Back to the Top
The following message was posted to: PharmPK
dear all
we perform our pharmacokinetic studies in mice. groups of
three mice are injected per time point and the
concentration-time curve is plotted using the mean of the
three plasma concentrations.
we want to derive a statistical method which will allow us
to compare the pharmacokinetic profile of
- compound X compared to compound Y
- compound X dosed singly to compound X dosed in
combination with other compounds (cassette dosing)
as each mouse is sampled only once we do not have a
pharmacokinetic profile from an individual subject.
i would really appreciate if anyone could comment on the
method that we have come up with as shown below or suggest
an alternative method to use
- the triplicate concentrations from three
individual animals at each time point are ranked
low, medium and high
- three concentration-time curves are plotted using
all of the low concentrations, all of the medium
concentrations and all of the high concentrations
- pharmacokinetic parameters are calculated from
the low, medium and high concentration-time
curves
- the low, medium and high parameter (eg.AUC) of
compound X is compared to the low, medium and
high parameter of compound Y using an unpaired t
test
any suggestions welcome
many thanks in advance
nicola
Nicola F Smith
Cancer Research UK Centre for Cancer Therapeutics
The Institute of Cancer Research
McElwain Building
15 Cotswold Road
Belmont
Sutton
Surrey
SM2 5NG
Tel: +44 (0)20 8643 8901 ext 4551 (office)
+44 (0)20 8722 4222 (lab)
nfsmith.-a-.icr.ac.uk
Back to the Top
The following message was posted to: PharmPK
The route you are taking is incorrect. You should fit the data using
compartmental analysis and the resulting PK parameters with its CV used
for statistic. N is derived from the degree of freedom + 1 and SEM is
part of the CV calculation.
Back to the Top
The following message was posted to: PharmPK
Nicola:
I would suggest using a mixed-effect method such as NONMEM. With it
you could obtain samples at less discreet (but accurately recorded)
time points. In an extreme example, you could use one mouse per sample
to characterize the concentration vs time curve for compound X or Y.
Similarly, you can fit data simultaneously for X vs (X plus other
compounds) to determine if there is an effect of the "other compounds".
The timing of the samples collected would be a function of the PK (or
expected PK) of your compound, and should also reflect the PK
parameters of greatest interest. As always, adequate duration of
sampling to characterize the terminal phase is helpful.
Good Luck
Paul
Paul Hutson, Pharm.D.
Associate Professor (CHS)
UW School of Pharmacy
777 Highland Avenue
Madison, WI 53705-2222
Tel: (608) 263-2496
FAX: (608) 265-5421
Pager: (608) 265-7000, #7856
Back to the Top
The following message was posted to: PharmPK
dear all
i forgot to point out that we are using non-compartmental
analysis
thanks
Nicola F Smith
Cancer Research UK Centre for Cancer Therapeutics
The Institute of Cancer Research
McElwain Building
15 Cotswold Road
Belmont
Sutton
Surrey
SM2 5NG
Tel: +44 (0)20 8643 8901 ext 4551 (office)
+44 (0)20 8722 4222 (lab)
nfsmith.aaa.icr.ac.uk
Back to the Top
Hi Nicola,
I would like to suggest the following approach:
- treat the 3 analyte levels for each time-point obtained from
different animals as a kind of "replicate" measures
- apply a compartmental analysis to this data set: one
concentration-time profile based on triplicate measures for each
data-point
- the nonlinear fitting will lead to estimate and standard
error for the PK parameters
- The estimate and its standard error can be used to contrast
results from different experiments
I hope this help,
radu
Radu D. Pop
Director Biopharmaceutics
Pharma Medica Research Inc.
966 Pantera Drive
Mississauga, Ontario
Canada, L4W 2S1
Back to the Top
The following message was posted to: PharmPK
My understanding is that the standard error from compartmental analysis
output is for the purpose of determining how well the data fits to the
chosen model, and are not really meant for between-group comparisons. I
like the NONMEM idea better for mixed-effect testing. Nicola, do you
really
expect to be able to statisticly prove anything with 1 point per mouse,
3
mice per time point? With these types of studies, I usually see
descriptive
stats, and discussions of apparent differences.
Shelly Dunnington R.Ph., Ph.D.
DunningtonShelly.-at-.praintl.com
913-577-2767
16400 College Blvd.
Lenexa, KS 66219
Back to the Top
The following message was posted to: PharmPK
Dear Nicola,
I feel the analysis what you have suggested is incorrect.
(we want to derive a statistical method which will allow us to
compare the pharmacokinetic profile of compound X compared to
compound Y)
As i understand from the protocol which you have mentioned you are
taking three samples for a single time point from 3 different
animals you have to take mean concentration and plot this mean
conc. vs time. This is done to rule out the biological variation
for that particular time point. (for a single timepoint you will
have mean+/-SD or mean+/-SEM). Once you have the pk profile of
compound X & Y, say for Ex. compound X is your test and compound Y
is your std, if you want to check the AUC difference, do
AUC(std)-AUC(test)/AUC(std)*100, see if you get the value between
80 to 125%, if you get b/w these values then you can conclude that
there is no significant difference b/w Compound X & Y's AUC. Like
wise you can compare for other pk parameters.
(compound X dosed singly to compound X dosed in combination with
other compounds (cassette dosing))
For this you need to have the pk of compound X dosed alone, so by
looking at the pk profile of compound X (dosed alone and dosed in
combination) you can see whether there is any sort of drug-drug
interaction (given in combi with respect to absorption or
elimination). Then you can compare the pk parameters as explianed
above.
(as each mouse is sampled only once we do not have a
pharmacokinetic profile from an individual subject).
To overcome this problem you can do pk in rats where you can take
serial sampling from a single animal.
These are my personal thoughts, any comments on this from pharmpk
members is highly appreciated.
Regards,
B.L.Suresh
Back to the Top
The following message was posted to: PharmPK
Dear Nicola,
You may find a smart way to deal with estimation of both AUC and its
variability when you use destructive sampling. Please see the following
article: Bailer AJ. Testing for the equality of area under the curves
when
using destructive
measurement techniques. J Pharmacokinet Biopharm 1988 Jun;16(3):303-9
This so-called Bailer's approach has been further improved by several
authors, including Gibiansky E, Nedelman J, Tse F...
Best regards,
Henri Merdjan
+33 (0)1 45 78 06 72
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)