of parameter estimates from the final pop- ulation PK and PK/PD models [25]. Model-based simula- tions were performed using Simulo 8.0 version Expert [26], a JAVA-based software that creates and executes R scripts to simulate the different scenarios in an efficient manner. 2.2.6 Ponesimod Formulations The two polymorphic forms of ponesimod used during clinical studies were Form A and Form C. Form A is a 3.1 Population PK Modeling In total, 15,342 plasma concentrations of ponesimod col- lected from 340 healthy volunteers and 905 MS patients were included in the PK analysis. The previously devel- oped population PK model [17] was successfully updated to include all data from the B301 study, with the same structural and statistical model. The parameter estimates of the ponesimod population PK model for patients with MS, including the clinically relevant covariate effects on model parameters, are presented in Table 3. All fixed and random effects were estimated with adequate precision, as measured by relative standard error (RSE, %), except the between- subject variability in apparent intercompartmental drug flow (Q/F), and were comparable with the previously published results [17]. Within the range of covariate values evaluated, body weight, disease status, race, and hepatic impairment were statistically significant covariates of the ponesimod population PK parameters; however, among those covari- ates, only the hepatic impairment was found to be a clini- cally relevant covariate. Figure 2 includes the forest plots created using the final population PK model, including the statistically significant covariates. As can be seen, except for subjects with moderate and severe hepatic impairment, none of the tested covariates had a clinically relevant impact on ponesimod exposure. As the phase III study included the to-be-marketed drug formulation (Tablet of Form C) and additional information regarding food effect, food and formulation effects on pone- simod PK were re-evaluated. Three formulations (Capsule of Form A, Capsule of Form C, and Tablet of Form C) [17] and three food conditions (fasted, fed, and uncontrolled) were considered when analyzing the full available PK data. Of the available plasma concentrations, 12.8% were from patients receiving Capsule of Form A, 25.5% were from Capsule of Form C, and 61.7% were from Tablet of Form C. Of the total available PK data, 63.0% were from fasted conditions, 6.00% were from fed conditions, and 31.0% were from uncontrolled conditions. The inclusion of food and formulation effects for duration of the zero-order absorption [TK0] (δOFV = 633, df = 4; p < 0.001), fraction of the drug absorbed via zero order [Fr] (δOFV = 331, df = 4; p < 0.001), absorp- tion lag time [Tlag] (δOFV = 490, df = 4; p < 0.001), first- order absorption rate constant [ka] (δOFV = 277, df = 4; p < 0.001), and relative bioavailability (δOFV = 531, df = 4; p < 0.001) significantly improved the fit. The simultaneous inclusion of food and formulation effects for Tlag, ka, and relative bioavailability provided the largest improvement of the model fit (δOFV = 795, df = 12; p < 0.001) with respect to the model that did not include any food or formulation effects over the absorption PK parameters. The food and formulation effects relative to ponesimod tablets of Form C for the absorption parameters are presented in Table 4, indicating no relevant food effect and formulation differ- ences as all the 90% CIs remained within the bioequivalence limits. Overall, the updated PK model accurately described the covariate effect from previously reported phase I studies. 3.2 Population PK/PD Modeling A total of 5062 ponesimod plasma concentrations and 14,362 lymphocyte counts from 81 healthy volunteers and 1026 MS patients were used for the PK/PD analysis on lymphocyte counts. Due to the pharmacological effect of ponesimod, mean lymphocyte count decreased during treat- ment with ponesimod and returned to baseline range around 1 week after the end of treatment. In the B201 study, at pre- dose levels after 12 weeks of treatment, a decrease of 46.8%, 63.1%, and 70.0% from baseline was observed for the 10, 20, and 40 mg ponesimod daily doses, respectively. No patient discontinued study treatment due to lymphopenia. Similar results were obtained from the B301 study, where there was a rapid decrease in lymphocyte count from baseline in week 2 (42.3%), week 4 (59.2%), and week 12 (63.6%) in patients treated with ponesimod 20 mg. From week 12 to week 108, the mean lymphocyte counts remained stable. The previously developed PK/PD model in healthy individuals [18] was successfully applied to fit the data from MS patients. As the mean lymphocyte count at baseline value determined in the B201 and B301 stud- ies (1.97 × 109/L and 1.89 × 109/L, respectively) were lower than those previously reported from healthy vol- unteers (2.18 × 109/L), the previous PK/PD model was updated to include two different factors accounting for the baseline lymphocyte count values observed in each MS study. The inclusion of these factors was associated with a statistically significant reduction in minimum OFV by 33.1 points (df = 2; p < 0.001). The final PD parameters are summarized in Table 2. All PD parameters were esti- mated with good precision (RSE <20%). The inclusion of the age effect on kout was not statistically significant (δOFV = 3.56, df = 1; p = 0.059). Sex and body weight effect over PD model parameters were also negligible (electronic supplementary Fig. 1). IIV was estimated for all PD parameters, except Rin, and the estimated IIV values were relatively low (< 27.0%), in agreement with previ- ously reported IIV in healthy volunteers. Moreover, IIV for amplitude was close to 0, indicating that the data did not support the estimation of this IIV. Indeed, fixing this parameter to 0, the change on the OFV was negligible (0.02 points). Residual error was estimated at 23.1% and, as expected, was slightly higher than the previous value reported in the phase I trials. 3.3 Model Evaluation The GOF plots for the final population PK or PK/PD models (electronic supplementary Figs. 2 and 3) show that there was good agreement in the observed, individ- ual, and population-predicted ponesimod plasma concen- trations and lymphocyte counts in MS patients. The GOF plots of the final models against the population and indi- vidual model predictions showed a normal random scat- ter around the identity line and indicated the absence of significant bias. The distribution of conditional weighted residuals as a function of the population predictions and time, did not show any relevant trend of evidenced model inadequacy. The mean and standard deviation (SD) of the normalized prediction distribution errors (NPDE) for the ponesimod plasma concentrations were 0.069 (95% CI 0.055–0.086) and 0.953 (95% CI 0.942–0.965), respec- tively. For the total lymphocyte counts, the mean and SD of the NPDE was −0.039 (95% CI − 0.053 to − 0.024) and 0.930 (95% CI 0.920–0.945), respectively. These results confirm the model accuracy because the mean of the NPDE for both plasma concentrations and lymphocyte counts was very close to 0; however, there is a trend to slightly overpredict the variability of plasma concentra- tions and lymphocyte counts since the upper limits of the 95% CIs of the NPDE SD do not include 1. The pcVPC evidenced and supported the ability of the PK/PD model to describe the time-course of ponesimod PK and lym- phocyte count, and the variability in MS patients included in the analysis across the dose range evaluated (Fig. 3). 3.4 Model Simulations Simulations from the final model show the temporary (daily) evolution of total lymphocyte counts after a 20 mg once- daily regimen initiation and discontinuation after 10 days of treatment in individuals with normal hepatic function and mild hepatic impairment (Fig. 4). At steady-state, at trough the total lymphocyte count is reduced to a median value of 41% of baseline in patients with normal hepatic function,which is similar to values observed in other clinical studies [18]. In patients with mild hepatic impairment, simulations showed that total lymphocyte count is reduced to 34% of the baseline value. After stopping treatment, lymphocyte counts were restored to normal levels in 2 days in 50% of patients, 4 days in 90% of patients, and 8 days in 99% of patients with normal hepatic function. As a result of mild hepatic impair- ment, lymphocyte counts were restored to normal levels in 3 days in 50% of patients, 7 days in 90% of patients, and 12 days in 99% of patients. 4 Discussion A previously elucidated population PK model based on an open, linear, two-compartment disposition model follow- ing oral absorption described the time course of ponesimod plasma concentrations and its variability in MS patients included in the B301 study. The parameter estimates were in line with those previously found in healthy volunteers. After oral administration of Tablet of Form C, ponesimod absorption in MS patients was relatively fast. After an ini- tial lag time of 0.376 h, 18.0% of the administered dose was absorbed through a zero-order process of approximately 0.50 h duration, while the remaining 82.0% of the dose administered was absorbed following a first-order process immediately after the zero-order process was completed. Consistent with previous results, ponesimod apparent vol- ume of distribution at steady state for a White patient with MS was estimated to be 239 L/72 kg, reflecting a moderate distribution to peripheral tissues. A single apparent clear- ance (CL/F) parameter, estimated as 5.52 L/h/72 kg, quan- tified the elimination of ponesimod through all the elimi- nation routes. Ponesimod plasma concentrations observed after the peak declined in a bi-exponential manner, with an α and β t½ of approximately 2 and 33 h, respectively. The PK of ponesimod appears to be dose-proportional and time- independent after oral administration of both Capsules and Tablets of Forms A and C, administered under fasting and fed conditions. Fig. 2 Forest plot for the covariate evaluation for ponesimod using the final population PK model, which included body weight, race, disease status, hepatic impartment, formulation, and food effects. Note that formulation and food effects have not been included in the forest plot as their impact over PK parameters have been detailed in Table 4. AUC area under the concentration–time curve, CI confidence interval, GMR geometric mean ratio, PK pharmacokinetics. Consistent with previous results, within the range of covariate values evaluated in MS patients, renal impair- ment had no discernible impact on the PK parameters of ponesimod. Food, formulation, body weight, and hepatic impairment were statistically significant covariates of the ponesimod population PK parameters in MS patients; how- ever, among those covariates, only hepatic impairment was a clinically relevant covariate. The increase in body weight lead to increased CL/F and both apparent central volume of distribution (Vc/F) and apparent peripheral volume of distri- bution (Vp/F). An increase in body weight from 72 kg (typi- cal value) to 146 kg (maximal value) provided an increase of 25%, 47%, and 40% for CL/F, Vc/F, and Vp/F, respectively. Black race was associated with a 3% decrease in CL/F. Only mild hepatic impairment, which caused an 11% reduction in the apparent elimination clearance compared with indi- viduals with normal hepatic impairment, was considered a clinically meaningful covariate. The PK/PD model describing total lymphocyte count in healthy volunteers was successfully applied to the data from the B201 and B301 studies, with limited changes in model parameters; for example, the adjustment by study-specific total lymphocyte count at baseline. These findings suggest no differences in the ponesimod effect on total lymphocyte counts between healthy volunteers and MS patients. Total lymphocytes were characterized by a maximum reduction of 88.0% and an IC50 of 54.9 ng/mL. Following steady-state dosing of ponesimod 20 mg, the inhibition of Rin in MS patients is expected to range from 60.0 to 75.0% at trough concentrations (approximately 100 ng/mL) and from 70 to 85% at maximum concentration (approximately 168 ng/mL). The parameter estimate for kout was consistent with the findings from healthy volunteers (kout = 0.379 h-1, t½ = 1.80 h), suggesting that the lymphocyte turnover is rate- limited by the PK of ponesimod (t½ = 32.0 h). Indeed, in the single-dose AC-058-101 study, total lymphocyte counts after a washout period of 96 h following ponesimod administra- tion were indeed similar to those observed in placebo-treated patients [19]. The circadian shift was estimated at 13.9 h, a result that is consistent with the circadian pattern reported for endogenous plasma glucocorticoids (i.e. cortisol, which Fig. 3 pcVPC based on the final population PK/PD model for pone- simod plasma concentrations (upper panels) and lymphocyte counts (lower panels), stratified by study. Blue circles denote the observed ponesimod plasma concentrations (upper panels) or total lympho- cyte counts (lower panels); solid lines represent the median (red) and 2.5th and 97.5th percentiles (blue) of the binned observations; red band denotes confidence intervals of the median of the model pre- diction and blue bands denote confidence interval of the 2.5th and 97.5th percentiles of the model predictions (1000 replicates). For the B201 study, stratification by treatment group is displayed: 10–10 mg ponesimod qd dose level; 20–20 mg ponesimod qd dose level; and 40–40 mg ponesimod qd dose level. Patients randomized to the pone- simod treatment groups received 10 mg daily for 7 days before being uptitrated, at weekly intervals, to the dose associated with the group at which they were randomized. pcVPC prediction-corrected visual predictive check, PD pharmacodynamic, PK pharmacokinetic, qd once daily. Fig. 4 Simulated absolute total lymphocyte counts following admin- istration of ponesimod 20 mg once daily for 10 days in individuals with normal hepatic function and mild HI. Colored solid lines denote the median, and dotted and dashed lines denote the 5th and 95th per- centiles, respectively, from 1000 patients. Solid vertical black lines denote the time to treatment discontinuation, and dashed horizon- tal black lines (bottom) denote 90% of baseline values. Left panels: patients with normal hepatic function; right panels: patients with mild HI. HI hepatic impairment has immunosuppressive effects and attains nadir plasma concentrations in the evening) [27]. The amplitude of the circadian variation in total lymphocytes was determined to be 16.0%. IIV was determined for all PD parameters, expect Rin, and the amplitude and was moderate (< 27.0%). Within the range of values of the covariates explored (age, sex, and body weight), it was not possible to deter- mine a relationship between these covariates and the PD parameters using the data included in the current analysis (see electronic supplementary Fig. 1). Model-based simulations indicated that after stopping ponesimod maintenance treatment of 20 mg in patients with normal hepatic function, the total lymphocyte count returns above the lower limit of the normal value within 2, 4, and 8 days in 50%, 90%, and 99% of patients, respectively. In patients with mild hepatic impairment, total lymphocyte count returns above the lower limit of the normal value within 3, 7, and 12 days in 50%, 90%, and 99% of patients, respectively. The model predicted that at steady state following a 20 mg once-daily dose, an average maximal reduction of about 69% (i.e. lymphocyte count accounting for 31% of baseline) can be reached in healthy patients for total lym- phocyte counts [18], which is consistent with the results observed in the B201 and B301 studies in patients with MS, in which an average total lymphocyte count reduction of 63.1% and 63.6% (i.e. lymphocyte counts accounting for 36.9% and 36.4% of the baseline values, respectively) was observed at predose following a 20 mg once-daily dose at steady state, respectively. Simulating different dosing scenarios leading to maintenance doses of 20 and 40 mg showed that at steady state, an average reduction of up to 69% and 77% can be reached for total lymphocyte count for the two regimens, respectively. Therefore, dou- bling the steady-state dose from 20 to 40 mg is expected to yield only a further 8% additional reduction for total lymphocytes. 5 Conclusion The population PK/PD model presented in this study ade- quately characterized the PK of ponesimod and the time course of total lymphocyte counts in MS patients, which were shown to be similar to those previously reported in healthy volunteers. Except for hepatic impairment, none of the evaluated covariates, including age, body weight, sex, race, food, formulation, or renal impairment, had a clinically relevant impact on the PK of ponesimod. After stopping ponesimod 20 mg treatment in patients with normal hepatic function and mild hepatic impairment, the total lymphocyte count returned above the lower limit of the normal value within 8 and 12 days, respectively. This should be taken into consideration when assessing the risk of infection, administration of live-attenuated vaccines, and concomitant use of immunosuppressants. Supplementary Information The online version contains supplemen- tary material available at https://doi.org/10.1007/s40262-021-01019-9. Acknowledgements The authors thank the study participants, without whom this study would never have been accomplished, and all the investigators and their medical, nursing and laboratory staff for their participation in this study. The authors also thank Colleen Elliott for writing assistance and editorial support. Declarations Funding This study was supported by Actelion Pharmaceuticals Ltd, Part of Janssen Pharmaceutical Companies, Allschwil, Switzerland. Conflicts of interest Belén Valenzuela and Juan Jose Perez-Ruixo are employees of Janssen-Cilag Spain, part of the Janssen Pharmaceutical Company of Johnson & Johnson, and hold stock in Johnson & John- son. Quentin Leirens was an employee of SGS Exprimo, part of SGS Belgium NV, at the time this analysis was conducted. Sivi Ouwerkerk- Mahadevan is an employee of Janssen NV, part of the Janssen Pharma- ceutical Company of Johnson & Johnson, and holds stock in Johnson & Johnson. Italo Poggesi is an employee of Janssen-Cilag Italy, part of Janssen Pharmaceutical Company of Johnson & Johnson, and holds stock in Johnson & Johnson. Ethics approval All procedures performed in studies involving human participants were in accordance with the ethical standards of the insti- tutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Protocols were reviewed and approved by an Institutional Review Board. Consent to participate Freely given, informed consent to participate was obtained for all human participants in this study. Consent for publication Not applicable. Availability of data and material The data sharing policy of the Jans- sen Pharmaceutical Companies of Johnson & Johnson is available at https://www.janssen.com/clinical–trials/transparency. As noted on that site, requests for access to the study data can be submitted through the Yale Open Data Access (YODA) Project site at http://yoda.yale.edu. Code availability Not applicable. Author contributions All authors participated in the original design of the studies and monitoring of data quality, and contributed to data interpretation, and development and review of this manuscript. They confirm that they have read the journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines. All authors meet the International Committee of Medi- cal Journal Editors (ICMJE) criteria and all those who fulfilled those criteria are listed as authors. All authors had access to the study data, provided direction and comments on the manuscript, made the final decision on where to publish these data, and approved submission to this journal. References 1. Reich DS, Lucchinetti CF, Calabresi PA. Multiple Sclerosis. N Engl J Med. 2018;378(2):169–80. 2. GBD 2016 Motor Neuron Disease Collaborators. Global, regional, and national burden of multiple sclerosis 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2016;18(3):269–85. 3. Dargahi NKM, Tselios T, Androutsou ME, de Courten M, Mat- soukas J, Apostolopoulos V. Multiple Sclerosis: Immunopathol- ogy and Treatment Update. Brain Sci. 2017;7(7):78. 4. David OJ, Kovarik JM, Schmouder RL. Clinical pharmacokinetics of fingolimod. Clin Pharmacokinet. 2012;51(1):15–28. 5. Sartori A, Carle D, Freedman MS. Teriflunomide: a novel oral treatment for relapsing multiple sclerosis. Expert Opin Pharma- cother. 2014;15(7):1019–27. 6. Tovey MG, Lallemand C. Immunogenicity and other problems associated with the use of biopharmaceuticals. Ther Adv Drug Saf. 2011;2(3):113–28. 7. Piali L, Froidevaux S, Hess P, Nayler O, Bolli MH, Schlosser E, et al. The selective sphingosine 1-phosphate receptor 1 agonist ponesimod protects against lymphocyte-mediated tissue inflam- mation. J Pharmacol Exp Ther. 2011;337(2):547–56. 8. Bolli MH, Abele S, Binkert C, Bravo R, Buchmann S, Bur D, et al. 2-imino-thiazolidin-4-one derivatives as potent, orally active S1P1 receptor agonists. J Med Chem. 2010;53(10):4198–211. 9. Chun J, Goetzl EJ, Hla T, Igarashi Y, Lynch KR, Moolenaar W, et al. International Union of Pharmacology XXXIV. Lysophospho- lipid receptor nomenclature. Pharmacol Rev. 2002;54(2):265–9. 10. Cyster JG, Schwab SR. Sphingosine-1-phosphate and lym- phocyte egress from lymphoid organs. Annu Rev Immunol. 2012;30:69–94. 11. Matloubian M, Lo CG, Cinamon G, Lesneski MJ, Xu Y, Brink- mann V, et al. Lymphocyte egress from thymus and periph- eral lymphoid organs is dependent on S1P receptor 1. Nature. 2004;427(6972):355–60. 12. Olsson T, Boster A, Fernandez O, Freedman MS, Pozzilli C, Bach D, et al. Oral ponesimod in relapsing-remitting multiple sclero- sis: a randomised phase II trial. J Neurol Neurosurg Psychiatry. 2014;85(11):1198–208. 13. ClinicalTrials.gov. Oral Ponesimod Versus Teriflunomide in Relapsing MUltiple Sclerosis (OPTIMUM), ClinicalTrials.gov; 2015. 14. Kappos L, Fox RJ, Burcklen M, et al. Ponesimod compared with teriflunomide in patients with relapsing multiple sclerosis in the active-comparator phase 3 OPTIMUM Study. JAMA Neurol. 2021. https://doi.org/10.1001/jamaneurol.2021.0405 15. Juif PE, Kraehenbuehl S, Dingemanse J. Clinical pharma- cology, efficacy, and safety aspects of sphingosine-1-phos- phate receptor modulators. Expert Opin Drug Metab Toxicol. 2016;12(8):879–95. 16. Boehler M, Juif PE, Hoch M, Dingemanse J. Absolute bio- availability of ponesimod, a selective S1P1 receptor modula- tor, in healthy male subjects. Eur J Drug Metab Pharmacokinet. 2017;42(1):129–34. 17. Lott D, Lehr T, Dingemanse J, Krause A. Impact of demo- graphics, organ impairment, disease, formulation, and food on the pharmacokinetics of the selective S1P1 receptor modulator ponesimod based on 13 clinical studies. Clin Pharmacokinet. 2017;56(4):395–408. 18. Lott D, Krause A, Seemayer CA, Strasser DS, Dingemanse J, Lehr T. Modeling the effect of the selective S1P1 receptor modu- lator ponesimod on subsets of blood lymphocytes. Pharm Res. 2017;34(3):599–609. 19. Brossard P, Derendorf H, Xu J, Maatouk H, Halabi A, Dinge- manse J. Pharmacokinetics and pharmacodynamics of ponesimod, a selective S1P1 receptor modulator, in the first-in-human study. Br J Clin Pharmacol. 2013;76(6):888–96.
20. Zhang L, Beal SL, Sheiner LB. Simultaneous vs. sequential analy- sis for population PK/PD data I: best-case performance. J Phar- macokinet Pharmacodyn. 2003;30(6):387–404.
21. Mandema JW, Verotta D, Sheiner LB. Building population phar- macokinetic–pharmacodynamic models. I. Models for covariate effects. J Pharmacokinet Biopharm. 1992;20(5):511-28.
22. Bergstrand M, Hooker AC, Wallin JE, Karlsson MO. Predic- tion-corrected visual predictive checks for diagnosing nonlinear mixed-effects models. AAPS J. 2011;13(2):143–51.
23. Sheiner LB, Steimer JL. Pharmacokinetic/pharmacodynamic modeling in drug development. Annu Rev Pharmacol Toxicol. 2000;40:67–95.
24. R Core Team, R: A language and environment for statistical com- puting. Vienna: R Foundation for Statistical Computing; 2017.
25. Bauer RJ. ICON Development Solutions. NONMEM Users Guide: Introduction to NONMEM 7.2.0. 2011. Ellicott City, MD. https:// nonmem.iconplc.com/nonmem720/guides/nm720.pdf. Accessed 20 Jan 2021.
26. Simulo 8.0 version Expert. https://www.page-meeting.org/defau
lt.asp?abstract=9182. Accessed 20 Jan 2021.
27. Mager DE, Lin SX, Blum RA, Lates CD, Jusko WJ. Dose equiv- alency evaluation of major corticosteroids: pharmacokinetics and cell trafficking and cortisol dynamics. J Clin Pharmacol. 2003;43(11):1216–27.
28. Patel H, Egorin MJ, Remick SC, Mulkerin F, Takimoto CHM, Doroshow JH, et al. Comparison of Child-Pugh (CP) criteria and NCI organ dysfunction working group (NCI-ODWG) criteria for hepatic dysfunction (HD): Implications for chemotherapy dosing. J Clin Oncol. 2004;22(14 Suppl):6051.
29. NCI. Protocol Template for Organ Dysfunction Studies. 2019. Available at: https://ctep.cancer.gov/protocoldevelopment/docs/ ctep_organ_dysfunction_protocol_template.docx. Accessed 20 Jan 2021.