3.1. The characteristics and quality of the included studies
After the duplicate removal, title and abstract assessment, and full-text evaluation, we finally included 46 studies involving 4,326 PC patients and 4,277 non-PC controls. The characteristics of included studies were listed in Table 1. Among these original studies, 34 studies were conducted in Asia[13-46], 6 in Europe[47-52], 4 in North America[53-56], 1 in Africa[57], and 1 in South America[58]. The publication year were 2019 (n=2), 2018 (n=5), 2017 (n=4), 2016 (n=6), 2015 (n=7), 2014 (n=11), 2013 (n=4), 2012 (n=2), 2011 (n=4), and 2009 (n=1). The flow diagram of literature search and study selection was detailed in Figure 1 (A).
We found that there was a high risk of bias in the domain of "Patient Selection" after the process of quality assessment using the QUADAS-2 tool. According to the statement of the QUADAS-2 group, an ideal diagnostic study should enroll a proportion of suspected patients ("difficult-to-diagnose patients") to reduce the risk of bias[12]. But all our included studies included patients with a definitive diagnosis, which resulted in a high risk of bias in this domain. Besides, there was a large proportion of the unclear risk of bias in the domain of "Index Test", because the researchers of these included studies did not describe how they determined the threshold. The risk of bias was low in the domain of "Reference test" and "Flow and Timing". All domains exhibited low concerns of applicability. The results of the quality assessment were shown in Figure 1 (B-C).
3.2. The diagnostic performance of circulating miRNAs
Circulating single miRNA, which means that only one kind of miRNA was used for diagnosis, distinguished PC patients from non-PC controls with a SEN of 0.78 (0.76-0.81), a SPE of 0.78 (0.75-0.80), and the PLR, NLR, DOR and AUC were 3.55 (3.13-4.02), 0.28 (0.25-0.31), 12.78 (10.19-16.03) and 0.85 (0.82-0.88), respectively. The circulating miRNAs panel, which means multiple miRNAs were applied for diagnosis, discriminated cases with PC from case of non-PC with SEN of 0.79 (0.76-0.82), SPE of 0.75 (0.72-0.78), PLR of 3.16 (2.74-3.65), NLR of 0.28 (0.23-0.33), DOR of 11.40 (8.55-15.20), and AUC of 0.84 (0.80-0.87). There is no significant difference in the diagnostic efficacy between single miRNAs and miRNA panels. Overall, the SEN, SPE, PLR, NLR, DOR and AUC of circulating miRNAs (including single miRNAs and miRNA panel) in differentiating patients with PC from non-PC controls were 0.79 (0.77-0.81), 0.77 (0.75-0.79), 3.38 (3.08-3.72), 0.28 (0.25-0.31), 12.22 (10.23-14.60) and 0.85 (0.81-0.87), respectively. The results were shown in Table 2 and Figure 2 (A-C).
In addition, we also summarized the SEN, SPE, PLR, NLR, DOR and AUC of miRNAs in distinguishing PC patients from healthy controls (HC) or patients with chronic pancreatitis (CP). The data was listed in Table 2. In general, the diagnostic accuracy of miRNAs in discriminating PC from HC was higher than that in discriminating PC from CP.
A total of 58 different single miRNAs and 23 miRNA panels were involved in the 46 included studies. For the single miRNAs and miRNA panels being studied in one data set, we extracted the diagnostic SEN, SPE, PLR, NLR and DOR from the original literature. For those being studied in more than 2 data sets, we performed a meta-analysis and obtained pooled diagnostic SEN, SPE, PLR, NLR and DOR. The results were listed in Table S1 and Table S2. Among these single miRNAs, miR-122, 212, 22-3p, 483-3p, 642b-3p and 885-5p yielded a high SEN of more than 90%, while the SPE values of miR-25, 223, 17-5p, 223-3p, 30c and 409-3p were greater than 90%. The SEN and SPE of miR-451, 106b, 10b, 181a, 196b, 20a and let-7a were both greater than 90%. For miRNA panels, the SEN of the combination of let-7b-5p, miR-192-5p, 19a-3p, 19b-3p, 223-3p and 25-3p exceeded 90% while the SPE of the combination of miR-1246, 4464, 3976 and 4306 was over 90%. The combination of miR-196a and 196b, and the combination of miR-451 and 409-3p, as well as the combination of 885-5p, 22-3p and 642b-3p, all exhibited high diagnostic accuracy, with SEN and SPE values both greater than 90%.
3.3. Circulating miRNAs for the diagnosis of early-stage PC
Early-stage PC was defined with TNM stage of 0-IIa. For this group of patients, the SEN, SPE, PLR, NLR, DOR and AUC of circulating miRNAs were 0.79 (0.76-0.82), 0.74 (0.68-0.79), 2.60 (2.19-3.10), 0.35 (0.30-0.41), 8.14 (5.85-11.33) and 0.81 (0.77-0.84), respectively. MiR-196b and the combination of miR-196a and 196b exhibited high diagnostic accuracy with both SEN and SPE greater than 90%. The results were listed in Figure 2 (D) and Table 3.
3.4. The diagnostic performance of conventional biomarkers
Besides circulating miRNAs, some researchers have also evaluated the diagnostic efficacy of conventional biomarkers, such as CA19-9, CEA, and CA242. Among these conventional biomarkers, CA19-9 was the most frequently studied. The SEN, SPE, PLR, NLR, DOR and AUC of CA19-9 in discriminating PC from non-PC were 0.78 (0.75-0.80), 0.90 (0.85-0.94), 7.90 (5.14-12.13), 0.25 (0.22-0.28), 31.89 (18.96-53.62), and 0.85 (0.82-0.88), respectively. The SEN of CEA and CA242 was similar to that of CA19-9, but the SPE was significantly lower than that of CA19-9. CEA distinguished PC from non-PC with SEN and SPE of 0.79 (0.39-0.96) and 0.32 (0.08-0.72), respectively. The PLR, NLR, DOR and AUC of CEA were 1.17 (0.82-1.65), 0.65 (0.26-1.60), 1.80 (0.55-5.88) and 0.59 (0.54-0.63), respectively. The SEN, SPE, PLR, NLR, DOR and AUC of CA242 were 0.79 (0.52-0.93), 0.46 (0.21-0.74), 1.47 (0.95-2.27), 0.45 (0.21-0.97), 3.25 (1.14-9.32) and 0.68 (0.63-0.71), respectively. The results were listed in Figure 2 (E) and Table 2.
3.5. The diagnostic performance of circulating miRNAs combined with CA19-9
The combination of circulating miRNAs and CA19-9 for the diagnosis of PC exhibited a significantly higher diagnostic accuracy than that of using of circulating miRNAs or CA19-9 alone. The SEN, SPE, PLR, NLR, DOR, AUC of miRNAs combined CA19-9 in differentiating PC from non-PC were 0.84 (0.80-0.87), 0.84 (0.80-0.87), 9.77 (7.65-12.47), 0.17 (0.14-0.22), 56.01 (37.70-83.20) and 0.94 (0.92-0.96), respectively. The results were listed in Figure 2 (F-H) and Table 2.
The combination of miR-196, 200 and CA19-9 exhibited a high SEN of more than 90%. Combinations with SPE greater than 90% included the combination of miR-1290 and CA19-9, the combination of miR-16 and CA19-9, the combination of miR-16, 196a and CA19-9, the combination of miR-145, 150, 223, 636 and CA19-9, and the combination of miR-26b, 34a, 122, 126, 145, 150, 223, 505, 636, 885-5p and CA19-9. There were 4 combinations with SEN and SPE both exceeding 90%, which were the combination of miR-210 and CA19-9, the combination of miR-25 and CA19-9, the combination of miR-196a, 210 and CA19-9, and the combination of miR-181a, 181b, 210 and CA19-9. The results were listed in Table S3.
3.6. Subgroup analysis and threshold effect analysis
Since significant heterogeneity appeared in our meta-analysis (I2>50%), random-effects model was applied for pooled analysis. Meanwhile, subgroup analysis of five potential sources of heterogeneity including region, conference test, miRNAs profiling, non-PC control population and specimen, was conducted to identify the source of heterogeneity. However, the results suggested that the I2 value of most subgroups was still greater than 50%, indicating that these factors were not associated with heterogeneity. The results were listed in Table S4.
The value of the Spearman correlation coefficient was -0.276 (P=0.000) in the threshold effect analysis, suggesting the existence of a threshold effect, which might be the main source of heterogeneity in the present meta-analysis.
3.7. Sensitivity analysis and publication bias
A sensitivity analysis was performed to validate the reliability of our results. Any removal of the original studies did not have a significant impact on the results and corresponding 95% CI, suggesting that the results were stable. Deeks' funnel plots provided no evidence for publication bias (P>0.05).