Multi-Omics: Immunometabolic Signature in Endometriosis
Confirming endometriosis still usually means laparoscopy, and most patients wait years to get it. A multi-omics study in Frontiers in Endocrinology (August 2026) asked how much of that signal survives in a plain blood draw. Wenwei Pan and colleagues profiled serum from 88 women (44 with endometriosis, 22 with benign ovarian cysts, 22 healthy controls), pairing the Olink Target 96 Inflammation panel with untargeted metabolomics on a UHPLC-Q Exactive Orbitrap. The metabolites carried most of the discriminatory weight; the proteins mainly explained why.
Key Takeaways
Multi-omics profiling of serum from 88 women tied endometriosis to coordinated inflammatory and metabolic shifts measurable without surgery.
Metabolites outperformed inflammatory proteins as standalone classifiers. N-lactoylvaline reached an AUC of 0.815 against healthy controls after FDR correction.
Combined protein plus metabolite panels beat any single marker, the practical case for cross-omics analysis in diagnostic development.
The cohort was single-center with no external validation, so these are leads for replication, not a clinic-ready test.
What Multi-Omics Integration Added to a Small Serum Cohort
Both layers came off the same 88 samples in a single experimental batch, which matters more than it sounds. Proximity extension assays of the kind described by Assarsson and colleagues (2014) give tight NPX reproducibility but see only the 92 proteins on the panel. The untargeted side was broader and messier. Together the datasets yielded 1,464 quantified features, each corrected feature-wise with Benjamini-Hochberg before any diagnostic claim.
Key Findings
Only two proteins separated endometriosis from benign cysts: CCL23 up, FGF21 down at |log2FC| > 0.58 and q < 0.05, a thin margin across 92 assayed proteins.
Metabolic disturbance ran far broader: 381 differential metabolites split endometriosis from healthy controls, with 42 altered across all three group comparisons.
N-lactoylvaline was the strongest single feature: AUC 0.815 (95% CI 0.698 to 0.925, FDR 3.38e-4); four other prespecified metabolites fell between 0.748 and 0.815.
Integration recovered what proteins lost: the four-protein panel managed 0.571 in cross-validation against the five-metabolite panel's 0.829, with FGF21, CCL23, CD8A and CDCP1 anchoring the network modules.
Figure 1. Serum proteomic differences across the three study groups. Panels A to C are volcano plots of differentially expressed Olink proteins for benign versus healthy, endometriosis versus benign, and endometriosis versus healthy, with significant proteins in red. Panel D is a Venn diagram of overlap among those comparisons. Panel E plots NPX distributions for twelve immune and inflammatory proteins, including CCL23, CD8A, CDCP1 and FGF-21. Adapted from Pan et al. (2026), Frontiers in Endocrinology.
Where the Inflammatory Protein Signal Ran Out
Twenty proteins shifted between endometriosis and healthy controls, including IL6, IL8, CCL3 and EN-RAGE. Against benign ovarian cysts, that number collapsed to two. Random Forest ranking put CD8A first by contribution score, yet no protein cleared an AUC of 0.67 in any pairwise comparison. A 92-plex inflammation panel describes a systemic inflammatory state shared by several gynecological conditions; it does not tell endometriosis apart from a benign cyst.
Energy Metabolism Carried the Discriminatory Weight
Fatty acid beta-oxidation, acylcarnitine transport and amino acid carbon flux were the pathways that moved. Joint KEGG enrichment across both layers converged on nicotinate and nicotinamide metabolism and the intestinal IgA immune network. Mantel testing linked the protein and metabolite blocks, though the coefficients stayed modest, and the authors say so plainly instead of overselling the network. Serum rather than plasma was the matrix, and coagulation shifts platelet-derived mediators, so several features need a targeted assay before anyone builds a panel around them.
From One Sample to a Combined Panel
Neither layer stands alone here: four proteins reached 0.571 for endometriosis versus healthy while the metabolite panel hit 0.829, and the pairing is what makes the immunometabolic story readable. Sample volume is the quiet constraint in designs like this, because a 96-plex immunoassay and an untargeted LC-MS run normally compete for the same aliquot and get processed on separate days. The Omni-MS workflow we run at Dalton reads proteins and metabolites from a single injection of one sample, holding both layers on the same batch structure and removing the aliquot-splitting tradeoff that multi-omics biomarker discovery programs usually absorb. When the endpoint is a combined panel rather than one marker, that is why we favor integrated proteomics and metabolomics from the outset.
Frequently Asked Questions
What is multi-omics analysis used for?
Multi-omics analysis measures two or more molecular layers, such as proteins and metabolites, in the same samples and models them together. It is used when one layer alone cannot explain a phenotype. The usual payoff in biomarker work is a panel that classifies better than any single analyte.
Can a multi-omics blood test diagnose endometriosis?
Not yet. This study reached a cross-validated AUC near 0.83 using five serum metabolites, promising but short of a diagnostic standard. With 44 cases at one center and no independent validation set, replication comes first.
How much sample does a proteomics and metabolomics study need?
Split-aliquot designs need enough serum or plasma for two separate preparations, which becomes limiting in pediatric, longitudinal or biobank cohorts. Single-injection workflows that read both layers from one aliquot roughly halve that requirement and remove a source of between-assay batch variation.
Conclusion
Endometriosis leaves a coordinated immune and metabolic footprint in serum, and metabolite features track it better than a fixed inflammation panel does. What isn't yet proven is that the four protein nodes hold up in an independent cohort. For teams building non-invasive diagnostics in gynecology, budget for both omics layers early, since the protein-only model here would have failed.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Pan, W., Lyu, G., Wang, Y., Chen, T., Liu, S., Fan, X., Liu, R., Lin, H., Li, G., Su, X., & Zhou, P. (2026). Multi-omics profiling identifies an immunometabolic signature associated with endometriosis. Frontiers in Endocrinology, 17, 1912463. https://doi.org/10.3389/fendo.2026.1912463
Note
This blog post summarizes findings from the above-cited research. Figures are adapted from the original publication. For full details, please refer to the source article.
By Seungjun Yeo, CEO at Dalton Bioanalytics. Specializing in multi-omics mass spectrometry for drug discovery and biomarker research.