Experimental Comparison of the Different Approaches To Estimate LOD and LOQ of an HPLC Method

Detection and quantification limits (LOD and LOQ) are two fundamental elements of method validation. Rigorous statistical definitions exist, but in HPLC they could not be implemented. Nevertheless there are several estimation methods for these limits. The most commonly used is the signal-to-noise ratio criterion. Others are based on the dispersion characteristics of the regression line, either simple or weighted. For LOQ, Eurachem proposed an alternate approach based on the use of a target value for the area RSD. Since official guidelines imposed no particular modus operandi, an experimental methodology was set up to investigate the compatibility of the different approaches and their respective reliabilities. Several samples prepared in a concentration range close to the limits were analyzed. It appeared that, both for values and their reliabilities, the different approaches were far from equivalent. In our opinion, the best way to handle the problem of detection and quantification limits was a methodology based on the use of the residual standard deviation of a weighted regression for LOD and on a Eurachem approach for LOQ. Values obtained by these means had the advantage of being reliable, i.e., with a small dispersion, and were still compatible with those obtained with the usual signal-to-noise ratio approach.