Population Pharmacokinetics of the Active Metabolite ...

Journal of Pharmacokinetics and Pharmacodynamics (© 2005) DOI: 10.1007/s10928-005-0049-8

Population Pharmacokinetics of the Active Metabolite of Lef lunomide in Pediatric Subjects with Polyarticular Course Juvenile Rheumatoid Arthritis Jun Shi,1,4,∗ Steven J. Kovacs,2 Yaning Wang,1 Thomas M. Ludden,3 and Vijay O. Bhargava1 Received May 6, 2005—Final May 6, 2005 Leflunomide is a pyrimidine synthesis inhibitor used in the treatment of rheumatoid arthritis. Data from two clinical studies were used to establish a population pharmacokinetic (PPK) model for the active metabolite (M1) of leflunomide in patients with juvenile rheumatoid arthritis (JRA) and determine appropriate pediatric doses. Seventy-three subjects 3–17 years of age provided 674 M1 concentrations. The PPK model was derived from nonlinear mixed-effects modeling and qualified by cross-study evaluation and predictive check. A one-compartment model with first-order input described M1 PPK well. Body weight (WT) correlated weakly with oral clearance (CL/F = 0.020·[WT/40]0.430 ) and strongly with volume of distribution (V/F = 5.8·[WT/40]0.769 ). Steady-state concentrations (Css ) of M1 in JRA were compared for a variety of leflunomide dose regimens using Monte–Carlo simulation. To achieve comparable Css values in pediatric patients with JRA to that in adult patients, doses of leflunomide should be adjusted modestly: 10 mg/d for 10–20 kg, 15 mg/d for 20–40 kg, and 20 mg/d for > 40 kg. KEY WORDS: leflunomide; population pharmacokinetics; juvenile rheumatoid arthritis; pediatrics; NONMEM.

1 Global Biopharmaceutics, Drug Metabolism and Pharmacokinetics, Aventis Pharmaceuti-

cals, Bridgewater, NJ, USA. 2 US Medical Research, Clinical Pharmacology, Aventis Pharmaceuticals, Bridgewater, NJ,

USA.

3 Pharmacometric Research and Development, GloboMax, A Division of ICON, Hanover,

MD, USA. 4 Clinical Pharmacology and Drug Dynamics, Forest Laboratories Inc., Jersey City, NJ,

USA.

∗ To whom correspondence should be addressed. Telephone: +1-201-4278044; e-mail:

[email protected]

© 2005 Springer Science+Business Media, Inc.

Shi et al.

INTRODUCTION Leflunomide is a disease modifying anti-rheumatic drug (DMARD), which effectively reduces the signs and symptoms of active rheumatoid arthritis (RA) in adults, while inhibiting joint damage and improving physical function (1). In clinical trials of adults, leflunomide (20 mg/day) demonstrated efficacy equivalent to that of methotrexate (7.5–15 mg/wk) and sulfasalazine (2 g/day) for improving individual signs and symptoms of active RA (2–5) and slowing disease progression as assessed by radiographic analysis of joint damage in the hands and feet (6–8). Leflunomide also demonstrated sustained efficacy and safety over a 5-year period in subjects continuing beyond 2 years of treatment (9). The pharmacokinetics (PK) of leflunomide have been characterized in healthy adults and adults with active rheumatoid arthritis (10–12). Following oral administration, leflunomide, a highly non-polar prodrug is rapidly and completely absorbed and is converted largely to its active metabolite A77 1726 (M1) during the first pass. Peak concentrations of M1 occur between 6 and 12 hr after dosing. M1 displays linear PK at the doses from 5 to 25 mg/d. M1 has a low volume of distribution (Vss = 0.13 L/kg) and it binds extensively to albumin (>99.3%) in healthy subjects; protein binding is linear at therapeutic concentrations. M1 has a long half-life (∼2 weeks) but can be much longer in some patients. The elimination is complex, involving both biliary and renal excretion, with biliary recycling contribution to the drug’s long half-life. M1 is also an inhibitor of CYP 2C9. M1 has anti-inflammatory and immunomodulatory properties and is responsible for essentially all anti-rheumatic activity in vivo. The primary mechanism of action for M1 is the selective inhibition of dihydroorotate dehydrogenase, the rate-limiting enzyme in de novo pyrimidine synthesis (13,14). Thus, the M1 metabolite of leflunomide inhibits the proliferation of activated CD4 lymphocytes by preventing pyrimidine generation required for DNA and RNA synthesis (15). Juvenile rheumatoid arthritis (JRA) is an inflammatory disease defined as a chronic, idiopathic arthritis with onset before the 16th birthday. Its worldwide prevalence has been estimated to be between 0.07 and 4.01 cases per 1000. Clinical manifestation has three major types: pauciarticular, polyarticular, and systemic. Polyarticular JRA (≥ 5 joints involved) affects approximately 30% of children with JRA (16). Although the etiology of JRA is unknown, many of the etiological factors associated with adult RA also are associated with JRA. Similarities in T cell, B cell, and macrophage abnormalities have been demonstrated (17). Evidence of complement activation and abnormal production and regulation of cytokines are common features of both diseases (17). The relationship between

Population Pharmacokinetics of the Active Metabolite of Leflunomide

dosage and clinical effect suggests that leflunomide should be given at a daily rate of 20 mg to obtain near maximum probability of clinical success in adult RA patients (60%) (18). At this dose, the median value of the average steady-state concentration in adults was about 34 mg/L. Two clinical trials were undertaken sequentially to evaluate the PK, efficacy and safety of leflunomide in the treatment of JRA. This study assessed data from these trials to establish a model describing the population pharmacokinetics (PPK) of M1 in pediatric patients with JRA, to examine the influence of demographic covariates on the PPK model, and to perform a trend analysis on the relationship between exposure and efficacy in order to establish appropriate dose adjustments for leflunomide in patients with JRA.

METHODS Patients and Study Design The PK data were analyzed from pediatric subjects with polyarticular course JRA who received oral leflunomide in one Phase I (Study I) and one Phase III (Study II) clinical trials. The protocols of both studies were approved by an Institutional Review Board (IRB) at each study center. Informed consent was obtained from all participants after the study was explained to the parents or guardians and the child, as appropriate. Study I, open-label, non-controlled, multicenter study included a 6-month treatment period with up to a 24-month extension phase in pediatric subjects (6–17 years) with active, polyarticular course JRA who had previously failed or were intolerant of methotrexate therapy. Leflunomide treatment was initiated with a loading dose equivalent to 100 mg/d in a standard adult with a body surface area (BSA) of 1.73 m2 (57.8 mg/m2 per day), administered on Days 1 through 3. Because of the restricted number of mg that can be reliably administered using the 10 mg tablets, BSA categories were used to determine the dose. Thereafter, maintenance doses of leflunomide were given daily, based on an equivalent of 10 mg/d/1.73 m2 (the low adult maintenance dose). Again, BSA categories were used to determine the exact dose to be given. In subjects without clinical response on or after 8 weeks based on Definition of Improvement responder analysis for juvenile rheumatoid arthritis subjects published by Giannini et al. (19), escalation to the equivalent of leflunomide 20 mg/d per 1.73 m2 BSA was allowed at the discretion of the investigator (Appendix A). Study II was a randomized, double-blind, parallel-group, 16-week trial comparing leflunomide to methotrexate in pediatric subjects (age 3–17 years) with active, polyarticular course JRA, irrespective of onset

Shi et al. Table I. Weight-Based Leflunomide Dosing Regimen in Study II Study (26) Actual body weight

Loading dose

Maintenance dose

< 20 kg 20–40 kg > 40 kg

100 mg/d × 1 100 mg/d × 2 100 mg/d × 3

5 mg/d 10 mg/d 20 mg/d

5 mg/d was given as 10 mg every other day.

type, who were na¨ıve to both methotrexate and leflunomide. A simplified treatment regimen was developed based on the PK results of Study I. Loading doses with 100 mg tablets and maintenance doses with 10 mg tablets were assigned based on actual body weight as described in Table I. Subjects unable to swallow the tablet(s) whole were permitted to crush the tablet(s) and mix it in applesauce or jam.

Pharmacokinetic Sampling and Bioanalysis Study I A baseline blood sample for PK analysis was taken prior to dosing during the screening visit. The first follow-up PK analysis was on Day 3 of treatment, after three loading doses of leflunomide. Subsequent PK analyses occurred on Weeks 4, 12, and 26 while the subject was on maintenance therapy. During these four occasions, serial blood samples were collected prior to the dose and at 2, 4, 8, and 24 hr following dosing. In addition, single blood samples were collected on several pre-defined event related occasions: (a) at the 16-week follow up visit for subjects not continuing in the extension phase, (b) prior to and 2 weeks after a dose increase or decrease. During the extension phase, a single blood sample was collected on Weeks 50 and 74. If a subject discontinued at or before Week 74, a single PK sample was also drawn 16 weeks after the subject discontinued. If there was a suspicion of a severe leflunomide-related adverse event, an extra blood sample was taken for M1 concentration measurements. The time since the last intake of leflunomide and the dose were documented. For each PK analysis during leflunomide treatment, the dose was to be administered orally, preferably at 8:00 a.m. following an overnight fast of 10 hr. Subjects were not permitted to eat for 2.5 hr after the dose of leflunomide, but they were allowed to drink milk or water. At the Week 42 follow-up visit, blood samples were collected prior to the 8:00 a.m. dose of leflunomide following an overnight fast of 10 hr.


Plasma samples were analyzed for M1 using a validated high-performance liquid chromatography (HPLC) method with UV detection and a limit of quantification (LOQ) of 100 ng/ml (0.1 µg/mL). Batchto-batch accuracy of the calibration standards and quality control samples ranged from 96.7% to 103.5% and 94.8% to 109.5%, respectively. Batchto-batch imprecision of the calibration standards and quality control standards ranged from 3.9% to 6.2% and 1.7% to 6.5%, respectively. Bioanalysis for Study I was performed at Aventis Pharma Deutschland GmbH. Study II In Study II, a pair of whole-blood samples was obtained for determination of plasma concentrations of leflunomide, M1, and 4-trifluoromethylaniline (TFMA) at Weeks 2, 4, 8, 12, and 16. One sample provided plasma for leflunomide analysis and the other sample provided plasma for the determination of M1 and TFMA. At their discretion, investigators could collect two pairs of PK samples at a single visit: one pair of blood samples was drawn before the dose and represented trough concentrations (elimination phase) from the preceding dose, and the other pair of samples were drawn at least 1 hr after the dose (absorption phase). Fixed sample collection times were not specified. The M1 concentrations in plasma were quantified using the same methods as Study I. Samples were analyzed for leflunomide concentration using a gas chromatography (GC) method with a nitrogen-selective detector and TFMA concentration using a GC method with mass-selective detection. Batch-to-batch accuracy of the calibration standards and the quality control samples (for M1) ranged from 98.2% to 101.2% and 101% to 104.5%, respectively. Statistical Procedures Data were analyzed by a nonlinear mixed-effect model using the NONMEM system (NONMEM version V Level 1.1, NONMEM Project Group, UCSF/GloboMax) to evaluate the population mean parameters, inter-subject and residual random effects, and covariate effects. Calculations were performed on a Pentium III 900 MHz computer with a Microsoft Windows 2000 operating system using the compiler Digital Visual Fortran V6.5. The first-order conditional estimation (FOCE) method with interaction was used. SYSTAT Version 10 (SPSS, Chicago) and S-PLUS Professional 6.1 (Insightful Corporation, Seattle) were used for data handling and for numerical and graphical analyses of the relevant NONMEM output (20).

Shi et al.

Individual PK parameter estimates were obtained from the best structural (base) PK model. Individual covariates were then screened by scatter plot matrices to determine if there was any relationship between them and the individual PK parameter estimates (colinearity). Any significant covariates identified were entered into the structural model to identify the optimal population model for describing the data. The performance of each model was evaluated by standard goodness-of-fit criteria and visual inspection of residual plots to guide selection of the final population model. The impact of a covariate on the PK of M1 was evaluated by means of the change in the objective function value (OFV). During the model building, an individual covariate was considered to improve the model fit significantly if its addition resulted in a reduction in the OFV by a factor greater than the critical value based on a chi square test with p = 0.05. The general modeling approach was based on the principle of parsimony; i.e., simpler models were chosen over more complex models when statistically justified (21). A previously established structural PK model for M1 concentrationtime data from adults was fitted to the M1 data from the pediatric population as the base model. The structural PK parameters were: ka (first-order input-rate constant representing overall rate of absorption of parent leflunomide, biotransformation of leflunomide to M1, and appearance of M1 in the circulation); CL/F (the total oral clearance); and V/F (apparent volume of distribution), where F is a product of the absolute bioavailability of parent leflunomide, the fraction of formation of M1 from leflunomide expressed as molar ratio between leflunomide and M1, and (1- fraction of M1 metabolized). Intersubject variability (random effects) on PK parameters was based on the assumption of a log-normal distribution (exponential error model). The nondiagonal elements of the variance-covariance matrix were set to zero. For a one-compartment model, random effects were modeled on CL/F, V/F and ka . For residual error (intrasubject variability), a series of error models was tested including additive error model, proportional error model and combined additive and proportional error model The best residual error model was selected based on OFV and goodness-of-fit plots.

Covariate Submodels The influence of individual demographic characteristics was examined by NONMEM stepwise regressions. The primary foci of the covariate screening were factors that are commonly used for dose adjustment


in pediatric practice: BSA, WT, and age. Once significant covariates were identified by trends in the scatter matrix plot, they were added to the base model sequentially and tested by NONMEM to determine if they were indeed statistically significant. The covariate with the strongest apparent correlation was entered first into the model. If a covariate was continuous in nature, an exponential covariate model was tested by adding one covariate at a time to the model in a median-normalized manner. For example, for the parameter CL/F and the covariate WT: TV(CL/F)i = θ1 · (WTi /WTmedian )θ2

(1)

where TV(CL/F)i represents the typical value of the clearance for the ith subject, θ1 represents the population typical parameter estimate at the median of the covariate (i.e., at WTmedian ), θ2 reflects the degree of correlation between the PK parameter and the covariate, and the WTi and WTmedian are the WT for ith subject and the median for the population, respectively. The medians in the pediatric population for age, weight, and BSA were 12 years, 40 kg, and 1.2 m2 , respectively. If a covariate was a dichotomous variable (e.g., sex), a step model was to be tested as follows: CL/F = θx + θ y · factor

(2)

where the factor is a discrete variable (0 = male, 1 = female). Model Qualification Because the two studies were conducted independently, examining the similarity of the model fits to the individual study datasets was used to evaluate and qualify the final model. The ability of the final population PK model to realistically mimic the observed data was investigated with a predictive check (22). This modelchecking technique is based on the premise that a model derived from and fitted to a set of observed data should produce similar data when used in a simulation mode. Monte Carlo simulations using the final “optimal” PPK model (Model 10), including final fixed effect and random effect parameters (inter-subject and residual variances), were conducted using NONMEM to create 500 replicates of the observed data set with identical sample collection time points and body weights (73 × 500=36500 subjects). Predicted median concentrations (50th percentile) and 95% predicted intervals (from the 2.5th to 97.5th percentile) of the simulated data were calculated at each sample collection time point. Observed data were plotted against these simulated medians and predicted intervals.

Shi et al.

Graphic explorations were conducted using the relationship between WT and CL/F and variability associated to predict Css values for M1 using the equation: Css,i = Dose/[(CL/F)i • exp(ηi )]

(3)

The 95% predication intervals were constructed by simulation of 2000 JRA patients with a uniform distribution of body weight from 10 to 80 kg. Acceptable dosage regimens should provide Css in the majority of patients falling within these intervals at appropriately partitioned WT groups. This process was repeated with visual assessment of the plots for various dosing regimens at different WT cuts (e.g. one cut at 20 kg or 30 kg) that used the available formulation strengths of leflunomide. To facilitate the visual assessment, the Css of the real patients in the two studies were plotted overlaying on the 95% prediction intervals using the proposed dosage regimens and individual posthoc CL/F estimates. Dose Determination Using the same simulation as described above, the validity of different dosing regimens on Css values in pediatric patients was further tested by comparing them to the Css distribution of adult patients receiving the standard regimen, i.e., 20 mg/d. The dose regimens tested included (a) the regimen from Study II, (b) a simplified regimen of 10 mg/d for pediatric patients 40 kg

10 mg/d 15 mg/d 20 mg/d

150

15 mg/d was given as 10 mg and 20 mg rotation every other day.

0

50

Css (mcg/ml)

100

Study I Study II

10

20

30

40

50

60

70

80

Body Weight (kg)

Fig. 4. Predicted steady-state concentrations (Css ) of M1 using the refined leflunomide dose regimen in children (circles), and the 95% prediction interval simulated by the model (shaded area). The data points are the Css of the real JRA patients in the two studies predicted using the refined regimens and their Posthoc CL/F estimates.

and mean values of M1 Css for the dosing regimen used in Study II (Table V). Simulation of Css values using a single leflunomide dose adjustment for a body weight 40 kg

10 mg every other day 8 6.1 30.6 12.6 14.5 7.2 0.50

10 mg/d 19 12.0 98.9 26.2 30.0 19.3 0.64

20 mg/d 20 8.9 86.4 36.7 38.9 20.4 0.52

C.V. = Coefficient of variation. Css was calculated as dose /(CL/F). One subject had no M1 data available. Her Css was calculated according to her body weight.

optimal regimens from the graphic exploration (Fig. 5, right panel). With this dosing regimen, median Css for M1 was around 34 µg/mL in each of the weight groups, matching the median Css previously observed in adult patients with rheumatoid arthritis who received leflunomide 20 mg daily in Phases II and III studies (10). Exposure to M1 in Pediatric Responders and Non-Responders Among the 47 pediatric subjects (one female subject had no M1 data and her Css was predicted based on the WT) treated with leflunomide for 16 weeks in Study II, 32 were categorized as responders and 15 were categorized as non-responders when assessed by JRA Definition of Improvement >30% (19). Comparison of M1 concentrations to response or non-response revealed a trend for lower exposures in the group of subjects who failed to respond to leflunomide (Fig. 6, left panel). The majority of subjects (80%) in the non-responder group had concentrations of M1 that were less than the median exposure in the responder group. Lighter subjects also tended to be less likely to respond to leflunomide (Fig. 6, right panel), but the trend was less pronounced than that for the M1 concentrations.

DISCUSSION The relationship between the Css of M1 and clinical effect previously determined in adult RA patients showed that the maximum probability of clinical success (60%) would be obtained by choosing a dose rate that maintains a Css above the target concentration of 13 mg/L for the majority

100

100

80

80 Body Weight (kg)

Average Css (µg/mL)

Shi et al.

60 40 20 0

60 40 20

r de on sp e r n No

r de on sp e R

0

er nd po s e nr No

r de on sp e R

Fig. 6. Steady-state concentration (Css ) of the M1 metabolite and body weight in pediatric responders and non-responders to leflunomide. Data were taken from a Phase III study in subjects with juvenile rheumatoid arthritis (26). Response was defined >30% improvement as defined by Giannini et al. (19). An outside value is defined as a value that is smaller than the lower quartile minus 1.5 times the interquartile range, or larger than the upper quartile plus 1.5 times the interquartile range (inner fences). These values are plotted with an asterisk. A far out value is defined as a value that is smaller than the lower quartile minus three times the interquartile range, or larger than the upper quartile plus three times the interquartile range (outer fences). These values are plotted with an open circle.

of the patients. To achieve the maximum probability of clinical success in 95% of patients treated with leflunomide, the dose rate would have to be adjusted to achieve a median plasma concentration of 30 mg/L (95% CI: 13; 67), requiring a daily dose of 16 mg leflunomide. The dose rate recommended is 20 mg daily, this dose rate should achieve the maximum probability of clinical success in 99% of patients. The plateau effect in the relationship between the Css and clinical effect shows that higher dosages of leflunomide would not increase the probability of clinical success to any significant degree (18). This present study assessed PK data collected during an uncontrolled Phase I trial (Study I) and a controlled Phase III trial (Study II) of leflunomide for the treatment of pediatric subjects with JRA to create a model for the PPK of the active metabolite, M1, and then use this model and in conjunction with efficacy and safety data obtained to determine appropriate dosing regimens in pediatric patients with JRA. Analysis of M1 concentrations and treatment response in Study II revealed a trend for lower exposures to M1 in the group of children who failed to respond to leflunomide. The majority of subjects (80%) in the non-responder group had exposures to M1 that were less than the median exposure in the responder group. A similar but less pronounced


trend was observed for lower WT and non-response to leflunomide. Those analyses suggest that the leflunomide doses in Study II were suboptimal among subjects who weighed less than 40 kg, providing supporting evidence that the concentration versus response relationship between pediatrics and adult patients is similar. Because of the common etiological factors and similar responses to the drug between adult and pediatric RA patients, it is reasonable to assume that the M1 concentration versus response relationship (efficacy and safety) is similar between the two populations. Therefore, the dosage regimens for JRA patients are based on the comparability of the exposure of M1 to that observed in adult RA patients. Chemical inhibition studies in human liver microsomes suggested P450 CYP 1A2 as the principal P450 isozyme responsible for leflunomide metabolism. In addition to CYP 1A2, studies with human recombinant CYP isozymes also indicated a role for CYP 2C19 and CYP 3A4 in leflunomide metabolism (23). The M1 is eliminated by further metabolism and subsequent renal excretion as well as by direct biliary excretion. After oral administration of leflunomide approximately 43% of the dose is excreted in urine, primarily as leflunomide glucuronides and an oxalinic acid derivative of M1, and 48% is excreted in feces, primarily as M1 (1). Considering that multiple pathways may be involved in the clearance of M1 and that it has an extremely long t1/2 , largely due to enterohepatic recycling, the PK of M1 in pediatrics cannot be reliably predicted from the PK in adults simply by normalizing a body size covariate (WT or BSA) in an empirical way. Hence, a sequential design was implemented in the program to first derive the PK information in JRA patients in a small scale and then to examine the efficacy and safety in a large scale with regimens recommended from the earlier study. In the final PPK model derived in JRA patients from Studies I and II, age and gender did not influence PPK, but body size correlated strongly with V/F and it correlated weakly with CL/F. The PPK model predicted a CL/F of 0.0254 L/h for M1 in a pediatric subject weighing 70 kg, which is consistent with the findings of previous PPK analyses in adults (10). The strong correlation that was observed between body size and V/F, reducing the intersubject variation in V/F from 35.9% to 18.6%, indicates that a loading dose regimen for pediatric patients should be adjusted, if indeed a loading dose is administered. Due to the very long half-life of M1 (∼ 2 weeks), a loading dose of 100 mg for 3 days was used in clinical studies of adult patients to facilitate the rapid attainment of steady-state concentrations of M1. However, since leflunomide is a chronic treatment for a chronic disease for which full response to therapy may take several months, the true

Shi et al.

benefit of the loading dose has been debated and has been shown to be associated with an increased risk of treatment discontinuation in adults (24,25). Furthermore, a disproportionate number of gastrointestinal adverse events, headaches, and cases of alopecia were reported within the first 4 weeks of leflunomide therapy in Study II, the Phase III study of JRA (26). Beginning leflunomide therapy for JRA without a loading dose may delay the initial onset of response in some patients, but it may permit a better opportunity for pediatric patients with JRA to tolerate early leflunomide therapy. In addition, the gradual accumulation provides an opportunity for a routine clinical monitoring of the safety during the accumulation. In pediatric patients with polyarticular course JRA, CL/F was correlated weakly to body size indicating that maintenance dose adjustments for body weight should be more modest than the 50% dose reductions often applied empirically in practice. Even with the most influential covariate identified (WT) included in the model, the intersubject variation in CL/F was reduced only slightly (from 53.6% to 50.4%, expressed as %CV), indicating that dosage reduction by half should only be required when the WT distribution is wide, i.e., in JRA patients and only for those with a low body weight. Either measure of body size (WT or BSA) provided similar prediction power on CL/F according to the OFVs, but WT can be measured more easily and more accurately, and it was selected to be the sole covariate in the final PPK model. Using the commercially available formulation strengths of leflunomide, the adjusted dosage regimen derivation was a trial and error process. First, the number of body weight groups was determined based on (a) the therapeutic index of the drug, and (b) the possible regimens given the available formulation strengths. Second, the partitioning of the body weight groups was determined by considering (a) the correlation between the CL/F and WT and the variability in CL/F, (b) the body weight distribution of JRA patients and (c) clinical convenience (20,30, and 40 kg). Finally, simulation was used to examine the performance of each possible dosage regimens. Through this process, the optimal weight-based protocol for dose adjustment was identified (Table IV). Although 15 mg/d would mean 10 and 20 mg on alternating days due to the lack of a 5 mg tablet, the inclusion of this regimen resulted in comparable exposure to M1 (median and range) across WT categories. This may trade some convenience for a gain in a better overall balance between efficacy and safety compared to other dose adjustment protocols that were tested. In conclusion, a one-compartment model with first-order input described M1 concentrations well. Age and gender of pediatric subjects did not influence M1 concentrations significantly, but body size (WT or BSA) correlated strongly with V/F and weakly with CL/F in children. An observed tendency of lower M1 concentrations to be associated with non-response to


leflunomide in Study II suggested that dose regimens in children need to provide adequate exposure to M1 to optimize the efficacy of leflunomide for JRA. The PPK modeling based on WT determined that the doses summarized in Table IV was predicted to provide exposure to M1 comparable to that achieved in adults with a dose of leflunomide 20 mg/d.

ACKNOWLEDGMENT We gratefully acknowledge Drs. W. Weber and L. Harnish, Aventis Germany, for their analysis of adult PPK and assembly of the PPK dataset of Study I; Dr. V. Strand, Division of Immunology, Stanford University, USA and Dr. K. Simpson, Medical Research, Aventis, USA for their expert clinical insight and advice.

APPENDIX A. Dose Regimen of Leflunomide Applied in Study I Oral treatment with leflunomide was initiated with a loading dose equivalent to 100 mg/day in the 1.73 m2 adult (57.8 mg/m2 ), administered on days 1 through 3. Body surface area (BSA), calculated according to the method of DuBois and DuBois (1916) BSA (m2 ) = 0.007184·[Weight (kg)]0.425 ·[Height (cm)]0.725 , categories were used to determine the loading dose as follows: BSA (m2 ) 0.45–0.60 0.61–0.75 0.76–0.90 0.91–1.05 1.06–1.20 1.21–1.35 1.36–1.50 > 1.50

Loading dose (mg/day)

Loading dose (tablets/day)

30 40 50 60 70 80 90 100

3 × 10 mg 4 × 10 mg 5 × 10 mg 6 × 10 mg 7 × 10 mg 8 × 10 mg 9 × 10 mg 1 × 100 mg

Patients began oral maintenance dose therapy on day 4 at the equivalent of 10 mg/day in the 1.73 m2 adult according to the following BSA categories BSA (m2 ) 0.45–1.00 > 1.00

Maintenance dose (mg/day)

Maintenance dose (tablets)

5 10

1 × 10 mg QOD 1 × 10 mg QD

Shi et al.

If a patient failed to respond to treatment after 8 weeks as defined by the Giannini criteria for clinically important improvement, dose escalations were permitted as follows BSA (m2 )

Maintenance dose (mg/day)

0.45–0.50 0.51–1.00 1.01–1.35

5 10 15

> 1.35

20

Maintenance dose (tablets) 1 × 10 mg 1 × 10 mg 1 × 10 mg 1× 10 mg 2 × 10 mg

QOD QD QD plus QOD QD

REFERENCES 1. Prescribing Information for Arava Tablets (Leflunomide). March 2004. 2. V. Strand, S. Cohen, M. Schiff, A. Weaver, R. Fleischmann, and G. Cannon, et al. Treatment of active rheumatoid arthritis with leflunomide compared with placebo and methotrexate. Arch. Intern. Med. 159:2542–2550 (1999). 3. P. Tugwell, G. Wells, V. Strand, A. Maetzel, C. Bombardier, and B. Crawford, et al. Clinical improvement as reflected in measures of function and health-related quality of life following treatment with leflunomide compared with methotrexate in patients with rheumatoid arthritis: sensitivity and relative efficiency to detect a treatment effect in a twelve-month, placebo-controlled trial. Arthritis. Rheum. 43:506–514 (2000). 4. J. R. Kalden, D. L. Scott, J. S. Smolen, M. Schattenkirchner, B. Rozman, and B. D. Williams, et al. Improved functional ability in patients with rheumatoid arthritis: long-term treatment with leflunomide versus sulfasalazine. J. Rheumatol. 28:1983–1991 (2001). 5. J. S. Smolen, J. R. Kalden, D. L. Scott, B. Rozman, T.K. Kvien, and A. Larsen, et al. the European Leflunomide Study Group. Efficacy and safety of leflunomide compared with sulfasalazine in active rheumatoid arthritis: a double-blind, randomized, multicenter trial. Lancet 353:259–266 (1999). 6. A. Larsen, T. K. Kvien, M. Schattenkirchner, R. Rau, D. L. Scott, and J. S. Smolen, et al. Slowing of disease progression in rheumatoid arthritis patients during long-term treatment with leflunomide or sulfasalazine. Scand. J. Rheumatol. 30:135–142 (2001). 7. D. L. Scott, J. S. Smolen, J. R. Kalden, L. B. A. van de Putte, A. Larsen, and T. K. Kvien, et al. Treatment of active rheumatoid arthritis with leflunomide: two year follow up of a double blind, placebo controlled trial versus sulfasalazine. Ann. Rheum. Dis. 60:913–923 (2001). 8. J. T. Sharp, V. Strand, H. Leung, F. Hurley, and I. Loew-Fridrich. Treatment with leflunomide slows radiographic progression of rheumatoid arthritis: results from three randomized controlled trials of leflunomide in patients with active rheumatoid arthritis. Leflunomide rheumatoid Arthritis Investigators Group. Arthritis Rheum. 43:495–505 (2000). 9. J. R. Kalden, M. Schattenkirchner, H. Sorensen, P. Emery, C. Deighton, and B. Rozman, et al. The efficacy and safety of leflunomide in patients with active rheumatoid arthritis: a five-year follow-up study. Arthritis. Rheum. 8:1513-1520 (2003). 10. B. Rozman. Clinical pharmacokinetics of leflunomide. Clin. Pharmacokinet. 41:421–430 (2002). 11. M. E. Weinblatt, J. M. Kremer, J. S. Coblyn, A. L. Maier, S. M. Helfgott, and M. Morrell, et al. Pharmacokinetics, safety, and efficacy of combination treatment with methotrexate and leflunomide in patients with active rheumatoid arthritis. Arthritis. Rheum. 42:1322–1328 (1999).

Population Pharmacokinetics of the Active Metabolite of Leflunomide 12. J. Li, H. W. Yao, Y. Jin, Y. F. Zhang, C. Y. Li, Y. H. Li, and S. Y. Xu. Pharmacokinetics of leflunomide in Chinese healthy volunteers. Acta. Pharmacol. Sin. 23:551–555 (2002). 13. J. P. Davis, G. A. Cain, W. J. Pitts, R. L. Magolda, and R. A. Copeland. The immunosuppressive metabolite of leflunomide is a potent inhibitor of human dihydroorotate dehydrogenase. Biochemistry 35:1270–1273 (1996). 14. L. D. Fairbanks, M. Bofill, K. Ruckemann, and H. A. Simmonds. Importance of ribonucleotide availability to proliferating T-lymphocytes from healthy humans: disproportionate expansion of pyrimidine pools and contrasting effects of de novo synthesis inhibitors. J. Biol. Chem. 270:29682–29689 (1995). 15. R. R. Bartlett, H. Anagnostopulos, T. Zielinski, T. Mattar, and R. Schleyerbach. Effects of leflunomide on the immune responses and models of inflammation. Springer. Semin. Immunopathol. 14:381–394 (1993). 16. J. T. Cassidy. Juvenile rheumatoid arthritis. In W. N. Kelley, E. D. Harris, S. Ruddy, and C. B. Sledge (eds.) Textbook of Rheumatology. Fifth Edition. WB Saunders, 1996, 1207–1254. 17. B. Nepom. The immunogenetics of juvenile rheumatoid arthritis. Rheum. Dis. No. Amer. 17: 825–842 (1991). 18. W. Weber and L. Harnisch: Use of a population approach to the development of leflunomide: A new disease-modifying drug in the treatment of rheumatoid arthritis. Hoechst Marion Roussel. COST B1 medicine, Geneva 1997, 239–244. 19. E. H. Giannini, N. Ruperto, A. Ravelli, D. J. Lovell, D. T. Felson, and A. Martini. Preliminary definition of improvement in juvenile arthritis, Arth. Rheum. 40:1202–1209 (1997). 20. S. L. Beal and L. B. Sheiner (Eds.) NONMEM Users Guides. GloboMax, a Division of ICON plc, Hanover, MD, 1989–1998. 21. E. I. Ette and T. M. Ludden. Population pharmacokinetic modeling: the importance of informative graphics. Pharma. Res. 12:1845–1855 (1995). 22. A. Gelman, J. B. Carlin, H. S. Stern, and D. B. Rubin. (eds.) Bayesian Data Analysis. Chapman & Hall, London, 1995. 23. A. S. Kalgutkar, H. T. Nguyen, A. D. Vaz, A. Doan, D. K. Dalvie, D. G. McLeod, and J. C. Murray. In vitro metabolism studies on the isoxazole ring scission in the anti-inflammatory agent lefluonomide to its active alpha-cyanoenol metabolite A771726: mechanistic similarities with the cytochrome P450-catalyzed dehydration of aldoximes. Drug Metab Dispos. 31:1240–1250 (2003). 24. G. Geborek, M. Crnkic, I. M. Peterson, and T. Saxne. Etanercept, infliximab, and leflunomide in established rheumatoid arthritis: Clinical experience using a structured follow up program in southern Sweden. Ann. Rheum. Dis. 61:793–798 (2002). 25. C. Siva, S. A. Eisen, R. Shepherd, F. Cunningham, M. A. Fang, and W. Finch, et al. Leflunomide use during the first 33 months after Food and Drug Administration Approval: Experience with a national cohort of 3,325 patients. Arthritis Rheumatism (Arthritis Care & Research) 49:745–751 (2003). 26. E. Silverman, L. Spiegel, and L. K. Jung, et al. Efficacy and safety of leflunomide (LEF) versus methotrexate (MTX) in the treatment of pediatric patients with juvenile rheumatoid arthritis (JRA). Late-Breaking Abstracts. Program and abstracts of the American College of Rheumatology 67th Annual Scientific Meeting, October 23–28, 2003; Orlando, Florida. Abstract LB2.