Back to the Top
Hi,
as per the study protocol, any subject having the predose
concentration more than 5% of the Cmax shall be considered for PK
analysis,
in one of our study, the predose concentration is above 5% of Cmax and
yet less than LOQ, (Predose concentrations are observed both in period
1 and period 2).
The PK analysis is at planned at another location, do we need to
report the predose concentrations (even though less than LOQ) which
are yet above 5% of Cmax?
since as per our procedures, all the concentrations below the LOQ will
be replaced by "0" in the data submitted for PK analysis.
if i do that, the PK personnel/Statistician does not know about the
predose concentrations above 5% of Cmax.
please advice to handle the situation.
Regards,
Back to the Top
The following message was posted to: PharmPK
There seems to be an issue.
How can you determine that the values are above 5% of the Cmax if 5%
of the
Cmax is LLOQ and the values are LLOQ?
Without knowing your exact circumstances, several options seem available
such as:
1. Reset (revalidate) the assay such that it can do what you ask it to
do!
Set the LLOQ to 1% of the Cmax? Extend the lower end of the range or
change
the dilution, or both. A little range finding exercise is important
upfront.
2. Amend the protocol to handle the 5% issue (remove it).
3. Leave things as they are, report the predose values as LLOQ which
is what
they are. LLOQ does not imply zero however, but your protocols do
describe
this conversion.
Back to the Top
Hi Debbie,
At my job what we do when we get drug conc in predose sample more
than 5% of C-max is identifie that sample for repeat analysis. I
suggest you to do the same at confirm that whether conc found in
predose is below LOQ, upon repeat analysis if it comes Below LOQ than
you may report them as '0'. Let me know whether I am wrong or Correct.
Varun Kapoor
Veeda Clinical Research
Back to the Top
Hi Debbie
Now it is not clear from your email if the carryover was in all the
subjects or in a few of them. My answer to your response may vary
depending on your clarification.
On a general note: If the carryover is in a few subjects and the
carryover is greater than 5% of Cmax then you could drop the subject
in the statistical analysis. However, you would still require to
perform a supplemental statistical analysis including all the subjects
irrespective of the pre-dose concentration being greater than the Cmax
(as additional data analysis).
If all subjects have a pre-dose concentration then it may be hard to
convince the regulatory authority (exception being endogenous
compounds where the data is handled differently).
Good Luck
Manish Issar, Ph.D
Applied Biopharmaceutics
Back to the Top
The following message was posted to: PharmPK
In general GLP like bioanalytical practice, values below LQ are
reported as BLQ (or BQL) and should not be considered for PK analysis.
How to treat BQL values in PK analysis is based on study SAP or PK
SOP. We had treated BQL value as absent value, or assigned BLQ value
as 0 or half of the LQ, depending on the study design and/or SAP.
In some research environments (not GLP like), concentrations below LQ
may be used for PK analysis to obtain critical information. In this
case, there is little concerns for confirmed pre-dose concentration
that is higher than 5% of the Cmax.
An LQ is >5% of the C-max value, suggesting that your bioanalytical
method can only support upto 3 half lives of the test article. You
may want to consider to re-develop your method. We normally would
prefer a bioanalytical method that can support at least 5 to 6 half
lifes.
Hope this help.
TK
Back to the Top
The following message was posted to: PharmPK
Hi Debbie
i think
if we got conc in predose greater than 5%, of Cmax but if it is less
than LLOQ,
there is no need to take as repeat, and report 0 for further evaluation.
kanchan
Back to the Top
The following message was posted to: PharmPK
Also, is this a crossover design and you did not wait long enough to
clear or is the drug a copy of an endogenous compound
Back to the Top
Dear Debbie
What is the approach you using to determine your drug, HPLC? If it
was, may be you should change to LC-MS/MS. May be you should try to
find another way with lower LOQ.
Jiu
Back to the Top
Dear All:
About the problem of the predose conc of drug being above
5% of the peak, and yet less that the LOQ. All this is an illusion
perpetrated by the lab, and the obsolete way they report their results
as a percent CV. Instead of this, they would do MUCH better if they
would simply report the standard deviation
of the measurement, uncorrupted by the measurement itself.
There is NO statistics book that describes the CV% as a valid measure
of the error of a measurement. Instead of this, you find the standard
deviation (SD), the variance (V, the SD squared), and the reciprocal
of the variance (1/V). This is and has been a well known measure of
the credibility of a measurement. And it avoids all the problems of LOQ.
As the measurement gets smaller and approaches zero, the CV% gets
greater and greater, and eventually becomes infinite as the
measurement approaches zero. When the CV% gets to be about 15 or 20 %,
people think the measurement is no longer "acceptably" precise.
Because of this, they censor that data and say it is below some
politically or socially acceptable limit. This is NOT science. This is
intuitive judgment. This is what is obsolete.
Actually, as the measurement approaches zero, the SD usually gets
smaller as well, though it may not decrease any more, and in some
assays, may increase a little bit. However, while the CV% gets
infinite, the SD always stays finite at a measurable value. Because of
this, there is NO NEED TO CENSOR LOW DATA. You can track the SD of an
assay measurement all the way down to and including the blank.
If you are making a PK model and fitting your data, weight your data
by the reciprocal of the variance of each measurement. This is easy.
Simply ask the lab to rearrange its CV% reports to give the SD
instead. Then you can fit the SD data with a polynomial (usually up to
2nd or 3rd order is quite enough). If you don't have the software to
do this, you can go to our Laboratory of Applied Pharmacokinetics (www.lapk.org
) and, after you sign in and we give you access, download the
demonstration MM-USCPACK software which has that feature.
Then there is no need to censor low data and the whole problem is
easily overcome. You can also click around among teaching topics and
new advances and get more information about all of this. CV% is out.
In the trash. SD and 1/V are much better.
Very best regards,
Roger Jelliffe
Back to the Top
The following message was posted to: PharmPK
Roger: The precision of a method is part of a lab's restriction in
reporting, the accuracy of a method is the other. Usually, not all the
time, the accuracy (as measured by %Bias or % Difference from
expected) of a
method fails when pushed too low. And is a great part of the
determination
of LLOQ and ULOQ. Most methods require an accuracy of +/-15% at each
of the
calibration points and QCs, with excursions to +/- 20% at the LLOQ and
the
ULOQ. Similarly the CV%s are accepted at 15% overall with excursion
to 20%
at LLOQ and ULOQ. While we can possibly substitute SD for %CV for
precision, what do we substitute for %bias in assessing accuracy?
Back to the Top
Dear Ed:
I would suggest the mean error, which is the usual
definition of bias, and not % bias.
All the best,
Roger
Back to the Top
The following message was posted to: PharmPK
We really need to get to a consensus on all of this
Back to the Top
Dear Ed:
Yes we do. We need science, not intuition.
All the best,
Roger
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Predose conc above 5% of Cmax yet less than LOQ" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)