First, we selected 4 patients with IPMNs (2 benign IPMNs and 2 IPMCs) and 4 controls without any type of tumor for microarray analysis to identify differences in the gene expression of serum EV-miRNAs between these groups. To validate the microarray results, we enrolled 38 consecutive patients who were diagnosed with IPMNs by imaging modalities at Fukushima Medical University Hospital and 21 patients without any neoplastic lesions as controls between June 2015 and November 2019. The diagnostic criteria were as follows: i) Dilation of the MD and/or a cystic dilation of the BD, and ii) secretion of mucin from the major or minor papilla identified by endoscopic retrograde cholangiopancreatography or duodenoscopy.

Clinical data including age, sex, background diseases in controls, presence of symptoms (jaundice, body weight loss, etc.), history of diabetes mellitus, pancreatitis, smoking, family history of pancreatic tumor, and serum tumor markers, including carcinoembryonic antigen (CEA) and CA 19-9, were retrieved from electronic medical records. Patients with co-existing tumors or active infection (i.e., cholangitis or cholecystitis) were excluded from the analysis. All recruited patients were evaluated with CT or MRI, and the location of the lesion, maximum diameter of cyst and main pancreatic duct (MPD), and the presence of mural nodules were determined. Medical management of the disease was also evaluated. Classification of BD-IPMN and others was performed according to the 2017 International Consensus Guideline (1).

The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the institutional review board of Fukushima Medical University (IRB #2387). All participants provided written informed consent.

To obtain the serum samples, 8 ml of blood was collected and incubated at room temperature for at least 60 min to allow clotting. Samples were then centrifuged at 1,000 × g for 10 min. The serum was collected and stored in aliquots at −80°C.

Serum miRNA was extracted from serum using the exoRNeasy Serum/Plasma Midi kit (Qiagen) according to the manufacturer's protocol. RNA quantity and quality were determined using an Agilent Bioanalyzer (Agilent Technologies), as recommended. miRNA microarrays were manufactured by Agilent Technologies. Briefly, RNA was dephosphorylated using calf intestinal alkaline phosphatase (CIP) master mix incubated at 37°C for 30 min. Dephosphorylated RNA was denatured with DMSO incubated at 100°C for 5 min and then immediately transferred to ice for 2 min. These products were mixed with a ligation master mix for T4 RNA ligase and Cy3-pCp (Cyanine 3-Cytidine biphosphate) and incubated at 16°C for 2 h. Labeled RNA was dried using a vacuum concentrator at 55°C for 1.5 h. Cy3-pCp-labeled RNA was hybridized on an Agilent SurePrint G3 Human miRNA 8×60K Rel.21 (design ID: 070156) array at 55°C for 20 h. After washing, microarrays were scanned using an Agilent SureScan Microarray Scanner System (G4900DA). The intensity values of each scanned feature were quantified using Agilent Feature Extraction software version 12.1.1.1, which performs background subtractions. We only used features that were flagged as no errors (detected flags) and excluded features that were not positive, not significant, not uniform, not above background, saturated, and population outliers (not detected flags). These expression analyses were performed with Agilent GeneSpring GX version 14.9.1.

Digital PCR and quantification of the absolute levels of serum miRNAs were performed using the Quant-Studio 3D Digital PCR system (Thermo Fisher Scientific, Inc.). Data were analyzed using QuantStudio 3D Analysis Suite Cloud Software (Thermo Fisher Scientificc, Inc.). The digital PCR mixture contained 5.0 µl of the RT product, 1.0 µl of nuclease-free H₂O, 7.50 µl of the QuantStudio™ 3D Digital PCR Master Mix, and 0.75 µl of the TaqMan MiRNA Assay-1 (20X) for let-7d. Samples were individually loaded onto the QuantStudio 3D digital PCR 20K chip kit v2 using the QuantStudio 3D digital PCR Chip Loader. Digital PCR was performed in a Proflex 2X flat block thermal cycler (Applied Biosystems) using standard conditions: 96°C for 10 min followed by 39 cycles of 60°C for 2 min, 98°C for 30 sec, and 60°C for 2 min. Chips were read on the QuantStudio 3D digital PCR instrument, and the number of FAM-positive and FAM-negative (empty) wells was quantified (13,14).

Target gene prediction was performed using DIANA-miRPath software (15). Specifically, we investigated whether tumor suppressor genes were targeted by these 3 miRNAs because aberrant tumor suppressor gene methylation was observed in malignant IPMN (16).

To predict cell signaling pathways that were potentially influenced by the EV-miRNAs, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed via the Database for Annotation, Visualization and Integrated Discovery (DAVID 6.8; http://david.abcc.ncifcrf.gov/) (17,18). KEGG pathways with a P-value <0.05 were considered significantly enriched.

Continuous variables (i.e., age, cyst size, serum CEA and CA 19-9 levels and GS) are reported as the median and range and were compared using Mann-Whitney analysis. Categorical variables (i.e., sex and location of disease) were analyzed using Fisher's exact test. The diagnostic yield of CEA and CA 19-9 levels and EV-miRNAs in distinguishing whole IPMN (benign IPMN and IPMC) from the control and IPMC from benign IPMN was assessed using the area under the receiver operating characteristic (ROC) curve. Statistical analyses were performed using SPSS version 26.0 for Windows (IBM Corp.), and figures were generated with Prism 7 (GraphPad Software, Inc.). P<0.05 was considered to indicate statistical significance.

We first identified miRNAs showing altered expression in 2 IPMNs,2 IPMCs and 4 controls (Table I). Among the 2588 mature miRNAs evaluated in the microarray, we found 3 EV-miRNAs [hsa-miR-6132 (EV-miR-6132), hsa-miR-22-3p (EV-miR-22) and hsa-miR-4539 (EV-miR-4539)] that could discriminate IPMN, IPMC and control (Table II). As shown in Fig. 1, target prediction using DIANA-miRPath software revealed that these 3 miRNAs could target various genes. Regarding tumor suppressor genes, we found that miR-22-3p could target TP53INP1 and mir-6132 could target 3 genes (tuberous sclerosis complex 2: TSC2, tumor protein p53-inducible protein 11: TP53I11 and protein phosphatase 2 regulatory subunit B'Delta: PPP2R5D).

In addition, pathway enrichment analyses found that these 3 miRNAs could influence the NOD-like receptor signaling pathway (6 target genes involved), glycosphingolipid biosynthesis-lacto and neolacto series-(1 target gene involved) and fat digestion and absorption (5 target genes involved) (Table III).

To validate whether the 3 miRNAs could be diagnostic markers for IPMN, we conducted a validation study using digital PCR. The clinical characteristics of patients in the IPMN (n=38) and control (n=21) groups are presented in Table IV. Briefly, there were no significant differences in age, sex, absence of symptoms, history of diabetes, pancreatitis, smoking, or family history of pancreatic cancer between the IPMN group and the control group. With regard to management, a total of 9 patients underwent surgery among 38 IPMN patients. We confirmed the pathological diagnosis of the resected cases according to the latest classification (1): Low-grade dysplasia (n=4), high-grade dysplasia (n=1), and invasive carcinoma (n=4). Twenty-eight patients did not require surgery and were continuously observed. One patient was treated with chemotherapy.

We also divided IPMNs into 2 groups, namely, IPMC (n=11) and benign IPMN without any suspicious findings of malignancy (n=27), and compared clinical characteristics as shown in Table V. Clinical symptoms were observed more frequently in IPMC than IPMN (36.3 vs. 0.0%, P=0.0001). Additionally, the median diameter of the MPD was significantly larger in IPMC than in IPMN (11.3 vs. 4.4 mm, P=0.0003).

Fig. 2 shows the differences in each biomarker between the control and IPMN (both benign IPMN or IPMC) groups (Fig. 2) and between benign IPMN and IPMC groups (Fig. 3). With regard to discriminating IPMNs from control, only EV-miR-4539 showed a significant difference (P=0.004) (Fig. 2A). ROC analysis showed that 5 markers could discriminate patients with IPMN and from control patients, with areas under the curve (AUCs) of 0.72 for EV-miR-4539 (95% confidence interval [CI]: 0.59–0.85), 0.55 for CEA (95% CI, 0.39–0.71), 0.55 for CA19-9 (95% CI, 0.41–0.71), 0.59 for EV-miR-22 (95% CI, 0.42–0.76) and 0.64 for EV-miR-6132 (95% CI, 0.47–0.81). As shown in Fig. 2B, EV-miR-4539 had the highest diagnostic yield compared with other markers at the cutoff value of 3.2 copies/µl, and the sensitivity and specificity were 60.5 and 95.2%, respectively (Fig. 2B).

In the comparison between benign IPMN and IPMC, CA19-9 and EV-miR-6132 showed significant differences (P=0.01 and 0.007, respectively) (Fig. 3A). ROC analysis showed that the markers could discriminate patients with IPMC from patients with benign IPMN, with an AUC of 0.77 for EV-miR-6132 (95% CI, 0.61–0.93), 0.74 for CA19-9 (95% CI, 0.52–0.97), 0.61 for EV-miR-22 (95% CI, 0.41–0.81), 0.58 for EV-miR-4539 (95% CI, 0.40–0.77) and 0.55 for CEA (95% CI, 0.34–0.77). EV-miR-6132 showed the highest diagnostic yield compared with other markers at the cutoff value of 1.4 copies/µl and the sensitivity and specificity were 88.3 and 65.4%, respectively (Fig. 3B).

With regard to the serum levels of biomarkers and the existence of high-risk indicators, the serum level of CA19-9 was higher in patients with obstructive jaundice (Fig. 4). The serum level of EV-miR-6132 was higher in patients with high-risk indicators (mural nodules and MPD diameter more than 10 mm) than in patients without them.