Beacuse LLOD/LLOQ determination using signal to noise ratio gives a value that is generally specific to the instrument on which it is run, I find it very difficult to consider this as a useful tool. Is there a better way to determine LLOD/LLOQ of a particular analyte? Also, when determining the signal to noise ratio, what is the best way to measure noise? RMS...?

The FDA's comments on the matter are as follows (taken from "Reviewer Guidance
Validation of Chromatographic Methods" available on the www.fda.gov website):

"Though USP expresses detection limit and quantitation limit in terms of 2 or 3, and 10 times noise level respectively, this concept is not very practical. Noise level on a detector during the method development phase may be different when samples are assayed on different detectors, etc. The use of standard(s) in the test method at the quantitation limit level (proposed by the applicant) is assurance that the impurity can be observed and quantitated."

I've attached a copy (I hope!)

As far as measuring noise goes, there is an ASTM practice for noise measurement embedded in their standard practice for testing UV detectors (ASTM E-1657-94). I can't attach that one because it is copyrighted by ASTM, but you should be able to order a copy from their web site: www.astm.org.

-- Tom Jupille / LC Resources Inc.

fda-valid.pdf fda-valid.pdf (82 k)

You can look up the ICH guideline which describes several acceptable ways to determine LOD and LOQ. Regardless of which way you are going to use, you have to demonstrate by injecting the standard solution at that concentrations. One simple way is to measure the peak height of the signal and noise of your analyte injection manually. Another way is based on the standard deviation of your noise in the blank/solvent injection. If you are using the software to calculate the noise, you have to be aware of the algorithm it used e.g. the AccessChrom includes the drift in the noise value!