Lipidomics: EV Lipids Catch Ovarian Tumors CA-125 Misses
Over half of early-stage ovarian cancers present with a normal CA-125, and the marker is weakest where surgeons need it most: separating borderline and low-grade tumors from benign adnexal masses. Rani and colleagues (Journal of Extracellular Vesicles, 2026) tested whether lipidomics could fill that gap. They profiled total plasma and matched small extracellular vesicles (sEVs) from 96 women, including 30 with invasive cancer spread across four histotypes.
Key Takeaways
Plasma and vesicle lipidomics added diagnostic information for ovarian masses where CA-125 is weakest, especially borderline, low-grade serous and mucinous tumors.
Whole-plasma lipids and vesicle lipids performed best in different diagnostic comparisons, which argues for profiling both compartments from the same blood draw.
The ovarian cancer lipid panels remain exploratory; several subgroups held only five or six patients, and no external cohort was tested.
Inside the Plasma and Vesicle Lipidomics Workflow
The team first benchmarked four liquid-liquid extraction (LLE) chemistries against solid-phase extraction (SPE), then isolated vesicles from 100 µL of plasma by size-exclusion chromatography (SEC). Extracts were directly infused into a SCIEX 5600 qTOF, so this is effectively shotgun lipidomics with sum-composition annotations, and the spiked internal standards were not used for quantitation. For every comparison, 14 classifiers were trained on 70% of samples and tested on the rest.
Key Findings
SPE beat liquid-liquid extraction: Replicates clustered more tightly, and vesicle-associated lipid classes concentrated in SEC fractions 1 to 3.
CA-125 held up only for advanced disease: It scored an AUROC of 1.00 for HGSOC versus controls, but 0.65 for borderline tumors and 0.61 for benign lesions.
Combined models narrowed the borderline gap: Lipids plus CA-125 reached accuracy up to 0.89 for borderline and low-grade serous tumors versus benign masses, against 0.67 for CA-125 alone.
When CA-125 reads normal, compartment matters: Below 35 U/mL, sEV lipids separated borderline from benign masses with an AUROC of 0.82, compared with 0.48 for plasma lipids.
Figure 1. Lipid extraction method development for plasma and small extracellular vesicles. Panel A outlines the SPE and liquid-liquid protocols, and Panel B clusters lipid classes by method across triplicate preparations. Panels C through E track lipid class composition, particle and protein concentration, and unsaturated lipid content across 13 SEC fractions. Panel F shows immunoblots for vesicle and non-vesicle markers, while Panel G presents cryo-TEM images of fractions 1 to 3. Adapted from Rani et al. (2026), Journal of Extracellular Vesicles.
Plasma Versus Vesicle Lipid Profiling
Each compartment won different contests. Plasma lipids dominated case-control comparisons (AUROC 0.96 for controls versus benign lesions), while vesicle lipids did better at differential diagnosis, scoring 0.74 for benign versus borderline masses. Buas and colleagues (2021) had shown that plasma diacylglycerols sharpen CA-125 in adnexal masses; the advance here is a matched plasma and vesicle comparison by histotype, echoing recent plasma vesicle proteomics work in pancreatic cancer.
Reading the Synthetic Cohorts with Care
To estimate performance at scale, the authors simulated 10,000- and 100,000-woman cohorts from a Gaussian copula model, and sEV lipids alone reached 0.94 accuracy for controls versus HGSOC in the screening scenario. Treat that cautiously. Synthetic patients inherit every quirk of 96 real ones, and each reported model was the best of 14 algorithms chosen on the test split. With five LGSOC and six borderline cases, a 30% hold-out holds one or two patients per class, so perfect AUROCs mean less than they seem.
In Practice: Extraction and Quantitation Choices
Extraction chemistry sets the ceiling for any lipid biomarker program, and this SPE-versus-LLE benchmark makes that ceiling visible. When we run plasma and EV lipidomics at Dalton, class-matched internal standards go in before extraction and drive quantitation, because relative peak areas rarely survive a move to a second site. We'd also add chromatography before any panel heads toward multi-omics biomarker discovery and validation, since a sum-composition call like PA 36:2 can hide isomers that don't behave alike.
Frequently Asked Questions
What is lipidomics used for in cancer biomarker research?
Lipidomics measures hundreds of lipid species at once, usually by mass spectrometry, to find patterns that shift with disease. In oncology it feeds diagnostic panels, tumor subtyping and treatment monitoring, because tumors rewire lipid metabolism to grow.
Can plasma lipidomics detect ovarian cancer when CA-125 is normal?
In this 96-patient study, lipid panels kept useful accuracy below the 35 U/mL cutoff, with plasma lipids reaching an AUROC of 0.79 for controls versus invasive cancer. Independent cohorts are still needed before clinical use.
How much plasma does an EV lipidomics study need, and what drives cost?
This study used 100 µL of plasma for vesicle isolation plus 10 µL for total lipids. Cost rises mainly with the vesicle isolation step and with absolute quantitation, while 96-well SPE plates keep large cohorts manageable.
Conclusion
It's now reasonable to believe that vesicle and plasma lipids carry signal CA-125 misses in borderline and low-grade disease. Whether any specific panel replicates isn't yet established. For ovarian biomarker programs, the practical move is banking enough plasma per patient to profile both compartments.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Rani, S., Guanzon, D., Lai, A., Scholz-Romero, K., Perrin, L., Lourie, R., Joshi, V., McCart Reed, A. E., Ferguson, K., Handberg, A., Möller, A., & Salomon, C. (2026). Small extracellular vesicle and total lipid profiles provide a diagnostic advantage over CA-125 across ovarian cancer subtypes. Journal of Extracellular Vesicles, 15(9), e70348. https://doi.org/10.1002/jev2.70348
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.