Received date: October 20, 2015; Accepted date: November 12, 2015; Published date: November 19, 2015
Citation: Ali S, Dubaybo H, Brand RE, Sarkar FH (2015) Differential Expression of Micrornas in Tissues and Plasma Co-exists as a Biomarker for Pancreatic Cancer. J Cancer Sci Ther 7:336-346. doi:10.4172/1948-5956.1000372
Copyright: © 2015 Ali S, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Visit for more related articles at Journal of Cancer Science & Therapy
Objective: Pancreatic cancer (PC) is a lethal disease with disappointing results from current treatment modalities, suggesting that novel therapeutic strategies are urgently needed. Since microRNAs (miRNAs) are important player in biology, the clinical utility of miRNAs for designing novel therapeutics is an active area of research. The objective of the present study was to examine differentially expressed miRNAs between normal and tumor tissues, and in plasma samples obtained from PC patients, chronic pancreatitis (CP) patients and healthy subjects (HC).
Material and methods: The miRNA expression profiling using formalin-fixed paraffin embedded (FFPE) tissues from normal and tumor specimens was accomplished using miRBase version 19 (LC Sciences, Houston, TX, USA). Quantitative real-time PCR (qRT-PCR) was subsequently performed in individual samples for 7 selected miRNAs. In addition, qRT-PCR was also performed for assessing the expression of 8 selected miRNAs in plasma samples.
Results: A significant difference in the expressions of miR-21, miR-205, miR-155, miR-31, miR-203, miR-214 and miR-129-2 were found in tumor tissue samples. Lower expression of miR-214 was found to be associated with better overall survival. We also observed differential expression of 8 miRNAs in plasma samples of CP and PC patients compared to HC. Interestingly, over expression of miR-21, and miR-31 was noted in both tumor tissues and in the plasma.
Conclusion: We found deregulated expression of miRNAs that could distinguish normal from PC in two different types of samples (tissues and plasma). Interestingly, lower expression of miR-214 was found to be associated with better overall survival. Although not statistically significant, we also observed higher expression of let-7a and lower expression of miR-508 to be associated with overall better survival. We conclude that our study nicely lays the foundation for detailed future investigations for assessing the role of these miRNAs in the pathology of pancreatic cancer.
miRNAs; Chronic pancreatitis; Pancreatic cancer; Plasma; FFPE; qRT-PCR
Despite considerable progress made in understanding the biology of pancreatic cancer (PC) during the past decade, PC still remains the most lethal cancer with a five-year overall survival of 7% . There has been no measurable improvement in early detection and diagnosis, suggesting that identification of sensitive and precise non-invasive biomarkers at an early stage would be beneficial in distinguishing PC patients from healthy individual. To that end, differential expression of microRNAs (miRNAs) could become a promising strategy especially because miRNAs are more stable than mRNAs, known to regulate multiple target genes post-transcriptionally, and reported to play important roles in oncogenesis and tumor metastasis . The miRNAs are valuable biomarker for investigating their levels in blood serum, plasma and archived material such as formalin-fixed paraffin embedded (FFPE) tissues. In addition, Shimizu et al. has reported that carcinoma of the pancreas when resected earlier at stage I before the tumor size reaches 2 cm in diameter and remains confined to the pancreas has a better prognosis , which clearly suggest that early diagnosis and surgical management of this deadly disease is crucial because the treatment outcome with conventional chemotherapeutic agents are disappointing for patients diagnosed with PC at late stages.
One of the risk factors for PC is chronic pancreatitis (CP), a benign inflammatory disease which is also difficult to differentiate from early stage PC. An interesting observation was made by Bloomston et al., who observed a distinct pattern of miRNA expression from FFPE samples that may differentiate PC from CP and normal pancreas . In line with this and other studies, Schultz et al. reported the identification of two investigative panels based on the expression levels of miRNA from blood samples in a large study with the probability of distinguishing PC patients from CP and healthy individuals . In addition, detection of a three-protein biomarker panel in urine samples of early stage PC patients compared to healthy individuals has been suggested to become a valuable non-invasive and inexpensive screening test for PC . Moreover, the same study suggested that the three proteins REG1A, TFF1 and LYVE1 when combined with CA19-9 (the only PC biomarker in clinical use) may increase the accuracy of the screening test, confirming the biological importance of three-panel protein and CA19-9 as an early diagnostic biomarker for PC .
The expressions of miRNAs such as miR-21, miR-31, and miR- 155 have been extensively studied by many investigators showing that the expression of these miRNAs in PC could be important [7-12]. The expression of the above three miRNAs was quantified in normal and tumor tissues of PC patients as well as in plasma samples of HC, CP and PC in the current study with an aim to differentiate their expression levels between different groups. In contrast, the expression of miR-129-2 has been less extensively studied in PC . The same group identified miR-129-2 down-regulation occurs in most pancreatic carcinoma samples compared to tumor-uninvolved tissue samples from the same patient which was further correlated with increased expression of SOX4 mRNA . These limited studies suggest that further investigations for assessing the role of differential expression of miRNAs between benign and tumor specimens and more importantly in serum or plasma samples is warranted. If confirmed, miRNAs could become a useful tool for early diagnosis and prognosis, and for designing and developing novel miRNA-based targeted therapeutics.
The aim of the present study was to assess differential expression of miRNAs among PC patients and healthy individuals in formalin-fixed paraffin-embedded (FFPE) tissues and plasma samples. The selection of seven miRNAs as discussed below was based on our miRNA microarray results for their differential expression, and also based on some limited literature. The relationship between overall survival and the differential expression of seven miRNAs in PC was examined quantitatively using FFPE samples that included the expressions of miR-21, miR-205, miR- 155, miR-31 miR-203, miR-214 and miR-129-2-3p. We also found that 5 miRNAs were up-regulated and 1 miRNA was down-regulated in PC compared to CP and normal plasma samples. Up-regulated miRNAs included miR-21, miR-221, miR-181a, miR-935, and miR-508, and down-regulated miRNA was let-7a. Interestingly we also observed two miRNAs (miR-155, miR-31) that were significantly up-regulated in CP compared to PC plasma samples. Although our study had limitations on sample size, nevertheless our results using tumor and normal tissues, and plasma led to the identification of three miRNAs (miR-21, miR-155 and miR-31) that were commonly expressed in PC compared to normal. The panel of these three miRNAs may facilitate in differentiating HC from PC and that the expression of miR-214, let-7a and miR-508 may also predict overall survival of PC patients. The knowledge gained from this study may lead to the development of miRNA-based tailored therapeutic strategy for improving the treatment outcome of patients diagnosed with PC.
Histopathology slides (HandE stained slides) from pancreatic adenocarcinoma (PC) patients who underwent surgical resection was evaluated by a pathologist, and the representative slides having predominantly tumors were selected (n=37) from the database of the Wayne State University. In addition, slides with only normal pancreatic tissue (n=24) were also selected. The representative formalin-fixed paraffin embedded (FFPE) tissue blocks corresponding to the pathology slide were pulled from the storage. Benign adjacent normal pancreatic tissue was not available from all 37 PC patients. For paired comparison of selected miRNAs, 24 cases of paired pancreatic non-tumor and tumor tissues were obtained. The institutional human investigation review board approved the study protocol. For normal and tumor tissue sections, four sections, each of 10 microns in thickness, were cut from the selected blocks and were used for the study.
The plasma samples from 20 healthy controls (HC), 20 chronic pancreatitis (CP) and 20 adenocarcinoma (PC) (Stage IIB) were processed and stored in the laboratory of Dr. Randall Brand at the University of Pittsburgh, Pennsylvania. Samples included about half male and half female. The institutional human investigation review board approved the study.
RNA Isolation from FFPE tissue blocks
The RNA was isolated from FFPE tissues using the RNeasy Kit (Qiagen, Valencia, CA, USA) in accordance with the manufacturer’s protocol. Briefly, four tissue sections were placed in micro tubes, and 1 ml xylene was added and RNA was isolated as described previously . RNA was eluted in a final volume of 25 μl, quantified and its purity was evaluated by the absorption ratio at 260/280 nm using NanoDrop 2000 (Thermo Scientific, Pittsburgh, PA, USA).
RNA Isolation from plasma samples
About 200 μl of plasma from each sample was mixed with 750 μl of QIAzol (Life Technology) and 1μg of carrier RNA (MS2 RNA, Roche Diagnostic) and was incubated for 5 minutes at room temperature. The addition of carrier RNA preceding to RNA extraction improves recovery of miRNA from plasma samples and allows use of very small amount of clinical samples. To this 200 μl of chloroform was added, mixed well, and centrifuged for 15 minutes at 12,000 × g which was described previously . About 1.5 volume of ethanol was added to the upper aqueous phase and the solution was then transferred onto the RNeasy Mini Spin Column and centrifuged at 13,000 × g for 30 seconds. The flow-through was discarded, and the steps were repeated until the entire sample was used. The RNeasy Mini Column was then washed with buffers provided with the kit, centrifuged for 13,000 × g for 1 minute and the RNA (containing miRNAs) was then eluted with about 25 μl of water. The RNA obtained from 200 μl of plasma samples cannot be quantified due to low yield compared to tissue samples. Hence, the miRNAs of interest will be reverse transcribed using standard curve from the template mature miRNAs available commercially (Applied BioSystems).
Initial screening of miRNAs in normal and tumor samples from FFPE using miRNA microarray profiling provides a comprehensive way of determining a large number of miRNAs from very small amounts of difficult samples such as FFPE. Equal quantity of extracted RNA from all samples was pooled into two tubes for example pancreas normal and pancreas tumor. LC Sciences then qualitatively and quantitatively analyzed the RNA for miRNA microarray profiling, using miRBase version 19 (LC Sciences, Houston, TX, USA). The data was normalized using selected housekeeping genes. Furthermore, the web-based Ingenuity pathway analysis software was used to perform network analysis (Ingenuity Systems, Redwood City, CA, USA).
Quantitative Real-time Polymerase Chain Reaction (qRTPCR) from FFPE tissue samples
Quantitative-RT-PCR was performed on the individual samples in order to validate the miRNA profiling results of 7 selected miRNAs using TaqMan Universal PCR Master Mix, no AmpErase UNG. Selected miRNAs were miR-21, miR-205, miR-155, miR-31, miR- 203, miR-214 and miR-129-2-3p. The High Capacity cDNA Reverse Transcription Kit (Applied BioSystems, Foster City, CA, USA) was used per manufacturer’s protocol. Approximately 10 ng of RNA from the respective tissue specimens was reverse transcribed using 7 μl of master mix and 3 μl of respective RT primers as described earlier . PCR reactions were then carried out using resulting cDNA, miRNA specific probes and the TaqMan Universal PCR Master Mix in triplicate using Step One Plus Real-Time PCR (Applied BioSystems, Foster City, CA, USA) as described previously . Relative expression of miRNAs was analyzed using the Ct method and was normalized by RNU48 expression.
Quantitative Real-time Polymerase Chain Reaction (qRTPCR) using plasma samples
Quantitative-RT-PCR was performed on the individual plasma samples (60) from three different groups (HC, CP and PC) with 8 selected miRNAs using Exiqon-Universal cDNA Synthesis kit (Exiqon, Woburn, MA). Three of the selected miRNAs were common from FFPE tissue samples miR-21, miR-155 and miR-31, and the other five miRNAs were chosen based on other investigators findings let- 7a, miR-221, miR-181a, miR-935 and miR-508. The quantification of all the above recognized 8 miRNAs that are known to express at different levels was investigated in all 60 plasma samples from three different groups. The cDNA for standard curve was synthesized by reverse transcriptase using the template mature miRNAs from Applied BioSystems. The RT reaction included 20 μl of sample containing 4 μl of 5X RT buffer, 2 μl of enzyme mixture, 10 μl of water and 4 μl of either the plasma miRNA or 4 μl (250 nM) of standard miRNA. The mixture was incubated for 60 minutes at 42°C and 5 minutes at 95o C. PCR reactions were performed in equivalent with standard miRNAs to evade batch effects. The standard cDNA prepared earlier were diluted with water and the standard curve was set up in triplicate with five points starting at 10,000, 5000, 2500, 1250 and 625 copy numbers. The plasma cDNA was diluted to 20 folds. The PCR reaction was set up with a total volume of 10 μl containing 5 μl of SYBR Green (Applied BioSystems), 1 μl of PCR primer mix, 4 μl of either the diluted standard cDNA or the plasma cDNA using standard curve model. The plasma miRNA concentration was calculated in 10-2 pM units using the quantity value *3.125/6.02/1000.
Differences in the expression levels of miRNAs between normal and tumor paired samples were statistically evaluated by non-parametric Wilcoxon matched pairs test using GraphPad StatMate software (GraphPad Software Inc.). The p values that represent differences between normal and tumor samples are displayed in the graph. In addition, differences in the expression levels of miRNAs between HC vs CP, and HC vs PC were statistically evaluated by GraphPad StatMate software using unpaired t-test. The p values between those two comparisons are shown above the bar.
The correlation between survival outcome and the miRNAs expression levels was determined by the Kaplan-Meier survival analysis of PC patients from both study sites. The survival was defined as time from diagnosis to any cause of death. The Log-rank test was used to evaluate the survival difference among the low vs high groups defined by median value of miRNA expression.
Description of study patients
Archival FFPE tissue blocks of 37 patients were both clinically and pathologically confirmed for pancreatic cancer (PC). We also collected FFPE tissue blocks of normal pancreas from 24 patients out of 37 from the same study group. Two patients were removed from the survival curve analysis due to lack of the availability of the survival data. The median age was 63 and gender count was 40.5% male and 59.5% female. For the miRNA study from the blood samples, 20 healthy controls (HC), 20 chronic pancreatitis (CP) and 20 pancreatic adenocarcinoma (PC) (Stage IIB) were selected, and the plasma were isolated in Dr. Randall Brand’s laboratory at the University of Pittsburgh, Pennsylvania. Samples included about half male and half female.
The results of miRNA expression profiling
Purified RNA samples were pooled (equal amount of RNA from each patients material) separately from FFPE tissue samples of tumor and normal, and were analyzed by LC Sciences for miRNA microarray profiling using miRBase version 19 (LC Sciences, Houston, TX). Expression profiling revealed 291 miRNAs that were differentially expressed between normal and tumor of PC patients. Based on the above results, we selected best seven miRNAs that were found to be significantly deregulated for further validation using qRT-PCR in each samples. These miRNAs included miR-21, miR-205, miR-155, miR-31, miR-203, miR-214 and miR-129-2-3p. The analyses of the profiling results on these miRNAs are illustrated in the following sections.
Ingenuity pathway analysis for deregulated miRNAs
Network analysis was performed with the web-based bioinformatics tool Ingenuity pathway Analysis (IPA) software (Ingenuity Systems) to understand the target genes and pathways involved in PC. The analysis revealed the influence of many frequently studied pathways such as NF- κB, MAP kinase, Pro-inflammatory Cytokines, insulin and EIF2C2, as illustrated in supplementary Figure 1A and 1B.
Figure 1: Comparative expression analysis of miR-21 and miR-205 in 37 pancreatic cancer patient’s tumor specimens compared to 24 pooled normal samples of FFPE tissue blocks (1A&1D), comparative expression analysis in 24 paired samples of FFPE tissue blocks of tumor and normal tissue samples from the same patient (1B&1E) quantitated individually using qRT-PCR. The Kaplan-Meier curve and Log-ranks tests for miR-21 (1C) and miR-205 (1F) expression and survival of patients are also presented. There was a significant up-regulation of miRNA-205 in almost all tumor samples when compared to Normal (1D&1E). Overall, a larger percentage of miR-21 appears to be up-regulated, in tumor samples than normal samples (1A&1B). The miRNAs expression was normalized using RNU48 miRNA. P values represent comparison between normal and tumor paired samples (1B,1E) using Wilcoxon matched pairs t-test.
Seven potential miRNAs to differentiate normal from tumor of the FFPE samples as assessed by qRT-PCR
Seven miRNAs were selected for further validation individually in 37 PC patient samples compared to pooled normal tissue (with no signs of cancer present in the specimens) obtained from FFPE tissue block that were further away anatomically from the pancreatic tumor and the results are shown in Figures 1-3 (A and D) and Figure 4A using qRTPCR. The expression analyses of seven miRNAs were also performed using 24 paired samples of normal and tumor tissues from the same PC patient as presented in Figures 1-3 (B and E) and Figure 4B. The relative differences in the expression level of the selected miRNAs in PC were established by setting the expression level of normal samples at unit value (1.0).
The miRNA expression analysis exhibited 6 miRNAs that were up-regulated in most of the tumor samples compared to the normal samples which included miR-21, miR-205 (Figure 1A and 1D), miR- 155, miR-31 (Figure 2A and 2D) and miR-203, miR-214 (Figure 3A and 3D). Only one miRNA (miR-129-2-3p) out of seven showed significant down-regulation (Figure 4A) in tumor tissue of most of the PC patient samples tested compared to normal samples. The expression levels in 24 paired tumor and normal tissue samples of miR-21, miR-205 (Figure 1B and 1E), miR-155, miR-31 (Figure 2B and 2E), miR-203, miR-214 (Figure 3B and 3E) showed significant up-regulation of all 6 miRNAs when compared to their paired normal tissue samples, suggesting the oncogenic role of all six miRNAs. The overall trend of up-regulation of all six miRNAs between the paired samples is clear from the figures even though the sample size was small. The expression level of miR- 205, miR-155 and miR-203 was significantly enhanced in all except few PC patients as depicted in Figure 1D, 2A and 3A.
Figure 2: Comparative expression analysis of miR-155 and miR-31 in 37 pancreatic cancer patient’s tumor specimens compared to 24 pooled normal samples of FFPE tissue blocks (2A&2D), comparative expression analysis in 24 paired samples of FFPE tissue blocks of tumor and normal tissue samples from the same patient (2B&2E) quantitated individually using qRT-PCR. The Kaplan-Meier curve and Log-ranks tests for miR-155 (2C) and miR-31 (2F) expression and survival of patients are also presented. There was a significant up-regulation of miRNA-155 in all except two tumor samples when compared to normal (2A). Similarly, miR-31 appears to be up-regulated in most of the tumor samples compared to normal samples (2D&2E). The miRNAs expression was normalized using RNU48 miRNA. P values represent comparison between normal and tumor paired samples (2B, 2E) using Wilcoxon matched pairs t-test.
Figure 3: Comparative expression analysis of miR-203 and miR-214 in 37 pancreatic cancer patient’s tumor specimens compared to 24 pooled normal samples of FFPE tissue blocks (3A&3D), comparative expression analysis in 24 paired samples of FFPE tissue blocks of tumor and normal tissue samples from the same patient (3B&3E) quantitated individually using qRT-PCR. The Kaplan-Meier curve and Log-ranks tests for miR-203 (3C) and miR-214 (3F) expression and survival of patients are also presented. There was a significant up-regulation of miRNA-203 and miR-214 in all except few tumor samples when compared to normal (A,B,D and E). Furthermore, Kaplan-Meier curve for miR-214 (3F) demonstrated a significant longer survival of patients with low miR-214 expression. The miRNAs expression was normalized using RNU48 miRNA. P values represent comparison between normal and tumor paired samples (3B,3E) using Wilcoxon matched pairs t-test.
Conversely, miR-129-2-3p showed down-regulation in its expression in most of the PC patient samples when compared to normal pooled samples (Figure 4A). This general trend of lower miR-129-2-3p expression was also observed in paired tumor samples when compared to their paired normal tissue specimens (Figure 4B), indicating that miR-129-2-3p may be a tumor suppressor miRNA. RNU48 was used as control miRNA to normalize the miRNA expression of all samples.
Figure 4: Comparative expression analysis of miR-129-2-3p in 37 pancreatic cancer patient’s tumor specimens compared to 24 pooled normal samples of FFPE tissue blocks (4A), comparative expression analysis in 24 paired samples of FFPE tissue blocks of tumor and normal tissue samples from the same patient (4B) quantitated individually using qRT-PCR. The Kaplan-Meier curve and Log-ranks tests for miR-129-2-3p expression and survival of patients are also presented in 4C. There was a significant down-regulation of miRNA-129-2-3p in most of the tumor samples when compared to normal (4A). The miRNAs expression was normalized using RNU48 miRNA. P values represent comparison between normal and tumor paired samples (4B) using Wilcoxon matched pairs t test.
Correlation between miRNAs expression from FFPE samples and the survival of 35 PC patients
The correlation between survival outcome and the seven miRNAs expression levels was determined by the Kaplan-Meier survival analysis for 35 PC patients by using median miRNA expression value which has been represented in Figures 1-3 (C and F) and Figure 4C. The survival was defined as time from diagnosis to any cause of death. The Log-rank test was used to evaluate the survival difference among the low vs high expressing miRNA relative to the mean expression of each miRNA by qRT-PCR. Interestingly, one miRNA (miR-214) showed the Log-rank p value as statistically significant when compared between low miRNA vs high miRNA (Figure 3F), suggesting that miR-214 may serve as an important prognostic marker for PC patients. Although miR-21 (Figure 1C) Log-rank p value was not significant, it had a wider curve and the patients with low miRNA expression tend to survive longer than the PC patients with high miR-21 expression. This data is consistent with our previous finding on plasma miRNA study . No such observation was observed for miR-205 (Figure 1F), miR-155 (Figure 2C), miR-31 (Figure 2F), miR-203 (Figure 3C) and miR-129-2-3p (Figure 4C).
Eight potential miRNAs to differentiate PC from CP and HC in plasma samples as assessed by qRT-PCR
Eight miRNAs were selected to measure their expression levels, and quantitated the amount of miRNAs present in each of the 20 HC, 20 CP and 20 PC plasma samples. Three of the selected miRNAs (miR- 21, miR-155, and miR-31) were based on findings showing that their expressions were found to be common in FFPE tissue samples, and the other five miRNAs were chosen (let-7a, miR-221, miR-181a, miR-935 and miR-508) based on our previous research and other investigators findings for further quantification individually in 20 samples in each of the group.
As can be seen in Figure 5A, the plasma samples from CP and PC showed significant up-regulation of miR-221 when compared to HC. This miRNA is known to be an oncogenic miRNA in PC and other cancers as reported by our group and also by other investigators [7,18- 21]. In contrast, let-7a was significantly down-regulated in CP and PC plasma samples when compared to HC (Figure 5B) which is consistent with published results, suggesting the global tumor suppressive role of let-7a [22-25]. Expression levels were also compared for miR-155 and miR-31 between all HC, CP and PC plasma samples, as shown in Figure 5C and 5D. The expression levels of these two miRNAs surprisingly showed overwhelming differences in CP samples compared to HC which were then decreased in PC samples. This interesting finding deserves further in-depth research. When miR-31 was examined in tumor samples isolated from FFPE tissue samples (Figure 2) and also PC plasma samples (Figure 5D) there appeared to be an overall increase in their expression when compared to controls, suggesting that miR- 31 may serve as oncogenic miRNA and can be measured and detected in a variety of patient samples especially in plasma.
Figure 5: Comparative expression analysis of miR-221 (5A), let-7a (5B), miR-155 (5C) and miR-31 (5D) from plasma samples of HC, CP and PC quantitated individually using qRT-PCR (n=20/group). In miR-221, there appears to be an increase in the expression in CP and PC compared to HC, suggesting an oncogenic role of miR-221 (5A). In contrast, the expression of let-7a was significantly decreased in PC compared to both HC and CP patient’s plasma samples indicating a tumor suppressor role of let-7a (5B). Both miR-155 and miR-31 expression was mostly found to be up-regulated in CP patient samples followed by PC and primarily lower expression was found in HC plasma (5C&5D). The plasma miRNAs concentration was calculated using the standard miRNA concentration in 10-2 pM units using the Quantity value *3.125/6.02/1000. P values represent comparison between HC vs CP and HC vs PC using t-test.
As depicted in Figure 6(A-D) the expression levels of miR-21, miR- 181a, miR-935, and miR-508 gradually increased from HC to CP to PC, suggesting their roles in tumor progression. It’s noteworthy that the differential expression from HC to CP was not significant; however, as we move from CP to PC, the enhancement in their expression became very significant. Further assessment of the expression levels of miR-21 from FFPE tissue samples (Figure 1 A and 1B) appeared to show overall up-regulation in FFPE tumor samples and also in PC plasma samples (Figure 6A), suggesting that miR-21 is a global oncogenic miRNA that can be discovered in patient samples, and these results are consistent with our previous findings on plasma samples .
Figure 6: Comparative expression analysis of miR-21 (6A), miR-181a (6B), miR-935 (6C) and miR-508 (6D) in plasma samples of HC, CP and PC subjects quantitated individually using qRT-PCR (n=20/group). There appears to be a gradual increase in the expression in CP and PC compared to HC in all four miRNAs as presented in Figure 6. The plasma miRNAs concentration was calculated using the standard miRNA concentration in 10-2 pM units using the Quantity value *3.125/6.02/1000. P values represent comparison between HC vs CP and HC vs PC using t-test.
Correlation between miRNAs expression in plasma samples and the survival of 20 PC patients
The correlation between survival outcome and the expression of eight miRNAs from plasma samples was determined by the Kaplan- Meier survival analysis for 20 PC patients by using median miRNA expression value as depicted in supplementary Figure 1. The survival was defined as time from diagnosis to any cause of death. The Log-rank test was used to evaluate the survival difference among the low vs high expressing miRNA relative to the mean expression of each miRNA by qRT-PCR. Based on the sample size limitation in the plasma study, the survival analysis does not claim to provide definitive conclusion on any of the miRNAs that were tested. However, it does aim to draw scientific attention to the possibility of revealing differences in two miRNAs of which one is oncogenic (miR-508) and other one is tumor suppressor (let-7a), and their correlation with overall survival is presented (Supplementary Figure 1). Although the Log-rank p value was not statistically significant when compared between low and high expression of miR-508, a wider curve was observed and the patients with low miR-508 expression tend to survive longer than the PC patients with high miR-508 expression. In contrast, let-7a higher expression had a wider curve and the patients with high let-7a expression tend to survive longer than patients with low let-7a expression. No such observation was observed for miR-221, miR-155, miR-31, miR-21, miR-181a and miR-935 (Supplementary Figure 1). Even though our sample size was small our results may facilitate further studies in the determination of which miRNAs may serve as biomarkers and are essential for future development of miRNA-targeted therapies to optimize personalized treatments of PC patients.
Despite intensive efforts and innovative discoveries in the developments of early stage biomarkers, none has yet been accurately translated in the clinical setting. Due to the lack of early clinical symptoms of the disease, the majority of PC patients are diagnosed very late when the disease is much advanced either locally or show metastasis. Hence, developments of early stage biomarkers are crucial for PC patients. Differential expression of specific miRNAs between normal and PC have been studied in recent years showing that it could serve as biomarkers to identify PC patients at an early stage. The miRNAs are known to regulate cell proliferation, differentiation and carcinogenesis for a number of malignancies including PC [2,15].
In this study, we assessed and found comprehensive expression profile of several miRNAs that are altered during tumor progression. We have previously reported the activation of many frequently studied signaling pathways such as NF-κB, and MAP kinase in other malignant diseases . Similarly, we observed the activation of NF-κB, MAP kinase, Pro-inflammatory Cytokines, insulin and EIF2C2 in PC as demonstrated in supplementary Figure 2Aand2B. This result supports the notion that NF-κB activation may be involved in the progression of several malignancies including PC.
Of the several miRNAs that were found to be altered from our profiling results between normal and tumor samples, we chose seven miRNAs for qRT-PCR validation individually in each of the samples. Six miRNAs appear to be up-regulated and one down-regulated in tumor tissue in most of the samples in our cross validation study by qRT-PCR. We also observed a significantly lower expression of miR-129-2 in many tumor samples which is consistent with another published study . The same published study demonstrated decreased expression of miR-129-2 in PC tumor samples compared to that of non-tumor samples, which was associated with over expression of SOX4 transcription factor with shorter survival patients diagnosed with PC. These authors further demonstrated that re-expression of miR-129-2 in PC cells by transient transfection had minimal effect on SOX4 expression, suggesting the involvement of other miRNAs along with miR-129-2 in the regulation of SOX4 .
The miR-21 and miR-221 has been extensively studied in PC and other cancers [7,19,21,26]. In our current study, we found a significant increase in the expression of both miR-21 and miR-221 in PC samples. We and others have demonstrated over expression of the above two miRNAs in different cells derived from PC patients such as like stellate cells, cancer-associated fibroblast (CAF) and cancer stem-like cells [7,26]. The inhibition of these two miRNAs by antagomir transfection significantly reduced cell migration, invasion,  cell differentiation and downstream gene regulation, suggesting that the assessment of these two miRNAs in the plasma could be useful marker of early disease and tumor progression.
One less studied miRNA in PC is miR-205 [27,28]. We observed a significant increase in the expression of miR-205 in most of the tumor samples. One study examined miR-205 expression level in exocrine pancreatic secretions from PC patients and observed a marked increase in the expression of miR-205 along with miR-210 compared to nonpancreatic, non-healthy as controls, which was further correlated with decreased overall survival, and differentiated tumors and lymph node metastasis .
Other miRNAs, for example miR-31 and miR-155 were found to be up-regulated in PC and CP regardless of sample type. The observed up-regulation of miR-31 and miR-155 reported by Wang et al. in next generation sequencing studies of pancreatic cyst fluid from low grade-benign and high grade-invasive lesions suggested that these miRNAs may serve as early detection biomarkers of PC developing from pancreatic cystic lesions . A meta-review of published studies comparing miRNA expression between PC tissues and neighboring non-cancerous tissues identified 10 miRNAs to be used as diagnostic markers and therapeutic targets . Amongst the 10 miRNAs identified, 7 miRNAs are up-regulated and 3 are down-regulation. Some of the up-regulated miRNAs such as miR-21, miR-31, miR-221 and miR- 155 are consistent with the findings reported in this study. Not only in PC patient samples but in a comprehensive study involving 16 PC cell lines, miR-31 was identified as a unique miRNA whose expression level was extremely high in 10 of the 16 PC cell lines. Interestingly, both the inhibition and re-expression of miR-31 in AsPC-1 PC cell line reduced cell proliferation, migration and invasion associating miR-31 as an exceptional miRNA whose level is critical for these phenotypes . The expression of miR-155 when inhibited with antagomir transfection increased the expression of Suppressor of cytokine signaling 1(SOCS1) and simultaneously decreased the invasion and migration of PC cells as demonstrated by Huang et al. . Previous research has demonstrated miR-155 as a negative regulator of Mut homologue 1 (MLH1) protein expression, and the over expression of this protein was found to be associated with less lymph node metastasis . The observed up regulation of miR-155 in tumor samples by the above investigators is consistent with our findings in both the samples types tested.
Another over expressed miRNA was miR-203, which showed increased expression in most of the tumor samples. Up regulation of miR-203 was also detected by Ikenaga et al. in FFPE tissue samples of PC compared to CP and normal pancreas using qRT-PCR, which was associated with shorter survival time . Although we also observed over expression of miR-203 in PC samples there was no change in the survival difference which may be due to limited sample size.
Interestingly, the increased expression of miR-214 was observed in human heart failure . PTEN, TP53, TWIST1 are some of the representative targets of miR-214 as highlighted in the same review article. The miR-214 was distinguished as a bi-functional cardiomiR that plays good and bad roles . Over expression of miR-214 was also demonstrated in gastric cancer and PTEN expression was negatively regulated through a miR-214 binding site within the 3’- UTR at the posttranscriptional level . Our current pilot study showed increased expression of miR-214 in tumor samples which was associated with overall survival, and thus deserves further future study in a larger patient population.
We also have identified over expression of miR-181a, miR-508 and miR-935 in plasma samples of PC patients. Most notably, miR- 181 was uniquely increased in CP and PC compared to HC. The other two miRNAs (miR-508 and miR-935) were over expressed in PC only compared to CP and HP, suggesting that these two miRNAs are specific for PC. Our findings of increased expression of miR-181a in tumor samples were consistent with another study relative to normal samples . In hepatocellular carcinoma (HCC), miR-181a was also identified as circulating miRNA that would differentiate HCC patients from chronic liver disease and from normal controls . Interestingly, miR-181a was negatively correlated with tumor suppressors PTEN and MAP2K4 expression as proposed by another study in PC . Both previous study and the results of our current study suggesting that miR-935 appear to be up regulated in PC compared to HC . The expression of miR-508 was also found to be up regulated in PC plasma samples. When compared to other miRNAs, miR-508 has been briefly studied in cancer. Lin et al. observed direct suppression of multiple phosphatases such as PTEN with miR-508 expression in esophageal squamous cell carcinoma consistent with activated PI3K/Akt signaling pathway .
Finally, we found that let-7a was significantly down regulated in PC plasma samples comparted to HC and also in CP compared to HC. Although not significant but over expression of let-7a in PC plasma samples showed association with overall better survival. Alterations of this miRNA in our study were similar to those described in the literature earlier by our group as well as by other investigators [22,23,40-42]. The aberrant expression of let-7a in many types of human cancers including PC indicates its tumor suppressor role [22,24,43,44]. For example, over-expression of let-7a with diflourinated-curcumin (CDF) mediated tumor regression with reduced EZH2, Notch-1, and EpCAM expression in PC in our previous study . Moreover, the expression of doublecortin-like kinase 1 (DCLK1) in PC was inhibited by XMD8- 92, a kinase inhibitor via up regulation of tumor suppressor miRNA let-7a .
Although our study had limitations on sample size, the present report contributes to the rising understanding of the role that miRNAs may play in differentiating PC patients from healthy individuals or patients with CP. Our results from both sample resources identified three miRNAs (miR-21, miR-155, and miR-31) that were common and presented at similar expression levels in PC compared to normal. The panel of these three miRNAs may facilitate in differentiating PC from HC, and that miR-214 expression may predict overall survival. In conclusion, we anticipate that plasma miRNA research being non-invasive will have a prevailing clinical consequence for the development of miRNA-based targeted therapies that may lead to a much improved therapeutic outcome in patients diagnosed with PC. Finally, we conclude that our study nicely lays the foundation for detailed future investigations for assessing the role of these microRNAs in the pathology of pancreatic cancer.
This work was funded by grant from the National Cancer Institute, NIH 5R01CA154321 05 awarded to FHS. We also acknowledge the generous funding of Puschelberg Foundation. We sincerely appreciate the financial support.