Article Text

Original article
Detection of miR-92a and miR-21 in stool samples as potential screening biomarkers for colorectal cancer and polyps
  1. Chung Wah Wu1,
  2. Simon S M Ng2,
  3. Yu Juan Dong2,
  4. Siew Chien Ng1,
  5. Wing Wa Leung2,
  6. Chung Wa Lee1,
  7. Yee Ni Wong2,
  8. Francis K L Chan1,
  9. Jun Yu1,
  10. Joseph J Y Sung1
  1. 1Institute of Digestive Disease and Department of Medicine and Therapeutics, Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Hong Kong, China
  2. 2Department of Surgery, The Chinese University of Hong Kong, Hong Kong, China
  1. Correspondence to Professor Jun Yu, Institute of Digestive Disease, Department of Medicine and Therapeutics, Room 707, Li Ka Shing Institute of Health Sciences, Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, NT, Hong Kong, China; junyu{at}cuhk.edu.hk

Abstract

Objective The detection of molecular markers in stool samples is a potential strategy for colorectal cancer (CRC) screening. This study evaluated the feasibility of detecting miR-21 and miR-92a in stool samples of patients with CRC or polyps.

Methods The reproducibility of detection and stability of stool-based microRNA were evaluated. Stool samples were collected from 88 patients with CRC, 57 patients with colorectal polyps and 101 healthy controls. MiRNA levels in CRC tissues and stool samples were detected by real-time quantitative reverse transcription PCR. Stool miR-21 and miR-92a levels were compared before and after the removal of tumour or advanced adenoma.

Results The study demonstrated that stool-based miRNA were stable with highly reproducible detection. The expression of miR-21 and miR-92a was significantly higher in CRC tissues compared with their adjacent normal tissues (p<0.0001). Patients with CRC had a significantly higher stool miR-21 level (p<0.01) and miR-92a level (p<0.0001) compared with normal controls. Stool miR-92a, but not miR-21, was significantly higher in patients with polyps than in controls (p<0.0001). At a cut-off value of 435 copies/ng of stool RNA, miR-92a had a sensitivity of 71.6% and 56.1% for CRC and polyp, respectively, and a specificity of 73.3%. In addition, the stool miR-92a level demonstrated a higher sensitivity for distal CRC than proximal CRC (p<0.05), and a higher sensitivity for advanced adenoma than minor polyps (p<0.05). Removal of tumour resulted in reduced stool miR-21 and miR-92a levels (p<0.01), and the removal of advanced adenoma resulted in a reduction of the stool miR-92a level (p<0.05).

Conclusion Stool miRNA are useful for screening CRC and polyps.

  • Colonic polyps
  • colorectal cancer
  • molecular biology
  • non-invasive biomarkers
  • screening
  • stool markers
  • stool microRNA

Statistics from Altmetric.com

Significance of this study

What is already known about this subject?

  • The detection of molecular markers in stool represents a promising non-invasive approach to screen CRC.

  • MiR-21 and miR-92a exhibit enhanced expression in CRC tissues compared with adjacent normal tissues.

  • A high expression level of miR-92a is detectable in plasma from patients with CRC.

What are the new findings?

  • Stool-based miRNA show good stability and highly reproducible detection.

  • The stool-based miR-92a level drops after the removal of tumour or advanced adenoma.

  • Stool-based miR-92a demonstrates moderate sensitivity and specificity for the detection of CRC and colorectal adenoma.

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

  • MiR-92a represents a new non-invasive stool-based molecular marker for the screening of CRC or adenoma.

  • Stool-based miRNA can be used alone or be integrated into currently existing marker panels for higher sensitivity.

Colorectal cancer (CRC) is the third most common cancer worldwide.1 The incidence of CRC is increasing rapidly in Hong Kong, and it represents the second most common malignancy in this locality.2 Colonoscopy is the current gold standard for diagnosing CRC. However, its invasive nature, the cost of equipment and the demand for manpower have hampered the wide application of this procedure. The most widely adopted screening test is the fecal occult blood test (FOBT). This test detects blood in the stool that has leaked from disrupted vessels on the tumour or adenoma surface. FOBT has a low sensitivity, and a substantial proportion of colorectal malignancy may be missed.3

Colonocytes exfoliate consistently and shed into the lumen. Molecular alterations found in tumours can also be detected in stool. The stool DNA test is a non-invasive and established test for the detection of genetic alteration in the stool. A study in an average-risk population showed that the individual marker based on DNA alteration of APC, TP53, KRAS, MSI or DNA integrity has a sensitivity ranging from 3.2% to 25.8% for the detection of CRC. A combined panel of 23 DNA markers has a sensitivity and specificity of 52% and 94%, respectively, for the detection of CRC.3 4 Stool-based messenger RNA is another frequently exploited analyte. Several reports have shown that detecting stool-based mRNA such as cyclin D1,5 cyclo-oxygenase 2 (COX-2), 6–9 or matrix metalloproteinase 7 (MMP-7)8 were able to discriminate CRC patients from healthy individuals. Notably, COX-2 mRNA was reported to be able to detect 26 out of 29 CRC cases (90% sensitivity) with 100% specificity in a Japanese study.9 Although some mRNA markers could achieve high sensitivities, the lack of stability of mRNA in stool samples has limited its wide application. In addition, neoplasm-derived proteins such as minichromosome maintenance proteins,10 carcinoembryonic antigen,9 11 M2 pyruvate kinase12 and secreted clusterin isoform13 in stool samples were also reported to be able to discriminate CRC patients from controls. Among them, stool carcinoembryonic antigen showed a sensitivity of 86% and a specificity of 93% for CRC.11 Compared with the stool DNA test, testing for RNA or protein in stool is relatively less well established, and validations in larger numbers of patients, including patients with adenomas, are warranted.

MicroRNA are 18–25-nucleotide non-coding RNA molecules that regulate the translation of genes. Altered miRNA expression is found in most tumour types including CRC.14–18 Previously, we and others had reported that miR-21 and miR-92a exhibit enhanced expression in CRC tumour tissues compared with their adjacent normal tissues.19 20 We have also demonstrated that miR-92a was detectable at a higher level in plasma of patients with CRC.20 The use of stool-based miRNA markers for the screening of CRC or precancerous polyps is less well established. In two small studies, Ahmed et al21 and Link et al22 demonstrated the feasibility of miRNA detection in stool of patients with CRC. More recently, Koga et al23 have shown reasonable sensitivities and specificities of a panel of miRNA of exfoliated colonocytes isolated from 206 CRC patients and 134 healthy controls. Their study, however, did not include patients with polyps.23 In the current study, we investigated the feasibility of two well-established oncogenic miRNA, miR-21 and miR-92a, in the stool as a screening tool for CRC and premalignant polyps.

Patients and methods

Determining the stability of miRNA in stool samples

Two grams of a freshly collected human stool sample was suspended in phosphate-buffered saline to obtain a 20% stool suspension. The suspension was homogenised by electronic homogeniser. An equal amount of the stool suspension aliquot was transferred to seven 1.5 ml tubes, which were incubated for 0, 12, 24, 36, 48, 60 and 72 h at room temperature. At the end of the incubation period, Trizol LS reagent was added to each aliquot to stop ribonuclease activity. Levels of miR-92a and β-actin mRNA across the incubation period were analysed. The miRNA (miR-92a) or mRNA (β-actin) level at each time point was expressed relative to the level at time 0 and calculated by the ΔCt method. The β-actin mRNA level was quantified using the TaqMan gene expression assay (assay ID: Hs99999903m1; Applied Biosystems, Foster City, California, USA). The stability of miRNA and mRNA in stool samples was compared thereafter.

Testing the reproducibility of miRNA detection in stool samples

To test the reproducibility of miRNA detection in stool, the miR-92a level was repeatedly detected in a set of stool samples (n=17) in two independent experiments separated by 1 month. RNA and stool samples were stored at −80°C between experiments. The correlation of the Ct values was compared.

MiRNA extraction in tissue and stool samples

Frozen tissue of 10–30 mg was added to 0.5 ml Trizol reagent (Invitrogen, Carlsbad, California, USA) in a 1.5 ml tube. The tissue was homogenised by RNase-free pestles and vortexed for 30 s to allow for complete homogenisation; 200 μl of chloroform was added to the 1.5 ml tube.

Fresh human stool sample was collected with a 50 ml specimen cup. Four categories of stool consistency were defined: ‘firm’ (the stool has clear-cut edges, maintains its own shape during handling but deforms with pressure); ‘soft’ (the stool has a uniform consistency but few or less apparent natural edges, and it maintains its own shape but deforms with minimal handling); ‘loose’ (the stool has a semi-solid consistency and can take over the shape of the container); ‘watery’ (no solid pieces, completely liquid). Only ‘firm’, ‘soft’ and ‘loose’ stools were analysed.

Stool samples stored at −80°C were thawed at room temperature for 20 min. Stool of 200–300 mg (wet weight) was added to 1 ml Trizol LS reagent in a 2 ml tube (Invitrogen). Stool sample was homogenised mechanically by RNase-free pestles (USA Scientific, Woodland, California, USA) to deform completely in a 2 ml tube, which was vortexed for 30 s to allow homogenisation in Trizol LS reagent; 300 μl of chloroform was added to the 2 ml tube.

The Trizol–chloroform mixture was vortexed for 15 s and then centrifuged at 12 000g for 15 min at 4°C. The upper aqueous phase was transferred to a fresh 2 ml tube, combined with 100% ethanol of 1.5 times. The volume was mixed thoroughly by pipetting and transferred to an RNeasy mini spin column from the miRNeasy mini kit (Qiagen, Valencia, California, USA). The subsequent total RNA extraction was carried out according to the manufacturer's instructions. Total RNA was eluted in 50 μl nuclease-free water. The RNA concentration was measured with a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific, Wilmington, Delaware, USA).

MiRNA quantitation by real-time qRT–PCR

Quantitative reverse transcription (qRT)–PCR of individual miRNA was performed using the TaqMan miRNA reverse transcription kit (Applied Biosystems) and the TaqMan human MiRNA assay (assay ID: miR-21: 000397, miR-92a: 000431 and RNU6B: 001093) based on a modified protocol. Briefly, 2 ng total RNA, 0.3 μl TaqMan MiRNA RT primer, 3 nmole dNTP (with dTTP), 10 units reverse transcriptase, 0.6 units RNase inhibitor and 0.3 μl RT buffer were used in one RT reaction with a total volume of 3 μl. The thermal cycling condition was as follows: 16°C for 30 min, 42°C for 30 min, 85°C for 5 min and hold in 4°C. The RT product was diluted fourfold by adding 9 μl nuclease free water.

The PCR reaction mix contains 10 μl TaqMan Universal PCR Master Mix with no AmpErase UNG, 0.5 μl miRNA TaqMan assay, 4 μl diluted RT product and 5.5 μl nuclease free water. Real-time PCR was carried out using the 7500 real-time PCR system (Applied Biosystems). The PCR profile was as follow: 95°C for 10 min, 50 cycles of 95°C for 15 s and 60°C for 1 min. Data collection was carried out at the 60°C step. The cycle threshold (Ct) values, which is defined as the number of cycles required for the fluorescent signal to cross the threshold in qRT–PCR, were converted to the absolute number of copies/ng RNA based on standard curves obtained from dilution series of known input quantities of synthetic target miRNA (Sigma Aldrich, St Louis, Missouri, USA). All miRNA expression level was expressed in the number of copies per nanogram of extracted RNA. The detection of tissue and stool-based miRNA was carried out in a blinded fashion.

Subjects and sample collection

Stool samples were collected from 239 subjects including 88 patients with CRC (mean age 67.2±12.0 years), 57 patients with polyps (61.7±7.8 years) and 101 normal controls (60.5±7.0 years) between 2002 and 2011 (table 1). Exclusion criteria included inflammatory bowel disease, a family history of familial adenomatous polyposis or hereditary non-polyposis colon cancer or previous colonic surgery. Normal controls were asymptomatic individuals recruited from a colonoscopy screening programme at the Prince of Wales Hospital, Hong Kong. Patients with polyps were recruited from the endoscopy centre. Stool samples from patients with polyps and normal controls were collected before colonoscopy. Stool samples of CRC patients were collected before surgical resection and at least 2 weeks after the initial colonoscopy. Patients who had undergone adjuvant therapy before surgery were excluded. To investigate the changes in stool-based miRNA levels after the removal of lesion, stool samples were also collected at least 1 month after operation or at least 7 days after the removal of advanced adenoma.

Table 1

Clinicopathological characteristics of subjects

Minor polyp includes hyperplastic polyp or adenoma less than 1 cm in diameter. Advanced adenoma is defined as adenoma 1 cm or greater in diameter, adenoma with more than 25% villous feature, or adenoma with high-grade dysplasia. Proximal lesions include tumours at or proximal to the splenic flexure, and distal lesions are those distal to the splenic flexure. All stool samples were stored at 4°C immediately and transferred to −80°C within 24 h.

Forty pairs of tissue samples were collected from CRC patients. The tumour and adjacent normal samples (at least 4 cm apart from the tumour) were biopsied during initial colonoscopy or during surgical resection. Samples were snap frozen upon collection and stored in a −80°C freezer.

All participants had signed informed consent for obtaining stool or tissue samples. The study protocol was approved by the institutional review board of the Chinese University of Hong Kong and the Hong Kong Hospital Authority.

Statistics

The difference in miRNA levels between paired tissue samples was determined by the Wilcoxon matched-pairs test. The difference in stool miRNA levels between two groups was determined by the non-parametric t test. Correlations between independent samplings and independent qRT–PCR were determined by the Pearson correlation test. The selection of cut-off values was based on the receiver operating characteristics (ROC) curve. The association between miRNA levels and clinicopathological features was analysed by Fisher's exact test; p<0.05 was taken as statistical significance. All the tests were performed by Graphpad Prism 5.0.

Results

Stability of miR-92a detection in stool samples

The stability of miR-92a in stool samples over a 72-h time span was evaluated. Stool miR-92a maintained a high detection level throughout the incubation period, with a 30% reduction from the original level in the first 12 h but showed no further degradation in the following 60 h of incubation. On the other hand, the level of β-actin mRNA degraded to less than 10% of the original concentration in the first 12 h and became undetectable after 72 h (figure 1A).

Figure 1

Stability and reproducibility of stool-based miRNA as biomarkers. (A) Degradation of miRNA and mRNA in stool was compared across time. The remaining level of miR-92a and β-actin mRNA is presented as a ratio to the level at time 0. Data represent mean±SEM of three independent quantitative reverse transcription PCR. (B) Scatter plots showing the association of Ct values from two independent detections of miR-92a on the same stool sample.

Reproducibility of miRNA detection in stool samples

Two independent detections of miR-92a in the same stool samples correlated strongly and yielded a Pearson r=0.82 (p<0.0001; figure 1B).

Detection and normalisation of miRNA levels

For accurate quantitation of stool miRNA levels with qRT–PCR, normalisation is an important step. There is currently no consensus on the method of normalisation for the miRNA level. RNU6B was used as an internal control for normalising the miRNA level in plasma.20 24 However, in a testing set of stool samples (n=30), RNU6B could only be detected in 83.3% (25/30) of stool samples, whereas miR-21 and miR-92a could be detected in all 30 (100%) samples (see supplementary table 1, available online only). In addition, under an equal amount of input RNA and detection threshold, RNU6B was detected at a much lower abundance compared with miR-21 and miR-92a (see supplementary figure 1, available online only). RNU6B may thus not be an ideal internal control for stool-based miRNA. In this regard, based on a previously described method, absolute quantitation with standard curve calibration was adopted for miRNA quantitation in the current study.24

Enhanced expression of miR-21 and miR-92a in CRC tissue samples

Among the 40 pairs of CRC tissue samples, 37 (92.5%) pairs had higher miR-21 expression in tumour than in adjacent normal tissue (p<0.0001; figure 2A), with a median difference of 2.24-fold (IQR 1.57–4.32). Thirty-three (82.5%) pairs had higher miR-92a expression in the tumour than in adjacent normal tissue (p<0.0001; figure 2B), with a median difference of 2.18-fold (IQR 1.23–3.93).

Figure 2

Levels of (A) miR-21a and (B) miR-92a in 40 pairs of colorectal tumours and adjacent normal tissues. The miRNA level is expressed in the number of copies per nanogram of extracted RNA. p Values indicate significant differences in miRNA level between paired samples determined by the Wilcoxon matched pairs test.

Elevated levels of stool miR-21 and miR-92a in CRC and polyp patients

MiR-21 was detected in all 246 (100%) stool samples and miR-92a was detected in 245 (99.6%) stool samples. The stool miR-21 level was significantly higher in CRC patients than in normal controls (p<0.01; figure 3A). No significant difference was found between polyp patients and normal controls. In addition, the stool miR-92a level was significantly higher in CRC (p<0.0001) and polyp patients (p<0.0001) compared with normal controls, (figure 3B), but no significant difference was found between CRC and polyp patients.

Figure 3

Comparison on levels of (A) miR-21 and (B) miR-92a in stool samples from colorectal cancer (CRC) patients (n=88), polyp patients (n=57) and normal controls (n=101). The miRNA level is expressed in the number of copies per nanogram of extracted RNA. The lines denote the medians. Dot lines at the y axis denote cut-off values: 356 000 for miR-21 and 435 for miR-92a. A significant difference between groups was tested by the non-parametric t test. NS denotes no significant difference between groups. Receiver operating characteristics (ROC) curves based on using (C) miR-21 and (D) miR-92a were plotted to discriminate normal and CRC patients. MiR-21 and miR-92a yield an area under the curve (AUC) value of 0.64 and 0.78, respectively.

Sensitivity of stool miR-21 and miR-92a towards the detection of CRC and polyp

The ROC curve based on the miRNA level in CRC and normal samples had an area under the curve value of 0.64 for miR-21 (figure 3C) and 0.78 for miR-92a (figure 3D), respectively. A cut-off value that maximises the sum of sensitivity and specificity was selected. At a cut-off value of 435 copies per nanogram of extracted stool RNA, miR-92a was detected positive in 63 (71.6%) patients with CRC, 32 (56.1%) patients with polyps and 27 (26.7%) healthy controls. The corresponding sensitivities for patients with CRC and polyps were 71.6% and 56.1%, respectively, and the specificity was 73.3% (table 2). At a cut-off value of 356 000 copies/ng of stool RNA, miR-21 was detected positive in 48 (55.7%) patients with CRC, 25 (43.9%) patients with polyps and 27 (26.7%) healthy controls. MiR-21 thus had a sensitivity of 55.7% and 43.9% for CRC and polyps, respectively, and a specificity of 73.3%. Combining the two markers, the test has a sensitivity of 81.8% and 68.4% for CRC and polyp, respectively, and a specificity of 57.4% (table 2).

Table 2

Sensitivity and specificity of stool-based miR-21 and miR-92a

Association of stool-based miR-21 and miR-92a with clinicopathological features

The association between stool-based miR-21/miR-92a and clinicopathological features is shown in table 3. Stool miR-92a, but not miR-21, detected significantly more cancers in the distal region (sensitivity 80.3%) than in the proximal region of the colon (sensitivity 51.9%, p=0.01; table 3 and figure 4A). Stool miR-92a also detected more advanced adenoma (sensitivity 84.6%) than minor polyps (sensitivity 47.7%). No significant association was found between the stool miR-92a and/or miR-21 level with tumour stages or the number of polyps (table 3 and figure 4B).

Table 3

Sensitivities of miRNA based on patient characteristics

Figure 4

Association between stool miR-92a level and lesion location or stage. (A) Scatter plot of stool miR-92a level in different tumour location (proximal colon, n=27; distal colon, n=61) (B) Scatter plot of stool miR-92a level across lesion stages. Groups include normal controls (n=101), minor polyps (MP; n=44), advanced adenoma (AA; n=13), tumour–node–metastasis (TNM) stages 1 and 2 (T1, 2; n=18) and TNM stages 3 and 4 (T3, 4; n=65). The lines denote the medians. A significant difference between groups was tested by the non-parametric t test.

Follow-up on stool miR-21 and miR-92a levels after removal of lesion

Stool miR-21 and miR92a levels were measured in a subgroup of patients with CRC and advanced adenoma after the removal of the lesions. There was a significant drop in miR-21 (p<0.01) and miR-92a (p<0.01, figure 5) levels after the removal of tumour. There was also a significant drop in stool miR-92a but not miR-21 levels after the removal of advanced adenoma (p<0.05, figure 5).

Figure 5

Changes in stool (A) miR-21 and (B) miR-92a levels after removal of tumour or advanced adenoma (AA). p Values indicate significant differences determined by the Wilcoxon matched pairs test.

Discussion

Although it has been well established that colorectal tumours have an altered miRNA expression profile, only few studies have assessed the feasibility of stool-based miRNA as a biomarker for screening for CRC or precancerous lesions. In this study, we have established systematically the feasibility of two miRNA, miR-92a and miR-21, in the stool as non-invasive biomarkers for screening for CRC or premalignant colorectal lesions. Stool-based miRNA have high stability and their detection is reproducible. An ideal class of biomarker should remain stable in the biological sample. MiRNA has been shown to exhibit a higher stability over mRNA. It maintains a high portion of its original level in a 72-h period. This allows for a sufficient amount of time for transfer of the sample to a laboratory for testing, and ensures that subsequent quantitative analysis is less likely to be affected by biomarker degradation. FOBT is widely used to screen for CRC but the degradation of haemoglobin in stool is associated with an increased false negative rate of FOBT.25 26 The exact process that stabilises miRNA in eukaryotic cells is not clearly understood. It has been reported that certain miRNA are protected from degradation by binding to a specific DNA/RNA-binding protein (Translin).27 In plasma, miRNA were found to exist in a form that resists plasma RNase activity.28 It is not known whether the small size of miRNA contributes to its stability, but previous study has shown that differences in stability exist among miRNA of similar sizes in a miRNA-specific manner.29 Stool-based miRNA detection is also highly reproducible. Unlike blood, which is lost intermittently from cancer and precancerous polyps into the stool, colonocytes are released continuously and mixed homogenously with the stool. Repeated sampling on the same stool thus generally results in a similar test result. These features favour the use of stool-based miRNA as screening biomarkers.

Colonocytes exfoliate consistently and swab off onto stool. MiRNA that are elevated in cancerous or even precancerous colonocytes may therefore be detectable at high levels in stool samples. We have shown that miR-92a and miR-21 are overexpressed in colorectal tumour tissues compared with their corresponding adjacent non-tumour tissues. In line with their enhanced expressions in primary tumour specimens, miR-92a and miR-21 were also readily detected at high levels in the stool samples of CRC patients. As both miR-21 and miR-92a are not exclusively expressed in tumour cells but are also expressed at lower levels in normal mucosa, they could therefore be detected in the stool of healthy individuals. Based on ROC curves, we selected cut-off values that best discriminated CRC patients from healthy individuals. MiR-92a was found to have a sensitivity of 71.6% and a specificity of 73.3%, whereas miR-21 had a sensitivity of 55.7% and a specificity of 73.3% for CRC. MiR-92a demonstrated a higher discriminating ability compared with miR-21. Combining the two miRNA markers increased test sensitivity but compromised specificity. The overall performance of the test was not significantly improved because of the over-fitting nature of these two markers. Therefore, as a stand-alone marker, stool miR-92a demonstrated reasonable sensitivity and specificity towards CRC. Recently, Koga et al23 showed that miR-17-92 cluster members and miR-135 in the stool may be useful for CRC screening. In their study, miR-92a showed a sensitivity and specificity of 21.8% and 90.8%, respectively, whereas the expression profile of miR-21 did not differ between CRC patients and healthy controls.15 The discrepancy in the performance of these markers between their study and ours is likely to be attributable to the difference in miRNA extraction and detection protocol. It also remains to be determined if miRNA expression in cancers differs in different ethnic groups as the study be Koga et al23 was based on a Japanese cohort. However, it is noteworthy that both studies identified miR-92a as a better stool marker than miR-21 for CRC screening.

In contrast to previous stool miRNA studies that have focused only on the detection of CRC,23 we have also investigated whether stool-based miR-21 and miR-92a can discriminate patients with polyps from normal controls. Notably, the miR-92a level was significantly higher in stool of polyp patients than in that of normal individuals (p<0.0001; figure 3). More importantly, miR-92a was able to discriminate patients with advanced adenoma from those with any other lower-risk polyps such as hyperplastic polyps or adenomas less than 1 cm in size. A recent study has shown that chromosomal copy number gain at the miR17-92 cluster region (miR-92a is located at this region) was associated with the neoplastic progression from adenoma to adenocarcinoma.30 This observation may partly explain the higher expression of stool-based miR-92a in stools of patients with precancerous lesions. The stool miR-92a level may represent a potentially useful biomarker for the screening of precancerous lesions, especially in high-risk groups including those with a family history of CRC or advanced adenoma. Due to the limited sample size, we were not able to establish an association between miR-92a in the stool and tumour stage. A further association study with a larger sample size of CRC samples with different stages is needed to assess for such a correlation.

MiR-92a demonstrated a higher sensitivity towards distal cancer than proximal cancer (sensitivity 80.3% vs 51.9%, p=0.01). We did not observe a disparity in miR-92a expression between tissue samples obtained from tumour located in the proximal and distal colon (data not shown); therefore, we postulated that the difference in stool miR-92a between proximal and distal cancer may be attributable to degradation during luminal transit. As colonocytes exfoliated from the distal colon have a shorter transit distance in the lumen before being evacuated with the stool, they have less exposure to the cytolytic agents and are more concentrated on the surface, and are more likely to be sampled. In this regard, a higher sensitivity of miR-92a was shown to be associated with CRC in the distal rather than the proximal colon.

To confirm that the enhanced miR-92a and miR-21 levels detected in the stool were derived from the primary malignancy, miRNA were measured in follow-up stool samples collected at least 1 month after surgery or at least 7 days after the removal of advanced adenoma. MiR-92a dropped significantly after the removal of tumour or advanced adenoma, whereas MiR-21 dropped significantly only after the removal of tumour. These findings suggest that the high levels of miR-92a and miR-21 in the stool of patients with CRC are derived from the neoplastic cells. Moreover, miR-92a is likely to be a relevant precancerous polyp marker.

A stool-based miRNA test also exhibits other advantages. It avoids dietary restriction before the test and repetitive testing as required by FOBT. The detection of a single type of variable (miRNA levels) instead of a panel of different types of markers such as DNA mutation or methylation status reduces the amount of sample required for the test and the cost incurred by different detection methods and laboratory expertise. The use of specifically designed primers and probes to detect miRNA in stool is highly sensitive and specific and can allow for the precise detection of miRNA of human origin.

In conclusion, this study has demonstrated the feasibility of using stool miRNA markers as non-invasive tools for the detection of CRC and precancerous polyps. Stool-based miRNA showed good stability and reproducibility. Stool miR-92a has an acceptable sensitivity for the detection of CRC and precancerous polyps. As the current trend in stool-based tests is to integrate multiple heterogeneous markers to increase test sensitivity, this study provides the rationale for further investigations on highly sensitive and specific stool-based miRNA panels. An alternative strategy would be to integrate stool-based miRNA markers such as miR-92a into an already existing molecular marker panel. Such a strategy needs to be validated in future studies.

Acknowledgments

The authors would like to thank Dr Janet FY Lee, Dr KY Ngo, Dr Jimmy CM Li and Ms BY Suen for their assistance in the collection of clinical samples, and Dr Martin CW Chan for optimisation of laboratory techniques.

References

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.

    Files in this Data Supplement:

Footnotes

  • Funding This study received financial support from the Institute of Digestive Disease and Li Ka Shing Institute of Health Sciences.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval This study was conducted with the approval of the institutional review board of the Chinese University of Hong Kong and the Hong Kong Hospital Authority.

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

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.

Linked Articles

  • Digest
    Emad El-Omar William Grady Alexander Gerbes