Back to the Top
Hello Every one,
I need your views for below matter. Please provide me your valuable
comments on it.
1. Decision tree for reporting values after completion of
reanalysis under the reason "Anomalous concentration".
a. If the difference between original, duplicate-1 and
duplicate-2 is within 20 % then report original value.
b. A) If the difference between original & one duplicate is
within 20% and Original & another duplicate is within 30%.
B) Both the duplicate values are within 30 %.
Then report median value of Original, duplicate-1 and duplicate-2.
c. If the difference between duplicate values is within 30 % but
with original more 30 % then report mean of duplicate values.
d. If the difference between duplicate values is more than 30%
then report as " NR"
2. CC standards and QC samples which shows significant ISTD
variation (more than stated in the SOP), should be excluded from
linearity generation and also from acceptance of batch.
With Regards,
Falguni Patel
Back to the Top
The following message was posted to: PharmPK
Dear Falguni:
Is this decision tree designed for LC/MS?MS assays? If it is, could
you share the rational(s) for selecting 20% and 30%, respectively in
various places of the decision tree knowing that the precision and
accuracy of the majority of the QC samples should not be over 15%
except at LLOQ? Thanks.
TK
Back to the Top
> I need your views for below matter. Please provide me your valuable
> comments on it.
>
> 1. Decision tree for reporting values after completion of
> reanalysis under the reason "Anomalous concentration".
>
> a. If the difference between original, duplicate-1 and
> duplicate-2 is within 20 % then report original value.
If that is what your sop says!
> b. A) If the difference between original & one duplicate is
> within 20% and Original & another duplicate is within 30%.
>
What is the acceptable agreement (cv or sd) between qc or cc if you
have two curves? Why would you make this more or less stringent?
> B) Both the duplicate values are within 30 %.
> Then report median value of Original, duplicate-1 and duplicate-2.
You are going to perform this re-analysis when and why? Is this
routine re-analysis or for cause?
> c. If the difference between duplicate values is within 30 %
but
> with original more 30 % then report mean of duplicate values.
>
Is this for routine re-analysis or for cause re-analysis??
> d. If the difference between duplicate values is more than 30%
> then report as " NR"
And repeat???
>
> 2. CC standards and QC samples which shows significant ISTD
> variation (more than stated in the SOP), should be excluded from
> linearity generation and also from acceptance of batch.
If that is what your sop says. But you may want to re-examine and
optimize addition of is to samples/stds. Positive displacement
pipettes work with almost any solvent, air diplacement single and
multi channels are limited to aqueous.
Back to the Top
Dear Falguni,
Regarding the following decision tree:
1. Decision tree for reporting values after completion of
reanalysis under the reason "Anomalous concentration".
a. If the difference between original, duplicate-1 and
duplicate-2 is within 20 % then report original value.
b. A) If the difference between original & one duplicate is
within 20% and Original & another duplicate is within 30%.
B) Both the duplicate values are within 30 %.
Then report median value of Original, duplicate-1 and duplicate-2.
c. If the difference between duplicate values is within 30 % but
with original more 30 % then report mean of duplicate values.
d. If the difference between duplicate values is more than 30%
then report as " NR"
First, you need to define how to calculate a % difference between two
values: (A-B)/A, (A-B)/B, or (A-B)/mean of A&B. (or something else).
Second, the decision tree seems objective. As long as they are
written in an SOP they should be OK. In my current/past groups we
reasoned that if a nominal value is allowed +/- 15% by the standard
method acceptance criteria, then a value of 85 should confirm a value
of 115, and we would allow a difference of 30%, calculated by (A-B)/
mean and expressed as a percentage. For simplicity I would suggest
choosing a single %difference value, rather than 20% for some part of
the tree and 30% for another part of the tree.
We generally have first checked duplicates within a run for
agreement. If the values agree by a set rule, then we compared them
to another value (e.g., from a previous run) for agreement. The
reasoning was that if duplicates within a run don't agree, then there
may be a reproducibility problem with the method or specific sample,
and the likelihood of agreement between runs is even lower.
Third, I would say that if the values do not conform to the decision
tree (e.g., three discordant values determined in a specific run) then
a plan for repeat assay and an ad hoc decision tree will be documented
in advance of the repeat assay.
-Tom
Thomas Tarnowski, Ph.D.
Bioanalytical Development / Analytical Development
Elan Pharmaceuticals
800 Gateway Blvd
South San Francisco, CA 94080
thomas.tarnowski.aaa.elan.com
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 "Anomalous concentration decision tree" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)