Charl ● 2007-07-19 09:37 (6098 d 08:06 ago) Posting: # 907 Views: 7,104 |
|
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 |
|
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:
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 ★★★ Vienna, Austria, 2007-07-19 15:16 (6098 d 02:27 ago) @ Charl Posting: # 912 Views: 6,693 |
|
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:
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. 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! Helmut Schütz The quality of responses received is directly proportional to the quality of the question asked. 🚮 Science Quotes |