Abstract
Background: Pancreatic cancer (PC) is a highly malignant tumor with no effective early diagnostic biomarkers. This study was performed to screen and identify serum microRNAs (miRNAs) as noninvasive biomarkers for PC diagnosis. Methods: Two upregulated miRNAs were selected by integrated analysis of three independent GEO datasets. Then, the expressions of two miRNAs in serum were determined by quantitative reverse-transcription PCR among 120 PC patients, 40 benign disease controls and 40 healthy controls. The correlation between serum miRNAs and clinical characteristics was analyzed. The diagnostic utility of miRNAs was compared to CA19-9 using receiver operating characteristic curve analysis. Results: We discovered miR-1290 and miR-1246 were upregulated in PC patients through GEO datasets analysis. Serum miR-1290 and miR-1246 expression levels were elevated in PC patients compared to all controls and dramatically decreased after tumor resection (all P<0.001). The area under the curve (AUC) for miR-1290 was larger than miR-1246 and CA19-9 (miR-1290: 0.91; miR-1246: 0.81; CA19-9: 0.82). The combined diagnosis of individual or both miRNAs with CA19-9 was more effective for discriminating PC from all controls than the single CA19-9 assay (miR-1290+CA19-9: 0.96, miR-1246+CA19-9: 0.93, miR-1290+miR-1246+CA19-9: 0.97). The abundance of serum miR-1290 and miR-1246 was associated with tumor stage and size respectively and logistic modeling proved that both of them were independent risk factors for PC. Conclusion: Serum miR-1290 and miR-1246 might be promising biomarkers for PC diagnosis and the combined detection of CA19-9, together with miR-1290 or miR-1246, could improve the diagnostic accuracy of PC.