Back to the Top
Dear Sir,
We have a few questions concerning the patient exclusion in
bioequivalence studies.
First of all, we all know that hemolysis of the blood affects the drug
concentration and we cannot realize that the blood is hemolyzed until we
make the centrifugation. We would like to know if we should exclude the
subject from the study if we obtain hemolyzed serum or plasma for only
1-2 sample collection points.
The other question is concerning the delay in time collection points. We
would like to know upto how many minutes of delay in collecting blood is
considered valid and how we should report the delay. Should we exclude
the subject completely if one sample point is missing or collected for
example 5 minutes late.
It will really be appreciated if you can help us...
Sebnem Apaydin, M.D.
Associate Professor of Pharmacology
Ege University, Center for Drug R&D and Pharmacokinetic Applications
35100 Bornova IZMIR TURKEY
Tel: +90-232-3392754
Fax: +90-232-3425088
e-mail: sapaydin.-at-.bornovva.ege.edu.tr
Back to the Top
The following message was posted to: PharmPK
Dear Sebnem,
just my private opinions:
> We would like to know if we should exclude the subject from the study
> if we
obtain hemolyzed serum or plasma for only 1-2 sample collection points.
Not in general.
If the erythrocyte/plasma-ratio of your analyte is not high and you are
using a
specific assay - not spectrophotometry at 400nm ;-) - you probably
won't see
anything strange in the concentration-time curve.
It would be nice to validate your assay beforehand with hemolytic
plasma. A set
of triplicates at low/intermediate/high concentrations would do the job.
If you experience strange values ("pharmacokinetic outliers") in your
hemolytic
samples, drop the samples and not the subject! It should be possible to
continue
the BE-analysis in the same way as you would handle "missing values"
(i.e.,
broken vial, poor chromatography,...).
> The other question is concerning the delay in time collection points.
> We would
like to know up to how many minutes of delay in collecting blood is
considered
valid and how we should report the delay. Should we exclude the subject
completely if one sample point is missing or collected for example 5
minutes
late.
I remember a nice article by Willi Cawello about that subject, maybe he
can
refer it to the group.
I would suggest to document all deviations, then you have two options:
a) Pragmatic way: In bioequivalence studies these deviations normally
mean out
(they occur both with test and reference preparations) and primarily
increase
variability. A positive bias is also introduced, but again: in BE we are
interested in the T/R-ratio. We recalculated a few BE-studies using the
actual
time points (as reported in the CRF) and the planned time points,
respectively.
Unless we had very small sample sizes (less than 12) the deviations in
point
estimates and confidence intervals were below 0.5%.
Nevertheless, some authorites may not be happy with this.
b) Scientific way: Use the actual time points. Strongly preferred by
many
authorities and mandatory if your doing pharmacokinetic modelling!
Don't exclude the subject. The only problematic situation arises if
your drug
has a long half-life, you have a large deviation (i.e., subjects in an
outpatient status report to the clinical facility 3 hours to late...),
and your
target parameter is AUC0-72h (truncated area). In that case you should
estimate
C72h from C75h and use that value in the calculation of AUC0-72h.
Last remark. Whatever you plan to do in the study, lay it down in the
protocol.
It's not a good idea to impovise...
Best regards
Helmut Schütz
Head Biostatistics
Biokinet GmbH
Nattergasse 4
A-1070 Vienna/Austria
Tel +43(0)1 4856969-62
Fax +43(0)1 4856969-90
mailto:helmut.schuetz.aaa.chello.at
Back to the Top
The following message was posted to: PharmPK
Thanks for your kind answer. We will really be glad if you can also help
us
about what we should do to handle the missing values "broken vial, poor
chromatography). We have seen that we can only have a maximum of % 10
missing values per each subject in Canada guidance however we could not
find
detailed information about how to handle them. For example at which
points
should we accept the data as a "missing value" or at which points should
we
exclude the subject. We really need a documented guidance or a reference
article to work with. Would you also give the name of the article you
have
mentioned by "Willi Cavello". Thanks a lot for your help beforehand
Sebnem APAYDIN
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)