Back to the Top
The following message was posted to: PharmPK
Dear colleagues,
During the AAPS conference I talked to at least 20
scientists with regard to what method they used to
treat the BQL data for PK analysis. About 60% used
omitting method (not using the BQL data), 20% used
half of LOQ, believe it or not 10% used zero and about
10% they were not sure what was used or should be
used. One pharmacokineticist from the Agency
mentioned that they definitely look at other methods
(such as LOQ/2 or more complex methods) besides
omitting the data point as it may provide useful info
and help interpreting PK and Tox data. I hope more
senior scientists in the field consider giving
talks/seminars in this type of conferences and start
educating people to move away from a simple solution
of just omitting the BQL data.
One interesting topic in the conference was the use of
whole blood instead of plasma which attracted lots of
audiences and very lively discussion.
Rostam
Back to the Top
The following message was posted to: PharmPK
Dear Rostam,
These are interesting figures! I conclude that
1) about 20% of the scientists use a 100% arbitrary guess (half of BLQ).
2) about 0% of the scientists use an appropriately weighted (i.e. using
its
reciprocal variance, based on the assay error) 'best estimate' of the
concentration. Yes, I know that about 100% of analytical scientists
refuse
to provide such 'best estimate' value, and simply return 'BLQ'.
Interesting, but not a hopeful result (except for the teachers under us,
e.g. Stuart Beal and Roger Jelliffe).
Please do not consider these one-liners as a personal attack. On the
contrary, it is really important to know what is done in practice. And
to be
honest, in most situations I belong also to the 'other 100%' (but never
to
the 20% of #1).
Best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
Email: j.h.proost.at.farm.rug.nl
Back to the Top
The following message was posted to: PharmPK
Dear Jay and Hans,
Thanks for your comments. I guess Hans second
option would be an alternative
(also I like Rogers approach which he previously
discussed it with the group and the info is available
on their web site= http://www.lapk.org). We always
calculate the %RSD and %bias during analytical method
development but we never use this info to their full
capacity neither the BQL data. With regard to LOQ/2,
I agree with Hans that this is somewhat arbitrary
guess and I believe it should be practice by care but
it does not mean it is not useful. It is very simple
and somewhat reasonable (you know your value is not
zero and you know is less than LOQ, so would be
reasonable to assume it is in the middle considering
you do not need to go complex stat or modeling to find
a value for it). Of course, when we use this approach
we must sit back and look at the data and use our
professional judgment to see whether the data make
sense or not considering pharmacology, toxicology and
preliminary PK data. At the end of the day we need
something simple and practical.
However, assigning a value to BQL data may not always
be needed or improve the dataset. Perhaps this should
be viewed as a case-by-case or compound specific
decision. One needs to look at the pharmacology (mode
of action, IC50, MIC), toxicology and preliminary PK
data (if available) of the compound they are dealing
with to determine how important the BQL data would be
in their analysis. For example if you are dealing with
a drug that its MIC is below the LOQ then becomes very
important to assign a value to BQL data rather than
omitting it. For many drugs (anticancer agents) IC50
drops dramatically by time (e.g., two order of
magnitude over 72 h) so omitting the BQL data when LOQ
is higher than MIC is not fair.
Rostam
Back to the Top
The following message was posted to: PharmPK
I think these are purely personal opinions.
Allow me to add my own one liner.
No data is less dangerous than having misleading data.
Needless to say BLQ is a value below which you can not provide data with
some degree of confidence. so Analytical chemists are right within their
right. Finally you validate your PK or clinical conclusions by
calculating
some confidence intervals. Data that generates Confidence intervals
should
come out of data that comes out with some degree of confidence.
IMHO trying to assign some value for BLQ values is sort of splitting
hair.
This hair splitting uses regression model or probability model there by
introdcuing some subjectivity with no consequence on the final result,
thereofre learned folks should really look into the fact whether this
issue
of assigning arbitrary values to BLQ values is a worthy excercise or
not.
Regards,
Prasad Tata
Back to the Top
The following message was posted to: PharmPK
Tata,
Please read the literature on this topic. "learned folks" have looked
into this problem and given useful advice to those who dont want to
throw away data by discarding BLQ values.
Beal S. Ways to fit a pharmacokinetic model with some data below the
quantification limit. Journal of Pharmacokinetics and Pharmacodynamics
2001;28(5):481-504
Hing JP, Woolfrey SG, Greenslade D, Wright PMC. Analysis of
Toxicokinetic Data using NONMEM: Impact of Quantification Limit and
Replacement Strategies for Censored Data. Journal of Pharmacokinetics
and Pharmacodynamics 2001;28(5):465-479.
--
Nick Holford, Divn Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.at.auckland.ac.nz
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
[There has been a problem with DNS on campus so this message may not
have got out onto the list. I am sending it out 'again' partly as a
test message. I'm sorry if this has been sent out before - db]
The following message was posted to: PharmPK
Tata,
Please read the literature on this topic. "learned folks" have looked
into this problem and given useful advice to those who dont want to
throw away data by discarding BLQ values.
Beal S. Ways to fit a pharmacokinetic model with some data below the
quantification limit. Journal of Pharmacokinetics and Pharmacodynamics
2001;28(5):481-504
Hing JP, Woolfrey SG, Greenslade D, Wright PMC. Analysis of
Toxicokinetic Data using NONMEM: Impact of Quantification Limit and
Replacement Strategies for Censored Data. Journal of Pharmacokinetics
and Pharmacodynamics 2001;28(5):465-479.
--
Nick Holford, Divn Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-a-.auckland.ac.nz
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)