Proteomics: Plasma Vesicle Proteins Flag Pancreatic Cancer
A jaundiced patient with a mass near the pancreatic head may have cancer or a benign biliary stricture, and routine labs such as CRP, ALP, and GGT climb in both. Telling them apart often waits on imaging and biopsy. A team in Johannesburg and Pretoria asked whether plasma proteomics could do some of that work earlier. Writing in the Journal of Proteome Research, they enriched extracellular vesicles (EVs, the membrane-bound particles cells shed into blood) from 178 people and found LRG1 and a stromal protein signature that separated pancreatic ductal adenocarcinoma (PDAC) from benign disease.
Key Takeaways
EV-enriched plasma proteomics quantified about 2,339 protein groups per sample from only 50 µL of plasma.
Plasma LRG1 rose with pancreatic cancer stage and separated PDAC from benign biliary disease with an AUC of 0.85.
Benign biliary disease sat between healthy controls and cancer, making it the comparator that matters for a pancreatic cancer blood test.
The cohort is single-center and unvalidated, so the candidate panel needs independent testing before any clinical use.
How the Plasma Proteomics Study Was Built
The cohort came from one Soweto hospital: 81 PDAC patients (60 resectable, 15 locally advanced, 6 metastatic), 62 with benign biliary pathologies (BBP), and 35 healthy controls, all of self-reported African ancestry. Each 50 µL EDTA plasma sample went through Mag-Net, a magnetic anion-exchange bead method that captures membrane particles and sheds much of the soluble background. Peptides ran on an Evosep One into a timsTOF HT using dia-PASEF, searched in Spectronaut 19. Age and sex adjustment was built into every comparison, which was necessary with a median age of 31 in controls and 62 in PDAC.
Key Findings
Depth from a small draw: an average of 2,339 protein groups was quantified per sample, and PCA placed BBP between healthy controls and PDAC.
Cancer versus healthy produced the widest gap. At a 2-fold threshold, 72 proteins rose and 71 fell in PDAC relative to controls, against 12 up and 12 down for BBP versus controls.
Stromal programs dominated the cancer signal: epithelial-mesenchymal transition was the top Hallmark set (adjusted p = 2.0 × 10-13), carried by SPP1, THBS2, VCAN, PTX3, and SDC1.
LRG1 tracked disease severity from healthy controls through BBP to resectable, locally advanced, and metastatic PDAC, with an AUC of 0.94 against healthy controls and 0.85 against BBP.
Figure 1. Mag-Net workflow for plasma proteome profiling. Fifty microliters of plasma are processed on a KingFisher Flex, where the first cycle (C1) captures extracellular vesicles on MagReSyn SAX beads, washes them, and reduces and alkylates the bound proteins. The second cycle (C2) aggregates proteins on the beads, washes with acetonitrile and ethanol, and digests with LysC and trypsin. Peptides are acquired with the Evosep One 60SPD method on a Bruker timsTOF HT and processed by directDIA in Spectronaut v19. Adapted from Buthelezi et al. (2026), Journal of Proteome Research.
Why Benign Biliary Disease Is the Control That Matters
The most useful design choice here is the BBP arm. Gallstones and strictures raise bilirubin, ALP, and GGT much as a pancreatic head tumor does. A marker that only separates cancer from healthy 31-year-olds has limited clinical value. So the drop in LRG1 performance, from an AUC of 0.94 against controls to 0.85 against BBP, is the honest figure to quote. LRG1 and SAA1 both respond to IL-6 driven inflammation, and some of their signal may reflect cholestasis rather than tumor biology.
A Stromal Signature With a Platelet Question
Five proteins (GPNMB, H4C1, PRSS2, SAA1, and CCDC47) increased across all three comparisons, the stepwise pattern expected of progression markers. Correlation mapping grouped LRG1 with C1QC, ORM1, and the myeloid phosphatase PTPRJ, which the authors read as vesicles from activated innate immune cells. The downregulated side deserves a harder look. Platelet and integrin proteins such as ITGA2B, ITGB3, GP1BB, and VWF fell in PDAC, and after a single 1,700 g spin, residual platelets are a plausible contributor. Geyer and colleagues (2019) showed how platelet and erythrocyte contamination can pass for disease signal in plasma studies, a caution that also applied to our earlier summary of a urine protein panel for kidney cancer.
In Practice: Vesicle Capture on Clinical Plasma
Bead-based vesicle capture is attractive because it automates on a plate handler and needs only 50 µL, but the same beads also pull down platelet microparticles along with tumor-derived material. In our own lab we run depleted and vesicle-enriched plasma proteomics for biomarker programs, and at Dalton a double-spun, platelet-poor collection protocol plus a contamination-marker check comes before any candidate is carried forward. For teams planning multi-omics biomarker discovery in pancreatic or biliary disease, that preanalytical discipline usually matters more than another few hundred protein groups.
Frequently Asked Questions
What is plasma proteomics used for in cancer biomarker discovery?
Plasma proteomics uses mass spectrometry to measure hundreds to thousands of circulating proteins from a small blood sample. In cancer research it surfaces candidate markers for detection and staging, which are then confirmed with targeted assays.
Can a blood test tell pancreatic cancer apart from benign bile duct disease?
Not yet in routine care, although this study points in that direction. Plasma LRG1 separated PDAC from benign biliary pathologies with an AUC of 0.85 in 178 South African participants. The result needs validation in larger, independent cohorts before it can guide diagnosis.
How much plasma do you need for an EV-enriched proteomics study?
The Mag-Net protocol in this study started from 50 µL of EDTA plasma per sample and quantified about 2,339 protein groups on average. Consistent handling matters as much as volume, especially double centrifugation to remove platelets and randomized plate layouts.
Conclusion
It's now reasonable to carry LRG1 and the stromal EMT proteins into targeted assays as PDAC candidates. What isn't shown is whether they add anything to CA19-9, which the study didn't benchmark, or whether they hold outside a single hospital. The sensible next step is a platelet-controlled validation cohort with benign obstruction as the main comparator.
Related Reading
See how we run these analyses in one lab: Dalton's multi-omics CRO services.
Citation
Buthelezi, S., Elebo, N., Naicker, P., Mokoena, R., Dubazana, S., Govender, I., Mamputha, S., Ellero, A., Stoychev, S., Mazibuko, J., Ojo, D., Candy, G., Devar, J., Cacciatore, S., Omoshoro-Jones, J., & Nweke, E. E. (2026). Proteomic profiling of extracellular vesicle-enriched plasma using Mag-Net for biomarker discovery in pancreatic ductal adenocarcinoma. Journal of Proteome Research, 25(8), 4120-4131. https://doi.org/10.1021/acs.jproteome.6c00172
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.