Purpose: Esophageal adenocarcinoma is a highly aggressive malignancy that frequently develops from Barrett's esophagus, a premalignant pathologic change occurring in the lower end of the esophagus. Identifying Barrett's esophagus patients at high risk of malignant transformation is essential to the prevention of esophageal adenocarcinoma. Although microRNA (miRNA) expression signatures have been associated with the etiology and prognosis of several types of cancers, their roles in the development of esophageal adenocarcinoma have not been extensively evaluated.

Experimental Design: In this study, we analyzed the expression patterns of 470 human miRNAs using Agilent miRNA microarray in 32 disease/normal-paired tissues from 16 patients diagnosed with Barrett's esophagus of either low- or high-grade dysplasia, or esophageal adenocarcinoma.

Results: Using unsupervised hierarchical clustering and class comparison analyses, we found that miRNA expression profiles in tissues of Barrett's esophagus with high-grade dysplasia were significantly different from their corresponding normal tissues. Similar findings were observed for esophageal adenocarcinoma, but not for Barrett's esophagus with low-grade dysplasia. The expression patterns of selected miRNAs were further validated using quantitative reverse transcription real-time PCR in an independent set of 75 pairs of disease/normal tissues. Finally, we identified several miRNAs that were involved in the progressions from low grade-dysplasia Barrett's esophagus to esophageal adenocarcinoma.

Conclusions: We showed that miRNAs were involved in the development and progression of esophageal adenocarcinoma. The identified significant miRNAs that may become potential targets for early detection, chemoprevention, and treatment of esophageal cancer. (Clin Cancer Res 2009;15(18):5744–52)

Translational Relevance

This is the first study to show that specific microRNA expression signatures are associated with the progression from Barrett's esophagus to esophageal adenocarcinoma. The validation and incorporation of the identified significant miRNAs with the currently available clinical variables may further stratify Barrett's esophagus patients, which will allow for the selection of those with the highest risk of malignant progression to receive targeted chemoprevention and more aggressive therapies.

Esophageal cancer is the sixth most common cause of cancer-associated death worldwide (1). In the United States, especially in Caucasian males, both the incidence and the mortality of esophageal cancer have been steadily increasing during the past several decades (2, 3). Despite the wide application of radical esophagectomy and systemic chemoradiotherapy, the overall 5-year survival rate of esophageal cancer remains <20%, mainly due to the fact that a large proportion of patients are diagnosed at an advanced stage (2, 4). Therefore, it is critical for the control and treatment of this dreaded malignancy to identify clinically applicable biomarkers for early detection and targeted prevention.

There are two major histologic types of esophageal cancer: esophageal squamous cell carcinoma and esophageal adenocarcinoma. Esophageal adenocarcinoma is one of the fastest growing malignancies in the United States with a 6-fold increased incidence in the past three decades (3, 5, 6). Esophageal adenocarcinoma is usually associated with symptomatic gastroesophageal reflux disease and frequently develops from Barrett's esophagus, a condition of pathologic metaplasia or dysplasia in which the epithelial cells in the lining esophageal mucosa are replaced by premalignant columnar epithelium (7, 8). Patients with Barrett's esophagus are at 30- to 125-fold increased risk of developing esophageal adenocarcinoma (57, 9). The malignant transformation rate of Barrett's esophagus varies, depending on the presence of dysplasia in Barrett's esophagus tissues, and specific host factors such as race, gender, and environmental exposures (9, 10). Benign Barrett's metaplasia without dysplasia or with low-grade dysplasia exhibit a much lower malignant transformation rate compared with those with high-grade dysplasia (8, 11, 12). There has been considerable debate regarding the follow-up and radical treatment of Barrett's esophagus patients due to low conversion rate and low cost-effectiveness (13, 14). This highlights the importance of developing novel biomarkers that may further stratify Barrett's esophagus patients, which will allow for the selection of those with the highest risk for malignant progression to receive more aggressive therapies.

MicroRNAs (miRNA) are a group of small noncoding RNA molecules that are involved in a wide spectrum of basic cellular activities through their negative regulations of gene expression (1522). Moreover, miRNAs have also been extensively associated with the etiology and clinical outcome of many human cancers (2325). Previous genome-wide studies have reported that miRNA expression signatures could be used in cancer risk prediction, early diagnosis, histologic classification, and prognosis assessment (16, 2629). To date, however, there have been only a few studies on global miRNA expression profiling in esophageal cancers (30, 31). To our knowledge, this is the first study to examine miRNA expression profiles using paired tissues of various stages of Barrett's esophagus and esophageal adenocarcinoma to evaluate the role of miRNAs in the development and progression of esophageal adenocarcinoma in Caucasians.

Tissue samples

A total of 91 pairs of disease tissues and adjacent noncancerous normal tissues of the surrounding esophagus from 91 patients were included in this study, which consisted of 16 pairs for the initial miRNA microarray experiments and 75 pairs for the subsequent quantitative reverse transcription real-time PCR (qRT-PCR) validations. For the microarray experiments, the 16 patients consisted of 5 Barrett's esophagus patients with low-grade dysplasia, 5 Barrett's esophagus with high-grade dysplasia, and 6 esophageal adenocarcinoma patients. For qRT-PCR validation, the 75 patients included 26 Barrett's esophagus with low-grade dysplasia, 24 Barrett's esophagus with high-grade dysplasia, and 25 esophageal adenocarcinoma. All Barrett's esophagus samples were snap-frozen tissues obtained from the Division of Gastroenterology and Hepatology at the Mayo Clinic, Rochester, Minnesota (32). An objective standard was used to distinguish low-grade dysplasia and high-grade dysplasia (33). The collection of esophageal adenocarcinoma and adjacent surrounding normal tissues was as previously described (34, 35). Briefly, all esophageal adenocarcinoma tissues were snap-frozen tumor specimens collected at the time of diagnostic or therapeutic endoscopic biopsy procedures through an approved tissue collection protocol at The University of Texas M.D. Anderson Cancer Center. Corresponding normal esophageal squamous mucosa tissue was obtained for each esophageal adenocarcinoma tissue. All tumors were staged based on the system described in the 6th edition of the American Joint Commission on Cancer Atlas (36). All Barrett's esophagus and esophageal adenocarcinoma specimens were reviewed by at least one experienced gastrointestinal pathologist before total RNA extraction.

MiRNA microarray

Total RNAs including small RNAs were extracted from each tissue using the mirVana miRNA extraction kit (Ambion) according to the manufacturer's protocol. Total RNA concentrations were measured using a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies). The miRNA microarray experiments were done at the Cancer Genomics Core Laboratory at M.D. Anderson Cancer Center, using the Agilent Human miRNA Microarray Kit version 1.0, including 470 human miRNAs, 17 Kaposi Sarcoma-associated herpesvirus miRNAs, 14 human cytomegalovirus, and 32 EBV miRNAs (Agilent Technology; ref. 37). For each miRNA, multiple probes were spotted on the array and the average intensity of these probes was calculated to represent the expression value of the miRNA. In addition, multiple spots were included as negative controls. For each tissue sample, 100 ng total RNA were hybridized with the miRNA array and further processed in accordance with the manufacturer's instructions. The arrays were scanned using an Agilent Technology G2565BA scanner and the scanned images were processed using the Feature Extraction software package version 9.5 (Agilent Technology). Coefficient of variation (CV) within groups of replicate probes was used as a quality control measure to reflect the intra-array reproducibility. The range of CV for highest quality, acceptable quality, and failure was determined by Agilent as ≤8%, 8% to 15%, and >15%, respectively. In our study, the average CV ± SD (range) of all 32 arrays was 6.46% ± 1.17% (5.10-10.07%). Among the 32 arrays, 29 arrays had the highest quality with a CV < 8%. Three arrays had an acceptable quality with a CV between 8% and 15% (8.17%, 9.31%, and 10.07%).

qRT-PCR

To select the miRNAs for validation, we first compared our results with those of the study by Guo et al. that identified seven miRNAs through a stepwise selection procedure as the most relevant miRNAs differentiating tumor and normal esophageal tissues in esophageal squamous cell carcinoma (30). In that study, three (hsa-miR-25, hsa-miR-424, and hsa-miR-151) of the seven miRNAs exhibited an increased expression in tumor tissues whereas four (hsa-miR-100, hsa-miR-99a, hsa-miR-29c, and hsa-miR-140) showed a decrease. In our study, five of these seven miRNAs showed the same direction of expression change and had at least a borderline statistical significance (P values ranged from 0.0009 to 0.07, Supplementary Table S1). However, two miRNAs, hsa-miR-29c and hsa-miR-140, which showed a significantly reduced expression in the study of Guo et al., exhibited a significantly increased expression in our study (Supplementary Table S1). For the comparison and validation purpose, we conducted qRT-PCR to determine the expression level of 4 of the 7 miRNAs in 25 pairs of esophageal tumor and adjacent normal tissues, including 1 of the 3 consistently up-regulated miRNA (hsa-miR-25), 1 of 2 consistently down-regulated miRNA (hsa-miR-100), and the 2 inconsistent miRNAs (hsa-miR-29c and hsa-miR-140). We also validated hsa-miR-146a, which was up-regulated in both high-grade dysplasia and esophageal adenocarcinoma tissues in our initial microarray experiments. TaqMan MicroRNA Assay (Applied Biosystems) was used to quantify miRNA expression with a protocol slightly modified from the one provided by the manufacturer. In brief, 100 ng of DNase I-treated total RNA were reverse-transcribed in 1× RT buffer containing pooled stem-loop RT primers for all the miRNA genes and one endogenous control (5.6 nmol/L each), dNTPs (each at 250 μmol/L), 0.6 U of RNase inhibitor, and 50 U of MultiScribe Reverse Transcriptase (Applied Biosystems). Reverse transcription reactions were incubated at 16°C for 30 min, 42°C for 45 min, and 85°C for 5 min. Real-time PCR was done in duplicates in 10 μL volume containing 0.4 μL of microRNA assay mix and 5 μL of TaqMan 2× PCR Master Mix. PCR conditions were 50°C for 2 min, 95°C for 10 min, 40 cycles of 95°C for 15 s, and 60°C for 1 min. Expression data for miRNA were acquired and analyzed using ABI PRISM 7900HT Sequence Detection System and SDS 2.1 software (Applied Biosystems). Small RNA U48 was used as internal control for input normalization. The cycle number at which the real-time PCR reaction reached an arbitrarily determined threshold (CT) was recorded for both the miRNAs and U48, and the relative amount of miRNA to U48 was described as 2−ΔCT where ΔCT = (CTmiRNA − CTU48).

Statistical analyses

The microarray data were analyzed using the BRB-ArrayTools version 3.70, developed by Dr. Richard Simon and Amy Peng Lam (38). Data points flagged as absent or low-quality signals by the Feature Extraction software were removed from further analyses. Data points flagged as valid signals were log2-transformed after thresholded at 6.67, the average intensity of all negative control spots on the arrays. After log2 transformation, normalization was conducted using the median over entire array method and normalization to median array as reference. Gene filters were then applied to exclude an miRNA from all arrays if (a) <20% of the expression values of the miRNA had ≥1.5-fold change in either direction from the median expression value, or (b) the miRNA had >50% missing or filtered out data points. Unsupervised hierarchical clustering was carried out to generate the tree structures of both arrays and median-centered genes, using the Cluster 3.0 program that uses the average linkage clustering algorithm with centered correlation as similarity metric (39). The clustering results were visualized using the TreeView program (39). For each disease group (low-grade dysplasia, high-grade dysplasia, or esophageal adenocarcinoma, determined by pathologic review), differentially expressed miRNAs between disease and normal tissues were identified using the Significance Analysis of Microarrays (40) with the global false discovery rate controlled at 5%. The miRNAs showing significant differences in the ratio of disease/normal expression values between different progression stages were identified by unpaired Student's t test implemented in the class comparison module in BRB-ArrayTools. The disease/normal intensity ratios were calculated by subtracting the value of log2-transformed normal tissue intensity from the log2-transformed disease tissue intensity. The derived P values were adjusted by 10,000 univariate permutation tests. For each gene, the permutation P value was calculated as a proportion of permutations for which the P value of the univariate test was less than the P value of the Student's t test. The expression data for each miRNA of the qRT- PCR was calculated using paired Student's t test. All P values reported in this study were two-sided.

MiRNA signatures discriminated between disease tissues and paired normal tissues

To identify miRNA signatures that are associated with esophageal adenocarcinoma development, we did miRNA microarray analysis on the tissues of 16 Caucasian patients consisting of 5 Barrett's esophagus with low-grade dysplasia, 5 Barrett's esophagus with high-grade dysplasia, and 6 esophageal adenocarcinoma patients (Table 1). A total of 170 miRNAs had a signal intensity level above the threshold value in ≥75% of samples, which was comparable with the percentage of reliably detectable miRNAs reported in previous studies (27, 30). After applying gene filters to further screen out the miRNAs that were unlikely to be informative, we identified 111 miRNAs that remained in the final analyses. We used unsupervised hierarchical clustering to differentiate the disease tissues from their paired normal tissues (Fig. 1). We first analyzed all the tissues together and found that these tissues clustered into two major groups with 16 tissues in each group. Most normal tissues and disease tissues were separately grouped except for three normal tissues [low-grade dysplasia (LGD)-N1, LGD-N2, high-grade dysplasia (HGD-N1)] that were grouped with other diseases tissues, whereas three disease tissues (LGD-T1, LGD-T2, LGD-T3) were grouped with other normal tissues (Fig. 1A). It was noted that all the misclassified disease tissues were low-grade dysplasia tissues, indicating that there might not be significant differences in terms of miRNA expression patterns between normal tissues and Barrett's esophagus with low-grade dysplasia. To confirm these observations, we further carried out unsupervised hierarchical clustering within each specific disease type (Fig. 1B-D). We found that 4 of 10 tissues were misclassified in the group of low-grade dysplasia and corresponding normal tissues (Fig. 1B). In comparison, only one sample was misclassified in the high-grade dysplasia group (Fig. 1C). In the esophageal adenocarcinoma groups, the samples were classified with 100% accuracy between disease and normal tissues (Fig. 1D).

Table 1.

Host characteristics of the 16 study subjects in the MiRNA microarray experiments

Sample pairID of disease tissueID of normal tissueDisease typeGradeAge at diagnosis (y)GenderEthnicitySmoking status
LGD1 2312 (LGD-T1) 2311 (LGD-N1) Barrett's esophagus Low-grade dysplasia 61 Male Caucasian Current 
LGD2 1782 (LGD-T2) 1781 (LGD-N2) Barrett's esophagus Low-grade dysplasia 82 Male Caucasian Never 
LGD3 1503 (LGD-T3) 1502 (LGD-N3) Barrett's esophagus Low-grade dysplasia 75 Male Caucasian Former 
LGD4 3437 (LGD-T4) 3436 (LGD-N4) Barrett's esophagus Low-grade dysplasia 71 Male Caucasian Never 
LGD5 3660 (LGD-T5) 3659 (LGD-N5) Barrett's esophagus Low-grade dysplasia 58 Male Caucasian Former 
HGD1 2732 (HGD-T1) 2731 (HGD-N1) Barrett's esophagus High-grade dysplasia 62 Female Caucasian Former 
HGD2 3495 (HGD-T2) 3494 (HGD-N2) Barrett's esophagus High-grade dysplasia 79 Male Caucasian Former 
HGD3 5307 (HGD-T3) 5305 (HGD-N3) Barrett's esophagus High-grade dysplasia 80 Male Caucasian Former 
HGD4 6470 (HGD-T4) 6469 (HGD-N4) Barrett's esophagus High-grade dysplasia 74 Female Caucasian Never 
HGD5 8887 (HGD-T5) 8885 (HGD-N5) Barrett's esophagus High-grade dysplasia 50 Male Caucasian Former 
EAC1 34686-3 (EAC-T1) 34686-8 (EAC-N1) Esophageal adenocarcinoma 44 Male Caucasian Former 
EAC2 40771-3 (EAC-T2) 40771-13 (EAC-N2) Esophageal adenocarcinoma 72 Male Caucasian Former 
EAC3 42743-7 (EAC-T3) 42743-12 (EAC-N3) Esophageal adenocarcinoma 60 Male Caucasian Never 
EAC4 38521-7 (EAC-T4) 38521-13 (EAC-N4) Esophageal adenocarcinoma 44 Male Caucasian Current 
EAC5 36141-1 (EAC-T5) 36141-7 (EAC-N5) Esophageal adenocarcinoma 57 Male Caucasian Former 
EAC6 40839-4 (EAC-T6) 40839-13 (EAC-N6) Esophageal adenocarcinoma 78 Male Caucasian Never 
Sample pairID of disease tissueID of normal tissueDisease typeGradeAge at diagnosis (y)GenderEthnicitySmoking status
LGD1 2312 (LGD-T1) 2311 (LGD-N1) Barrett's esophagus Low-grade dysplasia 61 Male Caucasian Current 
LGD2 1782 (LGD-T2) 1781 (LGD-N2) Barrett's esophagus Low-grade dysplasia 82 Male Caucasian Never 
LGD3 1503 (LGD-T3) 1502 (LGD-N3) Barrett's esophagus Low-grade dysplasia 75 Male Caucasian Former 
LGD4 3437 (LGD-T4) 3436 (LGD-N4) Barrett's esophagus Low-grade dysplasia 71 Male Caucasian Never 
LGD5 3660 (LGD-T5) 3659 (LGD-N5) Barrett's esophagus Low-grade dysplasia 58 Male Caucasian Former 
HGD1 2732 (HGD-T1) 2731 (HGD-N1) Barrett's esophagus High-grade dysplasia 62 Female Caucasian Former 
HGD2 3495 (HGD-T2) 3494 (HGD-N2) Barrett's esophagus High-grade dysplasia 79 Male Caucasian Former 
HGD3 5307 (HGD-T3) 5305 (HGD-N3) Barrett's esophagus High-grade dysplasia 80 Male Caucasian Former 
HGD4 6470 (HGD-T4) 6469 (HGD-N4) Barrett's esophagus High-grade dysplasia 74 Female Caucasian Never 
HGD5 8887 (HGD-T5) 8885 (HGD-N5) Barrett's esophagus High-grade dysplasia 50 Male Caucasian Former 
EAC1 34686-3 (EAC-T1) 34686-8 (EAC-N1) Esophageal adenocarcinoma 44 Male Caucasian Former 
EAC2 40771-3 (EAC-T2) 40771-13 (EAC-N2) Esophageal adenocarcinoma 72 Male Caucasian Former 
EAC3 42743-7 (EAC-T3) 42743-12 (EAC-N3) Esophageal adenocarcinoma 60 Male Caucasian Never 
EAC4 38521-7 (EAC-T4) 38521-13 (EAC-N4) Esophageal adenocarcinoma 44 Male Caucasian Current 
EAC5 36141-1 (EAC-T5) 36141-7 (EAC-N5) Esophageal adenocarcinoma 57 Male Caucasian Former 
EAC6 40839-4 (EAC-T6) 40839-13 (EAC-N6) Esophageal adenocarcinoma 78 Male Caucasian Never 

Abbreviations: LGD-T, tissues of Barrett's esophagus with low-grade dysplasia; LGD-N, paired normal tissue of LGD-T; HGD-T, tissues of Barrett's esophagus with high-grade dysplasia; HGD-N, paired normal tissue of HGD-T; EAC-T, tissues of esophageal adenocarcinoma; EAC-N, paired normal tissue of EAC-T.

Fig. 1.

Unsupervised hierarchical clustering of different tissue groups including (A) all tissue samples, (B) Barrett's esophagus with low-grade dysplasia, (C) Barrett's esophagus with high-grade dysplasia, and (D) esophageal adenocarcinoma. LGD-T, Barrett's esophagus with low-grade dysplasia; LGD-N, normal paired tissue of LGD-T; HGD-T, Barrett's esophagus with high-grade dysplasia; HGD-N, normal paired tissue of HGD-T; EAC-T, esophageal adenocarcinoma; EAC-N, normal paired tissue of EAC-T. Blue, disease tissues clustered by unsupervised hierarchical clustering; yellow, normal tissues clustered by unsupervised hierarchical clustering; in red font, samples misclassified by unsupervised hierarchical clustering. Gray squares, missing or filtered out data points.

Fig. 1.

Unsupervised hierarchical clustering of different tissue groups including (A) all tissue samples, (B) Barrett's esophagus with low-grade dysplasia, (C) Barrett's esophagus with high-grade dysplasia, and (D) esophageal adenocarcinoma. LGD-T, Barrett's esophagus with low-grade dysplasia; LGD-N, normal paired tissue of LGD-T; HGD-T, Barrett's esophagus with high-grade dysplasia; HGD-N, normal paired tissue of HGD-T; EAC-T, esophageal adenocarcinoma; EAC-N, normal paired tissue of EAC-T. Blue, disease tissues clustered by unsupervised hierarchical clustering; yellow, normal tissues clustered by unsupervised hierarchical clustering; in red font, samples misclassified by unsupervised hierarchical clustering. Gray squares, missing or filtered out data points.

Close modal

Significantly differentially expressed miRNAs between disease tissues and paired normal tissues in different disease groups

The miRNAs that were differentially expressed between disease and paired normal tissues in each disease group, as well as their chromosome locations and host or overlapping genes, are summarized in Table 2. When the global false discovery rate was controlled at 5%, we could not identify any such miRNA in the low-grade dysplasia group (data not shown). In comparison, we found 32 and 39 miRNAs for the group of high-grade dysplasia and esophageal adenocarcinoma, respectively. Among these miRNAs, 24 miRNAs exhibited the same trend of expression change in both groups, including 14 miRNAs (hsa-miR-126, hsa-miR-143, hsa-miR-145, hsa-miR-146a, hsa-miR-181a, hsa-miR-181b, hsa-miR-195, hsa-miR-199a, hsa-miR-199a*, hsa-miR-199b, hsa-miR-28, hsa-miR-29c, hsa-miR-30a-5p, and hsa-miR-424) that were up-regulated in disease tissues and 10 miRNAs (hsa-miR-149, hsa-miR-203, hsa-miR-205, hsa-miR-210, hsa-miR-221, hsa-miR-27b, hsa-miR494, hsa-miR-513, hsa-miR-617, and hsa-miR-99a) that were down-regulated in disease tissues. For 9 of the 14 up-regulated miRNAs (hsa-miR-126, hsa-miR-143, hsa-miR-145, hsa-miR-181a, hsa-miR-181b, hsa-miR-199a, hsa-miR199a*, hsa-miR-28, and hsa-miR-30a-5p), the ratio of expression level in disease tissues to that of normal tissues was higher in esophageal adenocarcinoma than in high-grade dysplasia. For 7 of the 10 down-regulated miRNAs (hsa-miR-149, hsa-miR-203, hsa-miR-210, hsa-miR-27b, hsa-miR-513, hsa-miR-617, and hsa-miR-99a), the ratio of disease to normal expression level was lower in esophageal adenocarcinoma than in high-grade dysplasia.

Table 2.

Differentially expressed MiRNAs between disease and normal esophageal tissues by disease groups

graphic
 
graphic
 

Validation by qRT-PCR

Five miRNAs were validated using qRT-PCR in an independent set of 75 normal/disease paired tissues, including 26 Barrett's esophagus with low-grade dysplasia, 24 Barrett's esophagus with high-grade dysplasia, and 25 esophageal adenocarcinoma. The replication results were consistent with those of our initial microarray experiments, and all P values were more significant (P values ranged from 0.00006 to 0.25) than those in the microarray experiments except for hsa-miR-29c (P = 0.25; Table 3).

Table 3.

Validation of microarray experiments using quantitative reverse transcription real-time PCR

miRNATissue group*Microarray experimentsQuantitative reverse transcription real-time PCR
Pair number (n)Fold change (disease/normal)PChange in disease tissuePair number (n)Fold change (disease/normal)PChange in disease tissueConsistency
hsa-miR-25 EAC-T/EAC-N 1.56 0.07 Up 25 3.42 0.01 Up Yes 
hsa-miR-100 EAC-T/EAC-N 0.48 0.02 Down 25 0.60 0.005 Down Yes 
hsa-miR-29c EAC-T/EAC-N 2.61 0.02 Up 25 1.69 0.25 Up Yes 
hsa-miR-140 EAC-T/EAC-N 2.05 0.002 Up 25 2.55 0.00006 Up Yes 
hsa-miR-146a EAC-T/EAC-N 2.61 0.04 Up 25 2.98 0.003 Up Yes 
hsa-miR-146a HGD-T/HGD-N 4.54 0.02 Up 24 11.51 0.01 Up Yes 
miRNATissue group*Microarray experimentsQuantitative reverse transcription real-time PCR
Pair number (n)Fold change (disease/normal)PChange in disease tissuePair number (n)Fold change (disease/normal)PChange in disease tissueConsistency
hsa-miR-25 EAC-T/EAC-N 1.56 0.07 Up 25 3.42 0.01 Up Yes 
hsa-miR-100 EAC-T/EAC-N 0.48 0.02 Down 25 0.60 0.005 Down Yes 
hsa-miR-29c EAC-T/EAC-N 2.61 0.02 Up 25 1.69 0.25 Up Yes 
hsa-miR-140 EAC-T/EAC-N 2.05 0.002 Up 25 2.55 0.00006 Up Yes 
hsa-miR-146a EAC-T/EAC-N 2.61 0.04 Up 25 2.98 0.003 Up Yes 
hsa-miR-146a HGD-T/HGD-N 4.54 0.02 Up 24 11.51 0.01 Up Yes 

*We did not include the results of the Barrett's esophagus with low-grade dysplasia group because we could not identify any significantly differentially expressed miRNA in this group after multiple comparison adjustment.

P value of paired t test.

Consistency between the results of qRT-PCR and microarray experiments.

MiRNA signatures associated with the progression of esophageal adenocarcinoma

Using the class comparison module implemented in the BRB-ArrayTools, we identified 11 significant miRNAs that may be important in the progression from low-grade dysplasia to high-grade dysplasia, with 5 of them (hsa-miR-200a*, hsa-miR-513, hsa-miR-125b, hsa-miR-101, and hsa-miR-197) up-regulated and 6 of them (hsa-miR-23b, hsa-miR-20b, hsa-miR-181b, hsa-miR-203, hsa-miR-193b, and hsa-miR-636) down-regulated (Table 4). Seven miRNAs were potentially important in the progression from high-grade dysplasia to esophageal adenocarcinoma, and all of them were down-regulated in esophageal adenocarcinoma, including four members of the let-7 miRNA family (hsa-let-7b, hsa-let-7a, hsa-let-7c, hsa-let-7f), hsa-miR-345, hsa-miR-494, and hsa-miR-193a. All the P values remained significant after adjusted by 10,000 univariate permutation tests (Table 4).

Table 4.

MiRNAs with significantly different expression patterns between different progression stages from Barrett's esophagus to esophageal adenocarcinoma

miRNAP*Permutation PFold change
LGD-T to HGD-T 
    hsa-miR-23b 0.007 0.02 0.69 
    hsa-miR-20b 0.01 0.02 0.64 
    hsa-miR-200a* 0.02 0.02 13.47 
    hsa-miR-181b 0.03 0.008 0.45 
    hsa-miR-203 0.03 0.03 0.67 
    hsa-miR-513 0.03 0.03 1.58 
    hsa-miR-193b 0.04 0.008 0.37 
    hsa-miR-125b 0.04 0.05 9.20 
    hsa-miR-101 0.04 0.05 1.83 
    hsa-miR-197 0.04 0.05 1.61 
    hsa-miR-636 0.04 0.03 0.24 
HGD-T to EAC-T 
    hsa-let-7b 0.009 0.009 0.63 
    hsa-let-7a 0.01 0.004 0.57 
    hsa-miR-345 0.02 0.02 0.50 
    hsa-let-7c 0.03 0.03 0.59 
    hsa-let-7f 0.03 0.009 0.59 
    hsa-miR-494 0.03 0.03 0.58 
    hsa-miR-193a 0.05 0.05 0.44 
miRNAP*Permutation PFold change
LGD-T to HGD-T 
    hsa-miR-23b 0.007 0.02 0.69 
    hsa-miR-20b 0.01 0.02 0.64 
    hsa-miR-200a* 0.02 0.02 13.47 
    hsa-miR-181b 0.03 0.008 0.45 
    hsa-miR-203 0.03 0.03 0.67 
    hsa-miR-513 0.03 0.03 1.58 
    hsa-miR-193b 0.04 0.008 0.37 
    hsa-miR-125b 0.04 0.05 9.20 
    hsa-miR-101 0.04 0.05 1.83 
    hsa-miR-197 0.04 0.05 1.61 
    hsa-miR-636 0.04 0.03 0.24 
HGD-T to EAC-T 
    hsa-let-7b 0.009 0.009 0.63 
    hsa-let-7a 0.01 0.004 0.57 
    hsa-miR-345 0.02 0.02 0.50 
    hsa-let-7c 0.03 0.03 0.59 
    hsa-let-7f 0.03 0.009 0.59 
    hsa-miR-494 0.03 0.03 0.58 
    hsa-miR-193a 0.05 0.05 0.44 

*P value of t test in the class comparison module of BRB-ArrayTools to compare the disease/normal signal intensity ratios between different disease groups.

P value adjusted after 10,000 univariate permutations.

Value of HGD-T/LGD-T or EAC-T/HGD-T.

In this study, we assessed the miRNA expression patterns in patients with different stages of Barrett's esophagus and esophageal adenocarcinoma. To our knowledge, this is the first study using paired tissues of various stages of Barrett's esophagus and esophageal adenocarcinoma to evaluate the global miRNA expression patterns in the development and progression of esophageal adenocarcinoma. We found a large number of differentially expressed miRNAs between high-grade dysplasia and their paired normal tissues as well as between esophageal adenocarcinoma and their paired normal tissues. Unsupervised hierarchical clustering using the complete set of 111 miRNAs differentiated the esophageal adenocarcinoma tissues with 100% accuracy. In addition, we also observed that for a large majority of overlapped miRNAs in these two tissue groups, the disease/normal intensity fold-changes were more dramatic with more significant P values in the esophageal adenocarcinoma group (Table 2), suggesting that the oncogenic changes mediated by these miRNAs may be further intensified during the progression to esophageal adenocarcinoma. In contrast, we could not identify any significantly differentially expressed miRNAs between low-grade dysplasia and paired normal tissues. These observations were in line with literature showing the indistinguishable pathologic features between low-grade dysplasia and regenerative changes of normal esophageal cells, and the significantly lower malignant transformation rate in low-grade dysplasia compared with high-grade dysplasia (8, 1012, 41). Taken together, our data strongly indicate that miRNA may play a more prominent role in the progression from low-grade dysplasia to esophageal adenocarcinoma than in the initial transformation from normal esophagus cells to low-grade dysplasia.

Guo et al. have evaluated the miRNA expression differences between normal esophagus tissues and esophageal squamous cell carcinoma (30). Using a stepwise elimination procedure, they identified a signature with a minimum set of seven miRNAs to differentiate normal and tumor tissues with the highest accuracy. Five of the seven miRNAs also showed the same direction of expression change in our microarray analysis with statistical significance (Supplementary Table S1). For the two miRNAs that exhibited an opposite direction of expression change, we confirmed our results in an additional set of 25 pairs of esophageal adenocarcinoma samples. These findings were biologically plausible because some miRNAs may carry out similar functions in the development of both esophageal adenocarcinoma and esophageal squamous cell carcinoma, whereas other miRNAs may play distinctive roles in these two histologies. It is also possible that the differences are due to the racial disparity between Chinese and Caucasian populations. Further studies are warranted to compare the roles of miRNAs in these two subtypes of esophageal cancer.

In another recent report on miRNA expression in esophageal cancer, Feber et al. (31) used the mirVana miRNA Bioarrays (Ambion) to evaluate the expression of 287 human miRNAs in 31 esophageal specimens, including 10 esophageal adenocarcinoma, 10 esophageal squamous cell carcinoma, 5 Barrett's esophagus without dysplasia, 1 high-grade dysplasia, and 9 normal squamous epithelium tissues. Among the 13 human miRNAs that showed differential expression between esophageal adenocarcinoma and normal squamous epithelium tissues, 8 (hsa-miR-27b, hsa-miR-203, hsa-miR-205, hsa-let-7c, hsa-miR-342, hsa-miR-100, hsa-miR-21, and hsa-miR125b) passed the filtering criteria of our study and were included in the 111 miRNAs that remained within our final analysis. All of the eight miRNAs exhibited a consistent direction of expression change between the data of Feber's and ours (Supplementary Table S1; P values ranged from 0.00005 to 0.12), further supporting the robustness of our results.

We also identified miRNAs showing statistically significant differences between the different progression stages of esophageal adenocarcinoma. Among the 11 miRNAs implicated in the progression from low-grade dysplasia to high-grade dysplasia, hsa-miR-200a* and hsa-miR-125b showed a >13-fold and 9-fold increase in the disease/normal signal intensity ratio, respectively. Hsa-miR-200a* has not been associated with cancer, but its complementary miRNA, hsa-miR-200a, has been linked to the etiology and prognosis of many cancers including esophageal squamous cell carcinoma (30, 4244). Hsa-miR-125b was found to be up-regulated in colon cancers, but down-regulated in ovarian cancers, suggesting a cancer-specific functional pattern (43, 45). Two miRNAs (hsa-miR-181b and hsa-miR-193b) had the most significant permutation-adjusted P value and both were down-regulated in high-grade dysplasia. This was consistent with the potential role of hsa-miR-181b as a tumor suppressor in brain tumorigenesis, but opposite to the increased expression of this miRNA in pancreatic and colon cancers (4547), again suggesting a cancer-specific function. Another interesting finding was that among the seven miRNAs showing significantly different disease/normal intensity ratios between high-grade dysplasia and esophageal adenocarcinoma, four belong to the let-7 miRNA family, most members of which function as tumor suppressors through negative regulation of the RAS gene (48). Accordingly, RAS mutations and amplifications have been frequently identified in Barrett's esophagus and esophageal adenocarcinoma tissues (49, 50). Taken together, the miRNAs identified in our progression signatures are biologically plausible, and further functional dissections of these miRNAs and identification of their target genes are highly warranted.

A strength of our study was that all the patients were Caucasians and most (87.5%) were males, greatly reducing the confounding effects of race and gender. Although it should be cautioned that the limited sample size (16 pairs) in our initial microarray experiments could result in potential chance findings, we used strict statistical approaches to control for type I errors in each of the subsequent analyses. In addition, our microarray results are highly consistent with our real-time PCR validation, as well as with the results of two previous reports (30, 31), indicating that the chances are small for our data to be false positive. Nevertheless, our current results and additional significant miRNAs need to be further validated using larger-size prospectively collected patient samples before the clinical application of these biomarkers. Another limitation of our study was that we only compared miRNA expression patterns between tumor and normal tissues but did not relate these data to the clinical outcome of the esophageal adenocarcinoma patients. Additional collection of clinical follow-up data and analyses of the correlations between miRNA signatures and patient outcome would shed more light into the functions of miRNAs not only in cancer development and progression, but also in treatment response and prognosis. Moreover, it will provide additional clinical significance if we include Barrett's esophagus tissues without dysplasia or indefinite for dysplasia in future studies, because these are also major conditions frequently encountered in clinics.

In conclusion, we report for the first time that miRNAs may play important roles in the development and progression of esophageal adenocarcinoma. The list of promising miRNAs identified in our study could provide potential therapeutic targets for early detection, chemoprevention, and treatment of esophageal cancer.

No potential conflicts of interest were disclosed.

1
Parkin
DM
,
Bray
F
,
Ferlay
J
,
Pisani
P
. 
Global cancer statistics, 2002
.
CA Cancer J Clin
2005
;
55
:
74
108
.
2
Jemal
A
,
Siegel
R
,
Ward
E
, et al
. 
Cancer statistics, 2008
.
CA Cancer J Clin
2008
;
58
:
71
96
.
3
Brown
LM
,
Devesa
SS
,
Chow
WH
. 
Incidence of adenocarcinoma of the esophagus among white Americans by sex, stage, and age
.
J Natl Cancer Inst
2008
;
100
:
1184
7
.
4
Daly
JM
,
Fry
WA
,
Little
AG
, et al
. 
Esophageal cancer: results of an American College of Surgeons Patient Care Evaluation Study
.
J Am Coll Surg
2000
;
190
:
562
72
,
discussion 72–3
.
5
Blot
WJ
,
Devesa
SS
,
Kneller
RW
,
Fraumeni
JF
 Jr
. 
Rising incidence of adenocarcinoma of the esophagus and gastric cardia
.
JAMA
1991
;
265
:
1287
9
.
6
Devesa
SS
,
Blot
WJ
,
Fraumeni
JF
 Jr
. 
Changing patterns in the incidence of esophageal and gastric carcinoma in the United States
.
Cancer
1998
;
83
:
2049
53
.
7
Lagergren
J
,
Bergstrom
R
,
Lindgren
A
,
Nyren
O
. 
Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma
.
N Engl J Med
1999
;
340
:
825
31
.
8
McManus
DT
,
Olaru
A
,
Meltzer
SJ
. 
Biomarkers of esophageal adenocarcinoma and Barrett's esophagus
.
Cancer Res
2004
;
64
:
1561
9
.
9
Bani-Hani
K
,
Sue-Ling
H
,
Johnston
D
,
Axon
AT
,
Martin
IG
. 
Barrett's oesophagus: results from a 13-year surveillance programme
.
Eur J Gastroenterol Hepatol
2000
;
12
:
649
54
.
10
Avidan
B
,
Sonnenberg
A
,
Schnell
TG
,
Chejfec
G
,
Metz
A
,
Sontag
SJ
. 
Hiatal hernia size, Barrett's length, and severity of acid reflux are all risk factors for esophageal adenocarcinoma
.
Am J Gastroenterol
2002
;
97
:
1930
6
.
11
Weston
AP
,
Sharma
P
,
Topalovski
M
,
Richards
R
,
Cherian
R
,
Dixon
A
. 
Long-term follow-up of Barrett's high-grade dysplasia
.
Am J Gastroenterol
2000
;
95
:
1888
93
.
12
Schnell
TG
,
Sontag
SJ
,
Chejfec
G
, et al
. 
Long-term nonsurgical management of Barrett's esophagus with high-grade dysplasia
.
Gastroenterology
2001
;
120
:
1607
19
.
13
Streitz
JM
 Jr.
,
Ellis
FH
 Jr.
,
Tilden
RL
,
Erickson
RV
. 
Endoscopic surveillance of Barrett's esophagus: a cost-effectiveness comparison with mammographic surveillance for breast cancer
.
Am J Gastroenterol
1998
;
93
:
911
5
.
14
Montgomery
E
,
Goldblum
JR
,
Greenson
JK
, et al
. 
Dysplasia as a predictive marker for invasive carcinoma in Barrett esophagus: a follow-up study based on 138 cases from a diagnostic variability study
.
Hum Pathol
2001
;
32
:
379
88
.
15
Lee
RC
,
Feinbaum
RL
,
Ambros
V
. 
The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14
.
Cell
1993
;
75
:
843
54
.
16
Lu
J
,
Getz
G
,
Miska
EA
, et al
. 
MicroRNA expression profiles classify human cancers
.
Nature
2005
;
435
:
834
8
.
17
Lim
LP
,
Lau
NC
,
Garrett-Engele
P
, et al
. 
Microarray analysis shows that some microRNAs downregulate large numbers of target mRNAs
.
Nature
2005
;
433
:
769
73
.
18
Schratt
GM
,
Tuebing
F
,
Nigh
EA
, et al
. 
A brain-specific microRNA regulates dendritic spine development
.
Nature
2006
;
439
:
283
9
.
19
Poy
MN
,
Eliasson
L
,
Krutzfeldt
J
, et al
. 
A pancreatic islet-specific microRNA regulates insulin secretion
.
Nature
2004
;
432
:
226
30
.
20
Chen
JF
,
Mandel
EM
,
Thomson
JM
, et al
. 
The role of microRNA-1 and microRNA-133 in skeletal muscle proliferation and differentiation
.
Nat Genet
2006
;
38
:
228
33
.
21
Li
QJ
,
Chau
J
,
Ebert
PJ
, et al
. 
miR-181a is an intrinsic modulator of T cell sensitivity and selection
.
Cell
2007
;
129
:
147
61
.
22
Care
A
,
Catalucci
D
,
Felicetti
F
, et al
. 
MicroRNA-133 controls cardiac hypertrophy
.
Nat Med
2007
;
13
:
613
8
.
23
Esquela-Kerscher
A
,
Slack
FJ
. 
Oncomirs - microRNAs with a role in cancer
.
Nat Rev Cancer
2006
;
6
:
259
69
.
24
Tricoli
JV
,
Jacobson
JW
. 
MicroRNA: Potential for Cancer Detection, Diagnosis, and Prognosis
.
Cancer Res
2007
;
67
:
4553
5
.
25
Hammond
SM
. 
MicroRNAs as tumor suppressors
.
Nat Genet
2007
;
39
:
582
3
.
26
Meng
F
,
Henson
R
,
Lang
M
, et al
. 
Involvement of human micro-RNA in growth and response to chemotherapy in human cholangiocarcinoma cell lines
.
Gastroenterology
2006
;
130
:
2113
29
.
27
Yanaihara
N
,
Caplen
N
,
Bowman
E
, et al
. 
Unique microRNA molecular profiles in lung cancer diagnosis and prognosis
.
Cancer Cell
2006
;
9
:
189
98
.
28
Calin
GA
,
Liu
CG
,
Sevignani
C
, et al
. 
MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias
.
Proc Natl Acad Sci U S A
2004
;
101
:
11755
60
.
29
Chen
Y
,
Stallings
RL
. 
Differential patterns of microRNA expression in neuroblastoma are correlated with prognosis, differentiation, and apoptosis
.
Cancer Res
2007
;
67
:
976
83
.
30
Guo
Y
,
Chen
Z
,
Zhang
L
, et al
. 
Distinctive microRNA profiles relating to patient survival in esophageal squamous cell carcinoma
.
Cancer Res
2008
;
68
:
26
33
.
31
Feber
A
,
Xi
L
,
Luketich
JD
, et al
. 
MicroRNA expression profiles of esophageal cancer
.
J Thorac Cardiovasc Surg
2008
;
135
:
255
60
,
discussion 60
.
32
Sato
F
,
Jin
Z
,
Schulmann
K
, et al
. 
Three-tiered risk stratification model to predict progression in Barrett's esophagus using epigenetic and clinical features
.
PLoS ONE
2008
;
3
:
e1890
.
33
Montgomery
E
,
Bronner
MP
,
Goldblum
JR
, et al
. 
Reproducibility of the diagnosis of dysplasia in Barrett esophagus: a reaffirmation
.
Hum Pathol
2001
;
32
:
368
78
.
34
Luthra
R
,
Wu
TT
,
Luthra
MG
, et al
. 
Gene expression profiling of localized esophageal carcinomas: association with pathologic response to preoperative chemoradiation
.
J Clin Oncol
2006
;
24
:
259
67
.
35
Luthra
MG
,
Ajani
JA
,
Izzo
J
, et al
. 
Decreased expression of gene cluster at chromosome 1q21 defines molecular subgroups of chemoradiotherapy response in esophageal cancers
.
Clin Cancer Res
2007
;
13
:
912
9
.
36
Singletary
SE
,
Greene
FL
,
Sobin
LH
. 
Classification of isolated tumor cells: clarification of the 6th edition of the American Joint Committee on Cancer Staging Manual
.
Cancer
2003
;
98
:
2740
1
.
37
Wang
H
,
Ach
RA
,
Curry
B
. 
Direct and sensitive miRNA profiling from low-input total RNA
.
RNA
2007
;
13
:
151
9
.
38
Wright
GW
,
Simon
RM
. 
A random variance model for detection of differential gene expression in small microarray experiments
.
Bioinformatics
2003
;
19
:
2448
55
.
39
Eisen
MB
,
Spellman
PT
,
Brown
PO
,
Botstein
D
. 
Cluster analysis and display of genome-wide expression patterns
.
Proc Natl Acad Sci U S A
1998
;
95
:
14863
8
.
40
Tusher
VG
,
Tibshirani
R
,
Chu
G
. 
Significance analysis of microarrays applied to the ionizing radiation response
.
Proc Natl Acad Sci U S A
2001
;
98
:
5116
21
.
41
Alikhan
M
,
Rex
D
,
Khan
A
,
Rahmani
E
,
Cummings
O
,
Ulbright
TM
. 
Variable pathologic interpretation of columnar lined esophagus by general pathologists in community practice
.
Gastrointest Endosc
1999
;
50
:
23
6
.
42
Nam
EJ
,
Yoon
H
,
Kim
SW
, et al
. 
MicroRNA expression profiles in serous ovarian carcinoma
.
Clin Cancer Res
2008
;
14
:
2690
5
.
43
Iorio
MV
,
Visone
R
,
Di Leva
G
, et al
. 
MicroRNA signatures in human ovarian cancer
.
Cancer Res
2007
;
67
:
8699
707
.
44
Murakami
Y
,
Yasuda
T
,
Saigo
K
, et al
. 
Comprehensive analysis of microRNA expression patterns in hepatocellular carcinoma and non-tumorous tissues
.
Oncogene
2006
;
25
:
2537
45
.
45
Bloomston
M
,
Frankel
WL
,
Petrocca
F
, et al
. 
MicroRNA expression patterns to differentiate pancreatic adenocarcinoma from normal pancreas and chronic pancreatitis
.
JAMA
2007
;
297
:
1901
8
.
46
Shi
L
,
Cheng
Z
,
Zhang
J
, et al
. 
hsa-mir-181a and hsa-mir-181b function as tumor suppressors in human glioma cells
.
Brain Res
2008
.
47
Schetter
AJ
,
Leung
SY
,
Sohn
JJ
, et al
. 
MicroRNA expression profiles associated with prognosis and therapeutic outcome in colon adenocarcinoma
.
JAMA
2008
;
299
:
425
36
.
48
Jerome
T
,
Laurie
P
,
Louis
B
,
Pierre
C
. 
Enjoy the silence: The story of let-7 microRNA and cancer
.
Curr Genomics
2007
;
8
:
229
33
.
49
Lord
RV
,
O'Grady
R
,
Sheehan
C
,
Field
AF
,
Ward
RL
. 
K-ras codon 12 mutations in Barrett's oesophagus and adenocarcinomas of the oesophagus and oesophagogastric junction
.
J Gastroenterol Hepatol
2000
;
15
:
730
6
.
50
Galiana
C
,
Lozano
JC
,
Bancel
B
,
Nakazawa
H
,
Yamasaki
H
. 
High frequency of Ki-ras amplification and p53 gene mutations in adenocarcinomas of the human esophagus
.
Mol Carcinog
1995
;
14
:
286
93
.

Competing Interests

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.

Supplementary data