Back to the Top
Hello everyone,
i am a student in PK-PD and i've been trying to solve a problem for
days. Actually, we have Plasmatic concentrations, effects and time and
we have to start with the modelization of the PK first before doing a
deconvolution and a simulation of a PATCH and BID. I could completely
match the disposition of the drug with the generalized minimal
residual method, finding that it is a tri-compartimental model, but
when it comes to the Ka (absorption constant rate), it just doesnt
work. I think the drug release is not of 1st order, but some sigmoidal
shape. My question is, whenever you have that case and you want to
work out the PKPD relationships, how can you calculate the Effect site
Concentration (Ce) without a Ka? I would really appreciate any
comments and helps. I am feeling really defeated by this problem.
Thank you
-Camille V.Pharmacokinetical studies at Paris Descartes, JM
Scherrmann, Paris, FRANCE
Back to the Top
The following message was posted to: PharmPK
Dear Camille,
You wrote:
"I could completely match the disposition of the drug with the
generalized
minimal residual method, finding that it is a tri-compartimental
model, but
when it comes to the Ka (absorption constant rate), it just doesnt
work. I
think the drug release is not of 1st order, but some sigmoidal shape. My
question is, whenever you have that case and you want to work out the
PKPD
relationships, how can you calculate the Effect site Concentration (Ce)
without a Ka?"
I'm afraid you have fallen for the "constant Ka myth".
For orally absorbed drugs, Ka is never a constant. There are times when
assuming a constant Ka provides a useful model - typically for drugs
that
are highly soluble and permeable and are absorbed rapidly (e.g.,
propranolol). But in general, Ka does not follow zero-order, first-
order,
sigmoid, or any other simple function. It is more complex, often
taking the
shape of a concentration-time curve, starting at zero and rising
rapidly,
then falling and decaying with occasional secondary peaks that do not
necessarily show up as secondary peaks in the Cp-time data.
For low solubility drugs, drugs with low permeability, drugs that are
subject to regional metabolism and/or transport in the gut wall, and for
formulations designed to release drug over extended periods, a
constant Ka
model is a gross simplification. In such cases, you can sometimes fit
model
parameters that appear to describe the drug's behavior with a constant
Ka,
but that is done by adjusting other PK parameters to compensate for the
incorrect absorption model.
If you see that you need to fit different model parameters for different
dose levels, that is your clue that something is wrong. A good model
should
provide reasonable explanation of the observations for all dose levels
without changing model parameters (other than those that are specific
to a
dosage form, such as release rate for a controlled release dose).
Best regards,
Walt Woltosz
Chairman & CEO
Simulations Plus, Inc. (NASDAQ: SLP)
42505 10th Street West
Lancaster, CA 93534-7059
U.S.A.
http://www.simulations-plus.com
E-mail: walt.-a-.simulations-plus.com
Back to the Top
The following message was posted to: PharmPK
Greetings Camille,
Like Walt stated, it looks like you gave up on Ka too soon.
Deconvolution of plasma data are very sensitive to accurate
estimations of
Kel. As such, the key to obtaining your Ka is likely, more often than
not,
dependent a proper estimation of Kel ... i.e., you must make sure you
have
a terminal phase independent of absorption (i.e., certainly after
absorption
is complete) or accurate (and precise) clearance estimations to perform
various methods of deconvolution analysis.
If you have estimated a Kel that is greater than what is "true"
according to
proper analysis of the data, or better yet, IV info, you will get a
sigmoidal appearance (i.e., concave up) in your Fabs vs time plot on
rectilinear scales, which should make sense to you intuitively.
You could alternatively, or concurrently attempt to add a lag phase,
which
depending on the data, can give the appearance of sigmoidal input, or a
general deviation from linearity (i.e., zero order) on Fabs vs time.
The above however, depends on your method of deconvolution, which you
did
not mention.
Happy Holidays.
SHAWN SPENCER, PhD.
Assistant Professor of Biopharmaceutics
College of Pharmacy and Pharmaceutical Sciences
Florida A&M University
Tallahassee, FL 32307
shawn.spencer.at.famu.edu
Back to the Top
Dear Sir,
thank you for your reply. I do understand that in "real life", there
is no Ka, constant rate to exist. Nevertheless, my problem is that i
have to modelise some database, and without a good PKmodel, i am
unable to move on to the PKPD steps. If that is really the case that
my absorption is not of 1st Order, what are the procedures to follow?
Should i still go on with a deconvolution in order to find an
absorption profil (rate and quantity absorbed). For the PKPD, i
already know i have a hysteresis loop, so it is an indirect
relationships that would need a Sheiner model. If that is the case, as
i know so far, i have only one equation for the Effect Site
concentration (Ce) that contains a Ka (that i dont have!). If you do
know any publication that can have an example of how to handle a
situation alike, please please let me know it.
I really thank you for your attention.
Best regards,
Camille
Back to the Top
The following message was posted to: PharmPK
Greetings Camille.
You are probably keen on all the details of your work at this point,
but I
will summarize the points to avoid anything misleading.
1. Good idea to try the two compartment model in lieu of three
compartments
as you acknowledged.
2. By "real life" I meant, when confronted with a problem, we start
simple
(called a rule of parsimony, or "occam's razor")and work our way to more
rigorous approaches as things don't fit; however I considered your
problem
may have an academic teaching purpose, not intended to test every
hypothesis. I did not intend to leave the impression that your work
was less
significant.
3.Regarding the absorption, one generally does not know if the
absorption is
first or zero order, because its raw data, and it depends on the drug
structure. For example, if a patch releases quickly and constantly,
however
the drug slowly leaks from the administration site into the plasma,
then a
zero order patch can present with first order absorption. As such, we
generally model zero order input, (looks like a short IV infusion
equation)
using a set infusion time, but may have to return to a first order
parameterization.
4. With regards to a PKPD link model, its important not to be limited to
equations in the literature, but to be able to derive them based on
one's
model to be tested. I see three possible approaches: A) if you have
Winnonlin, you can simply ask the software to link two library models
(numbers 10 and 105 in an older version); B) Using laplace transforms,
the
differential equation for the concentration in the effect compartment
with
zero order input will yield four exponential terms, for input into any
nonlinear regression software program. If you have trouble with laplace
techniques, I have class notes you are free to peruse if you like. C)
You
can model the data with a differential equations link model, directly in
Winnonlin using the Fortran code. The advantage is if you wish to forego
library models for academic reasons, or forego the laplace transform.
Cheers
-Prof. Spencer
Back to the Top
The following message was posted to: PharmPK
Dear Shawn,
Your statement:
"3.Regarding the absorption, one generally does not know if the
absorption
is first or zero order, because its raw data, and it depends on the drug
structure."
implies that it must be one or the other - i.e., first or zero order.
Perhaps that was not your intention, but it's important to note that, at
least for oral doses, it is never really first or zero order, although
there
are special conditions under which such models can be useful.
Students need to learn that simplifying assumptions are just that, and
under
what conditions various simplifications are appropriate, and when they
are
not. The real mechanistic behavior of how drug molecules transit the
gastrointestinal tract, get into solution, get absorbed, survive first
pass
extraction in the gut wall and liver, and reach the systemic
circulation is
an extremely complex process.
Best regards,
Walt
Walt Woltosz
Chairman & CEO
Simulations Plus, Inc. (NASDAQ: SLP)
42505 10th Street West
Lancaster, CA 93534-7059
U.S.A.
http://www.simulations-plus.com
E-mail: walt.at.simulations-plus.com
Back to the Top
The following message was posted to: PharmPK
Hello Walt,
Thanks for your interesting comments.
I agree, the determination of order of a process (especially PK
processes of
membrane transfer and transformation) should always be done with the
disclaimer the process is "under observation", and says not very much
(if
anything at all) about the real underlying mechanisms and/or driving
forces
involved.
Perhaps you would agree since we ordinarily derive clinical PK
parameters
from global systems in lieu of local processes (e.g., Ka from plasma not
jejunal Peff), then the myriad of complexities in oral drug
absorption, when
not distracting from the at-hand application, can be forgiven. But
you are
right; at worst, such discussions should be prefaced with "...from a
practical point of view only..." or vice versa.
Kind regards,
-Shawn
SHAWN SPENCER, PHD.
Assistant Professor of Biopharmaceutics
College of Pharmacy and Pharmaceutical Sciences
Florida A&M University
Tallahassee, FL 32307
shawn.spencer.at.famu.edu
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Drug input sigmoidal" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)