Back to the Top
Hi All,
I am curious about what the best practice is for dealing with early time-point concentrations that
are reported as greater than the upper limit of quantification, particularly when your primary
parameter of interest is AUC. Ideally, you would go back to your assay, but this is not an option
in this case.
In the past I have handled this by “censoring” the value to be equal to the ULQ, or
back-extrapolating from the initial compartment curve (whether it be distribution or elimination).
I tend to favor the later approach, but wanted to get input on any other alternatives that are well
accepted.
Thanks,
Parag
Parag Kumar, PharmD
Pharmacokineticist, Clinical Pharmacokinetics Research Laboratory
National Institutes of Health
Clinical Center Pharmacy Department
10 Center Drive, Bldg. 10, 1N 257
Bethesda, MD 20892
Back to the Top
Hi Parag
Why not just dilute the sample and try again?
Can you be more specific about ULQ?
Steve
--
Professor Stephen Duffull
Chair of Clinical Pharmacy
School of Pharmacy
University of Otago
PO Box 56 Dunedin
New Zealand
E: stephen.duffull.at.otago.ac.nz
Back to the Top
The assay validation should include a component describing the accuracy and precision of QC samples
prepared at5 to 6 steps above the ULOQ. This can be done during validation or on the plate
containing the previously established ULOQ samples. This component is described as dilutional
linearity or dilutional integrity, it means that samples above the ULOQ may be diluted into the
assay with acceptable precision and accuracy. Making the value equal to the ULOQ is even more
corrupting to the "good" data than is the practice of setting values less than the LLOQ equal to the
LLOQ!
Edward O'Connor"
--
Hello: Yes; I agree with Steve. I think first you should perform a validation exercise to show
that the dilution of spiked plasma samples of similar high drug concentrations with plasma gives the
anticipated result when assayed.
Angus McLean
Angusmdmclean.-a-.aol.com
--
Hi Steve,
Parag said that going back to the assay was not an option in his case.
ULQ should stand for upper limit of quantification.
Jing
Jing Li
Back to the Top
Hi all,
Thank you for the responses. Unfortunately the assay was not developed in-house and we no longer
have the sample to re-run. I've tried to contact the CRO that performed the bioanalytical work, but
it appears that we will be stuck with a reported value of > the upper limit of quantification (20uM
is the ULQ and the drug is busulfan) for two early time points (post-dose and 15-mins post dose).
I agree with you Ed, that reporting the value as equal to the ULQ is definitely far from ideal. The
only alternative that I have done in the past is back-extrapolated based on the first compartment
curve, which seems pretty intuitive to me but I have no reference that this is acceptable, much less
the most accurate technique to deal with this situation.
Parag
Parag Kumar, PharmD
Pharmacokineticist, Clinical Pharmacokinetics Research Laboratory
National Institutes of Health
Clinical Center Pharmacy Department
10 Center Drive, Bldg. 10, 1N 257
Bethesda, MD 20892
[No sample, no re-assay potential? I don't think you have any results (for those samples) that can
be used pharmacokinetically. You can only report as above quantitation range. - db]
Back to the Top
Hi Parag-
For purely exploratory purposes, you could plug the response for the above-quantification-limit
samples into the regression curve and generate an estimated value. If the CRO gave you only the
sample concentrations and will not provide anything else (curve equation, actual responses, both
concentration and response results from calibration standards and QC samples), then I would have
concerns about the entire data set.
For regulatory purposes, such extrapolation would not be acceptable, since there is no evidence
available showing whether or not the calibration curve becomes nonlinnear (and if so, by what
amount) above the ULOQ.
-Tom
Thomas L. Tarnowski, Ph.D.
ttarnowski1.-a-.aol.com
--
Parag:
It all depends on what you want to do with this study data. If it is meant for a regulatory
submission only course of action is is to exclude that data point and treat it as a missing value.
If you are trying to understand what is going on and involved in modeling then you can do
extrapolation and assign a value and move on.
Any sensible CRO would have diluted the sample and reported the re-analyzed value following their
SOPs, just giving the ULOQ in a PK study is kind of putting the burden back the sponsor.
Hope this helps,
Prasad NV Tata, Ph.D., FCP
Raleigh, NC 27616
Prasad Tata
--
Better not to extrapolate as well! The CRO should only have indicated a >ULOQ value so that you
should not be able to extrapolate. Hook effect is another real argument against it.
Without samples best to leave it, as hard as that is. The lesson is to include not only dilutional
linearity but also an investigation of where the response deviates from linearity. Applies to MS
as well as ligand assays.
Edward O'Connor
Back to the Top
Thanks Tom and Ed. As a note, this data is not for any regulatory submission. Dose is being
adjusted based on AUC in a phase 1 investigator-initiated study.
Parag Kumar
--
In response to Parag's question, I think some of the respondents are misunderstanding Parag's
suggestion of extrapolation. I believe that he is suggesting a PK back extrapolation, not an
analytical extrapolation above the curve. Such a back extrapolation to a theoretical t0 is a
standard option in most PK software. It obviously should be taken with a grain of salt and is
likely to underestimate the "true" theoretical C0, but otherwise is a reasonable approach for
certain applications.
David Wells
David Wells
Back to the Top
David, That is correct! I apologize for not being clearer in my explanation.
Parag
|
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@lists.ucdenver.edu with "Concentrations above upper limit of quantification" as the subject |
Copyright 1995-2014 David W. A. Bourne (david@boomer.org)