Sage Journals: Discover world-class research

Abstract

BACKGROUND:

Circulating miRNAs are promising biomarkers for detection of various cancers. As a “developmental” disorder, cancer showed great similarities with embryos.

OBJECTIVE:

A comprehensive analysis of circulating miRNAs in umbilical cord blood (UCB) and pan-cancers was conducted to identify circulating miRNAs with potential for cancer detection.

METHODS:

A total of 3831 cancer samples (2050 serum samples from 15 types of cancers and 1781 plasma samples from 13 types of cancers) and 248 UCB samples (120 serum and 128 plasma samples) with corresponding NCs from Chinese populations were analyzed via consistent experiment workflow with Exiqon panel followed by multiple-stage validation with qRT-PCR.

RESULTS:

Thirty-four serum and 32 plasma miRNAs were dysregulated in at least one type of cancer. Eighteen serum and 16 plasma miRNAs were related with embryos. Among them, 9 serum and 8 plasma miRNAs with consistent expression patterns between pan-cancers and UCB were identified as circulating oncofetal miRNAs. Retrospective analysis confirmed the diagnostic ability of circulating oncofetal miRNAs for specific cancers. And the oncofetal miRNAs were mainly up-regulated in tissues of pan-cancers.

CONCLUSIONS:

Our study might serve as bases for the potential application of the non-invasive biomarkers in the future clinical.

Keywords

Circulating miRNAs UCB pan-cancers oncofetal biomarkers

1. Introduction

Cancer is one of leading causes of death worldwide. Despite increased understanding of the molecular and clinical characteristics of cancers as well as improvements in treatment strategies, the prognosis of cancer patients is still poor [1]. Due to lack of obvious symptoms and insufficiently sensitive biomarkers, many patients are diagnosed with advanced stages and do not have a choice of surgery which still remains to be the only treatment strategy to cure the disease. Thus, early detection and intervention is the key strategy for improving the long-term survivals of cancer patients. Nowadays, pathological examination of biopsy is still the gold standard for cancer diagnosis. But the invasiveness, inconvenience and potentially subjective discrepancy caused by different operators limit its effectiveness in screening the diseases in large-scale populations. Existing tumor diagnostic biomarkers, such as carcinoembryonic antigen (CEA), carbohydrate antigen 19-9 (CA19-9), squamous cell carcinoma antigen (SCC) and CYFRA21-1, did not show sufficient sensitivity and specificity in diagnosis of cancers [2, 3, 4]. Hence, more novel and reliable biomarkers to detect the diseases for early intervention are urgently needed.

MicroRNAs (miRNAs) are small non-coding RNAs (18–25 nucleotides in length) which could regulate gene expression by inhibiting translation or encouraging transcript degradation. Hundreds of miRNAs are dysregulated in various cancers and could regulate processes related to all major hallmarks of cancers by targeting various mRNAs [5]. Since Mitchell et al. firstly reported the significance of serum miR-141 in diagnosis of prostate cancer, many studies have demonstrated that miRNAs could be stably detected in circulating plasma/serum, and could act as non-invasive biomarkers for early detection of many types of cancers [6]. But the results of identified circulating miRNAs for the same type of cancers were always heterogeneous among the published studies. Differences in research methods and populations might partly explain the inconformity. Meanwhile, expression profiles of circulating miRNAs in serum and plasma samples were also distinctive [7]. Therefore, we have committed to identifying circulating serum and plasma miRNAs as candidate diagnostic biomarkers for various types of cancers with Chinese populations in the past few years. With the almost identical workflow and appropriate normalization methods, serum miRNAs in esophageal squamous cell carcinoma (ESCC) [8], gastric cardia adenocarcinoma (GCA) [9], gastric non-cardia adenocarcinoma (GNCA) [10], colon adenocarcinoma (COAD), rectum adenocarcinoma (READ) [11], pancreatic adenocarcinoma (PAAD) [12], lung squamous cell carcinoma (LUSC) [13] and ovarian cancer (OV) [14], and plasma miRNAs in ESCC [15], GNCA [16], PAAD [17], LUSC [18], lung adenocarcinoma (LUAD) [19], thyroid cancer (THCA) [20], COAD, READ [21] and OV [22] were confirmed and could aid in cancer detection. As cancer is a disease with great heterogeneity as well as similarity, a pan-cancer analysis to show a global view of circulating miRNAs in various cancers is critical and conducive for us to better apply the potential diagnostic tools. To date, no study has conducted a pan-cancer analysis of circulating miRNAs in cancers. To address this issue, we additionally analyzed circulating miRNAs with serum samples in nasopharyngeal carcinoma (NPC), THCA, LUAD, breast cancer (BRCA), uterine corpus endometrial carcinoma (UCEC), prostate adenocarcinoma (PRAD) and cervical squamous cell carcinoma (CESC), and plasma samples in NPC, BRCA, CESC and UCEC according to the consistent criteria. Thus, with the data we previously reported (re-analyzed with the consistent criteria), the pan-cancer analysis of circulating miRNAs was presented in the study.

More and more studies indicated that embryos and tumors possess many similarities in pathophysiological processes, such as immune escape, angiogenesis, gene expression, invasion characteristic between embryo implantation and neoplasm invasion and metastasis [23, 24]. In addition, some sets of genes, proteins or metabolites shared between embryogenesis and tumorigenesis were applied as biomarkers for diagnosis of cancer, such as oncofetal proteins alpha-fetoprotein (AFP) and carcino-embryonic antigen (CEA). Therefore, it enlightens us to study cancers from the view of developmental biology, which may provide a new clue for cancer detection. Umbilical cord blood (UCB) is a newborn’s blood obtained after delivery from umbilical cord and represents the most accessible source of neonatal blood. In the present study, we comprehensively analyzed circulating miRNAs in UCB and pan-cancers and identified oncofetal circulating miRNAs which showed similar expression patterns in UCB and various types of cancers. The study could promote our deeper understanding of circulating miRNAs in cancers and further identification of novel cancer biomarkers.

2. Materials and methods

2.1 Sample collection and preparation

Peripheral blood samples of 15 types of cancer patients as well as age and gender matched normal control (NC) blood samples, and UCB samples were collected during 2010 and 2018 at First Affiliated Hospital of Nanjing Medical University. Blood samples of cancer patients analyzed in the study were all obtained before any clinical treatment and intervention. NC blood samples were collected from adult healthy donors who conducted routine health checkup. UCB samples were obtained immediately after the fetus was expelled.

The blood samples were put in an ethylene diamine tetra acetic acid (EDTA)-containing tubes (Becton, Dickinson and Company) or SST Advance tubes (Becton, Dickinson and Company) for separation of plasma and serum samples within 12 hours. Cell-free plasma samples were isolated using a two-step centrifugal process of 350 reactive centrifugal force (RCF) for 10 min and 20,000 RCF for 10 min (Beckman Coulter, USA), while serum samples were obtained by centrifugation at 1,500 RCF for 10 min and 12,000 RCF for 2 min after allowing blood to clot for 30 min. All the samples were then stored at $-$ 80 ${}^{\circ}$ C until further processing.

The study was approved and conducted in accordance with the guidelines by medical ethics committee of the First Affiliated Hospital of Nanjing Medical University. Each participant received written informed consent.

2.2 Study design

To identify differential expression of circulating miRNAs in different types of cancer patients and UCB samples, plasma/serum samples of each type of cancer and UCB as well as their own age and gender matched NC samples were analyzed independently in according to the experiment workflow (Fig. 1; GCA serum, GNCA serum, OV serum and OV plasma samples were evaluated through training and testing stages after the screening phase). In brief, each independent experiment according to different sample types was designed into four stages: screening phase, training, testing and external validation stage. In the screening phase, 2 to 4 case pool samples VS. 1 NC pool sample (every 10 samples were pooled as 1 pool sample) was applied to initially screen potential differential expressed circulating miRNAs using Exiqon miRCURY-Ready-to-Use PCR-Human-panel-I+II-V1.M (Exiqon miRNA qPCR panel, Vedbaek, Denmark; 168 miRNAs). The process of arrays and analyses was conducted consistent with the previous study [16]. Then, those miRNAs fulfilled the criteria (miRNAs with Ct $>$ 37 or 5 higher than NTC in the panel was omitted; miRNAs $>$ 1.5-fold or $<$ 0.67-fold in all pooled case samples were selected) were subjected to further analyzes. Subsequently, the miRNAs (with $>$ 1.5-fold or $<$ 0.67-fold as well as $p$ value $<$ 0.05 between cases and NCs) confirmed through the training, testing and external validation stage with qRT-PCR assays were identified as signature miRNAs.

In total, circulating samples including NPC (serum: 208 cases VS. 238 NCs; plasma: 200 cases VS. 189 NCs), THCA (serum: 100 cases VS. 96 NCs; plasma: 120 cases VS. 114 NCs), ESCC (serum: 200 cases VS. 188 NCs; plasma: 178 cases VS. 205 NCs), GCA (serum: 102 cases VS. 84 NCs), GNCA (serum: 203 cases VS. 167 NCs; plasma: 133 cases VS. 109 NCs), COAD (serum: 84 cases VS. 138 NCs; plasma: 65 cases VS. 132 NCs), READ (serum: 112 cases VS. 138 NCs; plasma: 74 cases VS. 132 NCs), PAAD (serum: 159 cases VS. 137 NCs; plasma: 216 cases VS. 220 NCs), LUSC (serum: 102 cases VS. 108 NCs; plasma: 102 cases VS. 101 NCs), LUAD (serum: 170 cases VS. 170 NCs; plasma: 141 cases VS. 124 NCs), BRCA (serum: 216 cases VS. 214 NCs; plasma: 257 cases VS. 257 NCs), UCEC (serum: 92 cases VS. 102 NCs; plasma: 93 cases VS. 79 NCs), PRAD (serum: 86 cases VS. 86 NCs), CESC (serum: 108 cases VS. 108 NCs; plasma: 101 cases VS. 91 NCs), OV (serum: 110 cases VS. 116 NCs; plasma: 101 cases VS. 100 NCs) and UCB (serum: 120 cases VS. 120 NCs; plasma: 128 cases VS. 127 NCs) were analyzed in the study. Additionally, the signature miRNAs showed consistent expression patterns between cancer patients and UCB were identified as oncofetal miRNAs.

2.3 RNA extraction

Total RNA from 200 $\mu$ l plasma or serum was extracted with the mirVana PARIS Kit (Ambion, Austin, TX, USA) following the manufacturer’s protocol. After the addition of denaturing solution (Ambion, Austin, TX, USA), 5 $\mu$ l of synthetic C.elegans miR-39 (5 nM/L, RiboBio, Guangzhou, China) was spiked into each sample for normalization. Total RNA was lysed in 100 $\mu$ l RNase-free water and stored at $-$ 80 ${}^{\circ}$ C. The concentration and purification of the RNA was evaluated using the ultraviolet spectrophotometer Nanodrop 2000 (NanoDrop Technologies, Wilmington, DE, USA).

2.4 Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Amplification of miRNAs was accomplished using Bulge-Loop ${}^{\text{TM}}$ miRNA qRT-PCR Primer Set including specific primers for RT and PCR (RiboBio, Guangzhou, China). RT reactions were performed at 42 ${}^{\circ}$ C for 60 min followed by 70 ${}^{\circ}$ C for 10 min. The qRT-PCR reactions were carried out at 95 ${}^{\circ}$ C for 20 sec, 40 cycles of 95 ${}^{\circ}$ C for 10 sec, 60 ${}^{\circ}$ C for 20 sec and then 70 ${}^{\circ}$ C for 10 sec on 7900HT real-time PCR system (Applied Biosystems, Foster City, CA, USA) or the LightCycler ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ 480 Real-Time PCR System (Roche Diagnostics, Mannheim, Germany). The expression levels of miRNAs were quantified by evaluating the level of fluorescence emitted by SYBR Green dye (SYBR ${}^{\@setsize{\scriptsize}{8pt}{\viipt}{\@viipt}\textregistered}$ Premix Ex Taq ${}^{\text{TM}}$ II, TaKaRa, Dalian, China) and determined using the 2 ${}^{-\Delta\Delta}$ Ct method relative to spiked in cel-miR-39 (The expression levels of plasma miRNAs in GNCA and LUAD assessed as absolute concentration based on standard curve method were also normalized to cel-miR-39).

2.5 Expression profiles of oncofetal miRNAs in tumor tissues

The expression levels of the identified oncofetal miRNAs were evaluated using miRNA isoform data from the TCGA database (http://cancergenome.nih.gov/). As no data of normal ovarian tissues was presented, 13 types of cancers including head neck squamous cell carcinoma (HNSC; 523 cases VS. 44 NCs), THCA (506 cases VS. 59 NCs), ESCA (esophageal cancer; 184 cases VS. 13 NCs), STAD (stomach adenocarcinoma; 436 cases VS. 41 NCs), COAD (444 cases VS. 8 NCs), READ (161 cases VS. 3 NCs), PAAD (178 cases VS. 4 NCs), LUSC (478 cases VS. 45 NCs), LUAD (513 cases VS. 46 NCs), BRCA (578 cases VS. 60 NCs), CESC (307 cases VS. 3 NCs), UCEC (538 cases VS. 33 NCs) and PRAD(494 cases VS. 52 NCs) were analyzed.

2.6 Statistical analysis

The differential expression levels of circulating miRNAs between cases and NCs were assessed by Mann-Whitney test. Multiple logistic regression analysis was used to set up the miRNA panel. ROC curves and the area under the ROC curve were applied to estimate the diagnostic value of the oncofetal miRNAs in detecting specific cancers. Binary logistic regression analyses with ROC curves were performed to evaluate the combination of oncofetal miRNAs in predicting specific types of cancers. SPSS16.0 software (SPSS Inc., Chicago, IL, USA) was used for statistical analyses. A two-sided $P$ value $<$ 0.05 was considered as statistical significance.

2.7 Functional enrichment analysis of the oncofetal miRNAs

MicroRNA Data Integration Portal (mirDIP) version 2 (http://ophid.utoronto.ca/mirDIP) was applied to predict potential target genes of the oncofetal miRNAs identified from our analysis. The interaction network of oncofetal miRNAs and the most overlapping target genes was displayed with Cytoscape 3.6.1 (https://cytoscape.org/). Functional enrichment and similarity analysis was performed on the web server MISIM v2.0 (http://www.lirmed.com/misim/).

3. Results

3.1 Subjects analyzed in the study

According to different sample types, each study (except GCA serum, GNCA serum, OV serum and OV plasma samples) analyzed was divided into four stages to identify dysregulated circulating miRNAs in various cancer patients and UCB compared to NCs (Fig. 1). After the initial screening phase with Exiqon panel, the cases and NCs were divided into three stages: the training stage, the testing stage and the external validation stage. The number of samples in each stage was listed in the Table S1.

Figure 1.

The workflow of each study by different cases and sample types analyzed in the study (except GCA serum, GNCA serum, OV serum and OV plasma samples which were evaluated through training and testing stages after the screening phase). UCB: umbilical cord blood; NC: normal control; NTC: no template control.

3.2 Differentially expressed circulating miRNAs in UCB compared with adult NC

To identify miRNAs with significantly different expression in UCB compared with adult NC, 120 serum and 128 plasma samples isolated from UCB samples were explored respectively in accordance to the experimental workflow. With Exiqon miRNA qPCR panels used in the screening phase, 53 serum miRNAs (42 up-regulated and 11 down-regulated miRNAs; Table S2) and 65 plasma miRNAs (52 up-regulated and 13 down-regulated miRNAs; Table S2) was identified separately in line with the criteria and were subjected to further verification. Consequently, 18 miRNAs including miR-140-5p, miR-144-3p, miR-185-5p, miR-197-3p, miR-19a-3p, miR-19b-3p, miR-20a-5p, miR-20b-5p, miR-210-3p, miR-223-3p, miR-25-3p, miR-382-5p, miR-409-3p, miR-425-5p, miR-483-5p, miR-584-5p, miR-92a-3p and miR-93-5p were confirmed consistently up-regulated in serum samples of UCB compared with adult NC through the training, testing and the external validation stages (Table S3 & Fig. 2; the other miRNAs were shown in Table S4). Meanwhile, sixteen highly expressed plasma miRNAs (miR-106a-5p, miR-130a-3p, miR-140-5p, miR-143-3p, miR-146a-5p, miR-18a, miR-197-3p, miR-20b-5p, miR-210-3p, miR-21-5p, miR-320a, miR-382-5p, miR-409-3p, miR-483-5p, miR-543 and miR-584-5p) were identified in UCB (Table S5 & Fig. 3; the other miRNAs were shown in Table S6).

Figure 2.

Expression levels of the eighteen miRNAs in serum of 120 UCB samples and 120 adult NCs. UCB: umbilical cord blood. ${}^{**}P$ -value $<$ 0.001; ${}^{*}P$ -value $<$ 0.05.

Figure 3.

Expression levels of the sixteen miRNAs in plasma of 128 UCB samples and 127 adult NCs. UCB: umbilical cord blood. ${}^{**}P$ -value $<$ 0.001; ${}^{*}P$ -value $<$ 0.05.

3.3 Circulating miRNAs in pan-cancers

In our previous studies, we demonstrated dysregulated serum miRNAs in ESCC [8], GCA [9], GNCA [10], COAD, READ [11], PAAD [12] LUSC [13] and OV [14], and plasma miRNAs in ESCC [15], GNCA [16], PAAD [17], LUSC [18], LUAD [19], THCA [20], COAD, READ [21] and OV [22] with normalization to different reference genes. To identify circulating miRNAs in pan-cancers, we re-analyzed the data of the serum and plasma miRNAs with the consistent control genes (the spiked in cel-miR-39). Additionally, serum samples of NPC, THCA, LUAD, BRCA, UCEC, PRAD and CESC, and plasma samples of NPC, BRCA, CESC and UCEC were analyzed in accordance with the criteria. In total, serum samples from 15 types of cancers including 2050 patients and plasma samples from 13 types of cancers including 1781 cases with their own age and gender matched NCs were explored to identify circulating miRNAs in pan-cancers.

According to the workflow (GCA serum, GNCA serum, OV serum and OV plasma samples were evaluated through training and testing stages), circulating miRNAs confirmed through the training, testing and the external validation stages by cancer and sample types were determined (Table 1). As shown in Fig. 4A, dysregulated circulating miRNAs identified from serum and plasma samples in the same type of cancers were not identical. However, some miRNAs showed consistent dysregulation in both serum and plasma samples, such as let-7b-5p and miR-140-3p for NPC, miR-106a-5p and miR-93-5p for LUSC, miR-192-5p for PAAD and let-7b-5p for BRCA (“bold” in Fig. 4A). Interestingly, miR-221-3p was down-regulated in serum samples but up-regulated in plasma samples in LUAD. Meanwhile, some miRNAs were dysregulated consistently among different types of pan-cancers. For serum samples, high expression of let-7b-5p, miR-106a-5p, miR-192-5p, miR-195-5p, miR-19a-3p, miR-19b-3p, miR-20a-5p, miR-20b-5p, miR-21-5p, miR-223-3p, miR-25-3p, miR-296-5p, miR-425-5p, miR-92a-3p and miR-93-5p were found in at least two types of pan-cancers. Up-regulation of plasma let-7b-5p, miR-20a-5p, miR-20b-5p and miR-21-5p was confirmed in at least two types of pan-cancers.

Figure 4.

Overview of signature miRNAs with potential for cancer detection. (A): Signature miRNAs presented by different human organs. S: serum; P: plasma; Marked red: oncofetal miRNAs; bold: miRNAs with consistent expression pattern in serum and plasma samples. NPC: nasopharyngeal carcinoma; THCA: thyroid cancer; ESCC: esophageal squamous cell carcinoma; GCA: gastric cardia adenocarcinoma; GNCA: gastric non-cardia adenocarcinoma; COAD: colon adenocarcinoma; READ: rectum adenocarcinoma; PAAD: pancreatic adenocarcinoma; LUSC: lung squamous cell carcinoma; LUAD: lung adenocarcinoma; BRCA: breast cancer; UCEC: uterine corpus endometrial carcinoma; PRAD: prostate adenocarcinoma; CESC: cervical squamous cell carcinoma; OV: ovarian cancer; NA: not available. (B): Circos plots of serum (left) and plasma (right) miRNAs in pan-cancers and UCB. Left circos plot: Eight tracks from outer to inner: 1. locations of serum miRNAs on chromosome; 2. serum miRNAs in UCB (red: oncofetal miRNAs); 3. serum miRNAs in ESCC (green), BRCA (blue) and LUAD (violet); 4. serum miRNAs in UCEC (green) and COAD (blue); 5. serum miRNAs in GNCA (green) and PAAD (blue); 6. serum miRNAs in LUSC (green) and NPC (blue); 7. serum miRNAs in PRAD (green) and THCA (blue); 8. serum miRNAs in READ (green) and GCA (blue). Right circos plot: Six tracks from outer to inner: 1. locations of plasma miRNAs on chromosome; 2. plasma miRNAs in UCB (red: oncofetal miRNAs); 3. plasma miRNAs in LUSC (green) and NPC (blue); 4. plasma miRNAs in LUAD (green), UCEC (blue) and BRCA (violet); 5. plasma miRNAs in PAAD (green) and CESC (blue).

3.4 Identification of circulating oncofetal miRNAs

Among 34 dysregulated miRNAs in pan-cancers and 18 miRNAs in UCB from serum samples, up-regulated miR-19b-3p, miR-20a-5p, miR-223-3p, miR-93-5p, miR-25-3p, miR-425-5p, miR-19a-3p, miR-92a-3p and miR-20b-5p showed consistent trend in expression levels were identified as serum oncofetal miRNAs (marked red in Fig. 4A). Similarly, up-regulated miR-106a-5p, miR-146a-5p, miR-18a-5p, miR-20b-5p, miR-210-3p, miR-21-5p, miR-409-3p and miR-584-5p were demonstrated as plasma oncofetal miRNAs among 32 dysregulated miRNAs in pan-cancers and 16 miRNAs in UCB from plasma samples (marked red in Fig. 4A).

All the miRNAs evaluated in serum and plasma samples of pan-cancers and UCB were visualized in circos plots. As shown in Fig. 4B, dysregulated serum miRNAs including oncofetal miRNAs (red bars) mainly located in chromosome 7, 13 and X. And eight of nine serum oncofetal miRNAs (except miR-223-3p) were members of four miRNA clusters (miR-17-92 cluster: miR-19a-3p, miR-20a-5p, miR-19b-3p and miR-92a-3p; miR-191/425 cluster: miR-425-5p; miR-106a-363 cluster: miR-20b-5p, miR-19b-3p and miR-92a-3p; miR-106b-25 cluster: miR-93-5p and miR-25-3p). On the other hand, mainly dysregulated plasma miRNAs was found on chromosome 11, 13, 17 and X (Fig. 4B). And plasma oncofetal miR-409-3p belongs to miR-379-410 cluster, miR-18a-5p to miR-17-92 cluster, and miR-106a-5p and miR-20b-5p to miR-106a-363 cluster.

3.5 Diagnostic performance of circulating oncofetal miRNAs in cancer patients

To evaluate the diagnostic capacity of circulating oncofetal miRNAs in cancer patients, ROC analyses were conducted with the specific oncofetal miRNA/miRNA panel (more than one oncofetal miRNA in a type of cancer) in the specific cancers. For serum samples, oncofetal miR-223-3p for NPC, panel of miR-92a-3p and miR-25-3p for THCA, signature of miR-19a-3p and miR-425-5p for COAD, combination of miR-19a-3p, miR-425-5p and miR-92a-3p for READ, panel of miR-19a-3p, miR-19b-3p and miR-25-3p for PAAD, signature of miR-20a-5p and miR-93-5p for LUSC, combination of miR-19b-3p, miR-20a-5p, miR-223-3p, miR-93-5p, miR-25-3p, miR-425-5p, miR-19a-3p and miR-92a-3p for BRCA, and panel of miR-92a-3p and miR-20b-5p for UCEC were explored. And the AUCs for the eight specific types of cancers ranged from 0.668–0.987 (Fig. 5A–H). For plasma samples, AUCs of oncofetal miR-20b-5p for NPC, panel of miR-18a-5p and miR-20b-5p for ESCC, miR-210-3p for GNCA, signature of miR-106a-5p and miR-21-5p for LUSC, combination of miR-21-5p, miR-409-3p and miR-584-5p for LUAD, and miR-146a-5p for UCEC ranged from 0.606-0.76 (Fig. 5I–N).

Figure 5.

Receiver-operating characteristic (ROC) curve analyses of the specific serum oncofetal miRNAs (A-H) and plasma oncofetal miRNAs (I-N) to discriminate patients with specific cancers from normal controls. A: miR-223-3p for NPC; B: combination of miR-92a-3p and miR-25-3p for THCA; C: combination of miR-19a-3p and miR-425-5p for COAD; D: combination of miR-19a-3p, miR-425-5p and miR-92a-3p for READ; E: combination of miR-19a-3p, miR-19b-3p and miR-25-3p for PAAD; F: combination of miR-20a-5p and miR-93-5p for LUSC; G: combination of miR-19b-3p, miR-20a-5p, miR-223-3p, miR-93-5p, miR-25-3p, miR-425-5p, miR-19a-3p and miR-92a-3p for BRCA; H: combination of miR-92a-3p and miR-20b-5p for UCEC. I: miR-20b-5p for NPC; J: combination of miR-18a-5p and miR-20b-5p for ESCC; K: miR-210-3p for GNCA; L: combination of miR-106a-5p and miR-21-5p for LUSC; M: combination of miR-21-5p, miR-409-3p and miR-584-5p for LUAD; N: miR-146a-5p for UCEC.

Figure 6.

Expression levels of the identified serum and plasma oncofetal miRNAs in 13 types of tumor tissues comparing with normal tissues using data from TCGA. Presented as log(fold change). HNSC: head neck squamous cell carcinoma; THCA: thyroid cancer; ESCA: esophageal cancer; STAD: stomach adenocarcinoma; COAD: colon adenocarcinoma; READ: rectum adenocarcinoma; PAAD: pancreatic adenocarcinoma; LUSC: lung squamous cell carcinoma; LUAD: lung adenocarcinoma; BRCA: breast cancer; CESC: cervical squamous cell carcinoma; UCEC: uterine corpus endometrial carcinoma; PRAD: prostate adenocarcinoma.

3.6 Assessment of the oncofetal miRNAs in tissues of pan-cancers

To evaluate the expression levels of the oncofetal miRNAs in tumor tissues, miRNA isoform data of 13 types of cancers and matched normal controls were retrieved from the TCGA database. As shown in Fig. 6, except in PAAD and THCA, the majority of the oncofetal miRNAs showed higher expression in tissues of pan-cancers than those in normal tissues.

3.7 Enrichment analysis of the oncofetal miRNAs

To identify genes potentially targeted by the oncofetal miRNAs, mirDIP was assessed. A total of 73 genes were predicted to be common target genes of at least 8 of the 9 serum oncofetal miRNAs. Among them, CPEB3, LCOR, PAPD5, PGM2L1, PTEN and SMG1 could be targeted by all the nine serum oncofetal miRNAs. Meanwhile, among the 28 common target genes (at least 7 of the 8 plasma oncofetal miRNAs), FBXO28, KLF12 and NFAT5 could be targeted by any one plasma oncofetal miRNA (Fig. S1A).

In addition, the web server MISIM v2.0 was used to explore functional enrichment and similarity analysis of the identified oncofetal miRNAs. The associated function, diseases and transcription factors of miRNAs were predicted (Tables S7 and S8; The top 10 terms of function were shown in Fig. S1B). Regardless serum or plasma oncofetal miRNAs could act as onco-miRNAs, play important roles in cancer associated processes such as cell proliferation, apoptosis, and epithelial-to-mesenchymal transition, and could be regulated by well known oncogene MYC, Stat5 and so on. Similarity analyzes showed that miRNAs, especially the serum oncofetal miRNAs, exhibited well functional similarity (Fig. S1C).

4. Discussion

Most of previous studies have made efforts to identified circulating miRNAs with potential in detecting various types of cancers so as to early diagnose the disease and prolong the survival of patients. However, the investigators always focused on single or few types of cancers with different methods in diverse populations and yielded inconsistent findings among studies. Cancer is a category of diseases shared great similarity (e.g. hallmarks of cancer) and heterogeneity. Pan-cancer analyzes could promote us better understand the complex relationship and interactions of the diseases. Nowadays, a global view of circulating miRNAs in pan-cancers with consistent methods in the same population to show the similarity and heterogeneity of circulating miRNAs in the diseases is lacking. In addition, the similarity and interaction between embryo development and tumorigenesis also encouraged us to explore the cancer related circulating miRNAs from the view of embryo. Thus, we performed the comprehensive analysis of circulating miRNAs in UCB and pan-cancers to identify the potential broad-spectrum or/and specific biomarkers for cancer detection.

In the current study, according to different cancer categories as well as sample types, independent exploration of circulating miRNAs was performed in consistent multiple-stages and analytic methods with 2050 serum samples from 15 types of cancers and 1781

Table 1

Expression levels of the signature miRNAs in pan-cancers (presented as mean $\pm$ SD; $\Delta$ CT)

Type (serum)	Number	miRNA	Case		Control		Fold	$P$ value		Type (plasma)	Number	miRNA	Case		Control		Fold	$P$ value
NPC	208 VS. 238									NPC	203 VS. 204
		let-7b-5p	4.71	$\pm$ 6.96	6.09	$\pm$ 2.56	2.61	$<$	0.001			let-7b-5p	3.02	$\pm$ 1.36	4.42	$\pm$ 1.43	2.63	$<$	0.001
		miR-140-3p	7.62	$\pm$ 9.41	9.53	$\pm$ 1.72	3.76	$<$	0.001			miR-140-3p	5.31	$\pm$ 2.16	6.31	$\pm$ 2.04	2.00	$<$	0.001
		miR-192-5p	8.12	$\pm$ 9.79	9.13	$\pm$ 1.94	2.01	$<$	0.001			miR-144-3p	4.38	$\pm$ 1.88	5.16	$\pm$ 2.03	1.72	$<$	0.001
		miR-223-3p	2.61	$\pm$ 4.13	5.27	$\pm$ 1.41	6.29	$<$	0.001			miR-17-5p	1.65	$\pm$ 1.98	3.35	$\pm$ 1.77	3.25	$<$	0.001
		miR-24-3p	5.68	$\pm$ 7.21	7.12	$\pm$ 1.88	2.70	$<$	0.001			miR-205-5p	8.32	$\pm$ 1.97	9.12	$\pm$ 1.99	1.74	$<$	0.001
THCA	100 VS. 96											miR-20a-5p	5.56	$\pm$ 2.29	6.48	$\pm$ 3.47	1.90	$<$	0.001
		miR-25-3p	1.34	$\pm$ 2.55	1.94	$\pm$ 2.5	1.51		0.014			miR-20b-5p	4.8	$\pm$ 1.83	6.33	$\pm$ 1.69	2.89	$<$	0.001
		miR-296-5p	4.31	$\pm$ 2.25	5.02	$\pm$ 2.21	1.64		0.034			miR-451a	2.51	$\pm$ 1.65	3.87	$\pm$ 1.48	2.57	$<$	0.001
		miR-92a-3p	2.14	$\pm$ 1.32	2.84	$\pm$ 1.16	1.62	$<$	0.001	ESCC	178 VS. 205
ESCC	200 VS. 188											miR-18a-5p	10.46	$\pm$ 2.46	11.24	$\pm$ 1.8	1.71		0.011
		miR-192-5p	6.91	$\pm$ 1.99	7.51	$\pm$ 1.54	1.51		0.013			miR-20b-5p	8.63	$\pm$ 1.96	9.33	$\pm$ 1.76	1.63		0.002
		miR-296-5p	8.4	$\pm$ 2.1	9.24	$\pm$ 1.54	1.80	$<$	0.001			miR-486-5p	1.29	$\pm$ 1.52	2.02	$\pm$ 1.58	1.55	$<$	0.001
GCA	102 VS. 84											miR-223-3p	3.82	$\pm$ 2.53	2.17	$\pm$ 1.92	0.32	$<$	0.001
		miR-296-5p	7.28	$\pm$ 1.64	7.87	$\pm$ 1.32	1.51		0.028	GNCA	133 VS. 109
		miR-485-3p	9.12	$\pm$ 2.93	10.1	$\pm$ 2.89	1.94		0.021			miR-185-5p	128	$\pm$ 154	69	$\pm$ 44	1.84	$<$	0.001
GNCA	203 VS.167											miR-20a-5p	5619	$\pm$ 7167	2795	$\pm$ 2340	2.01	$<$	0.001
		miR-195-5p	9.73	$\pm$ 1.44	10.41	$\pm$ 1.14	1.60	$<$	0.001			miR-210-3p	37	$\pm$ 55	14	$\pm$ 85	2.50	$<$	0.001
		miR-296-5p	8.89	$\pm$ 1.76	9.49	$\pm$ 1.71	1.52		0.001			miR-25-3p	1773	$\pm$ 2560	904	$\pm$ 854	1.96	$<$	0.001
COAD	84 VS. 138											miR-92b-3p	31	$\pm$ 33	15	$\pm$ 13	2.04	$<$	0.001
		miR-122-5p	7.93	$\pm$ 1.54	8.53	$\pm$ 1.32	1.52		0.002	PAAD	216 VS. 220
		miR-19a-3p	8.59	$\pm$ 1.46	9.46	$\pm$ 1.73	1.83		0			miR-122-5p	4.25	$\pm$ 2.44	6.85	$\pm$ 1.56	6.08	$<$	0.001
		miR-21-5p	5.15	$\pm$ 1.02	6.12	$\pm$ 0.85	1.96		0			miR-125b-5p	8.92	$\pm$ 2.37	9.51	$\pm$ 2.34	1.51		0.015
		miR-425-5p	8.05	$\pm$ 1.63	8.8	$\pm$ 2.15	1.67		0.029			miR-192-5p	4.8	$\pm$ 2.59	5.57	$\pm$ 2.45	1.70	$<$	0.001
READ	112 VS. 138											miR-193b-3p	9.58	$\pm$ 4.48	10.81	$\pm$ 4.7	2.35	$<$	0.001
		miR-19a-3p	8.79	$\pm$ 1.42	9.46	$\pm$ 1.73	1.59	$<$	0.001	LUSC	102 VS. 101
		miR-21-5p	5.45	$\pm$ 0.93	6.12	$\pm$ 0.85	1.60	$<$	0.001			miR-181a-5p	6.83	$\pm$ 1.51	7.97	$\pm$ 1.34	2.21	$<$	0.001
		miR-425-5p	7.75	$\pm$ 1.93	8.8	$\pm$ 2.15	2.07	$<$	0.001			miR-21-5p	3.12	$\pm$ 1.34	4.34	$\pm$ 1.9	2.34	$<$	0.001
		miR-92a-3p	3.43	$\pm$ 1.2	4.16	$\pm$ 1.21	1.65	$<$	0.001			miR-106a-5p	4.9	$\pm$ 1.45	6.11	$\pm$ 1.54	2.31	$<$	0.001
PAAD	157 VS. 137											miR-93-5p	5.04	$\pm$ 3.14	6.06	$\pm$ 3.08	2.02	$<$	0.001
		let-7b-5p	3.96	$\pm$ 2.02	5.07	$\pm$ 1.28	2.16	$<$	0.001	LUAD	141 VS. 124
		miR-192-5p	6.02	$\pm$ 1.93	7.37	$\pm$ 1.92	2.55	$<$	0.001			miR-19b-3p	576	$\pm$ 566	326	$\pm$ 247	1.76	$<$	0.001
		miR-19a-3p	6.13	$\pm$ 1.95	7.83	$\pm$ 1.14	3.25	$<$	0.001			miR-21-5p	1209	$\pm$ 1371	551	$\pm$ 539	2.20	$<$	0.001
		miR-19b-3p	3.87	$\pm$ 2.03	5.5	$\pm$ 1.31	3.10	$<$	0.001			miR-221-3p	159	$\pm$ 252	60	$\pm$ 68	2.65	$<$	0.001
		miR-223-5p	3.19	$\pm$ 3.48	5.53	$\pm$ 2.92	5.09	$<$	0.001			miR-409-3p	12	$\pm$ 24	5	$\pm$ 13	2.38	$<$	0.001
		miR-25-3p	3.31	$\pm$ 2.17	4.81	$\pm$ 1.09	2.83	$<$	0.001			miR-425-5p	234	$\pm$ 253	148	$\pm$ 70	1.58		0.002
LUSC	102 VS. 108											miR-584-5p	43	$\pm$ 57	18	$\pm$ 21	2.36	$<$	0.001
		miR-106a-5p	7.17	$\pm$ 1.62	8.13	$\pm$ 1.33	1.94	$<$	0.001	BRCA	272 VS. 272
		miR-20a-5p	6.69	$\pm$ 1.55	8.81	$\pm$ 1.36	4.35	$<$	0.001			let-7b-5p	4.66	$\pm$ 1.39	5.3	$\pm$ 1.24	1.56	$<$	0.001
		miR-93-5p	3.96	$\pm$ 1.42	6.1	$\pm$ 1.21	4.43	$<$	0.001	CESC	101 VS. 91

Table 1, continued
Type (serum)	Number	miRNA	Case		Control		Fold	$P$ value		Type (plasma)	Number	miRNA	Case		Control		Fold	$P$ value
LUAD	170 VS. 170											miR-2110	8.85	$\pm$ 3.32	9.46	$\pm$ 2.02	1.53		0.016
		miR-133a-3p	8.28	$\pm$ 1.71	9	$\pm$ 1.29	1.65	$<$	0.001	UCEC	107 VS. 81
		miR-10b-5p	8.85	$\pm$ 2.43	9.55	$\pm$ 1.79	1.62		0.004			miR-142-3p	8.26	$\pm$ 2.14	8.91	$\pm$ 1.72	1.57		0.014
		miR-221-3p	7.17	$\pm$ 1.45	6.54	$\pm$ 0.99	0.65	$<$	0.001			miR-151a-5p	10.47	$\pm$ 2.36	11.48	$\pm$ 2.04	2.01		0.01
BRCA	228 VS. 226											miR-146a-5p	6.5	$\pm$ 2.52	7.47	$\pm$ 2.29	1.96		0.008
		let-7b-5p	5.24	$\pm$ 1.84	7.66	$\pm$ 1.99	5.33	$<$	0.001	THCA	120 VS. 131
		miR-106a-5p	6.44	$\pm$ 1.89	8.3	$\pm$ 1.58	3.61	$<$	0.001			Null
		miR-16-5p	0.94	$\pm$ 1.2	3.81	$\pm$ 1.81	7.32	$<$	0.001	COAD	65 VS. 132
		miR-19a-3p	7.35	$\pm$ 2.12	9.98	$\pm$ 2.07	6.21	$<$	0.001			Null
		miR-19b-3p	4.59	$\pm$ 1.49	7.52	$\pm$ 1.85	7.62	$<$	0.001	READ	74 VS. 132
		miR-20a-5p	7.71	$\pm$ 3.27	8.62	$\pm$ 3.1	1.87	$<$	0.001			Null
		miR-223-3p	3.86	$\pm$ 2.73	5.2	$\pm$ 1.92	2.54	$<$	0.001	OV	101 VS. 100
		miR-25-3p	4.03	$\pm$ 1.42	7.19	$\pm$ 1.75	8.90	$<$	0.001			Null
		miR-425-5p	7.34	$\pm$ 1.49	9.34	$\pm$ 1.53	4.00	$<$	0.001
		miR-451a	4.43	$\pm$ 2.14	7.84	$\pm$ 2.18	10.67	$<$	0.001
		miR-92a-3p	3.41	$\pm$ 2.07	5.56	$\pm$ 2.3	4.46	$<$	0.001
		miR-93-5p	4.66	$\pm$ 1.1	7.65	$\pm$ 1.37	7.94	$<$	0.001
UCEC	92 VS. 102
		miR-143-3p	8.33	$\pm$ 1.42	9.26	$\pm$ 1.49	1.91	$<$	0.001
		miR-151a-3p	8.75	$\pm$ 1.45	9.58	$\pm$ 1.12	1.77	$<$	0.001
		miR-195-5p	7.99	$\pm$ 1.42	8.84	$\pm$ 1.28	1.81	$<$	0.001
		miR-204-5p	9.39	$\pm$ 1.75	10.51	$\pm$ 1.78	2.19	$<$	0.001
		miR-20b-5p	8.62	$\pm$ 1.53	10.5	$\pm$ 2.1	3.68	$<$	0.001
		miR-29c-3p	11.5	$\pm$ 1.55	12.22	$\pm$ 1.76	1.64		0.007
		miR-423-3p	7.89	$\pm$ 1.36	8.55	$\pm$ 1.34	1.58		0.001
		miR-484	4.71	$\pm$ 1.25	5.65	$\pm$ 1.28	1.92	$<$	0.001
		miR-590-5p	11.94	$\pm$ 2.36	12.98	$\pm$ 2.17	2.06	$<$	0.001
		miR-92a-3p	3.27	$\pm$ 1.32	4.47	$\pm$ 1.57	2.30	$<$	0.001
		miR-92b-3p	9.16	$\pm$ 1.09	9.8	$\pm$ 1.14	1.56	$<$	0.001
PRAD	87 VS. 91
		miR-146a-5p	6.18	$\pm$ 3.1	6.91	$\pm$ 3.14	1.65	$<$	0.001
		miR-24-3p	5.83	$\pm$ 3.25	6.58	$\pm$ 3	1.68		0.003
		miR-93-5p	7.06	$\pm$ 3.29	7.76	$\pm$ 3.42	1.63	$<$	0.001
CESC	108 VS. 108
		Null
OV	110 VS. 116
		Null

NPC: nasopharyngeal carcinoma; THCA: thyroid cancer; ESCC: esophageal squamous cell carcinoma; GCA: gastric cardia adenocarcinoma; GNCA: gastric non-cardia adenocarcinoma; COAD: colon adenocarcinoma; READ: rectum adenocarcinoma; PAAD: pancreatic adenocarcinoma; LUSC: lung squamous cell carcinoma; LUAD: lung adenocarcinoma; BRCA: breast cancer; UCEC: uterine corpus endometrial carcinoma; PRAD: prostate adenocarcinoma; CESC: cervical squamous cell carcinoma; OV: ovarian cancer.

plasma samples from 13 types of cancers. In total, 34 serum miRNAs and 32 plasma miRNAs were dysregulated in at least one type of cancer and showed potential in aiding of cancer detection. Analyzes of UCB also demonstrated that 18 serum miRNAs and 16 plasma miRNAs were closely related with embryos. A qualified biomarker should be sensitive and specific to the disease with reliable concentrations to be detected easily. Reportedly, more than 2500 mature human miRNAs have been identified. Concerning sample type, only a small part of miRNome showed relatively high abundance could be stably detected in circulating samples. Thus, we applied Exiqon miRNA qPCR panels to screen potential miRNAs from 168 miRNAs which were evaluated to be stable in serum/plasma with relatively high abundance by the producer (Vedbaek, Denmark) in the initial screening phase of each independent study.

Serum and plasma are two of the most commonly used sample types in clinical diagnosis and investigations. Due to different isolation methods, serum and plasma contain different components including miRNA expression profiles [7, 25]. No study has demonstrated the superiority of application of serum or plasma sample in cancer detection with circulating miRNAs. In our study, except GCA and PRAD, we analyzed the other 13 types of cancers with serum and plasma samples (not matched sample from the same individual) independently. Overall, 15 miRNAs including let-7b-5p, miR-106a-5p, miR-122-5p, miR-140-3p, miR-146a-5p, miR-192-5p, miR-19b-3p, miR-20a-5p, miR-20b-5p, miR-21-5p, miR-25-3p, miR-425-5p, miR-451a, miR-92b-3p and miR-93-5p (miR-221-3p and miR-223-3p showed different expression pattern) were consistently dysregulated in circulation of cancer patients among the 34 serum and 32 plasma miRNAs identified from pan-cancer analyzes (the venn diagram was shown in Fig. S2). For the same sample type, 15 serum miRNAs (let-7b-5p, miR-106a-5p, miR-192-5p, miR-195-5p, miR-19a-3p, miR-19b-3p, miR-20a-5p, miR-20b-5p, miR-21-5p, miR-223-3p, miR-25-3p, miR-296-5p, miR-425-5p, miR-92a-3p and miR-93-5p) and 4 plasma miRNAs (let-7b-5p, miR-20a-5p, miR-20b-5p and miR-21-5p) were respectively found in at least two types of cancers. Among the 13 types of cancers with both serum and plasma samples, NPC (let-7b-5p and miR-140-3p), LUSC (miR-106a-5p and miR-93-5p), PAAD (miR-192-5p) and BRCA (let-7b-5p) possessed consistently dysregulated circulating miRNAs in both types of samples. These miRNAs might show relative commonality among different types of samples and cancers, and might act as broad-spectrum biomarkers to some extent. The other circulating miRNAs might be specific to the type of cancer or/and sample and could be identified as specific biomarkers. The appropriate combination of the broad-spectrum and specific biomarkers could improve the sensitivity and specifity of cancer detection. However, it was noted that except miR-106a-5p in LUSC, let-7b-5p and miR-140-3p in NPC, miR-93-5p in LUSC, miR-192-5p in PAAD and let-7b-5p in BRCA all showed higher fold changes in serum samples than those in plasma samples. Meanwhile, compared to NCs, the mean fold change of 34 serum miRNAs (2.96 $\pm$ 2.1) was also larger than that of 32 plasma miRNAs (2.12 $\pm$ 0.85). The results might indicate the better sensitivity of application of serum miRNAs as diagnostic markers for cancers.

Compared to adult NCs, 18 and 16 miRNAs were identified to be up-regulated in serum and plasma samples of UCB, respectively. Among them, miR-140-5p, miR-197-3p, miR-20b-5p, miR-210-3p, miR-382-5p, miR-409-3p, miR-483-5p and miR-584-5p were consistently up-regulated in both sample types of UCB. Similarity of embryos and tumors prompted us to identify the miRNAs which showed consistent expression pattern between fetus and cancer samples as oncofetal miRNAs. Thus, a total of 16 oncofetal circulating miRNAs including 9 serum oncofetal miRNAs (miR-19b-3p, miR-20a-5p, miR-223-3p, miR-93-5p, miR-25-3p, miR-425-5p, miR-19a-3p, miR-92a-3p and miR-20b-5p) and 8 plasma oncofetal miRNAs (miR-106a-5p, miR-146a-5p, miR-18a-5p, miR-20b-5p, miR-210-3p, miR-21-5p, miR-409-3p and miR-584-5p) were confirmed in the study. It seemed that big divergence existed between the expression profiles of oncofetal miRNAs from serum and plasma samples. Only oncofetal miR-20b-5p was overlapped in the expression profiles from the two types of samples. The diagnostic capacity of oncofetal miRNAs was retrospectively evaluated in specific cancers. The results showed that the specific circulating oncofetal miRNA(s) could effectively distinguish specific types of cancer patients from NCs. However, with larger AUCs, serum oncofetal miRNAs might demonstrate better diagnostic performance than plasma oncofetal miRNAs for cancer patients (e.g. AUCs for NPC: serum: 0.904, plasma: 0.732; LUSC: serum: 0.881, plasma: 0.76; UCEC: serum: 0.765, plasma: 0.614). These results showed difference in application of serum and plasma oncofetal miRNAs in diagnosis of cancer patients.

Among the nine serum oncofetal miRNAs, eight miRNAs were identified as members of four miRNA clusters (miR-17-92, miR-191/425, miR-106a-363 and miR-106b-25 cluster), while 4 of 8 plasma oncofetal miRNAs were confirmed from miR-379-410 cluster, miR-17-92 cluster and miR-106a-363 cluster. As the three paralogous clusters, miR-17-92, miR-106a-363 and miR-106b-25 were highly conserved across species, played indispensable roles in processes of embryonic development and were demonstrated to be prooncogenic in a wide range of malignancies [26, 27, 28]. Application of circulating miRNAs from the three clusters as diagnostic biomarkers was also assessed in several types of cancers [29, 30]. Originated from miR-191/425 cluster, miR-425-5p could act as an oncogene in colorectal cancer and gastric cancer, and might serve as non-invasive indicators for detection of colorectal cancer, gastric cancer and so on [31, 32]. As a member of miR-379-410 cluster and a miRNA expressed by embryonic stem cells, miR-409-3p seemed to play controversial roles in tumorigenesis. Oncogenic roles of miR-409-3p were reported in most types of cancers except in prostate cancer [33, 34, 35]. But high expression of circulating miR-409-3p was common in cancer patients such as individuals with breast cancer or colorectal cancer [36, 37]. These findings showed inconsistency between gene ontology and expression levels of circulating miRNAs which should be paid attention to. The other one serum oncofetal miRNAs, miR-223-3p, also showed discrepancy of gene function in various cancers and even displayed the opposite expression pattern in ESCC plasma samples in our study [38, 39, 40]. Among the other 4 plasma oncofetal miRNAs (miR-146a-5p, miR-210-3p, miR-21-5p and miR-584-5p), miR-21-5p was a well known onco-miRNA and was studied thoroughly in various cancers including its capacity as a cancer biomarker [41]. And miR-21-5p was also essential for embryo implantation and development [42]. MiR-210-3p, a typical hypoxia-related miRNA, might function as an oncogene or tumor suppressor during tumorigenesis. But most evidence showed that circulating miR-210 was up-regulated in many solid tumors including breast cancer, pancreatic cancer, lung cancer and esophageal cancer [43]. Similarly, contradictory results of mechanistic function and expression level of miR-146a-5p in cancers were also described by previous studies [44]. Few studies have explored the roles of miR-584-5p in cancer, thus need to be further investigated. Though controversial roles of some miRNAs in cancer were inconclusive, the oncofetal miRNAs confirmed in our study could broad the view of application of circulating miRNAs in cancer detection.

In addition, we also performed bioinformatic analyzes to evaluate the roles of the serum and plasma oncofetal miRNAs might play. Both the serum and plasma oncofetal miRNAs identified in the study could target many targets. Target predication showed that common targets (at least targeted by n-1 miRNAs) of serum oncofetal miRNAs were much more than those of plasma oncofetal miRNAs. Among the six genes which were targeted by all the nine serum oncofetal miRNAs, PTEN, SMG1 and CPEB3 were confirmed as tumor suppressors in various cancers [45, 46, 47]. As for the three genes commonly targeted by eight plasma oncofetal miRNAs, only KLF12 could act as a tumor suppressor in lung cancer [48]. Majority of serum oncofetal miRNAs from the three paralogous clusters, which could regulate a similar set of genes and have overlapping functions, might partly explain the difference of common target genes between the serum and plasma oncofetal miRNAs. Similarity analysis presented similar findings that functional similarity was found among oncofetal miRNAs, yet more closely among the serum oncofetal miRNAs than those in plasma. However, most of serum and plasma oncofetal miRNAs colud be classified as onco-miRNAs in perspective of functional mechanism. These miRNAs were also closely related with immune system, involved in processes of angiogenesis and cell death, regulated vital pathways (e.g. Akt and NF-kB pathways) both in tumorigenesis and embryo development. Predicted associated diseases for the oncofetal miRNAs also covered most of solid tumors. It might show the universality of the oncofetal miRNAs for pan-cancers. Some well known prooncogenic transcription factors, such as MYC, Stat5 and E2F1, were predicted to be upstream regulators of the oncofetal miRNAs. These results showed the close relationships of oncofetal miRNAs and cancer in mechanism, and could promote us better use the potential diagnostic biomarkers.

Currently, it is not completely clear about the precise mechanism of the release of miRNAs into the extracellular environment. Circulating miRNAs were considered as passive leakage or active secretion with protection of protein complexes or membrane-bound vesicles (e.g. exosomes) by various cells. Once absorbed by recipient cells, circulating miRNAs could elicit corresponding regulatory functions and act as messengers among cells [49]. As cancer is a complex systemic disease, dysregulated circulating miRNAs in cancer were not only derived from tumor cells but also from other cells involved in tumorigenesis and development of cancer. Thus, circulating miRNAs could not only reflect the status of tumor itself but also systemic responses (such as local immune response and systemic inflammation) of cancer patients. Understanding relationship between circulating miRNAs and tumor tissue specific miRNAs is important for us to delineate their roles in cancer pathobiology and cancer prevention. In our previous studies, we have analyzed the expression levels of some identified serum/plasma miRNAs in tumor tissues and demonstrated that some miRNAs were not consistently dysregulated in circulation and tissues. To investigate the expression levels of oncofetal miRNAs in pan-cancers, we retrieved and analyzed miRNA isoform data of 13 types of cancers from the TCGA database. The results showed that except in PAAD and THCA, most of oncofetal miRNAs displayed consistent expression pattern between circulation and tissue in pan-cancers. The findings indicated that the oncofetal miRNAs in circulation might be closely related to tumor. Future efforts should be paid to focus on the mechanism of the circulating miRNAs in cancers.

Over the past decade, few studies have reported oncofetal miRNAs in some types of cancers. Yamamoto et al. identified miR-500 as an oncofetal miRNA in hepatocellular carcinoma, due to its consistently high expression in tumor tissues and fetal liver compared to adult normal liver [50]. Becker-Santos et al. determined 37 oncofetal miRNAs which shared similar expression patterns in fetal lung and NSCLC tumors, versus adult non-malignant lung tissues. And the predicted target genes of oncofetal miRNAs were involved in lung development and carcinogensis [51]. However, the two studies only assessed oncofetal miRNAs from tissue samples in single type of cancer. In the present study, we performed pan-cancer analyzes of circulating miRNAs and identified a set of oncofetal miRNAs in serum/plasma samples. The findings might serve as bases in the field of oncofetal miRNAs and contribute to the potential application of the potential non-invasive biomarkers in the future clinical.

Some issues should be addressed and settled before the former establishment of circulating miRNAs as non-invasive biomarkers for cancer detection in clinical. Firstly, it is critical to choose appropriate sample type for examining circulating miRNAs. Wang et al. performed a meta-analysis to assess the diagnostic performance of circulating miRNAs in detection of gastrointestinal cancer and found that plasma-based miRNA assay reached a higher accuracy than serum-based one in gastric cancer, while opposite conclusion was drawn for colorectal cancer [52]. In our study, regardless of the specific type of cancer, we speculated that serum samples might be superior to plasma samples for miRNAs detection when comprehensively take account in expression levels, AUCs in diagnosis of cancer patients and predicated functional mechanism. Secondly, current methodologies for circulating miRNA quantification were varied among studies. Normalization of miRNA concentrations, in order to minimize the impact of biological and technical variabilities, is important for the reliable quantification of circulating miRNAs in serum and plasma samples. The input sample volume, endogenous controls (e.g. miR-16 and miR-1228) and spike-in controls (e.g. ce-miR-39) were the most frequently used controls for normalization. In our previous studies, we have applied endogenous controls (according to other investigators or evaluated by geNorm software), spike-in controls or combination of the two controls as references for identification of circulating miRNAs. However, it is not comparable for the expression profiles using different normalization genes. It was recommended that the simultaneous processing of all samples collected from identical input volumes with the same spiked-in miRNA concentrations for correcting the variability among circulating samples [6]. Thus, we analyzed the expression profiles of circulating miRNAs with consistent volume and the same spiked-in miRNA cel-miR-39 in the study and compared the circulating miRNAs between pan-cancers and UCB. Meanwhile, each study with different types of cancer/UCB and sample types was performed independently which further make our findings more credible. Thirdly, the specificity and sensitivity of circulating miRNAs for diagnosing patients with different types of cancers and different populations are worth to be further explored. In the study, we identified some miRNAs (especially oncofetal miRNAs) might be broad-spectrum, while the other miRNAs specific for some specific types of cancers in Chinese populations. The appropriate combination of these miRNAs might further improve the diagnostic performance of cancer.

Several limitations of the study should be considered. Firstly, most of serum samples and plasma samples of the same type of cancer/UCB were not from the same individuals. Secondly, the diagnostic capacity of the identified circulating miRNAs especially oncofetal miRNAs was not additionally evaluated in all cancer patients to identify the appropriate combination of the miRNAs and cutoff values for discriminating patients with different types of cancers as well as NCs. Thirdly, the existing form of the identified miRNAs in circulation should be further evaluated. Finally, multicenter studies including more patients with different population backgrounds were needed to be further conducted.

In conclusion, we performed a comprehensive analysis of circulating miRNAs in pan-cancers and UCB, and identified circulating miRNAs (especially oncofetal miRNAs) which could act as potential broad-spectrum or/and specific biomarkers for cancer detection. More efforts should be made before the application of circulating miRNAs for the detection of various cancers in the clinical.

Author contributions

Conception: P.L., W.F.C., and W.Z.

Interpretation or analysis of data: X.Z., C.L., Y.Y.

Preparation of the manuscript: C.Z., X.Z., T.S.X., and X.N.G.

Revision for important intellectual content: X.Z., P.L., W.F.C, and W.Z.

Supervision: W.Z.

Supplementary data

The supplementary files are available to download from http://dx.doi.org/10.3233/CBM-203085.

sj-doc-1-cbm-10.3233_CBM-203085.doc - Supplemental material

Supplemental material, sj-doc-1-cbm-10.3233_CBM-203085.doc

Footnotes

Acknowledgments

This work was supported by the National Natural Science Foundation of China [Grant number: 81672400; 81672788; 81702364] and the Natural Science Foundation of Jiangsu Province [Grant number: BK20171085].

References

Hoadley

K.A.

Yau

Hinoue

Wolf

D.M.

Lazar

A.J.

Drill

Shen

Taylor

A.M.

Cherniack

A.D.

Thorsson

Akbani

Bowlby

Wong

C.K.

Wiznerowicz

Sanchez-Vega

Robertson

A.G.

Schneider

B.G.

Lawrence

M.S.

Noushmehr

Malta

T.M.

Stuart

J.M.

Benz

C.C.

and Laird

P.W.

, Cell-of-origin patterns dominate the molecular classification of 10,000 tumors from 33 types of cancer, Cell 173 (2018), 291–304 e6.

Hanash

S.M.

Baik

C.S.

and Kallioniemi

, Emerging molecular biomarkers-blood-based strategies to detect and monitor cancer, Nat Rev Clin Oncol 8 (2011), 142–150.

Mroczko

Kozlowski

Groblewska

Lukaszewicz

Niklinski

Jelski

Laudanski

Chyczewski

and Szmitkowski

, The diagnostic value of the measurement of matrix metalloproteinase 9 (MMP-9), squamous cell cancer antigen (SCC) and carcinoembryonic antigen (CEA) in the sera of esophageal cancer patients, Clin Chim Acta 389 (2008), 61–66.

Okamura

Takayama

Izumi

Harada

Furuyama

and Nakanishi

, Diagnostic value of CEA and CYFRA 21-1 tumor markers in primary lung cancer, Lung Cancer 80 (2013), 45–49.

Jansson

M.D.

and Lund

A.H.

, MicroRNA and cancer, Mol Oncol 6 (2012), 590–610.

Cheng

, Circulating miRNAs: roles in cancer diagnosis, prognosis and therapy, Adv Drug Deliv Rev 81 (2015), 75–93.

Foye

Yan

I.K.

David

Shukla

Habboush

Chase

Ryland

Kesari

and Patel

, Comparison of miRNA quantitation by Nanostring in serum and plasma samples, PLoS One 12 (2017), e0189165.

Huang

Zhang

Zhu

Shan

Zhou

L.W.

Zhu

Cheng

Zhang

Chen

Zhu

Wang

and Liu

, A novel serum microRNA signature to screen esophageal squamous cell carcinoma, Cancer Med 6 (2017), 109–119.

Wang

Zhang

Zhou

Wang

Zhang

Zhu

and Cheng

, Five serum-based miRNAs were identified as potential diagnostic biomarkers in gastric cardia adenocarcinoma, Cancer Biomark 23 (2018), 193–203.

10.

Huang

Zhu

Zhou

Zhang

Wang

Zhu

Cheng

Chen

Fan

Yin

Zhu

Shu

and Liu

, Six serum-based miRNAs as potential diagnostic biomarkers for gastric cancer, Cancer Epidemiol Biomarkers Prev (2016).

11.

Zhu

Huang

Zhu

Zhou

Shan

L.W.

Cheng

Zhu

Zhang

Chen

Zhu

Wang

and Liu

, A panel of microRNA signature in serum for colorectal cancer diagnosis, Oncotarget 8 (2017), 17081–17091.

12.

Zou

Wei

Huang

Zhou

Zhu

and Miao

, Identification of a six-miRNA panel in serum benefiting pancreatic cancer diagnosis, Cancer Med 8 (2019), 2810–2822.

13.

Zhang

Shan

Wang

Zhu

Huang

Zhang

Zhou

Cheng

Shu

Zhu

and Liu

, A three-microRNA signature for lung squamous cell carcinoma diagnosis in Chinese male patients, Oncotarget 8 (2017), 86897–86907.

14.

Wang

L.R.

Zhou

Liu

Jia

Chen

and Zhu

, Five serum microRNAs for detection and predicting of ovarian cancer, Eur J Obstet Gynecol Reprod Biol X 3 (2019), 100017.

15.

Zhou

Wen

Zhu

Huang

Zhang

L.W.

Shan

Wang

Cheng

Zhu

Yin

Chen

Zhu

Shu

and Liu

, A six-microRNA signature in plasma was identified as a potential biomarker in diagnosis of esophageal squamous cell carcinoma, Oncotarget 8 (2017), 34468–34480.

16.

Zhou

Zhu

Wen

Cheng

Wang

Fan

Chen

Ding

Qian

Huang

Wang

Zhu

Shu

and Liu

, Diagnostic value of a plasma microRNA signature in gastric cancer: a microRNA expression analysis, Sci Rep 5 (2015), 11251.

17.

Zhou

Wang

Huang

Zhu

and Miao

, Plasma miRNAs in diagnosis and prognosis of pancreatic cancer: a miRNA expression analysis, Gene 673 (2018), 181–193.

18.

Shan

Zhang

Zhou

Wang

Zhang

Shu

Zhu

Wen

and Liu

, Identification of four plasma microRNAs as potential biomarkers in the diagnosis of male lung squamous cell carcinoma patients in China, Cancer Med 7 (2018), 2370–2381.

19.

Zhou

Wen

Shan

Zhu

Guo

Cheng

Wang

L.W.

Chen

Huang

Wang

Zhu

Liu

and Shu

, A six-microRNA panel in plasma was identified as a potential biomarker for lung adenocarcinoma diagnosis, Oncotarget (2016).

20.

Wang

Zou

Huang

Zhang

Liu

Jiang

Zhou

and Zhu

, A three plasma microRNA signature for papillary thyroid carcinoma diagnosis in Chinese patients, Gene 693 (2019), 37–45.

21.

Zhang

Zhu

Shan

Zhou

Wang

Zhang

Tao

Cheng

Chen

Liu

Wang

and Zhu

, A panel of seven-miRNA signature in plasma as potential biomarker for colorectal cancer diagnosis, Gene 687 (2019), 246–254.

22.

Wang

Yin

Shan

Zhou

Liu

Cao

Zhu

Zhang

and Zhu

, The value of plasma-based MicroRNAs as diagnostic biomarkers for ovarian cancer, Am J Med Sci (2019).

23.

Monk

and Holding

, Human embryonic genes re-expressed in cancer cells, Oncogene 20 (2001), 8085–8091.

24.

Shah

Patel

Mirza

and Rawal

R.M.

, Unravelling the link between embryogenesis and cancer metastasis, Gene 642 (2018), 447–452.

25.

Shen

Tian

Bai

and Lu

, Profiling circulating microRNAs in maternal serum and plasma, Mol Med Rep 12 (2015), 3323–3330.

26.

Gruszka

and Zakrzewska

, The oncogenic relevance of miR-17-92 cluster and its paralogous miR-106b-25 and miR-106a-363 clusters in brain tumors, Int J Mol Sci 19 (2018).

27.

Mendell

J.T.

, miRiad roles for the miR-17-92 cluster in development and disease, Cell 133 (2008), 217–222.

28.

Khuu

Utheim

T.P.

and Sehic

, The three paralogous MicroRNA clusters in development and disease, miR-17-92, miR-106a-363, and miR-106b-25, Scientifica (Cairo) 2016 (2016), 1379643.

29.

Guo

Liu

and Zhang

, The significance of elevated plasma expression of microRNA 106b∼25 clusters in gastric cancer, PLoS One 12 (2017), e0178427.

30.

Zhou

Xia

Zhou

Huang

Zhang

Zhu

Ding

and Wang

, Circulating microRNAs from the miR-106a-363 cluster on chromosome X as novel diagnostic biomarkers for breast cancer, Breast Cancer Res Treat 170 (2018), 257–270.

31.

Tan

Lin

J.J.

Yang

Gou

D.M.

F.R.

and Yu

X.F.

, A panel of three plasma microRNAs for colorectal cancer diagnosis, Cancer Epidemiol 60 (2019), 67–76.

32.

Zhu

X.L.

Ren

L.F.

Wang

H.P.

Bai

Z.T.

Zhang

Meng

W.B.

Zhu

K.X.

Ding

F.H.

Miao

Yan

Wang

Y.P.

Liu

Y.Q.

Zhou

W.C.

and Li

, Plasma microRNAs as potential new biomarkers for early detection of early gastric cancer, World J Gastroenterol 25 (2019), 1580–1591.

33.

Cao

Zhang

Wei

Jiang

and Jia

, MicroRNA-409-3p represses glioma cell invasion and proliferation by targeting high-mobility group nucleosome-binding domain 5, Oncol Res 25 (2017), 1097–1107.

34.

Ren

Yao

and Tian

, MicroRNA-409-3p regulates cell invasion and metastasis by targeting ZEB1 in breast cancer, IUBMB Life 68 (2016), 394–402.

35.

Josson

Gururajan

Shao

Chu

G.Y.

Zhau

H.E.

Liu

Lao

C.L.

Y.T.

Lichterman

Nandana

Rogatko

Berel

Posadas

E.M.

Fazli

Sareen

and Chung

L.W.

, miR-409-3p/-5p promotes tumorigenesis, epithelial-to-mesenchymal transition, and bone metastasis of human prostate cancer, Clin Cancer Res 20 (2014), 4636–4646.

36.

Cuk

Zucknick

Heil

Madhavan

Schott

Turchinovich

Arlt

Rath

Sohn

Benner

Junkermann

Schneeweiss

and Burwinkel

, Circulating microRNAs in plasma as early detection markers for breast cancer, Int J Cancer 132 (2013), 1602–1612.

37.

Wang

Xiang

Gao

Wang

Chen

and Zhu

, A plasma microRNA panel for early detection of colorectal cancer, Int J Cancer 136 (2015), 152–161.

38.

Fang

Liu

Wang

Huang

Yang

Pang

and Yang

, MicroRNA-223-3p regulates ovarian cancer cell proliferation and invasion by targeting SOX11 expression, Int J Mol Sci 18 (2017).

39.

Luo

Wang

Zhang

Cheng

Zhou

Xie

and Wang

, MiR-223-3p functions as a tumor suppressor in lung squamous cell carcinoma by miR-223-3p-mutant p53 regulatory feedback loop, J Exp Clin Cancer Res 38 (2019), 74.

40.

Chai

Guo

Cui

Liu

Suo

Dou

and Li

, MiR-223-3p promotes the proliferation, invasion and migration of colon cancer cells by negative regulating PRDM1, Am J Transl Res 11 (2019), 4516–4523.

41.

Pfeffer

S.R.

Yang

C.H.

and Pfeffer

L.M.

, The role of miR-21 in cancer, Drug Dev Res 76 (2015), 270–277.

42.

Mondou

Dufort

Gohin

Fournier

and Sirard

M.A.

, Analysis of microRNAs and their precursors in bovine early embryonic development, Mol Hum Reprod 18 (2012), 425–434.

43.

Qin

Furong

and Baosheng

, Multiple functions of hypoxia-regulated miR-210 in cancer, J Exp Clin Cancer Res 33 (2014), 50.

44.

Iacona

J.R.

and Lutz

C.S.

, miR-146a-5p: expression, regulation, and functions in cancer, Wiley Interdiscip Rev RNA 10 (2019), e1533.

45.

Song

M.S.

Salmena

and Pandolfi

P.P.

, The functions and regulation of the PTEN tumour suppressor, Nat Rev Mol Cell Biol 13 (2012), 283–296.

46.

Roberts

T.L.

Luff

Lee

C.S.

Apte

S.H.

MacDonald

K.P.

Raggat

L.J.

Pettit

A.R.

Morrow

C.A.

Waters

M.J.

Chen

Woods

R.G.

Thomas

G.P.

St Pierre

Farah

C.S.

Clarke

R.A.

Brown

J.A.

and Lavin

M.F.

, Smg1 haploinsufficiency predisposes to tumor formation and inflammation, Proc Natl Acad Sci U S A 110 (2013), E285–294.

47.

Chen

Tsai

Y.H.

and Tseng

S.H.

, Regulation of the expression of cytoplasmic polyadenylation element binding proteins for the treatment of cancer, Anticancer Res 36 (2016), 5673–5680.

48.

Godin-Heymann

Brabetz

Murillo

M.M.

Saponaro

Santos

C.R.

Lobley

East

Chakravarty

Matthews

Kelly

Jordan

Castellano

and Downward

, Tumour-suppression function of KLF12 through regulation of anoikis, Oncogene 35 (2016), 3324–3334.

49.

Arroyo

J.D.

Chevillet

J.R.

Kroh

E.M.

Ruf

I.K.

Pritchard

C.C.

Gibson

D.F.

Mitchell

P.S.

Bennett

C.F.

Pogosova-Agadjanyan

E.L.

Stirewalt

D.L.

Tait

J.F.

and Tewari

, Argonaute2 complexes carry a population of circulating microRNAs independent of vesicles in human plasma, Proc Natl Acad Sci U S A 108 (2011), 5003–5008.

50.

Yamamoto

Kosaka

Tanaka

Koizumi

Kanai

Mizutani

Murakami

Kuroda

Miyajima

Kato

and Ochiya

, MicroRNA-500 as a potential diagnostic marker for hepatocellular carcinoma, Biomarkers 14 (2009), 529–538.

51.

Becker-Santos

D.D.

Thu

K.L.

English

J.C.

Pikor

L.A.

Martinez

V.D.

Zhang

Vucic

E.A.

Luk

M.T.

Carraro

Korbelik

Piga

Lhomme

N.M.

Tsay

M.J.

Yee

MacAulay

C.E.

Lam

Lockwood

W.W.

Robinson

W.P.

Jurisica

and Lam

W.L.

, Developmental transcription factor NFIB is a putative target of oncofetal miRNAs and is associated with tumour aggressiveness in lung adenocarcinoma, J Pathol 240 (2016), 161–172.

52.

Wang

Wen

Chen

Luo

Lin

and Xu

,Circulating microRNAs as a novel class of diagnostic biomarkers in gastrointestinal tumors detection: a meta-analysis based on 42 articles, PLoS One 9 (2014), e113401.

Diagnostic value of oncofetal miRNAs in cancers: A comprehensive analysis of circulating miRNAs in pan-cancers and UCB

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.1 Sample collection and preparation

2.2 Study design

2.3 RNA extraction

2.4 Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

2.5 Expression profiles of oncofetal miRNAs in tumor tissues

2.6 Statistical analysis

2.7 Functional enrichment analysis of the oncofetal miRNAs

3. Results

3.1 Subjects analyzed in the study

3.5 Diagnostic performance of circulating oncofetal miRNAs in cancer patients

3.7 Enrichment analysis of the oncofetal miRNAs

4. Discussion

Author contributions

Supplementary data

sj-doc-1-cbm-10.3233_CBM-203085.doc - Supplemental material

Footnotes

Acknowledgments

References