Estimates of LOQ using regression vs S/N for SEC methods

[oldtimer]

I am trying to apply regression data (RMS of line and slope; 10 x RMS/slope = LOQ) to LOQ estimates for SEC minor components. Results vs using S/N ratios are 10-20 fold higher than S/N estimates (10 x S/N = LOQ) of LOQ. Is application of regression data reliable using these minor components that have USP peak widths in the range of 1.5-1.8 minutes and higher, naturally, precision values?

[tom jupille]

For LLOQ, I would think the regression-based determination would be more reliable (certainly, more conservative) than S/N-based.

[zokitano]

I agree with Tom.
Determining the LOQ from regression line is better.
LOQ = Xav + (10 x sd)

Xav = average response gained from the blank
sd = standard deviation of the samples' responses