Pharmacokinetics

The Journal of Clinical Pharmacology

The Antibody Drug Absorption Following 53(3) 314–325
© The Author(s) 2013
Subcutaneous or Intramuscular Administration DOI: 10.1002/jcph.4

and Its Mathematical Description by Coupling

Physiologically Based Absorption Process with
the Conventional Compartment
Pharmacokinetic Model

Liang Zhao, PhD, Ping Ji, PhD, Zhihong Li, PhD, Partha Roy, PhD, and
Chandrahas G. Sahajwalla, PhD

Abstract
The main objective of this paper is to propose a quantitative model to describe the absorption process for monoclonal antibody (mAb) following
subcutaneous (SC) or intramuscular (IM) administration. A hybrid model was established by coupling the physiologically based absorption process with a
conventional pharmacokinetic–pharmacodynamic (PK–PD) or PK model associated with intravenous infusion. Key physiological parameters evaluated
include the volume distribution before systemic absorption, mAb drug clearance during lymphatic transport, the neonatal Fc receptor (FcRn) capacity,
the intrinsic drug clearance via lysosomal proteolytic process, and the lymphatic flow rate. Sensitivity analyses were performed to identify those
physiological parameters significantly impacting time to peak concentration (Tmax) or drug bioavailability. Simulation results showed that lymphatic flow
rate is the only influential factor to Tmax. Lymphatic transit time and drug clearance during lymphatic transport are most influential to bioavailability but
not identifiable in their effects. Bioavailability is positively related to the lymphatic flow only at its low range (i.e., at <0.5‐fold of the selected value of
0.043 mL/min). The rest of physiological parameters only have marginal effects on either drug bioavailability or Tmax. Finally, simulation results confirmed
lymphatic transport as the major route of mAb delivery.

Keywords
absorption, bioavailability, monoclonal antibody, pharmacokinetics, physiologically based pharmacokinetic model

Monoclonal antibodies (mAbs) and fusion proteins are the serum albumin, and erythropoietin) were absorbed via the
two most common classes of biologic agents approved by lymph. Charman et al.6 have demonstrated that the major
FDA. Administration route for these agents include determinant of the SC bioavailability of human growth
intravenous (IV), subcutaneous (SC), intramuscular hormone in sheep was the presystemic catabolism during
(IM), and local injections.1 Unlike small molecule drugs, the course of lymphatic transport, instead of protein
which are mostly administered orally, biologic agents degradation at the injection site. Due to experimental
have not been formulated to allow oral administration due limitations that can be conducted in humans, the role of
to denaturation of protein by acidic pH in the stomach, lymphatic catabolism on the bioavailability of other
proteolytic degradation within the gastrointestinal (GI) proteins, including mAbs, is poorly understood following
tract, and minimal diffusion through the GI epithelium.2
The mechanism of antibody drug absorption after SC
and IM injection is not fully understood, but has been
Supplementary material for this article is available on the journal’s
proposed to occur mainly via lymphatic drainage.3,4 The website at www.wileyonlinelibrary.com.
two primary pathways for systemic absorption are the
convective transport of antibody through lymphatic Office of Clinical Pharmacology, Office of Translational Sciences,
Center for Drug Evaluation and Research, US Food and Drug
vessels and diffusion of antibody across blood vessels. Administration (FDA), Silver Spring, MD, USA
Findings in sheep by Supersaxo et al.3 revealed that the
majority of IgG1 administered SC might be absorbed Submitted for publication 16 January 2012; accepted 21 June 2012.
through lymphatic vessels. In contrast, findings in rat by
Corresponding Author:
Kagan et al.5 suggest that the role of diffusion into blood Liang Zhao, 10903 New Hampshire Ave, Silver Spring, MD 20993,
vessels may be underestimated by the sheep studies with a USA
finding that <3% of administered proteins (insulin, bovine Email: liang.zhao@fda.hhs.gov
Zhao et al. 315

SC or IM route of administration. Therefore, the primary including the distribution of small molecules and bio-
determinant of antibody absorption remains uncertain. logics.7–15 A series of publications in the area of PBPK
The major objective of this paper is to propose a hybrid models describe antibody drug tissue distribution.16–21
model to predict mAb bioavailability and to quantitatively These publications mainly focused on the mathematical
describe its major routes of delivery to systemic description of paracellular transport through convection
circulation post SC and IM administration. We will first and transcellular movement of antibodies in various
propose a mathematical model coupling the physiologi- human tissues and organs. Given the size and surface
cally based absorption process with the conventional charges of antibodies, the paracellular transport of
compartment PK model to describe mAb drug concentra- antibodies in human tissues has been considered as less
tion profile. Sensitivity analyses will follow to evaluate the prevalent. In comparison, mechanisms involving the
effects of model parameters on Tmax and bioavailability. transcellular movement include phagocytosis, fluid‐phase
Finally, inferences about the extent of drug absorption via pinocytosis, and the receptor‐mediated endocytosis via
lymphatic transport versus other delivery pathways to binding of mAbs to cell surface antigens or Fcg
systemic blood circulation will be made based on the receptors.22 Once endocytosized, IgG antibodies can
Compartment Expansion approach. bind to the Brambell receptor, FcRn, in endosome. FcRn‐
bound IgGs are protected from intracellular degradation,
and can be recycled back to the plasma and/or interstitial
Methods space. It has been proposed that FcRn be an important
A Mini Survey of Antibody Drugs Approved for SC contributor to the transport of IgGs from plasma to the
or IM Administration interstitial fluid of tissues.22,23 Physiological parameters
There is a paucity of research in understanding the associated with such processes have been reported for
physiological basis for antibody drug absorption when human and animal organs. However, no report to our
administered via SC or IM route in humans. Based on our knowledge has used PBPK approach to characterize the
review of the FDA‐approved biologics over the past two mAb absorption process post‐SC or ‐IM administration.
decades (n ¼ 33 based on our survey; n ¼ 19 for IV;
n ¼ 12 for SC; n ¼ 2 for IM, n ¼ 12 for both IV and SC, Assuming SC and IM Administrations as Two
http://www.accessdata.fda.gov/scripts/cder/drugsatfda/), Intravenous Routes of Drug Administrations
the bioavailability range for mAbs following SC Based on the anatomical understanding of protein drug
administration is 52–80% (Table 1). For fusion proteins, transport post SC and IM administration,24 protein drugs
the bioavailability ranges from 43% to 65%. The reported are made bioavailable to systemic circulation via two
time to peak concentration (Tmax) values are 1.7–13.5 days mechanistic pathways, the pathway of interstitial space
for mAbs and 1.0–3.6 days for fusion proteins, ! endothelial space ! vascular space ! systemic cir-
respectively. culation or the pathway of interstitial space ! lymphatic
system ! the systemic circulation (Figure 1, with the
Physiologically Based Pharmacokinetic Model (PBPK) diagram for antibody absorption process adapted from
as a Tool to Describe mAb Distribution ref.15). The two entries to blood circulation from either the
Physiologically based pharmacokinetic (PBPK) modeling vascular space or the lymphatic transport are viewed in
is a useful tool in describing physiological processes this paper as two sources of IV infusions. Table 2 provides

Table 1. FDA‐Approved mAbs and Fusion Proteins Administered via SC Route

Route Antibody Type Trade Name Generic Name BLA No. FDA Approval Date Half Life (day) Tmax (day) Fa (%)

IM Fusion Protein Amevive Alefacept 125036 1/30/2003 11.25 3.6 63

IM hIgG1 Synagis Palivizumab 103770 6/19/1998 20 5
SC Fusion protein Enbrel Etanercept 103795 11/2/1998 4.25 2.9 58
SC Fusion protein Arcalyst Rilonacept 125249 2/27/2008 6 – –
SC Fusion protein Nplate Romiplostim 125268 8/22/2008 3.5 0.3‐2 –
SC Fusion protein Orencia Abatacept 125118 7/29/2011 13.1 2‐7 79
SC hIgG1 Stelara Ustekinumab 125261 9/25/2009 14.9–45.6 13.5 (45 mg); 7 (90 mg) 57
SC hIgG2 Prolia Denosumab 125320 6/1/2010 25.4 10 –
SC hIgGk Simponi Golimumab 125289 4/24/2009 14 2–6 53
SC rhIgG1 Ilaris Canakinumab 125319 06/17/2009 26 7 70
SC rhIgG1 Raptiva Efalizumab 125075 10/27/2003 7 1.7 50
SC rhIgG1 Xolair Ornalizumab 103976 6/20/2003 26 7–8 62
SC rhIgG1 Humira Adalimumab 125057 12/31/2002 14 5.5 64
SC rhpegylated Fab Cimzia Certolizumab pegol 125160 4/22/2008 14 2‐7 80
316 The Journal of Clinical Pharmacology / Vol 53 No 3 (2013)

Figure 1. Physiological process for antibody drug absorption. There are two routes of inputs to blood stream post SC or IM administration, as indicated
by Q–L and 1/t. These two routes of absorption are depicted by actual physiological parameters. Once drug enters the blood streams, the PK profile is
depicted by the empirical PK parameters (schematics of the structure of the absorption site adopted from Garg and Balthasar16). a: The overall scheme.
b: PK model for IV Administration. Left: the conventional two‐compartment PK model; Right: the mechanism based binding model. c: Simplified scheme.

the definitions of physiological and PK parameters used to Garg and Balthasar was intended to describe the tissue/organ
describe the drug absorption, distribution, and disposition distribution of antibody drugs post IV administration, the
dynamics following these two pathways. same physiological framework of human tissue can be
adapted to describe the absorption process.
Describing the Absorption Processes Mathematically Following SC or IM distribution, the administered
The absorption process can be described mathematically mAb molecules were assumed to be confined in a
with the intention to infer on extent and the rate of protein distribution space prior to their systemic absorption. As
drug absorption based on the dynamic physiological depicted in Figure 1a, the interstitial, endothelial, and
parameters. There are mainly two types of computational vascular compartments within this distribution space were
approaches to predict antibody biodistribution in the field of assumed to be well‐stirred systems with no concentration
physiologically based pharmacokinetcs, the one pore gradients present in each of them. The volume of this
model16 and the two pore model.17 Based on the one pore distribution space is referred to as the distribution volume
model, human organs and tissues can be compartmentalized throughout this paper. The ratio of the distribution volume
into the vascular space, the endosomal/endothelial space, the to the intended tissue volume (i.e., the total skin volume
interstitial space. Although the one pore model developed by for SC or the total muscle volume for IM), r, was used to
Zhao et al. 317

Table 2. PBPK Physiological Parameters That are Used to Describe Antibody Drug Absorption Post SC/IM Administration

Parameters (Unit) Symbol Muscle Skin Note

Plasma flow rate (mL/min) Q 41318 22018

Lymphatic flow rate (mL/min) L 2.5 0.043 Tmax is most sensitive to L. Parameter value
was chosen by Tmax mapping. Its value
reported in17 was evaluated during
sensitivity analysis
Total organ volume (mL) V 3500018 680018
Vascular volume (mL) VV 70018 46218
Interstitial volume (mL) VI 455818 22718
Endosomal volume (mL) VE 0.5% V16,26 Not a sensitive parameter to PK profile as
predicted by simulation
Recycling fraction of FcRn bound antibody FR 0.71516 This value is consistent with speculation that
2/3 antibody are recycled
Clearance of antibody from endosomal CL 0.000059227 Not a sensitive parameter to PK profile as
layer (mL/min) predicted by simulation
Endosomal uptake rate of antibody (1/hour) R1 0.0000092627
Endosomal return rate of antibody (1/hour) R2 0.2627 Simulation confirmed that PK profile is not
sensitive to R2 in this range
Transit time for drug from lymph system to t 35,25,28 Tansit time only impact on bioavailability but
systemic circulation (hour) not on Tmax
Elimination rate during lymphatic transport Klymph 0.00338 Value chosen based on t ¼ 3 hr. It is
(per minute) confounded with t
FcRn concentration in organ (nM) nPt 40 µM21,29 PK profile is not sensitive to nPt when it is 20
times of administered dose
Dissociation constant for antibody fcrn Kd 63026
bindiNG (nM)
Unbound antibody fraction fu Calculated
Lymphatic reflection coefficient sL 0.216 Based on the assumption that lymphatic flow is
less reflective than vascular flow
Vascular reflection coefficient sV 0.9516
Endogenous endosomal antibody CEndogenous, Endo_Or- 6711422,25 Converted from 10 mg/mL for human plasma
concentration at steady state (nM) gan endogenous concentration; PK profile is not
sensitive to this parameter when > 10000
nM; Assumed to have a high concentration
at endosomal space
Bioavailability (%) f 62 From label
Pharmacokinetic clearance (mL/h) CLs 5.4 Deduced from label
Central volume of distribution (mL) Vs 1810 Deduced from label
Transfer rate from central to peripheral K12 0.0108 By fitting and by referencing to the PK
compartment (/h) parameter from other IgG1 antibodies
Transfer rate from peripheral to central K21 0.0097 By fitting and by referencing to the PK
compartment (/h) parameter from other IgG1 antibodies
Clearance of free omalizumab (mL/h) CLX/f 7.3230 Value for 61.1 kg subject
Clearance of complex (mL/h) CLC/f 5.8630 Value for 61.1 kg subject
Clearance of free IgE (mL/h) CLE/f 7130 Value for 61.1 kg subject
Endogenous production rate of IgE (µg/h) PE 30.330
Distribution volume of free omalizumab VX/f 590030 Value for 61.1 kg subject
(mL)
Distribution volume of IgE (mL) VE/f 590030 Value for 61.1 kg subject
Distribution volume of complex(mL) Vc/f 363030 Value for 61.1 kg subject
Dissociation Constant for IgE and Kd 1.07 (XTX/XTE)30 XTX: total omalizumab; XTE: total IgE
omalizumab(nM)

scale all other volume and flow parameters of the intended

tissue to calculate the counterpart parameter values within dCV
VV ¼ ðFR  R2  ð1  fuÞ  VE  CE Þ
the distribution space. The following differential equa- dt
tions were correspondingly employed to describe the  ðR1  CV  VV Þ  ðð1  s V Þ  L  CV Þ
absorption process and predict the corresponding PK  ðQ  LÞ  CV
profile:
318 The Journal of Clinical Pharmacology / Vol 53 No 3 (2013)
 
dCE dXTX 1
VE ¼ ðR1  CV  VV Þ  ðfu  CL  CE Þ ¼  XL þ rðQ  LÞ  CV  CLX CfX
dt dt t
 ðð1  fuÞ  R2  VE  CE Þ þ R1  CI  VI  CLC CC

dCI dX T E
VI ¼ ðð1  s V Þ  L  CV Þ þ ðð1  FRÞ ¼ PE  CLE C f E  CLC C C
dt dt
 R2  ð1  fuÞ  VE  CE Þ  ðð1  s L Þ
where XTX represents the total amount of omalizumab, Cfx
 L  CI Þ  R1  CI  VI
the concentration of free omalizumab, CC the concentra-
dX L tion of IgE‐omalizumab complex, XTE the total amount of
¼ ðð1  s L Þ  r L  C I Þ  ðK Lymph þ 1=tÞ  X L IgE, CfE the free concentration of IgE. And
dt
dCS XC
VS ¼ ð1=tÞ  XL  ðK12 þ K10 Þ  VS  CS CC ¼
dt CC
þ K21  XP þ ðQ  LÞ  CV CT X ¼ Cf X þ CC
dX P CT E ¼ Cf E þ CC
¼ K 12  V S  C S  K 21  X P
dt
X TX  X C
The first three equations were adapted from literature.16 Cf X ¼
Although the scaling factor r was initially applied to the VX
volume and flow parameters to describe the mAb X TE  X C
absorption process within the distribution space, it was Cf E ¼
VE
canceled out from both sides of these equations. The last
two equations are the pharmacokinetic differential where CTX and CTE are concentrations (nM) of total
equations for a two compartmental open model. All the omalizumab and total IgE, respectively. XC, the amount of
associated parameters are defined in Table 2. CV complex can be calculated as follows:
represents the total mAb drug concentration in vascular
space, CE the total mAb drug concentration in endothelial ( !
1 Kd VX VE
space, CI the total mAb drug concentration in interstitial XC ¼ þ XTX þ XTE
space, XL the total mAb drug amount in lymph during 2 VC
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
transport, CS the total mAb drug serum concentration, XP u0 !2 )
u
the total mAb drug amount in peripheral compartment, u Kd VX VE
t@  4XTX XTE
and Kij the transfer rate from compartment i to VC þ XTX þ XTE
compartment j (compartment 0 ¼ system exterior, com-
partment 1 ¼ blood, compartment 2 ¼ peripheral). fu is Definitions of Kd and the rest of parameters can be
calculated as follows: found in Table 2. If the drug uptake to the endothelial

space is negligible compared to the lymphatic collection,
1
fu ¼ 1   ðKd þ nPt þ CETotal Þ the endothelial and vascular compartments can be
2  CE Total
eliminated from the model. Based on the parameter
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
values shown in Table 2, the percent of mAbs transported
 ðKd þ nPt þ CETotal Þ2  4  CETotal  nPt
to the endothelial space can be calculated as follows:

Here, C TE otal ¼ C Endogenous

E þ C E . At time 0, CI is % of Drug to Endothelial Space
calculated as Dose=ðr  V I Þ and C TE otal ¼ C Endogenous . R1 V I
E ¼
Initial concentrations in all other compartment were set to R1 V I þ ð1  s L Þ  L
be 0. In all of the calculations, drug concentration and drug ¼ 0:102% f or skin or 0:035% f or muscle
amount were set to be nM and nmole, respectively. Drug
concentration was converted to µg/mL for PK profile Therefore, under the physiological environment as
plots. defined, only trivial amount of drug can enter the
Note that if mechanism based binding model instead of endothelial space and the drug absorption scheme as
the conventional two compartment PK model, in the case depicted in Figure 1a can be simplified to the one as
of omalizumab,30 is to be used to describe PK profile post depicted in Figure 1c.
absorption, the last two differential equations above Caveat should be given that the model described above
should be replaced by the following ones: has the following limitations. First, a well stirred
Zhao et al. 319

compartment with fixed distribution volume may not TMDD model.30 At the absence of a published conven-
perfectly reflect the scenario of drug distribution before tional two‐compartment model for omalizumab, we
systemic absorption post SC or IM injection. The deduced its PK parameter values based on both its
existence of concentration gradient is possible and labeling information (i.e., clearance, distribution volume,
the drug distribution space can be expanded over time. intercompartmental transfer rates, slightly fine‐tuned
However, it will be demonstrated in the paper that PK within ½ of their standard errors) and profile matching
profile is not sensitive to distribution volume and thus (i.e., absorption rate). The deduced parameter values were
these deviations from current assumptions about distribu- reported in Table 2 for interested audience.
tion space are not expected to significantly change the The omalizumab PK profile can thus be described by
simulation results. Second, target binding is not consid- three models: the original mechanism based model with
ered at the inject site or during drug transport. parameter values as reported in literature,30 the conven-
Heterogeneity in Tmax and bioavailability of mAbs can tional two‐compartment PK model, and a hybrid model
be attributable to antibody‐target binding affinity, target coupling physiologically based drug absorption process
capacity, and target biodistribution. The target effect on with either the conventional mechanism based or two‐
mAb absorption is out of the scope of this paper. Third, the compartment PK model as described by this paper. For the
lymphatic flow rate is assumed to be constant over time. mechanism based PK model, parameters as reported30
For instance, it is likely that the flow rate post SC or IM were used to simulate the expected PK profile of
injection can increase due to physiological responses. omalizumab following a 225 mg SC dose for a 75 kg
adult. For the proposed hybrid model, the distribution
Parameter Determination volume following SC or IM injection before systemic
Unless stated specifically, the physiological parameter absorption was assumed to be 10 mL. The impact of
values used to describe the absorption process are shown distribution volume, when deviating from 10 mL, on PK
in Table 2. Given the heterogeneity in PK and complexity profile was further evaluated by sensitivity simulations.
in mechanism associated with target mediated drug The drug bioavailability, the absorption rate for the
disposition (TMDD), our proposed antibody absorption conventional two compartment model, and the physiolog-
model will not consider the effects of target capacity and ical parameter values for mAb drug absorption process
distribution on the absorption process. By design, all mAb such as lymphatic flow rate (L) and lymphatic drug
drugs should exhibit TMDDs to a certain extent. However, elimination constant (Klymph) for the proposed hybrid
the PK profile of a mAb drug resembles that of the model, were all estimated to be those values that render the
endogenous mAb when the target effect becomes same omalizumab PK profile as predicted by the original
negligible. The endogenous human IgG1 mAb typically mechanism based model (Supplemental Figure S1).
has a half life of 23 days and a shorter half life normally Noteworthy, it is unclear whether mAb drugs are
indicates an accelerated clearance as a result of TMDD. “eliminated” in lymphatic transport and to what extent.
Supplemental Table S1 summarizes humanized IgG1 Here, “eliminated” means either cleared via proteolytic
mAb drugs that are associated with a half life similar to processes or distributed to other tissues other than
that of an endogenous IgG1, demonstrating dose‐ blood.
proportionality in drug exposure and having published
linear two‐compartment models. Canakinumab and Sensitivity Analysis
palivizumab were left out from the table because they Conventionally, the sensitivity of to model parameters was
do not have published two‐compartment PK models. assessed by evaluating the percentage change in serum
Out of the three mAb drugs as shown in Supplemental bioavailability with 10% change in the physiological
Table S1, none has been approved for SC administration. parameter in evaluation, which has been defined as
Instead, the omalizumab PK profile, with a serum sensitivity coefficient. Given potential nonlinear pattern in
elimination half life of 26 days, apparent volume of model output (i.e., bioavailability) versus model parame-
distribution of 2.4  1.1 mL/kg/day, a Tmax of 7–8 days, ter and large range of parameter fluctuation, the parameter
and linear PK at doses of >0.5 mg/kg, as reported from its sensitivities were evaluated by plotting bioavailability
label (http://www.accessdata.fda.gov/drugsatfda_docs/la- against a series of parameter values (i.e., at 0.1‐, 0.3‐,
bel/2007/103976s5102lbl.pdf), is the only one that 0.5‐, 0.7‐, 1‐, 3‐, 5‐, 7‐, 10‐, 50‐, and 100‐fold of its
represents an IgG1 based antibody profile without demonstrated value as shown in Table 2). Therefore, the
exhibiting significant TMDD following SC administration sensitivity of PK profile to certain physiological parameter
and has a published model to describe its PK profile.30 was evaluated by assuming extreme values for parameters
Indeed, the PK profiles of omalizumab showed a of interest (i.e., nPt, elimination rate during lymph
prominent dose‐proportional pattern for both AUC and transport, preabsorption distribution volume, and lymph
Cmax, although its PK profile following SC administration flow rate). For those parameters that were shown as
was originally characterized by a quasi‐equilibrium insensitive to the simulated PK profile, they were assigned
320 The Journal of Clinical Pharmacology / Vol 53 No 3 (2013)

with arbitrary values which only exert marginal influence by the conventional two‐compartment model (plot not
on simulation results. shown in Supplemental Figure S1 for clarity of presenta-
tion). It is noteworthy that the PK parameters for a two
Case Application compartment model, either deduced or estimated by
This application is an expanded presentation of a regression, are highly comparable between omalizumab
published case where calculation details have been carried (shown in Table 2) and the other three IgG1 based mAb
out with a different parameter set.15 In this paper, the drugs (i.e., bevacizumab, pertuzumab, and trastuzumab;
physiologically based parameter for general IgG1 mAbs shown in Supplemental Table S1).
and the PK parameters for omalizumab were adopted
(Table 2). Regarding the bioavailability post SC adminis- Sensitivity Analysis
tration, the following quantities for mAb drugs similar to The sensitivity of time to peak concentration (Tmax) to
omalizumab were calculated: (1) the percentage of the lymphatic flow rate (L) was shown in Figure 2a. With
total subcutaneously administered drug that will be other model parameters fixed (Table 2), the curves from
delivered to systemic circulation via lymphatic transport, left to right correspond to increasing flow rates. Figure 2a
(2) the probability for a subcutaneously administered drug reveals that Tmax is highly sensitive to lymphatic flow rate.
molecule to get absorbed to systemic blood circulation via In fact, L is the only influential parameter to Tmax among
lymphatic transport before traveling anywhere else (i.e., to all screened parameters. The lymphatic flow rate (L) for
be directly collected into lymphatic vessel before traveling skin was determined to be 0.043 mL/min by an iterative
to the endothelial space or vascular space), and (3) the process for Tmax matching. The counterpart rate for muscle
probability for a subcutaneously administered drug
molecule to enter the systemic circulation via the
lymphatic transport after traveling out and into interstitial
space via the FcRn mediated recycling pathway.
The above questions can be addressed by the
Compartment Expansion approach to determine drug
concentration following certain traveling inter‐compart-
mental route in a nonlinear system. The theoretical details
can be found in our earlier publication.15

Simulation Software
All mathematical models and simulations were con-
structed with Mathematica (Wolfram Research, Inc.,
Mathematica, Version 7.0, Champaign, IL (2008)).

Results
Using Three Different Models to Simulate the Same
Omalizumab PK Profile
As shown in Supplemental Figure S1, the PK profile for a
75 kg adult following a 3 mg/kg SC dose were simulated
using the published mechanism based binding PK
model.30 In this simulation, the baseline IgE level was
assumed to be the reported population median. We do not
expect significant change in PK profile corresponding to
different IgE levels given the fact that the covariate effect
has been shown to be marginal and the omalizumab label
states that it has a linear PK profile at doses >0.5 mg/kg.
The same PK profile was simulated by replacing the Figure 2. Sensitivity analysis for Tmax. The green line indicates the
absorption equation in the mechanism based binding simulated omalizumab PK profile without parameter value change. The
model with the ones used to describe the physiologically broken lines indicate PK profiles corresponding to a chosen parameter
based absorption process as proposed by this paper. All of with 0.1‐, 0.3‐, 0.5‐, 0.7‐, 1‐, 3‐, 5‐, 7‐, 10‐, 50‐, and 100‐fold of its original
the physiologically based parameter values are shown in value as reported in Table 2. a: Lymphatic flow rate (L). Broken lines from
left to right correspond to decreasing lymphatic flow rates from 100 to
Table 2 with references and descriptive notes for their
0.1 fold of its original value. b: Transit time for drug from lymph system
origins. With a bioavailability of 62% and an absorption to systemic circulation (t). Broken lines from top to bottom correspond
rate of 0.22 day1, the PK profile again can be duplicated to increasing t from 0.1 to 100 folds of its original value.
Zhao et al. 321

was determined by using the same human to mouse ratio in addition, the sensitivity of PK profile to the clearance of
L as previously reported.16 antibody from endosomal layer (CL) at the injection site is
As shown in Figure 2b, Tmax is not sensitive to the found to be marginal. This finding can be attributed to the
elimination rate during lymphatic transport (Klymph). negligible fraction of unbound IgG molecules (fu) that is
Klymph is not identifiable in its effect on Tmax from the available for proteolytic clearance from the endosomal
transit time for a drug molecule from lymph system to layer at the injection site.
systemic circulation (t). Fluctuations in both parameters As discussed in Methods Section, in a physiological
do not alter the simulated curve shape although they do environment as defined by parameters shown in Table 2,
alter the simulated bioavailability. Other physiological the mAb absorption process can be simplified as depicted
parameters as shown in Table 2 have not been found to in Figure 1c. Based on the drug disposition scheme in
significantly change Tmax over a wide range. lymphatic system, the drug bioavailability to systemic
As shown in Figure 3, the sensitivities of bioavailability circulation can be calculated by
to physiological parameters are in the order of: the
elimination rate during lymphatic transport (Klymph) ¼ 1=t 1
Bioavailability ðf Þ ¼ ¼
transit time for drug from lymph system to systemic 1=t þ K lymph 1 þ t K lymph
circulation (t) > lymphatic flow rate (L)  endosomal
uptake rate of antibody (R1) ¼ FcRn concentration Therefore, t and Klymph, are not identifiable regarding
(nPt) ¼ endosomal return rate of antibody (R2). Bio- their impacts on f. Based on a simple calculation, the
availability will increase when t or Klymph decreases or above equation predicts a bioavailability of 62%, which
when L increases at its low range (i.e., up to 0.5‐folds of its verifies the simulation outcome.
initial value of 0.043 mL/min). As numerically demon-
strated in Methods Section, the percent uptake of drug to Comparison of PK Profile Following SC or IM Route
the endothelial space can be calculated by of Administrations
Based on physiological parameters as shown in Table 2,
R1 V I the PK profiles following SC and IM routes of
R1 V I þ ð1  s L Þ  L administration were simulated. As shown in Supplemental
Figure S2, IM route of administration is associated with
The lack of correlations between bioavailability and higher maximum drug concentration (Cmax) and shorter
parameters such as R1, R2, nPt, or L at its high range (i.e., Tmax than SC route. This is consistent with expectations
>0.5‐folds of the original L value) is because the that IM administration exhibits a relatively faster absorp-
lymphatic draining rate, as reflected by (1  sL)L, tion process.34 Caveat should be given that the simulated
overwhelms the FcRn mediated recycling process, as PK profile for IM administration is subjected to change
reflected by R1V1, in an physiological environment as given lack of actual human PK data.
defined by the parameter values shown in Table 2. In
Effect of Distribution Volume before Systemic
Absorption on PK Profile
Post SC injection, the impact of volume of distribution
before systemic absorption on PK profile and bioavail-
ability is demonstrated in Figure 3, where the PK profiles
are shown to be identical at a wide range of assumed
volumes of distribution (1–200 mL). Therefore, the
distribution volume of the injected drug at the injection
site does not affect the PK exposure. This is because the
physiological environment for drug absorption is set as a
linear system and the volume and flow rate parameters at
the injection site are scaled by the same factor r (i.e.,
Distribution Volume/V).
Although it is found that distribution volume does not
affect PK profile in a physiological system where the
Figure 3. Sensitivity analysis for bioavailability. Green line: Transit time target effect is assumed to be negligible, the same
for drug from lymph system to systemic circulation (t) or the elimination conclusion cannot be extrapolated to a system where
rate during lymphatic transport (Klymph) (overlapped). Red line: FcRn target binding can significantly affect drug trafficking. In
concentration (nPt), endosomal return rate of antibody (R2), the
endosomal uptake rate of antibody (R1), clearance of antibody from
that case, with a fixed target concentration, the distribution
endosomal layer (CL), or the presystemic distribution volume volume is positively correlated with the target binding
(overlapped). Blue line: the Lymphatic flow rate (L). capacity and can thus affect the PK parameters by
322 The Journal of Clinical Pharmacology / Vol 53 No 3 (2013)

changing the drug absorption behavior. The potential overwhelming lymphatic drainage over drug transfer from
effects of target binding and other factors on drug interstitial space to endothelial space.
absorption, such as saturable presystemic elimination To evaluate the combined effect of R1, R2, and nPt, an
that relates injection volume to bioavailability, the extreme parameter set was chosen as a means to
existence of drug concentration gradient, and the coexis- exaggerate the FcRn mediated recycling process (i.e.,
tence of SC or IM absorption, are out of the scope of this R1, R2, and nPt set at 100‐folds of their physiological
paper. values as shown in Table 2). In this case, as shown in
Figure 4b, the bioavailability increases to 65%. The
Case Application probability for a drug molecule that is directly made
The physiological system associated with antibody available to systemic circulation via lymphatic flow is
absorption after subcutaneous administration was used 58%. The probability for a drug molecule to enter systemic
to address the questions as raised in the case application circulation through lymphatic transport after paying extra
section. By using Compartment Expansion method15 and visit(s) to endothelial space, via the FcRn mediated
assuming a bioavailability of 62% in the case of recycling pathway, is estimated to be 2%. About 5% of the
omalizumab, almost 100% of the drug molecules travel drug molecule is delivered via the vascular space. This
through the lymphatic route to blood circulation before hypothetical case shows that, even with an exaggerated
traveling anywhere else (Figure 4a), as a result of the FcRn mediated recycling process, direct lymphatic
transport is still the dominating route of drug delivery.
Caveat should be given that the established hybrid
PBPK model only intends to provide a relative measure on
the absorption process of IgG1 based mAb drugs. Firm
values for bioavailabilities through different traveling
routes can only be determined by further experimental or
clinical study results.

Discussion
This paper demonstrated that a hybrid PBPK model along
with the physiologically based absorption process coupled
with a conventional PK–PD or PK model can be
constructed to describe the absorption process of mAb
drugs following SC and IM administration. By solving
differential equations as defined for the absorption process
and viewing SC or IM injection as two IV routes of
infusions, bioavailability against time can be estimated
and lymphatic transport is found to account for the
majority of the drug made bioavailable to blood
circulation.
We have identified lymphatic flow rate as the only
influential factor to Tmax within all the factors in
consideration by this paper for mAb SC administration.
In fact, as shown in Table 1, omalizumab and canakinu-
mab are the only two humanized IgG1 based mAbs
approved for SC administration that are associated with a
half life similar to that of an endogenous IgG1 and are
dose‐proportional in drug exposure within a wide range of
Figure 4. Bioavailability against time. a: The overlapped line indicates
dose levels without exhibiting a pattern of TMDD.
the (1) the overall bioavailability of subcutaneously administration, (2) the
bioavailability via lymphatic transport, and (3) the bioavailability via Interestingly, their Tmaxs are highly comparable (i.e.,
lymphatic transport before traveling anywhere else (i.e., to be directly 7 days for canakinumab vs. 7–8 days for omalizumab).
collected into lymphatic vessel before traveling to the endothelial space The heterogeneity in Tmaxs for the approved mAb drugs
or capillary space). b: The solid lines from top to bottom represent (1) may stem out from several other factors that affect the
the overall bioavailability of subcutaneously administration, (2) the lymphatic flow rate at the site of injection. First, it can be
bioavailability via lymphatic transport, (3) the bioavailability via lymphatic
transport before traveling anywhere else, and (4) the bioavailability via
attributed to differences in target capacity and distribution
the lymphatic transport after traveling out and into interstitial space via which may affect drug trafficking in its absorption route.
the FcRn mediated recycling pathway. Compared to omalizumab and canakinumab, all of the
Zhao et al. 323

other IgG1 mAbs approved for SC administration either that the PK profile is not sensitive to R1 over a wide range,
have a shorter half life than 23 days or are not dose a R1 value not different from 9.26  106/hour in two
proportional in PK, which may be a manifestation of target orders will not lead to significant deviation in simulation
effect in drug delivery and/or elimination. Second, outcomes. In addition, there is abundant literature
lymphatic flow rate may change due to difference in supporting the notion of limited antibody drug entry to
ingredients in the injection solution. For example, adding the endothelial space and lymphatic transport as the major
hyalase to the injection can greatly increase the lymphatic route of delivery.4 For example, the injected particle size is
flow, to such an extent that the exposed trunk lymphatics at found to be a significant factor that influences the particle
the injection site can be identified, as a result of distension uptake and retention in lymph nodes. Particles with a
by the flow. This is caused by marked local edema at diameter of 500 nm are associated with slow clearance
the site of injection.35 Some ingredients may have the from the injection space40 or can be trapped in the first
potential to change the physiological environment of the node of lymphatic chain.41 In contrast, smaller particles
injection site. For example, certain vaccine adjuvants with a diameter of 10 nm are optimal for quick lymphatic
work by stimulating the innate immune system. Co- drainage and colloidal particles of this size were found to
injection with monophosphoryl lipid A, a “detoxified” be ideal for lymphoscintigraphy.42 Interestingly, the
version of lipolysaccharide, caused a significantly diameter of monoclonal antibodies has been reported to
increased recruitment of neutrophils and monocytes to be in a range of 7–10 nm, the ideal size for lymphatic
the injection site. This in turn significantly modulated the drainage. All the above reports support the simulation
antigen transport from the injection site into the afferent finding that the lymphatic uptake of antibody drug from
lymph.36 Third, lymphatic flow can vary as a result of the injection site overwhelms the drug migration to the
differences in injection volumes and other factors that can endothelial space and serves as the major route of drug
affect the interstitial pressure. Without external interfer- transport.
ence, the arterial pulse pressure serves as the basic driving In this paper, the effects of pre‐systemic clearance
power for lymph transport in the skin. Similarly, the (catabolism) and target binding on drug absorption are not
interstitial hydrostatic pressure in the injection site can be considered. Following extravascular administration, ther-
dramatically changed by a high volume of injection which apeutic biologics can undergo pre‐systemic clearance by
yields significant changes in local lymphatic flow rate.37 It soluble peptidases in the interstitial space, which
has also been reported that the level of lymph flow negatively impacts the bioavailability. However, the
generated could be enhanced by tissue compression or by pre‐systemic clearance is less likely to influence the curve
elevation of the venous pressure. For example, the lymph shape parameters such as Tmax. Therefore, the derivation
flow rate could be elevated 22‐fold by the skin massage of lymphatic flow rate based on the observed Tmax will not
and threefold by venous pressure elevation in rabbit, be affected by the potential presystemic clearance. There
which reflects the significant effect of interstitial pressure are a number of factors affecting the pre‐systemic
on drug lymphatic transport.38 In addition to raise clearance, such as extracellular degradation (e.g., via
interstitial pressure, large volume of injection can dilate proteolysis), rate of endocytosis (e.g., receptor‐mediated,
the lymphatic channels and enlarge apertures of other fluid phase), and rates of recycling through interaction
lymphatic capilleries,39 which can also lead to increase in with the neonatal Fc‐receptor (FcRn, a.k.a Brambell
lymph flow rate. receptor).43 Dose and concentration of the biologic
Although an endosomal uptake rate R1 of products may also influence the bioavailability of the
9.26  106/hour corresponds to a half life of 8.5 years, therapeutic biologics. Higher dose delivered at the
its physical significance should not be interpreted without injection site can saturate pre‐systemic proteolytic
considering the injected drug amount. For example, with degradation and may lead to increase in bioavailability.
an injection of 200 mg antibody in our case for These considerations are out of the scope of this paper.
simulation, the initial drug transfer rate to the endothelial Given our current model does not consider target binding
space is calculated to be 1.85 µg/hour. Additional effect, it may describe the linear portion of pharmacoki-
sensitivity analysis reveals that a 100‐folds increase R1 netic profile of fusion protein drugs as well.
does not change the simulated PK profile. Therefore, the
amount of drug that can make an entry to endothelial space
can be as high as 185 µg/hour. Given the size of Conclusion
endothelial space, which is 0.5% of the distribution Through simulation approach, physiological factors that
volume, a transfer rate of 1.85–185 µg/hour to the can impact mAb Tmax and bioavailability following SC or
endothelial space is not an unreasonably low number. IM administration were evaluated. Based on our simula-
We cannot rule out the possibility that R1 may be tion results, Tmax is most sensitive to lymphatic flow rate
concentration dependent (e.g., a sigmoidal function of but not sensitive to the rest of evaluated physiological
interstitial drug concentration). However, based on the fact parameters within a wide range around their chosen
324 The Journal of Clinical Pharmacology / Vol 53 No 3 (2013)

values. The insensitivity between Tmax and the parameters 7. Laplanche R, Meno‐Tetang GM, Kawai R. Physiologically based
associated with FcRn mediated mAb recycling, including pharmacokinetic (PBPK) modeling of everolimus (RAD001) in rats
involving non‐linear tissue uptake. J Pharmacokinet Pharmacodyn.
the endosomal uptake rate of antibody, the endosomal
2007;34(3):373–400.
return rate of antibody, the FcRn concentration at the 8. Kristensen NR, Madsen H, Ingwersen SH. Using stochastic
injection site, and the clearance of antibody from differential equations for PK/PD model development. J Pharmaco-
endosomal layer, can be explained by the overwhelming kinet Pharmacodyn. 2005;32(1):109–141.
lymphatic drainage as opposed to the FcRn mediated drug 9. Beck BD, Mattuck RL, Bowers TS, Cohen JT, O’Flaherty E. The
delivery via capillary blood vessels. The insensitivity development of a stochastic physiologically‐based pharmacokinetic
model for lead. Sci Total Environ. 2001;274(1–3):9–15.
between Tmax and other parameters, such as the transit 10. Willmann S, Höhn K, Edginton A, et al. Development of a
time for drug from lymph system to systemic circulation, physiology‐based whole‐body population model for assessing the
t, and the elimination rate during lymphatic transport, influence of individual variability on the pharmacokinetics of drugs.
Klymph, come as a result of the relatively short transit time J Pharmacokinet Pharmacodyn. 2007;34(3):401–431.
from injection site to systemic blood circulation as 11. Lánský P. A stochastic model for circulatory transport in
pharmacokinetics. Math Biosci. 1996;132(2):141–167.
opposed to Tmax (2–3 hour vs. several days). In 12. Matis JH, Kiffe TR, Matis TI, Stevenson DE. Nonlinear stochastic
comparison, bioavailability is most sensitive and nega- modeling of aphid population growth. Math Biosci. 2005;198
tively related to t and Klymph, whose effects are not (2):148–168.
identifiable with the current model. With a lesser extent, 13. Jang SH, Wientjes MG, Au JL. Determinants of paclitaxel uptake,
accumulation and retention in solid tumors Investigational New
bioavailability is also sensitive to the lymphatic flow rate
Drugs. 2001;19(2):113–123.
as characterized by a positive relationship when at <0.5‐ 14. Zhao L. New developments in using stochastic recipe for multi‐
fold of its original value. This can be explained by the compartment model: inter‐compartment traveling route, residence
weakening dominance of lymphatic drainage, as opposed time, and exponential convolution expansion. Math Biosci Eng.
to drug clearance in the endosomal layer, at decreasing 2009;6(3):663–682.
lymphatic flow rate. 15. Zhao L, Li N, Yang H. A new stochastic approach to multi‐
compartment pharmacokinetic models: probability of traveling route
Overall, the established hybrid PBPK model provided and distribution of residence time in linear and nonlinear systems. J
critical insight in quantitatively understanding the absorp- Pharmacokinet Pharmacodyn. 2011;38(1):83–104.
tion process of mAb drugs, which could be a useful 16. Garg A, Balthasar JP. Physiologically‐based pharmacokinetic
approach during development process for biologics. We (PBPK) model to predict IgG tissue kinetics in wild‐type and
hope that future research efforts in this area can yield FcRn‐knockout mice. J Pharmacokinet Pharmacodyn. 2007;34
(5):687–709.
further model improvement and use of this approach. 17. Baxter LT, Zhu H, Mackensen DG, Jain RK. Physiologically based
pharmacokinetic model for specific and nonspecific monoclonal
antibodies and fragments in normal tissues and human tumor
Disclaimer xenografts in nude mice. Cancer Res. 1994;54(6):1517–1528.
18. Baxter LT, Zhu H, Mackensen DG, Butler WF, Jain RK.
Views expressed in the article are those of the authors not
Biodistribution of monoclonal antibodies: scale‐up from mouse to
the FDA and should not be construed to represent FDA’s human using a physiologically based pharmacokinetic model.
views or policies. Cancer Res. 1995;55(20):4611–4622.
19. Friedrich SW, Lin SC, Stoll BR, Baxter LT, Munn LL, Jain RK.
References Antibody‐directed effector cell therapy of tumors: analysis and
1. Daugherty AL, Mrsny RJ. Formulation and delivery issues for optimization using a physiologically based pharmacokinetic model.
monoclonal antibody therapeutics. Adv Drug Deliv Rev. 2006; Neoplasia. 2002;4(5):449–463.
58(5–6):686–706. 20. Ferl GZ, Kenanova V, Wu AM, DiStefano JJ III. A two‐tiered
2. Reilly RM, Domingo R, Sandhu J. Oral delivery of antibodies. Future physiologically based model for dually labeled single‐chain Fv–Fc
pharmacokinetic trends. Clin Pharmacokinet. 1997;32(4):313–323. antibody fragments. Mol Cancer Ther. 2006;5(6):1550–1558.
3. Supersaxo A, Hein WR, Steffen H. Effect of molecular weight on the 21. Ferl GZ, Wu AM, DiStefano JJ III. A predictive model of therapeutic
lymphatic absorption of water‐soluble compounds following monoclonal antibody dynamics and regulation by the neonatal Fc
subcutaneous administration. Pharm Res. 1990;7(2):167–169. receptor (FcRn). Ann Biomed Eng. 2005;33(11):1640–1652.
4. Tang L, Meibohm B. Pharmacokinetics of peptides and proteins. 22. Gessner JE, Heiken H, Tamm A, Schmidt RE. The IgG Fc receptor
Pharmacokinetics and pharmacodynamics of biotech drugs: family. Ann Hematol. 1998;76(6):231–248.
principles and case studies in drug development. Bernd Meibohm 23. Lencer WI, Blumberg RS. A passionate kiss, then run: exocytosis
(Editor). Weinheim, Germany: WILEY‐VCH Verlag GmbH & Co. and recycling of IgG by FcRn. Trends Cell Biol. 2005;15:5–9.
KGaA; ISBN: 978‐3‐527‐31408‐9. 2007; Chapter 2: p 15–23. 24. Flessner MF, Lofthouse J, Zakaria el‐R. In vivo diffusion of
5. Kagan L, Gershkovich P, Mendelman A, Amsili S, Ezov N, Hoffman immunoglobulin G in muscle: effects of binding, solute exclusion,
A. The role of the lymphatic system in subcutaneous absorption and lymphatic removal. Am J Physiol. 1997;273(6 Pt 2):H2783–
of macromolecules in the rat model. Eur J Pharm Biopharm. 2007; H2793.
67(3):759–765. 25. Urva SR, Yang VC, Balthasar JP. Physiologically based pharmaco-
6. Charman SA, Segrave AM, Edwards GA, Porter CJ. Systemic kinetic model for T84.66: a monoclonal anti‐CEA antibody. J Pharm
availability and lymphatic transport of human growth hormone Sci. 2010;99(3):1582–1600.
administered by subcutaneous injection. J Pharm Sci. 2000;89(2): 26. Zhou J, Johnson JE, Ghetie V, Ober RJ, Ward ES. Generation of
168–177. mutated variants of the human form of the MHC class I‐related
Zhao et al. 325

receptor, FcRn, with increased affinity for mouse immunoglobulin monoclonal antibody. Clin Pharmacokinet. 2011;50(4):215–
G. J Mol Biol. 2003;332(4):901–913. 227.
27. Davda JP, Jain M, Batra SK, Gwilt PR, Robinson DH. A 35. Turner‐Warwick RT. Intravital techniques for the demonstration of
physiologically based pharmacokinetic (PBPK) model to characterize lymphatics. Ann R Coll Surg Engl. 1959;24(2):101–109.
and predict the disposition of monoclonal antibody CC49 and its single 36. de Veer M, Kemp J, Chatelier J, Elhay MJ, Meeusen EN. Modulation
chain Fv constructs. Int Immunopharmacol. 2008;8(3):401–413. of soluble and particulate antigen transport in afferent lymph by
28. McLennan DN, Porter CJ, Edwards GA, Brumm M, Martin SW, monophosphoryl lipid A. Immunol Cell Biol. 2012;90(4):404–410.
Charman SA. Pharmacokinetic model to describe the lymphatic DOI: 10.1038/icb.2011.53
absorption of r‐metHu‐leptin after subcutaneous injection to sheep. 37. McMaster PD. Changes in the cutaneous lymphatics of human
Pharm Res. 2003;20(8):1156–1162. beings and the lymph flow under normal and pathological
29. Garg A, Balthasar JP. PhD Thesis. Department of Pharmaceutical conditions. J Exp Med. 1937;65(3):347–372.
Science SUNY Buffalo. 2007; Chapter 3: 71–111. 38. Ikomi F, Hunt J, Hanna G, Schmid‐Schönbein GW. Interstitial fluid,
30. Hayashi N, Tsukamoto Y, Sallas WM, Lowe PJ. A mechanism‐ plasma protein, colloid, and leukocyte uptake into initial lymphatics.
based binding model for the population pharmacokinetics and J Appl Physiol. 1996;81(5):2060–2067.
pharmacodynamics of omalizumab. Br J Clin Pharmacol. 2007; 39. Swartz MA, Berk DA, Jain RK. Transport in lymphatic capillaries. I.
63(5):548–561. Macroscopic measurements using residence time distribution theory.
31. Lu JF, Bruno R, Eppler S, Novotny W, Lum B, Gaudreault J. Clinical Am J Physiol. 1996;270(1 Pt 2):H324–H329.
pharmacokinetics of bevacizumab in patients with solid tumors. 40. Uren RF, Thompson JF, Howman‐Giles RB. Lymphatic drainage of
Cancer Chemother Pharmacol. 2008;62:779–786. the skin and breast – locating sentinel nodes. Singapore: Harwood
32. Ng CM, Lum BL, Gimenez V, Kelsey S, Allison D. Rationale Academic Publishers; 1999;15:17–49, 66.
for fixed dosing of pertuzumab in cancer patients based on 41. Uren RF, Howman‐Giles R, Thompson JF, et al. Lymphoscintig-
population pharmacokinetic analysis. Pharm Res. 2006;23(6): raphy to identify sentinel lymph nodes in patients with melanoma.
1275–1284. Melanoma Res. 1994;4(6):395–399.
33. Bruno R, Washington CB, Lu JF, Lieberman G, Banken L, Klein P. 42. Pedersen B, Kristensen K. Evaluation of methods for sizing of
Population pharmacokinetics of trastuzumab in patients with colloidal radiopharmaceuticals. Eur J Nucl Med. 1981;6(11):521–
HER2þ metastatic breast cancer. Cancer Chemother Pharmacol. 526.
2005;56(4):361–369. 43. Wang W, Wang EQ, Balthasar JP. Monoclonal antibody pharmaco-
34. Smith DA, Minthorn EA, Beerahee M. Pharmacokinetics and kinetics and pharmacodynamics. Clin Pharmacol Ther. 2008;84:
pharmacodynamics of mepolizumab, an anti‐interleukin‐5 548–558.