Abstract

BIOGENESIS OF MIRNAs AND MECHANISM OF REGULATION OF GENE EXPRESSION

MicroRNAs are transcribed by RNA polymerase II/III in the nucleus, and the primary miRNAs are then processed by Drosha into hair-loop pre-miRNAs before export to the cytoplasm by exportin ⁵. In the cytoplasm, pre-miRNA is cleaved by Dicer into a mature single-stranded miRNA hairpin loop, which regulates its cognate targeted gene mRNA by 2 primary mechanisms (Fig. 2). MicroRNAs utilize the RNA-induced silencing complex (RISC) to regulate target genes by binding the 3' UTR (untranslated region). When miRNA is perfectly matched with the target mRNA, it will induce cleavage, thus inhibiting gene expression. When the miRNA is imperfectly matched, it will induce translational repression. Thus, the overall mRNA remains unchanged while gene expression is inhibited. MicroRNA can induce translational repression by (i) translation initiation inhibition, (ii) postinitiation inhibition, (iii) mRNA decay in removal foci, and (iv) mRNA storage in stress granules.²¹ In brief, miRNA can induce translation initiation inhibition by repressing the 48S translational complex assembly,²² competing the m7G of mRNA binding site with eIF4E ^23,24 (miRNA binds to the Ago2 complex to bind to m7G) or blocking poly(A)–binding protein to affect translation initiation.²⁵ Messenger RNA can induce postinitiation inhibition by leading to higher rates of ribosome drop-off, leading to immature termination during the elongation step.²⁶ Argonaute proteins are part of the catalytic components in RISC and are able to bind to small noncoding RNAs (including miRNAs, small interfering RNAs, and Piwi-interacting RNAs).^27,28 Some of the argonaute proteins have endonuclease activity to enable degradation of perfectly complementary mRNA.²⁹ In eukaryotes, argonaute proteins have been identified in high concentration in regional foci within the cytoplasm known as P bodies.^30,31 MicroRNA induces sequestration of mRNA within P bodies.³² MicroRNAs can also induce temporary storage of mRNA in stress granules, which can either be degraded or derepressed later within the cell.³³

Open in a separate window

Figure 2

miRNA biogenesis and miRNAs that are distinctly different in pancreatic cancer tissues and blood. Primary miRNA (Pri-miRNA) is transcribed in the nucleus and then processed by Drosha into Pre-miRNA before export to the cytoplasm by Exportin ⁵. In the cytoplasm, the pre-miRNA, a hairpin/stem loop, is further processed by the enzyme Dicer, producing a double-stranded miRNA. One strand of the mature miRNA is then incorporated with the RISC to regulate its gene targets. There are 2 distinct regulatory mechanisms: (1) mRNA cleavage if the miRNA is perfectly matched with its targets 3’ UTR and (2) protein synthesis inhibition if the miRNA is imperfectly matched with the targets’ 3’ UTR. The targeted genes’ mRNA will be transported to P-bodies or stress granules and are thus inaccessible to the translational machinery. Some of the miRNAs found in tumor cells will, with necrosis, be released into plasma/serum or alternatively released in exosomes.

ORIGIN OF MIRNAs PRESENT IN BLOOD

MicroRNAs can be isolated directly from blood (PBMCs are especially sensitive to microenvironmental changes including those arising in the setting of cancer), plasma, or serum. Studies of whole blood or PBMC miRNA expression to detect tumor (eg, ovarian cancer and melanoma) are developing rapidly ^11,19 (Table 1). There are also circulating miRs normally present in the serum or plasma.³⁵ Many scenarios have been formulated to explain how miRNA can survive endogenous ribonucleases that are present within blood. These include miRNA binding to DNA for protection from RNases and DNases,³⁶ as well as gaining protection by envelopment within lipid or lipoprotein carriers or vesicles,³⁷ perhaps derived from exteriorized autophagosomes (exosomes). The latter appears to be the most likely mechanism that preserves miRs in plasma and serum.³⁵ The circulating miRs in plasma and serum might originate from tumor-derived exosomes (eg, miR-21, miR-106, miR-141, miR-14, miR-155, mir-200 family, miR-203. MicroRNA-205, miR-214, etc). Note that only miR-21 and miR-18 families are found to be up-regulated in more than 2 cancer types (Table 1). Perhaps blood miR markers may be more cancer type–specific than tissue miRNA markers.^38–41 The lack of appropriate endogenous controls (a miRNA that does not change with disease stage) limits the predictive power, and further validation of the biological role of such circulating miRNAs is needed.

For therapeutic purposes, it would be useful to identify pancreatic cancer miRNAs that are shared between clinical samples and cancer cell lines (cancer cell lines are more readily available for therapeutic target validation than clinical samples). One study compared the expression profiles between individual pancreatic cancer cell lines and clinical specimens using polymerase chain reaction (PCR) (95 miRNA primers). Eight miRNAs were found to be commonly expressed in both cell lines and clinical samples (miR-196a, mIR-190, miR-186, miR-221, miR-222, miR-200b, miR-15b, miR-95).⁴⁴ When examining the clinical specimens, 20 miRNAs were overexpressed in all 5 specimens, and 11 miRNAs were overexpressed in at least 4 specimens. The results suggest that although there are similarities between pancreatic cancer cell lines and clinical specimens, the miRNA expression patterns are not identical.

MicroRNA expression profiles in normal pancreatic tissue (referred to as pancreatic miRNome), pancreatic ductal adenocarcinoma (PDAC), pancreatitis, and pancreatic cancer cell lines have been recently examined.⁴⁷ This study first created a pancreatic miRNome by clustering miRNAs that are highly expressed in pancreatic normal tissue compared with other tissues. The group used this miRNome as the parameter to measure miRNA expression changes in pancreatitis and PDAC miRNA. Twenty miRNAs were differentially expressed when comparing PDAC, chronic pancreatitis, and normal tissues. Twelve of 20 miRNAs are also differentially expressed in cancer cell lines. Furthermore, 2 potential miRNA (miR-196a and miR-217) markers are overexpressed in both primary neoplastic ductal cells and in PDAC cell lines. A similar study found that 23 (15 overexpressed and 8 underexpressed) miRNAs could be used to distinguish pancreatic cancer from pancreatitis with an extraordinary 93% accuracy.⁴⁴ These similar studies identified divergent sets of miRs, possibly because of the differences in comparison strategies and the patient populations utilized by the 2 groups. One method compared expression with normal tissue, whereas the other group compared expression with a pancreatic tissue–specific gene expression file.

Pancreatic cancer–specific miRNAs are commonly expressed in both clinical specimens and pancreatic cancer cell lines, but the expression profiles are not identical to each other. Because pancreatic tumors are indeed more than just pancreatic cancer cells, examining more stage- and cell type-specific miRNA profiles should provide a more refined result.

Pancreatic cancer is a dynamic disease. Understanding the difference between stages of pancreatic cancer utilizing miRNA profiles is very important. A murine RT2 pancreatic neuroendocrine tumor model study identified pancreatic cancer miRNA markers by stage.⁷ The study identified primary tumor stage miRNA signatures and metastasis-specific miRNA signatures by comparing the normal islets with primary tumor, liver metastases, and tumor pools. They identified miRNA signatures for hyperproliferation and angiogenesis using flow cytometry to sort hyperproliferating islets and angiogenic islets. The result of the study provides more detail on tumor stage-specific and cell type–specific miRNA signatures in pancreatic tumors.

Two other studies compared pancreatic cancer tissue with the adjacent tissue to identify miRNA markers.^43,48 One study identified 20 miRNAs that are differentially expressed in both pancreatic adenocarcinoma and cancer cell lines compared with normal pancreatic tissue miRNA.⁴³ The in situ result showed that miR-221 and miR-376a are localized to tumor cells but not to the benign pancreatic acini or stromal cells. Deregulation of miR-15a and up-regulation of miR-214 are also potential pancreatic cancer markers.⁴⁸ Microsectioning to allow in situ hybridization on epithelial cells was also compared with matched normal pancreatic tissues.⁴⁵ Ten miRNAs were differentially expressed, and 2 miRNAs (miR-21, and miR-155) had the highest fold change with miR-21 and miR-155 expression correlating with precursor lesions. The results are congruent with murine RT2 studies demonstrating that miR-21 and miR-155 are overexpressed in hyperproliferating and angiogenic islets.

Nominally specific pancreatic cancer miRNAs could be shared with other cancer types. One study compared solid tumor samples (breast, colon, lung, pancreas, prostate, stomach) miRNA expression with normal tissues (stomach, lung) from patients or individuals with no cancer (for the breast, colon, pancreas, and prostate cancer specimen).⁴² Twenty-one miRNAs were shared among 6 individual solid cancer types. Twenty of the pancreatic cancer miRNAs were shared with more than 1 solid tumor type. Most of the targets of these 21 shared miRs are identifiable tumor suppressors and/or oncogenes. Seventeen miRs were up-regulated, and 3 were down-regulated. A possible reason for variation between individual clinical pancreatic cancer profiling studies might be attributable to the stage of the patient sample and the type of cell that makes up the tumor. Therefore, a more refined classification of pancreatic cancer with cell type–specific isolation before miRNA profiling could be important for identifying suitable pancreatic miRNAs.

Another extensive study performed with human pancreatic cancer tissue identified miRs that are differentially expressed in individual patient groups.49 Ten miRs (miR-10a, miR-21, miR-143, miR-145, miR-155, miR-222, miR-223, miR-224, and miR-373) are up-regulated, whereas 7 miRs (miR-148, miR-216, miR-217, miR-211, miR-345, miR-596, and miR-708) are down-regulated. The study also characterized some nonoverlapping miRs: 9 miRs to distinguish tumor stage, 16 miRs to distinguish tumor grade, 4 miRs to distinguish the lymph node status, and miR-21 and miR-34a serving as survival predictive miRs.

COMPARING MIRNA EXPRESSION TO IDENTIFY PROGNOSTIC, SURVIVAL, AND CHEMORESISTANT MARKERS

Because the current 5-year survival rate for patients with pancreatic cancer is less than 5%, and surgical resection remains the most effective therapy, identifying markers to predict survival and determine chemoresistance may improve our ability to define subsets of pancreatic cancer patients most suitable for aggressive therapy. Some groups have combined clinicopathologic, follow-up data and miR expression to identify useful biomarkers to help predict survival and clinical outcome. Two independent studies found that miR-21 is a potential marker for survival.49,50 One group extracted RNA from fresh frozen samples, whereas the other group used in situ hybridization to profile the miRNA. Both groups found that pancreatic cancer patients with high miR-21 expression have a low median survival time (13.7 and 14.3 months), whereas patients with lower miR-21 expression have a longer median survival time (25.7 and 23.1 months, respectively). The first group also identified potential markers for better prognosis (high expression of miR-29c, miR-30d, and miR-34a) and determined that patients who have high miR-21 expression are more effectively treated with chemotherapy than those who have lower miR-21 expression. Pancreatic cancer patients with high miR-196a expression in their serum are correlated with poor survival with 100% sensitivity and 75% specificity (6.1 vs 12 months for the low miR-196a expression group).⁵¹ One study showed that patient tissue specimens that have high expressions of miR-142-5p and miR-204 correlate with a better patient survival rate (45 and 33 months vs 16.3 and 16.3 months for lower-expression group) when receiving gemcitabine treatment. Patients whose tumors express higher levels of miR-125a and miR-34a seemed to be more effectively treated by gemcitabine, although it did not reach statistical significance.⁵² The miR-200 family and miR-21 are also predictive markers for an apparent increased benefit of chemotherapy.^53,54

Sadly, based on the current literature, there is thus no common pancreatic cancer signature identified among the 8 studies summarized above. Four miRNAs are commonly overexpressed; however, in 5 studies, 3 more miRNAs are commonly overexpressed in at least 4 studies, and 2 additional miRNAs are commonly overexpressed in at least 3 studies. MicroRNA-142p and miR-141 are commonly down-regulated in pancreatic cancer in at least 2 studies, whereas the expressions of 2 other miRNAs (miR-200, miR-145) are contradictory when comparing these 2 studies (Table 3). This reflects the current disarray in the field, and reproducing results is difficult based on variation in sampling of clinical specimens, platforms used to identify miRs, and bioanalytic tools.

MIRNA PROFILING STUDIES IN PANCREATIC CANCER PATIENTS’ BLOOD

Tissue miRNA markers could do more not only to help us understand cancer biology, but also to advance therapeutic options in treating the disease. Such markers have clear limitations as early diagnostic tools for monitoring drug response and defining disease prognosis. First, there are limited solid tumor samples available to scientists. Second, such an approach requires invasive procedures to obtain biopsies from solid tumors if they are identifiable. Thus, tissue is not an ideal approach as an early-stage diagnostic method (before symptoms develop). More importantly, it is not practical to repetitively obtain solid tumor tissue miRNA to monitor disease progression. On the other hand, patients’ blood is readily available. Blood samples can easily be obtained (pretreatment/posttreatment) and may be a more appropriate sample source to establish a miRNA based biomarker for early diagnosis of cancer, prediction of drug responsiveness, and definition of prognosis. Studies have shown promising proof of concept to utilize cancer patients’ blood miRNA profile as a diagnostic and prognostic tool in pancreatic cancer. MicroRNAs can be isolated from the PBMCs, serum, or plasma components of blood specimens. Three individual studies ^12,13,34 found 6 miRNAs expressed in pancreatic cancer patients’ serum and plasma as potential biomarkers. MicroRNA-18a, miR-21, miR-210, miR-155, and miR-196a are overexpressed in a majority of the pancreatic cancer patients’ plasma examined with at least 2-fold increases.¹³ Sensitivity of greater than 40% and specificity of greater than 70% (Table 4) can be realized. When categorizing the patient population by age and sex, compared with healthy individuals, miR-200 a/b is overexpressed in primary pancreatic cancer and cancer cell lines, as well as pancreatic cancer patients’ serum.¹² A sensitivity and specificity of 84.4% and 87.5%, respectively, for miR-200a and 71.1% and 96.9% for miR-200b were found. MicroRNA-18a (one of the miR-17-92 gene cluster families) is up-regulated in primary pancreatic cancer tissue and cancer cell lines.³⁴ miR-18a expression in patient's serum was significantly reduced following surgical excision. Another study examined pancreatic cancer patient serum and investigated whether miR-21, miR-155, miR-196a, miR-181a, miR-181b, miR-221,and miR-222, which are differentially expressed in cancer tissues, can serve as biomarkers.⁵¹ Higher expressions of miR-21, miR-155, and miR-196a are observed in pancreatic cancer patients’ serum, but both miR-155 and miR-196a are also up-regulated in chronic pancreatitis. The group also found that patients who have higher miR-196a expression in the serum have a lower median survival (6.1 vs 12 months). Because immune cells respond to the cancer microenvironment and macroenvironment, our group hypothesizes that, in the presence of pancreatic tumor, the miRNA expression in patients PBMCs will be altered. Our laboratory has profiled pancreatic cancer patients PBMCs miRNA with Taqman Low Density Array (From Life Technologies, Grand Island, NY) in an age- and sex-matched study (unpublished data) and found that miR-125a-5p is up-regulated in patients, whereas miR-29c and miR-146b are down-regulated when compared with controls. We are currently continuing the study to determine if these individual miRs will change following receipt of chemotherapy and surgical extirpation. Most of the miRNAs circulating in the blood are thought to be present in membrane-bound vesicles or exosomes.⁵⁵ Differences in plasma and serum miRNA could in part reflect liberated platelet-derived miRNAs.⁵⁶

Although the serum and plasma studies provide proof of concept that circulating miRNAs could be used to diagnose pancreatic cancer, we need to use caution before applying to broader application because the prevalence of pancreatic cancer adjusted to rate is 12 per 100,000 individuals, and many false positives could be expected.⁵⁷ Positive and negative predictive values of 0.00016 and 0.9988, respectively, and in another study positive and negative predictive values of 0.05 and 0.9998, respectively, were found. Both studies provide a relatively good test to identify individuals who are negative for the disease, but the test is not useful to identify individuals who have pancreatic cancer.

Interestingly, most of those pancreatic cancer serum/plasma miRNAs have also been found in pancreatic cancer tissues in more than 1 study. MicroRNA-21, miR-155, and the miR-200 family are differentially expressed in both pancreatic cancer tissue and blood in more than 1 study. These particular miRs are involved in cell cycle, cell migration, and cell survival (Fig. 3) and are also clinically significant in cancer biology. The following section discusses the significance of these 3 best validated miRs.

Open in a separate window

Figure 3

Three commonly differentially expression MicroRNAs in pancreatic cancer tissue and patient blood: MicroRNAs 21, 155, and 200. miRNA 21, 155, and 200 are commonly differential expressions in pancreatic cancer tissue and patient blood. They play important roles in tumorgenesis and drug resistance by repressing their targeted genes. miR-21–regulated genes are involved in apoptosis, tumor invasion, metastasis, and cell cycle progression. Up-regulation of miR-21 is commonly found in cancers and the blood. Here, a schematic diagram of predicted miR-21 target genes important in cancer biology is presented. ²¹⁸ miR-21 targets PTEN (regulating expression of PI3K, FAK, Akt, matrix metalloproteinases, p53, and caspase 8), BCL2 (via PI3K targeting), PDCD4, tropomyosin, matrix metalloproteinases, RECK, and TIMPs, which promotes tumors resistant to apoptosis, metastatic growth, and increase their migratory capability. miR-21 also correlated with activation of BCL2, which reduced pancreatic cancer cell apoptosis and chemosensitivity to gemcitabine.miR-155 is found often up-regulated in cancer tissues and/or patient's serum. Potential targets of miR-155 include TP53IP1 (via caspase-3), PU.1, SOCS1, MYD88 (via JAK/STAT), which are involved in apoptosis, cell cycle progression, immune cell differentiation, and proinflammatory cytokine production all of which are important for tumor development. miR-155 overexpression is also associated with increased mutation in BRAF, RAS, and RET/PTC, genes that are commonly dysregulated in solid tumors. miR-200 family members include miR200 b/c, miR-429 (subgroup2), and miR-141. miR-200a (subgroup 1) forms a negative feedback loop with ZEB1/2, involved in the TGF[beta] signaling pathway. ZEB1 is a critical EMT activator in human cancers . ZEB2 interact with p300, pCAF to induce EMT. miR-200 family regulate EMT by regulating ZEB1/2. ZEB 1/2 also plays an important role in control stem cell–like properties, senescence, apoptosis, and proliferation (by repressing cyclin D1). miR-200 family members also target FAP1, one of the apoptosis inhibitors.219

PATHWAYS OF COMMONLY EXPRESSED PANCREATIC CANCER MIRNA IN BOTH TISSUE AND BLOOD

Currently there is no unique set of miRNA biomarkers contrasting pancreatic cancer tissue and blood miRNA profiling studies from other cancer profiles. However, there are potential miRNA biomarkers (miR-21, miR-155, and miR-200) that are identified in both pancreatic cancer tissue and patients’ blood. Are there any unique characteristics shared between those miRNAs that make them potential markers for both tissue and blood? Following the pathways that those miRNAs are involved in may provide clues to explain why these individual miRNAs can serve as suitable biomarkers.

CONNECTIONS BETWEEN MIR-21, MIR-200a/b, MIR-155, AND DEFINED GENETIC LESIONS IN PANCREATIC CANCER

Pancreatic cancer progression is associated with several defined genetic mutations or loss, and because miRNAs can regulate oncogene and tumor suppressor genes, these can in turn be also regulated by other genes. It is of interest to examine if there is any connection between commonly altered pathways, such as transforming growth factor [beta] (TGF-[beta])/SMAD4, Kras, BCRA, p53, and p16,116 and miRNAs. In our estimation, molecules released from necrotic tumor cells, especially damage-associated molecular pattern (DAMP) molecules may also alter the miRNA expression in pancreatic cancer tissue/blood. We discuss the linkage between known alterations in pancreatic cancer genetic pathways and these differentially expressed miRNAs in the following sections.

INTERPLAY OF DAMP MOLECULES AND MIRNA IN PANCREATIC CANCER

Many studies have focused on investigating the mutations that are directly responsible for cancer development. However, recent evidence demonstrates that changes in the microenvironment such as inflammation also play an important role in tumorigenesis.¹⁸³ Tumor cells have limited apoptosis and emergent pronounced increases in autophagy that results in necrosis under conditions of heightened stress. In this instance, they release so-called DAMP molecules that trigger inflammation and immunity. Damage-associated molecular pattern molecules such as HMGB1 and expression of one of its cognate receptors, RAGE (receptor for advanced glycation end products), play important roles in cancer biology.^184,185 In brief, HMGB1 is a highly conserved nuclear protein that bends DNA and promotes access to transcriptional protein assemblies on specific DNA targets.^186,187 High-mobility group box 1 can also serve as an extracellular signaling molecule during inflammation, cell differentiation, cell migration, and wound repair driving acute inflammatory response and tumor metastasis.^186–189 High-mobility group box 1 is released from necrotic and stressed autophagic cells and is actively secreted by inflammatory cells binding with receptors including RAGE, Toll-like receptor family (TLR2, TLR4, TLR9), and CD24 mediating response to infection or injury to regulate inflammation.^186–190 High-mobility group box 1 is also induced by chemotherapy and radiotherapy, and the release of HMGB1 contributes to the disordered tumor microenvironment. RAGE is a member of the immunoglobulin superfamily encoded within the class III major histocompatibility complex that can be activated by advanced glycation end products and several DAMP molecules. RAGE ligands including DNA, HMGB1, S100B, Mac1, and S100A6 activate intracellular signaling molecules (eg, nuclear factor [kappa]B, MAP kinases) to induce proinflammatory response.^191,192 Overexpression of RAGE lowers cell proliferation, and down-regulation promotes development of advanced-stage lung tumors.¹⁹³ However, in non–lung cell tumors, RAGE ligands are overexpressed.¹⁹⁴ Currently, there are few miRNA studies on HMBG1 and RAGE, but those miRNAs being investigated share some common pathways with the 3 potential pancreatic cancer miRNA markers (miR-200, miR-155, and miR-21). MicroRNA-16 is down-regulated by S100b (one of the RAGE ligands) in THP-1 monocyte cell lines.¹⁹⁵ MicroRNA-16 targets BCL2,¹⁹⁶ an antiapoptotic gene, which is differentially expressed in many tumors.¹⁹⁷ Wild-type p53 in diffuse large-B-cell lymphoma can target overexpressed BCL2 and induce cell arrest and apoptosis, but cell death is reduced when p53 is inhibited.¹⁹⁸ Our laboratory is currently investigating “DAMP-miRs” with freeze-thaw cell lysates from HMBG1 wild-type cells and HMGB1 knockout cells. MicroRNA-34c has been identified as up-regulated in human PBMCs following stimulation.¹⁹⁹ MicroRNA-34 family members are transactivation targets of p53,²⁰⁰ and miR-34 targets various cell cycle and apoptosis proteins including BCL2 and c-Myc.²⁰¹ Ectopic miR-34 expression induces apoptosis and, in the absence of miR-34c, promotes apoptosis induced by p53 activating agents.²⁰² Kras and the DAMP/RAGE pathway are connected by the p53 signaling pathway, which forms a signaling network with these 3 potential pancreatic cancer miRNA markers (Fig. 4).

Open in a separate window

Figure 4

Predicted network linking pancreatic cancer miRNA markers, known pancreatic cancer genetic lesions, and DAMPs. miR-21, miR-155, and miR-200 family members are connected to known genetic lesions found in pancreatic cancers including mutant K-ras, BRCA1, TGF[beta], and DAMPs (HMGB1) and DAMP-receptors (RAGE) as well as the p53 pathway. Note that the role played by c-myc in apoptosis regulation is paradoxical; under some conditions, c-myc promotes proliferation, and under other condition c-myc promotes apoptosis.221 c-myc is crucial for apoptosis and requires p53 to activate apoptosis.

UTILITY OF HYPOMETHYLATED OR HYPERMETHYLATED MIRNA GENES AS SPECIFIC EARLY DIAGNOSTIC MARKERS FOR PANCREATIC CANCER

The identification of specific miRNA markers is important for the early diagnosis of pancreatic cancer. DNA methylation is a process that involves the addition of a methyl group to the 5 position of the cytosine pyrimidine ring or the number 6 nitrogen of the adenine purine ring. Although methylation is essential for normal cell development and gene transcription, aberrant methylation is associated with carcinogenesis. Unmethylated CpGs are often grouped in clusters called CpG islands, which are present in the 5' regulatory regions of many genes. In many disease processes, such as cancer, gene promoter CpG islands acquire abnormal hypermethylation, resulting in transcriptional silencing that can be inherited by daughter cells following cell division.

Hypermethylation of miRNA genes leads to decreased expression of the associated mature miRNAs, whereas hypomethylation leads to enhanced expression. Hypermethylation is one of the major epigenetic modifications that repress transcription via the promoter region of tumor suppressor genes.²⁰⁴ The majority of miRNAs in tumors are repressed, indicating that they play essential tumor suppressor functions.²⁰⁵ Such aberrantly methylated miRNAs could serve as early diagnostic markers in multiple cancer types, especially in pancreatic cancer.

Multiple miRNAs have been reported to be hypermethylated and hence repressed in pancreatic cancer. One of these is miR-148a, which is down-regulated in early PDAC ²⁰⁶ and can potentially be used as an early diagnostic marker. Another epigenetically silenced miRNA in pancreatic cancer is miR-107.²⁰⁷ In contrast to repressed miRs in pancreatic cancer patients, both miR-200a and miR-200b are hypomethylated and hence are significantly elevated in patient sera with respect to healthy controls.¹² The methylation status of all the previously mentioned miRs could be utilized as early diagnostic markers for pancreatic cancer.

CURRENT CHALLENGES IN MIRNA EXPRESSION STUDIES

Many challenges remain in the field for establishing pancreatic cancer biomarkers. First, it is quite a challenge to compare the miRNA expression between individual studies. Microarray and quantitative reverse transcriptase (qRT)–PCR techniques were used in the pancreatic tissue and biofluid miRNA profiling studies, but the techniques have limitations. Prior knowledge about individual miRNAs is often required, suggesting that novel miRNAs are often omitted. Background levels may be high owing to cross-hybridization, a low dynamic expression range, and complicated normalization methods to compare individual studies.²⁰⁸ The next-generation sequencing (NGS) technology can overcome some of these limitations. Next-generation sequencing has greater expression range compared with microarray ^209,210 and has higher reproducibility between experiments.²¹¹ Furthermore, NGS provides digital readouts that make direct comparisons between studies from different laboratories possible. Some groups have used the NGS technology to discover novel miRNAs species that are not in current microarray platforms.^212,213 Therefore, applying the NGS technology to pancreatic cancer tissue and patients’ biofluid may result in a more quantifiable and comparable miRNA biomarker signature when compared with microarrays. Specific pancreatic cancer miRs could be the ones that have low copy numbers and are not expressed in any other cancer types. The NGS technology may be used as a sensitive tool to reveal these low-copy-number, cancer-specific miRs.

Second, some miRNAs are released from tumors into the serum and plasma by tumor-derived exosomes,³⁸ which may serve as another diagnostic tool. However, the mechanism by which tumor miRNAs are released by exosomes and identifying which miRNAs are released remain unclear. Studies have shown that 46% of the miRNAs in the tumor-derived exosomes are correlated with the tumor tissue miRNAs. Why the rest of 54% tumor tissue miRNAs are not released into the serum/plasma remains unclear.^38–41 It is possible that all miRNAs in the tumor are released into the serum, but the abundance is too low for qRT-PCR and microarray detection. Perhaps using the NGS technique (such as RNA-seq) can overcome some of the limitations to identify suitable miRNA markers. Currently, it is still unclear what mechanism controls the release of miRNAs from the tumor into the patients’ serum. It is possible that only metastatic tumors release the miRNAs via exosomes. Based on the current serologic miRNA studies, 2 of 3 miRNAs markers are found to be associated with metastasis. If we take a closer look at the commonly overexpressed miRNA markers in pancreatic tissues, miR-107, miR-221, and miR-16 also play a role in metastasis.^214–216 It is possible that they are indeed present in the patients’ blood, below the detection range of microarrays.

Optimizing the comparison strategy may improve the development of pancreatic cancer miRNA biomarker. Currently there are 3 widely used comparison strategies: (1) comparing to other healthy individuals, (2) comparing to adjacent normal tissues, and (3) comparing the gene expression to a pancreatic tissue–specific profile. One might argue that comparing with other healthy individuals to find differentially expressed pancreatic cancer miRNAs will identify genetic variations due to differences between individuals. On the other hand, comparisons with normal adjacent tissue removes shared common genetic variations; it also introduces problems associated with miRNA expression patterns associated with disease, as stressors derived from the tumor can promote genetic and phenotypic alterations within the surrounding tissues.^184,217 Comparing gene expression with a pancreatic tissue– specific profile can provide a more tissue-specific miRNA biomarker but could fail to identify the biomarkers that are commonly expressed in multiple cancer types.⁴² Therefore, optimizing comparison strategies is needed to improve pancreatic cancer miRNA biomarker development.

Lastly, developing a noninvasive early diagnostic strategy is crucial for patients with pancreatic cancer. Early diagnosis is rare, and surgical extirpation is thought to be most beneficial before the cancer becomes locally invasive or metastatic. Although currently there are many potential biomarkers distinguishing normal pancreatic tissues and cancer, it is not useful as an early diagnostic tool. A few circulating miRNA biomarkers are being validated and developed to distinguish healthy individuals from pancreatic cancer patients. Because the prevalence of pancreatic cancer is 12 of 10,000 in the United States, it is very difficult to develop a high positive predictive value test to screen for pancreatic cancer patients. It is necessary for any test to have at least 0.99995 specificity and a 95% positive predictive value. If individuals are screened and tests show that they are negative for pancreatic cancer with the current circulating miRNA assays available, there is only 0.2% likelihood that they have pancreatic cancer. Therefore, if individuals are screened and results show that they are positive for pancreatic cancer, although the positive predictive value is only between 0.016% and 5% that they do have pancreatic cancer, it can allow them to undergo further examination to confirm if they have the disease as an early diagnostic test. MicroRNA-18a and miR-200a/b might serve as biomarkers to monitor the disease following treatment as they show promising sensitivity and specificity when the individual is confirmed to have pancreatic cancer.

SUMMARY

Pancreatic cancer miRNA biomarker signatures appear to be a protean area of investigation for future diagnostic or therapeutic purposes. Although there are potential pancreatic miRNA biomarkers in pancreatic tissue and patients’ blood, those biomarkers are not pancreatic cancer–specific, but could be quite useful in studying recurrence or progression. It is possible to establish a miRNA cancer biomarker signature, but distinguishing the site of origin of the cancer also remains challenging. Furthermore, because cancer is a dynamic disease, presorting the patients’ sample based on disease stages, ethnicity, and age before miRNA profiling might facilitate the identification of unique pancreatic cancer signatures for individual stages of cancer. It will also be interesting to apply the NGS technology to profile the cancer tissue and biofluid miRNAs to develop a more quantifiable and comparable, cancer type–specific miRNA signature for pancreatic cancer diagnosis and therapeutic target development. What is quite clear is that as our deeper understanding of the tumor microenvironment and macroenvironment reveals complexities of genetic and epigenetic control mechanisms, the frequent occurrence of aberrant forms of cell death in response to chronic stress requires that more holistic approaches integrating the known genetic changes and miR expression patterns now be considered.^218–224 Integrating the host response with the panoply of genetic changes the occurrence within the tumor are now necessary for a full explication of cancer biology and the development of effective diagnostic tests and therapies.

Acknowledgments

Footnotes

REFERENCES