Back to the Top
Dear colleagues,
My questions are relative to n vivo tox models of monoclonal
antibodies that are intended for therapy of human solid tumors.
How important, for the planning of an in vivo tox study of a
monoclonal antibody in a large animal (e.g. non-human primate), is it
to understand how the tumor burden affects the PK of the antibody in
a PK/PD study (in an in vivo efficacy model of that antibody) and how
is the data analyzed, i.e. what types of models should be used to
analyze the data obtained in these conditions? And what are the
guidelines for scaling up the dosages to be tested on the in vivo tox
model from those found effective in the in vivo efficacy model of
that antibody?
I truly appreciate any information or suggestions on this area.
Best regards to all,
Luis da Cruz
Arius Research Inc.
55 York Street, 16th floor
Toronto, Ontario
M5J 1R7
CANADA
Back to the Top
Dear Luis,
In my opinion it is very important to understand whether or not a
drug will exhibit target mediated disposition. For biotherapeutics,
which, unlike conventional organics are often given at low molar
doses and may therefore be present at levels similar to the molar
levels of the target, a significant quantity of drug can be bound to
the target. I recommend reading a recent paper from Don Mager (2006)
Biochem Pharmacol 72; 1-10 where he shows how nonlinear exposure
generated by significant binding to a target can be interpreted.
Clearly one should not blindly use conventional non-compartmental
analysis, especially for half-life, as there can often be
nonlinearities such as faster half-lives at later times which may not
be apparent when only studying high doses. This was very nicely shown
by Chee Ng and colleagues in their paper on TRX1 binding CD4 (2006,
Pharm Res 23; 95).
The key point to address for scaling is the level of expression and
turnover (half-life) of the target in man compared with the
toxicology or pharmacology species. Except for target binding and
subsequent fate of the drug-target complex, many other aspects of
protein therapeutics such as monoclonal antibodies follow in a
straightforward manner from small animal to man. You may not even
have to scale, at least for free IgG, if you have sufficient clinical
knowledge of monoclonals and the new antibody follows endogenous IgG
kinetics. Scaling for the target is less easy as often little is
known or published (Meno-Tetang 2005 Basic Clin Pharm Tox 96; 182).
One should review all pharmacology, in vitro and in vivo, plus the
toxicology, prior to deciding on doses to be tested in man. As was
discovered earlier this year with TGN1412, it is not sufficient to
rely on a safety margin from a toxicology study, even 500 fold. One
must understand interspecies differences both for the drug and the
target, plus the potential for the drug to find and bind said target
in vivo. Lest it be forgotten, expression levels of target vary in
humans as well, as is evident from the tumour burdon dependent
kinetics of Herceptin (Bruno 2005 Canc Chemo Pharmacol 56; 361). The
general principles are documented in ICH E8 http://www.emea.eu.int/
pdfs/human/ich/029195en.pdf. Thorough scientific understanding of the
basic principles is the way to interpret the guidance.
Best regards, Phil.
Philip Lowe PhD
Senior Fellow, Modelling & Simulation
Novartis Pharmaceuticals AG
4002 Basel
Switzerland
Back to the Top
Dear Luis,
Choice of the animal model (for tox studies) for monoclonal
antibodies depends upon the senstivity results obtained from in vitro
tox studies. ICH S6 guidelines clearly mention the use of only
senstive speices for tox studies (for both large and small animals).
In case there is no senstive speices available then tox studies in
primates (being closest to man) can be asked for by regulators.
The data generated from tox studies is helpful in knowing the First
in man dose and safety profile of the drug. Toxicokinetic data also
helps in giving idea about the toxicity related to serum
comcentrations of antibodies.
The only guidelines that I'm aware of for monoclonal antibodies is
issued by FDA and is available on the following link.
http://www.fda.gov/cber/gdlns/ptc_mab.pdf
Also the ICH S6 guidelines are helpful. Kindly let me know if some
other guidelines are available. Another article that can help is
"Mahmood I and Balian JD. The Pharmacokinetic principles behind
scaling from preclinical results to Phase I protocols. clinical
Pharmacokinet 1999 : 36(1); 1-11"
Hope this little information helps a bit.
Regards
Dr. Shaloo Pandhi, M.D.
Back to the Top
Dear Philip,
Thank you so much for your reply. I'll definitely gather the
references that you indicate as those are important issues to be
aware of.
But following up on the topic of PK/PD studies of monoclonal
antibodies I have another question:
What is the best type of assay to monitor the PK of the antibodies in
the serum? Is there a big significance, from a pharmacokinetics or
pharmacodynamics perspective, between assays that measure the serum
concentration of free antibody (not bound to target antigen if this
is present in the serum), such as capture by anti-Id antibodies (I am
referring to those antibodies that block the binding of the
therapeutic antibody to the target antigen) and assays that measure
total antibody?
As an example:
- if one wants to do a PK/PD study in a mouse with an
antibody from a different species, is is preferable to quantitate the
antibody in serum with an assay that measures free antibody in the
serum or an assay that measures total antibody of that species. Which
result is more relevant from a PK perspective? A priori I would think
that quantification of the free antibody is more important as that is
the fraction of antibody that is available to mediate the therapeutic
effect. But in many instances people measure total antibody (by using
radiolabeled antibody).
Is this a valid assumption?
How would one deal with a situation where the target molecule is a
serum protein and the objective of the therapy is to bind that
molecule. What is the relevant data for that PK/PD study, the level
of total therapeutic antibody (bound + unbound) or the level of the
free antibody (unbound). Does it matter? Or is it just a practical
issue of which assay is the easier to develop?
How would that differ from the PK/PD study where the target molecule
is in a different tissue altogether, such as expressed on solid tumor?
Best regards to all,
Luis
Arius Research Inc.
55 York Street, 16th floor
Toronto, Ontario
M5J 1R7
phone: 416.862.2323 ext.237
email: ldacruz.-a-.ariusresearch.com
fax: 416.862.9696
www.ariusresearch.com
Back to the Top
Dear Luis,
In terms of the assay required to monitor the serum or plasma PK, one
must consider the nature of the antibody target. If the target is not
soluble, but for example a receptor on a cell, then one will
automatically measure the free antibody if the target is in a tissue,
or due to the separation by centrifugation (if a blood cell target).
The more interesting case is when the target is soluble in the
plasma, such as IgE or cytokine.
With omalizumab, assays were developed for total antibody, i.e. both
free omalizumab and the omalizumab-IgE complex. When working to
understand the system, a binding model was used where the free
omalizumab and complex were summed. However, there was a separate
clearance for free and complex, and the two were not the same. Free
omalizumab has the standard IgG elimination rate, but the complex is
cleared faster (possibly due to interference with FcRn binding?).
Given that the clearance of the free is not the same as that of the
complex, the observed PK of the sum of the two then becomes nonlinear
with respect to dose and with respect to how much IgE there is in the
system (measured as baseline IgE). This was evident in a population
PK model where both of these factors were covariates, but explainable
by a mechanistic PKPD binding model which fitted, simultaneously,
total omalizumab, free and total IgE. The basic model structure is in
Meno-Te! tang 2005, but another version is due out soon (Hayashi et
al Br J Clin Pharmacol).
For a soluble target, there is no preference for free antibody or
total, as long as the assay is properly specified. With a model based
analysis either can be used. With a non-compartmental analysis, one
must interpret appropriately. In a PKPD system, it is very useful to
have is a separate assay for the target. In the omalizumab case this
was IgE, both free and total. Since the free decreases, the total is
mainly the complex. Due to the fact that the clearance of the complex
is slower than that of free IgE, the complex accumulates, therefore
total IgE increases from baseline. The same can be seen in the
literature with infliximab and TNFalpha and we have observed similar
results with ACZ885 and total IL-1beta (see two abstracts in
http://www.ersnet.org/learning_resources_player/abstract_print_06/
files/51.pdf).
The binding model gives a neat method of directly assessing in vivo
Kd and can also allow estimation of the production rate, or
expression, of a soluble target.
A tissue target is more difficult as one cannot measure the drug-
target complex directly, unless one uses receptor microscopic
autoradiography. The alternative is to infer target occupancy through
pharmacodynamic effects. In this case the binding model reduces to
the effect site or an indirect response model as commonly used in
PKPD. Key here, however, is to remember that with high molecular
weight potent drugs such as biotherapeutics, the molar dose is small
so one should not be surprised if one has target mediated disposition
with consequent dose and time dependent nonlinearities.
Best regards, Phil.
Philip Lowe PhD
Senior Fellow, Modelling & Simulation
Novartis Pharmaceuticals AG
4002 Basel
Switzerland
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)