Back to the Top
Dear All
Could you please suggest me
If pre dose concentration in subject is less than LLOQ, but the
concentration is more than 5% of Cmax of that subject, can the
subject be included in PK and statistical analysis in case of BA/BE
studies.
Regards
Imran Khan
Scitec Lab.Pvt. Ltd
Mumbai
Back to the Top
Dear Imran,
Values below LLOQ can not be reported as the method for
quantitation is validated only up to the LLOQ level. Also as per your
statement the Cmax obtained will be less then 5% of LLOQ. If the same
is observed for majority of the study samples then the values
obtained will be in-between the few lower calibration levels only. As
per my suggestion the values obtained should reflect the whole
calibration range and not only a few points of the calibration curve.
As per USFDA guidance document:
A sufficient number of standards should be used to adequately define
the relationship between concentration and response. The relationship
between response and concentration should be demonstrated to be
continuous and reproducible. The number of standards used should be a
function of the dynamic range and nature of the concentration-
response relationship. In many cases, six to eight concentrations
(excluding blank values) can define the standard curve. More standard
concentrations may be recommended for nonlinear than for linear
relationships.
With regards
Kuldeep Sharma
Pliva Research India Pvt.Ltd.
Goa
Back to the Top
Dear Imran
if pre dose concentration in subject is greter than 5% of Cmax of
that subject, then FDA guid line clearly state that exclude that
subject from PK and statistical analysis.
Yogesh
Biostatistician
Mumbai
Back to the Top
Dear Imaran
Since the value in predose sample is below the LLOQ, it should be
considered as zero. Check your SOP and Protocol because the statement
regarding this generally predefined in SOP/Study Protocol. Hence if
it is zero you can you use the subject data for PK analysis.
Regards
Manish Jain
Group Leader
Bioanalytical
Wellquest Clinical Research, Mumbai
Back to the Top
Dear Imran,
If the pre-dose concentrations are below LLOQ, and if you have
defined in your PK SOP that it will be replaced with zero for PK
analysis. Then this subject can be included for PK analysis. But this
should be documented in your SOP's.
Regards
Dr. Tausif Ahmed
Metabolism and PK department
Ranbaxy Research Laboratories
Back to the Top
The following message was posted to: PharmPK
Dear Imarn,
As Kuldeep Sharma said
"Values below LLOQ can not be reported as the method for quantitation is
validated only up to the LLOQ level".
Normally, if the pre dose concentration is below LLOQ, it is set zero
for Pharmacokinetic and statistical analysis. Provided that in your
protocol it is pre defined.
If it is not predefined in the protocol, still you can carry the
Pharmacokinetic and statistical analysis by setting the below LLOQ vales
Zero but you have to have a proper documentation for the same. Example:
Project memo/Amendment to protocol/ect...
You have to worry only in this case where in, the pre dose concentration
is above LLOQ and the concentration is greater than 5% of Cmax. In this
case we can't include that particular subject for the Pharmacokinetic
and statistical analysis.
Looking forward for member's suggestions
Regards,
Kanimozhi.A
Biostatistician
Back to the Top
Dear All:
This question of LLOQ never seems to die. The precision of
the result is the important thing, and the % CV is NOT the way to
look at it, but rather the SD. If you do this, you can track the
result all the way down to and including a blank. THERE IS NO LLOQ.
If the result is for toxicology, you cannot be confident that
something is there unless the result is enough SD's above the blank
to make you feel good about the result. How many SD's? Two? Three?
This is totally negotiable, and is not a question of science at all.
Why do you have a lower limit of quantification at all? It
is true, for toxicology, where the specimen itself is the only source
of information, you may need it, as stated above. However, for PK
work, where you also know the time of the sample and the time of the
last dose, you know the drug is there, since it usually decreases
with a half-time. Because of this, you are not asking if the drug is
present or not, as in toxicology. You know, from the other
information, that the drug IS there, and you are asking HOW MUCH is
there? You might look at pp 136-139 in Handbook of Analytical
Separations, ed by Georg Hempel, Vol 5, Drug Monitoring and Clinical
Chemistry, Elsevier, 2004, ISBN 0-444-50972-0, for more on this, and
also in Jelliffe RW, Schumitzky A, Van Guilder M, Liu M, Hu L, Maire
P, Gomis P, Barbaut X, and Tahani B: Individualizing Drug Dosage
Regimens: Roles of Population Pharmacokinetic and Dynamic Models,
Bayesian Fitting, and Adaptive Control. Therapeutic Drug Monitoring,
15: 380-393, 1993.
Let's get with it in the 21st century. There really is NO
LOQ. Comments? Screams? or just more inertia? Certainly, values below
the so-called LOQ can truly be reported, as the best measure of what
in is the sample, AND THEY SHOULD BE! The people for whom you do the
supposed service can also think. Report the result and the SD for it.
Then everyone knows what is what, even for toxicology. Come on, guys,
either refute me of get with it. To set it to zero or to something
other that the reported result is simply NEITHER SCIENCE OR GOOD
STATISTICS.
How about it?
All the best,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
I agree with Kuldeep Sharma, Kanimozhi and other distinguished
colleagues that concentrations below assay LLOQ should not be reported
and pre-dose LLOQ values should be treated as stated in the
pre-defined SOP.
Per US-FDA "Guidance for Bioanalytical Method Validation" published in
May 2001, the LLOQ is defined as: The lowest amount of an analyte in a
sample that can be quantitatively determined with suitable precision
and accuracy. The Agency also stated in the same guidance "Estimation
of concentration in unknown samples by extrapolation of standard
curves below LLOQ or above the highest standard is NOT recommended".
It is clear, per Guidance, that any concentration below LLOQ should be
reported as BQL (or BLQ, Below Limit of Quantification) and not to
have any numerical concentration value associated with the analyte in
the sample. Ideally, we will develop an assay with an LLOQ at least 5
half life lower than estimated Cmax to ensure LLOQ value is less than
5% of the C-max, but this may be difficult in some cases.
Since BQL is not a value, a numerical value or "absent value" should
be assigned to the analyte concentration for the sample per
established SOP for modeling/ plotting/ data processing purpose. We
prefer to assign "0" value to BQL for modeling purpose especially for
the first-day pre-dose sample. However, our SOP allows us to assign
other values (absent, half of LQ, etc.) to BQL sample provided there
is a sound and documented scientific rationale. In the case that a
"0" value is assigned to the BQL sample, 0 concentration is less than
5% of any C-max, the subject(s) should be included in the statistics
unless specifically excluded by the study analysis protocol.
In a few rare cases, our SOP allows us to include concentration values
below LQ for modeling if the sample is critical to the interpretation
of the results from a Research study but never for Clinical or GLP
studies.
I am eager to hear any inputs/ comments from my distinguished
colleagues in the field.
Ta Kung Chen
Back to the Top
The following message was posted to: PharmPK
Dear Dr Chen, Dear All,
Why not 1) report the concentration value for ALL samples, and
2) add whichever comment(s) you like in another data column?
Even if you doubt the sound arguments from Dr Jelliffe and many
others, the decision of how to handle an LOQ-sample should be
taken by the person who will be responsible for the PK analysis.
If you do not agree, would you please tell us the scientific rationale
for ultimately deleting a value below the LOQ from the dataset
(or any other procedure you proposed)? Please feel free to
include citations.
We may need more papers on this issue to reach a final
consensus, although a lot has been done already. However,
no SOP is bullet-proof to new knowledge or a new consensus
to be jointly achieved in the future.
Throwing away data for samples below the LOQ seems like
throwing away the hard work of many analysts to me.
All the best
Juergen
PS: Even a reported comment like: -There was a small interfering
peak in the chromatograms of all pre-dose samples.- are highly
welcome by the data analyst.
---
Juergen Bulitta, PhD, Chemist, Post-doctoral Fellow
Pharmacometrics, University at Buffalo, NY, USA
Phone: +1 716 645 2855 ext. 281, j.-at-.bulitta.com
Back to the Top
Dear Juergen,
You will not be able to measure samples below LOQ with acceptable
accuracy and precision.
Back to the Top
Dear all
If predose concentration is less than LLOQ many Pharm PK scientist
says that put it zero. If i am including in PK parameter calculation
how we can ensure the below LLOQ value is it accurate and precise.
plz explan about this question.
I look forward your advices. thank you.
by
A.Karthik
Back to the Top
Dear Dr. Karthik:
There is no science in setting a LLOQ sample to zero. The
best thing you can to is to use the actual value measured. Don't
worry about the CV%. The SD and the variance is the thing to focus
on. Look at any statistics book under Fisher information of a data
point. Why the lab community is so stubbornly blind on this point
escapes me. The error is not a percent. The error is the ERROR ITSELF.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
Dear Katie Chan:
I really think you have not thought enough about the
problem. I believe you are still thinking of accuracy and precision
as a percent error. That is really NOT the correct way. The %CV is
the result of two values, the result aand the SD with which it was
measured. Suppose you have one result of 10 units, with an SD of 0.5
units. The CV% is 0.5/10 = 0.05 = 5%. But the actual SD is 0.5. Now,
suppose you have another result. It is 0.5 units, and is measured
with the same SD. Now the CV% is 0.5/0.5 = 1.0 = 100%. It actually
has exactly the same precision. Look at just about any statistics
book and look up Fisher information. The Fisher information of a data
point is the value times 1 / the variance. This is the point. The %CV
depends on 2 values. The actual prediction is the SD or the variance.
There is no scientific reason at all for a lab to cover up
its errors by behaving as most do at present. When you get any value,
also report the SD. You will not only educate those who ask for your
analysis, but you will also serve them much better.
Best regards to all, and thanks, Juergen, for also jumping in.
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.at.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Dear Group,
I have been following the discussion on this issue and listening to the
comments of the experts. I just want to add few things from my
experience. Different organizations follow different strategies to
handle conc. which are below LLOQ.
Some replace it by LLOQ/2, LLOQ/5, LLO, zero etc.
I think whatever procedure we follow, it should be written in our SOP's.
With regard to pre-dose conc < 5% of Cmax, I think to avoid this, we
should always try to keep the LLOQ of our analytical mehtod to be less
than 1/25 of Cmax (if known).
Hope, experts can throw more light on this.
Regards
Tausif Ahmed, Ph.D.
Metabolism and PK department
Ranbaxy Research Laboratories, India
Back to the Top
Dear Tausif:
You have listed many practices people do when a value is
below the so-called LLOQ. Actually, there is no science to any of
those policies at all, nor to your thoughtful suggestion. The best
estimator of the value is simply the measured value itself. The
precision of the measurement is NOT the CV%, which is the relation
between the measurement and the actual precision with which it was
measured, namely the assay SD or variance at the value of the
measurement. Most people, even physicians who order the tests, are
able to realize that in toxicology the result has to be enough SD's
above the blank to make one reasonably confident that something is
there. That value is totally negotiable, and is not science but
rather is one's choice, just as you choose to like one wine and not
another.
For PK work, you KNOW the drug is there because you know
when it was given and when the sample was drawn. So there really is
NO LLOQ at all, and most people are not as dumb as the lab may think
when they treat people this way with results. Once again, look at
Jelliffe RW, Schumitzky A, Van Guilder M, Liu M, Hu L, Maire P, Gomis
P, Barbaut X, and Tahani B: Individualizing Drug Dosage Regimens:
Roles of Population Pharmacokinetic and Dynamic Models, Bayesian
Fitting, and Adaptive Control. Therapeutic Drug Monitoring, 15:
380-393, 1993. You might also look at almost ANY statistics book on
the credibility of a data point. For example, look at Morris de
Groot, Probability and Statistics, 2nd edition, 1989, Addison-Wesley,
under Fisher Information, pp. 420-424. It is not just me, it is real
statistics. I do not understand why the lab assay community has fixed
on the CV% instead of the very well documented Fisher information of
a data point. In this way, you can track the credibility of a lab
result all the way down to, and including, a blank. There really is
NO LLOQ.
Very best regards,
Roger Jelliffe
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.aaa.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
I would like to strongly endorse Roger Jelliffe's remarks about the
use and abuse of the lower limit of quantitation (LLOQ).
The LLOQ is a bioanalytical method validation parameter e.g. as
described in the FDA Guidance http://www.fda.gov/CDER/GUIDANCE/
4252fnl.htm. It is a criterion used to satisfy those concerned with
assay reliability. In the sense that the LLOQ is a defined parameter
of a bioanalytical method then it has an existence and is useful for
tracking assay performance. But as a concept for understanding and
interpreting the consequences of measurement error for
pharmacokinetic analysis it often misinterpreted.
The FDA Guidance does not recommend that concentration measurements
that are less then the LLOQ should be discarded. The advice is
"Estimation of concentration in unknown samples by extrapolation of
standard curves below LLOQ or above the highest standard is not
recommended". The FDA is to be congratulated on this sensible
position because it allows the responsibility for interpretation and
use of measurements less than the LLOQ to be in the hands of the
pharmacokineticist who knows how to deal with residual error. The FDA
document is unfortunately naively silent on the problem of selective
reporting of results.
Typically the non-pharmacokinetically orientated chemical analyst
invents the idea that the Guidance dictates that values less than the
LLOQ should not be reported i.e. the chemical analyst hides valuable
information from the pharmacokineticist.
The consequence of this misinterpretation and consequent deliberate
concealment of data is that pharmacokinetic analyses will necessarily
be biased. This is an inescapable fact when the random distribution
of potential measured values is truncated by not using values less
than the LLOQ. The bias will mean that terminal rate constants will
be estimated to be slower than they really are. It also means that
compartmental based analyses may detect what seems to be an extra
compartment because of the upwardly biased 'measurements' close to
the LLOQ.
If you are forced to work with a chemical analyst who will not give
you measurements less than the LLOQ then you can minimize the bias by
discarding values reported by the analyst that are less than 150% of
the LLOQ e.g. if the LLOQ is 100 ng/mL then do not use values less
than 150 ng/mL. This empirical advice is based on the assumption that
the LLOQ is defined by a proportional random error coefficient of
variation of around 20% plus a typical size of the additive non-
proportional random error which when combined gives a combined CV% of
20% at the LLOQ. It is reported values in the range of the LLOQ to
150% of the LLOQ that will be misleading if the LLOQ truncation is
applied.
The alternative is to ask your chemical analyst to read these
postings on PharmPK and ask them to consult a competent statistician
about the consequences of truncating a random distribution.
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
email:n.holford.-a-.auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/
Back to the Top
The following message was posted to: PharmPK
Dear Tausif
In my opinion and practice Cmax/64 is an ideal LLOQ. And it is
possible with the avaliability of LCMSMS techniques. Many times the
reported mean concentration has wide range of variability and Cmax
range in one study or multiple studies is quite wide (i.e. Max vlue
of Cmax minus min value of Cmax). In such cases we can take LLOQ as
(minimum reported value of the Cmax)/ 64 or 32 or 16, which ever is
suitable and necessary to support the study objectives. Of course you
need lot of considerations in terms of utility of such treatment of
the data to the objectives of the study.
with best regards
Dr.Prashant
Back to the Top
Dear Nick:
Many thanks! I often feel as though I am talking to the
wind. Thanks so much for adding your valuable input. There is no
getting around the fact that the measured value, even if below the so-
called LLOQ, is the best estimate of the true value in that specimen.
Other tactics, such as those suggested by many, are reactions to the
CV% and not to the real statistical issues. Assay precision is given
by the SD or the variance at that point, NOT by the CV%. The
laboratory community needs to hear this more, and especially to think
about it, not just to follow someone else's so-called guidelines and
tell themselves they are good guys. That is not science. As you say,
good statistics is what matters.
Very best regards,
Roger
Roger W. Jelliffe, M.D. Professor of Medicine,
Division of Geriatric Medicine,
Laboratory of Applied Pharmacokinetics,
USC Keck School of Medicine
2250 Alcazar St, Los Angeles CA 90033, USA
Phone (323)442-1300, fax (323)442-1302, email= jelliffe.-a-.usc.edu
Our web site= http://www.lapk.org
Back to the Top
The following message was posted to: PharmPK
Dear All
I agree with "The error is the ERROR ITSELF".
Also practically it is not possible to have zero LLOQ. If we agree
to have this, then can we measure 0.000001 pg/ml of analyte in the
matrix by using any analytical technique available commonaly.
Moreover can we distinguish between 10,000 pg/ml and 10,001 pg/ml
samples repeatedly and accurately? Practically No. And there is no
need to correlate to such values as this kind of data may not be
therapeutically significant.
Primarily we are trying to quantify reasonably the therapeutic
response to quantifyable levels of drug or metabolites. In most of
the cases there is a threshold (of dose) for initiation of
significant (not the statistical but literal) and measurable
therapeutic response. It therfore, is not necessary to have zero LLOQ.
Is it going to be significant to measure 10,000 and 10,001 pg/ml? I
sincerely believe no. The analytical systems we have can not adjust
themselves to the biological variation to predict such minor
variation. The therapeutic response at both these concentrations is
NOT going to be different. I am sure those in the group who deal with
patients directly or have dealt in the past can vouch for this.
Thanks and regards
Dr.Prashant
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)