Whole transcriptome expression profiling of pancreatic cancers patients with T4 stage but diametrically opposing survival outcomes

Pancreatic cancer is a highly malignant tumor in the digestive tract, with its incidence increasing year by year. Due to the insidious onset and rapid progression, most patients have already advanced disease at the time of the initial diagnosis, and the overall 5-year survival rate was only 11%. Tumors in T4 stage often have the invasion of key upperabdominal vessels, mainly referring to artery-involved pancreatic cancer with a worse prognosis. Resection of pancreatic ductal adenocarcinoma (PDAC) with T4 stage is more challenging and usually needs advanced surgical techniques including treatment of the involved artery including sub-adventitial divestment technique (SDT) and arterial resection with reconstruction, and periarterial dissection such as the Heidelberg Triangle Operation. SDT means a technique to separate the tumor from the artery by dissecting into the plane between the external elastic lamina (EEL) and the tunic media when the tumor was not invaded to EEL of the involved artery. Our previous study revealed that SDT procedure appears to be relatively safe, without excessive mortality, as compared to arterial resection with reconstruction. However, survival times can vary widely among different patients. So a strategy to select the T4 patients who may benefit from the surgery with SDT procedure is urgently needed. Our previous RNA-seq dataset has identified a series of differentially expressed circRNAs including circNEIL3 and circRHOBTB3 in PDACs. These results suggested that circRNAs might be prominent prognostic markers for PDAC. So to explore the prognostic value of circRNAs in pancreatic cancer and seek a useful biomarker for distinguishing specified population with a good prognosis from the whole T4 patients, we further performed whole transcriptome sequencing including miRNAs, mRNAs, lncRNAs and circRNAs in ten PDAC cancer samples with T4 stage but diametrically opposing survival outcomes. They all underwent radical surgery and received the assessments of pre- and post-operative contrast-enhanced computed tomography, five of these patients developed newly discovered metastasis soon after surgery and died within 4-9 months. In contrast, the other five patients had no recurrence or metastasis and had long-term survival for more than 39 months. We systematically identified differentially expressed circRNAs in our previous dataset and in patients with T4 stage through combining three popular circRNAs detection algorithms (CIRI2 v2.0.6, CIRCexplorer2 v2.3.8, and find_circ v1.2) and two differential analysis methods (DEGseq and DEseq). Results were validated by using quantitative RT-PCR. circRNA alternative splicing events and highly expressed circRNAs with high circular ratio were also identified by CircSplice and CIRCexplorer3-CLEAR pipeline. Overall design: RNA was extracted from frozen pancreatic tissues for RNA-seq analysis. Total RNAs were isolated from 10 patients with T4 pathological stage using TRIzol Reagent (Invitrogen, CA, USA). After passing the assessment of RNA purity and RNA integrity (RIN), two different RNA sequencing strategies (ribo- RNA-seq, poly(A)-/ribo- plus RNase R-treated RNA-seq) were applied to profile circRNAs and mRNAs/lncRNAs. Small RNA library was also prepared according to the prescribed procedure and standards of the Illumina Sample Preparation Protocol. Poly(A)-/ribo- plus RNase R-treated RNA-seq dataset was used to detect differentially expressed circRNAs by three popular circRNAs detection algorithms and detect circRNA alternative splicing events by CircSplice algorithm. Ribo- RNA-seq dataset was used to identify highly expressed circRNAs with high circular ratio by CIRCexplorer3-CLEAR pipeline. Ribo- RNA-seq dataset can also be used to detect the expression profile of mRNAs/lncRNAs by Hisat2 (version 2.1.0) and featureCounts function from the Subread package (version 1.6.3).