Szafranska et al[27] performed the first analysis comparing miRNA expression in normal pancreas tissue, chronic pancreatitis (CP) tissue and PDAC tissue. On imaging, it can be challenging to distinguish CP from PDAC given the thick stroma and inflammation that may be found in both of these conditions. Furthermore, it is unclear if the aberrant expression of particular miRNAs is secondary to the desmoplastic reaction in CP and PDAC, and not related to tumorigenesis itself. They and others have found that miRNA-216 and miRNA-217 are significantly down-regulated in PDAC and miRNA-143, miR-145, miR-146a, miR-148a, miR-150, miR-155, miR-196a, miR-196b, miR-210, miR-222, miR-223, miR-31 are up-regulated in PDAC[24,27-29]. However, this study also demonstrated that dysregulation of miRNA-196a, miR-196b, miR-203, miR-210, miR-222, miR-217, and miR-375 were found only in PDAC, whereas miRNA-29c, miR-96, miR-143, miR-145, miR-148b, and miR-150 were abnormally expressed in both CP and PDAC. This may suggest that the latter are responsible for causing the desmoplastic reaction as opposed to tumorigenesis. MiR-1290 is elevated in early stage PDAC compared to normal controls[30]. Additionally, miR-135b has been shown to be an effective biomarker for distinguishing PDAC from CP with high sensitivity and specificity[31].

MiR-21, MiR-155, and miR-196 have been demonstrated by multiple groups to differentiate PDAC from non-cancerous lesions of the pancreas[20-24,27,32]. Special attention has been placed on the role of miR-21 in PDAC, as it has been implicated in tumorigenesis, TC invasion, the desmoplastic reaction, and metastasis of TC[33-36]. Further studies did not demonstrate that miR-21 expression in stromal cells correlated with tumor stage.

MiR-192 has also been found to be present in pancreatic TC, but is seldom seen in stromal cells and not found in adjacent normal pancreas tissue[37]. In this same study, miR-194 expression was detected in PDAC tissue, but not found in the surrounding normal pancreatic tissue. Unfortunately, despite these findings, no significant difference was found between the serum levels of miR-194 in patients with PDAC and healthy controls.

One proposed mechanism for PDAC development includes signaling between the molecular markers of the desmoplastic reaction and TCs[38-41]. Liffers et al[29] demonstrated that miR-148a is down-regulated in microdissected PDAC tissue and when over-expressed prevents tumor growth. This suggests that miR-148a may have a crucial role in the molecular signaling by which tumorigenesis occurs. While it is important to find biomarkers that are deregulated in PDAC, it is also important to understand which miRNAs are involved in these aberrant signaling pathways.

MiRNAs are known to have organ-specific expression in many human cancers[42,43]. Less than a decade ago, studies found that miRNA could reliably be detected in the serum in both animal models and humans[44,45], and since that time, there has been much research dedicated to identifying which miRNAs have differential expression and the implications of these findings in the detection, staging, treatment, and prognosis of cancers[46-50].

Attempts to use miRNA biomarkers in conjunction with CA19-9 have yielded mixed results. A study examining 847 different miRNAs in patients with PDAC found increased expression of miR-375 in PDAC as opposed to controls. MiR-375 did not improve detection nor predict prognosis in patients with PDAC when compared to CA19-9 alone[51]. Liu et al[52] found that using serum miR-16 and miR-196a in combination with CA19-9 increased detection of PDAC and Stage I lesions when compared to either modality alone, which suggests that miR-16 and miR-196a may be deregulated early in PDAC. These biomarkers were also up-regulated in studies performed on pancreas tissue, demonstrating that miR-16 and miR-196a can be used as peripheral biomarkers of PDAC. Gao, et al[53] also demonstrated that miR-16, when combined with CA19-9, served as a potential biomarker for detection of PDAC when compared to patients with CP.

One limitation of CA19-9 as a biomarker is that it is elevated in a large portion of patients with benign pancreatic diseases. Because of this limitation, studies have evaluated the miRNA expression of patients with PDAC compared to those with benign diseases such as CP or choledocolithiasis. They found that miR-10b, miR-155, and miR-106b were consistently elevated in the serum of patients with PDAC but not in those with benign pancreatic disease[54]. Liu et al[55] have demonstrated that up-regulation of miR-20a, miR-21, miR-24, miR-25, miR-99a, miR-185, and miR-191 can be used to distinguish PDAC from healthy controls and CP. Additionally, miR-135b has been shown to be an effective biomarker for distinguishing PDAC from CP with high sensitivity and specificity[31]. MiR-18a levels have also been shown to have increased expression in patients with PDAC and interestingly decrease after surgical resection suggesting that miR-18a levels may be a good marker to not only detect disease but also to monitor disease recurrence[29,56,57]. Zhang et al[37] also demonstrated that miR-194, miR-192, miR-602, miR-801, miR-212, miR-34a are up-regulated in PDAC, while miR-923, miR-139-3p, miR-513a-5p, miR-630, miR-30c-1, miR-887, miR-508-5p, and miR-139a-5p were down-regulated in PDAC specimens[37,58]. From these data, they demonstrated that miR-192 is neither present in the stromal cells of the pancreas nor the serum, but it is up-regulated in PDAC TCs and is involved in cell proliferation of PANC-1 TC lines in vitro[58]. Lin et al[59] performed microarray on 1711 serum miRNAs and found that 23 were down-regulated and 22 were up-regulated in the serum of PDAC patients when compared to normal controls. Of these, miR-492 and miR-663a were found to have decreased expression that was statistically significant in PDAC; however, only miR-663a was found to have a positive correlation with stage of disease[59]. Further studies are needed to determine which miRNAs will be clinically relevant.

Pancreatic juice sampling requires an invasive endoscopic procedure, but studying the miRNA concentrations of patients with PDAC, benign pancreatic lesions, and healthy controls can shed light on potential biomarkers for detecting disease as they are found in high concentration in cyst fluid. As EUS and endoscopic retrograde cholangiopancreatogram (ERCP) are two methods by which pancreatic masses are frequently detected and sampled, these specimens could be sent for miRNA analysis in order to determine the malignant potential of these lesions. Wang et al[60] performed microarray of 49 miRNAs on secretin-stimulated pancreatic juice of a group of patients with PDAC, CP, and normal controls. They demonstrated that miR-205, miR-210, miR-492, and miR-1427 are all significantly elevated in PDAC when compared to controls; however, this statistical significance does not exist when compared to patients with CP[60]. Additionally, by using ROC curves, they determined that combining these 4 miRNAs with serum CA19-9 the sensitivity and specificity of PDAC detection is 91% and 100% respectively, though this analysis was limited by a sample size of 6. Other groups have evaluated the pancreatic juice of patients with PDAC pre-operatively via ERCP and from post-operative specimens[61,62]. Sadakari et al[61] analyzed the expression of miR-155 and miR-21 in pancreatic juice sampled via ERCP and found that these miRNAs were significantly elevated when compared to patients with CP and healthy controls, though the levels did not correlate with pancreatic juice cytology[61]. Again these findings are consistent with those from pancreatic tissue and serum. Hong et al[62] evaluated 158 miRNAs in post-operative fine needle aspiration specimens and found by qRT-PCR that miR-21, miR-27a, miR-146a, and miR-186a were significantly over-expressed in PDAC tissue and miR-217, miR-20a, and miR-96 were significantly down-regulated in PDAC tissue when compared to normal controls[62]. These two studies have demonstrated the feasibility of detecting miRNA from pancreatic juice, thus indicating the potential for using pancreatic juice biomarkers to detect early lesions given the higher concentration of miRNA in this fluid sample.

Frozen stool specimens may serve as potential non-invasive biomarker samples for PDAC. Over-expressed miRNAs from gastrointestinal cancers are shed from the exfoliative cells of the gastrointestinal tract. Intraluminal release of pancreatic juice also allows for detection of miRNAs in the stool[63-69]. Previous studies have largely focused on genetic markers of tumorigenesis and not miRNA[70]. Yang et al[63] performed a feasibility study of using stool miRNAs as a potential screening tool for detection of PDAC. They evaluated expression of 5 miRNAs that had been previously shown to be over-expressed in PDAC and found that miRNA-21 and miR-155 were over-expressed and miR-216 was under-expressed in all PDAC stool specimens when compared to normal controls and CP patients. These findings are consistent with what has been found in whole pancreas, pancreatic cyst fluid, and serum specimens. Additionally, with ROC analysis they demonstrated that combining miR-21 and miR-155 in stool samples there was a sensitivity of 93.33% and specificity of 66.67%. When they combined all 3 miRNAs (miR-21, miR-155, and miR-216), the sensitivity and specificity were 83.33% each. Link et al[71] selected 7 miRNAs (miR-21, miR-210, miR-143, miR-155, miR-196a, miR-216a, miR-375) and determined that like Yang’s group miR-216a was found in lower concentrations in the stool of patients with PDAC. However, unlike Yang’s group, they found that miR-155 was down-regulated in this population and miR-21 was unchanged in the stool of controls compared to CP or PDAC[71].

Ren et al[72] also evaluated the expression of miR-21, miR-155, miR-181a, miR-196a, and miR-210 and found that miR-181b, miR-196a, and miR-210 were significantly over-expressed in PDAC patients when compared to controls, but only miR-181b and miR-210 were elevated in CP patients, though these elevations were not significant. Ren et al[72] established a positive correlation between miR-196a levels and tumor size, which had not been previously described in studies of the serum or stool. Overall, while studies of fecal miRNA have demonstrated feasibility, conflicting data have emerged on which miRNAs are differentially expressed in the stool of PDAC, benign pancreatic disease, and normal controls.

The field of salivanomics has been developing since blood molecules have been found in saliva[73]. As with stool miRNA, salivary miRNA may serve as a non-invasive biomarker for PDAC. Xie et al[74] is the only group to have evaluated salivary miRNA in PDAC. They conducted a microarray of 2006 miRNAs and noted that 10, including miR-4433-5p, miR-4665-3p, miR-940, miR-1273g-3p, miR-3676-5p, miR-3679-5p, miR-3940-5p, miR-4327, miR-4442, and miR-5100, were up-regulated or down-regulated in salivary samples. Of these, only miR-940 was significantly up-regulated and miR-3679-5p was significantly down-regulated in the PDAC specimens during the validation phase of the study. Until now, neither has been implicated in PDAC in the serum, stool, whole pancreas, or pancreatic juice. More studies are needed in this area.

Given the increased use of abdominal imaging, more pancreatic cystic lesions are being detected. There are guidelines in place to help guide management based on cystic characteristics that are consistent with malignancy[77]. The first study looking at miRNA expression levels in precursor lesions of the pancreas was performed by Habbe et al[23] who determined that miR-155 and miR-21 were over-expressed in the IPMN neoplastic epithelium, specifically those with carcinoma-in-situ[23,87]. MiR-155 was also significantly elevated in the pancreatic juice of these patients. While the levels of up-regulation of miRNA-21 and miR-155 correlated with the degree of cellular atypia found in the IPMN lesions, the study lacked long-term outcome data, highlighting the need for large and more longitudinal studies. Caponi et al[32] established a relationship between expression of miR-21 in invasive IPMNs and clinical outcome and observed that higher levels of miR-21 were correlated with worse overall and disease-free survival. Furthermore, they demonstrated that miR-155 and miR-21 had higher expression levels in invasive IPMN lesions when compared to non-invasive lesions, suggesting that these miRNAs may serve as early markers of malignant transformation[32,88]. MiR-101 has been shown to be down-regulated in invasive IPMNs when compared to non-invasive IPMNs and normal tissue. This deregulation of MiR-101 has not been described in PDAC samples, which may indicate that miR-101 plays a role in tumor invasion[32,89]. As with studies of miRNA in the serum and tissue of PDAC patients, miR-196a was found to be up-regulated in the pancreatic juice of intestinal-type IPMN[90]. In a recent study miR-100, miR-99b, miR-99a, miR-342-3p, miR-126, miR-130a were found to be down-regulated were up-regulated in high-risk vs low-risk IPMN lesions[91]. Furthermore, low miR-99b in IPMN fluid was associated with MD involvement, which is associated with a greater risk for transformation into a malignant neoplasm. Abue et al[88] found that miR-483-3p was up-regulated in PDAC cells and plasma when compared to IPMN lesions and may also serve as a useful biomarker in differentiating IPMN lesions with malignant potential from normal tissue and PDAC. The down-regulated miRNAs correlated with high-risk IPMNs and may be involved in cyst invasion and progression. Lee et al[92] found that miRNA expression varied amongst pancreatic cystic neoplasm. Specifically, miR-31-5p, miR-4830-5p, miR-99a-5p, and miR-375 were characteristic of serous cyst adenomas (SCA), whereas miR-10-5p, miR-202-3p, miR-210, and miR-375 differentiated MCN from SCA, IPMN, and PDAC[92]. It is unclear why this overlap in miR-375 occurs[92]. Henry et al[93] found that miR-92a, miR-99a, miR-100, miR-125b, miR-145, miR-212 and miR-483 were differentially expressed between benign and pre-malignant and malignant lesions of the pancreas and they suggested that a high amount of RNA present in the cystic fluid may suggest the presence of malignant transformation[93]. As previously described miR-21, miR-155, miR-196a have been implicated in both PDAC and IPMN and given these widely replicated results, studies aimed at detecting these biomarkers in serum, saliva, and stool could help to determine, in a non-invasive way, if they are increased in pre-malignant lesions.

Currently, PanIN lesions are found in the neighboring pancreatic tissue of patients with PDAC; however, there is no consistent way to detect the presence of these lesions[10,94-96]. Identifying a biomarker to detect PanIN lesions may be critical in early detection of PDAC. Slater et al[97] demonstrated that miRNA-196a and miR-196b were elevated in PDAC and PanIN-2/3 lesions in both animal models of PDAC and humans with PDAC. Ryu et al[98] demonstrated that miR-155 is up-regulated in PanIN-2 and PanIN-3 lesions when compared to neighboring healthy pancreatic tissue, but not in PanIN-1 lesions. Furthermore, miR-21 has been shown to be over-expressed in PanIN-2[98,99] and PanIN-3[98,99], but not PanIN-1 lesions suggesting that this is a marker for later disease. These are significant findings as miR-155 and miR-21 have been shown to be up-regulated in IPMN lesions and PDAC suggesting that they are early markers for cells with malignant potential. Yu et al[100] found that miR-196b was up-regulated in PanIN-3 lesions, which correlates with previous studies that have found that miR-196b is up-regulated in PDAC lesions[100,101]. Importantly, miR-21, miR-155, miR200a, miR-200b, miR-182, and miR-296-5p were deregulated as early as PanIN-1 lesions and remained deregulated until progression to PDAC, with the exception of miR-200c that normalized in PanIN-3 lesions. A recent publication describing miRNA expression in PanIN lesions found that miRNA-148a and miR-217 were down-regulated while miR-10b was up-regulated in PanIN-2 and PanIN-3 lesions[101]. While miR-21 has been shown repeatedly to be over-expressed in PDAC, there are conflicting studies on its deregulation in early PanIN lesions suggesting that it may represent a later and more aggressive dysregulation in the progression to PDAC. A non-invasive method to detect advanced PanIN lesions would represent a significant advance in the field.