Untargeted, semi-targeted and targeted analytical approaches

There are three analytical chemistry strategies applied in metabolomics – untargeted, semi-targeted and targeted.

Both untargeted and semi-targeted approaches are applied in hypothesis-generating studies, while targeted assays are usually applied to validate and translate the novel discoveries of a hypothesis-generating study. The major differences between untargeted, semi-targeted and targeted studies are:

• the level of sample preparation required,
• the number of metabolites detected, and
• the level of quantification of the metabolites.

The objectives of a study define which of the three analytical strategies are applied.

Untargeted approaches

Untargeted approaches provide the most appropriate route to detect unexpected changes in metabolite concentrations; the aim is to maximise the number of metabolites detected and therefore provide the opportunity to observe unexpected changes. Hundreds to thousands of metabolites can typically be measured. However one single analytical method cannot detect all of the metabolites in a biological system. It’s therefore desirable to combine multiple analytical approaches (such as complementary UPLC methods) to maximise the number of metabolites detected and increase coverage of the metabolome.

Sample preparation in untargeted studies typically involves the extraction of the metabolites from the biological sample in to a suitable solvent for the analytical analysis. The extracted sample is analysed by an appropriate analytical method (for example, reversed phase liquid chromatography mass spectrometry). The peak area of each metabolite (determined in the liquid chromatography mass spectrometry analysis) is used as the parameter in the statistical analysis to define the concentration differences between the different biological samples being measured. This is referred to as relative quantification as there is no comparison to calibration curves constructed with chemical standards. The use of calibration curves is required for full quantification. The biological importance of each metabolite is determined during data analysis and metabolite identification, and biological interpretation is performed at the end of the experimental pipeline. The chemical identity of each metabolite in the study is not known at the start of the study. Data acquired during the data acquisition is applied to identify metabolites after data acquisition.

Currently, one of the major limitations in untargeted approaches is the identification of metabolites. It may not be possible to identify the metabolites highlighted in the statistical analysis as significantly changing between the biological classes in the study. Metabolite identification is currently one of the hot topics in metabolomics.

Targeted approaches

Targeted studies analyse a relatively small and specific number of metabolites, typically up to twenty metabolites. These metabolites are chemically characterised and biochemically annotated with established biological importance at the start of the study before data acquisition is performed. Targeted methods have a greater selectivity and sensitivity than untargeted methods. A targeted study can only be performed if an authentic chemical standard of the metabolite is available. Quantification of the metabolite is performed through the use of internal and the chemical standards to construction calibration curves for each of the metabolite in the study.

The sample preparation in targeted studies applies methods that can be optimised to retain the metabolites of interest and to remove other biological species and analytical artefacts which are not carried through to the downstream analysis. The analysis of the data and interpretation of biological significance is much simpler in targeted studies.

Semi-targeted approaches

Semi-targeted methods fall in the middle between untargeted and targeted approaches. This approach aims to quantify hundreds of metabolites whose identity is known before data acquisition. The method typically applies one calibration curve for a number of metabolites of similar chemical structure, rather than apply a single calibration curve for each metabolite. So in a semi-targeted approach targeting 100 metabolites you may have 10-15 calibration curves.