Exposure-Response Relationship of Tocilizumab, an Anti IL-6 Receptor Monoclonal Antibody, in a Large Population of Patients With Rheumatoid Arthritis

Transcription

1 al The Journal of Clinical Pharmacology / Vol XX No XX (20XX) 2012 XX X / Levi et JCP Pharmacokinetics and Pharmacodynamics Exposure-Response Relationship of Tocilizumab, an Anti IL-6 Receptor Monoclonal Antibody, in a Large Population of Patients With Rheumatoid Arthritis The Journal of Clinical Pharmacology 53(2) The Author(s) 2012 DOI: / Micha Levi, PhD 1, Susan Grange, PhD 2, and Nicolas Frey, PharmD 2 Abstract Relationships between tocilizumab exposure and response were evaluated using data from 4 phase III studies. Increased tocilizumab exposure was associated with improvements in Disease Activity Score using 28 joints (DAS28) and American College of Rheumatology (ACR) criteria and with a decrease in inflammation markers. A population pharmacokinetic/pharmacodynamic (PKPD) model was developed to describe data from 2 studies. An indirect-response model with a sigmoid E max (maximal drug effect) inhibitory drug effect on DAS28 production rate adequately described the relationship between tocilizumab concentration and DAS28. Mean minimum serum tocilizumab concentration at steady state was greater than the EC 50 (concentration at which 50% of E max on DAS28 is reached) with the 8-mg/kg dose but not with the 4-mg/kg dose. Simulations within a large rheumatoid arthritis (RA) population showed that DAS remission rates were 38% for 8 mg/kg and 24% for 4 mg/kg. Tocilizumab was more potent in RA patients with higher baseline interleukin-6 levels, but this effect was not clinically significant. Other covariates (eg, presence of neutralizing antitocilizumab antibodies) did not demonstrate a clinically meaningful effect on tocilizumab DAS28 dose-response relationships. These data support clinical observations that tocilizumab 8 mg/kg is more effective than 4 mg/kg in reducing disease activity. Keywords pharmacodynamics, rheumatology/immunology, clinical pharmacology, immunopharmacology, clinical trials Interleukin-6 (IL-6) is one of the key inflammatory cytokines involved in the development of rheumatoid arthritis (RA). 1,2 Interleukin-6 levels are elevated in the serum and synovial fluid of patients with RA compared with healthy persons. Serum IL-6 levels correlate with disease activity, 3,4 and the presence of IL-6 in joints correlates with chronic synovitis 5 and joint destruction. 6 Tocilizumab is a humanized monoclonal antibody that competitively inhibits IL-6 binding to its soluble and membrane-expressed receptors in a dose-dependent manner, thus blocking multiple IL-6 mediated proinflammatory activities. 7 Tocilizumab was effective and well tolerated in the treatment of RA in phase II and III studies in Japanese 8-10 and European 11 patients. Furthermore, the efficacy and safety of tocilizumab in moderate to severe active RA in various patient populations were confirmed recently in an extensive international phase III clinical program, in which tocilizumab was evaluated in several 24-week clinical trials Actemra (tocilizumab; Roche, Nutley, New Jersey) versus Methotrexate double-blind Investigative Trial In monotherapy (AMBITION) evaluated tocilizumab monotherapy in patients who had not received methotrexate in the 6 months before the study and who had not previously discontinued methotrexate therapy because of toxicity or lack of response. 12 The tocilizumab Pivotal Trial in methotrexate Inadequate responders (OPTION) and Tocilizumab in combination With traditional disease-modifying AntiRheumatic Drug (DMARD) therapy (TOWARD) studies investigated tocilizumab in combination with conventional DMARDs in patients with a previous inadequate response to DMARDs. 13,14 In addition, the Research on Actemra Determining efficacy after Anti-TNF failures (RADIATE) trial investigated tocilizumab in combination with methotrexate in patients with a previous inadequate response to 3 anti tumor necrosis factor (anti-tnf) agents. 15 Secondary objectives of these studies included investigation of the pharmacokinetic (PK), pharmacodynamic (PD), and immunogenicity parameters of tocilizumab in these patient populations. Supplementary material for this article is available on the journal s website at 1 Hoffmann La Roche, Inc, Nutley, NJ, USA 2 Hoffmann La Roche AG, Basel, Switzerland Submitted for publication 08-Jul-2011 ; accepted 28-Dec-2011 Corresponding Author: Micha Levi, Pharma Research and Early Development/Translational Research Sciences, Hoffman La Roche, Inc, 340 Kingsland St, Nutley, NJ , USA micha.levi@roche.com

2 152 The Journal of Clinical Pharmacology / Vol 53 No 2 (2013) The PKPD analyses, which use data from clinical trials with dissimilar clinical populations and trial designs, can provide useful integrated information about relationships between drug doses or concentrations and efficacy or safety. 16 In the present evaluation, graphical analyses were used to assess relationships between tocilizumab and key clinical end points and between tocilizumab and biomarkers of inflammation and tocilizumab mechanism of action. A population PKPD model was developed to investigate the relationship between tocilizumab exposure and efficacy, as measured by the Disease Activity Score using a 28-joint count (DAS28). Methods Patient Population Patients with moderate to severe active RA from 4 randomized, double-blind, placebo-controlled, phase III clinical trials of tocilizumab (AMBITION, OPTION, RADIATE, and TOWARD) were included in this analysis In these studies, patients received an intravenous infusion of tocilizumab 4 or 8 mg/kg or placebo (controls) as monotherapy or in combination with methotrexate (10-25 mg/wk) or other DMARDs every 4 weeks for up to 24 weeks. Rescue therapy consisted of tocilizumab 8 mg/kg within 8 weeks in controls in AMBITION, 12 tocilizumab 8 mg/kg from week 16 in OPTION 13 and RADIATE, 15 or an addition or change in DMARD or corticosteroid therapy from week 16 in TOWARD. 14 Data from all studies were pooled for the graphical PKPD analyses. Data from OPTION and TOWARD were used for the development of the PKPD model. PD Biomarkers Levels of markers of inflammation (C-reactive protein [CRP], erythrocyte sedimentation rate [ESR], and serum amyloid A [SAA]) and markers of tocilizumab target engagement (serum IL-6 and soluble IL-6 receptor [sil- 6R]) were recorded at screening, baseline, and weeks 2, 4, 6, 8, 12, 14, 16, 20, and 24. Serum IL-6 and sil-6r were measured using Quantikine Human IL-6 and sil- 6R Immunoassays (R&D Systems, Inc, Minneapolis, Minnesota). Clinical End Points 17 DAS28-ESR and American College of Rheumatology (ACR) scores 18 were recorded at baseline and at weeks 2, 4, 8, 12, 16, 20, and 24. For patients who discontinued the study prematurely, efficacy assessments were performed 4 weeks after the last study dose. Graphical Exposure-Response Analysis of PD Biomarkers and Clinical End Points Patients from AMBITION, OPTION, RADIATE, and TOWARD who had available PK parameters when treated with tocilizumab, who had completed the studies, and who did not receive rescue therapy were included in the exposure-biomarker efficacy population. Patients were divided into 3 cumulative tocilizumab area under the curve (AUC) categories: low (AUC < µg/ ml h), medium (AUC 100 and < µg/ml h), and high exposure (AUC µg/ml h). Individual cumulative AUC was calculated using a previously developed population PK model and individual dosing history. 19 Relationships between the main PD biomarkers, efficacy end points, and cumulative tocilizumab AUCs were assessed graphically by comparing the time course of PD parameters for the control group and for groups of low, medium, and high tocilizumab AUCs. Pharmacodynamic parameters analyzed included ACR20/50/70 responses, DAS28, markers of inflammation (CRP, ESR, and SAA concentrations), and markers of tocilizumab target engagement (serum IL-6 and sil-6r concentrations). PKPD Model Development Based on the graphical evaluation of exposure-response relationships, DAS28 was selected as the principal clinical end point for further PKPD modeling because (1) its time course was less variable than that for ACR response, and (2) a continuous variable such as DAS28 is more suitable to nonlinear mixed-effects regression analysis than a dichotomous variable such as ACR response. The PKPD model was developed using patients from OPTION and TOWARD. Patients with available PK parameters and control patients with at least 1 DAS28 value were included in the analysis. Methods and development of the PKPD models are included in supplementary materials. Effect of Neutralizing Antitocilizumab Antibodies. The influence of neutralizing antitocilizumab antibodies (yes or no), a categorical time-varying covariate, on PKPD parameters was investigated using the final model. The importance of change induced by this covariate was assessed graphically by comparing the population- and individual-weighted residuals measured in the presence and absence of neutralizing antitocilizumab antibodies by using a logistic regression. Model Evaluation Visual Predictive Check. The PKPD model was qualified by a visual predictive check (VPC). Parameters were randomly sampled from distributions of interindividual variability. Dependent values (DAS28 scores) were simulated based on sampled PKPD parameters and residual variability. For each individual, covariate values, PK parameters, dosing information, and sampling times were as in the original data set used for model development. One hundred simulated data sets were produced. A 90th percentile prediction interval was created by taking the 5th

3 Levi et al 153 and 95th percentiles of the simulated data at each observation time point. The observed data used for model development were then compared with the prediction interval, and the percentage of observed data points lying outside the interval was computed. To qualify the model, the percentage of data outside the prediction interval had to be ~10%. Posterior Predictive Check. The ability of the model to simulate EUropean League Against Rheumatism (EULAR) improvement criteria 20 and DAS28 remission (DAS28 <2.6) after 24 weeks with placebo or tocilizumab 4 or 8 mg/kg was assessed with a posterior predictive check (PPC). The original data set was simulated 100 times based on sampled parameter values and the same covariate values, dosing information, and individual PK concentration estimates from the original data set. The following criteria of response were calculated at week 24 from the original data set and for each replicate: (1) the proportion of patients with various EULAR improvement criteria (good, moderate, and nonresponders) and (2) the proportion of patients with DAS28 remission. The 5th and 95th percentiles of simulated criteria of response were calculated from all replicates and compared with observed criteria of response. Simulation of DAS28 Response Following 6 Months of Therapy The DAS28 time profiles in 108, 500 patients were simulated after tocilizumab 4 or 8 mg/kg was administered every 4 weeks for 24 weeks. Pharmacodynamic parameters were sampled from distributions of interindividual variability. Dependent values (DAS28 scores) were calculated based on sampled parameter values, residual variability, and the same covariate values in the original data set. Subsequently, proportions of patients with the various EULAR improvement criteria and DAS28 remission were calculated. This process was repeated 100 times, and mean and standard deviation values for the EULAR improvement criteria and DAS28 remission were reported. Results Graphical Analyses: Tocilizumab Exposure Versus Efficacy Parameters and Biomarkers The efficacy-biomarker population comprised 2243 patients who completed the 4 phase III studies. An exposure-response relationship was observed for ACR20/50/70 and DAS28 clinical efficacy end points, with clear differences between the lowest category of tocilizumab exposure (AUC < µg/ml h) and the other categories (Figure 1 ). Across all 4 studies analyzed, levels of markers of inflammation (CRP, ESR, and SAA) decreased with increasing tocilizumab exposure (Figure 2 ). Low levels of these biomarkers (ie, normalization) were sustained between infusions with increasing tocilizumab exposure. Tocilizumab exposure had an effect on other PD biomarkers (IL-6 and sil-6r). IL-6 levels increased initially and then decreased over time, whereas IL-6 fluctuations over time decreased with increasing tocilizumab exposure. Levels of sil-6r increased with higher tocilizumab exposure and did not normalize between consecutive infusions. Similar trends were observed for ACR, DAS28, and markers of inflammation for each clinical trial when analyzed separately (data not shown). PKPD Structural Model The PKPD population for model development comprised 12, 618 DAS28 observations from 1703 patients who had DAS28 data available in the OPTION and TOWARD trials. In general, similar values were observed for all continuous covariates (except duration of RA) across all studies included in the PKPD model (Supplementary Table S1). Approximately 80% of patients were female, with a similar female to male ratio across studies, and patients were predominantly white, again with a similar proportion of races across studies. The distribution of previous RA treatments (DMARDs, methotrexate, and corticosteroids) differed across studies, as would be expected given the different study designs. The structural PKPD model that best described the data was an indirect-response model with a sigmoid inhibition of DAS28 production (K in ) by tocilizumab. The following parameters were estimated: DAS28 first-order rate loss ( ), DAS28 at baseline (BASE), the maximal effect of tocilizumab on K (EC50 in ), the tocilizumab concentration at which 50% of the E max is reached (EC 50 ), the E max mode sigmoidicity coefficient (GAMMA), and a background effect of concomitant therapy with methotrexate or nonmethotrexate DMARDs (DMARD). Additional information on the structural model is provided in the supplementary material. Final Model The final PKPD model estimated population values for the non drug-related parameters BASE and of 6.8 DAS28 units and day 1, respectively. Population mean values for the drug-related parameters EC 50, Emax, and GAMMA were 3.7 µg/ml, 0.73, and 0.64, respectively. The population mean value of background therapy with DMARDs expressed in tocilizumab concentration units was 0.30 µg/ml. The E max effect on K in translates into a maximum decrease in DAS28 scores of 73%, using a typical initial DAS28 baseline score of 6.8. The following 8 covariate relationships were retained in the final model: an inverse relationship between EC 50 and log-transformed IL-6; higher E max in males; higher in blacks and whites combined compared with other

4 154 The Journal of Clinical Pharmacology / Vol 53 No 2 (2013) Figure 1. Effects of tocilizumab exposure on American College of Rheumatology (ACR)20/50/70 responses and Disease Activity Score using 28 joints (DAS28). Abbreviation: AUC, area under the curve. races; positive relationships between BASE and Health Assessment Questionnaire (HAQ), log-transformed IL-6, PAIN, and physician s global score of disease activity (VASP); and an inverse relationship between background DMARD effect in the control group and log-transformed IL-6. Inclusion of these 8 covariates in the final model reduced the objective function by points and decreased interindividual variability of BASE, E max, EC 50, DMARD, and by 42%, 32%, 22%, 15%, and 1.6%, respectively. A correlation term between EC 50 and E max was estimated in the final model. This term did not have a statistically significant effect; however, it enabled successful convergence of the covariance step. The population PD parameters of the final PKPD model are presented in Table 1. The magnitude of covariate effects on the PKPD parameters is summarized in Table 2. The final model goodness-of-fit plots did not reveal any major deficiencies (Supplementary Figure S1). Neutralizing Antitocilizumab Antibodies Eighteen patients had neutralizing antitocilizumab antibodies, and of these patients, 14 had PK samples available for analysis. In an analysis of plots of weighted residuals and individual-weighted residuals versus time, no obvious differences between patients with and without neutralizing antitocilizumab antibodies were observed (data not shown). Model Validation The internal validation using VPCs and PPCs demonstrated good agreement between observed data and model predictions. For VPCs, the proportions of observations above the 95th percentile and below the 5th percentile were 5.9% and 5.8% for the control group, 10.0% and 3.2% for the tocilizumab 4-mg/kg group, and 10.0% and 5.4% for the tocilizumab 8-mg/kg group (Supplementary Figure S2). The PPCs revealed that most of the observed response criteria for EULAR responses and DAS28 remission were within the 90% simulated interval ( Figure 3 ). Tocilizumab 4 mg/kg Versus 8 mg/kg (Model-Based Simulation) The model-based simulation of DAS28 scores after 6 months of tocilizumab 4 mg/kg versus 8 mg/kg in the same large RA population predicted that the mean fraction of patients in DAS28 remission would be 24% (4 mg/kg) versus 38% (8 mg/kg). Mean proportions of

5 Levi et al 155 Figure 2. Effects of tocilizumab exposure on C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and serum amyloid A (SAA) concentrations. Thin gray lines represent individual CRP, ESR, and SAA profiles. The smooth black line represents the median. patients with a EULAR good response would be 32% versus 48%. Corresponding proportions of patients with a moderate EULAR response would be 45% versus 38%, and proportions with no EULAR response would be 22% versus 13%. Simulations were undertaken to illustrate the clinical impact on DAS28 time course of the inverse relationships between baseline IL-6 level and EC 50 and baseline IL-6 level and DMARD background therapy. Predictions were performed for tocilizumab 8 mg/kg, for the effect of IL-6 on EC 50, and for standard-of-care treatment (ie, DMARD) for the effect of IL-6 on DMARD background therapy. To predict a covariate effect at the population level, PKPD parameters were fixed to population values, between-patient variability was set to zero, patients were assumed to have the same tocilizumab exposure (generated by population PK parameters) for the 8-mg/ kg dose, and patients were assumed to have the same set of covariate values expected for baseline IL-6. Figure 4 illustrates the impact of baseline IL-6 on median DAS28 time course by categories of baseline IL-6 level (low, medium, and high). The observed difference between

6 156 The Journal of Clinical Pharmacology / Vol 53 No 2 (2013) Table 1. Model Parameters of Final Population PKPD Model Parameter Unit Estimate RSE, % Fixed effects EC 50 μg/ml E max Day GAMMA Baseline DAS DMARD effect expressed μg/ml as TCZ serum concentrations Covariate effects Effect of log IL-6 on EC Effect of SEX on E max Effect of RACE on Effect of HAQ on BASE Effect of log IL-6 on BASE Effect of PAIN on BASE Effect of VASP on BASE Effect of log IL-6 on DMARD effect Random effects IIV EC 50 CV% a E max CV% a CV% a Baseline CV% a DMARD effect CV% a Correlation EC 50 -E max a Error model Additive DAS Abbreviations: BASE, baseline DAS28; CV, coefficient of variation; DAS28, Disease Activity Score using 28 joints; DMARD, disease-modifying antirheumatic drug; EC 50, concentration at which 50% of E max is reached; E max, maximum tocilizumab effect on ; GAMMA, sigmoidicity coefficient; HAQ, Health Assessment Questionnaire; IIV, interindividual variability; IL-6, interleukin 6;, first-order loss rate of DAS28; PAIN, global pain score; PKPD, pharmacokinetic/pharmacodynamic; RSE, relative standard error of estimate; TCZ, tocilizumab; VASP, physician s global score of disease activity. a RSE of variance. the 2 extreme categories of baseline IL-6 was around the known measurement error of DAS28 (0.60 DAS28 units). Discussion Increasing tocilizumab exposure decreased DAS28 scores, increased ACR responses, and decreased serum concentrations of inflammatory biomarkers (CRP, ESR, and SAA). At exposure levels corresponding to the highest dose of tocilizumab (8 mg/kg), a more sustained suppression of inflammatory biomarkers could be observed over a dosing interval compared with the observed fluctuations in exposure levels corresponding to lower dose tocilizumab (4 mg/kg). These results are consistent with those from individual clinical trials that evaluated both tocilizumab regimens. In OPTION 13 and RADIATE, 15 tocilizumab 8 mg/kg had greater effects on clinical end points (ACR responses and DAS28 remission) and inflammatory biomarkers than did tocilizumab 4 mg/kg. A clear relationship was also evident between serum tocilizumab concentration and biomarkers of tocilizumab target engagement (IL-6 and sil-6r). The observed increase in sil-6r level with increasing tocilizumab concentration is consistent with data in the literature 21 and reflects tocilizumab binding to sil-6r. The sil-6r bioanalytical assay measures free sil-6r, tocilizumab/sil- 6R complexes, and IL-6/sIL-6R complexes. The increase in sil-6r in serum after tocilizumab administration suggests that tocilizumab/sil-6r complexes clear slowly. The observed initial increase in IL-6 level after the first tocilizumab infusion is most likely because IL-6 is normally cleared via membrane-expressed IL-6R (mil-6r), and tocilizumab inhibits this process. The subsequent decrease in IL-6 after the next tocilizumab dose probably reflects reduced IL-6 production because RA disease activity is controlled by tocilizumab. A delay was noted between increased serum tocilizumab concentration and reduced DAS28 score. This indirect relationship was described successfully using an indirect-response model with inhibition of the DAS28 production rate K in. The typical serum tocilizumab concentration at which 50% of E max was reached (EC 50 ) was estimated to be 3.7 µg/ml and was associated with high interindividual variability of 170%. The maximum effect of tocilizumab on DAS28 production (E max ) was estimated to be 73%; using a typical initial DAS28 baseline of 6.8, this corresponds to a maximum reduction of 5 units. The effect of DMARD background therapy in the control group (expressed in tocilizumab concentration units) was equivalent to constant exposure to a serum tocilizumab concentration of 0.30 µg/ml, which translates to a 12% reduction in DAS28 at steady state (ie, a decrease of 0.80 DAS28 units from a typical baseline of 6.8). This effect represents only a small fraction of the total effect observed (for the 2 tocilizumab doses) on background DMARD therapy. The standard deviation of the residual error was estimated to be 0.68 DAS28 units. This value agrees with the known measurement error of DAS28 (0.60 DAS28 units). 20 The relationship between serum tocilizumab concentration and inhibition of K in is displayed in Supplementary Figure S3. After administration of tocilizumab 8 mg/kg, the mean minimum serum concentration at steady state was higher than the EC 50. In contrast, after tocilizumab 4 mg/kg, the mean minimum serum concentration at steady state was less than the EC 50. Baseline DAS28 scores increased with increasing baseline HAQ score, VASP, PAIN, and serum IL-6 concentration. Such positive correlations were expected because these covariates are markers of RA disease activity. The

7 Levi et al 157 Table 2. Covariates With a Statistically Significant Effect in the Final PKPD Model Statistically Significant Effect Relationship Covariate Range (min, max) Percent Change of PKPD Parameters From Typical Value (min, max) Log IL-6 on EC 50 EC 50 = 3.7 *(log(il-6* 1000)/9.9) , , +415 SEX on E max E max = 0.72 * 1.1: male E max = 0.72 * 1.0: Female/male 0/+11 female RACE on = * 1.0: black and white Black and white/asian and others 0/ 25 = * 0.75: Asian and others HAQ on BASE BASE = 6.8 *(HAQ/1.6) , , +2.6 Log IL-6 on BASE BASE = 6.8 *(log(il-6* 1000)/9.9) , , +4.4 PAIN on BASE BASE = 6.8 *(PAIN/60) , , +3.2 VASP on BASE BASE = 6.8 *(VASP/65) , , +5.8 Log IL-6 on DMARD effect DMARD effect = 0.30 *(log (IL-6* 1000)/9.9) , , Abbreviations: BASE, baseline DAS28; DMARD, disease-modifying antirheumatic drug; EC 50, concentration at which 50% of E max is reached; E max, maximum tocilizumab effect on ; HAQ, Health Assessment Questionnaire; IL-6, interleukin 6;, first-order loss rate of DAS28; PAIN, global pain score; PKPD, pharmacokinetic/pharmacodynamic; VASP, physician s global score of disease activity. Figure 3. Posterior predictive check: incidence of EUropean League Against Rheumatism (EULAR) categories and Disease Activity Score using 28 joints (DAS28) remission at week 24. Histograms represent the predicted distribution of the proportion of patients with a EULAR response and DAS28 remission at week 24 (from 100 simulated studies). Dashed vertical lines represent the 5th and 95th percentiles of the simulated criteria of response. The solid vertical line represents the observed criteria of response calculated from the original data set. As seen in the figure, most of the observed responses are within the 90% simulated interval.

8 158 The Journal of Clinical Pharmacology / Vol 53 No 2 (2013) Figure 4. Median Disease Activity score using 28 joints (DAS28) time course (per category of baseline interleukin 6 [IL-6] level) in terms of the relationship between baseline IL-6 level and tocilizumab concentration at which 50% of the E max is reached. observed difference between the 2 extreme categories of baseline IL-6 (low vs high; Figure 4 ) was around the known measurement error of DAS28 (0.60 DAS28 units); this effect therefore was considered to have minor clinical impact. The clinical impact of the relationship between baseline IL-6 level and DMARD background therapy on DAS28 time course also was found to be minor (data not shown). The tocilizumab EC 50 value was lower in patients with high IL-6 levels at baseline. This infers that tocilizumab is more potent in RA patients with high IL-6 levels, which may result from either marked overproduction of IL-6 or lower expression of mil-6r and thus slower clearance of IL-6. However, this effect has a minor clinical impact because the predicted difference in DAS28 between the patients with high or low baseline IL-6 was around the known measurement error of DAS28 (0.6 DAS28 units). Increased baseline IL-6 levels also were found to decrease the effect of background DMARD therapy in the control group. This relationship very likely is an artifact of the way the DMARD effect was accounted for in the PKPD model. The DMARD effect was expressed in serum tocilizumab concentration units and therefore was dependent on the population value for tocilizumab EC 50. The inverse relationship between baseline IL-6 and the DMARD effect possibly is a correction of the inverse relationship between baseline IL-6 and EC 50. The tocilizumab E max value was 11% higher in males than in females. For a typical initial DAS28 baseline of 6.8, this corresponds to a maximum reduction of 5.0 units in males and 4.6 units in females. It is unlikely that the model-based covariate analysis was gender biased because males and females were similarly represented in the data set. This difference is lower than the known measurement error of DAS28 20 and therefore was considered to have no clinical impact. The presence of neutralizing antitocilizumab antibodies did not affect the PKPD model outcomes. The DAS28 population PKPD model was used to compare the effects of tocilizumab 4 mg/kg versus 8 mg/kg on DAS28 scores during 6 months of treatment. Simulations of EULAR response criteria and DAS28 remission rates were performed in the same virtual RA population, composed of more than patients. The simulations assumed full compliance to assigned treatment and no patients dropping out of the study. Response rates in the simulations were higher than observed values because simulated (rather than actual) observed dosing histories were used. Consequently, the predicted effect is larger than the observed effect in the clinical studies. Nevertheless, the simulations are useful to compare the effect of tocilizumab 4 mg/kg versus 8 mg/kg and to capture our best understanding of the inter- and intrasubject variability. Model predictions clearly showed the superiority of tocilizumab 8 mg/kg over 4 mg/kg, especially regarding a greater rate of DAS28 remission (38% vs 24%) and a greater percentage of patients with a good EULAR response (48% vs 32%). In summary, this exposure-response analysis, by using either graphics or a population-modeling approach, supports the observation that tocilizumab 8 mg/kg is more effective than 4 mg/kg. The higher tocilizumab dose is associated with greater tocilizumab exposure, and such increased exposure is associated with greater improvement in DAS28 scores. No covariate had a clinical impact on the effect of tocilizumab on DAS28. Declaration of Conflicting Interests Support for third-party writing assistance for this manuscript was provided by F. Hoffmann La Roche Ltd. All authors are employees of F. Hoffmann La Roche Ltd. Funding This study was funded by F. Hoffmann La Roche Ltd. References 1. McInnes IB, Schett G. Cytokines in the pathogenesis of rheumatoid arthritis. Nat Rev Immunol. 2007;7: Lipsky PE. Interleukin-6 and rheumatic diseases. Arthritis Res Ther. 2006;8(suppl):S4. 3. Madhok R, Crilly A, Watson J, Capell HA. Serum interleukin 6 levels in rheumatoid arthritis: correlations with clinical and laboratory indices of disease activity. Ann Rheum Dis. 1993;52: Dasgupta B, Corkill M, Kirkham B, Gibson T, Panayi G. Serial estimation of interleukin 6 as a measure of systemic disease in rheumatoid arthritis. J Rheumatol. 1992;19: Sack U, Kinne RW, Marx T, Heppt P, Bender S, Emmrich F. Interleukin-6 in synovial fluid is closely associated with chronic synovitis in rheumatoid arthritis. Rheumatol Int. 1993;13: Kotake S, Sato K, Kim KJ, et al. Interleukin-6 and soluble interleukin-6 receptors in the synovial fluids from rheumatoid arthritis patients are responsible for osteoclast-like cell formation. J Bone Miner Res. 1996;11: Mihara M, Kasutani K, Okazaki M, et al. Tocilizumab inhibits signal transduction mediated by both mil-6r and sil-6r, but not by

9 Levi et al 159 the receptors of other members of IL-6 cytokine family. Int Immunopharmacol. 2005;5: Nishimoto N, Miyasaka N, Yamamoto K, et al. Study of active controlled tocilizumab monotherapy for rheumatoid arthritis patients with an inadequate response to methotrexate (SATORI): significant reduction in disease activity and serum vascular endothelial growth factor by IL-6 receptor inhibition therapy. Mod Rheumatol. 2009;19: Nishimoto N, Miyasaka N, Yamamoto K, Kawai S, Takeuchi T, Azuma J. Long-term safety and efficacy of tocilizumab, an antiinterleukin-6 receptor monoclonal antibody, in monotherapy, in patients with rheumatoid arthritis (the STREAM study): evidence of safety and efficacy in a 5-year extension study. Ann Rheum Dis. 2008;68: Nishimoto N, Hashimoto J, Miyasaka N, et al. Study of active controlled monotherapy used for rheumatoid arthritis, an IL-6 inhibitor (SAMURAI): evidence of clinical and radiographic benefit from an X ray reader-blinded randomised controlled trial of tocilizumab. Ann Rheum Dis. 2007;66: Maini RN, Taylor PC, Szechinski J, et al. Double-blind randomized controlled clinical trial of the interleukin-6 receptor antagonist, tocilizumab, in European patients with rheumatoid arthritis who had an incomplete response to methotrexate. Arthritis Rheum. 2006;54: Jones G, Sebba A, Gu J, et al. Comparison of tocilizumab monotherapy versus methotrexate monotherapy in patients with moderate to severe rheumatoid arthritis: the AMBITION study. Ann Rheum Dis. 2010;69: Smolen JS, Beaulieu A, Rubbert-Roth A, et al. Effect of interleukin-6 receptor inhibition with tocilizumab in patients with rheumatoid arthritis (OPTION study): a double-blind, placebo-controlled, randomised trial. Lancet. 2008;371: Genovese MC, McKay JD, Nasonov EL, et al. Interleukin-6 receptor inhibition with tocilizumab reduces disease activity in rheumatoid arthritis with inadequate response to disease-modifying antirheumatic drugs: the Tocilizumab in Combination with Traditional Disease-Modifying Antirheumatic Drug Therapy Study. Arthritis Rheum. 2008;58: Emery P, Keystone E, Tony HP, et al. IL-6 receptor inhibition with tocilizumab improves treatment outcomes in patients with rheumatoid arthritis refractory to anti tumour necrosis factor biologicals: results from a 24-week multicentre randomised placebo-controlled trial. Ann Rheum Dis. 2008;67: US Department of Health and Human Services, Food and Drug Administration, Center for Drug Evaluation and Research, Center for Biologics Evaluation and Research. Guidance for Industry: Population Pharmacokinetics. Drugs/Guidancecomplianceregulatoryinformation/Guidances/ Ucm pdf. Accessed November 21, Prevoo MLL, van t Hof H, Kuper H, van Leeuwen MA, van de Putte LBA, van Riel PLCM. Modified disease activity scores that include twenty-eight-joint counts. Arthritis Rheum. 1995:38: Felson DT, Anderson JJ, Boers M, et al. The American College of Rheumatology preliminary core set of disease activity measures for rheumatoid arthritis clinical trials. Arthritis Rheum. 1993:36: Frey N, Grange S, Woodworth T. Population pharmacokinetic analysis of tocilizumab in patients with rheumatoid arthritis. J Clin Pharmacol. 2010;50: van Gestel AM, Prevoo MLL, van t Hof MA, van Rijswijk MH, van de Putte LBA, van Riel PLCM. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis: comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism criteria. Arthritis Rheum. 1996;39: Nishimoto N, Terao K, Mima T, Nakahara H, Takagi N, Kakehi T. Mechanisms and pathologic significances in increase in serum interleukin-6 (IL-6) and soluble IL-6 receptor after administration of an anti-il-6 receptor antibody, tocilizumab, in patients with rheumatoid arthritis and Castleman disease. Blood. 2008;112: