Urinary miRNA-based assay shows high accuracy, potentially transforming pancreatic cancer screening.
Study: A noninvasive urinary microRNA-based assay for the detection of pancreatic cancer from early to late stages: a case control study. Image Credit: Krakenimages.com/Shutterstock.com
In a recent study published in the eClinicalMedicine, a group of researchers evaluated the effectiveness of a novel noninvasive urinary extracellular vesicle micro Ribonucleic Acid (miRNA)-based assay for detecting pancreatic cancer across all stages, focusing on improving early-stage detection.
Background
Pancreatic cancer is one of the deadliest cancers globally, with a 5-year survival rate of only 12%, largely due to late-stage diagnosis and the lack of effective screening tools. Early detection is crucial, as localized pancreatic cancer can significantly improve the 5-year survival rate to 44%.
Currently, Cancer antigen 19-9 (CA19-9) serves as a clinical biomarker, but its sensitivity is limited in the early stages.
Liquid biopsy, especially through blood-based detection of cancer-associated circulating tumor cells (CTCs) and circulating tumor DNA (ctDNA), shows promise but lacks early-stage sensitivity. Further research is essential to enhance early detection methods.
About the study
Conducted as a case-control study, the present research collected urine samples from individuals with pancreatic cancer and non-cancer controls from multiple locations in Japan between September 2019 and July 2023. Patients with confirmed pancreatic cancer diagnoses based on imaging or pathological examination were included, excluding those with multiple primary cancers.
Reflecting Japanese cancer demographics, the cohort predominantly featured participants with pancreatic ductal adenocarcinoma (PDAC) (cancer arising in pancreatic ducts), which comprises over 90% of pancreatic cancer cases.
The sample also captured both early- and late-stage pancreatic cancers defined according to the Union for International Cancer Control (UICC) Tumor, Node, Metastasis (TMN) classification system.
Urine samples were processed to isolate extracellular vesicles, followed by Ribonucleic acid (RNA) extraction for micro (mi)RNA analysis using a polymer-based precipitation method. RNA libraries were prepared for sequencing, and sequencing data was analyzed to identify miRNA profiles associated with pancreatic cancer.
Using a support vector classifier, a machine-learning model was trained and validated to detect pancreatic cancer based on these miRNA signatures, achieving significant accuracy.
Statistical analyses accounted for demographic factors, and differential expression analysis identified miRNAs relevant to pancreatic cancer biology. Further research is essential to refine and validate this approach across broader populations and stages of pancreatic cancer.
Study results
To enrich the concentration of miRNAs, urinary extracellular vesicles were precipitated and processed before RNA extraction. This approach significantly improved the miRNA mapping rate, rising from 5.22% ± 4.72% in cell-free urine (cfUrine) to 24.5% ± 15.6% in vesicle-derived samples, showing a considerable increase in detectable miRNA species. A higher unique miRNA read count, approximately eightfold, was observed with extracellular vesicle-derived RNA, enhancing the identification of miRNA types, from an average of 200.7 to 385.9 unique miRNAs.
The study cohort was characterized by demographic and clinical features, including age, body mass index (BMI), sex, and smoking history.
Pancreatic cancer participants were typically older than non-cancer controls, had a lower BMI, and included fewer current smokers and more past smokers. Clinical staging distribution showed that 33.3% of patients in the training set had early-stage cancer, compared to 16.7% in the test set.
Using miRNA profiles from training participants, differential expression analysis identified 45 differentially expressed miRNAs (DEMs) associated with pancreatic cancer. A binary prediction algorithm was developed with a support vector classifier using miRNA profiles, achieving high predictive accuracy with an area under the curve (AUC) of 0.972.
This model effectively detected early-stage pancreatic cancer, with a sensitivity reaching 97.0% for stage I/IIA cases. In test sets, performance remained accurate, with an AUC of 0.963 overall and 0.983 for the early stages. This miRNA-based model outperformed the CA19-9 biomarker, particularly in early detection, where CA19-9 showed limited sensitivity.
Functional enrichment analysis of the 45 DEMs linked to pancreatic cancer revealed significant associations with pathways in cancer, including Protein Kinase B (PI3K-Akt), Mitogen-Activated Protein Kinase (MAPK0, Janus Kinase-Signal Transducer and Activator of Transcription (Jak-STAT), and Wnt.
These pathways, identified using Kyoto Encyclopedia of Genes and Genomes (KEGG) and Gene Ontology annotations, are integral to pancreatic cancer pathogenesis and progression. Further, urinary miRNA profiles were compared to miRNA expressions from pancreatic cancer organoids.
In these organoids, several DEMs matched those in the urinary miRNA profile, reinforcing the potential of urinary miRNAs to reflect tumor characteristics.
The miRNA-based assay demonstrated promising sensitivity and specificity for pancreatic cancer detection across stages, including early stages where conventional markers often fail.
Comparison with the CA19-9 biomarker highlighted the superior early-detection capabilities of the miRNA assay, emphasizing its potential as a noninvasive diagnostic tool. Continued investigation into urinary extracellular vesicle-derived miRNAs could enhance early intervention options, potentially improving survival outcomes for pancreatic cancer patients.
Conclusions
To summarize, this study developed a noninvasive assay using urinary extracellular vesicle-derived miRNA to detect pancreatic cancer.
Low miRNA concentrations in urine pose challenges for small RNA sequencing, but this method enriched miRNA via vesicle precipitation, achieving a five-fold increase in mapping rates.
This miRNA-based approach demonstrated high performance, with an AUC of 0.972 in the training set and 0.963 in the test set, achieving sensitivities of 93.9% and 77.8%, respectively, and specificities of 91.7% and 95.7%.