Expression of selected MicroRNAs in pancreatic ductal adenocarcinoma in relation to tumor progression and patient’s outcome.

Background : pancreatic ductal adenocarcinoma (PDAC) remains a disease with extremely poor prognosis and limited effective available treatment. Differential expression of miRNAs isolated from tumor tissue has been proposed as a marker for tumor diagnosis, progression and prognosis. Nevertheless, prognostic value of miRNAs expression in PDACs for patient outcome still remains unclear. Methods : expression of 7 selected miRNAs, isolated from FFPE samples of 54 PDAC patients, was quantified using RT-qPCR. The relationship of miRNA expression levels with tumor histology, clinicopathological characteristics, patient overall survival (OS) and progress-free survival (PFS), was subsequently evaluated. Results : overexpression of miR-21, miR-155 and miR-210 was observed in PDACs (up to 72.62-fold, 232.36 and 181.38-fold respectively), in comparison with non-neoplastic tissues. On the other hand, miR-96 and miR-217 were significantly downregulated in PDACs (up to one hundred times). No differences were, however, noticed between cancer and normal tissues for the expression levels of miR-148a and miR-196a.

overall survival of 5-7% [4][5][6][7][8]. Despite dramatic progress in the management of malignancies, the outcome of PDAC patients failed to show significant improvement and by 2030, PDAC is estimated to be the second leading cause of cancer-related death [9]. Therefore, there is an intensive ongoing search for biomarkers permitting tumor detection, characterization of cancer progression and prediction of patient survival.
Dysregulation in miRNA expression profiles has been detected in a wide variety of neoplastic diseases [12][13][14]. They have been theorized to act as oncogenes and tumor suppressors, with aberrant miRNA expression being already presented in neoplastic precursor lesions [15][16][17]. MiRNAs can be isolated from plasma, tissue samples and excretions, while maintaining sample integrity due to their stability [18][19][20]. Moreover, miRNAs are preserved after formalin fixation and can be isolated from paraffin embedded tissue, yielding similar results to fresh material [21]. increased plasma levels of miR-210 were linked to better patient survival [28] In this study, the expression of 7 miRNAs (miR-21, miR-96 miR-148a, miR-155, miR-196a and miR-210 and miR-217), described to be dysregulated in PDAC, was analyzed. Three of the selected miRNAs, miR-21, miR-155 and miR-217 were described to be differentially expressed in relationship with tumor progression [23,25,29]. MiR-21, miR-155, miR-196a and miR-210 were selected, due to them having been proposed as prognostic markers [22,27,30]. The role of miR-96 as well as miR-148a and miR-217 expression in patient prognosis has not been analyzed extensively yet. The aim of this work was evaluation of the relationship of miRNA expression levels with tumor morphology, progression and patients´ survival.

Morphological analysis of tumors
Microscopic patterns of PDAC were classified into tubular, cribriform, solid-trabecular, mucinous, clear cell, dissociative and signet ring; quantified in increments of 5%, taking into account all the available slides with tumor. Tumor growth patterns are illustrated in Fig. 1. Nuclear atypia was semiquantitatively graded as low, medium or high. Mitotic count for each tumor was also performed in 10 HPF (Olympus microscope BX43 and objective Olympus Plan 40x/0.65).

MicroRNAs isolation and reverse transcription.
Three 6µm thick unstained sections from selected FFPE blocks where the tumor occupied at least two thirds, were procured for RNA extraction, using the miRNeasy FFPE kit (Qiagen), following the margin was negative (R0) in 40 and positive (R1) in 14 patients (Table 3).

Correlation between tumor growth patterns
Analysis of the relations of particular histomorphological patterns of PDACs using Spearman's correlation showed negative correlation of tubular tumor pattern (being present in tumors with lower grade) with solid trabecular (p<0.0001) and dissociative (p=0.0003) growth patterns, high nuclear atypia (p<0.0001) and mitotic count (p=0.0016). Hallmarks of high tumor grade, which are solid trabecular and dissociative growth patterns, were, on the other hand, associated with higher degrees of nuclear atypia (p=0.0002, p<0.001). High mitotic count in PDAC was also related with nuclear atypia (p=0.0004).

Survival
Data on progression-free survival (PFS) were available in 42 patients, with a median of 13 months.
Overall survival (OS) of the entire group ranged between 1-81 months, with a median of 19 months.
Seven patients have shown no recurrence of the disease and have survived for 20-81 months, still being alive at the end of the follow-up period ( Table 3).

Relationship of tumor parameters and patient survival.
Evaluation of prognostic significance of tumor stage was limited by the number of the patients. No difference in prognosis could be demonstrated between grade 2 and grade 3 PDACs (p>0.05). The small number of grade 1 tumors (n=4) in our group prevented us from further characterizing deviation in survival according to this parameter. Positive resection margin was associated with shorter PFS (p=0.005). Vascular invasion was significantly correlated with poor patient OS (p=0.036).
No such relationship was apparent in perineural invasion and lymph node status. Microscopic tumor growth patterns, including tubular, cribriform, solid trabecular and dissociative, were not associated with patient prognosis. The strongest correlation of OS and PFS was with tumor mitotic count (p=0.093 and p=0.063). We identified a cut-off point of 3 mitoses on 10 HPF 40x to distinguish between patients with poor and good prognosis.

Abnormal miRNA expression in pancreatic cancers.
We observed significant overexpression of miR-21, miR-155 and miR-210 (up to 72.62-fold, 232.36 and 181.38-fold correspondently; p<0.01, Expression of the miR-217 often was inhibited, up to one hundred times and even was not detected in 15 PDAC samples (Table 4). However, we did not find any significant differences between cancer and normal tissues for the expression levels of miR-148a and miR-196a (p>0.05, Table 4).

MiRNAs expression and clinico-pathological characteristics
Comparison of miRNA expression with clinico-pathological characteristics of patients disclosed positive correlation of miR-210 expression with patient age (ρ=0.35, p=0.0094; Table 6). Expression levels of all 7 miRNAs, failed to demonstrate significant correlation with other parameters, such as tumor stage, grade, nodal involvement, perineural and vascular invasion ( Table 6).

MiRNA expression and patient's survival.
Analysis of a prognostic role of expression of tested miRNAs in PDAC did not discover any significant evidence for OS (p>0.05, Table 8). Correlation between miRNA levels and duration of PFS was also statistically insignificant for all seven selected miRNAs (p>0.05, Table 9).
Decrease of miR-148a levels has been consistently identified in PDAC tissue samples, across several studies [19,30,42]. In contrast with the cited literature, we found slightly decreased miR-148a mean level in tumors (-1.63-fold), but insignificantly in comparison with normal tissues (p > 0.05, Table 4).
The miR-155 overexpression in PDACs and pancreatic cancer cell lines, measured by microarray, ranged from 1.8 to 2.9-fold in different studies [20,30,35,43]. On the other hand, Zhang et al.
Abnormalities in miR-196a expression have been described in pancreatic cancer as well as  (Table 4). On the other hand, we did not find significant differences in miR-196a expression between cancer and normal tissues (mean was 0.913-fold, p > 0.05; Table 4).
Besides identifying abnormalities in the expression of single miRNAs, we have also discovered a positive correlation between high expression levels of the three onco-miRNAs: miR-21, miR-155a and miR-210 (Table 5). Acting together, these miRNAs may promote cancer development and progression [48][49][50][51]. Moreover, positive correlation was detected for downregulation of tumor suppressing miRNAs miR-148a and miR-217 ( Table 5). Both of them inhibit cell proliferation [52,53], therefore it may be necessary to deactivate them in tumors for successful cancerogenesis. Additionally, positive correlation was detected for downregulation of tumor suppressing miRNAs miR-96 and miR-196a (Table 5). Both of them inhibit cell proliferation [39,40,54], but miR-196a acts quite opposite, inhibiting apoptosis [55], promoting cell proliferation and migration [48]. So, downregulation of miR-96 and upregulation of miR-196a looks like a necessary condition for tumor survival.
Thus, we observed that selected miRNAs were abnormally up-or downregulated in pancreatic cancers. Five of seven selected miRNAs demonstrated significant differences of expression levels in tumors in comparison with adjacent normal tissues (Table 4). Therefore, differential miRNAs expression may be a very sensitive tool for the pancreatic cancer diagnostics.

Correlation of clinicopathological features of tumors with microRNA expression
In surgical resection specimens, the relationship of abnormal miRNA expression with tumor morphology and progression has been investigated less frequently, compared with miRNA diagnostic In our cases, we haven't detected significant correlation of miRNA expression with tumor progression, grade, perineural, vascular invasion and lymph node status. Positive association was, however, discovered between patient age and miR-210 levels ( Table 6). This finding could be in part related to the proposed role of miR-210 in cellular senescence [61]. Further investigation is warranted to analyze the potential of miRNA signatures in predicting the extent of tumor progression. Additionally, correlation of miRNA expression PDAC resection specimens and plasma samples in prospective studies could be beneficial for further understanding of the promise of miRNAs as predictors of tumor progression, therapy and prognosis.
Prognostic role of miRNAs expression profiles for PDAC patient's overall and progression-free survival.
In spite of significant progress in anticancer therapy in the recent decades, no breakthroughs in PDAC treatment have materialized. The only clinically available biomarker capable of assessing the prognosis of PDAC patients is CA19-9, yet its utility is limited by non-specific positivity and false negativity in multiple neoplastic and non-neoplastic diseases [62]. There is growing evidence that miRNA expression profiles have a potential to provide tumor-specific prognostic information. Several recent works have reported associations between microRNA expression and overall survival in PDAC patients [22,57,63,64]. In our study we observed no correlation between expression of all selected miRNAs and OS or PFS (Tables 8 and 9).
Among the dysregulated miRNAs in PDAC, miR-21 has been the most widely studied potential prognostic factor. Abnormal expression of miR-21 as a marker of poor patient outcome has also been demonstrated in many other malignancies, including melanoma, glioma, carcinomas of breast, colorectal, lung, stomach, prostate, hepatocellular carcinoma and renal cell carcinoma [65]. In The role of miR-96 expression levels in patient prognosis has not been analyzed extensively. Li et al. showed that downregulated expression of miR-96-5p was associated with a decreased overall survival in patients with PDAC [72]. MiR-96 was also associated with poor overall survival in liver and colorectal cancer [73,74]. On the other hand, Cai et al. did not demonstrate any significant correlation between the expression level of miR-96 and OS in the lung cancer [75].
The data utilizing miR-148a as a prognostic marker for PDAC are inconsistent, as well. Schultz et al.
identified low miR-148a expression as a predictor of short OS [60]. On the other hand, in a group of 78 patients, miR-148 expression levels where not statistically significant with regards to overall survival [47]. Also, miR-148a overexpression did not have any effect on cell proliferation and cell chemosensitivity in four PDAC cell lines [76]. On the other hand, patients showing elevated expression of 6 genes targeted by miR-148a-5p had a significantly poorer overall survival as well as a shorter DFS [77].  [47,67].
In our study, we could not confirm significant prognostic value for OS and PFS for our panel of miRNAs (Tables 8 and 9). Therefore, the data about prognostic role of miRNAs expression in PDACs are insufficient and sometimes controversial. These discrepancies may be ascribed to several factors.
Most of the studies have been performed on a wide variety of PDAC samples types, including frozen and FFPE tumor samples, plasma or blood serum. Tissue samples were analyzed with or without microdissection, which could affect the results by excluding miRNAs from tumor-associated fibroblasts. Detection of variations in miRNA expression was performed using several methods, including RT-PCR, FISH and microarrays, each requiring different protocols for sample preparation and varying greatly in their sensitivity and selectivity. The number of performed studies is low; as such the size and ethnicity of the PDAC patient cohort may also be regarded as a significant source of variability. Thus, further studies are needed to confirm whether miRNAs could serve as prognostic and predictive biomarkers for pancreatic cancer.

Conclusions
In conclusion, this study detected statistically significant differences in miRNA expression in PDAC, compared to normal pancreatic tissue. Thus, even though it has not been clinically implemented yet, detection of differential miRNAs expression may be a sensitive tool for the pancreatic cancer diagnostics. On the other hand, changes in miRNA expression in our group of patient could not be correlated with key tumor parameters, such as grade, stage, and lymph node status. We have, however, found significant correlation with histologic tumor growth patterns and miRNA expression, making this the first study to analyze this aspect of PDAC. Finally, our work could not confirm statistically significant relationship between miRNA expression and patient prognosis. Therefore, further investigations analyzing the potential of miRNAs to predict the extent of PDAC tumor progression and the possible prognostic role of miRNAs are needed.

Funding
This work was supported by the Charles University research program PROGRES Q 28 (Oncology).

Availability of Data
Data files, including raw CT values or fold change tables are available on request, please contact the corresponding author.

Competing interests
The authors declare that they have no competing interests.        Table 8. Overall survival and miRNAs expression levels in PDAC patients.