Abstract
BACKGROUND: Pancreatic cancer (PC) is a lethal disease, however current screening methods unable to achieve early diagnosis. Blood-based microRNAs (miRNAs) are promising molecular biomarkers for detecting PC. This meta-analysis summaries studies identifying serum- or plasma-based miRNAs dysregulated in PC patients compared to non-PC cases to evaluate their diagnostic accuracy for characterizing PC. METHODS: A systematically reviews and meta-analysis of published studies was conducted to compare the serum or plasma miRNAs expressions between PC patients and non-PC cases. Summary estimates for sensitivity, specificity, along with other measures of accuracy of miRNAs in the diagnosis of PC were pooled using the random-effects model. I and Q tests were used to assess the heterogeneity of included studies. The Spearman test was used to analyze the threshold effect. RESULTS: Twenty-seven eligible studies were identified after electronic search and literature selection. For single miRNA dysregulation, 32 miRNAs were found to be upregulated in PC patients, and 5 miRNAs were downregulated. Four studies identified a 2-miRNA panel, and 10 studies identified a panel consisting of 3 or more miRNAs which were used to detect PC patients. Additionally, 8 studies combined miRNA panels and carbohydrate antigen 19-9 (CA 19-9) to diagnose PC. The pooled sensitivities for these 4 groups were 0.77 to 0.85, and specificities were 0.70 to 0.87. The highest area under the curve (AUC), 0.9308, was identified using 2 miRNA panels with sensitivity and specificity of 0.79 (0.74-0.83) and 0.85 (0.81-0.89), respectively. There was great heterogeneity of these 4 miRNA groups. Results of Spearman test revealed that there existed a threshold effect on single miRNA group (r=-0.437, P=.001), and none of the other groups (P all>.05). CONCLUSIONS: Serum- or plasma-based miRNAs are capable of distinguishing PC from non-PC with relatively high sensitivity and specificity. In future, miRNAs may be used as promising diagnostic biomarkers for detection of PC.