Multi-Omics: Amniotic Fluid Markers Predict Persistent FGR
Two fetuses can both measure below the 10th percentile at 24 weeks and still diverge: one catches up before birth, the other keeps falling behind and is delivered early. Ultrasound cannot reliably tell them apart at that first diagnosis. Cao and colleagues tried a molecular route, running multi-omics profiling (untargeted LC-MS metabolomics plus data-independent acquisition, or DIA, proteomics) on residual amniotic fluid from clinical amniocentesis to stratify fetal growth restriction (FGR). Their best protein candidates were then retested by ELISA in an independent cohort of 69 pregnancies.
Key Takeaways
Amniotic fluid from routine amniocentesis carries protein signals that separate fetuses who recover growth from those who keep falling behind.
In this multi-omics study, the amniotic fluid proteome distinguished growth-restricted pregnancies from controls far better than the metabolome did.
Because amniocentesis is invasive, these markers suit pregnancies already having the procedure; routine screening would need a blood-based version.
Building a Multi-Omics Map of Amniotic Fluid
The discovery set was small: 13 controls plus 11 transient and 9 persistent FGR cases. Cases met the Delphi consensus definition first set out by Gordijn and colleagues (2016), which screens out fetuses that are simply constitutionally small. Metabolite profiling on a Q Exactive in both ionization modes annotated 1,873 metabolites, and DIA runs on an Orbitrap Q Exactive HF, processed in Spectronaut 19 at 1% FDR, quantified 4,132 proteins. Differential calls used a 1.2-fold cutoff. For 33 samples that is generous, and a few passengers have probably slipped onto the protein list.
Key Findings
The proteome did most of the diagnostic work: 188 proteins changed between FGR and controls, against only 8 metabolites.
Diagnostic accuracy shrank on validation: a PDGFA plus STAT5A panel scored an AUC (area under the ROC curve) of 1.000 in discovery but 0.780 when measured by ELISA in the validation cohort.
Three proteins flagged which cases would persist: a combined IREB2, HLA-C, and PLXNB2 model separated persistent from transient FGR with an AUC of 0.966.
WASHC2C as a severity gauge: this endosomal trafficking protein anchored the cross-omics network and rose stepwise from controls to transient to persistent FGR (p < 0.0001).
Figure 1. Study enrollment flowchart. Of 102 pregnant women referred for amniocentesis, 33 formed the discovery cohort (13 controls, 20 FGR) and 69 the validation cohort (27 controls, 42 FGR), with genetic abnormalities and intrauterine infection excluded. FGR cases were followed to delivery and labeled transient or persistent by whether birth weight reached the 10th percentile. Adapted from Cao et al. (2026), Annals of Medicine.
Why Proteins Outperformed Metabolites
Joint pathway analysis tied the two layers together through purine metabolism, and the protein changes alone pointed to the one-carbon pool by folate. The fall from an AUC of 1.000 to 0.780 is what usually happens when a model fitted on 33 samples meets new patients; a perfect discovery score says more about overfitting than about accuracy. In the discussion, the authors defend 0.780 because the fluid comes from procedures already indicated for other reasons. That argument holds.
Separating Catch-Up Growth From Progressive Restriction
Persistent cases differed from transient ones on 14 metabolites and 65 proteins, with the TCA cycle and pyrimidine metabolism disrupted. WASHC2C tracked with the TCA intermediate cis-aconitic acid, yet the authors call it an indicator of severity rather than a driver. With no reliable antibody on the market, they propose parallel reaction monitoring (PRM), a targeted mass spectrometry assay, for quantification. Validation came from one center with no cross-validation described, so treat the 0.966 prognostic AUC as optimistic.
One Aliquot, Two Omics Layers
Amniotic fluid is scarce, and this study split each sample between a FASP digest for proteomics and a separate solvent precipitation for metabolomics, adding prep-to-prep variance to exactly the cross-layer correlations the WASHC2C network rests on. Reading both layers from one aliquot in a single injection, as the Omni-MS workflow does in our integrated proteomics and metabolomics work at Dalton, removes that variable. It's the main reason we'd steer any multi-omics biomarker discovery project on residual biofluids toward a single-sample design from day one.
Frequently Asked Questions
What is multi-omics analysis used for in pregnancy research?
Multi-omics analysis measures two or more molecular layers, such as proteins and metabolites, in the same samples and links them statistically. In pregnancy research, it helps uncover amniotic fluid or maternal blood biomarkers for problems ultrasound misses.
Can amniotic fluid proteins predict persistent fetal growth restriction?
In one single-center study, an ELISA panel of IREB2, HLA-C, and PLXNB2 reached an AUC of 0.966 for separating persistent from transient cases. The panel still needs multicenter validation.
How much sample does a multi-omics biomarker study need?
Needs range from a few microliters of plasma to several hundred microliters of dilute fluids like amniotic fluid. Running proteomics and metabolomics from one aliquot, rather than separate preps, cuts that volume.
Conclusion
What's believable now is that the amniotic fluid proteome carries a growth-restriction signal that survives the move from mass spectrometry to ELISA, at least in one hospital. Whether the prognostic panel works at other centers, or whether WASHC2C does anything causal, isn't yet shown. Practically, banking residual amniocentesis fluid with birth-weight follow-up is what made this dataset predictive.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Cao, Y., Hao, W., Wang, Y., Chai, C., Li, Y., Liu, Y., Zhu, H., Lu, Y., Wang, D., Zhai, Y., Yan, Y., & Cao, Z. (2026). Integrated multi-omics profiling of amniotic fluid identifies predictive biomarkers for fetal growth restriction trajectories. Annals of Medicine, 58(1), 2718562. https://doi.org/10.1080/07853890.2026.2718562
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.