Back to the Top
Dear group,
In one of our PK study in rats, we found following plasma
concentrations in one time point :
Animal1-142ng/ml, Animal2-118 ng/ml and Animal3-1090 ng/ml. Pk
parameters estimated with and without 1090 ng/ml are completely
different.
THERE IS NO BIOANALYTICAL ISSUE. ON REPEAT ANALYSIS WE GET SAME RESULT.
My questions are:
How to deal with such data?
Is it allowed to drop a result because it is a statistically
significant outlier?
What percent of outliers can be rejected in a study?
How to ensure that variation is biological variation and not dosing
error? (analytical error is easy to rule out )
If there are too many outliers , will the repeat study allowed to get
better profile? (I say it is not correct)
In the repeat study, if there are no outliers, which profile to be
accepted-first one are second one? Or we need to do one more
confirmation study?
A detailed reply will be very good learning for me.
Thanks,
Vinayak
Back to the Top
Dear Vinayak, My first question would be, is the formulation a
suspension and is the dispersion stable. Have you compared the
semilog plot (Concentration vs time) of all the three animal data. If
it were a dosing error all the time points corresponding to animal 3
should not be overlapping (or proximal) with the other two animal
data and the profile should be completely different at least during
the absorption phase. I would rather be careful while repeating the
study and since this is a preclinical study I would repeat in at
least three animals after ensuring proper dispersion of the NCE in
the formulation if it is a suspension, hope ths helps, cheers, Jagannath
-- Jagannath Kota PhD Student Victorian College of Pharmacy Monash
University 381 Royal Parade Parkville Victoria Australia
Back to the Top
Dear Vinayak
As evident from your question, the profiles of the three animals are
not deviating to a greater extent except one time point of the third
animal. This clearly indicated you that it is an outlier and you went
for repeat analysis. The results were again surprising and it
revealed that the analysis was correct.
Now, from a PK profile perspective, you can judge that this value is
not acceptable. I guess that particular time point sample might have
contaminated during collection or analysis. Ensure outlier time point
is surrounded closely by the adjacent time points.
You can consider this as an outlier and do not consider for
calculation. Too many outliers are confusing and data is to be
rejected. In such instances. you can better opt for more number of
animals in one single study to rule out the biological variation.
Back to the Top
Dear Vinayak,
I think you have to drop that outlier value as it is significantly
different from others and if you consider that, its negative impact
would be more on your data.
I dont think there is a guideline for the percentage of outliers
consideration out of a data set, rather it will be more of a logical.
To rule out the dosing variation, ensure that your formulation is
good for uniformity in dosing. This can be ruled out with a repeat
study in another set of 4 animals. There are certain chemical classes
which will give us biological variation in PK studies.
Even after repeating the study by eliminating possible errors, If you
get the data with same kind of variations, you had left with no
options but accepting the data.
In repeat study if there is no outlier, its logical to go with the
repeat study data.
Hope this may help you
Regards
RajaReddy Kallem
Back to the Top
RR.Kalem,
> I think you have to drop that outlier value as it is significantly
different from others and if you consider that, its negative impact
would be more on your data.
You can do it in case you have previously defined in your SP the
criteria for considering a data point or whole subject/animal
outlier. The rules are defined elsewhere in different guidelines.
Best regards,
Dimiter Terziivanov
--
Dimiter Terziivanov, MD,PhD,DSc, Professor and
Head, Clinic of Clinical Pharmacology and Pharmacokinetics,
Univ Hosp "St. Ivan Rilski",
15 Acad. Ivan Geshov st, 1431 Sofia, Bulgaria
e-mail: dterziivanov.-at-.rilski.com; terziiv.-at-.yahoo.com
Back to the Top
The following message was posted to: PharmPK
Sometimes the simplest explanation is the one. You must present the
data inclusive and exclusive of the point. You must use a defensible
statistic for identifying the point as an outlier and offer some
interpretation of the deviation of the point.
1) Were all the formulations the same and were the animals all dosed the
same?
2) Rule out the possibility of a dilution error either in the actual
preparation or in the correction formula.
3) Rule out pre analytical error- an error in the collection time or
labeling.
4) If an internal standard was used is the response the same as for the
other animals at this point?
5) Where are these points relative to the LLOQ of the assay?
6) Need to investigate the basis and correct if there are too many
outliers. If you can identify a cause and correct this, then the study
may be repeated, if you just forge on without investigating or
correcting you are just digging a deeper hole.
7) If the first study or analysis is questionable and you repeat and the
second study is now acceptable you have 50% success. Would you base
further drug development and a submission on that?
It is not clear but are the rest of the concentration time point
acceptable for animal 3?
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 "Data point outlier" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)