Back to the Top
I am a veterinary internist who is interested in evaluating warfarin for
clinical use in cats with thromboembolic disease. Toward that end,
I have recently completed data collection in a study on the
pharmacokinetics and pharmacodynamics of warfarin in normal
cats. We applied the classical analysis (according to Nagashima
et al, 1969) to our data, and achieved a relationship between the
decline in P and time that was very much like that described for
humans. We then derived the degradation constant of P from the
slope of the plot of P vs. time. The kd=EDs were very consistent
between individuals, as was reported from humans. We calculated
Rsyn according to the Nagashima method, and then plotted Rsyn
against log[warfarin]. We have a small problem with
these plots though, and wondered if anyone else had a comment
on this problem.
During the time period where P was declining, Rnet was a
negative number sometimes of amplitude larger than Rdeg, so the
calculation for Rsyn yielded a negative number as well. When we
plotted Rsyn vs log[warfarin] for points in the time period when an
anticoagulant effect was apparent, we got a good linear-log
relationship, but =87 to 2/3 of the points were Rsyn<0 (ie negative).
In the Nagashima paper they only report Rsyn=EDs that were positive.
Do you think our analysis is appropriate (ie does the
classical analysis only include the Rsyn vs log[warfarin] points
where Rsyn>0) or does it sound to you like we have made an error
somewhere? It makes sense to me that Rsyn would be negative
during the time period where P is declining, but these plots
resulted in Cpmax and Cpmin that were quite
close to one another. Could this be just a species difference?
Thank you in advance for any suggestions you might pass on.
Stephanie A. Smith DVM, MS
Senior Resident, Small Animal Internal Medicine
Kansas State University
Back to the Top
To Stephanie: I have been involved for several years in modeling patients
response to warfarin and have a program that could be used to directly model
your cat data and predict future dosage adjustments ( DrugCalc). This is a
bayesian forecaster that use s a modified model of theophanous and barille .
If you are interested please contact me at thertch.-at-.aol.com. Dennis Mungall,
Pharm.D. , Associate Director of Clinical pharmacology and Research, TMH
Back to the Top
The Nagashima et al. method is indeed classical but also not to be trusted.
It relies on the awful
log-conc response model that cannot predict either end of the actual
response relationship and is only
valid over 60% of the response range.
I suggest you reconsider your data with something like an emax model e.g.
Holford NHG. Clinical pharmacokinetics and pharmacodynamics of warfarin.
Understanding the dose-effect
relationship. Clinical Pharmacokinetics 1986; 11:483-504
Nick Holford, Dept Neurology, L226
OHSU,3181 SW Sam Jackson Park Rd,Portland,OR 97201,USA
n.holford.aaa.auckland.ac.nz tel:+1 (503) 494-7228 fax:494-7242
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (firstname.lastname@example.org)