Metabolomics for Metabolic Disease

national
bioinformatics
metabolomics
Untargeted and targeted metabolomics to discover blood biomarkers detectable years before Type 2 Diabetes diagnosis, and lipidomic signatures of subclinical cardiovascular risk in diabetic patients.
Keywords

metabolomics, lipidomics, Type 2 Diabetes, biomarkers, early detection, cardiovascular risk, LC-MS, MAIT, FELLA, mWISE, UPC, Ministerio de Ciencia

Funding Ministerio de Ciencia e Innovación (AEI)
Programme Proyectos de Generación de Conocimiento / Proyectos I+D+i
Status Active
Clinical focus Type 2 Diabetes · Cardiovascular disease
B2SLab PI Alexandre Perera Lluna

Context: beyond glucose — the molecular signature of metabolic disease

Type 2 Diabetes (T2D) is typically diagnosed when blood glucose passes a clinical threshold. But the metabolic disruption that produces T2D develops over years or even decades before that threshold is crossed. By the time of clinical diagnosis, many patients already have early microvascular damage, elevated cardiovascular risk, and established insulin resistance — changes that could have been detected, and potentially reversed, had molecular biomarkers been measured years earlier.

Metabolomics — the comprehensive measurement of small-molecule metabolites in blood, urine, or tissue — provides a window onto the metabolic state of an organism at a resolution far greater than standard clinical chemistry. By measuring hundreds or thousands of metabolites simultaneously using liquid chromatography-mass spectrometry (LC-MS), it is possible to identify molecular signatures that distinguish pre-diabetic from normoglycaemic individuals years before any glucose abnormality appears.

B2SLab’s metabolomics programme pursues two distinct but complementary questions: what can we detect before T2D to prevent it? and what can we detect after diagnosis to prevent its most serious complications?

Research line 1: early biomarker discovery for Type 2 Diabetes

The pre-diagnostic window

Population cohort studies — where participants are followed prospectively from health to disease — create an opportunity to identify molecules that predict future T2D with a multi-year lead time. B2SLab analyses metabolomic data from such cohorts, applying both untargeted metabolomics (discovery of novel markers without prior hypotheses) and targeted metabolomics (precise quantification of pre-specified candidate compounds).

Note

What pre-diagnostic metabolomics can reveal

Untargeted LC-MS analysis of baseline samples from individuals who later developed T2D versus those who remained normoglycaemic identifies metabolites that diverge between groups years before diagnosis. These include:

  • Amino acid and nucleotide derivatives reflecting altered protein and purine metabolism in insulin-resistant tissues
  • Lipid species associated with impaired fatty acid handling in early insulin resistance
  • Gut microbiome-associated metabolites connecting intestinal microbial activity to systemic metabolic state

The goal is to identify candidates that are not only statistically discriminating but mechanistically interpretable — molecules that point to specific biological processes amenable to intervention.

Computational challenges

Pre-diagnostic metabolomics is computationally demanding. Raw LC-MS data requires extensive preprocessing before biological insight can be extracted: peak detection, alignment across samples, adduct annotation (distinguishing true metabolite signals from instrument-generated artefacts), batch correction, and missing value handling. B2SLab has developed and maintains a suite of open-source tools that address each step of this pipeline:

  • MAIT (Metabolite Automatic Identification Toolkit) — an R/Bioconductor package providing a complete end-to-end workflow for LC-MS untargeted metabolomics: peak detection, feature alignment, statistical analysis, and biological interpretation
  • FELLA — a network-based pathway enrichment tool that places metabolite hits in their biological context using KEGG pathway and reaction graphs, enabling richer interpretation than standard overrepresentation analysis
  • mWISE (Metabolite Wise Identification of Structural Entities) — a tool for adduct annotation and metabolite identification in untargeted LC-MS, reducing the high false positive rate that plagues automatic annotation

Research line 2: lipidomics and cardiovascular risk in diabetes

The lipidome as a risk predictor

Patients with T2D face a cardiovascular risk that is two to four times higher than age-matched controls, yet standard risk calculators — based on total cholesterol, LDL, and HDL — miss the majority of excess risk in this population. The reason is that cardiovascular risk in diabetes is not primarily driven by the lipid species that conventional assays measure, but by specific lipid classes and subclasses — ceramides, lysophosphatidylcholines, plasmalogens, diacylglycerols — whose concentrations in blood diverge significantly in diabetic patients with and without early cardiovascular disease.

Tip

Why lipidomics adds value over standard lipid panels

The human lipidome comprises thousands of distinct molecular species organised into dozens of structural classes. High-resolution LC-MS can quantify several hundred of these simultaneously. In the context of T2D:

  • Ceramide species (particularly C16:0, C18:0, C24:1) are emerging risk markers for cardiac events that outperform LDL in diabetic cohorts
  • Lysophosphatidylcholine levels correlate with endothelial dysfunction and inflammation independent of traditional lipid markers
  • Plasmalogen species, which protect cell membranes against oxidative damage, are depleted in patients with established cardiovascular disease

Identifying which lipid signatures predict cardiovascular events in T2D patients — and doing so before clinical events occur — could transform risk stratification for one of the most common chronic disease combinations in European healthcare.

Approach

B2SLab applies targeted lipidomic profiling of serum samples from diabetic patient cohorts, combining:

  1. High-resolution LC-MS/MS quantification of prioritised lipid classes
  2. Statistical models relating lipid profiles to subclinical markers of cardiovascular risk (intima-media thickness, arterial stiffness, silent ischemia)
  3. Integration with genomic and clinical data to identify patient subgroups with distinct lipid-driven risk profiles

Bioinformatics infrastructure

The tools developed for B2SLab’s metabolomics programme are freely available to the research community and have been used by groups internationally:

Tool Function Repository
MAIT Complete LC-MS untargeted metabolomics workflow Bioconductor
FELLA Network-based metabolomics pathway enrichment Bioconductor / software page
mWISE LC-MS adduct annotation and metabolite identification GitHub