An epidemiological introduction to human metabolomic investigations

Abstract

Metabolomics holds great promise for uncovering insights around biological processes impacting disease in human epidemiological studies. Metabolites can be measured across biological samples, including plasma, serum, saliva, urine, stool, and whole organs and tissues, offering a means to characterize metabolic processes relevant to disease etiology and traits of interest. Metabolomic epidemiology studies face unique challenges, such as identifying metabolites from targeted and untargeted assays, defining standards for quality control, harmonizing results across platforms that often capture different metabolites, and developing statistical methods for high-dimensional and correlated metabolomic data. In this review, we introduce metabolomic epidemiology to the broader scientific community, discuss opportunities and challenges presented by these studies, and highlight emerging innovations that hold promise to uncover new biological insights.

Keywords: epidemiology; high-dimensional statistical methods; integrative omics; metabolites; metabolomic epidemiology; quality control.

Figure 1.. Network analysis of scientific topics that co-occur with “metabolomics” in PubMed.

Data were collected from PubMed using “metabolomics” as the search keyword and filtering for co-occurring scientific keywords. The number of co-occurrences was used as the edge weight between keywords. A total of 1,542 publications, including clinical studies (Phases I-V) and randomized control trials were used in this analysis. From these, 6,283 keywords were extracted and 102 keywords with 30 co-occurrences were used to generate the figure using VOSviewer. Each node is a keywork and each edge represents the weight of the co-occurrence between keywords. Colors represent clusters of relevant keywords.

Figure 2.. Distribution of metabolomic studies over time by A) platform and B) tissue type.

PubMed search terms used for A included: metabolite*, metabolom*, or metabonom* with the key term NMR, GC-MS, or LC-MS. PubMed search terms used for B included: metabolite*, metabolom* or metabonom* with the key term urin*, plasma, serum, saliva/sputum, faecal/fecal, cerebrospinal fluid/CSF, or tissue. Years included are 1966-2021 with searches conducted on November 11, 2022.

Figure 3.. Drift and batch effect correction to increase the biology-to-noise ratio in metabolomic data.

LC-MS is shown here as an exemplary technique. A) We illustrate a common approach for metabolite sample quality control (QC) implemented in a study. Two pooled references for QC (column labels starting with “RA” shown in red and “RB” shown in blue) and internal standards (demonstrated by the first row) are used to capture trends introduced during the sequential processing of samples (column labels starting with “S”). Sample colors represent varying metabolite intensities. Three unique metabolites are represented by rows two, three, and four. The x-axis represents the order in which the samples are injected to the ionization source (injection order). B) Internal standards (also represented by row 1 of A) are specific metabolites that are added to samples as a baseline to adjust for intensity drift introduced by sample injection order. C) Pooled references are aliquots of all samples used as a baseline for all metabolites we expect to see among samples and are used complementarily with internal standards to adjust for drift introduced by sample injection order. D) Large studies require more than one column in LC-MS approaches to process samples and introduce a column effect combined with E) injection order effect that need to be considered during batch effect correction. F) Samples collected at two different time points can significantly differ and be a source of confounding. G) After correcting metabolites for batch effects and sample collection time, sex differences can still be observed. Principal coordinate analysis (PCo) reflects the metabolite variation explained by PCos, and t-distributed stochastic neighbor embedding (t-SNE) is used to visually demonstrate the metabolite similarity between pairwise samples.