A 29-Gene cfRNA Blood Test Flags Pancreatic Cancer before Biops

 

Figure: Moore, et al., Nat Commun (2025)

A team at OHSU reports a plasma cell-free RNA (cfRNA) signature (29 genes) that distinguishes pancreatic ductal adenocarcinoma (PDAC) from benign and look-alike pancreatic conditions before patients undergo EUS-FNA. In an external validation cohort, the model reached AUC ≈ 0.90, and it outperformed CA19-9. The clever bit is a cfRNA-specific normalization that corrects for pre-analytical noise (like platelet activation) — long a headache in liquid-biopsy RNA. 

Why this matters

PDAC is often found late, when treatment options are limited. EUS-FNA is sensitive but invasive, expensive, and not universally accessible; CA19-9 is widely used but poor at early diagnosis and has low positive predictive value, so it’s not recommended for screening asymptomatic people. A robust, non-invasive triage test could help decide who actually needs EUS-FNA and reduce delays.  

What’s new in this paper

• Pre-diagnostic sampling. Blood was drawn before EUS in high-risk or symptomatic patients — the real-world scenario where a triage test would be used. Two cohorts: CEDAR (n=153) for discovery and BCC (n=95) for external validation. 

• cfRNA-specific normalization. The authors model each gene’s signal as a mix of intrinsic (true circulating) and extrinsic (handling-induced) components, learned with non-negative matrix factorization, then normalize them separately and recombine (“cf-normalization”). This tackles the notorious effect of platelet activation on plasma RNA profiles. 

• 29-gene PDAC panel. Differential expression in discovery led to a 29-gene set. A random forest classifier (with SMOTE to handle class imbalance) produced a PDAC score. Cross-validation AUCs were ~0.95 inside CEDAR and ~0.90 on BCC (e.g., 0.896 when comparing PDAC vs all others). 

• Beating CA19-9. The cfRNA classifier outperformed CA19-9 alone; adding CA19-9 to the model didn’t meaningfully improve accuracy. 

 

Biology highlight: Why cfRNA is different (and powerful)

cfDNA reports mutations and methylation; cfRNA reports tissue programs — gene expression states that can reflect tumor activity, immune context, and distant organ responses. Here, several up-regulated markers traced to liver-related genes (e.g., SERPINA1, CRP), and deconvolution suggested a larger liver contribution to the cfRNA pool in PDAC — even in early-stage patients without liver metastasis. This hints at an early pancreas-liver axis (pre-metastatic signaling or systemic response) surfacing directly in cfRNA. 

Big picture: cfRNA complements cfDNA by capturing tissue-of-origin and pathway signals, which multiple studies show can aid cancer detection and typing.   

The pre-analytics problem — and a credible fix

In plasma RNA work, how you collect, spin, freeze, and thaw blood can swamp biological signal. Residual platelets release RNAs and EVs ex vivo, distorting profiles — a major reason cfRNA signatures have struggled to travel. The study’s intrinsic/extrinsic modeling and cf-normalization target exactly this pain point and likely explain the strong external performance. 

 How could this be used clinically?

Think of it as an EUS triage step for people at elevated risk or with concerning symptoms:

1. Blood draw → cfRNA PDAC score

2. High score: prioritize EUS-FNA and surgical/oncology consults

3. Borderline score: pair with imaging, repeat test, or add orthogonal markers (cfDNA methylation, EV proteins)

4. Low score: consider deferring invasive work-up, tighten surveillance

It’s not a population-wide screening test (and the paper doesn’t claim that); it targets the indicated group where decision support is most actionable. 

 

Key numbers at a glance

• Cohorts: 153 discovery / 95 validation (pre-EUS blood draws). 

• Signature: 29 cfRNA genes; RF classifier with SMOTE. 

• External validation: AUC ≈ 0.90 (e.g., 0.896 vs all non-PDAC diagnoses). 

• Against CA19-9: cfRNA > CA19-9; combining both didn’t add much. 

 

Caveats (the honest bits)

• Small PDAC n (20 and 21 cases per cohort) and single-center recruitment could inflate performance; needs multi-site prospective validation. 

• Intended use is pre-diagnostic triage among high-risk/symptomatic patients — not general screening. 

• Assay practicality: Whole-transcriptome sequencing is slow/pricey; a targeted panel will be needed for clinical deployment. 

• CA19-9 context: Long-standing limitations remain; it’s not recommended for asymptomatic screening and can be elevated in benign disease. 

My take for RNA folks

• The normalization strategy is the star: it squarely addresses ex vivo confounders, a key translational barrier in cfRNA. If reused across diseases, it could standardize multi-site cfRNA studies. 

• The liver signal is intriguing — cfRNA may capture organ cross-talk (pre-metastatic niche biology) years before imaging does. Follow-up mechanistic work would be fascinating. 

• Expect multi-analyte panels next (cfRNA + cfDNA methylation + EV proteins) to push sensitivity/specificity further. 

 Glossary (30-second refresher)

• cfRNA: RNA fragments circulating in plasma inside EVs, protein complexes, or lipoproteins; reflects active gene expression from multiple tissues. 

• EUS-FNA: Endoscopic ultrasound–guided fine-needle aspiration; sensitive for PDAC but invasive. 

• CA19-9: A serum glycan antigen used in PDAC management, not reliable for early detection or screening. 

Original Study: 

Moore, T.W., Spiliotopoulos, E., Callahan, R.L. et al. Cell free RNA detection of pancreatic cancer in pre diagnostic high risk and symptomatic patients. Nat Commun 16, 7345 (2025). https://doi.org/10.1038/s41467-025-62685-y

Keywords: cell-free RNA, cfRNA, liquid biopsy, pancreatic cancer, pancreatic ductal adenocarcinoma, PDAC, early detection, RNA sequencing

====

#cfRNA, #CellFreeRNA, #LiquidBiopsy, #PancreaticCancer, #PDAC, #EarlyDetection, #RNAseq, #Transcriptomics, #MolecularDiagnostics, #CancerDiagnostics, #NonInvasive, #BloodTest, #Biomarkers, #29GeneSignature, #PrecisionOncology, #OncologyResearch, #EUSFNA, #CA199, #MachineLearning, #RandomForest, #SMOTE, #Normalization, #PlateletActivation, #ExtracellularVesicles, #LiverSignal, #TissueDeconvolution, #HumanProteinAtlas, #NatureCommunications, #OHSU, #HighRiskPatients

Keywords: cell-free RNA, cfRNA, liquid biopsy, pancreatic cancer, pancreatic ductal adenocarcinoma, PDAC, early detection, RNA sequencing, transcriptome, biomarker panel, 29-gene signature, non-invasive blood test, EUS-FNA triage, CA19-9 comparison, AUC 0.90, random forest classifier, SMOTE oversampling, pre-analytical variation, platelet activation, intrinsic–extrinsic normalization, liver-derived cfRNA, tissue deconvolution, Human Protein Atlas, validation cohort, CEDAR cohort, BCC cohort, Knight Cancer Institute, Nature Communications 2025, molecular diagnostics, precision oncology, pancreatitis vs PDAC, IPMN, high-risk symptomatic patients;  moleculardiagnostics.blogspot.com

A Glimmer of Hope: Personalized RNA Vaccines Teach the Immune System to Fight Pancreatic Cancer

 

 Author: KuriousK. | Subscribe: Don’t miss updates—follow this blog! 

For decades, a diagnosis of pancreatic cancer has been one of the most feared in medicine. With a grim survival rate that has barely budged in over 60 years, it remains one of the deadliest cancers. Standard treatments like surgery and chemotherapy can help, but for the vast majority of patients, the cancer relentlessly returns.

But what if we could teach our own bodies to hunt and destroy this formidable enemy?

A groundbreaking study published in Nature has offered a powerful glimpse into this very possibility. Researchers from Memorial Sloan Kettering Cancer Center, in collaboration with BioNTech and Genentech, have demonstrated that a personalized mRNA vaccine can awaken a patient's immune system, sending a powerful army of T-cells to attack pancreatic cancer cells and significantly delay the disease's return.

The Challenge: A Cancer That Hides in Plain Sight

Our immune system’s T-cells are expert soldiers, constantly patrolling our bodies for foreign invaders like viruses and bacteria. They can also recognize and eliminate cancer cells, but notoriously "cold" tumors like pancreatic cancer are masters of disguise. They build fortress-like environments and have very few unique markers, or "neoantigens," on their surface, allowing them to hide from the immune system.

The researchers behind this study decided to turn this weakness into a weapon. They hypothesized that even a few neoantigens—which are unique to each patient's tumor—could be enough to act as a "most wanted poster" for the immune system.

The Breakthrough: A Custom-Made Weapon for Every Patient

In a revolutionary Phase I clinical trial, scientists developed a truly personalized treatment protocol. Here’s how it worked:

1. Surgery: First, a patient's pancreatic tumor was surgically removed.
2. Genetic Analysis: The tumor was immediately sent to a lab where scientists sequenced its DNA to identify its unique mutational fingerprint—the neoantigens.
3. Custom Vaccine Creation: Using this genetic blueprint, a personalized mRNA vaccine (named autogene cevumeran) was created for each patient. This vaccine contained instructions to teach the immune system to recognize up to 20 of that specific patient's neoantigens.
4. A Three-Pronged Attack: Patients first received a dose of immunotherapy (atezolizumab) to "take the brakes off" their immune system. Then, they received their personalized vaccine to direct the T-cells to their target. Finally, they underwent a standard course of chemotherapy.

The Stunning Results

The results were remarkable. The complex, time-sensitive process of creating and delivering a personalized vaccine was successful and safe. But more importantly, it was effective.

In 8 out of 16 patients, the vaccine triggered a massive and powerful T-cell response. These newly activated T-cells specifically targeted the neoantigens from the patient's own cancer.

The clinical impact was even more striking. The study measured recurrence-free survival—the length of time patients lived before their cancer returned.

• For the 8 patients who did not respond to the vaccine, the cancer returned after a median of 13.4 months.
• For the 8 patients who did respond, their median recurrence-free survival had not yet been reached at the 18-month follow-up mark.

This indicates a dramatic and meaningful delay in cancer recurrence for those whose immune systems were successfully activated by the vaccine.

In one incredible case, the researchers witnessed the vaccine in action. A patient developed a small lesion in their liver, suspected to be a metastasis. A biopsy revealed it was not a full-blown tumor, but a dense cluster of the very same T-cells that the vaccine had trained. On subsequent scans, the lesion had vanished, suggesting the vaccine-activated T-cells had traveled to the site and eliminated the microscopic spread of cancer.

What's Next?

This was an early-stage trial with a small number of patients, and it's not a cure. However, its findings are incredibly promising. It provides powerful evidence that personalized mRNA vaccines can turn "cold" tumors "hot," making them vulnerable to an immune attack.

The success of this trial has paved the way for a larger, global randomized trial to confirm these findings. For a disease that has seen so little progress for so long, this research represents a beacon of hope and a monumental step forward in the fight against pancreatic cancer. It signals that the era of personalized immunotherapy is not just coming—for some, it has already begun.

 

Reference:  Rojas, L.A., Sethna, Z., Soares, K.C. et al. Personalized RNA neoantigen vaccines stimulate T cells in pancreatic cancer. Nature 618, 144–150 (2023). https://doi.org/10.1038/s41586-023-06063-y 

Author: KuriousK. | Subscribe: Don’t miss updates—follow this blog!