Article Text

PDF

Original article
Circulating microRNA-203 predicts prognosis and metastasis in human colorectal cancer
  1. Keun Hur1,2,3,
  2. Yuji Toiyama1,4,
  3. Yoshinaga Okugawa1,
  4. Shozo Ide4,
  5. Hiroki Imaoka4,
  6. C Richard Boland1,
  7. Ajay Goel1
  1. 1Center for Gastrointestinal Cancer Research, Center for Epigenetics, Cancer Prevention and Cancer Genomics, Baylor Research Institute and Charles A Sammons Cancer Center, Baylor University Medical Center, Dallas, Texas, USA
  2. 2Department of Biochemistry and Cell Biology, Cell and Matrix Research Institute, School of Medicine, Kyungpook National University, Daegu, Republic of Korea
  3. 3BK21 Plus KNU Biomedical Convergence Program, Department of Biomedical Science, Kyungpook National University, Daegu, Republic of Korea
  4. 4Department of Gastrointestinal and Pediatric Surgery, Mie University Graduate School of Medicine, Tsu, Mie, Japan
  1. Correspondence to Dr Ajay Goel, Center for Gastrointestinal Cancer Research, Center for Epigenetics, Cancer Prevention and Cancer Genomics, Baylor Research Institute, Baylor University Medical Center, 3500 Gaston Avenue, Suite H-250, Dallas TX 75246, USA; ajay.goel{at}baylorhealth.edu

Abstract

Background and aims Distant metastasis is a major cause of deaths in patients with colorectal cancer (CRC), which is partly due to lack of robust metastasis-predictive biomarkers. In spite of the important function of microRNA (miR)-203 in cancer metastasis, its clinical significance in CRC metastasis remains unknown. Here, we evaluated the potential role of serum miR-203 as a non-invasive biomarker for CRC metastasis.

Methods MiR-203 expression was quantified by quantitative reverse-transcription PCR in 58 pairs of primary CRC (pCRC) and corresponding matched liver metastasis (LM), as well as 186 serum and 154 matched tissue specimens from patients with CRC in cohort 1. Next, we performed validation of miR-203 levels in serum from 144 patients with CRC in an independent cohort (cohort 2). Mouse models of CRC-associated metastases were established to identify the source of circulating miR-203. Expression patterns of miR-203 in tissues were determined by in situ hybridisation.

Results MiR-203 expression was significantly upregulated in LM compared with matched pCRC tissues. Serum miR-203 levels were significantly upregulated in a stage-dependent manner, and high miR-203 expression was associated with poor survival in patients with CRC in both patient cohorts. Increased miR-203 levels in serum indicated high risk for poor prognosis (HR=2.1), as well as metastasis to lymph nodes (OR=2.5), liver (OR=6.2), peritoneum (OR=7.2) and distant organs (OR=4.4). Serum miR-203 levels were significantly higher in animals with liver or systemic metastasis compared with controls.

Conclusions High levels of serum miR-203 associate with poor survival and metastasis, suggesting it to be a promising non-invasive prognostic and metastasis-predictive biomarker in patients with CRC.

  • COLON CARCINOGENESIS
  • COLORECTAL CANCER
View Full Text

Statistics from Altmetric.com

Significance of this study

What is already known on this subject?

  • MicroRNA (miR)-203 is involved in epithelial-to-mesenchymal transition in cancer cells.

  • MiRNA biomarkers are emerging as important substrates for developing tissue-based and non-invasive biomarkers for cancer diagnosis and prognosis.

What are the new findings?

  • In contrast to tissue expression, high serum miR-203 expression predicted prognosis as well as metastasis to lymph nodes, liver, peritoneum and distant organs.

  • MiR-203 expression is significantly upregulated in liver metastasis compared with primary colorectal cancer (CRC) tissues.

  • Culture media from CRC cell lines, and animal model of liver and systemic metastasis, demonstrated that miR-203 is a secretory miR, and hence may be exploited as a non-invasive biomarker for predicting metastasis in patients with CRC.

How might it impact on clinical practice in the foreseeable future?

  • Since high levels of serum miR-203 associate with poor survival and metastasis, these results suggest that this miR may serve as a promising non-invasive prognostic and metastasis-predictive biomarker in patients with CRC.

Introduction

Colorectal cancer (CRC) is one of the most frequent cancers, and the second leading cause of cancer-related mortality worldwide.1 Survival rates of patients with CRC have increased in the past few years, possibly as a result of earlier diagnosis and improved treatment regimens; nonetheless, approximately 30%–50% of patients who undergo curative resection subsequently experience local and systemic recurrence.2 Patients with tumour recurrence frequently receive cytotoxic chemotherapeutic regimens coupled with targeted monoclonal antibodies.3 However, the substantial financial costs associated with CRC treatment present an economic burden, and treatment of all patients with chemotherapy without a priori selection leads to overtreatment of patients with toxic agents that produce severe adverse effects.4 In addition, distant metastasis is the major cause of deaths in patients with CRC, and the liver is the most common metastatic site for patients with CRC and approximately 15%–25% of patients with CRC present with liver metastasis (LM) at the time of diagnosis.5 Although resection of hepatic metastases improves 5-year overall survival rates ranging from 25% to 40%, less than one-third of patients with CRC with LMs have a potentially resectable disease at the time of diagnosis.6 In order to overcome these clinical problems, there is a clear need for biomarkers that will facilitate the identification of patients with early recurrence or poor prognosis, and permit earlier diagnosis of the patients with systemic metastases.

Accumulating evidence indicates that circulating microRNAs (miRs) reflect physiological and pathological alterations in patients with cancer, and may serve as important surrogate minimally invasive biomarkers. Indeed, several studies have identified differential expression of key miRs directly involved in the pathogenesis of CRC as circulating diagnostic biomarkers of patients with CRC.7–14 Both miR-17-3p and miR-92 have been shown to be significantly elevated in preoperative plasma from patients with CRC compared with healthy controls, while their levels plummeted in postsurgery plasma specimens.11 Moreover, plasma miR-29a and miR-92a expression successfully differentiate patients with advanced adenomas and CRCs from healthy controls.9 Recently, we and others have found that serum or plasma miR-21 expression is significantly upregulated in patients with adenomas and CRCs compared with healthy subjects, suggesting miR-21 to be a promising non-invasive biomarkers for the early detection of colorectal neoplasia.10 ,13 Although significant gains have been made in terms of developing non-invasive biomarkers for the detection of early-stage lesions, limited evidence exists for the potential significance of circulating miRs as biomarkers for prognosis and predicting metastasis in patients with advanced CRC.

MiRs are the most abundant class of small non-coding RNAs that post-transcriptionally regulate gene expression in multiple cancer-related signalling pathways, including metastasis.15–17 In cancer metastasis, the epithelial–mesenchymal transition (EMT) is a key process that converts polarised immotile epithelial cells into motile, invasive mesenchymal cells, enabling cancer cells to gain stem cell characteristics and an aggressive malignant phenotype.18 ,19 Several miRs have been reported to regulate transcription factors involved in the EMT process.20–22 Recently, we have demonstrated that miR-200c plays a pivotal role in regulating EMT and metastatic behaviour in CRC.23 Moreover, miR-203 has been shown to directly suppress EMT activators such as zinc finger E-box binding homeobox 2 (ZEB2) and SNAI1/2.24–27 However, in spite of the key functional role of miR-203 in cancer metastasis, no previous studies have investigated the clinical significance of miR-203 in CRC metastasis.

In current study, we focused on the clinical significance of serum miR-203 (official name: miR-203a-5p, gene ID: 406986) and its origins in patients with CRC by analysing matched pairs of primary CRC (pCRC) and corresponding LM tissues, followed by analysis of serum and matched tissue specimens from patients with CRC. In addition, we addressed the potential mechanism for the origin of circulating miR-203 levels in a series of in vitro and in vivo experiments. Our systematic approach demonstrated that serum levels of miR-203 are significantly associated with a metastatic phenotype in the colon, and also serve as a potential biomarker for predicting systemic metastases to lymph node, liver and peritoneal cavity and poor prognosis in patients with CRC.

Materials and methods

Clinical specimens and cell lines

This study included examination of 762 serum and tissue specimens, including 58 formalin-fixed, paraffin-embedded (FFPE) pCRC tissues and 58 matched corresponding LM tissues that were enrolled at Okayama University, Toho University and Mie University Medical Hospital in Japan, as described in online supplementary table S1. In addition, 24 healthy controls, 186 preoperative serum samples and 154 matched pCRC tissues (cohort 1) and 144 preoperative serum samples from patients with CRC (cohort 2) were analysed from Mie University Medical Hospital (see online supplementary table S1). Furthermore, for evaluation of disease specificity, 162 preoperative serum samples from gastric cancer (n=113), oesophageal cancer (n=19), ulcerative colitis (UC) (n=9) and another healthy controls (n=21) were also analysed. Careful microdissection was performed to collect and enrich tumour cells from the FFPE tissue specimens. Patients treated with radiotherapy or chemotherapy prior to surgery were not included in this study. Patients with stage III and IV disease received 5-fluorouracil-based chemotherapy, whereas no adjuvant chemotherapy was given to patients with stage I and II CRC. Carcinoembryonic antigen (CEA) expression levels were also determined in the 186 serum samples by standard enzyme immunoassay. Written informed consent was obtained from all patients and the study was approved by the institutional review boards of all participating institutions.

Human CRC cell lines Caco-2, HCT116, HT29, LoVo, SW480 and SW620 were obtained from the American Type Culture Collection (ATCC, Rockville, Maryland, USA) for in vitro and in vivo analysis. These cell lines were maintained in roswell park memorial institute (RPMI)-1640 medium supplemented with 10% fetal bovine serum, 100 IU/mL penicillin and 100 μg/mL streptomycin at 37°C in a 5% humidified CO2 atmosphere. The authenticity of various cell lines was routinely monitored by analysing DNA profiles (short tandem repeats) specific for each cell line in an approved laboratory (last tested on 15 July 2014).

RNA isolation and quantitative reverse-transcription PCR

MiR extraction from serum and culture media samples was performed with miRNeasy RNA isolation kits (Qiagen, Valencia, California, USA), whereas miR extraction from FFPE samples was performed using RecoverAll Total Nucleic Acid Isolation Kits (Ambion, Austin, Texas, USA). TaqMan miR real-time quantitative reverse-transcription PCR (qRT-PCR) (Applied Biosystems, Foster City, California, USA) was used to detect and quantify miR expression using the 2–ΔCt method (for details, see online supplementary methods).

In situ hybridisation

For in situ hybridisation (ISH) analysis, 5 μm thick FFPE tissue sections were hybridised with the miR-203 probe (locked nucleic acid (LNA)-modified and 5′-DIG-labelled and 3′-DIG-labelled oligonucleotide; Exiqon, Woburn, Massachusetts, USA), as described previously.23 Positive control (U6 snRNA, LNA-modified and 5′-DIG-labelled and 3′-DIG-labelled oligonucleotide, Exiqon) and negative control (scrambled miR control, LNA-modified and 5′-DIG-labelled and 3′-DIG-labelled oligonucleotide, Exiqon) were included in each hybridisation reaction.23

In vitro studies and in vivo mice model of liver and systemic metastasis

To determine the secretory potential of miR-203, CRC cell lines were cultured and a fraction of the culture medium was collected at 0, 24 and 48 h after the initial seeding of cells in 10 cm dishes. Additionally, CRC cells (2.0×106/50 µL of phosphate-buffered saline (PBS)) were injected in the spleen of 6-week-old male athymic nude mice (Balb/c nu; Harlan Laboratories) during an open laparotomy to establish in vivo mice model of LM. After 10 weeks, mice were sacrificed. Liver and spleen tissues were resected, fixed in 4% paraformaldehyde, embedded in paraffin and sectioned at 5 µm. With regard to systemic metastasis models, CRC cells (1.0×106/200 µL of PBS) were injected into the tail vein of 6-week-old male athymic nude mice (Balb/c nu; Harlan Laboratories). After 2 months, mice were sacrificed and the degree of systemic metastasis were checked and recorded. The control mice were the ones without establishment of metastasis after injection of HT29 in spleen and tail vein. Thereafter, the tissue sections were stained with H&E. For analysis of serum miR-203 expression, whole blood was collected via cardiac puncture and serum was isolated using Vacutainer tubes with Serum Separator (BD Biosciences, San Jose, California, USA).

All mice were group-housed in standard cages under a controlled temperature (25°C) and photoperiod (12 h light/dark cycle). All experiments were approved by the Baylor Research Institute's Institutional Animal Care and Use Committee.

Statistical analysis

Paired t test, Mann–Whitney test, Kruskal–Wallis test and χ2 tests were used to analyse miR expression data and its relationship with various clinicopathological factors. Kaplan–Meier analysis and the log-rank test were used for survival analysis. Correlation between two continuous values was analysed by Spearman's correlation. The miR expression values were dichotomised into high-expression and low-expression groups based on receiver operating characteristic (ROC) curves with Youden's index correction.28 We estimated that 154 patients were needed to achieve 80% power to substantiate >20% differences in prognostic outcomes (n=184 in cohort 1 and n=144 in cohort 2). Univariate and multivariate Cox's proportional hazards models were used to identify independent prognostic factors dictating patient survival. Univariate and multivariate logistic regression models identified independent predictive factors for distant metastasis, LM, lymph node metastasis and peritoneal metastasis. Data are presented as mean±SD, and all statistical analyses were conducted using the MedCalc V12.3 (Broekstraat, Belgium) and the GraphPad Prism V5.0 (GraphPad Software, San Diego, California, USA).

Results

MiR-203 expression is significantly upregulated in LM compared with matched pCRC tissues

To determine the involvement of miR-203 in CRC metastasis, its expression was analysed in 58 pairs of matched pCRC and corresponding LM tissue specimens (figure 1A). The qRT-PCR analysis revealed that miR-203 expression was significantly upregulated in LM compared with pCRC tissues (mean=0.111 for pCRC vs mean=0.153 for LM; p=0.0002, paired t test).

Figure 1

Expression status of microRNA (miR)-203 in human clinical specimens and colorectal cancer (CRC) cell culture medium. (A) Expression status of miR-203 in 58 pairs of primary CRC (pCRC) and corresponding matching liver metastasis (LM). The grey horizontal bars represent mean expression levels; ***p<0.001, t test. (B) In situ hybridisation analysis of miR-203 expression in pCRC and corresponding LMs (positive control, U6 snRNA; negative control, scrambled miRs control). (C) Expression of miR-203 in CRC cell lines (Caco-2, HCT116, HT29, LoVo, SW480 and SW620). (D) The amount of miR-203 excreted in the HT29 culture medium was increased depending on cell number and duration of culture (E), whereas these levels were decreased following treatment with miR-203 inhibitor in HT29 cells. (F) Serum miR-203 levels in healthy control subjects (N; n=24) and different tumour, lymph node, metastasis (TNM) stages (I–IV) of CRC (n=186) in cohort 1. ***p<0.001, *p<0.05. (G) Tissue miR-203 levels in different TNM stages (I–IV) of CRCs (n=154) and adjacent normal mucosa (N; n=15) in cohort 1. NS, not significant; ***p<0.001.

The pathological expression pattern of miR-203 was confirmed by ISH staining in matched pairs of pCRC and LM tissues (figure 1B). ISH analysis showed that miR-203 expression was higher in the luminal regions compared with the invasive tumour front where metastasis originates in CRC. Of note, miR-203 was strongly expressed in the metastatic foci in liver, while normal hepatocytes barely expressed this miR.

Serum miR-203 is significantly upregulated in patients with CRC with LM

Since miR-203 was highly expressed in LM tissues, we hypothesised that this miR may be detectable in the circulation, which would provide a rationale for the development of a non-invasive biomarker for tumour metastasis. To determine whether miR-203 is a secretory miR, we examined if miR-203 is secreted by CRC cells cultured for various time periods. We first investigated the expression of miR-203 by qRT-PCR in human CRC cell lines (figure 1C). Of the six CRC cell lines, HT29 showed the highest expression level of miR-203. On the basis of these results, HT29 CRC cell line was cultured and miR-203 expression was investigated in the culture medium by qRT-PCR assays (figure 1D). We observed that HT29 cells secreted miR-203 into the culture medium and that its expression was directly proportional to cell number and cell culture duration. Moreover, treatment of miR-203 inhibitor in HT29 cells showed decreased level of miR-203 in the culture medium, suggesting culture medium miR-203 is directly secreted from CRC cells (figure 1E).

Next, the expression of miR-203 was analysed in 186 serum samples as well as 154 matched tissue specimens from patients with CRC in cohort 1, 24 serum samples from healthy controls and 15 adjacent normal colonic mucosa. In comparison with healthy controls, the expression levels of serum miR-203 demonstrated significantly increase in patients with CRC (p<0.001) (figure 1F). In addition, miR-203 expression was significantly upregulated in tumour, lymph node, metastasis (TNM) stage-dependent manner in serum samples obtained from patients with CRC (figure 1F; p=0.0070). On the other hand, although miR-203 expression in CRC was significantly higher than adjacent normal mucosa (figure 1G), tissue miR-203 expression did not change with CRC progression (figure 1G; p=0.2748). Collectively, these results suggest that miR-203 may be released by cancer cells, and high levels of circulating miR-203 could be a reflection of metastasised CRC cells in the systemic circulation, and indicate the specificity of serum miR-203 levels for identification of metastases in patients with CRC. Furthermore, we quantified serum miR-203 expression levels in patients with other gastrointestinal (GI) cancers, including gastric and oesophageal cancers, as well as patients with UC, to investigate the disease specificity of serum miR-203. Interestingly, the expression levels of miR-203 were significantly lower in patients with other diseases versus CRC (see online supplementary figure S1), highlighting that high serum miR-203 levels might be a specific feature of patients with CRC.

High serum miR-203 expression is associated with poor prognosis and metastasis progression in patients with CRC

To further demonstrate the clinical significance of miR-203 expression in patients with CRC, we investigated the association between miR-203 expression and various clinicopathological variables in cohort-1 patients (table 1). In contrast to tissue expression, high serum miR-203 expression was significantly associated with TNM stage (p=0.0071), lymph node metastasis (p=0.0139), distant metastasis (p=0.0008), LM (p=0.0004) and peritoneal metastasis (p=0.0030). Moreover, Kaplan–Meier survival analysis revealed that high miR-203 serum expression was significantly associated with poor survival in patients with CRC (figure 2A; log-rank p<0.0001), whereas tissue miR-203 expression did not associate with patient survival (figure 2B).

Table 1

Clinical significance of tissue and serum miR-203 expression in matched tissue and serum in cohort 1 patients with colorectal cancer*

Figure 2

Kaplan–Meier survival analysis for microRNA (miR)-203 expression in serum and matched tissue specimens from patients with colorectal cancer (CRC), and miR-203 expression status in different classifications of tumour metastasis in cohort 1. Overall survival analyses based on miR-203 expression of 186 serum specimens and 154 matched tissue specimens from patients with CRC. (A) Overall survival of patients with CRC (all stage) based on serum miR-203 expression. (B) Overall survival of all patients with CRC (all stage) based on tissue miR-203 expression. (C) Overall survival of patients with curative CRC (stages I–III) based on serum miR-203 expression. (D) Overall survival of patients with non-curative CRC (stage IV) based on serum miR-203 expression. Serum miR-203 expression levels based on the Japanese Society for Cancer of the Colon and Rectum classification system; (E) lymph node N-classification (N0, absent lymph node metastasis; N1, metastasis in 1–3 regional lymph nodes; N2, metastasis in 4 or more regional lymph nodes; N3, lymph node metastasis to the aorta), (F) liver metastasis (LM) H-classification (H0, no LM; H1, LM with <5 nodules smaller than 5 cm; H2, metastasis that does not involve H1 and H3; H3, LM with >5 metastasis larger than 5 cm) and (G) peritoneal metastasis P-classification (P0, no peritoneal metastasis; P1, metastasis localised to the adjacent peritoneum; P2, limited metastasis to the distant peritoneum; P3, diffuse metastasis to the distant peritoneum).

In subsequent analysis for determining associations between miR-203 expression and N-classification for lymph node involvement (N0, absent lymph node metastasis; N1, metastasis in 1–3 regional lymph nodes; N2, metastasis in ≥4 regional lymph nodes; N3, lymph node metastasis to the aorta), H-classification for LM (H0, no LM; H1, LM with <5 nodules smaller than 5 cm; H2, metastasis that does not involve H1 and H3; H3, indicating LM with >5 metastases larger than 5 cm) and P-classification for peritoneal metastasis (P0, no peritoneal metastasis; P1, metastasis localised to the adjacent peritoneum; P2, limited metastasis to the distant peritoneum; P3, diffuse metastasis to the distant peritoneum), serum miR-203 expression levels were significantly associated with progression of lymph node (figure 2E; p=0.0353), liver (figure 2F; p<0.0001) and peritoneal metastases (figure 2G; p=0.0090), respectively. Our findings first demonstrate that serum miR-203 expression reflects tumour progression and cancer cell dissemination to lymph nodes, distant organs, liver and peritoneal cavity—all the features that directly influence survival rates of patients with CRC. Next, to determine whether the effect of miR-203 on patient survival was affected by the presence of metastases, the prognostic potential of serum miR-203 was separately evaluated in patients with curative (stages I–III) and non-curative (stage IV) CRC. Interestingly, high miR-203 serum expression was significantly associated with poor survival in patients with CRC, regardless of the presence of metastasis (figure 2C; log-rank p=0.0038, figure 2D; log-rank p=0.0147).

Serum miR-203 expression is an independent prognostic and metastasis-predictive biomarker in CRC

Since serum miR-203 expression was associated with specific clinicopathological factors that relate to cancer progression and metastases, we examined whether miR-203 also helped to predict prognosis and metastasis in patients with CRC in cohort 1. ROC analysis revealed that serum miR-203 yielded area under the ROC curve ( AUC ) values of 0.678, with a corresponding sensitivity of 47.50% and a specificity of 85.52% in discriminating patients with poor prognosis in cohort 1 (see online supplementary figure S2A). Univariate Cox's proportional hazards analysis (table 2) revealed that high expression of miR-203 in serum (HR 5.8; 95% CI 3.1 to 10.8; p<0.0001), tumour size (HR 2.3; 95% CI 1.2 to 4.3; p=0.0081), lymph node metastasis (HR 22.4; 95% CI 7.0 to 72.1; p<0.0001), distant metastasis (HR 37.0; 95% CI 15.1 to 90.5; p<0.0001) and CEA expression (HR 5.0; 95% CI 2.2 to 11.3; p=0.0001) were associated with poor prognosis, while tissue miR-203 expression did not associate with any of these clinical factors. In multivariate analysis that included serum miR-203 expression, tumour size, lymph node metastasis, distant metastasis and CEA expression, high serum miR-203 expression (HR 2.1; 95% CI 1.1 to 4.2; p=0.0285) was significantly associated with poor survival, and was independent of other clinical factors. Logistic regression analysis was used to examine the potential of miR-203 to predict lymph node, liver, peritoneal and distant metastases, respectively (table 3). Univariate analyses revealed that serum miR-203 expression associated with lymph node metastasis (OR 2.9; 95% CI 1.4 to 6.1; p=0.0035), liver metastasis (OR 7.3; 95% CI 3.0 to 17.6; p<0.0001), peritoneal metastasis (OR 7.4; 95% CI 2.1 to 26.9; p=0.0022) and distant metastasis (OR 5.3; 95% CI 2.4 to 11.5; p<0.0001), while tissue miR-203 expression failed to associate with any of the metastasis-related factors. Our ROC analyses also revealed that serum miR-203 levels significantly differentiate patients who had CRC with lymph node metastasis, liver metastasis, peritoneal metastasis or distant metastasis, respectively (see online supplementary figure S2B–E). Each AUC value with associated sensitivity and specificity is as follows: lymph node metastasis (AUC=0.607, sensitivity=35.06%, specificity=85.05%; see online supplementary figure S2B), liver metastasis (AUC=0.719, sensitivity=61.54%, specificity=84.18%; see online supplementary figure S2C), peritoneal metastasis (AUC=0.767, sensitivity=72.73%, specificity=76.30%; see online supplementary figure 2SD) and distant metastasis (AUC=0.674, sensitivity=52.50%, specificity=85.42%; see online supplementary figure S2E). In multivariate analyses, serum miR-203 expression was an independent predictor of lymph node metastasis (OR 2.5; 95% CI 1.2 to 5.4; p=0.0172), liver metastasis (OR 6.2; 95% CI 2.1 to 18.3; p=0.0009), peritoneal metastasis (OR 7.2; 95% CI 2.0 to 26.5; p=0.003) and distant metastasis (OR 4.4; 95% CI 1.6 to 11.8; p=0.0037). Collectively, these results suggest that serum miR-203 expression status may be an important biomarker for predicting prognosis and metastases in patients with CRC.

Table 2

Association between microRNA (miR)-203 expression and prognosis in cohort 1 patients with colorectal cancer (CRC)

Table 3

Association between microRNA (miR)-203 expression and metastasis prediction in cohort 1 patients with colorectal cancer (CRC)

Validation of prognostic power and metastasis discriminating power of miR-203 in the validation cohort

To confirm the results from cohort 1 which showed serum miR-203 in CRC as a promising prognostic and metastatic predictive biomarker, we further quantified serum miR-203 levels in another validation cohort (cohort 2), which included 144 patients with CRC. Serum miR-203 levels increased significantly according to tumour stage (see online supplementary figure S3A; p=0.00029). In addition, we confirmed that serum miR-203 levels were significantly higher in patients with CRC with multiple metastases, including lymph node metastasis (see online supplementary figure S3B; p=0.0054), liver metastasis (see online supplementary figure S3C; p=0.029) or peritoneal dissemination (see online supplementary figure S3D; p=0.029), respectively. Kaplan–Meier survival curves showed that high miR-203 serum expression was significantly associated with poor overall survival in all patients with CRC (see online supplementary figure S3E; log-rank p<0.0001). Moreover, subanalyses of overall survival demonstrated that the patients with high miR-203 expression had a significantly worse survival than those with low expression in both curative (see online supplementary figure S3F; log-rank p=0.0001) and non-curative patients (see online supplementary figure S3G; log-rank p<0.0001), respectively. To validate the prognostic power of serum miR-203, we conducted ROC analysis which showed that serum miR-203 yielded an AUC value of 0.748, with 62.5% sensitivity and 77.68% specificity in discriminating patients with poor prognosis in the validation cohort (see online supplementary figure S4A). On the other hand, to confirm the potential diagnostic applicability of serum miR-203 for predicting metastasis development, we also generated ROC curves. Each AUC value with sensitivity and specificity was as follows: lymph node metastasis (AUC=0.636, sensitivity=49.15%, specificity=78.82%; see online supplementary figure S4B), liver metastasis (AUC=0.690, sensitivity=91.67%, specificity=46.97%; see online supplementary figure S4C), peritoneal metastasis (AUC=0.690, sensitivity=91.67%, specificity=46.97%; see online supplementary figure S4D) and distant metastasis (AUC=0.678, sensitivity=95.83%, specificity=34.17%; see online supplementary figure S4E).

Animal model for CRC-associated LM and systemic metastasis demonstrated increased serum miR-203 expression

In order to demonstrate that high levels of miR-203 expression observed in sera and CRC cells metastasised to the liver indeed originated from the metastasized tumour, we established an in vivo LM animal model. We injected HT29 cells in the spleens of animals, which provided unrestricted movement of these colon cancer cells through circulation and establishment in liver tissues as metastatic foci. As illustrated in figure 3, we analysed the expression levels of miR-203 in serum samples as well as liver tissues from these animals. The animals with successful development of metastasis in the liver (figure 3A, B) revealed significantly higher expression of miR-203 in metastatic foci, compared with adjacent regions with normal hepatocytes and spleen (figure 3C; LM vs H, p=0.0006; LM vs S, p<0.0001). More interestingly, miR-203 expression was significantly elevated in serum specimens from animals with CRC-associated LMs compared with controls (figure 3D; p=0.04).

Figure 3

MicroRNA (miR)-203 expression in animal models of liver and systemic metastasis (SM). Athymic nude mice for liver metastatic models were sacrificed 10 weeks after injection of HT29 colorectal cancer (CRC) cells (2.0×106/50 µL of PBS) in the spleen. (A) Representative images of non-treated control mouse. (B) Representative images of animals with established liver metastasis (LM). Yellow arrows indicate metastatic foci. The upper-right panel illustrates representative results of H&E staining in mouse liver. The lower-left panel displays representative results for in situ hybridisation staining for miR-203 in animal liver. (C) MiR-203 expression status in spleen (S), adjacent hepatocytes (H) and LM tissues of LM established mouse group (n=10; ***p<0.001). (D) Serum miR-203 expression status between non-LM (n=7) and animals with established LM (n=10; *p<0.05). Athymic nude mice for systemic metastatic models were sacrificed 8 weeks after injection of HT29 CRC cells (1.0×106/200 µL of PBS) in the tail vein. Representative images of non-SM (E) and established systemic metastatic mouse (F). Red arrows indicate metastatic foci. (G) Serum miR-203 expression status between non-SM (n=6) and animals with established SM (SM; n=6; **p<0.01). (H) Significant correlation between serum miR-203 expression and tumour volume in mice (r=0.829, p=0.0009).

Next, to further confirm the pathological expression of miR-203, we performed ISH staining in tissue samples from the mouse liver tissues (figure 3B). Analogous to human CRC-related LM tissues, animals with LMs also displayed high miR-203 expression compared with adjacent normal hepatocytes.

In addition, to further demonstrate that high serum miR-203 levels were contributed by the metastatic sites, we injected HT29 cells in the tail vein of mice, which provided free movement of these CRC cells through circulation and establishment in distant organs as metastatic foci. As shown in figure 3E and F, systemic metastatic (SM) foci mainly occurred in subcutaneous tissues. We quantified the expression levels of miR-203 in serum samples from non-treated controls, and animals with established metastatic foci, and analysed the correlation between serum miR-203 expression levels and tumour tissues. Consistent with the results of LM models, miR-203 expression was significantly elevated in serum specimens from animals with CRC-associated systemic metastases compared with controls (figure 3G; p=0.0064). Furthermore, significant positive correlation was recognised between serum miR-203 levels and total volume of metastatic nodules (figure 3H; r=0.829; p=0.0009).

Taken together these results indicate that CRC cells, when metastasise to the liver and/or other organs express high levels of miR-203, which may be released into the circulation and serves as a substrate for the development of a non-invasive biomarker for determining prognosis and metastasis in patients with CRC.

Discussion

This study demonstrates the potential role of serum miR-203 as a non-invasive biomarker for CRC metastasis. In this report, we first highlight the clinical significance of serum miR-203 expression for determining patient prognosis and predicting CRC metastasis. Herein, we demonstrate that circulating miR-203 in serum could originate from metastasised cancer cells by directly analysing miR-203 expression in pCRC tissues and matched serum. In addition, we provide further credence to our findings by supporting these results in vitro and in animal models of hepatic and/or systemic metastasis.

Accumulating data in recent years have convincingly demonstrated that the expression of various miRs is frequently dysregulated in CRC tissues compared with normal colonic mucosa. More importantly, recent studies have shown the promise that some of these can also be detected as circulating miRs in the systemic circulation, and their expression pattern in circulation can be directly related with physiological and pathological alterations in patients with CRC.7–14 However, with regards to CRC metastasis, only a few circulating miRs have been reported to potentially play a role in advanced disease. Plasma miR-141 expression is elevated in patients with metastatic stage IV compared with non-metastatic stage I–II CRC.8 More recently, Wang et al14 identified elevated serum miR-29a expression in patients who had CRC with LM compared with those who had CRC without liver metastasis. However, these previous studies did not address the origin of the circulating miRs, nor their clinical utility as possible biomarkers for determining prognosis and predicting distant metastasis.

MiR-203, which is known to be a putative tumour suppressor gene and a target of promoter hypermethylation, has been shown to inhibit cell proliferation and invasion, migration and tumour angiogenesis in a variety of tumour cells.29–31 Although miR-203 expression has been reported in CRC tissues previously, none of the prior studies examined the expression pattern of this miR in detail, or explored its feasibility as a minimally invasive biomarker in serum.32–38 This is the first study that explores the clinical relevance of miR-203 expression in CRC tissues and matched serum specimens from patients with CRC. In the current study, we observed significantly increased expression of miR-203 in CRC tissues compared with normal colonic mucosa and no significant change of miR-203 expression in CRC tissues in a stage-dependent manner. We observed a significant loss of methylation for miR-203 promoter in CRC, but we did not observe any correlation between expression and methylation levels of this miR (data not shown). In contrast, significant increase of miR-203 expression was observed in serum from patients with CRC compared with healthy volunteers, and a simultaneous elevation in miR-203 expression levels in the serum of CRC patients in a stage-dependent manner. In addition, miR-203 expression was significantly upregulated in LMs compared with matched pCRCs, which was consistent with markedly high intensity of miR-203 in the LM in the ISH analysis. Furthermore, high levels of miR-203 in serum were significantly associated with metastases to lymph node, liver, peritoneum and other organs. Interestingly, the increase in serum miR-203 levels occurred in accordance with the N-category in the TNM classification (indicating involvement of lymph nodes) and the H-system and P-system published from the Japanese Society for Cancer of the Colon and Rectum, in which metastases are classified according to the size of maximum metastasis diameter and the number of metastases in liver and peritoneum, respectively. In addition, high levels of serum miR-203 can predict curative CRC with poor prognosis, which indicates that serum miR-203 might be secreted from micrometastatic lesions that were not detected at the time of curative surgery. Furthermore, no correlation was observed between miR-203 expression in serum and matched pCRC tissues, and serum miR-203 levels did not change after primary resection in CRC (data not shown). Collectively, our data from clinical specimens implies that systemic metastatic lesions excrete miR-203 in circulation of patients with CRC.

Another unique strength of our study is that we were able to further substantiate our results from human clinical specimens with an in vivo metastasis animal model. We recognise that although our liver and/or systemic metastasis mouse models may not reproduce the complexity of metastatic CRC in human beings, our animal models result corroborate the clinical data. Furthermore, ISH analysis of the CRC metastases in animal livers displayed a specifically high expression of miR-203 in the metastasised foci in the mouse livers, but not in regions of normal hepatocytes. It is known that miR-203 induces the reversal of the EMT process, that is, mesenchymal–epithelial transition (MET), by directly targeting key EMT regulators such as ZEB2 and SNAI1/2 in breast cancer cells and prostate cells, respectively.24–27 Recently, we found that miR-200c is overexpressed in LM samples compared with matched pCRC, and it is a key regulator of the EMT–MET process.23 Furthermore, we observed high serum miR-200c expression correlated with distant and lymph-node metastasis, independently predicted tumour recurrence (especially with stage II CRC), and was an independent prognostic marker for CRC.39 In the same context, high miR-203 expression in localised metastasised CRC cells may promote metastatic colonisation at secondary sites (eg, lymph node, liver, peritoneum and other distant organs) by inducing MET, and the origin of serum miR-203 from patients with metastasis might be from the metastatic site, from which the cancer cells are secreted abundantly into the systemic circulation in patients with CRC. Future studies may address whether serum miR-203 levels may also be used as a disease monitoring marker for tumour recurrence after curative surgery.

In conclusion, current study demonstrates several novel pieces of evidence associated with serum levels of miR-203 in patients with CRC. First, miR-203 in serum was significantly associated with a metastatic phenotype in CRC and was an independent predictive marker for lymph node, liver and peritoneal metastases in CRC, respectively. In addition, significant diagnostic potential in differentiation of patients with metastatic CRC was evaluated by ROC analysis using independent two cohorts. Second, miR-203 in serum was an independent prognostic marker, whereas the prognostic value of CEA levels was statistically significantly compromised by other clinical factors by the multivariable Cox proportional hazards model. Finally, our data suggest that the source of miR-203 in serum might be contributed by foci of tumour metastasis within lymph nodes, liver, peritoneal space or other distant organs. Therefore, we propose that evaluation of serum miR-203 is a promising clinical tool for identification of patients with CRC with metastasis who need early administration of intensive chemotherapy and/or resection of metastases to gain a survival much longer.

Acknowledgments

We thank Dr Margaret M Hinshelwood for her skilful editing and revision of the manuscript.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • KH and YT contributed equally.

  • Contributors Study concept and design (KH, YT and AG); provision of samples (YT and YO); acquisition of data (KH, YT, SI and HI); analysis and interpretation of data (KH, YT, YO and AG); statistical analysis (KH, YT and YO) and drafting of the manuscript (KH, YT, CRB and AG).

  • Funding The present work was supported by grants R01 CA72851, CA18172, CA184792 and U01 187956 from the National Cancer Institute, National Institutes of Health, funds from the Baylor Research Institute and a pilot grant from Charles A Sammons Cancer Center. This study was supported by the National Research Foundation of Korea (NRF) grant, funded by the Korean government (2014R1A5A2009242).

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval Institutional Review Boards of Toho University and Mie University, Japan, and Baylor Research Institute, Dallas, Texas, USA.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data are already included in the article.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.