Roman‑Canal et al. J Transl Med (2019) 17:208 https://doi.org/10.1186/s12967-019-1954-8 Journal of Translational Medicine RESEARCH Open Access EV‑associated miRNAs from peritoneal lavage as potential diagnostic biomarkers in colorectal cancer Berta Roman‑Canal1,2,3, Jordi Tarragona1, Cristian Pablo Moiola1,4, Sònia Gatius1, Sarah Bonnin5, Maria Ruiz‑Miró1, José Enrique Sierra6, Maria Rufas6, Esperanza González7, José M. Porcel8, Antonio Gil‑Moreno4,9, Juan M. Falcón‑Pérez7,10, Julia Ponomarenko5,11, Xavier Matias‑Guiu1,2,12* and Eva Colas4* Abstract Background: Colorectal cancer (CRC) is the third leading cause of cancer-related mortality worldwide. Current sys‑ tematic methods for diagnosing have inherent limitations so development of a minimally-invasive diagnosis, based on the identification of sensitive biomarkers in liquid biopsies could therefore facilitate screening among population at risk. Methods: In this study, we aim to develop a novel approach to identify highly sensitive and specific biomarkers by investigating the use of extracellular vesicles (EVs) isolated from the peritoneal lavage as a source of potential miRNA diagnostic biomarkers. We isolated EVs by ultracentrifugation from 25 ascitic fluids and 25 peritoneal lavages from non-cancer and CRC patients, respectively. Analysis of the expression of EV-associated miRNAs was performed using Taqman OpenArray technology through which we could detect 371 miRNAs. Results: 210 miRNAs were significantly dysregulated (adjusted p value < 0.05 and abs(logFC) ≥ 1). The top-10 miR‑ NAs, which had the AUC value higher than 0.95, were miRNA-199b-5p, miRNA-150-5p, miRNA-29c-5p, miRNA-218-5p, miRNA-99a-3p, miRNA-383-5p, miRNA-199a-3p, miRNA-193a-5p, miRNA-10b-5p and miRNA-181c-5p. Conclusions: This finding opens the avenue to the use of EV-associated miRNA of peritoneal lavages as an untapped source of biomarkers for CRC. Keywords: Colorectal cancer, Biomarkers, Diagnostic, miRNAs, Ascitic fluid, Peritoneal lavage, Liquid biopsy, Extracellular vesicles, Colon cancer Background women [1]. The overall survival of colorectal cancer is Colorectal cancer (CRC) is the third most common 65%, but this is highly dependent upon the disease stage type of malignant tumor and the third leading cause at diagnosis, and ranges from a 90% of 5-year survival of cancer-related mortality worldwide among men and rate for cancers detected at the localized stage (40% of the cases) and 70% for regional (35% of the cases) to 15% for distant metastatic cancer (20% of the cases) *Correspondence: fjmatiasguiu.lleida.ics@gencat.cat; eva.colas@vhir.org † Xavier Matias-Guiu and Eva Colas are senior co-authors [2]. Current systematic methods for diagnosis, such as 4 Biomedical Research Group in Gynecology, Vall Hebron Research fecal occult blood test and flexible sigmoidoscopy, help Institute (VHIR), Universitat Autònoma de Barcelona, CIBERONC, Pg. Vall to reduce mortality by removing precursor lesions and Hebron 119‑129, 08035 Barcelona, Spain12 Oncologic Pathology Group, Department of Medicine UdL, Biomedical Research Institute of Lleida making diagnosis at an earlier stage. However, these (IrbLleida), Av. Rovira Roure 80, 25198 Lleida, Spain techniques have inherent limitations, such as low sen- Full list of author information is available at the end of the article sitivity and invasiveness for patients, respectively; and © The Author(s) 2019. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creat​iveco​mmons​.org/licen​ses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creat​iveco​mmons​.org/ publi​cdoma​in/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Roman‑Canal et al. J Transl Med (2019) 17:208 Page 2 of 14 the burden of disease and mortality is still high [3]. Ethics Committee of the hospital approved the study Serum tumor markers CA19-9 and CEA have been used and all the participating patients provided a signed for detection of many types of cancer, but their sensitiv- informed consent. Ascitic fluids and peritoneal lav- ity for the detection of CRC is low [4]. Therefore, devel- ages were extracted from a cohort of 50 patients, cor- opment of a minimally-invasive diagnosis, based on the responding to 25 control patients with decompensated identification of sensitive biomarkers in liquid biopsies cirrhosis, and 25 patients with CRC who underwent could therefore facilitate screening among population curative surgery. In control patients, the collection of at risk of CRC, impact on early detection, and thus, ascitic fluid was aspirated using 18 or 21G needles (for decrease CRC-related mortality. diagnostic paracentesis) or an over-the-needle catheter MiRNAs are a highly conserved family of endogenous device (for therapeutic paracentesis). The procedure non-coding and single-stranded RNAs that are 19–24 was performed under sterile conditions, the site of nee- nucleotides in length [5]. Generally, miRNAs negatively dle insertion was selected by ultrasound guidance, and regulate gene expression via binding to the 3′-untrans- skin and parietal peritoneum were previously anesthe- lated region (3′-UTR) of their target double-stranded tized with 2% mepivacine. A total of 100 mL of ascitic mRNA that results in transcriptional repression or fluid was gently aspirated, collected into a 50 mL tube mRNA degradation upon dicer complex [6]. miRNAs and stored at − 80 °C. In CRC patients, the collection have been implicated in development and progression of peritoneal lavage was performed before the surgery, of CRC by functioning as oncogenes and tumor sup- once the abdominal cavity has been opened and prior pressors [7]. Recent studies demonstrated that miRNAs to any manipulation of the colon. A total of 100 mL of are secreted from various cells, including cancer cells, physiological saline were instilled into the abdominal into bodily fluids such as blood, urine, breast milk, and cavity with a 50 mL syringe, mobilizing patients for the saliva, either as free miRNAs or via extracellular vesi- correct distribution of saline, which was then extracted cles (EVs) [4]. with a 50 mL syringe connected to a 14-gauge aspira- EVs are 20–200 nm membrane vesicles released by tion needle. The peritoneal lavage was gently aspirated. either directly from plasma membranes, or from intra- A volume ranging from 50 to 100 mL was collected and cellular multivesicular bodies by their fusion with the stored at − 80 °C. The clinical features of each patient cell membrane. Their function is to mediate intercel- are listed in Additional file 1: Table S1. lular communication, influencing the recipient cell function. Importantly, EVs have awakened the interest of the scientific community as a source of biomarkers, EVs isolation mainly because they carry a broad range of bioactive EVs were isolated with a differential centrifugation material (proteins, metabolites, RNA, miRNA, etc.) and method as previously described [9] with slight modi- this material is well-protected owing to the EVs lipid fications. Briefly, ascitic fluids and peritoneal lavages bilayer membrane, even if EVs are extracted from circu- were centrifuged at 300×g for 10 min, followed by a lating or proximal body fluids [8]. centrifugation at 2500×g for 20 min and a centrifuga- Herein, we investigated the use of EVs isolated from tion of 10,000g for 30 min (Thermo Scientific Heraeus the peritoneal lavage, a proximal fluid in CRC patients, MultifugeX3R Centrifuge (FiberLite rotor F15-8x-50c)). as a source of potential diagnostic biomarkers. To do The supernatant was then filtered through 0.22 µm fil- so, we conducted miRNA-profiling of EVs isolated from ters (Merck Millipore) and the sample obtained was peritoneal lavages of surgical CRC patients and ascitic transferred to ultracentrifuge tubes (Beckman Coul- fluids of non-cancer patients by using the TaqMan ter) and filled with PBS to perform two consecutive OpenArray Human MicroRNA Panel. We unveiled the ultracentrifugation steps at 100,000g for 2 h each on a most relevant individual miRNAs for diagnosing CRC Thermo Scientific Sorvall WX UltraSeries Centrifuge and characterized the biological and molecular land- with an AH-629 rotor. The pellet containing the EVs scape of the CRC milieu. The study was conceived as a was resuspended in 50 µL of PBS. From those, 5 µL proof of concept investigation to demonstrate the feasi- were isolated for nanoparticle tracking analysis (NTA) bility of peritoneal lavage as a source of EV-associated and quantification, and the rest was frozen at − 80 °C miRNAs in patients with CRC. with 500 µL of Qiazol for RNA extraction. Methods Nanoparticle tracking analysis Patients and ascitic fluid and peritoneal lavages collection Size and number of EVs was determined using a Participants in the study attended to the Hospital Arnau Nanosight LM10 instrument equipped with a 405 nm de Vilanova in Lleida, Spain. The Clinical Research laser and a Hamamatsu C11440 ORCA-Flash 2.8 camera Roman‑Canal et al. J Transl Med (2019) 17:208 Page 3 of 14 (Hamamatsu) with Nanoparticle Tracking Analysis interquartile range across samples. Differential expres- (NTA, Malvern Instruments, UK). Each sample was sion analysis was carried out with an empirical Bayes diluted appropriately with Milli-Q water (Milli-Q Synthe- approach on linear models, using the limma (version sis, Merck Millipore, Massachusetts, USA) to give counts 3.36) R Package [13]. Results were corrected for multiple in the linear range of the instrument. The particles in the testing using the False Discovery Rate (FDR) [14]. laser beam undergo Brownian motion, and a video was recorded for 60 s in triplicate. Analysis was performed Development of predictors following manufacturer’s instructions and data were ana- For predictive analysis, the whole patient cohort was ran- lyzed using the version 2.3 of the NTA-software. domly divided into training and validation sets with the 3:2 ratio. Calculated (with the limma R Package) rela- Total RNA extraction and OpenArray analysis tive miRNA expression values were used as input vari- Total RNA, including miRNAs and other RNAs, was iso- ables to a logistic regression model between groups. Each lated from the EVs samples using the miRNeasy Mini- miRNA (adjusted p-value < 0.05) was fitted in the logistic Kit (Qiagen) according to manufacturer’s protocol. RNA regression model to differentiate the CRC and the control from EVs was eluted with 30 µL of Nuclease-free water patients groups in the training set and its classification (Ambion). MiRNA expression was performed using ability was evaluated using the AUC (area under the ROC TaqMan OpenArray Human MicroRNA Panel, Quant- curve), accuracy, sensitivity, and specificity values on the Studio 12 K Flex (Catalog number: 4470187, Thermo validation set. The procedure from division into train- Fisher Scientific), a fixed-content panel containing 754 ing and validation sets and fitting the logistic model was well-characterized human miRNA sequences from the repeated 500 times and statistics were collected. Sanger miRBase v14 and according to the manufacturer’s miRNA target genes prediction and bioinformatics analysis on 2 µL RNA using Megaplex™ Primer Pools A and B and instructions. Reverse transcription (RT) was performed miRNAs target genes were obtained using the Predic- the supporting ­TaqMan® MicroRNA Reverse Transcrip- tive Target Module of miRWalk2.0 online software [15] tion Kit as follows: 15 min at 42 °C and 5 min at 85 °C. (https​://goo.gl/ajG9j​a), considering the following param- eters: 3´UTR localization, miRNA seed start at position to real-time PCR analysis using Megaplex™ PreAmp Then, 5 µL of the resulting cDNA was preamplified prior 1 and minimum 7 bp seed length. To improve the accu- Pools and the T ­ aqMan® PreAmp Master Mix using the racy of target gene prediction, only those transcripts that following conditions: one single step at 95 °C during were predicted in at least 8 out of the 12 databases were 5 min, 20 cycles of a two-steps program (3 s, 95 °C and considered (miRWalk, miRanda, MicroT4, miRDB, miR- 30 s, 60 °C) followed by a single cycle of 10 min at 99 °C Map, miRBridge, miRNAMap, PICTAR2, RNA22, PITA, to inactivate the enzyme. The preamplified products TargetScan, and RNAhybrid). Gene Ontology (GO) functional analysis were used to analyze the potential ­ aqMan® ­OpenArray® Real-Time PCR Master were diluted 1:20 in 0.1× TE buffer pH8.0, and mixed in functions of the predicted target genes, using the online Mix in the 384-well ­OpenArray® Sample Loading Plate. 1:1 with T Panther software [16] (http://www.panth​erdb.org/). Bio- ­TaqMan® ­OpenArray® MicroRNA Panels were automati- logical process (BP) and molecular function (MF) GO cally loaded using the AccuFill™ System. terms were analyzed and plotted. Preprocessing and differential expression analysis Results All bioinformatics analysis was performed with the Bio- We analyzed the miRNA profile of EVs isolated from Conductor (version 3.7) [10] project in the R statistical the ascitic fluid of 25 control individuals and peritoneal environment (version 3.5.0) [11]. For the data preproc- lavage of 25 CRC patients. Additional file 2: Figure S1 essing, the HTqPCR (version 1.34) R package [12] was illustrates the workflow that was followed in this study. used. Probes that had a “Cycle threshold” (Ct) value of 40 The quality of EVs isolated from the ascitic fluids and in all samples were removed. Further samples in which peritoneal lavages was measured by size distribution and more than 80% of the probes had a Ct value above 40 concentration by Nanoparticle Tracking Analysis, dem- were retained. To assure comparability across samples, onstrating that we analyzed a population mostly enriched the Ct values were delta normalized. The average Ct val- in small EVs but also containing a low representation ues of the probes hsa − miR − 150 − 5p, hsa − let − 7g-5p, of larger vesicles (Additional file 3: Figure S2). MiRNAs hsa − miR − 598 − 3p, and hsa − miR − 361 − 3p were were extracted from EVs for a systematic miRNA expres- used for normalization. These probes had the Ct values sion analysis using the Taqman OpenArray technology, of 40 in a maximum of three samples, and the lowest through which we detected 371 out of the 754 miRNAs (49.2%) present in the OpenArray. Probes that had the Roman‑Canal et al. J Transl Med (2019) 17:208 Page 4 of 14 Ct value of 40 in all samples and samples in which more To further understand the milieu generated by CRC than 80% of the probes had the Ct value above 40 were EVs, we performed a bioinformatics study to first unveil removed, resulting in 355 miRNAs from 22 control and the predicted transcripts that are regulated by all the 19 CRC patients analyzed for the differential expression differential miRNAs, and then assess the biological pro- analysis (Table 1). cesses and molecular functions in which they participate. The differential expression analysis between can- A total of 9358 transcripts were found to be regulated by cer and control cases yielded a list of 210 miRNAs that the 210 miRNA differentially expressed. Figure 2 shows were significantly dysregulated (adj. p-value < 0.05 and the number and most frequently regulated transcripts logFC lower or higher than 1). Among those, 207 miRNA predicted for each dysregulated miRNA. To comprehen- were found to be downregulated and 3 miRNA were sively integrate the properties of all target transcripts, upregulated in CRC patients. To evaluate whether these they were studied using Gene Ontology (GO). The most miRNAs can be used as diagnostics biomarkers, we per- enriched biological processes in CRC EVs were metabolic formed a predictive analysis using the logistic modeling. processes (24.3%), mostly including biosynthetic process, Ten miRNAs demonstrated predictive performance at the organic substance metabolomic process and cellular met- AUC values higher than 0.95: miRNA-199b-5p, miRNA- abolic process; biological regulation (22.5%); cellular pro- 150-5p, miRNA-29c-5p, miRNA-218-5p, miRNA-99a-3p, cesses (10.7%), signal transduction, cellular component miRNA-383-5p, miRNA-199a-3p, miRNA-193a-5p, organization and cellular metabolic process; and cellular miRNA-10b-5p and miRNA-181c-5p (Table 2; Fig. 1). component organization or biogenesis, including cellu- All those miRNAs were downregulated from 3.52 to lar component organization (Fig. 3a). In relation to the 12.82 in the ­log2 scale with adjusted p-value lower than most altered molecular functions in CRC EVs, the Gene 1.56E−05, except miRNA-150-5p which was upregulated Ontology (GO) analysis revealed that many targeted (adjusted p-value 3.41E−04). In Table 3, studies report- transcripts were found to be involved in binding (37.8%), ing an association between each of these top-10 miRNAs including protein binding and organic cyclic compound and CRC are described based on a search of Pubmed for binding; and in catalytic activity (31.2%), including cata- each miRNA and the word “colorectal cancer”. Although lytic activity, and protein and hydrolase activity (Fig. 3b). there are some controversies among the different stud- ies, most of the miRNA dysregulations observed in our Discussion study are concordant with the observations reported by In this study we investigated, for the first time, the other authors, either in tissue, plasma or stool samples. miRNA content of EVs isolated from peritoneal lavages MiRNA-199b-5p, miRNA-29c-5p, and miRNA-99a-3p and ascitic liquid of CRC and control patients, respec- have never been reported previously in association to tively. Our study shows that EV-associated miRNAs can CRC. be consistently extracted from peritoneal lavages and ascitic liquids and that miRNA expression profiles can indicate and represent the status of CRC patients. The Table 1 Clinicopathological characteristics of patients EV-associated miRNA were analyzed by Taqman Ope- nArray technology and the differential expression analy- Colon Cancer Control sis yielded a list of 210 miRNAs that were significantly Age dysregulated in CRC patients, being downregulated the Median 74 65 98.57% of the altered miRNAs. Minimum 50 52 The finding that miRNAs are dysregulated in CRC Maximum 88 90 patients is known, as many studies have reported this Gender previously, mostly in tissue specimens [17] but also Female 12 4 in different body fluids. In CRC, most of the studies Male 7 18 use plasma or serum as it is the most common, easy- Pathology to-handle, accessible liquid biopsy. The first report Colon cancer 19 – detected 69 miRNAs in serum of CRC patients but not ADC low grade 15 – in serum of normal controls [18]. Since then, several ADC other types 4 – studies have identified miRNA upregulation or down- Hepatic cirrhosis – 20 regulation in plasma or serum samples [17] including Others – 2 studies that have focus on the search of biomarkers Clinical characteristics of the final cohort of patients included in the study after in miRNAs dysregulated in the vesicular fraction of data normalization the serum or plasma of CRC patients. Hiroko Ogata- ADC adenocarcinoma Kawata et al. [4] analyzed the EV-associated miRNA Roman‑Canal et al. J Transl Med (2019) 17:208 Page 5 of 14 Table 2 miRNA transcripts displaying a significant differential expression in patients with CRC compared to control patients miRNA LogFC p-value Adj. p-value AUC​ AUC 95% AUC 95% Accuracy Sensitivity Specificity CI_lower CI_upper hsa-miR-199b-5p_478486_mir − 12.82 2.59E−08 9.85E−07 1 1 1 0.967 0.968 0.964 hsa-miR-150-5p_477918_mir 2.58 7.10E−05 3.41E−04 0.978 0.959 0.996 0.919 0.936 0.899 hsa-miR-29c-5p_478005_mir − 2.93 2.78E−08 9.85E−07 0.973 0.954 0.991 0.943 0.943 0.944 hsa-miR-218-5p_477977_mir − 8.16 6.51E−07 8.25E−06 0.97 0.945 0.995 0.913 0.905 0.921 hsa-miR-99a-3p_479224_mir − 4.89 1.51E−08 8.95E−07 0.97 0.95 0.99 0.94 0.976 0.9 hsa-miR-383-5p_478079_mir − 8.33 3.55E−15 1.26E−12 0.968 0.952 0.985 0.939 0.94 0.938 hsa-miR-199a-3p_477961_mir − 6.16 2.84E−09 2.65E−07 0.968 0.942 0.994 0.905 0.92 0.887 hsa-miR-193a-5p_477954_mir − 3.62 1.32E−06 1.56E−05 0.962 0.932 0.991 0.873 0.852 0.897 hsa-miR-10b-5p_478494_mir − 2.79 2.58E−07 4.17E−06 0.957 0.93 0.983 0.871 0.875 0.866 hsa-miR-181c-5p_477934_mir − 3.52 1.23E−05 8.74E−05 0.952 0.929 0.974 0.833 0.859 0.803 hsa-miR-708-5p_478197_mir − 6.45 9.72E−08 2.29E−06 0.946 0.917 0.975 0.877 0.834 0.926 hsa-miR-125b-5p_477885_mir − 2.35 9.27E−07 1.13E−05 0.946 0.918 0.974 0.885 0.884 0.885 hsa-miR-140-5p_477909_mir − 5.82 4.59E−05 2.39E−04 0.943 0.919 0.968 0.817 0.825 0.807 hsa-miR-451a_478107_mir − 8.35 1.34E−07 2.64E−06 0.942 0.913 0.972 0.843 0.881 0.8 hsa-miR-148b-3p_477824_mir − 3.05 1.42E−07 2.66E−06 0.942 0.916 0.968 0.834 0.853 0.813 hsa-miR-130a-3p_477851_mir − 2.62 1.85E−06 2.00E−05 0.94 0.909 0.972 0.861 0.884 0.835 hsa-miR-214-3p_477974_mir − 7.59 1.19E−07 2.49E−06 0.937 0.901 0.972 0.896 0.94 0.846 hsa-miR-10a-5p_479241_mir − 2.19 1.44E−02 2.82E−02 0.937 0.906 0.969 0.897 0.904 0.889 hsa-miR-497-5p_478138_mir − 3.68 2.13E−04 8.08E−04 0.936 0.911 0.961 0.814 0.829 0.797 hsa-miR-143-3p_477912_mir − 3.15 1.58E−06 1.81E−05 0.936 0.906 0.965 0.86 0.868 0.85 hsa-miR-20a-5p_478586_mir − 2.79 5.14E−08 1.30E−06 0.933 0.901 0.964 0.877 0.906 0.843 hsa-miR-29c-3p_479229_mir − 3.55 2.14E−04 8.08E−04 0.931 0.897 0.965 0.86 0.84 0.883 hsa-miR-17-5p_478447_mir − 3.68 4.83E−05 2.41E−04 0.93 0.893 0.966 0.874 0.898 0.847 hsa-miR-486-5p_478128_mir − 11.10 2.21E−07 3.73E−06 0.929 0.899 0.958 0.853 0.818 0.893 hsa-miR-145-5p_477916_mir − 3.07 6.47E−06 5.34E−05 0.929 0.899 0.958 0.877 0.901 0.851 hsa-miR-214-5p_478768_mir − 8.13 2.33E−08 9.85E−07 0.923 0.885 0.96 0.877 0.906 0.844 hsa-miR-20b-5p_477804_mir − 11.65 3.73E−09 2.65E−07 0.921 0.887 0.956 0.883 0.879 0.887 hsa-miR-551b-3p_478159_mir − 9.71 2.07E−10 3.68E−08 0.919 0.885 0.953 0.852 0.906 0.79 hsa-miR-107_478254_mir − 4.70 1.35E−03 3.73E−03 0.917 0.883 0.951 0.919 0.938 0.898 hsa-miR-202-5p_478755_mir − 7.19 5.11E−08 1.30E−06 0.915 0.876 0.954 0.855 0.867 0.842 hsa-miR-93-5p_478210_mir − 2.84 3.88E−04 1.30E−03 0.915 0.875 0.954 0.86 0.872 0.847 hsa-miR-483-3p_478122_mir − 7.01 3.69E−09 2.65E−07 0.913 0.877 0.949 0.837 0.884 0.784 hsa-miR-652-3p_478189_mir − 2.05 2.63E−05 1.70E−04 0.913 0.882 0.945 0.825 0.831 0.818 hsa-miR-29b-3p_478369_mir − 3.59 3.71E−06 3.47E−05 0.911 0.877 0.945 0.836 0.834 0.838 hsa-miR-328-3p_478028_mir − 5.03 2.86E−04 1.01E−03 0.908 0.874 0.941 0.819 0.798 0.842 hsa-miR-25-3p_477994_mir − 2.37 3.44E−05 1.99E−04 0.908 0.87 0.946 0.865 0.899 0.827 hsa-miR-26a-5p_477995_mir − 2.30 1.99E−05 1.38E−04 0.904 0.872 0.937 0.796 0.799 0.791 hsa-miR-296-5p_477836_mir − 6.51 3.42E−05 1.99E−04 0.903 0.865 0.941 0.878 0.796 0.973 hsa-miR-144-3p_477913_mir − 5.05 3.53E−05 1.99E−04 0.903 0.867 0.938 0.827 0.85 0.801 hsa-miR-769-5p_478203_mir − 3.98 3.53E−05 1.99E−04 0.903 0.864 0.942 0.899 0.901 0.896 hsa-miR-181a-5p_477857_mir − 2.13 5.21E−06 4.52E−05 0.902 0.865 0.939 0.843 0.895 0.783 hsa-miR-29a-3p_478587_mir − 3.12 1.16E−03 3.38E−03 0.901 0.86 0.942 0.878 0.854 0.905 hsa-miR-152-3p_477921_mir − 3.98 2.17E−05 1.48E−04 0.9 0.866 0.934 0.804 0.791 0.819 hsa-miR-125b-1-3p_478665_mir − 9.13 1.93E−07 3.42E−06 0.895 0.86 0.931 0.86 0.868 0.851 hsa-miR-30a-3p_478273_mir − 1.66 9.38E−05 4.27E−04 0.891 0.854 0.928 0.812 0.793 0.834 hsa-miR-449b-5p_479528_mir − 10.10 2.30E−08 9.85E−07 0.889 0.846 0.932 0.88 0.908 0.847 hsa-miR-219a-5p_477980_mir − 6.58 5.10E−07 7.54E−06 0.889 0.852 0.926 0.84 0.854 0.824 hsa-miR-125a-5p_477884_mir − 1.40 1.48E−03 4.02E−03 0.888 0.846 0.93 0.827 0.84 0.813 Roman‑Canal et al. J Transl Med (2019) 17:208 Page 6 of 14 Table 2 (continued) miRNA LogFC p-value Adj. p-value AUC​ AUC 95% AUC 95% Accuracy Sensitivity Specificity CI_lower CI_upper hsa-miR-374b-3p_479421_mir − 4.64 2.15E−06 2.24E−05 0.887 0.845 0.929 0.8 0.791 0.81 hsa-miR-101-3p_477863_mir − 4.19 6.98E−05 3.39E−04 0.886 0.845 0.927 0.832 0.896 0.759 hsa-miR-452-5p_478109_mir − 3.88 2.89E−05 1.80E−04 0.886 0.852 0.92 0.751 0.737 0.766 hsa-miR-193a-3p_478306_mir − 2.46 2.37E−03 5.97E−03 0.884 0.841 0.928 0.869 0.899 0.835 hsa-miR-148a-3p_477814_mir − 1.48 1.24E−03 3.52E−03 0.884 0.845 0.923 0.822 0.818 0.826 hsa-miR-133a-3p_478511_mir − 6.89 1.67E−04 6.73E−04 0.883 0.843 0.922 0.827 0.853 0.797 hsa-miR-675-5p_478196_mir − 4.22 5.46E−06 4.61E−05 0.883 0.845 0.92 0.766 0.721 0.817 hsa-miR-34a-5p_478048_mir − 2.04 3.41E−05 1.99E−04 0.882 0.841 0.922 0.823 0.801 0.848 hsa-miR-582-5p_478166_mir − 8.33 3.37E−07 5.20E−06 0.881 0.842 0.92 0.84 0.767 0.923 hsa-miR-2110_477971_mir − 4.46 2.45E−06 2.48E−05 0.879 0.837 0.92 0.782 0.794 0.769 hsa-miR-185-5p_477939_mir − 2.69 1.09E−02 2.18E−02 0.879 0.836 0.921 0.828 0.872 0.777 hsa-miR-144-5p_477914_mir − 11.46 4.47E−08 1.30E−06 0.877 0.833 0.921 0.873 0.944 0.791 hsa-miR-199a-5p_478231_mir − 7.83 1.24E−04 5.31E−04 0.877 0.831 0.923 0.772 0.712 0.84 hsa-miR-361-5p_478056_mir − 1.48 9.56E−04 2.83E−03 0.877 0.837 0.918 0.821 0.86 0.778 hsa-miR-195-5p_477957_mir − 2.80 2.19E−04 8.17E−04 0.875 0.832 0.919 0.835 0.822 0.849 hsa-miR-136-5p_478307_mir − 8.01 5.88E−07 8.03E−06 0.873 0.834 0.913 0.828 0.821 0.836 hsa-miR-548d-5p_480870_mir − 3.80 3.79E−03 8.96E−03 0.873 0.827 0.919 0.755 0.65 0.875 hsa-miR-30b-5p_478007_mir − 2.55 2.64E−03 6.47E−03 0.873 0.829 0.916 0.774 0.795 0.751 hsa-miR-363-3p_478060_mir − 8.98 3.58E−05 1.99E−04 0.869 0.825 0.914 0.782 0.776 0.789 hsa-miR-27b-3p_478270_mir − 2.46 2.87E−02 4.90E−02 0.868 0.828 0.908 0.76 0.815 0.698 hsa-miR-24-3p_477992_mir − 2.06 4.68E−05 2.40E−04 0.868 0.823 0.913 0.859 0.9 0.813 hsa-miR-499a-5p_478139_mir − 4.05 4.42E−05 2.34E−04 0.864 0.821 0.908 0.851 0.875 0.824 hsa-miR-15a-5p_477858_mir − 1.96 3.80E−04 1.28E−03 0.863 0.823 0.903 0.77 0.781 0.758 hsa-miR-31-3p_478012_mir − 5.27 2.92E−04 1.02E−03 0.86 0.817 0.904 0.799 0.766 0.836 hsa-miR-18a-3p_477944_mir − 4.70 5.55E−05 2.74E−04 0.859 0.817 0.901 0.761 0.747 0.777 hsa-miR-92a-3p_477827_mir − 1.48 5.48E−04 1.74E−03 0.859 0.811 0.907 0.813 0.822 0.804 hsa-miR-130b-3p_477840_mir − 5.57 3.72E−04 1.27E−03 0.858 0.815 0.901 0.748 0.644 0.866 hsa-let-7b-5p_478576_mir − 1.57 5.76E−04 1.81E−03 0.858 0.814 0.902 0.76 0.758 0.761 hsa-miR-30e-3p_478388_mir − 4.70 5.51E−03 1.22E−02 0.854 0.81 0.899 0.753 0.628 0.897 hsa-miR-23b-5p_477991_mir − 3.62 3.36E−05 1.99E−04 0.853 0.81 0.896 0.756 0.758 0.754 hsa-miR-29b-2-5p_478003_mir − 4.33 2.79E−06 2.75E−05 0.85 0.806 0.894 0.797 0.806 0.786 hsa-miR-30e-5p_479235_mir − 8.36 6.16E−04 1.92E−03 0.849 0.802 0.896 0.709 0.638 0.791 hsa-miR-200c-3p_478351_mir − 7.14 2.32E−05 1.53E−04 0.848 0.802 0.894 0.802 0.721 0.894 hsa-miR-1180-3p_477869_mir − 3.91 4.43E−05 2.34E−04 0.847 0.798 0.896 0.806 0.754 0.866 hsa-miR-190a-5p_478358_mir − 1.89 1.87E−02 3.48E−02 0.847 0.801 0.892 0.829 0.814 0.847 hsa-miR-151b_477811_mir − 9.82 2.25E−04 8.31E−04 0.846 0.801 0.892 0.761 0.733 0.792 hsa-miR-505-5p_478957_mir − 5.86 4.24E−05 2.32E−04 0.846 0.801 0.891 0.796 0.805 0.786 hsa-miR-196b-5p_478585_mir − 6.46 1.86E−06 2.00E−05 0.845 0.795 0.894 0.767 0.67 0.879 hsa-miR-324-5p_478024_mir − 1.51 8.13E−03 1.65E−02 0.843 0.797 0.889 0.777 0.754 0.802 hsa-miR-224-5p_477986_mir − 2.72 1.58E−04 6.45E−04 0.842 0.8 0.883 0.746 0.695 0.804 hsa-miR-139-5p_478312_mir − 5.12 2.43E−04 8.80E−04 0.839 0.794 0.885 0.727 0.725 0.729 hsa-miR-545-5p_479003_mir − 5.30 8.42E−06 6.79E−05 0.838 0.79 0.886 0.79 0.79 0.79 hsa-miR-222-3p_477982_mir − 2.08 7.82E−04 2.37E−03 0.838 0.791 0.884 0.728 0.762 0.689 hsa-miR-340-5p_478042_mir − 9.05 9.31E−06 7.18E−05 0.836 0.79 0.882 0.803 0.817 0.786 hsa-miR-504-5p_478144_mir − 3.54 1.86E−04 7.18E−04 0.836 0.787 0.886 0.778 0.729 0.833 hsa-miR-106a-5p_478225_mir − 10.76 4.30E−08 1.30E−06 0.834 0.784 0.883 0.816 0.924 0.693 hsa-miR-1271-5p_478674_mir − 9.10 3.87E−06 3.52E−05 0.834 0.785 0.883 0.823 0.9 0.734 hsa-miR-125b-2-3p_478666_mir − 6.80 9.11E−06 7.18E−05 0.834 0.782 0.886 0.804 0.866 0.735 hsa-miR-339-3p_478325_mir − 3.60 6.56E−03 1.39E−02 0.834 0.782 0.886 0.778 0.692 0.876 Roman‑Canal et al. J Transl Med (2019) 17:208 Page 7 of 14 Table 2 (continued) miRNA LogFC p-value Adj. p-value AUC​ AUC 95% AUC 95% Accuracy Sensitivity Specificity CI_lower CI_upper hsa-miR-483-5p_478432_mir − 6.45 5.42E−07 7.69E−06 0.832 0.783 0.882 0.785 0.871 0.687 hsa-miR-584-5p_478167_mir − 10.36 6.40E−07 8.25E−06 0.831 0.781 0.882 0.803 0.906 0.685 hsa-miR-17-3p_477932_mir − 8.43 2.22E−05 1.49E−04 0.831 0.779 0.883 0.805 0.861 0.741 hsa-miR-570-3p_479053_mir − 4.46 1.11E−04 4.79E−04 0.831 0.785 0.876 0.766 0.697 0.844 hsa-miR-625-5p_479469_mir − 10.93 4.36E−06 3.87E−05 0.83 0.781 0.88 0.806 0.865 0.738 hsa-miR-196a-5p_478230_mir − 7.37 1.01E−05 7.64E−05 0.83 0.777 0.883 0.812 0.879 0.735 hsa-miR-7-1-3p_478198_mir − 7.37 1.00E−04 4.52E−04 0.829 0.783 0.874 0.762 0.646 0.895 hsa-miR-450b-5p_478914_mir − 9.65 1.03E−07 2.29E−06 0.828 0.775 0.88 0.803 0.906 0.685 hsa-miR-221-5p_478778_mir − 5.01 3.08E−04 1.06E−03 0.827 0.779 0.875 0.745 0.64 0.865 hsa-miR-128-3p_477892_mir − 1.32 5.21E−03 1.16E−02 0.823 0.772 0.874 0.743 0.741 0.746 hsa-miR-491-5p_478132_mir − 3.50 1.88E−03 4.92E−03 0.822 0.774 0.87 0.743 0.703 0.79 hsa-miR-136-3p_477902_mir − 7.94 2.78E−05 1.76E−04 0.821 0.771 0.871 0.786 0.855 0.707 hsa-miR-101-5p_478620_mir − 7.44 1.09E−05 8.08E−05 0.819 0.766 0.873 0.812 0.877 0.738 hsa-miR-151a-3p_477919_mir − 1.93 2.85E−04 1.01E−03 0.819 0.769 0.87 0.803 0.854 0.745 hsa-miR-28-3p_477999_mir − 2.25 4.30E−03 1.00E−02 0.817 0.771 0.863 0.738 0.728 0.749 hsa-miR-489-3p_478130_mir − 4.40 1.33E−04 5.62E−04 0.815 0.766 0.865 0.76 0.736 0.787 hsa-miR-106b-3p_477866_mir − 2.41 6.48E−03 1.39E−02 0.815 0.765 0.866 0.72 0.693 0.751 hsa-miR-324-3p_478023_mir − 7.87 2.85E−04 1.01E−03 0.814 0.766 0.861 0.761 0.686 0.847 hsa-miR-125a-3p_477883_mir − 3.24 4.18E−04 1.39E−03 0.811 0.76 0.862 0.765 0.741 0.791 hsa-let-7i-3p_477862_mir − 6.66 4.69E−04 1.54E−03 0.81 0.759 0.861 0.747 0.665 0.841 hsa-miR-33b-5p_478479_mir − 5.49 2.06E−03 5.30E−03 0.81 0.758 0.862 0.704 0.626 0.793 hsa-miR-503-5p_478143_mir − 2.80 6.04E−03 1.31E−02 0.81 0.757 0.863 0.762 0.726 0.805 hsa-miR-301a-3p_477815_mir − 5.08 1.51E−03 4.05E−03 0.809 0.76 0.858 0.723 0.608 0.856 hsa-miR-330-3p_478030_mir − 5.72 7.79E−04 2.37E−03 0.805 0.754 0.856 0.761 0.645 0.892 hsa-miR-425-5p_478094_mir − 1.51 2.66E−02 4.63E−02 0.805 0.757 0.852 0.728 0.714 0.744 hsa-miR-16-2-3p_477931_mir − 3.32 1.12E−02 2.22E−02 0.804 0.747 0.862 0.78 0.769 0.793 hsa-miR-548k_479374_mir − 14.17 8.73E−03 1.76E−02 0.801 0.757 0.845 0.718 0.654 0.79 hsa-miR-429_477849_mir − 2.00 1.76E−02 3.37E−02 0.801 0.747 0.854 0.768 0.807 0.723 hsa-miR-598-3p_478172_mir − 1.51 4.51E−03 1.04E−02 0.8 0.754 0.847 0.696 0.726 0.661 hsa-miR-887-3p_479189_mir − 5.41 1.42E−04 5.95E−04 0.799 0.748 0.85 0.737 0.632 0.858 hsa-miR-93-3p_478209_mir − 4.60 2.28E−04 8.33E−04 0.798 0.745 0.852 0.749 0.702 0.801 hsa-miR-629-5p_478183_mir − 6.16 1.75E−04 6.90E−04 0.796 0.746 0.846 0.753 0.678 0.839 hsa-miR-21-5p_477975_mir 1.32 7.91E−03 1.61E−02 0.796 0.751 0.842 0.683 0.688 0.677 hsa-miR-140-3p_477908_mir − 3.42 6.10E−03 1.31E−02 0.793 0.743 0.843 0.734 0.752 0.713 hsa-miR-425-3p_478093_mir − 5.28 9.90E−04 2.90E−03 0.792 0.74 0.844 0.728 0.588 0.888 hsa-miR-200a-5p_478752_mir 3.05 2.34E−02 4.15E−02 0.792 0.735 0.85 0.713 0.693 0.736 hsa-miR-590-3p_478168_mir − 4.77 1.27E−03 3.56E−03 0.791 0.739 0.842 0.702 0.637 0.776 hsa-miR-30a-5p_479448_mir − 8.52 7.71E−04 2.37E−03 0.789 0.738 0.841 0.734 0.726 0.744 hsa-let-7 g-3p_477850_mir − 6.38 1.74E−04 6.90E−04 0.787 0.733 0.84 0.756 0.608 0.926 hsa-miR-542-3p_478153_mir − 8.15 2.88E−06 2.76E−05 0.786 0.729 0.842 0.778 0.901 0.638 hsa-miR-31-5p_478015_mir − 1.78 7.34E−03 1.51E−02 0.786 0.737 0.835 0.676 0.708 0.64 hsa-miR-379-5p_478077_mir − 5.15 5.46E−04 1.74E−03 0.78 0.724 0.835 0.755 0.773 0.734 hsa-miR-194-5p_477956_mir − 1.66 1.21E−03 3.48E−03 0.78 0.724 0.836 0.778 0.861 0.683 hsa-miR-34c-5p_478052_mir − 2.51 4.47E−03 1.04E−02 0.779 0.726 0.832 0.685 0.639 0.738 hsa-miR-576-5p_478165_mir − 6.85 8.70E−05 4.01E−04 0.778 0.721 0.836 0.774 0.853 0.684 hsa-miR-28-5p_478000_mir − 5.87 1.45E−03 3.96E−03 0.778 0.726 0.829 0.731 0.64 0.835 hsa-miR-708-3p_479162_mir − 2.77 2.27E−03 5.75E−03 0.772 0.715 0.828 0.738 0.74 0.736 hsa-miR-505-3p_478145_mir − 2.12 3.62E−03 8.67E−03 0.771 0.718 0.824 0.758 0.737 0.781 hsa-miR-26b-5p_478418_mir − 1.01 2.37E−02 4.19E−02 0.768 0.717 0.819 0.692 0.714 0.666 Roman‑Canal et al. J Transl Med (2019) 17:208 Page 8 of 14 Table 2 (continued) miRNA LogFC p-value Adj. p-value AUC​ AUC 95% AUC 95% Accuracy Sensitivity Specificity CI_lower CI_upper hsa-miR-365a-3p_478065_mir − 4.75 2.67E−02 4.63E−02 0.767 0.713 0.822 0.661 0.495 0.85 hsa-miR-423-3p_478327_mir − 1.63 1.58E−03 4.19E−03 0.765 0.716 0.815 0.65 0.687 0.607 hsa-miR-338-5p_478038_mir − 3.61 2.40E−03 6.01E−03 0.761 0.703 0.819 0.704 0.719 0.687 hsa-miR-210-3p_477970_mir − 1.17 1.67E−02 3.22E−02 0.761 0.708 0.815 0.672 0.698 0.643 hsa-miR-551a_478158_mir − 5.58 1.97E−03 5.10E−03 0.76 0.699 0.822 0.741 0.69 0.8 hsa-miR-889-3p_478208_mir − 8.68 1.14E−05 8.28E−05 0.759 0.698 0.82 0.789 0.922 0.637 hsa-miR-301b-3p_477825_mir − 5.75 1.87E−02 3.48E−02 0.758 0.7 0.816 0.751 0.851 0.637 hsa-miR-590-5p_478367_mir − 3.83 1.63E−02 3.15E−02 0.757 0.703 0.812 0.707 0.59 0.841 hsa-miR-548am-5p_480872_mir − 3.25 1.13E−02 2.22E−02 0.757 0.703 0.81 0.663 0.526 0.82 hsa-miR-187-3p_477941_mir − 4.47 6.97E−03 1.46E−02 0.754 0.701 0.807 0.704 0.644 0.772 hsa-miR-450a-5p_478106_mir − 7.12 1.26E−03 3.56E−03 0.753 0.692 0.814 0.743 0.646 0.853 hsa-miR-376a-5p_478859_mir − 5.94 1.04E−04 4.56E−04 0.75 0.688 0.811 0.738 0.832 0.631 hsa-miR-1296-5p_479451_mir − 4.26 2.49E−03 6.17E−03 0.75 0.695 0.806 0.703 0.719 0.686 hsa-miR-181c-3p_477933_mir − 3.64 9.29E−04 2.77E−03 0.748 0.685 0.811 0.715 0.805 0.613 hsa-miR-1247-5p_477882_mir − 3.36 2.43E−02 4.26E−02 0.748 0.692 0.804 0.698 0.586 0.827 hsa-miR-34a-3p_478047_mir − 2.32 1.77E−02 3.37E−02 0.748 0.694 0.803 0.658 0.727 0.579 hsa-miR-654-3p_479135_mir − 7.56 4.77E−04 1.55E−03 0.74 0.681 0.799 0.76 0.692 0.839 hsa-miR-411-5p_478086_mir − 5.12 7.16E−03 1.49E−02 0.738 0.68 0.796 0.715 0.647 0.793 hsa-miR-181d-5p_479517_mir − 5.67 4.10E−03 9.63E−03 0.733 0.675 0.791 0.686 0.554 0.838 hsa-miR-200b-3p_477963_mir − 3.43 7.26E−03 1.50E−02 0.733 0.674 0.792 0.683 0.609 0.767 hsa-miR-299-3p_478792_mir − 6.51 4.73E−05 2.40E−04 0.732 0.67 0.794 0.748 0.893 0.582 hsa-miR-182-5p_477935_mir − 8.68 8.37E−05 3.91E−04 0.73 0.666 0.795 0.753 0.9 0.585 hsa-miR-410-3p_478085_mir − 5.65 1.22E−03 3.48E−03 0.728 0.67 0.787 0.753 0.771 0.733 hsa-miR-744-5p_478200_mir − 3.55 1.81E−02 3.40E−02 0.725 0.67 0.781 0.657 0.637 0.68 hsa-miR-96-5p_478215_mir − 6.84 1.03E−04 4.56E−04 0.724 0.658 0.79 0.744 0.886 0.583 hsa-miR-133b_480871_mir − 3.18 1.98E−02 3.66E−02 0.724 0.664 0.783 0.686 0.651 0.726 hsa-miR-544a_478156_mir − 5.67 1.37E−03 3.77E−03 0.719 0.66 0.779 0.756 0.81 0.694 hsa-miR-497-3p_478946_mir − 6.92 1.44E−04 5.95E−04 0.718 0.652 0.784 0.753 0.899 0.585 hsa-miR-331-3p_478323_mir − 5.93 3.27E−03 7.90E−03 0.715 0.656 0.774 0.761 0.775 0.744 hsa-let-7f-2-3p_477843_mir − 3.26 1.19E−02 2.33E−02 0.715 0.655 0.776 0.682 0.567 0.814 hsa-miR-195-3p_478744_mir − 4.54 6.67E−03 1.40E−02 0.712 0.652 0.773 0.717 0.712 0.723 hsa-miR-378a-5p_478076_mir − 5.30 2.12E−03 5.41E−03 0.71 0.65 0.769 0.723 0.584 0.882 hsa-miR-1-3p_477820_mir − 6.27 6.10E−03 1.31E−02 0.706 0.644 0.768 0.732 0.778 0.679 hsa-miR-615-3p_478175_mir − 3.14 4.99E−03 1.13E−02 0.706 0.643 0.769 0.696 0.652 0.746 hsa-miR-545-3p_479002_mir − 2.99 1.99E−02 3.66E−02 0.704 0.64 0.767 0.675 0.678 0.672 hsa-miR-548a-3p_478157_mir − 3.75 4.83E−03 1.10E−02 0.703 0.641 0.765 0.716 0.572 0.881 hsa-miR-1248_478653_mir − 2.64 1.08E−02 2.17E−02 0.702 0.642 0.763 0.707 0.764 0.643 hsa-miR-381-3p_477816_mir − 5.10 5.97E−03 1.31E−02 0.701 0.639 0.763 0.734 0.727 0.742 hsa-miR-627-5p_478427_mir − 4.94 5.01E−04 1.62E−03 0.701 0.632 0.77 0.735 0.863 0.589 hsa-miR-1301-3p_477897_mir − 5.30 5.52E−03 1.22E−02 0.696 0.638 0.754 0.711 0.596 0.843 hsa-miR-486-3p_478422_mir − 6.30 1.86E−04 7.18E−04 0.692 0.626 0.758 0.725 0.891 0.535 hsa-miR-200a-3p_478490_mir − 4.37 1.79E−02 3.39E−02 0.691 0.626 0.756 0.719 0.792 0.635 hsa-miR-15a-3p_477928_mir − 3.53 6.65E−03 1.40E−02 0.688 0.62 0.755 0.703 0.808 0.584 hsa-miR-548 g-3p_479020_mir − 2.12 2.25E−02 4.04E−02 0.687 0.625 0.75 0.691 0.759 0.613 hsa-miR-432-5p_478101_mir − 3.66 2.70E−02 4.65E−02 0.683 0.621 0.746 0.691 0.724 0.652 hsa-miR-15b-5p_478313_mir − 6.53 2.89E−02 4.90E−02 0.682 0.619 0.745 0.666 0.692 0.637 hsa-miR-132-5p_478705_mir − 6.01 3.23E−03 7.86E−03 0.68 0.614 0.745 0.734 0.867 0.582 hsa-let-7e-5p_478579_mir − 5.40 2.04E−02 3.72E−02 0.679 0.614 0.744 0.655 0.556 0.769 hsa-miR-485-5p_478126_mir − 2.09 2.46E−02 4.30E−02 0.678 0.613 0.743 0.688 0.734 0.634 Roman‑Canal et al. J Transl Med (2019) 17:208 Page 9 of 14 Table 2 (continued) miRNA LogFC p-value Adj. p-value AUC​ AUC 95% AUC 95% Accuracy Sensitivity Specificity CI_lower CI_upper hsa-miR-299-5p_478793_mir − 5.05 8.70E−04 2.62E−03 0.672 0.603 0.741 0.702 0.858 0.524 hsa-miR-16-1-3p_478727_mir − 4.44 1.55E−03 4.13E−03 0.669 0.603 0.735 0.699 0.841 0.537 hsa-miR-215-5p_478516_mir − 7.23 8.19E−05 3.88E−04 0.668 0.601 0.735 0.718 0.896 0.514 hsa-miR-103a-2-5p_477864_mir − 4.32 2.05E−02 3.73E−02 0.664 0.603 0.724 0.664 0.529 0.818 hsa-miR-29a-5p_478002_mir − 6.48 2.54E−03 6.26E−03 0.643 0.568 0.717 0.723 0.896 0.524 hsa-miR-874-3p_478205_mir − 2.80 2.77E−02 4.74E−02 0.638 0.575 0.701 0.617 0.524 0.722 hsa-miR-502-5p_478954_mir − 4.03 2.18E−02 3.93E−02 0.637 0.573 0.701 0.662 0.53 0.813 hsa-miR-542-5p_478337_mir − 5.28 3.71E−03 8.83E−03 0.633 0.564 0.702 0.68 0.855 0.481 hsa-miR-362-3p_478058_mir − 5.81 1.73E−03 4.56E−03 0.632 0.564 0.701 0.69 0.879 0.475 hsa-miR-431-3p_478888_mir − 4.55 4.63E−03 1.06E−02 0.624 0.552 0.696 0.678 0.853 0.478 hsa-miR-192-3p_478741_mir − 4.08 5.15E−03 1.16E−02 0.61 0.535 0.685 0.674 0.85 0.474 hsa-miR-589-5p_479073_mir − 2.81 2.28E−02 4.07E−02 0.582 0.512 0.652 0.633 0.795 0.448 hsa-miR-888-5p_479192_mir − 3.79 2.06E−02 3.73E−02 0.559 0.489 0.63 0.656 0.861 0.422 hsa-miR-15b-3p_477929_mir − 3.73 1.76E−02 3.37E−02 0.529 0.459 0.599 0.629 0.86 0.365 hsa-miR-651-5p_479131_mir − 3.94 2.90E−02 4.90E−02 0.523 0.448 0.599 0.638 0.87 0.372 Log fold-change expression, p-value, adjusted p-value, AUC values, accuracy, sensitivity, specificity, and 95% of confidence intervals of the 210 dysregulated miRNAs CI confidence of interval profile of serum samples from CRC patients and representation of the molecular alterations that takes healthy controls and identified 16 miRNA that were place in the tumor. The peritoneal lavage is a proximal expressed in a significantly higher levels among CRC fluid with an unexploded value in biomarker research patients. Of these, 7 miRNAs (let-7a, miR-1229, miR- for cancers originating within the peritoneal cavity. 1246, miR-150, miR-21, miR-223, and miR-23a) were Tokuhisa et al. [23] showed that EV-associated miRNAs suggested as promising diagnostic biomarkers of CRC can be consistently extracted from this bodily fluid and with an AUC between 0.67 and 0.95. More recently, the that miRNAs expression profiles can indicate the status serum exosomal miRNA-19a was found to be upregu- of peritoneum in gastric cancer patients. lated in the serum of CRC patients compared to healthy To the best of our knowledge, our study is the first to volunteers, but also was associated with poor progno- report the value of this proximal fluid for the identifica- sis [19]. Finally, Zhao et al. [20], demonstrated that the tion of miRNAs associated to EVs in CRC. Importantly, exosomal miRNA-21 expression is associated with the this study unveiled the promising use of the top-10 early diagnosis of CRC. Although plasma and serum miRNA dysregulated (miRNA-199b-5p, miRNA-150-5p, have reported promising biomarkers for CRC diagno- miRNA-29c-5p, miRNA-218-5p, miRNA-99a-3p, sis, other approaches as it is the use of proximal bodily miRNA-383-5p, miRNA-199a-3p, miRNA-193a-5p, fluids as a source of biomarkers have aroused the atten- miRNA-10b-5p and miRNA-181c-5p) as diagnostic bio- tion of the biomarker research community. Proximal markers, all showing the AUC value higher than 0.95. bodily fluids, such as urine for prostate cancer [21], Those biomarkers should be validated as well as com- or uterine fluid for endometrial cancer [22] have dem- bined in order to increase the already excellent accuracy onstrated that this type of fluids offers an improved of individual miRNAs. However, this should be done Roman‑Canal et al. J Transl Med (2019) 17:208 Page 10 of 14 Fig. 1 Diagnostic performance of the top-10 differentially expressed miRNAs. a Relative dCT values of top differentially expressed miRNAs in patients with CRC (n = 19) compared to control patients (n = 22). **p < 0.05. b ROC-curves and AUC-scores the top-10 differentially expressed miRNAs Table 3 Published studies of the top-10 miRNAs dysregulated in CRC patients Tissue samples Other type of samples miR-199b-5p Not previously reported Not previously reported miR-150-5p Upregulated: [24, 25] Downregulated: serum [28] Downregulated: [26, 27] miR-29c-5p Not previously reported Not previously reported miR-218-5p Upregulated: [29] Not previously reported Downregulated: [30, 31] miR-99a-3p Not previously reported Not previously reported miR-383-5p Downregulated: [32] Not previously reported miR-199a-3p Upregulated: [26] Upregulated in stool [33, 34] miR-193a-5p Downregulated in CRC cell lines [35] Not previously reported miR-10b-5p Downregulated: [30, 36, 37] Not previously reported miR-181c-5p Upregulated: [38] Not previously reported Roman‑Canal et al. J Transl Med (2019) 17:208 Page 11 of 14 a b 210 miRNA dysregulated 0 200 400 600 800 1000 1200 Number of regulated transcripts Fig. 2 Prediction of miRNA target transcripts. a Boxplot of the number of transcripts regulated by the 210 dysregulated miRNAs. b Graphical representation in which the transcript ID is represented with a different font size accordingly to the times that is predicted to be regulated by the different miRNAs, i.e. the most frequently regulated transcripts are shown in a larger font size in an independent study including a larger cohort of untapped source of biomarkers. Specifically, we identi- patients. Moreover, further analysis should be performed fied miRNA-199b-5p, miRNA-150-5p, miRNA-29c-5p, to elucidate the prognostic value of the detection of the miRNA-218-5p, miRNA-99a-3p, miRNA-383-5p, different types of miRNAs in EVs isolated from perito- miRNA-199a-3p, miRNA-193a-5p, miRNA-10b-5p and neal lavages. miRNA-181c-5p as promising biomarkers of CRC diag- nosis with the AUC value higher than 0.95. Conclusions In this study, we have demonstrated that use of EV- associated miRNA of ascitic liquid from control patients and peritoneal lavages from CRC patients are an Roman‑Canal et al. J Transl Med (2019) 17:208 Page 12 of 14 a b Fig. 3 Sun projection plot of GO terms. Predicted transcripts regulated by the differentially expressed miRNAs. a GO analysis of up-regulated and down-regulated target genes according to biological process. b GO analysis of up-regulated and down-regulated target genes according to molecular function Roman‑Canal et al. J Transl Med (2019) 17:208 Page 13 of 14 Additional files Technology Park, Derio, Spain. 8 Pleural Medicine Unit, Arnau de Vilanova University Hospital, IRBLleida, Lleida, Spain. 9 Gynecological Oncology Depart‑ ment, Vall Hebron University Hospital, CIBERONC, Barcelona, Spain. 10 IKER‑ Additional file 1: Table S1. Clinicopathological characteristics of all BASQUE, Basque Foundation for Science, 48011 Bilbao, Spain. 11 University patients. Pompeu Fabra, Barcelona, Spain. 12 Oncologic Pathology Group, Department of Medicine UdL, Biomedical Research Institute of Lleida (IrbLleida), Av. Rovira Additional file 2: Figure S1. Workflow. Workflow of the study design. Roure 80, 25198 Lleida, Spain. Additional file 3: Figure S2. EVs characterization. (A) Box-plot repre‑ senting the average mode of EVs isolated from the peritoneal lavage Received: 2 April 2019 Accepted: 13 June 2019 and ascitic fluid of CRC and control patients, respectively (Mean ± SD); measured by Nanoparticle Tracking Analysis. (B) Size distribution and concentration of isolated EVs of a peritoneal lavage of a CRC patient (left) and a ascitic fluid of a control patient (right), measured by Nanoparticle Tracking Analysis. References 1. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mor‑ Abbreviations tality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. CRC​: colorectal cancer; EVs: extracellular vesicles; RT: reverse transcription; Ct: 2018;68(6):394–424. cycle threshold; FDR: False Discovery Rate; AUC​: area under the ROC curve; 2. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018: cancer statistics, GO: Gene Ontology; BP: biological process; MF: molecular function; 3′-UTR​: 2018. CA Cancer J Clin. 2018;68(1):7–30. 3′-untranslated region. 3. Schmoll HJ, Van Cutsem E, Stein A, Valentini V, Glimelius B, Haustermans K, et al. ESMO Consensus Guidelines for management of patients with Acknowledgements colon and rectal cancer. A personalized approach to clinical decision The authors would like to acknowledge the work of all who have participated making. Ann Oncol. 2012;23(10):2479–516. in the recruitment of clinical samples. We also thank the patients for their 4. Ogata-Kawata H, Izumiya M, Kurioka D, Honma Y, Yamada Y, Furuta K, et al. willingness to participate in the study. Circulating exosomal microRNAs as biomarkers of colon cancer. PLoS ONE. 2014;9(4):e92921. Authors’ contributions 5. Wang Y, Chen Z, Chen W. Novel circulating microRNAs expression Study conception and design: XMG, EC. Sample and data collection: SG, MRM, profile in colon cancer: a pilot study. Eur J Med Res. 2017. https​://doi. JT, JMP, AGM, JES, MR. Acquisition of data: BRC, CM, SB, EG. Analysis and inter‑ org/10.1186/s4000​1-017-0294-5. pretation of data: SB, JP, CM, BRC, JMFP, EG, XMG, EC. Drafting of manuscript: 6. Zeng M, Zhu L, Li L, Kang C. miR-378 suppresses the proliferation, migra‑ BRC, CM. Critical revision: CM, JMFP, EG, JMP, JP, XMG, EC. All authors read and tion and invasion of colon cancer cells by inhibiting SDAD1. Cell Mol Biol approved the final manuscript. Lett. 2017. https​://doi.org/10.1186/s1165​8-017-0041-5. 7. Yuan Q, Cao G, Li J, Zhang Y, Yang W. MicroRNA-136 inhibits colon Funding cancer cell proliferation and invasion through targeting liver receptor EC hold a postdoctoral fellowship from the Departament de Salut of the Gen‑ homolog-1/Wnt signaling. Gene. 2017;628:48–55. eralitat de Catalunya (SLT002/16/00274). This work was supported by grants: 8. Yáñez-Mó M, Siljander P-RM, Andreu Z, Zavec A, Borràs FE, Buzas EI, et al. Discovery, validation and implementation of biomarkers for Precision Oncol‑ Biological properties of extracellular vesicles and their physiological func‑ ogy (ISCIII PIE15/00029), CIBERONC (CB16/12/00231 and CB16/12/00328). tions. J Extracell Vesicles. 2015;4(1):27066. Grups consolidats de la Generalitat de Catalunya (2017SGR1368 and 9. Campoy I, Lanau L, Altadill T, Sequeiros T, Cabrera S, Cubo-Abert M, et al. 2017SGR1661). Work supported by IRBLleida BIOBANK (B.0000682) and Plata‑ Exosome-like vesicles in uterine aspirates: a comparison of ultracentrif‑ forma biobancos PT17/0015/0027. ugation-based isolation protocols. J Transl Med. 2016;14:1. https​://doi. org/10.1186/s1296​7-016-0935-4. Availability of data and materials 10. Huber W, Carey VJ, Gentleman R, Anders S, Carlson M, Carvalho BS, et al. All bioinformatics analysis was performed with the BioConductor (version 3.7) Orchestrating high-throughput genomic analysis with bioconductor. Nat [10] project in the R statistical environment (version 3.5.0). Methods. 2015;12(2):115–21. 11. R Foundation for Statistical Computing, Vienna, Austria. R: A language Ethics approval and consent to participate and environment for statistical computing. Participants in the study attended to the Hospital Arnau de Vilanova in Lleida, 12. Dvinge H, Bertone P. HTqPCR: high-throughput analysis and Spain. The Clinical Research Ethics Committee of the hospital approved the visualization of quantitative real-time PCR data in R. Bioinformatics. study and all the participating patients provided a signed informed consent. 2009;25(24):3325–6. 13. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. limma powers dif‑ Consent for publication ferential expression analyses for RNA-sequencing and microarray studies. All authors consent for publication. Nucleic Acids Res. 2015;43(7):e47. 14. BENJAMINIt Y, Hochberg Y. Controlling the false discovery rate: a practical Competing interests and powerful approach to multiple testing. 13. The authors declare that they have no competing interests. 15. Dweep H, Gretz N. miRWalk2.0: a comprehensive atlas of microRNA- target interactions. Nat Methods. 2015;12(8):697. Author details 16. Mi H, Muruganujan A, Casagrande JT, Thomas PD. Large-scale gene 1 Department of Pathology and Molecular Genetics/Oncologic Pathology function analysis with the PANTHER classification system. Nat Protoc. Group, Biomedical Research Institute of Lleida (IRBLleida), University of Lleida, 2013;8(8):1551–66. CIBERONC, Lleida, Spain. 2 Department of Pathology, University Hospital 17. Mazeh H, Mizrahi I, Ilyayev N, Halle D, Brücher BL, Bilchik A, et al. The of Bellvitge, Bellvitge Biomedical Research Institute (IDIBELL), L’Hospitalet diagnostic and prognostic role of microrna in colorectal cancer—a com‑ de Llobregat, Barcelona, Spain. 3 Universitat Autònoma de Barcelona (UAB), prehensive review. J Cancer. 2013;4(3):281–95. Barcelona, Spain. 4 Biomedical Research Group in Gynecology, Vall Hebron 18. Chen X, Ba Y, Ma L, Cai X, Yin Y, Wang K, et al. Characterization of micro‑ Research Institute (VHIR), Universitat Autònoma de Barcelona, CIBERONC, Pg. RNAs in serum: a novel class of biomarkers for diagnosis of cancer and Vall Hebron 119‑129, 08035 Barcelona, Spain. 5 Centre for Genomic Regulation other diseases. Cell Res. 2008;18(10):997–1006. (CRG), The Barcelona Institute or Science and Technology, Dr. Aiguader 88, 19. Matsumura T, Sugimachi K, Iinuma H, Takahashi Y, Kurashige J, Sawada G, Barcelona 08003, Spain. 6 Department of Surgery, Hospital Arnau de Vilanova, et al. Exosomal microRNA in serum is a novel biomarker of recurrence in Biomedical Research Institute of Lleida (IRBLleida), Lleida, Spain. 7 Exosomes human colorectal cancer. Br J Cancer. 2015;113(2):275–81. Laboratory and Metabolomics Platform, CIC bioGUNE, CIBEREHD Bizkaia Roman‑Canal et al. J Transl Med (2019) 17:208 Page 14 of 14 20. Zhao L, Yu J, Wang J, Li H, Che J, Cao B. Isolation and Identification of miR‑ 31. Ye Y, Gu B, Wang Y, Shen S, Huang W, et al. E2F1-mediated MNX1-AS1- NAs in exosomes derived from serum of colon cancer patients. J Cancer. miR-218-5p-SEC61A1 feedback loop contributes to the progression of 2017;8(7):1145–52. colon adenocarcinoma. J Cell Biochem. 2019;120(4):6145–53. 21. Sequeiros T, Rigau M, Chiva C, Montes M, Garcia-Grau I, Garcia M, et al. 32. Hamfjord J, Stangeland AM, Hughes T, Skrede ML, Tveit KM, Ikdahl T, et al. Targeted proteomics in urinary extracellular vesicles identifies biomarkers Differential expression of miRNAs in colorectal cancer: comparison of for diagnosis and prognosis of prostate cancer. Oncotarget. 2017; 8 (3). paired tumor tissue and adjacent normal mucosa using high-throughput http://www.oncot​arget​.com/fullt​ext/13634​. sequencing. PLoS ONE. 2012;7(4):e34150. 22. Martinez-Garcia E, Lesur A, Devis L, Cabrera S, Matias-Guiu X, Hirschfeld 33. Ahmed FE, Ahmed NC, Gouda MM, Vos PW, Bonnerup C. RT-qPCR for M, et al. Targeted proteomics identifies proteomic signatures in liquid fecal mature microRNA quantification and validation. In: Beaulieu J-F, biopsies of the endometrium to diagnose endometrial cancer and assist editor. Colorectal cancer. New York: Springer; 2018. p. 203–15. https​://doi. in the prediction of the optimal surgical treatment. Clin Cancer Res. org/10.1007/978-1-4939-7765-9_13. 2017;23(21):6458–67. 34. Ahmed FE, Ahmed NC, Vos PW, Bonnerup C, Atkins JN, Casey M, et al. 23. Tokuhisa M, Ichikawa Y, Kosaka N, Ochiya T, Yashiro M, Hirakawa K, et al. Diagnostic MicroRNA markers to screen for sporadic human colon cancer Exosomal miRNAs from peritoneum lavage fluid as potential prognos‑ in stool: I. Proof of principle. Cancer Genomics. 2013;21:93. tic biomarkers of peritoneal metastasis in gastric cancer. PLoS ONE. 35. Shirafkan N, Shomali N, Kazemi T, Shanehbandi D, Ghasabi M, Baghbani E, 2015;10(7):e0130472. et al. MicroRNA-193a-5p inhibits migration of human HT-29 colon cancer 24. Volinia S, Calin GA, Liu C-G, Ambs S, Cimmino A, Petrocca F, et al. A micro‑ cells via suppression of metastasis pathway. J Cell Biochem. 2018. https​:// RNA expression signature of human solid tumors defines cancer gene doi.org/10.1002/jcb.28164​. targets. Proc Natl Acad Sci. 2006;103(7):2257–61. 36. Arndt GM, Dossey L, Cullen LM, Lai A, Druker R, Eisbacher M, et al. Char‑ 25. Jacob H, Stanisavljevic L, Storli KE, Hestetun KE, Dahl O, Myklebust MP. acterization of global microRNA expression reveals oncogenic potential Identification of a sixteen-microRNA signature as prognostic biomarker of miR-145 in metastatic colorectal cancer. BMC Cancer. 2009. https​://doi. for stage II and III colon cancer. Oncotarget. 2017;8 (50). http://www. org/10.1186/1471-2407-9-374. oncot​arget​.com/fullt​ext/21237​. Accessed 25 Feb 2019. 37. Sarver AL, French AJ, Borralho PM, Thayanithy V, Oberg AL, Silverstein 26. Slattery ML, Wolff E, Hoffman MD, Pellatt DF, Milash B, Wolff RK. MicroR‑ KA, et al. Human colon cancer profiles show differential microRNA NAs and colon and rectal cancer: differential expression by tumor loca‑ expression depending on mismatch repair status and are characteristic tion and subtype. Genes Chromosomes Cancer. 2011;50(3):196–206. of undifferentiated proliferative states. BMC Cancer. 2009. https​://doi. 27. Ma Y, Zhang P, Wang F, Zhang H, Yang J, Peng J, et al. miR-150 as a poten‑ org/10.1186/1471-2407-9-401. tial biomarker associated with prognosis and therapeutic outcome in 38. Monzo M, Navarro A, Bandres E, Artells R, Moreno I, Gel B, et al. Overlap‑ colorectal cancer. Gut. 2012;61(10):1447–53. ping expression of microRNAs in human embryonic colon and colorectal 28. Wang Q, Huang Z, Ni S, Xiao X, Xu Q, Wang L, et al. Plasma miR-601 and cancer. Cell Res. 2008;18(8):823–33. miR-760 are novel biomarkers for the early detection of colorectal cancer. PLoS ONE. 2012;7(9):e44398. 29. Zhou X, Huang X, Liang S, Tang S, Wu S, Huang T, et al. Identifying miRNA Publisher’s Note and gene modules of colon cancer associated with pathological stage Springer Nature remains neutral with regard to jurisdictional claims in pub‑ by weighted gene co-expression network analysis. OncoTargets Ther. lished maps and institutional affiliations. 2018;11:2815–30. 30. Mosakhani N, Sarhadi VK, Borze I, Karjalainen-Lindsberg M-L, Sundström J, Ristamäki R, et al. MicroRNA profiling differentiates colorectal cancer according to KRAS status. Genes Chromosomes Cancer. 2012;51(1):1–9. Ready to submit your research ? Choose BMC and benefit from: • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations • maximum visibility for your research: over 100M website views per year At BMC, research is always in progress. Learn more biomedcentral.com/submissions