MicroRNA Expression Profiling of Exfoliated Colonocytes Isolated from Feces for Colorectal Cancer Screening Free

Colorectal cancer (CRC) is the third leading cause of cancer-related mortality (639,000 deaths in 2004) and cancer-related incidence (1,080,000 new cases in 2004) worldwide (1). Yet, CRC is curable by surgical resection if diagnosed at an early stage. Thus, to reduce the mortality rate of CRC, it is necessary to develop a screening test by which the cancer can be diagnosed at an early stage. To date, the fecal occult blood test has been widely used as the main screening test for CRC (2–4). However, large-scale studies have shown that the sensitivity of fecal occult blood test is not very high, using total colonoscopy as a reference standard in all subjects (5–8). Therefore, other methods have been reported for early detection of CRC, including those based on the detection of mutated DNA (9, 10), cancer-related methylation analysis (11, 12), and DNA integrity (13, 14) in fecal samples. A combination of these analyses has been reported as the stool DNA test (7). Guidelines for screening and monitoring for the early detection of adenomatous polyps and CRC in average-risk adults were recently updated (15). In these guidelines, the stool DNA test was recommended as a CRC screening method; however, the sensitivity and specificity of the stool DNA test was insufficient compared with fecal occult blood test (16). Several studies have reported an attempt to detect CRC by using reverse transcriptase-PCR in fecal samples (17–19). However, there is no evidence that the stool RNA test is useful for CRC screening. The major reason for these inaccuracy is that the nucleic acids in feces are derived from an enormous number and variety of bacteria.

Previously, we reported the presence of viable cancer cells in the feces and that viable cancer cells can be isolated from naturally evacuated feces by using cell isolation methods (20, 21). Moreover, we reported several methods for the detection of early CRC, detection of CRC-related gene mutation using direct sequence analysis (21) and single-strand conformational polymorphism analysis (22), and analysis of CRC-related gene expression (23). The gene expression analysis based on real-time reverse transcriptase-PCR was simple and cost-effective; however, ∼25% of the CRC patients and healthy volunteers could not be detected because of the low concentration or degradation of mRNA.

MicroRNAs (miRNA), which are small (18-25 nucleotides in size) noncoding RNA molecules, regulate the activity of specific mRNA targets and play a major role in cancer. The function of miRNA is the downregulation of multiple target gene expression by degrading the mRNA or blocking its translation into protein through RNA interference (24, 25). Several miRNAs, such as miRNA-21 (miR-21), miR-17-92 cluster, and miR-135, were highly expressed in CRC tissues (26–29). Several recent studies reported that the circulating miRNA in plasma was a potential marker for CRC detection (30, 31) and was remarkably stable in plasma and protected from endogenous RNase activity (32).

To the best of our knowledge, there have been no other studies investigating miRNA expression profiling of fecal-based RNA. In the present study, we analyzed the miRNA expression profile in CRC tissue and exfoliated colonocytes and investigated the potential of miRNA expression analysis of exfoliated colonocytes as a new method for early detection of CRC.

From August 2003 to March 2007, 206 patients with histologically confirmed CRC and 134 healthy volunteers were enrolled in the present study. The healthy volunteers consisted of 59 men and 75 women without any symptoms and evident abnormalities, such as adenoma or carcinoma (including hyperplastic polyps) as determined by screening colonoscopy done at the Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo. The median age of the volunteers was 60 (range 40-70 years). The characteristics of these patients are summarized in Table 1. All the patients had undergone surgical resection of their primary cancer at the National Cancer Center Hospital, Tokyo, Japan. The median age of the patients was 63 (range 32-83 years). There were 138 men and 68 women patients. The primary tumor was located at the following sites: rectum, 101 patients; sigmoid colon, 45 patients; descending colon, 9 patients; transverse colon, 11 patients; ascending colon, 31 patients; cecum, 9 patients. The median diameter of the primary tumor was 38 mm (range 7-160 mm). The clinical stage of the patients according to Dukes' classification was as follows: stage A, 57 patients; stage B, 56 patients; stage C, 79 patients; stage D, 14 patients. All participants were provided with detailed information about the study, and each gave written consent for participating in the study, which was approved by the institutional review board of the National Cancer Center, Japan.

Dynabeads Epithelial Enrich (Dynal), which are commercially available immunomagnetic beads conjugated with EpCAM monoclonal antibody (clone Ber-EP4) and measure 4.5 μm in diameter, were used in the present study. JSR beads (JSR), which are immunomagnetic beads conjugated with EpCAM monoclonal antibody (clone B8-4) and measure 3.0 μm in diameter and reported as optimal immunomagnetic beads for colonocyte isolation from stool (33), were also used in the present study.

Naturally evacuated fecal samples were obtained from 206 CRC patients before they underwent surgical resection. Fecal samples were also obtained from 134 healthy volunteers a few weeks after they had undergone a total colonoscopy. All patients and volunteers were instructed to evacuate the bowel at home, place the sample into a disposable 5 × 10-cm polystyrene tray (AsOne), and bring the sample to the reception counter at the outpatient clinic or the Cancer Prevention and Screening Center of the National Cancer Center. The fecal samples were prepared for the next step immediately after they were brought to our laboratory.

The samples were processed as described previously (21). Briefly, the fecal sample was homogenized with a buffer (40 mL) consisting of Hanks' solution, 10% fetal bovine serum, and 25 mmol/L HEPES buffer (pH 7.35) at 200 rpm for 1 minute using a Stomacher system (Seward). The homogenate was filtered through a nylon filter (pore size 512 μm). After the addition of 80 μL of the immunomagnetic beads, the sample mixture was incubated for 30 minutes under gentle rolling conditions at room temperature. The mixture on the magnet was incubated on a shaking platform for 15 minutes at room temperature (25°C). The supernatant was then removed and the colonocytes in the pellet were stored at −80°C until RNA extraction.

To clarify the difference of the expression of several miRNA between CRC and normal colorectal tissue, fresh tissue samples were obtained from the surgically resected specimens of 31 patients. The samples were frozen in liquid nitrogen within 20 minutes of their arrival at the pathologic specimen reception area and stored in liquid nitrogen until RNA extraction.

Total RNA was extracted from the colonocytes isolated from the fecal samples using an RNeasy Mini Kit (Qiagen), in accordance with the manufacturer's instructions. Each sample was eluted in 100 μL of RNase-free water. Fresh tissue samples were placed in tubes containing ceramic beads and were homogenized using a Precellys 24 device (Bertin Technologies) at 6,500 rpm for 50 seconds. Total RNA was also extracted from each homogenized tissue sample by using an RNeasy Mini Kit.

For miRNA expression analysis, we targeted 10 miRNAs: miR-17-92 cluster (includes miR-17, miR-18a, miR-19a, miR -19b, miR-20a, and miR-92a) and miR-21, miR-135 (miR-135a and miR-135b), and U6 snRNA as an internal control. For all of these miRNAs, we used the commercially available TaqMan MicroRNA Assay (Applied Biosystems).

cDNA was synthesized using the TaqMan MicroRNA RT Kit (Applied Biosystems), in accordance with the manufacturer's instructions. The reaction mixture consisted of 2 μL of total RNA, 0.5 μL of 10× reverse transcriptase buffer, 1 μL of 5× specific primer, 0.05 μL of deoxynucleotide triphosphates (100 mmol/L), 0.06 μL of RNase inhibitor (20 U/μL), and 0.33 μL of MultiScribe Reverse Transcriptase (50 U/μL) in a final reaction volume of 5 μL. The reaction mixture for the analysis consisted of 4 μL of a template cDNA, 10 μL of TaqMan Fast Universal PCR Master Mix (Applied Biosystems), and 1 μL of 20× primer/probe mixture in a total reaction volume of 20 μL. Real-time PCR was done with precycling heat activation at 95°C for 20 seconds, followed by 40 cycles of denaturation at 95°C for 3 seconds, and annealing/extension at 62°C for 30 seconds, in an Applied Biosystems 7500 Fast Real-Time PCR System.

The miRNA expression analysis was conducted using the comparative Ct (threshold cycle) method. In this analysis, the formulae for the relative quantification of each of the genes were as follows: (dCt of each miRNA) = (Ct of each miRNA) − (Ct of U6 snRNA), and (relative quantification of each miRNA) = 2^{−(dCt of each miRNA)}. Differences in relative quantification of the target miRNAs in tissue RNA or colonocyte RNA were analyzed by two-sided Mann Whitney's U test. Sensitivity of detection of the tumor location, tumor size, tumor depth, and Dukes' stage was analyzed by a two-sided Fisher's exact test. Statistical analyses were done using StatView Ver. 5 for Windows (Abacus Concepts). P < 0.05 was considered statistically significant.

The expression of the miR-17-92 cluster, considered as oncogenic miRNA including miR-17, miR-18a, miR-19a, miR-19b, miR-20a, and miR-92a, was significantly higher in CRC tissue compared with normal mucosal tissue (Fig. 1A). The mean expression level of each miRNA in CRC tissue and normal tissue was as follows: miR-17, 0.21 (0-0.62) in CRC tissue and 0.083 (0.020-0.198) in normal tissue (P = 0.0001); miR-18a, 0.0049 (0.0003-0.0307) in CRC tissue and 0.0012 (0-0.0040) in normal tissue (P < 0.0001); miR-19a, 0.0017 (0.0001-0.0130) in CRC tissue and 0.0004 (0-0.0013) in normal tissue (P = 0.0002); miR-19b, 0.0061 (0.0005-0.0416) in CRC tissue and 0.0023 (0.0005-0.0090) in normal tissue (P = 0.01); miR-20a, 0.068 (0.004-0.272) in CRC tissue and 0.022 (0.005-0.069) in normal tissue (P = 0.0003); and miR-92a, 0.65 (0.11-2.48) in CRC tissue and 0.36 (0.10-1.20) in normal tissue (P = 0.01), respectively. Also, the expression of miR-21 and miR-135 (miR-135a and miR-135b), considered as oncogenic miRNA, was significantly higher in CRC tissue compared with normal tissue (Fig. 1B and C). The mean expression level of each miRNA in CRC tissue and normal tissue were as follows: miR-21, 0.23 (0.01-0.88) in CRC tissue and 0.072 (0.008-0.448) in normal tissue (P = 0.0002); miR-135a, 0.0012 (0-0.0064) in CRC tissue and 0.0002 (0-0.0008) in normal tissue (P < 0.0001); and miR-135b, 0.015 (0.001-0.075) in CRC tissue and 0.0014 (0-0.0064) in normal tissue (P < 0.0001).

The detection rate of U6 snRNA was 97.1% (200/206) in the CRC patients and 97.8% (131/134) in the healthy volunteers (Table 2). The detection rate of miR-92a and miR-21 in the CRC patients and healthy volunteers exceeded 90%. Although the detection rate of miR-17, miR-19b, and miR-20a in the CRC patients exceeded 90%, the rate in the healthy volunteers was 60% to 80%. The detection rate of miR-18a, miR-19a, miR-135a, and miR-135b in healthy volunteers was <50%.

Of the 206 CRC patients and 134 healthy volunteers enrolled in this study, six patients and three volunteers were excluded because of undetected U6 snRNA. When the Ct value of U6 snRNA was more than 36 due to the small amount of RNA, such samples were not used for analysis of miRNA expression. Three patients and 12 volunteers were excluded from the miRNA expression analysis. Finally, the relative expression level of each miRNA compared with that of the U6 snRNA internal control was determined for 197 CRC patients and 119 healthy volunteers. The mean relative expression level of miR-17, miR-18a, miR-19a, miR-19b, miR-20a, miR-92a, miR-21, miR-135a, and miR-135b in the CRC patients was 5.60 (range 0-357.05), 0.43 (0-38.40), 1.11 (0-58.97), 4.01 (0-185.72), 2.44 (0-148.57), 9.38 (0-460.16), 25.26 (0-1206.82), 0.02 (0-2.46), and 0.29 (0-28.54), respectively, and the corresponding relative expression level in the healthy volunteers was 1.06 (range 0-33.11), 0.06 (0-3.29), 0.34 (0-10.45), 1.65 (0-49.87), 0.31 (0-10.05), 1.13 (0-42.61), 11.59 (0-484.38), 1.26 × 10⁻⁹ (0-5.07 × 10⁻⁷), and 0.02 (0-1.01), respectively. Although there was no significant difference in the relative expression level of miR-21 between the CRC patients and healthy volunteers (P = 0.6), significant differences were observed on the relative expression level of the miR-17-92 cluster and miR-135 between the patient and volunteer groups (P < 0.001; Table 3).

According to the expression analysis of nine miRNAs, the overall sensitivity of patients and specificity of healthy volunteers was 74.1% [146/197; 95% confidence interval (95% CI), 67.4-80.1] and 79.0% (94/119; 95% CI, 70.6-85.9), respectively (Table 4). The sensitivity obtained using the miR-17-92 cluster, miR-21, and miR-135 was 69.5% (137/197; 95% CI, 62.6-75.9), 14.7% (29/197; 95% CI, 10.1-20.5), and 46.2% (91/197; 95% CI, 39.1-53.5), respectively. The specificity obtained using the miR-17-92 cluster, miR-21, and miR-135 was 81.5% (97/119; 95% CI, 83.3-88.0), 91.6% (109/119; 95% CI, 85.1-95.9), and 95.0% (113/119; 95% CI, 89.3-98.1), respectively.

The sensitivity of detection of patients with cancers located on the right side of the colon and those located on the left side of the colon was 52.9% (27/51; 95% CI, 38.4-67.1) and 81.5% (119/146; 95% CI, 74.3-87.4), respectively (Table 5). There was a significant difference in sensitivity between the right and left side of the colon (P = 0.0001). On the other hand, there were no significant differences in sensitivity of detection among tumors of varying size (P = 0.32), invasion depth (P = 0.39), and Dukes' stage classification (P = 0.26).

In the present study, we reported that the sensitivity and specificity of miRNA expression assay was 74% and 79%, respectively (Table 4). To date, we have reported several CRC screening methods using colonocytes isolated from feces (21–23). The sensitivity and specificity of the cancer-related gene mutation analysis were relatively high (71% and 88%, respectively); however, the cost of direct sequencing analysis was expensive and the procedure was complicated. On the other hand, the cost of real-time reverse transcriptase-PCR was inexpensive and the procedure was simple; however, ∼25% of the subjects were excluded from gene expression analysis due to small concentration of total RNA. In the present miRNA expression assay, ∼97% of all subjects were detected (Table 2). This means that small RNAs, including miRNAs, in exfoliated colonocytes may be conserved in greater number compared with mRNA in exfoliated colonocytes, and the miRNA expression assay may be more useful as a CRC screening method compared with the gene expression assay.

Recently, several oncogenic miRNAs and tumor-suppressor miRNAs were reported. let7, miR-34 family, miR-126, miR-143, miR-145, and miR-200 family are considered to be tumor-suppressor miRNAs in CRC (34), inhibiting several oncogenes (35–38). The expression level of tumor-suppressor miRNAs in cancer tissue was lower than that in normal tissue. On the other hand, the miR-17-92 cluster, miR-21, and miR-135 are considered to be oncogenic miRNAs in CRC (34), inhibiting several tumor-suppressor genes (29, 39–42). The expression level of the oncogenic miRNAs in cancer tissue was higher than that in normal tissue. Although the detection of oncogenic miRNAs in serum could be used for CRC biomarkers (30, 43), this would not be possible for cancer screening because the miRNAs secreted from cancer cells were very few in serum. In addition, the optimal internal control miRNA of serum circulating miRNA has not yet been determined. In our miRNA expression assay, exfoliated colonocytes were used and three internal control miRNAs—U6 snRNA, RNU6B, and miR-16—were analyzed. Subsequently, we concluded that U6 snRNA was the optimal internal control miRNA for colonocytes isolated from feces (data not shown).

The controls were obtained from asymptomatic patients at screening colonoscopy; therefore, one drawback of this study is that patients with inflammatory conditions have not been tested to evaluate potential false positives. However, the sensitivity of miRNA expression assay was 74% and specificity was <80% (Table 4). These oncogenic miRNAs were highly expressed not only in CRC patients but also in healthy volunteers because they are related to carcinogenesis but not to cancer. The reason for the low specificity is difficult to explain. There is a possibility that these oncogenic miRNAs were already highly expressed in healthy subjects; thus, the specificity of miRNA expression assay was relatively lower. Another possible reason may be that there are some high-risk groups for CRC among the healthy volunteers. We have to perform follow-up on these false-positive healthy subjects to monitor for newly developed CRC.

We were able to detect cases with resectable CRC as well as those with advanced CRC (Table 5). Detection at an early stage is important in reducing the rate of mortality from CRC. Meanwhile, the detection rate for patients with the primary tumor located on the right side of the colon was lower than that for patients with the primary tumor located on the left side of the colon. A possible reason for this is that the exfoliated cancer cells from the right colon are exposed to the feces for a longer time than those from the left colon, resulting in a reduced number of surviving cells in the feces. Given that colon cancer incidence is increasing on the right side, we need to develop a new method that improves the recovery of exfoliated cancer cells even from cancer on the right side of the colon.

In this study, we reported a new CRC screening method using miRNA expression profiling of exfoliated colonocytes. In the next step, we have to detect more useful oncogenic miRNAs by analyzing exfoliated colonocytes in feces from CRC patients or healthy volunteers. Our miRNA expression assay using exfoliated colonocytes should be further evaluated for future CRC screening.

No potential conflicts of interest were disclosed.

We thank S. Miyaki, K. Inoue, J. Izumisawa, Y. Ishihara, and H. Koike for their technical assistance, and K. Shiina and K. Abe for their secretarial assistance.

Grant Support: Grant-in-Aid for Young Scientists (B) from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (Y. Koga); Program for Promotion of Fundamental Studies in Health Sciences of the National Institute of Biomedical Innovation of Japan (Y. Matsumura); Innovation Promotion Program from the New Energy and Industrial Technology Development Organization of Japan (Y. Matsumura); Mitsui Life Social Welfare Foundation (Y. Matsumura); and the Regional Innovation Cluster Program (Y. Matsumura).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.