Physiologically Based Pharmacokinetic
Model-Informed Drug Development for
Polatuzumab Vedotin: Label for Drug-Drug
Interactions Without Dedicated Clinical
Trials
Divya Samineni, PhD1, Hao Ding, MS1, Fang Ma, MS2, Rong Shi, PhD1, Dan Lu, PhD1,
Dale Miles, PhD1, Jialin Mao, PhD2, Chunze Li, PhD1, Jin Jin, PhD1,
Matthew Wright, PhD2, Sandhya Girish, PhD1, and Yuan Chen, PhD2
Abstract
Model-informed drug development (MIDD) has become an important approach to improving clinical trial efficiency, optimizing drug dosing, and
proposing drug labeling in the absence of dedicated clinical trials. For the first time, we developed a physiologically based pharmacokinetic (PBPK)
model-based approach to assess CYP3A-mediated drug-drug interaction (DDI) risk for polatuzumab vedotin (Polivy), an anti-CD79b-vc-monomethyl
auristatin E (MMAE) antibody-drug conjugate (ADC). The model was developed and verified using data from the existing clinical DDI study for
brentuximab vedotin, a similar vc-MMAE ADC. Analogous to the brentuximab vedotin clinical study, polatuzumab vedotin at the proposed labeled
dose was predicted to have a limited drug interaction potential with strong CYP3A inhibitor and inducer. Polatuzumab vedotin was also predicted to
neither inhibit nor induce CYP3A. The present work demonstrated a high-impact application using a PBPK MIDD approach to predict the CYP3A￾mediated DDI to enable drug labeling in the absence of any dedicated clinical DDI study. The key considerations for the PBPK report included in the
Biologics License Application/Marketing Authorization Application submission, as well as the strategy and responses to address some of the critical and
challenging questions from the health authorities following the submission are also discussed.Our experience and associated perspective using a PBPK
approach to ultimately enable a drug interaction label claim for polatuzumab vedotin in lieu of a dedicated clinical DDI study, as well as the interactions
with the regulatory agencies, further provides confidence in applying MIDD to accelerate the registration and approval of new drug therapies.
Keywords
antibody-drug conjugate, drug-drug interaction, drug labeling, model-informed drug development, physiologically based pharmacokinetics
A significant unmet medical need remains for patients
with transplantation-ineligible relapsed/refractory
(R/R) diffuse large B-cell lymphoma (DLBCL),
including those who experience autologous stem
cell transplantation treatment failure. Although 56%
to 60% of patients with DLBCL are curable with
standard frontline chemoimmunotherapy, 30% to
40% of patients remain refractory to or relapse
after treatment. Polatuzumab vedotin (Polivy) was
recently approved by the United States Food and
Drug Administration (FDA) and European Medicines
Agency (EMA) in combination with bendamustine
and rituximab for patients with R/R DLBCL after at
least 2 prior therapies.1,2
Polatuzumab vedotin (anti–CD79b–vc–monome￾thyl auristatin E [MMAE]) is an antibody-drug
conjugate (ADC) that contains a humanized im￾munoglobulin G1 antihuman CD79b monoclonal an￾tibody and a potent antimitotic agent, MMAE, linked
through a protease-labile linker, maleimidocaproyl￾valine-citrulline-p-aminobenzyloxycarbonyl (MC-VC￾PABC). The chemical structure is shown in Figure 1.
Polatuzumab vedotin selectively binds to the CD79b
portion of the B-cell receptor present on the surface
of malignant and nonmalignant B cells, triggering
internalization of the complex by the cell. After inter￾nalization, the linker is cleaved by lysosomal enzymes,
resulting in the intracellular release of MMAE, which
binds to tubulin to inhibit polymerization, trigger￾ing tumor cell death.3 Similar to typical monoclonal
1Clinical Pharmacology, Genentech, Inc., South San Francisco, California,
USA
2Drug Metabolism and Pharmacokinetics, Genentech, Inc., South San
Francisco, California, USA
Submitted for publication 13 June 2020; accepted 26 July 2020.
Corresponding Author:
Yuan Chen, PhD, Drug Metabolism and Pharmacokinetics, Genentech,
Inc., 1 DNA Way, South San Francisco, CA 94080
Email: [email protected]
Samineni et al S121
Figure 1. Chemical structure of polatuzumab vedotin. MC, maleimidocaproyl; MMAE, monomethyl auristatin E; PABC, p-aminobenzyloxycarbonyl; VC,
valine-citrulline.
antibodies, the antibody component of the ADC un￾dergoes catabolism via low-capacity, saturable pro￾cesses such as target-mediated drug disposition and
high-capacity, largely nonsaturable processes like non￾specific proteolytic degradation.3 Hence, the drug-drug
interactions (DDIs) involving the antibody component
of polatuzumab vedotin are at low risk. However,
the unconjugated MMAE formed from the catabolism
of polatuzumab vedotin is expected to behave like a
small molecule that could be metabolized and cleared
via cytochrome P450 enzymes (CYPs) and/or trans￾porters. Concomitant medications that are inhibitors
or inducers of the same metabolic isoenzymes and/or
transporters could alter the pharmacokinetics (PK)
of unconjugated MMAE impacting clinical outcomes.
In vitro data indicated that MMAE is a substrate of
CYP3A and also a weak competitive CYP3A inhibitor
with half -maximal inhibitory concentration (IC50) of
10 μM and a time-dependent CYP3A inhibitor with an
apparent maximum inactivation rate constant (kinact)
value of 0.10/min and an apparent inactivation con￾stant (KI) value of 1.12 μM.4 Although the circu￾lating unconjugated MMAE levels are relatively low
following administration of polatuzumab vedotin, with
mean maximum concentration (Cmax) of approximately
10 nM or 7 ng/mL at the therapeutic dose of 1.8 mg/kg,5
the clinical DDI risk still needs to be evaluated to
inform the label.6
Physiologically based pharmacokinetic (PBPK)
modeling has increasingly impacted various stages of
drug development, ranging from early compound
selection for first-in-human trials to supporting
dosing recommendations in product labeling. Based
on a recent publication by Grimstein et al,7 the
majority of PBPK applications fall into the category
of predicting DDI potential. However, until now there
has been no published evidence on using a PBPK
approach to assess the DDI liability for ADCs. We
have previously developed a PBPK model to predict
the CYP3A-mediated DDI risks for a vc-MMAE
containing ADC.8 The model’s predictability was
qualified using the available data from a clinical
DDI study9 for brentuximab vedotin (ADCETRIS,
CD30 monoclonal antibody covalently linked through
MC-VC-PABC to MMAE), which contains a mc￾vc-MMAE linker/payload identical to polatuzumab
vedotin. The developed PBPK model was able to
adequately describe the observed clinical DDI, in
which brentuximab vedotin did not affect the PK of
midazolam, a sensitive CYP3A substrate. Concomitant
administration of rifampicin (a strong CYP3A inducer)
and ketoconazole (a strong CYP3A inhibitor) did not
alter the PK of the ADC measured as conjugated
antibody. However, exposure of unconjugated MMAE
was reduced by ∼46% with rifampicin and increased
by ∼34% with ketoconazole coadministration.8,9
Consequently, patients who are receiving strong
CYP3A inhibitors concomitantly with brentuximab
vedotin will be closely monitored for adverse
reactions.10
S122 The Journal of Clinical Pharmacology / Vol 60 No S1 2020
Figure 2. Elucidation of the application of the acMMAE-MMAE-linked PBPK model for the prediction of DDI potential for polatuzumab vedotin.
acMMAE, antibody-conjugated monomethyl auristatin E; ADC, antibody-drug conjugate; ADME, absorption, metabolism, distribution, and excretion;
CL, clearance; DDI, drug-drug interaction; DI, drug interaction; MMAE, monomethyl auristatin E; PBPK, physiologically based pharmacokinetic; PK,
pharmacokinetics; vc, valine-citrulline; Vss, volume of distribution at steady state.
As polatuzumab vedotin contains the same linker￾drug (vc-MMAE) combination as brentuximab ve￾dotin, it is conceivable that a PBPK approach,
qualified using brentuximab vedotin clinical DDI data,
can be used to predict a CYP3A-mediated DDI for
polatuzumab vedotin in place of a dedicated clinical
DDI study and subsequently inform the drug label. Ac￾cording to recently issued FDA and EMA draft PBPK
guidelines,11,12 our strategy for implementing a PBPK
model-based approach to inform the polatuzumab ve￾dotin DDI label in lieu of a dedicated clinical DDI
study was considered a high-impact application. Fur￾thermore, no prior knowledge for the implementation
of such an approach was evident for an ADC platform.
In this article, we present our PBPK model-informed
drug development (MIDD) approach to enable po￾latuzumab vedotin drug labeling with the objectives
outlined as follows: (1) to develop a PBPK model for
polatuzumab vedotin to predict the magnitude of drug
interaction for polatuzumab vedotin with the known
CYP3A probe substrate, inhibitor, and inducer; (2) to
submit the PBPK assessment package in the Biologics
License Application (BLA)/Marketing Authorization
Application (MAA) submission to inform the drug
label in the absence of a dedicated DDI study; and (3)
to address the regulatory queries/comments following
the submission of the PBPK analyses. Given that
many ADCs share the same payload and linker, it is
conceivable that the PBPK approach presented here can
be used to predict the DDI risk for a new ADC with dif￾ferent payload/linker moieties following a clinical DDI
assessment for at least 1 ADC within the same class of
payload/linker chemistries. Ultimately, the perspective
and experience shared in this work can further increase
our confidence in the high-impact PBPK applications
to facilitate development of clinical recommendations
around concomitant medication use for new molecular
entities without requiring unnecessary clinical DDI
studies.
Methods
The Simcyp population-based ADME simulator (ver￾sion 12, release 2) was used to construct the PBPK
model and perform the PK and DDI simulations.
The PBPK modeling approach to assess MMAE-based
drug interaction potential for polatuzumab vedotin
is elucidated in Figure 2. An antibody-conjugated
MMAE (acMMAE)-MMAE PBPK model was de￾veloped using in silico, in vitro MMAE data, and
in vivo clinical PK data from anti-CD22-vc-MMAE
ADC, an in-house vc-MMAE ADC. Then the PBPK
model was verified using clinical PK and DDI data
for brentuximab vedotin, the only vc-MMAE ADC
that has reported results from a dedicated clinical DDI
study to date and whose data were not included in the
development of the PBPK model. Last, the verified
PBPK model was used to predict MMAE-based drug
interaction between polatuzumab vedotin and other
drugs, with MMAE considered a victim (interaction
with ketoconazole, rifampicin) or perpetrator (interac￾tion with midazolam). Prior to the DDI prediction, the
model was verified for its prediction of the observed
clinical PK data for polatuzumab vedotin.
PBPK Model Development and Verification
PBPK model development and verification were elab￾orated in greater detail in our earlier publication.8
Briefly, a PBPK model linking unconjugated MMAE to
acMMAE as metabolite-to-parent drug was developed
using in silico, in vitro, and clinical data from an
in-house ADC (anti-CD22-vc-MMAE ADC) using a
mixed “bottom-up” and “top-down” approach in the
Simcyp population-based ADME simulator. Model
Samineni et al S123
Table 1. PK Parameters for the acMMAE Used in PBPK Model
Simulation Following the Administration of Polatuzumab Vedotin
Parameter Value Source
CL 38.2 mL/day/kg,a 19.4 mL/day/kgb NCA of clinical datac
Vss 0.09 L/kg NCA of clinical datac
Vsac 0.04 L/kg Simcyp best fitd
kin/kout 0.00727/h/0.00249/h Simcyp best fit
acMMAE, antibody-conjugated monomethyl auristatin E; CL, clearance;
CLint, intrinsic clearance; kin, rate constant from systemic compartment
to SAC; kout, rate constant from SAC compartment to the systemic
compartment; MMAE, monomethyl auristatin E; NCA, noncompartmental
analysis; PK, pharmacokinetic; Vsac, apparent volume associated with single
adjusting compartment; Vss, volume of distribution at steady state.
Minimal PBPK distribution model. The partition coefficient Kp value of liver
was 0.1. Vsac (input as 0.08 L/kg), kin, and kout were adopted from a previously
developed and verified PBPK model. The input of CL was calculated using a
Simcyp retrograde model, as CLint = 0.000463 and 0.000233 μL/min/pmol in
the enzyme kinetic panel and directed to the metabolite MMAE formation
via a randomly selected P450, CYP2E1, for the polatuzumab vedotin doses of
1.8 and 2.4 mg/kg. Because CL and Vss from the clinical study were used as
inputs, there are no other physiochemical parameters entered into the model
for acMMAE.
aInput value of CL from NCA in clinical study DCS4968g at the 1.8-mg/kg
dose.
bInput value of CL from NCA in clinical study DCS4968g at the 2.4-mg/kg
dose.
cPK parameters from NCA for the acMMAE analyte in the polatuzumab
vedotin clinical study DCS4968g at the 1.8- and 2.4-mg/kg doses.
dOutput value from best-fitted model simulation (inputted as 0.08 L/kg).
development included 2 parts: the acMMAE PBPK
submodel and the MMAE PBPK submodel. The final
assembled PBPK model treats acMMAE as a parent
drug that leads to the formation of unconjugated
MMAE (modeled as a metabolite formed from acM￾MAE) associated with potential DDIs. The PBPK
model was verified using the clinical PK and DDI data
of brentuximab vedotin (containing the same linker￾drug combination, ie, vc-MMAE as anti-CD22 ADC),
which was not used for PBPK model development. The
simulated results were compared against the observed
clinical PK and DDI data of brentuximab vedotin.
The model development and verification process are
illustrated in Figure 2 and detailed in Chen et al.8
PBPK Model Application
PBPK Model for Polatuzumab Vedotin. The PBPK
model linking acMMAE to unconjugated MMAE
was employed for the prediction of PK and CYP3A￾mediated DDI potential for the unconjugated MMAE
analyte of polatuzumab vedotin. The key parameters
used in the PBPK model simulation for acMMAE and
MMAE following the administration of polatuzumab
vedotin are listed in Table 1 and Table 2, respectively.
Simulation of Polatuzumab Vedotin PK. Before the DDI
prediction for polatuzumab vedotin, the model’s per￾formance in capturing polatuzumab vedotin PK was
verified by comparing the simulated acMMAE and
unconjugated MMAE concentration-time profiles with
the observed clinical PK data obtained following the
intravenous administration of polatuzumab vedotin in
a phase 1 study (DCS4968g; NCT01290549) for the
patients with R/R non-Hodgkin’s lymphoma at the
clinically relevant dose of 1.8 mg/kg and higher, that
is, 2.4 mg/kg.
The PK model simulations used randomly selected
individuals aged 20 to 50 years old and a sex ratio of 1:1
from sim-healthy volunteers. The dose regimen and trial
size used in the simulations were matched to clinical
study DCS4968g. Briefly, under fasted conditions, a
total of 10 trials of 6 subjects at the 1.8-mg/kg dose
(acMMAE-equivalent dose of 2.3 mg at average body
weight of 70 kg) and 10 trials of 44 subjects at the
2.4-mg/kg dose (acMMAE-equivalent dose of 3 mg
at average body weight of 70 kg) of polatuzumab
vedotin infused over 90 minutes on day 1 at 9:00 am
were simulated. The PK data were collected over a
21-day duration following the intravenous infusion in
the simulation. The number of points in the simulation
toolbox of Simcyp was set to 200.
The acMMAE-equivalent doses corresponding to
1.8 and 2.4 mg/kg of polatuzumab vedotin were con￾verted using the average drug-to-antibody ratio (DAR)
from the following equation:
acMMAE equivalent dose = (dose/MWADC) × DAR
× MWMMAE × average body weight
where MWADC represents the molecular weight of po￾latuzumab vedotin, which is 145 001 Da; and MWMMAE
represents the molecular weight of the unconjugated
MMAE, which is 718 g/mol. The dose of ADC was con￾verted to acMMAE-equivalent dose using an average
DAR of 3.7; average body weight was 70 kg.
Prediction of DDI Potential for Polatuzumab Vedotin. The
polatuzumab vedotin PBPK model described above
was then used to predict the DDI between polatuzumab
vedotin at a clinical dose of 1.8 mg/kg and CYP3A
substrates (midazolam) and perpetrators (ketoconazole
and rifampicin). The in vitro CYP3A inhibition data
for MMAE were entered into the model to simulate the
DDI in which MMAE served as a CYP3A inhibitor
(Table 2). For the DDI involving MMAE as a CYP3A
victim drug, the contribution of CYP3A to MMAE
clearance (CL; fm,CYP3A) was entered as CLint-CYP3A
enzyme kinetics. The fm,CYP3A value of 0.4 for MMAE
was adopted from a previously verified model using
brentuximab vedotin clinical DDI data through sensi￾tivity analysis (as elaborated further in Chen et al).8
S124 The Journal of Clinical Pharmacology / Vol 60 No S1 2020
Table 2. Input Parameters for Unconjugated MMAE in Simcyp PBPK Model
Parameter Value Reference
MW 717.98 g/mol BLA 125388 and 1253994
clogP 2.6 In silico calculateda
Compound type Monoprotic base
cpKa 8.08 In silico calculateda
B/P ratio 1.45 In-house data
fu plasma 0.178 BLA 125399
Vss 8.4 L/kg Simcyp-predicted minimal distribution model used
Vsac 2.0 L/kg Simcyp best fit−adopted from the verified Simcyp model
kin/kout 8 × 10−4/h/1 × 10−8/h Simcyp best fit — adopted from the verified Simcyp model
CLint-CYP3A 0.04 μL/min/pmol Calculated using Simcyp retrograde model to achieve fm,CYP3A of 40% of total
CL (8 L/h, adopted from the verified model)
CLbiliary 2.15 μL/min/106 Simcyp retrograde model calculated to account for about 50% of total CL
(8 L/h, adopted from the verified Simcyp model
CLadditional, HLM 1.15 μL/min/mg Calculated by Simcyp retrograde model to account for rest of total CL (8 L/h)
CLR 0 Assumed based on clinical information
CYP3A4/5 reversible inhibition Ki = 5 μM EMA Assessment Report17
fumic 0.985 Simcyp calculated
Time-dependent CYP3A4/5 inhibition kinact = 0.10/min KI = 1.128 μM EMA Assessment Report
fumic 0.971 Simcyp calculated
BLA, Biologic License Application; B/P, ratio of concentration of drug in blood to plasma; CL, clearance; CLadditional, HLM, additional clearance from human liver
microsomes; CLbiliary, biliary clearance; CLint, intrinsic clearance; CLint-CYP3A, intrinsic clearance for CYP3A; CLR, renal clearance; cpKa, calculated logarithmic acid
dissociation constant; clogP, calculated logarithm of octanol: water partition coefficient; CYP, cytochrome P450; EMA, European Medicines Agency; fu, fraction
unbound in plasma;fumic, unbound fraction in microsomes;Ki, reversible inhibition constant, concentration causing half maximal inhibition;KI, apparent inactivation
constant, concentration causing half maximum inactivation; kin, rate constant from systemic compartment to SAC; kinact, apparent maximum inactivation rate
constant; kout, rate constant from SAC compartment to the systemic compartment;MMAE,monomethyl auristatin E;MW,molecular weight; PBPK, physiologically
based pharmacokinetic; Vsac, apparent volume associated with a single adjusting compartment; Vss, volume of distribution at steady state. aCalculated in MoKa, version 1.1.0 (Molecular Discovery, Perugia, Italy).
The trial design and sizes used for the DDI sim￾ulations are described in Figure 3, similar to those
described in the brentuximab vedotin clinical DDI
study.9 Briefly, the interaction between ketoconazole
(400 mg once daily for 24 days, starting on day 1
at 9:00 am) and MMAE (following an intravenous
infusion of polatuzumab vedotin at 1.8 mg/kg; 2.3-
mg MMAE-equivalent dose on day 4 at 9:00 am) was
simulated with a trial size of 10 × 16 (a total of 10
trials of 16 healthy volunteer subjects) for a study
duration of 24 days (Figure 3). The interaction between
rifampicin (600 mg once daily for 29 days, starting on
day 1 at 9:00 am) and MMAE (following an intra￾venous infusion of polatuzumab vedotin at 1.8 mg/kg;
2.3-mg MMAE-equivalent dose on day 9 at 9:00 am)
was simulated at a trial size of 10 × 14 (a total
of 10 trials of 14 healthy volunteer subjects) for a
study duration of 29 days. The inhibitory effect of
MMAE (intravenous infusion of polatuzumab vedotin
at 1.8 mg/kg; 2.3-mg MMAE-equivalent dose on day
4 at 9:00 am) and midazolam (1 mg intravenously
dosed over 2 minutes on days 1 and 6 at 9:00 am)
was simulated with a trial size of 10 × 15 (a total of
10 trials of 15 healthy volunteer subjects) for a study
duration of 6 days. To simplify trial design and limit
simulation run times for the simulations with rifampicin
and ketoconazole, a single cycle of intravenous infusion
of polatuzumab vedotin was simulated.
PBPK Report for the BLA/MAA Submission and Regula￾tory Interactions
To support our proposal for using the PBPK model
simulations to inform the drug label in the absence of a
dedicated DDI study for polatuzumab vedotin, a PBPK
report on the prediction of CYP3A-mediated MMAE￾based DDI potential for polatuzumab vedotin using the
PBPK model was submitted to the regulatory agencies
in accordance with the published draft guidance issued
by the EMA and FDA regarding PBPK analyses and
reporting for the industry. In addition, the executable
model files for workspaces, healthy volunteers, oncol￾ogy populations, treatments (.cmp, .lbs, and .wks),
and output files (.xls) for final simulations were also
submitted.
Results
The PBPK model was adequately verified using the
clinical DDI data for brentuximab vedotin and was
elaborated in greater detail in our earlier publication.8
PBPK Model Application for Polatuzumab Vedotin
Before the application of the PBPK model for the
DDI prediction, the model’s performance in capturing
Samineni et al S125
Figure 3. Design of DDI study between polatuzumab vedotin and CYP3A victim and perpetrators used in PBPK model simulation.
Mid, midazolam (1 mg intravenously administered over 2 minutes on days 1 and 6).
PK sampling for midazolam and polatuzumab vedotin.
Pola, polatuzumab vedotin (intravenous infusion at 1.8 mg/kg; 2.3 mg MMAE-equivalent dose) administered on days 4 and 9.
Rif, rifampicin (600 mg once daily for 29 days, starting on day 1).
Keto, ketoconazole (400 mg once daily for 24 days, starting on day 1).
MMAE, monomethyl auristatin E; PK, pharmacokinetics.
polatuzumab vedotin PK was verified by comparing
the simulated acMMAE and unconjugated MMAE
concentration-time profiles with the observed clinical
PK data.
Simulation of acMMAE Pharmacokinetics for Polatuzumab
Vedotin. The PK profiles of acMMAE following ad￾ministration of polatuzumab vedotin at 1.8 mg/kg
(MMAE-equivalent dose of 2.3 mg) were simulated.
The simulated mean Cmax (0.69 μg/mL) and area un￾der the plasma drug concentration-time curve (AUC;
1.1 μg·day/mL) were in good agreement with the ob￾served mean Cmax (0.80 μg/mL) and area under the
concentration-time curve from time zero to infinity
(AUC0-inf) (1.86 μg·day/mL), respectively, at the 1.8-
mg/kg dose (Figure 4). The model captured the shape
of the observed concentration-time curves, which ex￾hibited a multiexponential decline with a long terminal
half -life.
Simulation of Unconjugated MMAE Pharmacokinetics
for Polatuzumab Vedotin. The model was able to
describe the PK profile of MMAE formed from
Figure 4. Simulated and observed plasma concentration-time profiles
of acMMAE following administration of a 1.8-mg/kg dose of polatuzumab
vedotin (2.3-mg MMAE dose equivalent). The black line represents the
mean concentration for the simulated population (n = 6). The thin
gray lines represent simulated individual trials (10 trials of 6 subjects).
The circles denote the mean of the observed concentrations from the
patients receiving a 1.8-mg/kg dose of polatuzumab vedotin in the phase
1 study DCS4968g (n = 6). acMMAE, antibody-conjugated monomethyl
auristatin E; MMAE, monomethyl auristatin E.
S126 The Journal of Clinical Pharmacology / Vol 60 No S1 2020
Figure 5. Simulated and observed plasma concentration-time profiles
of unconjugated MMAE formed from acMMAE following the adminis￾tration of a 1.8-mg/kg dose of polatuzumab vedotin (2.3-mg MMAE
dose equivalent). The black line represents the mean concentration
for the simulated population (n = 6). The thin gray lines represent
simulated individual trials (10 trials of 6 subjects). The circles denote
the mean of the observed unconjugated MMAE concentrations from the
patients receiving a 1.8-mg/kg dose of polatuzumab vedotin in the phase
1 study DCS4968g (n = 6). acMMAE, antibody-conjugated monomethyl
auristatin E; MMAE, monomethyl auristatin E.
acMMAE reasonably well (Figure 5). At a 2.3-mg
acMMAE-equivalent dose of polatuzumab vedotin
(1.8 mg/kg), the simulated mean AUC of MMAE was
0.037 μg·day/mL, compared with the observed AUC0-inf
value of 0.052 μg·day/mL, respectively. The median
time for MMAE to reach the maximum concentration
as determined by the rate of acMMAE elimination
(responsible for unconjugated MMAE formation) and
the MMAE CL was approximately 42 hours, which
was comparable to the observed value of 60 hours.
The simulated mean Cmax of 0.0072 μg/mL was also
similar to the observed mean maximum concentration
of 0.0068 μg/mL (Figure 5).
Prediction of MMAE DDI for Polatuzumab Vedotin. The
CYP3A-mediated DDI potential was predicted for
polatuzumab vedotin dosed at 1.8 mg/kg (2.3 mg
MMAE equivalent). The predicted DDI for po￾latuzumab vedotin was compared with the observed
clinical DDI data for brentuximab vedotin, an ADC
that contains the same linker-drug (vc-MMAE) combi￾nation and the only vc-MMAE ADC that has reported
a dedicated clinical DDI study with the CYP3A sub￾strate, inhibitor and inducer.
Effect of Polatuzumab Vedotin on Midazolam Pharma￾cokinetics. The effect of polatuzumab vedotin on the
PK of midazolam was simulated. Predictions showed
that there was no change in midazolam AUC fol￾lowing a single 1.8-mg/kg intravenous dose of po￾Table 3. Predicted MMAE-Mediated DDI (AUC and Cmax Ratio) for
Polatuzumab Vedotin and Comparison With Observed DDI Data
Reported for Brentuximab Vedotin
Analyte PK Parameter
Simulated GMR
(90%CI)
Observed GMR
(90%CI)
Polatuzumab vedotin with or without
ketoconazole
AUC 1.48 (1.43-1.52)a
Cmax 1.18 (1.17-1.20)
Polatuzumab vedotin with or without
rifampicin
AUC 0.49 (0.47-0.52)a
Cmax 0.69 (0.67-0.71)
Midazolam with or without
polatuzumab vedotin
AUC 1.00a
Cmax 1.00
Brentuximab vedotin with or without
ketoconazole
AUC0-inf 1.47 (1.43-1.51)a 1.34 (0.98-1.84)
Cmax 1.21 (1.19-1.23) 1.25 (0.90-1.72)
Brentuximab vedotin with or without
rifampicin
AUC0-inf 0.49 (0.47-0.52)a 0.54 (0.43-0.68)
Cmax 0.67 (0.65-0.69) 0.56 (0.42-0.76)
Midazolam with or without
brentuximab vedotin
AUC0-inf 1.00a 0.94 (0.81-1.10)
Cmax 1.00 1.15 (0.76-1.74)
AUC, area under the plasma drug concentration-time curve; AUC0-inf, area
under the concentration-time curve from time zero to infinity; CI, confidence
interval; Cmax, maximum drug concentration in plasma; DDI, drug-drug
interaction; GMR, geometric mean ratio; MMAE, monomethyl auristatin E; PK,
pharmacokinetic.
The DDI for polatuzumab and brentuximab with or without rifampicin and
ketoconazole reported in this table is for the unconjugated MMAE analyte;
the DDI for polatuzumab and brentuximab with or without midazolam is for
the midazolam analyte.
aRatio of AUC.
latuzumab vedotin. The predicted AUC and Cmax ratios
of midazolam with and without polatuzumab vedotin
coadministration were well aligned with the observed
brentuximab vedotin clinical DDI data (Table 3).
Effect of Ketoconazole on the Pharmacokinetics of Po￾latuzumab Vedotin. The effects of ketoconazole on the
PK of acMMAE and unconjugated MMAE following
coadministration of polatuzumab vedotin were sim￾ulated. The model predicted a 1.18-fold increase in
the unconjugated MMAE Cmax and 1.48-fold increase
in the unconjugated MMAE AUC in the presence
of ketoconazole, which was similar to the clinically
observed brentuximab vedotin DDI with Cmax and
AUC ratio of 1.25 and 1.34, respectively (Table 3). For
the acMMAE analyte, the simulated Cmax and AUC
values were similar with and without the concomitant
ketoconazole.
Samineni et al S127
Effect of Rifampicin on the Pharmacokinetics of Po￾latuzumab Vedotin. Following coadministration of po￾latuzumab vedotin with rifampicin (8 days pretreatment
at 600 mg), the model predicted a 2.0-fold decrease
(geometric mean ratio [GMR], 0.49) in unconjugated
MMAE AUC and a 1.4-fold decrease (GMR, 0.69)
in unconjugated MMAE Cmax. The PBPK model pre￾dicted that the DDI potential for polatuzumab vedotin
is similar to brentuximab vedotin as observed in the
clinic (Table 3). For the acMMAE analyte, the sim￾ulated Cmax and AUC values were similar with and
without the concomitant rifampicin.
Submission of PBPK Report for Regulatory Review
Submission of the PBPK modeling and simulation is
essential to ensure an efficient assessment and timely
decision-making during regulatory review to support
our objective of using PBPK simulations for informing
the polatuzumab vedotin label in place of a dedicated
DDI study for polatuzumab vedotin. Table 4 summa￾rizes the suggested contents of the PBPK report based
on the EMA and FDA draft guidelines regarding PBPK
analyses and reporting for the industry.11,12 The specific
purposes of each section along with the key elements
that support high-impact PBPK application, such as a
qualified PBPK platform is built for a specific intended
use, drug model verification is performed using an
independent set of clinical data, sensitivity analyses of
uncertain parameters are evaluated, and appropriate
use of PBPK analyses results in product labels for
polatuzumab vedotin are highlighted in Table 4.
Impact of PBPK Modeling and Simulation on Polatuzumab
Vedotin Labeling
The submission of the PBPK report for the predic￾tion of the DDI risk for polatuzumab vedotin was
accepted and approved by all regulatory bodies (ie,
FDA, EMA, Medicines & Healthcare products Regu￾latory Agency, Health Canada, and other health agen￾cies [HAs] in several countries). More important, the
quantitative PBPK simulation results for polatuzumab
vedotin when administered in combination with strong
CYP3A modulators and sensitive CYP3A substrates
were directly used to inform the labeling language with
the label containing a description on the predicted ex￾posure changes within the US Prescribing Information
and the EMA Summary of Product Characteristics for
polatuzumab vedotin.1,2 This is especially intriguing, as
there was no dedicated clinical DDI study conducted
for polatuzumab vedotin to anchor the DDI prediction
and thus is considered a “high-impact” application of
PBPK. Specific key questions from the HAs following
our PBPK submission focused on: (1) the quantitative
predictive performance of the PBPK models including
the confirmation of the compound files (eg, midazo￾lam, ketoconazole, and rifampin) used in the simula￾tion; and (2) the justification for not incorporating P￾glycoprotein (P-gp) in the model and its impact on
the interpretation of ketoconazole (which can interact
with both CYP3A and P-gp pathways) DDI prediction
for unconjugated MMAE as a substrate of P-gp. In
response to the queries from the HAs, minimal work
was required to perform additional analyses, and our
responses to the HAs are elaborated in the Discussion
section.
Discussion
Polatuzumab vedotin (anti-CD79b-vc-MMAE) repre￾sents an important class of therapeutic agents with
both monoclonal antibody (anti-CD79b) and small￾molecule (cytotoxic agent, MMAE) characteristics. The
MMAE, on release from anti-CD79b, is of concern
for enzyme-based drug interactions. Here, we report a
first successful application of PBPK modeling for ADC
DDI prediction in a BLA submission to enable drug
labeling without any dedicated clinical trials.
For ADCs with the same vc linker, site of conjuga￾tion, and cytotoxic agent (MMAE), the formation
mechanisms and kinetics of unconjugated MMAE
from a vc-MMAE ADC is expected to be similar.
Polatuzumab vedotin contains mc-vc-MMAE linker/
payload identical to brentuximab vedotin. The drug￾to-antibody ratio (DAR) reported for brentuximab
vedotin (DAR, 4.0) and polatuzumab vedotin (DAR,
3.7) were largely similar. The unconjugated MMAE
released from polatuzumab vedotin is identical to that
released from brentuximab vedotin. This is in line with
the theoretical expectation based on the similar dose￾normalized exposures observed for the unconjugated
MMAE analyte following the administration of 1.8 mg/
kg of brentuximab vedotin (Cmax, 2.8 ng/mL/[mg/kg];
AUC0-inf, 20.5 ng·day/mL/[mg/kg]) and 1.8 mg/kg
of polatuzumab vedotin (Cmax, 3.8 ng/mL/[mg/kg];
AUC0-inf, 29 ng·day/mL/[mg/kg]).4 Therefore, it is
conceivable that drug interactions with polatuzumab
vedotin can be predicted using a PBPK model that
can accurately describe the brentuximab vedotin clin￾ical DDI with a good mechanistic understanding.
Accordingly, for the first time, a PBPK model was
developed to predict the CYP3A-mediated DDI risk
for polatuzumab vedotin. These modeling results were
submitted to the HAs to inform the drug label in
the absence of a dedicated clinical DDI study for po￾latuzumab vedotin. This submission was well received
and accepted by several HAs and consequently in￾formed the labeling recommendations using the quan￾titative predictions from the PBPK model.8
The increasing number of submissions to regula￾tory agencies has prompted the EMA and FDA to
S128 The Journal of Clinical Pharmacology / Vol 60 No S1 2020
Table 4. Highlights of Key Components in the Polatuzumab Vedotin PBPK Report
Title of Section Purpose Considerations
Executive summary Rationale for the CYP3A-mediated DDI assessment for
polatuzumab vedotin Succinct overview of the PBPK model-based approach to predict
the DDI potential for polatuzumab vedotin Objective of waiver of a dedicated clinical CYP3A-based DDI
study for polatuzumab vedotin
MMAE, the cytotoxic component of polatuzumab vedotin, could
be a substrate and inhibitor of CYP enzymes High-level summary of PBPK model development, verification,
and application PBPK prediction of limited DDI potential for polatuzumab
vedotin can be directly used in drug labeling without conducting
clinical DDI trials
Introduction and
objectives
DDI potential associated with cytotoxic component MMAE in
ADC is a clinical risk that needs to be evaluated to propose a
concomitant medication strategy in polatuzumab vedotin drug
labeling
Limited DDI potential for brentuximab vedotin study with strong
CYP3A inhibitors and inducers with no DDI impact on sensitive CYP3A
substrates PBPK model-based approach to predict the CYP3A-mediated
drug interaction risk for polatuzumab vedotin in lieu of a
dedicated clinical DDI study.
Background information on the in vitro and in vivo ADME and
pharmacokinetics of polatuzumab vedotin
Clinical PK and DDI data for brentuximab vedotin, a similar
vc-MMAE ADC, with strong CYP3A inhibitors and inducers and
sensitive CYP3A substrates Application of the verified PBPK model to predict the
CYP3A-based DDI potential for polatuzumab vedotin
Methods Detailed documentation of approaches used in the modeling
processa
Overview of the model development, model verification, and
model application Description of the model structure and parameters for the
acMMAE and unconjugated MMAE analytesa
Graphical workflow of PBPK model development, verification,
and application process for the prediction of DDI potential for
polatuzumab vedotin Model structure: PBPK model linking acMMAE with the
unconjugated MMAE as parent to metabolite Approaches for parameter estimation and tabulated summaries
of key model parameters For acMMAE submodel, minimal PBPK model with CL and Vss
parameter estimation using a top-down approach For unconjugated MMAE submodel, acMMAE
formation-rate-limited kinetics with CL and Vss predicted using
in vitro ADME and preclinical data for unconjugated MMAE
Model assumptions Incomplete understanding of the elimination mechanism of
vc-MMAE ADCsa
The model assumes all acMMAE is eventually converted to
unconjugated MMAE before being eliminated from the body The nonmetabolic pathway, CL-biliary, contributed substantially to
MMAE clearancea
Based on the rat ADME study, the majority of acMMAE was
catabolized to unconjugated MMAE prior to subsequent
elimination Based on human mass-balance data with other vc-MMAE ADC
and in-house bile duct-cannulated study in rats treated with
radiolabeled unconjugated MMAE, model estimated biliary
accounts for about 50% of total CL of MMAE
Model evaluation
and verification
Description of the methodology for clinical PK and DDI data
used in the model development and model verificationa
Verification of PBPK model for acMMAE and unconjugated
MMAE PK prediction (using clinical PK data for
anti-CD22-vc-MMAE, brentuximab vedotin, and polatuzumab
vedotin) Verification of PBPK model for MMAE DDI prediction (using
brentuximab vedotin clinical DDI data)
Simulation design Description of the simulation conditions for the model
development, model verification, and model applicationa
Characteristics of the virtual population (adult healthy volunteers
and oncology population, number of virtual subjects/trials,
dosing information [dose, route of administration, fed
condition], simulation duration)
Sensitivity analysis As part of model qualification to obtain accurate estimation of
fm,CYP3A4, a key parameter determining polatuzumab vedotin
DDI involving MMAE as a victim druga
Range of fm,CYP3A4 estimates assigned based on the mass-balance
study of MMAE in rats and brentuximab vedotin in humans fm,CYP3A4 value determined based on the simulation that provides
the best fit to the observed clinical DDI
Results Description of the model evaluation, verification, and application
results
Graphical and tabulated displays of the comparison of the simulated
versus observed concentration-time profiles for acMMAE and
MMAE
Key PK parameters (Cmax, Tmax, AUC) Predicted magnitude of DDI (AUC and Cmax ratios)
Conclusions A dedicated clinical DDI study for polatuzumab vedotin is not
considered necessary PBPK model simulations can be used to inform labeling (eg, USPI,
SmPC) for polatuzumab vedotin
Low magnitude of DDI with strong CYP3A modulators and
sensitive CYP3A substrates for polatuzumab vedotin Limited DDI risk for brentuximab vedotin observed in the
clinical study with strong CYP3A modulators and sensitive
CYP3A substrates
acMMAE, antibody-conjugated monomethyl auristatin E; ADC, antibody-drug conjugate; ADME, absorption, metabolism, distribution, and excretion; AUC, area
under the plasma drug concentration-time curve; CL, clearance; CLbiliary, biliary clearance; Cmax, maximum drug concentration in plasma; CYP, cytochrome P450;
DDI, drug-drug interaction; MMAE, monomethyl auristatin E; PBPK, physiologically based pharmacokinetic; PK, pharmacokinetic; SmPC, Summary of Product
Characteristics; Tmax, time taken to reach maximum concentration; USPI, United States Prescribing Information; vc, valine-citrulline; Vss, volume of distribution
at steady state.
aThe published PBPK model development for the vc-MMAE ADCs was referenced in the report (Chen et al).8
Samineni et al S129
issue draft guidelines regarding PBPK analyses and
reporting for the industry.11,12 These draft guidelines on
PBPK modeling and simulation analyses pave the way
for a broader application of PBPK modeling and sim￾ulation in drug development. Although both guidelines
focused on the format and content of reporting PBPK
analyses, the EMA guidance specifically emphasized
the qualification of the PBPK platform for regulatory
submissions. As stated in the recent EMA draft guide￾line on the qualification and reporting of PBPK model￾ing and simulation, the degree of regulatory scrutiny is
proportional to the level of impact of the modeling and
simulation exercise on regulatory decision-making. For
polatuzumab vedotin, this is considered a high-impact
PBPK application, as the PBPK simulations were used
in lieu of a dedicated clinical DDI study to inform the
label.
As expected, questions related to the PBPK model’s
performance and qualifications were raised. In gen￾eral, the qualifications for each procedure used in
the submission report including PBPK models for all
perpetrators and victims used in the PK and DDI
predictions are required to be provided. In this work,
for the acMMAE-MMAE PBPK model, the model’s
predictive performance was verified using the observed
acMMAE and unconjugated MMAE exposure data
from brentuximab vedotin clinical PK and DDI studies.
The model-predicted DDI (Cmax and AUC ratio) were
all within 1.2-fold of the observed data from the clinical
DDI study for brentuximab vedotin. When this model
was used to simulate the acMMAE and unconjugated
MMAE PK for polatuzumab vedotin, although it is
comparable to the observed PK, slight apparent un￾derprediction of polatuzumab vedotin PK (∼40% for
acMMAE and ∼30% for unconjugated MMAE) led
to concerns from the reviewers potentially invalidating
the use of the acMMAE-MMAE PBPK model for
the prediction of DDI for polatuzumab vedotin. For
evaluating MMAE as a perpetrator of DDIs with other
CYP3A substrates, it is important that unconjugated
MMAE exposure is predicted accurately. PK model
verification for polatuzumab vedotin was performed
by comparing the simulated acMMAE and uncon￾jugated MMAE concentration-time profiles and PK
parameters with the observed clinical data following
the intravenous administration of polatuzumab vedotin
at the 1.8-mg/kg clinical dose. To further demonstrate
the validity of the PBPK model for DDI prediction,
given the limited clinical data at the 1.8-mg/kg dose (n =
6), 2 additional analyses were included. (1) Additional
clinical PK data were included at a higher dose of
2.4 mg/kg (n = 45) from the dose-escalation study to
show the model’s predictive performance in capturing
the PK of acMMAE and MMAE. For the updated
45 patients at the 2.4-mg/kg dose, the PBPK model
described the data well, although the overall predicted
acMMAE and unconjugated MMAE exposures (AUC)
were numerically lower than the observed values by 8%
and 21%, respectively. (2) A sensitivity analysis was per￾formed to bring up the simulated unconjugated MMAE
exposure (AUC 0.05 mg·day/mL) closer to the observed
mean data (0.053 mg·day/mL) by increasing the dose of
polatuzumab vedotin in the simulation. At this higher
exposure, the predicted DDI with MMAE and sensi￾tive CYP3A substrate midazolam was 1.0 (Cmax and
AUC ratio), confirming that MMAE does not result
in clinically significant inhibition of a sensitive CYP3A
substrate. For MMAE as a victim drug (substrate of
CYP3A4/5), its interaction with the known CYP3A
inhibitor and inducer was predicted using a PBPK
model that was verified using the observed clinical DDI
data for brentuximab vedotin with ketoconazole or
rifampicin. At the higher simulated MMAE exposure,
the predicted DDI showed no changes for polatuzumab
vedotin when administered with strong CYP3A mod￾ulators (ketoconazole and rifampin), suggesting that
the apparent underprediction in the simulated acM￾MAE and unconjugated MMAE analyte PK at the
1.8-mg/kg dose does not underestimate the predicted
DDI risk for polatuzumab vedotin. The approach used
to address potential concerns was well received by the
agencies.
In addition, the documentation supporting the pre￾dictive performance of the interacting drugs used (ie,
ketoconazole and rifampin) was provided by the soft￾ware provider Simcyp. This included a comparison of
the simulated and observed PK minimally at steady
state and the simulated and observed DDI between
the perpetrator and sensitive substrate(s) for the path￾way of interest. For the substrate (midazolam) model,
model verification included a comparison of the simu￾lated and observed PK and the simulated and observed
DDI between the substrate and strong inhibitor(s) for
the pathway of interest. Furthermore, the historical
data were acceptable to the HAs to support the model
qualification.
Knowledge gaps, uncertainty of in vitro data be￾cause of technical challenges, and the lack of clinical
data sets can hamper the qualification of a PBPK
model, thereby warranting clarity and flexibility regard￾ing the process of qualification of the PBPK model to
advance the use and acceptability of the PBPK analyses
that are intended for regulatory decision-making. For
example, the in vitro data suggested that unconjugated
MMAE is a substrate of CYP3A and P-gp efflux
transporter, and ketoconazole can interact with both
CYP3A and P-gp pathways.13 In our unconjugated
MMAE model, P-gp was not incorporated. Therefore,
justification for not incorporating P-gp in the model
and its impact on the interpretation of ketoconazole
S130 The Journal of Clinical Pharmacology / Vol 60 No S1 2020
DDI prediction for unconjugated MMAE as a sub￾strate were requested by the agency.
Clinical DDI data for brentuximab vedotin are
considered the primary evidence for the assessment of
polatuzumab vedotin DDI using PBPK modeling. As
determined in the clinical brentuximab vedotin DDI
study and given that ketoconazole is both a CYP3A
and P-gp inhibitor, any impact of P-gp inhibition
by concomitant ketoconazole is subsumed into the
overall clinical DDI effect determined for brentux￾imab vedotin. Therefore, neither the P-gp nor CYP3A
interaction individually would exceed the combined
effect of the observed ∼34% increase in exposure of
unconjugated MMAE. It is known that the magnitude
of P-gp inhibition on an intravenously administered
P-gp substrate is expected to be low.14,15 Consistent
with this current knowledge, the clinical DDI study
with brentuximab vedotin showed that the effect of
ketoconazole on MMAE (following intravenous ad￾ministration of brentuximab vedotin) is small (34%-
46%) by taking in account both the P-gp and CYP3A
effect. The current PBPK model, accounting for 2 ma￾jor disposition pathways, that is, CYP3A metabolism
and biliary excretion, adequately captured clinically
observed DDI. Although the PBPK model could be
more mechanistic by incorporating P-gp into the biliary
excretion of MMAE, its impact on the predicted DDI
would be minimal, as the DDI observed in clinic has
reflected the total effect through interaction with both
CYP3A and P-gp.
From a mechanistic standpoint, we have considered
incorporating P-gp in biliary excretion to further un￾derstand MMAE interactions with P-gp in the PBPK
model. However, building a mechanistic model through
a bottom-up approach would require adequate in vitro
transporter data as well as a good understanding of
the translation of the in vitro data to in vivo for quan￾tification of P-gp contribution in MMAE elimination.
Currently, for PBPK modeling of transporter-mediated
interactions, in vitro to in vivo extrapolation is not
mature because of an inadequate body of information,
and the predictive performance of the PBPK has yet to
be adequately demonstrated.16 Many assumptions need
to be introduced, such as translation of ketoconazole
to in vitro P-gp inhibition to in vivo effects on P-gp
export of MMAE into bile, quantitative description
of P-gp contribution in MMAE biliary excretion, and
involvement of other transporters in MMAE biliary
excretion. Moreover, without relevant clinical data,
it would not be possible to verify these model as￾sumptions. Although it is still scientifically interesting
to further explore how to build a more mechanistic
PBPK model, we postulate that the current model built
and verified using clinical brentuximab vedotin DDI
study data (ie, a combined bottom-up and top-down
approach) provides an adequate DDI risk assessment
for polatuzumab vedotin.
In summary, primary evidence from a clinical DDI
study of brentuximab vedotin suggests that concomi￾tant strong CYP3A and P-gp inhibitors are unlikely
to have a clinically meaningful impact on the PK
of unconjugated MMAE released from polatuzumab
vedotin after intravenous administration. Although the
impact of ketoconazole on P-gp-mediated biliary excre￾tion of unconjugated MMAE cannot be ruled out, the
impact of P-gp inhibitors on the PK of unconjugated
MMAE is expected to be low and not clinically relevant.
Although incorporation of potential P-gp impact on
biliary excretion into the PBPK model remains of
scientific interest, that outcome would not be expected
to substantially alter our overall conclusions related to
potential DDI risk between polatuzumab vedotin and
strong CYP3A inhibitors.
Conclusions
In conclusion, PBPK is a mechanistic modeling frame￾work essential for extrapolation of prior knowledge to
new molecules. Both the pharmaceutical industry and
regulatory agencies are aggressively pursuing MIDD.
This work demonstrated the success of a PBPK MIDD
strategy that enabled a drug interaction label claim for
polatuzumab vedotin without the need to conduct a
dedicated clinical DDI study. The increased decision
quality through MIDD has led to resource saving for
the registration and accelerated approval of ADCs.
Acknowledgments
The authors acknowledge Amita Joshi and Marcel Hop for
their scientific input and managerial support to this work.
Third-party editorial assistance, under the direction of the
authors, was provided by Farah Dalwai of Gardiner-Caldwell
Communications and was funded by F. Hoffmann-La Roche.
Conflicts of Interest
All authors are employees of Genentech, Inc. and stockhold￾ers of the Roche group.
Funding
This study was funded by F. Hoffmann-La Roche Ltd.
Author Contributions
Participated in model design: D.S., D.L., C.L., Y.C.; collected
data and ran simulations: D.S., H.D., F.M., D.L., C.L., Y.C.;
performed data analysis and wrote the article: D.S., D.H.,
F.M., R.S., D.L., D.M., J.M., C.L., J.J., M.W., S.G., Y.C.;
reviewed the article and approved for submission: D.S., D.H.,
F.M., R.S., D.L., D.M., J.M., C.L., J.J., M.W., S.G., Y.C.
Samineni et al S131
Data Accessibility Statement
Qualified researchers may request access to individual
patient-level data through the clinical study data request
platform (https://vivli.org/). Further details on Roche’s
criteria for eligible studies are available at https://vivli.org/
members/ourmembers/. For further details on Roche’s
Global Policy on the Sharing of Clinical Information and
how to request access to related clinical study documents,
see https://www.roche.com/research_and_development/who_
we_are_how_we_work/clinical_trials/our_commitment_to_
data_sharing.htm.
References
1. Genentech, Inc. Polivy® (polatuzumab vedotin)
Prescribing Information. Initial U.S. approval. 2019.

https://www.accessdata.fda.gov/drugsatfda_docs/label/2019/

761121s000lbl.pdf. Accessed June 10, 2020.
2. Roche Pharma AG. Polivy (polatuzumab vedotin) Summary
of Product Characteristics. January 16, 2020. https://www.ema.
europa.eu/en/documents/product-information/polivy-epar￾product-information_en.pdf. Accessed June 10, 2020.
3. Shemesh CS, Agarwal P, Lu T, et al. Pharmacokinetics of
polatuzumab vedotin in combination with R/G-CHP in patients
with B-cell non-Hodgkin lymphoma. Cancer Chemother Phar￾macol. 2020;85(5):831-
4. U.S. FDA Center for Drug Evaluation and Research (CDER).
Clinical Pharmacology and Biopharmaceutics Review(s);
Application No. 12538 Orig1s000; BLA 125388 & 125399
Brentuximab vedotin. February 2011. https://www.accessdata.
fda.gov/drugsatfda_docs/nda/2011/
125388Orig1s000ClinPharmR.pdf. Accessed June 10, 2020.
5. U.S. FDA CDER. Administrative and Correspondence
Documents; Application No 761121Orig1s000. https://www.
accessdata.fda.gov/drugsatfda_docs/nda/2019/
761121Orig1s000AdminCorres.pdf. Accessed June 10, 2020.
6. U.S. FDA CDER. Clinical Drug Interaction Studies — Cy￾tochrome P450 Enzyme- and Transporter-Mediated Drug
Interactions Guidance for Industry. January 2020 Clinical Phar￾macology. https://www.fda.gov/media/134581/download. Ac￾cessed June 10, 2020.
7. Grimstein M, Yang Y, Zhang X, et al. Physiologically based
pharmacokinetic modeling in regulatory science: an update
from the U.S. Food and Drug Administration’s office of clinical
pharmacology. J Pharm Sci. 2019;108(1):21-25.
8. Chen Y, Samineni D, Mukadam S, et al. Physiologically
based pharmacokinetic modeling as a tool to predict drug VcMMAE
interactions for antibody-drug conjugates. Clin Pharmacokinet.
2015;54(1):81-93.
9. Han TH, Gopal AK, Ramchandren R, et al. CYP3A-mediated
drug–drug interaction potential and excretion of brentux￾imab vedotin, an antibody−drug conjugate, in patients with
CD30-positive hematologic malignancies. J Clin Pharmacol.
2013;53(8):866-877.
10. Seattle Genetics, Inc. ADCETRIS® (brentuximab vedotin)
Prescribing Information. Initial U.S. approval: 2011; Update
2014. https://www.accessdata.fda.gov/drugsatfda_docs/label/
2014/125388_S056S078lbl.pdf. Accessed June 10, 2020.
11. U.S. FDA CDER. Physiologically Based Pharmacokinetic
Analyses — Format and Content Guidance for Industry. August
2018. https://www.fda.gov/media/101469/download. Accessed
June 10, 2020.
12. European Medicines Agency (EMA) Committee for Human
Medicinal Products (CHMP). Guideline on the reporting
of physiologically based pharmacokinetic modelling and
simulation. December 13, 2018. https://www.ema.europa.eu/en/
documents/scientific-guideline/guideline-reporting-physiologi
cally-based-pharmacokinetic-pbpk-modelling-simulation_en.
pdf. Accessed June 10, 2020.
13. Wang EJ, Lew K, Casciano CN, Clement RP, Johnson
WW. Interaction of common azole antifungals with P
glycoprotein. Antimicrob Agents Chemother. 2002;46(1):
160-165.
14. Wessler JD, Grip LT, Mendell J, Giugliano RP. The P￾glycoprotein transport system and cardiovascular drugs
[published correction appears in J Am Coll Cardiol. 2014;
63(20):2176. Dosage error in article text]. J Am Coll Cardiol.
2013;61(25):2495-2502.
15. Greiner B, Eichelbaum M, Fritz P, et al. The role of intesti￾nal P-glycoprotein in the interaction of digoxin and rifampin
[published correction appears in J Clin Invest. 2002;110(4):571].
J Clin Invest. 1999;104(2):147-153.
16. Wagner C, Zhao P, Pan Y, et al. Application of physio￾logically based pharmacokinetic (PBPK) modeling to sup￾port dose selection: report of an FDA public workshop on
PBPK. CPT Pharmacometrics Syst Pharmacol. 2015;4(4):226-
17. EMA. Assessment report: Adcetris. EMA/702390/2012.
Procedure number EMEA/H/C/002455. July 19, 2012. http:
//www.ema.europa.eu/docs/en_GB/document_library/EPAR_-
_Public_assessment_report/human/002455/WC500135054.pdf.
Accessed June 10, 2020.