Chromatography Forum

In Related Substances Method, As per ICH limit for LOD is 3:1 or 2:1 and for LOQ 10:1 mentioned.
IS it require LOQ shlould always come more than 10? or if its coming around 8 is it acceptable? as the concentration of analyte remains fixed we cannot change the concentration.

IS it require LOQ shlould always come more than 10? or if its coming around 8 is it acceptable?

If they had meant 8 to be acceptable, they probably would have said so.

Actually LOQ is defined as the lowest level which can be determined with the requisite degree of accuracy and precision. A better way to verify LOQ is to run replicates at that level to confirm that the standard deviation is acceptable. If you do that, be sure to have an SOP to that effect, and to do it that way every time.

% RSD from Six replicate for LOQ should be NMT 10 % and s/n ration 10:1 is the acceptance criteria.
% RSD from Six replicated for LOQ level is within the acceptance criteria but average s/n is 8:1. is is require that the s/n ration should come more than 10:1

If the acceptance criteria say 10, then 8 is not acceptable.

I recommend not using S/N as a criterion when developing and validating a method, but in the case of an existing method you have to follow it as written.

Hi porens, Hi Tom

What method you suggest for LOD/LOQ Tom?

For LOD, the approved alternative is based on the standard error of the y-intercept.
LOD = 3.3 x SE / slope
In any case you must run a standard at approximately the LOD to verify that the analyte can be detected (regardless of how LOD was estimated).

For LOQ, the equivalent is
LOQ = 10 x SE / slope
This is better than S/N (if for no other reason than it avoids the question of where and for how long you measure noise), but it is inadequate in the sense that, like S/N, it's a single value and doesn't account for the "with the required level of accuracy and precision" part of the LOQ concept. Let's face it, if you must quantitate with a CV of +/- 1%, you will not be able to do that at S/N = 10.

The best way is to generate a CV vs A plot (CV is % RSD; A is the amount of analyte) by running 3 - 5 replicates at each of multiple levels starting below what you expect LOQ to be, and then plotting the %RSD as a function of the amount of analyte. It's usually done using logarithmic axes, and looks something like this (I've used arbitrary units for the X-axis; imagine that they represent nanograms):



Now you can look at the plot and relate the LOQ to the required CV. In this case, the LOQ for 2% RSD is about 10ng. The LOQ for 10% RSD is about 4 ng, and so on.

There is lots more you can do with the information, of course, by looking at the distribution of residuals from the regression, testing for goodness of fit, etc.