Back to the Top
The following message was posted to: PharmPK
Firstly, as a newcomer to the area of pharmacokinetics, I would like to
thank all of you who regularly contribute to this mailing list, I have
found
it to be an extremely valuable resource for helping to climb this
learning
curve.
But onto a different kind of curve, with which I've had a little more
experience. I have had modest exposure to what is known here as
multi-exponential "curve stripping" in MRI T2 analysis, where
amplitudes and
characteristic times have well established theoretical interpretations
as
the size of an aqueous pool, and the respective spin-spin relaxation
time
for the protons within that pool.
I was first drawn to pharmacokinetics by the prospect of an analogous
system
where my experience in curve fitting might be applicable.
As I understand it, one of the key assumptions behind 'classical'
multi-exponential clearance is that of 'well stirred compartments', an
assumption which is presumably satisfied at later timescales in an
experiment. However, sampling rates generally thin out at these
timescales,
where true multi-exponential behavior is theoretically expected. In my
preliminary observations, this leads to curve stripping using a very
limited
number of points that generally have low "signal to noise" ratios,
which is
a dangerous game regardless of the algorithm used.
I understand that there are important logistical concerns RE: sampling
rates, however I'm wondering if there might be an experimental protocol
that
I've overlooked that might be more compatible with the goals of a 'curve
stripping' experiment.
I am hoping that the experience on this list might point me to a
potential
system that:
1) Adheres to a classical "multi-exponential" clearance model.
2) Logistically allows for continued aggressive sampling throughout the
time
domain (pkin imaging, perhaps?).
Any references relevant to the issues I've raised would be very much
appreciated.
Sincerely,
Evan Minty
Back to the Top
The following message was posted to: PharmPK
Dear Evan,
For curve-stripping in pharmacokinetics, let me suggest you look at PK
Solutions. At www.SummitPK.com you will find a demo, a tour, and long
listing of the pharmacokinetic equations this program calculates.
Curve-stripping in PK Solutions is fun, easy, graphical, and flexible.
Once you have extracted the exponential terms from your plasma curve, a
click of the mouse will calculate, tabulate, and graph some 75
pharmacokinetic parameters, including projection of single does data to
multiple does kinetics. Try it. It will make you an instant expert in
pharmacokinetics!
Regards,
David S. Farrier
David S. Farrier, Ph.D. Phone: 970-249-1389
Summit Research Services Fax:: 970-249-1360
68911 Open Field Dr. Email: DFarrier.at.SummitPK.com
Montrose, CO 81401 Web: http://www.SummitPK.com
Back to the Top
Dear Evan and David:
I would be careful about curve stripping. How much credibility
do you give to each data point? Further, many say it is
ill-conditioned, and depends on the particular points which are
selected to belong to each phase. People tell me that it is much better
to use weighted nonlinear least squares, MAP Bayesian fitting, or
probably best, population modeling, to get the best parameter
estimates. You might consider going to our web site www.lapk.org, and
clicking around on teaching topics. Don't look for curve stripping,
though. We just don't do it.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Evan,
I fully agree with Roger Jelliffe's comments. In addition I would say
curve stripping should be used to provide initial estimates for more
intelligent methods of parameter estimation. Personally I cannot recall
the last time I used curve stripping. Adeqaute initial estimates of
parameters typically come from inspection of the curve and some back of
the envelope calculations.
I would also suggest that you think more carefully about what you want
to obtain from the data. Choice of model and parameterization is
important so that you learn what you want to know rather than what some
black box method spits out as output.
If your application is best interpreted in terms of amplitudes and
characteristic times then I suggest you use these as parameters.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-a-.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
In the resent message Dr. Farrier wrote:
> Curve-stripping in PK Solutions is fun, easy, graphical, and flexible.
> Once you have extracted the exponential terms from your plasma curve, a
> click of the mouse will calculate, tabulate, and graph some 75
> pharmacokinetic parameters, including projection of single does data to
> multiple does kinetics. Try it. It will make you an instant expert in
> pharmacokinetics!
I sincerely hope that the last sentence of the paragraph above
was indented to be funny. For those who are interested why,
I recommend the study by Aldo Rescigno:
The rise and fall of compartmental analysis, Pharmacol Res, 44, 2001,
335-340.
Similarly to the group of Prof. Jelliffe, my group also do not utilize
the curve striping method. The approach employed in our work is
outlined in
our recent paper:
Durisova M, Dedik L.: A system approach method for the adjustment
of time-varying continuous drug infusion in individual
patients: A simulation study. J Pharmacokin Pharmacodyn, 29,
2002,427-444.
With best regards,
Maria Durisova, PhD, DSc (Math/Phys),
Head of Department of Pharmacokinetics
and Scientific Secretary
Institute of Experimental Pharmacology
Slovak Academy of Sciences
841 04 Bratislava 4
Slovak Republic
Phone/Fax: +421 2 54775928
http://www.uef.sav.sk/durisova.htm
Back to the Top
The following message was posted to: PharmPK
Dear Roger,
I regret that I have not been clear in my line of questioning. I know
that
"curve stripping" refers to a particular method of parameter estimation,
however I was under the impression that in pharmacokinetics the term was
broadly used to refer to ANY method that seeks to fit clearance data and
extract relevant parameters. I apologize.
I agree completely with your assessment of the dangers of curve
stripping.
The techniques I've had experience with in MRI are also Least Squares
techniques.
My concerns regarding experimental methodology and the challenges it
poses
to accurate curve fitting still stand. Specifically, I would be
interested
in hearing about experiments on a system that:
1) Adheres to a classical "multi-exponential" clearance model.
2) Logistically allows for continued aggressive sampling throughout the
time
domain (pkin imaging, perhaps?).
I'm sorry I wasn't more clear. Thanks for any advice that might be
sent my
way.
Sincerely,
Evan Minty
Back to the Top
The following message was posted to: PharmPK
In the recent message Assoc. Prof. N. Holford wrote:
> I would also suggest that you think more carefully about what you want
> to obtain from the data. Choice of model and parameterization is
> important so that you learn what you want to know rather than what some
> black box method spits out as output.
For those who are not familiar with the connection between model
functions
which
are commonly used in pharmacokinetics and the dynamic-system theory and
for
those who are interested to learn about this connection I can recommend
e.g. the following classic pharmacokinetic resources:
1. Wagner JG. Fundamentals of clinical pharmacokinetics, Drug
Intelligence
Publication, Inc. Hamilton, Illinois, 1975, pp.231-245.
2. Gibaldi M, Perrier D. Pharmacokinetics, (Second edition), Marcel
Dekker,
Inc. New York and Basel, 1982, pp. 419-424.
In the practical use of tools of the dynamic-system theory for modeling
purposes,
the system under study is frequently portrayed by employing a
representation,
generally known as the black-box representation. In this case the
notation
"black-box" does not have any pejorative meaning.
It simply implies that this representation does not convey any
information
about components of the system and about the nature of their
interconnection.
Regards,
Maria Durisova, PhD, DSc (Math/Phys),
Head of Department of Pharmacokinetics
and Scientific Secretary
Institute of Experimental Pharmacology
Slovak Academy of Sciences
841 04 Bratislava 4
Slovak Republic
Phone/Fax: +421 2 54775928
http://www.uef.sav.sk/durisova.htm
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)