CENTER FOR DRUG EVALUATION AND
`CENTER FOR DRUG EVALUATION AND
`RESEARCH
`RESEARCH
`
`APPLICATION NUMBER:
`APPLICATION NUMBER:
`125166
`125166
`
`PHARMACOMETRICS REVIEW
`PHARMACOMETRICS REVIEW
`
`AMG1044
`
`1
`
`AMG1044
`
`

`

`Eculizumab (Soliris®)
`
`Pharmacometrics Review
`
`125166/0000 .
`BLA;STN
`9/15/06
`Submission Date(s)
`3/17/07
`PDUFA Due Date
`Brand Name
`Soliris®
`Eculizumab
`Generic Name
`Rajanikanth Madabushi, Ph.D.
`Pharmacometrics Reviewer
`Pharmacometrics Team Leader Joga Gobburu, Ph.D.
`Primary Reviewer
`Jang-lk Lee, Pharm.D., Ph.D.
`Primary Review Team Leader Hong Zhao, Ph.D.
`Sponsor
`Alexion PharmaGeuticals
`Relevant IND(s)
`BB-IND -
`Submission Type
`Original BLA (NME)
`Formulation
`Solution for intravenous infusion
`Proposed indication
`Treatment of paroxysmal nocturnal hemogloburinuria
`Proposed Dosage and
`600 mg qwk x4, 900 mg qwk x1, then 900 mg q2wk
`Administration
`
`======================================================~
`
`Appears This Way
`On Original
`
`2/28/2007
`
`Raj Madabushi
`
`1
`
`2
`
`

`

`Eculizumab (Soliris®)
`
`Executive Summary
`Eculizumab is a humanized monoclonal antibody that binds to the human C5
`complement protein and inhibits terminal complement-mediated activation. In the
`present submission, the sponsor is seeking for the approval of eculizumab for the
`treatment of paroxysmal nocturnal hemoglobinuria (PNH), an acquired genetic
`deficiency of endogenous complement inhibitors on the surface of blood cells.
`In
`the present submission,
`the sponsor has used pharmacokinetic
`modeling to derive the pharmacokinetic parameters from the sparse trough and
`peak serum concentration data collected in the pivotal trial. Relationship
`between eculizumab activity as assessed by a serum complement hemolysis
`assay and eculizumab concentrations has also. been evaluated using direct
`PKIPD model.
`The summary of the review 1s:
`
`1) Based on the population pharmacokinetic modeling, the clearance of
`eculizumab for a typical PNH patient weighing 70 kg was 0.022 Llhr and
`the volume of distribution was 7 7 L. The half-life was 272 ± 82 hrs (mean
`±SO).
`2) The mean observed peak and trough serum concentrations of eculizumab
`by week 26 were 194 ± 76 IJg/mL and 97 ± 60 mcg/ml respectively.
`3) Based on the PKIPD modeling,
`the
`total serum (free and bound)
`required
`concentration of eculizumab
`to block 50% of
`terminal
`complement activation is roughly 43 IJg/mL (EC50).
`4) Treatment with Eculizumab with the proposed dosing regimen results in
`immediate (from 2109 ± 965 U/L to 652 ± 278 U/L by week 1) and
`sustained (326 ± 438 U/L by week 26) lowering of LDH levels.
`
`Signatures:
`
`'
`
`·ir'~\ /
`
`\. ..... ·'
`'
`'
`L-.
`<
`. ~
`Rajanik~hth Madabushi, Ph.D.
`Pharmacometrics Reviewer
`
`Office of Clinical Pharmacology
`
`j}~
`
`Jqgarao V. Gobburu, Ph.D.
`
`Pharmacometrics Team Leader
`
`Office of Clinical Pharmacology
`
`2/28/2007
`
`Raj Madabushi
`
`2
`
`3
`
`

`

`Eculizumab (Soliris®)
`
`Table of Contents
`Pharmacometrics Review .................................................................................. 1
`Executive Summary ............................................................................................. 2
`Table of Contents ................................................................................................ 3
`List of Tables ...................................... ~ .................... : ........................................... 3
`List of Figures ..................................................................................................... 4
`Labeling Changes ............................................................................................... 5
`Introduction ......................................................................................................... 7
`Sponsor's Analysis ............................................................................................. 8
`Objectives ......................................................................................................... 8
`Data ................................................................................................................... 8
`Methods ............................................................................................................ 9
`Covariate Analyses ........... · ............................................................................. 9
`Population Variability- Monte Carlo Simulation .......................................... 1 0
`Model Assessment with Data from the Phase 3 PNH Study (C04-001 ) ....... 1 0
`PKIPD Relationship ..................................................................................... 1 0
`Results ............................................................................................................ 10
`Pharmacokinetic modeling ........................................................................... 1 0
`Analysis of Covariates ....................... : ......................................................... 13
`Population Variability - Monte Carlo simulation .................................... : ...... 14
`Analysis of PK in the Pivotal Trial ................................................................ 15
`PKIPD Relationship ..................................................................................... 16
`Conclusions ..................................................................................................... 17
`Reviewer's Comments ...................................................................................... 18
`Reviewer's Analysis ......................................................................................... 19
`Objective ......................................................................................................... 19
`Introduction ..................................................................................................... 19
`· Data .................................................................................................................. t9
`Method ............................................................................................................ 19
`Results ............................................................................................................ 19
`Conclusion ...................................................................................................... 21
`
`List of Tables
`Table 1: Patient demographics for PKIPD model development (per sponsor
`PKIPD report) ..................................................................................................... · .. 9
`Table 2: Mean parameter predictions for the final pharmacokinetic model (Mass-
`dependent linear volume expansion) from single dose patients only .................. 11
`Table 3: Mean parameter predictions for the final pharmacokinetic model (Mass-
`dependent linear volume expansion) from the entire database .......................... 12
`Table 4: Cluster summary statistics for different disease states ........................ 14
`Table 5: Parameter mean estimates of the data in PNH patients from pivotal trial
`(N=40, 1-compartment model) ............................................................................ 15
`Table 6: Summary of the LDH levels on placebo during the pivotal trial (C04-
`001) ........................... , ......................................................................................... 20
`
`2/28/2007
`
`Raj Madabushi
`
`3
`
`4
`
`

`

`Eculizum;::~b (Soliris®)
`
`Table 7: Summary of the LDH levels on eculizumab during the pivotal trial (C04-
`001 ) .. -------------------------------------------------------------------------------------------------------------------20
`
`List of Figures
`Figure 1: Schematic representation of the final pharmacokinetic model (2-
`compartment with mass-dependent linear volume expansion) ---------------------------11
`Figure 2: Bivariate fit of (a) Observed by Individual predicted and (b) Observed
`by Mean predicted ...... ___ ..... ____ ......... ___ ....... ___ .. ___ ..... ___ ......................................... 12
`Figure 3: Bivariate fit of observed and predicted eculizumab concentrations for
`the single-dose and multiple-dose studies for the different diseases .................. 13
`Figure 4: Bi-plot of K-means Cluster Analysis of Parameter Distributions:
`Cluster Centers Linked to Indication ................................................................... 14
`Figure 5: Monte carlo simulation of eculizumab concentrations in 50 simulated
`PNH patients based on PNH patient mean parameters
`from
`the
`final
`pharmacokinetic model .... ______ ............................................................................. 15
`Figure 6: Bivariate fit of (a) Observed by Individual predicted and Mean
`predicted ............................. ___ ............................................................................. 16
`Figure 7:
`Emax model of inhibition of % hemolysis by eculizumab
`concentrations from patients with IMG, RA and PNH ......................................... 17
`Figure 8: Treatment with Eculizumab results in immediate and sustained
`lowering of LDH levels ........................................................................................ 21
`
`peats lh\S 'NOV
`AP On Or\g\no\
`
`2/28/2007
`
`Raj Madabushi
`
`4
`
`5
`
`

`

`:L Page(s) Withheld
`
`- - - Trade Secret I Confidential
`
`___ Draft Labeling
`
`- - - Deliberative Process
`
`fhtttlf?t/CO!IJefvlC
`f.-WJM
`
`w iJJJMJA uAa Cit 1!Uur!h?J~_____.L-1_
`
`6
`
`

`

`Eculizumab (Soliris®)
`
`Introduction
`Eculizumab is a humanized monoclonal antibody that binds to the human C5
`complement protein and inhibits terminal complement-mediated cell lysis and
`activation. In the present submission, the sponsor is seeking for the approval of
`eculizumab for the treatment of paroxysmal nocturnal hemoglobinuria (PNH).
`PNH
`is an acquired genetic deficiency of endogenous complement
`inhibitors on
`the surface of blood cells, referred to as PNH cells.
`The
`pathophysiology of PNH is directly linked to complement-mediated destruction of
`the susceptible PNH red blood cells (RBSs), which result in intravascular
`hemolysis, the primary clinical manifestation in all PNH patients and contributes
`to mortality in these patients.
`A single phase 3, double-blind, placebo-controlled, comparative pivotal
`clinical trial (C4-001, n=87) combined with a phase 3 open-label safety and
`efficacy trial (C04-002, n=97) form the basis of the present marketing application.
`Additional support for the efficacy and safety of eculizumab as a treatment for
`PNH patients is provided in the associated PNH extension trial (E05-001 ).
`The primary objective of the pivotal trial (C4-001) was to evaluate the
`in patients with
`transfusion-dependent
`safety and efficacy of eculizumab
`hemolytic PNH. This trial was a randomized, double-blind, placebo-controlled,
`multicenter study of eculizumab or placebo administered by intravenous (IV)
`infusion
`to 87 patients with hemolytic,
`transfusion-dependent PNH.
`Randomization was stratified according to the number of packed red blood cell
`(PRBC) units transfused within 1 year prior to Screening. Patients randomly
`assigned to the placebo group received 1 dose of IV placebo weekly for 5 weeks,
`then once every 2 weeks for 21 weeks for a total of 26 weeks of treatment.
`Patients randomly assigned to the eculizumab group received 600 mg of
`eculizumab IV once a week for 4 doses, followed by 900 mg eculizumab IV 1
`week later for 1 dose, then 900 mg eculizumab IV every 2 weeks for 21 weeks.
`There was a total of 26 weeks of treatment. The co-primary endpoints were
`hemoglobin stabilization and number of PRBC units
`transfused. For
`the
`hemoglobin stabilization endpoint, a 2-sided Fisher exact test was performed.
`For the number of units of PRBCs transfused, each patient's units of PRBCs
`transfused after randomization to Week 26 (Visit 18) were calculated and the
`primary analysis method applied was the Wilcoxon rank sum test. Secondary
`objectives were transfusion avoidance, hemolysis as measured by lactate
`dehydrogenase (LDH) area under the curve (AUC) during the treatment period
`from Baseline to week 26, and Functional Assessment of Chronic Illness Therapy
`fatigue (FACIT-Fatigue) scale changes from Baseline to week 26.
`(PO) of
`The pharmacokinetics
`(PK) and
`the pharmacodynamics
`eculizumab were studied in clinical trials for six different indications: paroxysmal
`nocturnal hemoglobinuria (PNH), idiopathic membranous glomerulopathy (IMG),
`rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), dermatomyositis,
`and psoriasis. Single dose studies of eculizumab were performed in RA and SLE
`patients, which are reviewed by or. Jang-lk Lee, Clinical Pharmacology Reviewer.
`Data from multiple dose studies were available from phase-11 studies conducted
`in patients with RA, IMG, and PNH. Also PKIPD data were collected in the
`
`2/28/2007
`
`Raj Madabushi
`
`7
`
`7
`
`

`

`Eculizumab (Soliris®)
`
`pivotal trial (C04-001 ).
`The present Pharmacometrics review focuses on
`population PKIPD reports generated from the multiple dose studies and the
`pivotal trial.
`
`Sponsor's Analysis
`
`Objectives
`• To describe the PK of eculizumab and develop a compartmental PK
`model suitable for population PK analysis of sparsely sampled eculizumab
`serum concentration data.
`• To describe the PK of eculizumab based on data from patients with a
`variety of diseases.
`• To identify clinically important and statistically significant covariates with
`respect to the PK parameters.
`• To explore predicted eculizumab concentration variability in PNH patients
`using the derived model parameter estimates from PNH patients in a
`simulation analysis.
`• To describe the relationship between eculizumab activity (PD) (as
`assessed by a serum complement hemolysis assay) and eculizumab
`concentrations, and to link the PD data to the PK data in a direct PKIPD
`model.
`• To assess PKIPD data collected from PNH patients with the developed
`models.
`
`Data
`Total concentrations (bound + free) of eculizumab in human serum samples (PK)
`were measured using a validated enzyme-linked immunosorbent assay (ELISA)
`method. The complement hemolytic activity in serum (PD) were measure using
`an ex vivo hemolytic assay.
`These PKIPD data were collected from patients diagnosed with 6 different
`diseases {RA, IMG, psoriasis, PNH, and SLE). In these studies, 6 dose levels
`and 6 schedules were used. PKIPD data from a total of 209 patients who
`participated in 5 clinical trials was used to develop the PKIPD models. These
`studies included C97-001, C97-002, C01-004, C02-001, and C99-004.
`In addition, PKIPD data from patients with a diagnosis of PNH who
`participated in the pivotal trial (C04-001) was subsequently analyzed with the
`models after they had been developed.
`The PKIPD data in the PNH patients was obtained following multiple
`doses at trough and 1-hour post-dose. The PKIPD sampling for the 11 PNH
`patients in the C02-001 study included 4 sets of trough and 1-hour postdose
`peak samples for each patient. These were obtained at the first dose, at 2 doses
`in the middle portion of the study, and at the last dosing interval. A mid-interval
`{1-week postdose) sample was also obtained in mid study. PK and PO samples
`in the C04-001 (N=40) were drawn at baseline. Continuing into the treatment
`
`2/28/2007
`
`Raj Madabushi
`
`8
`
`8
`
`

`

`Eculizumab (Soliris®)
`
`period, trough and peak samples were drawn at Weeks 1, 4, 6, 12, and 26. A
`trough sample was obtained at end of study. The dosing regimen studied in the
`PNH studies were same and are described under Introduction.
`The summary of the patient demographics for the population PKIPD
`analysis is shown in Table 1 below.
`
`Table 1: Patient demographics for PK/PD model development (per sponsor
`PK/PD report)
`
`N
`Min
`Max
`Median
`Mean
`SD
`
`Age
`(y)
`209
`19.0
`84.0
`52.0
`52.0
`13.56
`
`Height
`(em)
`198
`132.0
`195.0
`168.0
`169.2
`10.46
`
`Weight
`(kg)
`209
`48.5
`157.2
`84.0
`85.4
`20.43
`
`SCI"
`(mg!tlL)
`209
`0.5
`2.2
`0.9
`0.9
`0.27
`
`Dose
`(mglkg)
`209
`2.0
`16.8
`8.0
`7.9
`1.89
`
`Methods
`Compartmental models were fitted to the data to obtain the estimates of the
`population-derived PK parameters using the USC*PACK suite of software.
`Specifically, the IT2B as well as the NPAG (an improved version of the
`nonparametric expectation maximization fitter [NPEM]) were used. Other
`software and programs used to analyze the data were: SAAM II and SAAM II
`Population Kinetics, GNU G-77 and G-95 Fortran Compiler, ActivePerl 5.8.7,
`JMP® version 5.01 for Windows®.
`All data fitted was weighted by the reciprocal of that datum estimated
`variance. Weighting was accomplished by making the assumption that total
`observation variance was proportional to assay variance. A rough estimate of the
`assay error, as computed by the USC*PACK. assay procedure was used
`throughout this analysis. The analysis could also be performed with adaptive
`gamma (y), a scalar that multiplies the polynomial described above and is
`optimized with each cycle. The objective functions, as well as other measures of
`goodness-of-fit
`including mean error, mean squared error, and Akaike
`information criterion (aic) were used to construct and select the most appropriate
`model.
`
`Covariate Analyses
`Covariates of eculizumab PK parameter estimates were explored using a cluster
`analysis in all 209 patients. In addition, a 1-way analysis of variance (ANOVA)
`was performed for each variable: gender, race, age, height, weight, and
`estimated creatinine clearance (CCr) with each of the PK parameters from the
`final model for the 209 patients in whom adequate data were suitable for analysis.
`Each of the PK parameters was subjected to analysis for each of the variables
`
`2/28/2007
`
`Raj Madabushi
`
`9
`
`9
`
`

`

`Eculizumab (Soliris®)
`
`and for relationships with each other. The model was fit using JMP 5.01
`statistical discovery software from SAS Institute, Inc.
`
`Population Variability- Monte Carlo Simulation
`A Monte Carlo simulation of estimated PNH patient PK parameters were derived
`from the model parameter means and SOs from the 11 PNH patients studied in
`study C02-001 and fit with the final population PK model. A Monte Carlo
`simulation was done exploring a maximum variability of 2 SOs of all 5 PK
`parameters conducted randomly and simultaneously. Three dosing regimens
`were simulated: 600 mg given weekly 4 times followed by 600 mg Q2wks, 900
`mg given weekly 4 times followed by 900 mg Q2wks, and 600 mg given weekly 4
`times followed by 900 mg Q2wks beginning 1 week after the last 600-mg dose.
`
`Model Assessment with Data from the Phase 3 PNH Study (C04-001)
`The population variability simulation for PNH patients was updated with sparsely
`sampled PK data from 40 PNH patients who participated ill the C04-001
`(TRIUMPH) pivotal study. This updated data were analyzed with the models.
`
`PK/PD Relationship
`The relationship of the eculizumab concentrations with the PO measurement of
`hemolysis was explored using maximum PO effect (Emax) model of hemolysis
`inhibition and eculizumab concentrations. The data set used included paired
`eculizumab concentration and % hemolysis data from 209 patients: 11 with PNH,
`121 with RA, 71 with IMG, and 6 with SLE. The model was fitted using JMP 5.01
`software.
`·
`The PO effect of eculizumab binding to C5 was evaluated using an assay
`that measures the % hemolysis of antibody-coated RBCs. Antibody-coated
`chicken RBCs are very sensitive to lysis by the presence of the membrane attack
`the
`formation of C5b-9,
`the
`terminal product of the
`complex formed by
`complement cascade. These RBCs lyse when small amounts of C5b-9 deposit
`on the cell membrane and thus are a sensitive measure of complement lysis
`activity. This complement hemolysis assay served as the primary assay to
`determine the effectiveness of eculizumab in blocking terminal complement
`activity in serum of ·patients. The assay had a steep dose-response curve
`because of the sensitivity of th? antibody-coated RBCs.
`Hemolysis data was inverted and normalized by subtracting every datum
`·
`from the maximum value recorded for that particular patient.
`
`Results
`
`Pharmacokinetic modeling
`A total of 39 models were explored to characterize total eculizumab serum
`time profile. None of the simple linear models tested could
`concentration -
`describe the complete data set without significant individual bias. This included
`
`2/28/2007
`
`Raj Madabushi
`
`10
`
`10
`
`

`

`Eculizumab (Soliris®)
`
`regular 2- and 3-compartment models, which gave a fairly good overall fit of the
`data, but showed considerable bias in selected single-dose (data-rich) patients.
`The PK of total concentrations of eculizumab were, best characterized by a
`model with 2-compartments with a nonlinear element expressed as the central
`volume of distribution expanding as a linear function of the mass in the central
`compartment (LVF model) using the single dose data. Mass exchange was
`parameterized either by rate constants or by the corresponding clearance (CL)
`and volume terms (V). The schematic representation of the model is shown in
`Figure 1 below. The mean parameter predictions are shown in Table 2 below.
`Bivariate fits of the observed and the predicted by individual parameter" estimates
`and mean estimates are shown in Figure 2. Further results from the application
`of this model to the 4 patient populations are shown in Table 3.
`The results of
`the ·fitting for the single-dose and the multiple-dose data are shown Figure 3.
`
`Figure 1: Schematic representation of the final pharmacokinetic model (2-
`compartment with mass-dependent linear volume expansion)
`
`Differential equations:
`
`dX 1,1 =Rm-[
`·]·Xl+[CLd2]·x,,,
`CLd2
`V2.
`VI+(A+X<>1 )
`dt
`d.¥,,, = [
`] . X"' _ [ CLd2 ] . Xc.,
`CLd2
`VI+(.~ +X,,,)
`V2
`dt
`
`Xo
`
`INPUT
`
`Output:
`
`y ~ [
`)
`X( I)
`- Vl+(A+X(l))
`
`CL
`
`Legends:
`n:
`Compartment number
`Xo
`Dose
`X(n) Mass in Compru1ment n (at timet)
`Vn
`Vohune associated with Comprutment
`n
`CL
`CLd2
`
`Clearance (elimination)
`Inter compartmental c leamnce
`
`Vt +(A· X(lJ)
`
`Table 2: Mean parameter predictions for the final pharmacokinetic model
`(Mass-dependent linear volume expansion) from single dose patients only
`
`2/28/2007
`
`Raj Madabushi
`
`11
`
`11
`
`

`

`Eculizumab (Soliris®)
`
`Model and
`Statistic
`2CLVFCLI
`Mean
`SD
`Median
`%SD
`
`AUC.,.
`(bi"*JlglmL)
`
`CL
`(a1·ea)
`(mlJhi")
`
`CL
`(mL/ha-lkg)
`
`CL.u
`(mLihr/kg)
`
`v!
`(mLikg)
`
`A
`(mL/pg)
`
`v.
`(mlJkg)
`
`26161
`12772
`28756
`48.8
`
`0.2589
`0.0837
`0.2315
`32.3
`
`0.3391
`0.2724
`0.2350
`80.3
`
`0.8256
`1.3097
`0.4756
`158.6
`
`25.777
`24.751
`25.019
`96.0
`
`0.003821
`0.001059
`0.004175
`27.7
`
`19.0635
`12.1166
`17.8632
`63.6
`
`Figure 2: Bivariate fit of (a) Observed by Individual predicted and (b)
`Observed by Mean predicted
`
`Bee.ed "n lnd1v:;.ciual Pat;a:uet;er el!lti:uat.e.=~
`
`250 ,----------------~
`
`'1 = l.0637x + ':l..<:l~t4:
`
`200
`
`~
`" M 1so
`
`o
`
`<Po
`0
`o8o
`
`g
`!!'.
`" ~ En)
`
`8
`
`/
`
`o
`0
`oo
`
`0
`
`CD
`
`oo
`o
`0 0
`0 •••
`n 8 .··
`.•
`
`,1$·'
`/o!f
`.o· g
`«<
`••
`o,.q
`il-
`u ••• o
`o~··g·
`oq,: . .d~B
`
`•• 0
`
`u!bo
`
`0
`
`9~
`~ .~ g
`
`0 0
`
`0
`
`100
`
`100
`
`15!1
`
`.
`
`ISO:•
`
`Pn~dicted fEcu)
`
`(mcg/mL)
`
`Predict.ed [Ecu}
`
`(mcg/mL)
`
`Table 3: Mean parameter predictions for the final pharmacokinetic model
`(Mass-dependent linear volume expansion) from the entire database
`R..4.
`SLE
`PNH
`Total
`H\'IG
`N=ll
`N=121
`N=6
`N=209
`~=71
`SD
`SD
`SD Mean
`SD
`SD Mean
`i\Ifall
`Mt>an
`Mt>an
`OJO
`OJO
`0.333
`0.134
`0.41
`0.139
`0.25
`0.069
`0.117
`0.106
`0.174
`0.54
`0.59
`0.43
`0.21
`0.055
`0.156
`0.562
`0.241
`0.21
`
`CL (mL'llrikg)
`Cl<12
`(mL!hrikg)
`V2 (mL!kg)
`A (mLipg)
`
`V1 (mL!kg)
`
`149.5
`0.000
`7
`64.9
`
`44.92
`0.000
`21
`31.53
`
`26.5
`0.002
`0
`32.5
`
`7.37
`0.001
`42
`16.50
`
`42.3
`0.004
`2
`12.7
`
`9.68
`0000
`75
`8.54
`
`20.6
`0.004
`0
`18.7
`
`10.84
`0.0004
`2
`4.84
`
`77.5
`58.70
`0.002
`0.001
`9
`8
`31.66 . 31.32
`
`2/28/2007
`
`Raj Madabushi
`
`12
`
`12
`
`

`

`Eculizumab (Soliris®)
`
`fit of observed and predicted eculizumab
`Bivariate
`Figure 3:
`concentrations for the single-dose and multiple-dose studies for the
`different diseases
`
`3 ! 9 . . - - - - - - - - - - - - - - 7 1
`
`CI70D1&C87002
`RAand aLE
`
`l xo
`• 8 IJO
`
`!0
`
`3!0
`
`);()
`
`:!:~0
`
`! XC
`§ 1~0
`
`](4}
`
`!0
`
`y • t.0(62t + D.8D69
`Rl- 0.97~8
`
`!0
`
`l 0)
`
`l »
`Pr&dldBd
`
`X(i
`
`~~Co
`
`lC:l
`
`3l0
`
`~ m
`~ ~ m
`Prel1lctttd
`
`~
`
`CB'!I-DD.tl.
`IYO
`
`CD2-DD1
`PNH
`
`0
`
`0
`
`0
`
`;:o
`
`oro
`
`"2~
`
`! X->
`J:
`8 l!C·
`
`RC·
`
`0
`
`!0
`
`PreoldBtl
`
`~- ~.{133'•- :1.5367
`
`R'· os-:.n
`
`!0
`
`0
`
`0
`
`0
`
`!0
`
`,. .. tz.t?7lr:- ~.!~s.-4
`~l"-:J.:t!l~S
`
`Precuct&a
`
`Analysis of Covariates
`Patient$ with a diagnosis of IMG were found to be different from patients with the
`other diseases studied with regard to the PK parameter estimates when analyzed
`in a cluster analysis. The cluster analysis included all parameters estimated (CL,
`Cld2, V2, A, V1) with the number of allowed clusters set to the number of tested
`indications (IMG, RA, PNH, and SLE). The IMG patient group had higher total CL
`and Vd than did the other patient groups as shown in Figure 4 and Table 4.
`Statistically significant correlations were observed
`for a number of the
`covariate/parameter pairs, including body weight vs CL in both IMG and the other
`these covariate/parameter pairs appear to be
`disease
`indications. While
`statistically correlated, there were PO observations which demonstrated these
`correlations were not clinically relevant. Consequently no covariates, apart from
`IMG were incorporated in the final compartmental model.
`
`2/28/2007
`
`Raj Madabushi
`
`13
`
`13
`
`

`

`Eculizumab (Soliris®)
`
`Figure 4: Bi-plot of K-means Cluster Analysis of Parameter Distributions:
`Cluster Centers Linked to Indication
`
`5
`
`....
`
`3
`
`2 -
`
`1-
`
`0-
`
`N
`
`~
`
`-1-
`
`-2-
`
`-3 _,.
`
`- -
`... ,..-·· : .. -;.-~----.:.
`--;..,..... .... . .
`l -... -...
`.......... -
`--;-_....,
`
`.1
`
`-:>
`
`•.
`
`Pri" 1
`
`Cluster 1- IMG; Cluster 2-PNH; Clu&ter 3-RA; Cln~ter 4-SLE
`
`Table 4: Cluster summary statistics for different disease states
`
`Indication
`
`CL
`(mVht•/kg)
`
`CLd!
`(mVbt•/kg)
`
`v!
`(mLikg)
`
`A
`(mU~tg)
`
`Yt
`(mL/kg)
`
`IMG
`
`PNH
`RA
`SLE
`
`0.41±0.139
`
`0.21±0.174
`
`149.57±44.924 0.000699±0. 000207
`
`64.908±31.529
`
`0.26±0.066
`
`0.543±0.168
`
`29.97±8.094
`
`0.0019±0.00148
`
`32.51±16.289
`
`0.29±0.117
`
`0.542±0.155
`
`42.30±9.680
`
`0.0042±0.00074
`
`12.72±8.541
`
`0.304±0.106
`
`0.591±0.562
`
`20.57±10.835
`
`0.00397±0.00042
`
`18. 74±4.843
`
`Data expressed as mean± SD
`
`Population Variability- Monte Carlo simulation
`A Monte Carlo simulation was conducted exploring a maximum variability of 2
`SDs of all 5 PK parameters conducted randomly and simultaneously. The
`simulated PK parameters were derived from the model parameter means and
`SDs from the 11 PNH patients studied in study C02-001 and fit with the final
`pharmacokinetic model (2-compatment with mass-dependent linear volume
`expansion). Three dosing regimens were simulated: 600 mg given weekly for 4
`weeks followed by 600 mg Q2wks (Figure 5: 1st panel); 900 mg given weekly for
`4 weeks followed by 900 mg Q2wks (Figure 5: 2nd panel), and 600 mg given
`weekly for 4 weeks followed by 900 mg Q2wks beginning 1 week after the last
`600 mg dose (Figure 5: 3rd panel; dose schedule used to treat PNH patients).
`The serum concentrations in the simulations show acceptable variability, as they
`are relatively well conserved in the population. Also the simulations indicate that
`excessive accumulation does not occur in PNH patients.
`
`2/28/2007
`
`Raj Madabushi
`
`14
`
`14
`
`

`

`Eculizumab (Soliris®)
`
`Figure 5: Monte carlo simulation of eculizumab concentrations in 50
`simulated PNH patients based on PNH patient mean parameters from the
`final pharmacokinetic model
`
`Panel1
`
`Panel2
`
`[PBnel3
`
`Concentrations in J.lg/mL are plotted on the y-axis in scientific notation as 10
`raised to the exponential. The time frame is plotted on the x-axis in hours raised
`to the exponent.
`
`Analysis of PK in the Pivotal Trial
`A standard 1-compartmental model analysis on the C04-001 patients' sparsely
`sampled PK data was conducted. The summary of the parameter mean
`estimates from the compartmental analysis are shown in Table 5 below.
`
`Table 5: Parameter mean estimates of the data in PNH patients from
`pivotal trial (N=40, 1-compartment model)
`.. ---------·····--·····-·---······-······-···-·---------
`CL
`(mL/ht·ikg)
`0.311183
`0.125091
`40.20
`
`l\Jean
`so
`'~o SD
`J...tin
`
`(mJ../l;:.g)
`ll(U
`17.9
`16 2
`
`K.r
`(1/hr)
`
`000081"/
`
`271 7
`
`.30.0
`
`0_002793
`}v:Ieclian
`0 289969
`108 3
`I C-p = 1 CP:NCLI: 1-companment clearance par:uneterization IT2B method
`
`2/28/2007
`
`Raj Madabushi
`
`15
`
`15
`
`

`

`Eculizumab (Soliris®)
`
`A 1-compartment model with mass-dependent volume expansion model did not
`result in improved fit of the data; hence the parsimonious model was preferred.
`Bivariate fits of the observed and the predicted by individual parameter estimates
`and mean estimates are shown in Figure 6.
`
`Figure 6: Bivariate fit of (a) Observed by Individual predicted and Mean
`predicted
`
`PRJ!O
`3;:- -~-~-----~---·------- "~· · - - ,
`
`I
`
`, : ..
`
`~=-;_·
`
`Ii?R£0
`
`Pt·cdic::cd
`
`PK/PD Relationship
`The
`relationship between eculizumab concentrations and C5 activity as
`measured by a serum complement hemolysis assay was described using a
`simple Emax model. The fit of the data to the Emax model was acceptable
`(Figure 7) given the limits of the assay, number of points, and patients involved.
`The x-axis in Figure 7 represents eculizumab concentrations and the y-axis the
`inverted normalized % hemolysis in the paired samples. The samples were
`obtained at anytime after dosing had begun during the relevant study. The 2
`offsets seen on the y-axis are caused by the BLQ samples being set to zero or to
`.. The Emax was estimated to be 118% ± 27% hemolysis (95% Cl for
`-
`the mean is 114-123%). The E50 was estimated to be 43 ± 2.7 (Jg/ml (95% Cl
`for the mean is 39-48 (Jg/ml). Because the PK assay used to determine
`eculizumab concentrations in human serum measures both bound and free
`eculizumab levels, an overestimated EC 50 of 43 1-Jg/ml reflects the total serum
`level of eculizumab required to block terminal complement activation.
`
`2/28/2007
`
`Raj Madabushi
`
`16
`
`16
`
`

`

`Eculizumab (Soliris®)
`
`Figure 7: Emax model of inhibition of % hemolysis by eculizumab
`concentrations from patients with IMG, RA and PNH
`
`:-
`
`-.-
`
`. .
`.... .-...
`. . -.
`
`-.
`
`: ......
`
`-. ·-
`
`.· ~
`
`. . . . ...
`....
`- ..
`. . ....
`- _·,.
`
`. . •
`
`The x-nxis represents eculizmnab concentmtions. The y-axis represents the inveJted normalized% hemolysis in the
`pai1·ed samples.
`
`Conclusions
`• A 2-compartmental LVF PK model suitable for population PK analysis of
`sparsely sampled eculizumab serum concentration data was developed
`and validated.
`• The PK of eculizumab was adequately described by the selected model
`based on data from patients with a variety of diseases.
`• No clinically important covariates with respect to the PK parameters were
`found.
`• Population eculizumab concentration variability in PNH patients was
`explored using the derived model parameter estimates from PNH patients
`in a Monte Carlo simulation. In this analysis, no excessive accumulation of
`eculizumab was observed.
`• The relationship between eculizumab concentration and hemolytic activity
`was described using a simple Emax PKIPD model.
`the
`• PKIPD data collected
`from PNH patients were assessed with
`developed models and were found to fit well to accurately predict PK
`parameters.
`• A successful link between the mass-dependent linear volume expansion
`model and PD response (hemolytic activity) was described.
`
`2/28/2007
`
`Raj Madabushi
`
`17
`
`17
`
`

`

`Eculizumab (Soliris®)
`
`Reviewer's Comments
`The conduct of the population pharmacokinetic analysis, PK/PD analysis and the
`interpretation of the results were reasonable.
`The following are the comments:
`(i)
`•
`It is unclear as to the purpose of two different population analyses -
`analysis conducted from the data derived from SLE. RA, IMG and phase II
`PNH data - a two compartment model with mass-dependent linear volume
`expansion model and (ii) one compartment analysis of the data derived
`from the phase Ill pivotal trial in PNH patients.
`• One of the objectives of the report was to asses the models developed
`from the data in different disease states with the data from PNH patients
`from the phase Ill C04-001 study. Instead of testing the two compartment
`model developed, the sponsor chose to develop a different model.
`
`Appears This Way
`On Original
`
`2/28/200