Charl
●    

2007-07-19 09:37
(6098 d 08:06 ago)

Posting: # 907
Views: 7,104
 

 LLOQ [Bioanalytics]

Dear all

Can I report concentrations that has 20% less values than the Lower Limit Of Quantitation in my final study report? once we already accepts these values in our validation as a part of accuracy measures...! ;-)
Jaime_R
★★  

Barcelona,
2007-07-19 12:37
(6098 d 05:06 ago)

@ Charl
Posting: # 909
Views: 6,257
 

 LLOQ

Dear Charl!

❝ Can I report concentrations that has 20% less values than the Lower Limit Of Quantitation in my final study report?


No, because they are below the ‘Lower Limit of Quantitation’.

❝ once we already accepts these values in our validation as a part of accuracy measures...! ;-)


That’s another story.
These values were back-calculated spiked (known) samples. Their purpose was to establish an LLOQ of accepted accuracy and precision.
Actually it’s a two step procedure:
  • you assume an LLOQ:
    setting up the calibration curve, the QCs, the LLOQ-samples accordingly
  • you run your validation: if at the assumed LLOQ you find:
    inaccuracy (bias) ≤20% and precision ≤20%, then your LLOQ is established
If you feel that your measurements at LLOQ were too accurate and/or precise you are free to repeat this procedure.
But once you have established the LLOQ, this represents a definite number you will have to follow for your unknown samples (fuzzy logic is not applicable).

To stay with your example: if we measure an unknown at exactly LLOQ, we know that it’s ‘true’ concentration may be even <80% of LLOQ due to the accepted inaccuracy/precision.
Leaving out any statistical issues, if you suggest reporting a value at 80% of LLOQ, the true value may be <0.8×0.8 = <64% of LLOQ (whereas you assume a ‘true’ value of >LLOQ, which was only brought down by chance).
And so on and so forth.

At least this is the ‘message from the frontline’ in BA/BE (in other words, the only way regulators will accept your study).

I know that there’s another point of view represented by pharmacokineticists (especially population PK modelers). They will fight for any single data point above LOD, argueing that they are able to incorporate measurement error into the model.
Besides statistical problems (negative concentrations are impossible, unproven distributional assumptions, etc.) there remains one unresolved issue:
We simply have no idea about the bias of values between LOD and LLOQ.

Regards, Jaime
Helmut
★★★
avatar
Homepage
Vienna, Austria,
2007-07-19 15:16
(6098 d 02:27 ago)

@ Charl
Posting: # 912
Views: 6,693
 

 LLOQ

Dear Charl,

we must not consider validation as an end in itself, but as the

“demonstration of the method’s suitability for use”.


We expect minimum concentrations (i.e., which will allow a reliable coverage of 80% of AUC) we develop a method suitable for achieving this goal we validate our method to demonstrate this assumption.

If during the course of a study you find out that your method is insufficient, IMHO you have two options:
  1. stop analysis, revalidate (e.g., for a lower LLOQ), and re-analysis
  2. simply ignore it; subjects with insufficient data may have to be excluded by the statistician from assessments
If you opt for method #2, you are risking insufficient power (e.g., in BE), or a drop below the minimum sample size (e.g., in BA- or PK-studies).
Please note: IMHO, it’s not a good idea to go with method #2, discover bioINequivalence, and afterwards switch back to method #1. Regulators may interpret such a procedure as data mining:cherry picking:
Therefore, I would recommend a blinded plausibility review as soon as possible (i.e, already during bioanalytics; not after its completion).

See also this post and followings.
The topic of values between LOD and LLOQ is leading to controversial discussions approximately every second month at David Bourne’s PKPD-list.
To get an overview use the search function; suggested keywords: ‘LOQ’, ‘limit of quantification’, etc.
There was a rather lengthy discussion in May/June 2007.

Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes
UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
105 visitors (0 registered, 105 guests [including 7 identified bots]).
Forum time: 16:44 CET (Europe/Vienna)

With four parameters I can fit an elephant,
and with five I can make him wiggle his trunk.    John von Neumann

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5