9.1. Definitions and important aspects

Limit of Detection

http://www.uttv.ee/naita?id=23290

https://www.youtube.com/watch?v=PnYWCM8lkzs

In this course we define the limit of detection (LoD) (also called detection limit) as the smallest amount or concentration of the analyte in the test sample that can be reliably distinguished from zero [ref 12]. It should be noted, however, that there are different approaches for defining a limit of detection which can cause confusion [ref 12]. One of these approaches (examined more thoroughly in section 9.2) is by defining two separate limits by probability of false positive and false negative errors. To avoid confusion in this course, these definitions will be called a decision limit (CCα) and a detection capability (CCβ).

LoD is used for two purposes:

• To estimate whether the analyte is detected with the used measurement procedure in the sample by comparing the LoD value and the result obtained from analysis of the sample.
• To characterize the analytical method in terms of its ability of detecting low levels of analytes and compare it to other methods, laboratories or standards.

It can be seen, that in the first case, LoD is used for making decisions on analyzed samples (whether we can claim that the analyte is detected or not), in the second case LoD is used for making decisions on the  analytical methods (how low analyte levels is the method able to detect).

In Table 1 (of Section 9.3) different approaches of how to estimate LoD with detailed information are presented. Only widespread approaches, taken from the prominent guidelines and articles, are presented, thus it is not an exhaustive list of all possible approaches. Approaches can be based on the different statistical assumptions and therefore, the obtained LoD values often cannot be strictly compared. In some cases even results from the same approach should not be compared if important differences exist in the estimation procedure. How to choose a suitable approach to estimate LoD is discussed in following sections.

It is also possible to calculate LoD and LoQ in software. This software allows the user to choose the guideline that must be followed, specific estimation approaches that the user wants to apply and to input the data. All calculations (including plotting the residuals) are done automatically.

Not all LoD estimation approaches are fit for all analytical methods. Therefore, as LoD is an important validation parameter, these problems must be understood by the analyst. LoD can then be estimated and used correctly for the particular method used. Section 9.3 gives guidance relevant for LC-MS methods.

In many cases regulations set the maximum allowed limits for the analytes (e.g. pesticides) in certain matrices (e.g. different food products). This limit is called the maximum residue limit (MRL).

Two aspects must be considered when carrying out analysis in the framework of a regulation with the aim of establishing whether the analyte level in the sample exceeds the MRL: (a) is the analytical method capable of detecting the presence of the analyte in the matrix at levels lower than MRL, and (b) from which analyte content in the sample, as determined by the method, we can with high probability conclude that the actual analyte content in the sample is over the MRL. To answer the first question, the analyst must compare the MRL value to the LoD value. For example by the European Commission standards the method LoD to estimate cadmium in drinking water should be 10 times lower than the MRL value [ref 13]. The second question can be answered by using decision limit (CC_α), which is discussed further in section 9.2.

Limit of Quantitation

http://www.uttv.ee/naita?id=23629

https://www.youtube.com/watch?v=SHZ4Zik3uFg

Limit of quantitation (LoQ) is defined as the lowest analyte concentration in the sample that can be determined with an acceptable repeatability and trueness. LoQ can also be called lower limit of quantification, limit of quantification, quantification limit, quantitation limit or limit of determination.
LoQ is determined by the required trueness and precision. Therefore, LoQ is not a limit set by nature and quantitative information at analyte concentrations below the LoQ exists. However, such quantitative information should be used with care as the relative uncertainty of the results below LoQ can be too high for further use.

Instrumental LoD and method LoD

http://www.uttv.ee/naita?id=23294

https://www.youtube.com/watch?v=PSHiS56U2bo

A distinction has to be made between the instrumental LoD and the method LoD. Instrumental LoD is found from the analysis of the analyte in pure solvent. Therefore, instrumental LoD only shows the capability of the instrument to detect the analyte and can only be used for comparing different instruments. Method LoD also takes into account the effects that sample preparation and measurement procedure have on the analysis result. The samples used to estimate a method LoD must be matrix-matched and must go through the whole sample preparation procedure. Any conclusions related to detection ability of a method have to be made using method LoD (not instrumental LoD). This course addresses only method LoD.

Interpretation of analysis results with LoD and LoQ

http://www.uttv.ee/naita?id=23246

https://www.youtube.com/watch?v=oEIG2cPy5fU

In simplified cases – i.e. if it is not critically important to accurately state whether the analyte can be detected or not (1)– the interpretation of the results obtained with a specific sample should be done in the following way:

• when the analyte cannot be detected or its concentration in the sample is found to be below LoD, it can be said that the analyte content is below LoD;
• when the analyte content is above LoD, but below LoQ, then it can be said that the analyte is present at trace levels, but usually no quantitative data are reported;
• when analyte content is at or above LoQ, then the result can be quantified.

It is also important to state the LoD value of the analysis method as a result below LoD does not indicate that there is no analyte in the sample, but only that the used analysis method is not capable of detecting the analyte in this sample.

If the analyte is found to be above LoD, but below LoQ, then simply stating that the analyte is present in the sample and not providing any quantitative information, only gives part of the information available from the analysis. However, it is often still suggested that the measured value and its uncertainty should be reported even if the results are below LoQ (or even LoD), so that the quantitative data can be used for further calculations if necessary. For example, in case of pharmacokinetic studies it has been suggested that results below LoQ can be given out so that better pharmacokinetic models could be built. Further discussion on the topic can be found at the following reference [ref 58].

It must be understood that the LoD value might not account for false positive and false negative results. Therefore, more correct interpretation of the results can be given with CC_α and CC_β values which are discussed further in the following section.