Metabolomics Strategies
Learn more about different metabolomics approaches and Bevital´s outstanding platforms
Which Metabolomics Strategy Should You Choose?
Untargeted, semi-targeted, and targeted metabolomics are three distinct strategies for studying metabolites. Selecting the most suitable approach depends on various factors, such as the research objectives and the study hypothesis.
Untargeted Metabolomics
Untargeted metabolomics is a hypothesis generating approach, analyzing as many metabolites as possible without prior knowledge of their identity. It’s used for broad, comprehensive profiling and discovery of new or unexpected metabolites.
Semi-Targeted Metabolomics
Semi-targeted metabolomics focuses on a subset of metabolites, often from a specific pathway or class, but without the strict quantification requirements of targeted approaches. It combines some level of metabolite selection with exploratory analysis.
Targeted Metabolomics
Targeted metabolomics specifically quantifies a predefined set of known metabolites. It is highly sensitive and precise, often used for hypothesis-driven research or validation of known metabolic pathways.
General Comparison
Targeted approaches are typically used for hypothesis testing, whereas untargeted methods are employed to generate hypotheses and discover new or unknown metabolites (Table 1). As a result, targeted metabolomics usually covers a limited range of analytes, typically around 50-200 metabolites, while untargeted approaches provide broader data, often identifying more than 10,000 so-called features.
When it comes to data quality, targeted approaches generally surpass untargeted methods for several reasons. Untargeted metabolomics provides relative quantification, limiting the ability to compare data across different experiments, labs, cohorts, or studies. In contrast, targeted metabolomics delivers absolute quantification (typically in nmol/L or µg/mL), which is essential for clinical diagnostics and especially valuable when validation and regulatory approval are required. Additionally, untargeted methods lack standardization, leading to lower assay reproducibility and often resulting in inconsistent findings across studies. Targeted approaches also typically offer higher specificity and sensitivity, as they are optimized for a defined set of well-characterized analytes. As a result, automation is more feasible in targeted workflows, with analytes measured under well-established parameters. Finally, data interpretation is faster and more cost-effective with targeted approaches, while untargeted methods require extensive and costly statistical analysis tools.
Semi-targeted metabolomics is a middle ground, suitable when researchers need both breadth and specificity, typically around defined biochemical pathways. It doesn’t fully replicate the discovery power of untargeted or the precision of targeted metabolomics, but it’s a versatile approach for studies with defined focus.
For a detailed comparison of targeted and untargeted metabolomics we recommend the talk by Dr. David S. Wishart from October 2022: “Why targeted metabolomics is essential for population health”.
| Feature | Targeted | Untargeted |
|---|---|---|
| Discovery | Low: Primarily focuses on validating or quantifying pre-defined metabolites. It is not designed to discover novel or unexpected compounds. | High: The primary strength of untargeted metabolomics. Aims to detect and identify a broad range of metabolites, including novel, unknown, or unexpected compounds, leading to new hypotheses and biomarker identification. |
| Coverage | Limited, focused: Measures a predefined, relatively small set (tens to a few hundreds) of known metabolites. | Broad, comprehensive: Aims to detect and identify as many metabolites as possible (hundreds to thousands), including novel or unknown compounds. |
| Quantification | Absolute Quantification: Typically uses stable isotope-labeled internal standards and calibration curves for precise and accurate quantification of specific metabolites. | Relative Quantification: Provides relative changes in metabolite levels between samples (e.g., fold changes). Absolute quantification is challenging due to the lack of standards for all detected compounds. |
| Standardization | High: Methods are highly standardized, with established protocols for sample preparation, instrument parameters, and data processing. Relies on validated analytical methods for specific analytes. | Lower: Due to the discovery-driven nature and broad scope, standardization is more challenging. Variations in sample preparation, analytical platforms, and data processing workflows can impact results. |
| Reproducibility | High: Excellent reproducibility due to well-defined analytical parameters, use of internal standards, and focus on a limited set of analytes. | Moderate to Good: Can be challenging due to the complexity of the data, potential for noise, and variability in metabolite identification. Robust quality control (QC) practices are crucial to improve reproducibility. |
| Specificity | High: Designed to specifically detect and quantify a particular metabolite, minimizing interference from other compounds. Uses specific precursor-product ion transitions. | Lower (for individual metabolites): While sensitive to many compounds, the high number of detected features can make precise identification and confirmation of individual metabolites challenging, especially for unknown compounds. |
| Sensitivity | High: Optimized for the sensitive detection of the targeted metabolites, often achieving low limits of detection (LODs) for specific compounds. | Variable: Overall sensitivity is good for detecting a wide range of compounds, but for any single metabolite, it might be lower than a highly optimized targeted assay for that specific compound. |
| Automation | High: Highly amenable to automation, from sample preparation to data acquisition and initial processing, due to defined workflows. | Moderate: While data acquisition can be automated, significant manual intervention is often required for data processing, peak annotation, and metabolite identification due to the complexity and sheer volume of data. |
| Data Interpretation | Straightforward: Data is typically presented as absolute or relative concentrations of known metabolites, making interpretation and pathway mapping relatively direct and hypothesis-driven. | Complex: Generates large, complex datasets requiring sophisticated bioinformatics tools, multivariate statistical analysis, and often manual curation for metabolite identification. Interpretation is more discovery-driven and can lead to new hypotheses. |
| FDA Approval | More Feasible (for specific assays): While there aren’t many broadly “FDA-approved metabolomics tests” as a whole, individual targeted metabolomics assays can and are being developed and validated for specific diagnostic purposes (e.g., inborn errors of metabolism screening). They benefit from the high specificity, quantification, and reproducibility that align with FDA’s requirements for in vitro diagnostics (IVDs) or Laboratory Developed Tests (LDTs), especially when adhering to Good Clinical Practice (GCP) and Good Clinical Laboratory Practice (GCLP) standards. | Highly Challenging (for broad application): Direct FDA approval for a general “untargeted metabolomics test” is currently very difficult, if not impossible, due to the inherent complexity, lack of comprehensive standardization across all detectable metabolites, and the discovery-oriented nature. While untargeted metabolomics data can be used in FDA submissions for drug development (e.g., to understand drug mechanism of action or safety), the assay itself for broad diagnostic claims would face significant hurdles. Untargeted approaches are primarily used in the discovery phase to identify potential biomarkers that can then be validated using targeted methods for potential regulatory approval. |
Targeted Metabolomics by Bevital
During the last 20 years Bevital´s scientists and technicians have specialized in targeted metabolomics and our panels have been applied in many hypothesis-driven projects and biomarker validation studies.
Bevital’s diverse analytical platforms are designed to be both analytically and biologically complementary, established across dedicated GC- and LC-MS/MS systems. As a result, our metabolomic platforms enable quantification of a wide range of related metabolite classes, spanning both low- and high-abundance compounds within physiologically relevant dynamic ranges from pmol/L to mmol/L. Unlike untargeted approaches, which often produce missing data for low abundance metabolites while “fishing on the surface of the ocean” (Fig. 1), targeted methods allow to “explore the deep sea ” and quantify metabolites within their physiological concentration range.
Fig. 1: Dynamic range. Exploring the “Deep Sea of Metabolomics” by targeted metabolomics. Various metabolites and biomarkers are plotted against their physiological concentrations and spread in the general population.
Isotope-labeled internal standards (ILIS) are considered the gold standard for quantification in metabolomics. However, because authentic ILIS (AILIS) are either unavailable or very expensive for certain metabolites, most targeted approaches and metabolomics labs use non-authentic ILIS for many of their analytes. In contrast, Bevital’s methods exclusively rely on authentic ILIS for each analyte, enabling us to provide both relative and absolute metabolite quantification of the highest quality, consistently validated through participation in external quality programs. As demonstrated by figure 2, AILIS offer higher precision with 3-7 times lower CVs (Ulvik et al.), which is crucial in small cohort studies often suffering from limited statistical power to detect true biological differences. Additionally, internal data from our group indicate that quantification using non-authentic ILIS can lead to false-positive results. These artifacts, known as spurious correlations since described by Pearson in 1897, can be avoided only through targeted approaches specifically designed to quantify each analyte using its AILIS.
Fig. 2: Precision. Comparison of CV (%) of selected analytes by use of authentic and non-authentic ILIS.
