Barrett's esophagus is the precursor lesion of esophageal adenocarcinoma, whose progression follows sequential stages. However, the low progression rate and the inadequacy and subjective interpretation of histologic grading in predicting Barrett's esophagus progression call for more objective biomarkers that can improve risk prediction. We conducted a genome-wide profiling of 754 human microRNAs (miRNA) in 35 normal epithelium, 34 Barrett's esophagus, and 36 esophageal adenocarcinoma tissues using TaqMan real-time PCR-based profiling. Unsupervised hierarchical clustering using 294 modestly to highly expressed miRNAs showed clear clustering of two groups: normal epithelium versus Barrett's esophagus/esophageal adenocarcinoma tissues. Moreover, there was an excellent clustering of Barrett's metaplasia (without dysplasia) tissues from normal epithelium tissues. However, Barrett's esophagus tissues of different stages and esophageal adenocarcinoma tissues were interspersed. There were differentially expressed miRNAs at different stages. The majority of miRNA aberrations involved upregulation of expression in Barrett's esophagus and esophageal adenocarcinoma tissues, with the most dramatic alterations occurring at the Barrett's metaplasia stage. Known oncomiRs, such as miR-21, miR-25, and miR-223, and tumor suppressor miRNAs, including miR-205, miR-203, let-7c, and miR-133a, showed progressively altered expression from Barrett's esophagus to esophageal adenocarcinoma. We also identified a number of novel miRNAs that showed progressively altered expression, including miR-301b, miR-618, and miR-23b. The significant miRNA alterations that were exclusive to esophageal adenocarcinoma but not Barrett's esophagus included miR-375 downregulation and upregulation of five members of the miR-17-92 and its homologue clusters, which may become promising biomarkers for esophageal adenocarcinoma development. Cancer Prev Res; 6(3); 196–205. ©2013 AACR.
Esophageal cancer is the seventh most common cancer and sixth leading cause of cancer death in the world (1). In the United States, esophageal cancer is a rare cancer with an estimated 17,460 new cases and 15,070 deaths in 2012 (2). More than 90% of esophageal cancers are either esophageal adenocarcinoma or squamous cell carcinoma (ESCC). There are striking geographic differences in the incidences of these 2 histologic subtypes, with ESCCs dominating in Asian countries and esophageal adenocarcinoma prevailing in Western countries (3). Once a rare cancer representing only 5% of all esophageal cancers in the United States, esophageal adenocarcinoma is the cancer with the fastest increasing incidence—6-fold increase in the past 3 decades—and comprises more than 80% of all new cases currently in this country (4–7).
Esophageal adenocarcinoma is a deadly cancer with an overall 5-year survival rate of below 20% (8). The dismal prognosis is due to two thirds of esophageal adenocarcinoma cases presenting with advanced regional and local involvement or metastasis, when the current treatment is largely ineffective. To improve patient survival and reduce disease burden, the best hope in the near term is to detect esophageal adenocarcinoma at its early stage, or even better, to prevent the progression of esophageal adenocarcinoma from its premalignant lesions. Barrett's esophagus is an established precursor of esophageal adenocarcinoma in which the squamous epithelium of the esophagus is replaced by a metaplastic, columnar-lined epithelium extending from the gastroesophageal junction, and patients with Barrett's esophagus exhibit a 30- to 60-fold increased risk of developing esophageal adenocarcinoma (9–11). The incidence of Barrett's esophagus increases with age and may be present in 1% to 2% of general population after the age of 60 years (9–11). It is generally believed that the progression of Barrett's esophagus follows a series of histologic evolvement: non-dysplastic Barrett's metaplasia, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and ultimately, adenocarcinoma (12, 13). The malignant progression rate of Barrett's esophagus varies depending on the presence of dysplasia in Barrett's esophagus tissues, and the progression rate of non-dysplastic Barrett's metaplasia and LGD is much lower than that of HGD (12, 14, 15). The risk of developing esophageal adenocarcinoma in patients with HGD may be as high as 10% per patient year (11, 14). However, the grading of dysplasia is subjective resulting in substantial interobserver variations among pathologists on the degrees of dysplasia (16). Independent objective biomarkers to predict malignant progression may complement and augment pathologic grading and improve risk stratification among patients with Barrett's esophagus, allowing cost-effective surveillance, screening, and treatment of patients with Barrett's esophagus.
miRNAs are a class of small noncoding endogenous RNAs of 18 to 25 nucleotides capable of simultaneous regulation of hundreds of genes through binding to the 3′-untranslated region (3′-UTR) of target mRNAs resulting in either mRNA degradation or translational inhibition (17–19). Aberrant miRNA expressions have been observed in almost all solid tumors and hematologic malignancies studied (19, 20). miRNAs may function as oncogenes or tumor suppressors depending on their gene targets. miRNAs have been suggested as better biomarkers than mRNAs because of its small size and stability, capability of regulating hundreds of mRNAs, and small total number of human miRNAs compared with mRNAs (19–21).
There have been several published small scale studies comparing miRNA expression between normal epithelium, Barrett's esophagus, and esophageal adenocarcinoma tissues (22–28). These studies used different array platforms followed by validation of limited number of candidate miRNAs by TaqMan real-time PCR method. The results were heterogeneous, with a few miRNAs consistently shown to be involved in Barrett's esophagus and esophageal adenocarcinoma progression, for example, miR-21 and miR-192 were upregulated, and miR-203 and miR-205 were downregulated in both Barrett's esophagus and esophageal adenocarcinoma (29). To confirm previously reported miRNAs and identify additional novel miRNAs involved in Barrett's esophagus and esophageal adenocarcinoma progression, in this study, we used the real-time PCR TaqMan assay to conduct genome-wide miRNA profiling in a large series of Barrett's esophagus and esophageal adenocarcinoma tissues of different stages. To our knowledge, this is the first study to use TaqMan real-time PCR platform to profile miRNA expression and the largest profiling study of Barrett's esophagus and esophageal adenocarcinoma tissues to date.
Materials and Methods
A total of 105 tissues (35 normal, 34 Barrett's esophagus, and 36 esophageal adenocarcinoma) were included in this study. The collection of esophageal tissues has been described previously (25). Briefly, esophageal adenocarcinoma and adjacent normal tissues were obtained from patients diagnosed with esophageal adenocarcinoma with no prior treatment at the University of Texas MD Anderson Cancer Center (Houston, TX). All Barrett's esophagus tissues were from patients from the Mayo Clinic who did not have a diagnosis of cancer, among which were 11 Barrett's metaplasia without dysplasia, 13 LGD, and 10 HGD. There were no significant differences in ages between patients with Barrett's esophagus and esophageal adenocarcinoma: the average age of the patients with Barrett's esophagus and esophageal adenocarcinoma were 64.0 (SD, 12.8; range, 41–92 years) and 62.2 years (SD, 13.9; range, 35–85 years), respectively (P = 0.56). All tissues were snap-frozen at the time of diagnostic or therapeutic endoscopic biopsies, and tumor tissues were appropriately staged. All tumor specimens were reviewed for histologic reading by at least one experienced gastrointestinal pathologist before total RNA extraction.
RNA extraction and TaqMan miRNA profiling
Total RNAs including small RNAs were extracted from tissues using the mirVana miRNA Extraction Kit (Ambion). We used NanoDrop Spectrophotometer (Thermal Scientific) to measure the concentrations and purities of extracted RNAs. Only those RNAs with optical density (OD)260/OD280 > 1.8 and OD260/OD230 > 2.0 were used for miRNA profiling. We also visually examined the spectra of each RNA sample and ensured that all RNAs for miRNA profiling had normal spectra profile (i.e., normal shape of peaks and trough at different wavelengths). We used the Applied Biosystems' real-time PCR-based TaqMan Human MicroRNA Card Set v3.0 that enables accurate quantitation of 754 human miRNAs. The assay was conducted according to the manufacturer's protocol. Small nuclear RNAs, U6, U44, and U48, were used as internal control for input normalization. The cycle number at which the real-time PCR reaction reached an arbitrarily determined threshold (Ct) is recorded for each miRNA and internal controls. miRNAs that were detectable in less than 20% of the samples were excluded. All the excluded miRNAs had very low expressions across the board and were not limited to any single histologic stage. There were no significant differences in expression for any of the excluded miRNAs between normal epithelium, Barrett's esophagus, and esophageal adenocarcinoma tissues. The relative amount of miRNA to internal control was described as where .
Hierarchical clustering analysis was conducted using the GenePattern version 3.2.3. Spearman rank correlation coefficient was used to compare the miRNA expression levels among normal, Barrett's esophagus, and esophageal adenocarcinoma tissues and identify differentially expressed miRNAs with increasing histologic stages. The Student t test was used to identify differentially expressed miRNAs between Barrett's esophagus and normal, between esophageal adenocarcinoma and normal, and between esophageal adenocarcinoma and Barrett's esophagus tissues. Spearman rank test was used to examine the correlation between individual miRNA with histologic grades. We used a false discovery rate (FDR)-based Q test to control for multiple testing. All tests were 2-sided, and P < 0.05 was considered statistically significant.
Tissue classification with unsupervised hierarchical clustering
To identify miRNA expression signatures that may predict Barrett's esophagus progression, we conducted a genome-wide miRNA profiling of 35 normal epithelium, 34 Barrett's esophagus (including 11 Barrett's metaplasia, 13 LGD, and 10 HGD), and 36 esophageal adenocarcinoma tissues. Among 754 human miRNAs on the array, 408 were detectable in more than 80% of samples. We then used the 294 miRNAs that showed modest to high expressions (mean amplification threshold cycle Ct < 30) to conduct an unsupervised hierarchical clustering of all 105 tissues (Fig. 1). There were 2 major clusters, normal squamous tissues (Fig. 1, left green bar) versus columnar (Barrett's esophagus and esophageal adenocarcinoma) tissues (Fig. 1, right, light blue, and purple bar), with only 3 normal epithelium tissues clustered with columnar tissues and a few columnar tissues clustered with normal epithelium. It should be pointed out that the normal epithelium tissues were adjacent normal squamous tissues obtained from patients with esophageal adenocarcinoma, and it is likely that the grossly normal tissues may already have molecular abnormalities. Notably, the Barrett's esophagus and esophageal adenocarcinoma tissues were interspersed without a clear separation, suggesting that miRNA alterations occur early. To confirm this observation, we further conducted unsupervised hierarchical clustering between specific histologic stages. There was clear clustering of normal epithelium and Barrett's metaplasia (no dysplasia) tissues (Fig. 2) with all Barrett's metaplasia tissues clustered together. We conducted similar unsupervised clustering of normal versus LGD, normal versus HGD, and normal versus esophageal adenocarcinoma, all of which showed clear clustering of normal squamous tissues versus different stages of Barrett's esophagus and esophageal adenocarcinoma tissues; however, the clustering of paired different stages of Barrett's esophagus (Barrett's metaplasia, LGD, and HGD) and esophageal adenocarcinoma tissues was interspersed (data not shown).
Differentially expressed miRNAs between normal epithelium, Barrett's esophagus, and esophageal adenocarcinoma tissues
There were a large number of differentially expressed miRNAs between normal epithelium and Barrett's esophagus/esophageal adenocarcinoma tissues. Pairwise comparison of Barrett's esophagus versus normal epithelium and esophageal adenocarcinoma versus normal epithelium showed that there were much more upregulated miRNAs than downregulated miRNAs in Barrett's esophagus and esophageal adenocarcinoma tissues than in normal tissues. In Barrett's esophagus versus normal epithelium, a total of 148 miRNAs were upregulated and 16 downregulated; similarly in esophageal adenocarcinoma versus normal epithelium, 122 were upregulated and 16 downregulated (Supplementary Table S1). The vast majority of differentially expressed miRNAs were significant in both Barrett's esophagus and esophageal adenocarcinoma tissues. There were only a few miRNAs that were significantly different in esophageal adenocarcinoma but not in Barrett's esophagus tissues compared with normal epithelium, including miR-375 (downregulated in esophageal adenocarcinoma only, not in Barrett's esophagus) and miR-106-3b, miR-18, miR-18-3p, miR-20b, and miR-92a-1-3p (upregulated in esophageal adenocarcinoma only, not in Barrett's esophagus; Supplementary Table S1). The differentially expressed miRNAs between esophageal adenocarcinoma and Barrett's esophagus, particularly upregulated, were much fewer (Supplementary Table S1). Interestingly, among 12 upregulated (>2-fold) miRNAs in esophageal adenocarcinoma versus normal epithelium, 5 were in human miR-17-92 cluster (miR-17-3p, miR-18a, miR-18a-3p, miR-92, miR-92a-1-3p) and 2 in its homologue miR-106b-25 (miR-106b-3p and miR-25) cluster (30). A substantial number of upregulated miRNAs in Barrett's esophagus had lower expression in esophageal adenocarcinoma than in Barrett's esophagus tissues, although still higher than normal epithelium tissues (Table 1, Supplementary Table S1).
The most biologically relevant miRNAs specifically involved in Barrett's esophagus progression are likely those with progressively altered expression along the different stages. Table 2 shows the top miRNAs with progressively altered expression in Barrett's esophagus and esophageal adenocarcinoma, which are likely oncomiRs or tumor suppressor miRNAs involved in esophageal adenocarcinoma carcinogenesis. Among these miRNAs were well-established oncomiRs such as miR-21 and miR-25 and tumor suppressor miRNAs such as miR-205 and miR-203. Notably, the miRNA cluster miR-99a/let-7c/miR-125b-2 was downregulated in Barrett's esophagus and esophageal adenocarcinoma.
We also conducted paired comparisons between normal epithelium and varying stages of Barrett's esophagus (Barrett's metaplasia, LGD, and HGD) as well as among varying stages of Barrett's esophagus/esophageal adenocarcinoma tissues (data not shown). There were significantly more differentially expressed miRNAs from normal epithelium to Barrett's metaplasia stage than the later progressive stages (i.e., Barrett's metaplasia to LGD, LGD to HGD, and HGD to esophageal adenocarcinoma), and more than 90% of the differentially expressed miRNAs from normal epithelium to Barrett's metaplasia were upregulated. All of the upregulated miRNAs in the overall analysis of Barrett's esophagus tissues exhibited marked increase of expression during Barrett's metaplasia stage and maintained high expression throughout later stages (Fig. 3). In contrast, the downregulated miRNAs started deregulation at different stages, for example, miR-205 showed dramatically reduced expression at Barrett's metaplasia stage, let-7c showed most significant reduction at LGD, whereas miR-375 started downregulation at HGD with the most dramatic reduction at esophageal adenocarcinoma stage (Fig. 3).
Potential mRNA targets of the differentially expressed miRNA
To identify potential genes whose mRNAs might be targeted by the differentially expressed miRNAs in Barrett's esophagus and esophageal adenocarcinoma, we used a published study of Nancarrow and colleagues (31), in which a list of 54 differentially expressed mRNAs (reported in at least 3 studies) were compiled from 16 esophageal expression profiling studies that compared mRNA expression in normal epithelium, Barrett's esophagus, and esophageal adenocarcinoma tissue groups. For each of the top 20 progressively differentially expressed miRNAs identified from our study (Table 2), putative mRNA targets were predicted by 5 different algorithms, TargetScan, PicTar, microrna.org, miRanda, and miRGen, and compared with the 54 mRNAs. As shown in Table 3, 19 of the top 20 differentially expressed miRNAs target one or more of these most frequently altered mRNAs, and a total of 77.8% (42 of 54) of mRNAs differentially expressed between these tissue types were potential targets for the top 20 differential miRNAs. miR-203 had 12 potential target mRNAs that are differentially expressed and let-7c had 11.
Barrett's esophagus is a metaplastic premalignant lesion with high likelihood for progression to esophageal cancer. Although the molecular basis of Barrett's esophagus progression to esophageal adenocarcinoma has been widely studied, the precise mechanisms for carcinogenesis remains not completely understood. miRNAs have been shown to have powerful potential as diagnostic, prognostic, and therapeutic tools in a variety of cancer indications. We hypothesize that miRNA expression signatures may help to define malignant progression in Barrett's esophagus and esophageal cancer.
This study provides the most comprehensive catalogue of miRNA expressions in Barrett's esophagus and esophageal adenocarcinoma tissues to date. A total of 754 miRNAs have been profiled, of which 408 have detectable expression in normal epithelium, Barrett's esophagus, or esophageal adenocarcinoma tissues. Previous studies of miRNA profiling in Barrett's esophagus and esophageal adenocarcinoma tissues mostly used microarrays with variable results. We used a TaqMan real-time PCR method for profiling, which gives more specific and accurate quantification of miRNA expression than array-based profiling technology. Importantly, many of our identified candidates have been previously reported to have altered expression in Barrett's esophagus and esophageal adenocarcinoma development, such as known oncomiRs, miR-21 and miR-25, and tumor suppressor miRNAs, let-7c, miR-205, and miR-203 (Table 2), showing the reliability of this technology. Several of the significant miRNAs are known to target genes that have been implicated in cancer-related activities (Supplementary Table S2). In studies that reported the significant markers, the expression of miR-21 and miR-25 uniformly showed progressively increased expression from normal epithelium to Barrett's esophagus and from Barrett's esophagus to esophageal adenocarcinoma. Similarly, the expression of let-7c, mir-205, and miR-203 displayed stepwise decrease in expression during the same stages of disease progression. Consistent with prior study by Wijnhoven and colleagues (26), miR-215, miR-143, and miR-145 have been shown to have large increases in miRNA expression (≥10-fold) along with miR-133a and miR-1 (which showed the largest gain) in transformed tissues. Interestingly, for most of the upregulated miRNAs, the greatest increase in expression was found comparing Barrett's esophagus with normal epithelium tissues, whereas the expression of these genes decreased in esophageal adenocarcinoma, although still higher than their counterparts in esophageal squamous tissues. Leidner and colleagues (32) applied next-generation sequencing followed by real-time PCR to identify miRNAs potentially associated with progression of Barrett's esophagus. They identified 26 miRNAs that were frequently deregulated in esophageal adenocarcinoma, among which 23 were deregulated at the earliest stage (Barrett's metaplasia), consistent with observations from our study. These observations might indicate the larger alteration of miRNA changes during the initial transformation toward premalignant lesion. These observations also suggest that miRNA alterations may precede genetic aberrations and mRNA changes, as the latter 2 events are not as dramatic as miRNA changes at Barrett's metaplasia stage and there are substantial differences of both DNA copy number and mRNA expression between later stages of Barrett's esophagus and esophageal adenocarcinoma tissues (13, 31). However, as Barrett's esophagus and esophageal adenocarcinoma are columnar tissues, compared with the squamous nature of normal tissues, it is possible that the difference in tissue origin may contribute partially to the greater similarity of miRNA expressions between Barrett's esophagus and esophageal adenocarcinoma tissues than between normal epithelium and columnar (Barrett's esophagus/esophageal adenocarcinoma) tissues. The large number of upregulated miRNAs compared with downregulated species suggests that the alterations tend to favor miRNAs that target genes which suppress growth-promoting activities. We cannot rule out that lesional content of the tissue may partially contribute to the larger number of miRNAs being upregulated than downregulated because it is more difficult to identify downregulated transcripts in a given tissue with a higher heterogeneous cellular content.
miR-223 and its alternative form, miR-223-5p, showed the most significant trend of progressively increased expression. A previous study has shown that miR-223 was overexpressed in esophageal adenocarcinoma tissues (33). This current study is the first one to show its upregulation in Barrett esophagus tissues. miR-223 is a myeloid-specific miRNA with important regulatory roles in hematopoiesis and oncogenesis and was shown to be overexpressed in colorectal adenocarcinoma patients with hereditary nonpolyposis colorectal cancer syndrome (34, 35). miR-223 displayed the largest increase in tissue expression in recurrent ovarian cancer compared with primary tumors consistent with its involvement in tumor progression (36). Predicted gene targets for miR-223 include several growth factor signaling pathways (ACVR2A, FGFR2, IGF1R, EGF, and TGFB2), oncogene KRAS, transcription factor E2F1, and cytokine IFNB1, suggesting that increased miR-223 expression might alter global balance of proliferative versus differentiation signaling to favor carcinogenesis.
Of the miRNAs that displayed progressively downregulation from normal epithelium to Barrett's esophagus and to esophageal adenocarcinoma, miR-99a-3p and miR-23b were the top 2 miRNAs. miR-99a is a putative tumor suppressor whose known gene targets include mTOR, FGFR2, and IGF-1R (Supplementary Table S2). Its expression was downregulated in prostate cancer and correlated with the prognosis of patients with hepatocellular carcinoma (HCC; refs. 37, 38). In in vitro experiments, expression of miR-99a in cancer cells suppressed proliferative activity leading to growth arrest (37, 38). miR-23b is a direct target of c-MYC–mediated transcriptional regression (39). miR-23b is downregulated in a few other cancer types, including prostate, endometrial, and colon cancers (40–42). miRNA-23b represses proto-oncogene Src kinase in prostate cancer cells and inhibits cell proliferation (43). A recent genome-wide functional screening found miR-23b as a pleiotropic modulator directly regulating a cohort of prometastatic genes or oncogenes, including FZD7, MAP3K1, PAK2, TGFβR2, RRAS2, or uPA (42). Loss of miR-23b may therefore confer proliferative advantage and promote esophageal carcinogenesis.
We found several hitherto unreported miRNAs related to Barrett's esophagus and esophageal adenocarcinoma carcinogenesis that showed progressive up- or downregulation. Many of the novel miRNAs in Table 2 have been anecdotally reported in other cancers. For example, the expression of miR-301b and miR-424 was reported upregulated and miR-378 downregulated in colorectal cancer compared with para-cancerous control tissue (44). For miR-618, a recent study showed that the level of this miRNA was elevated in urine samples from patients with HCC who were hepatitis C (HCV) carriers compared with normal individuals and patients with HCV-negative HCC (45). Further validation of these miRNAs in independent sample sets and then characterization of their biologic effects are warranted.
Paradoxically, we found miR-133a and miR-1, 2 putative tumor-suppressing miRNAs, having the largest fold increase in expression in patients with Barrett's esophagus. In addition, the expressions of these miRNAs were also elevated in esophageal adenocarcinoma compared with normal tissue. This increase of expression apparently is not due to genetic amplification. Fist, miR-1-1/miR-133a-2 and miR-1-2/miR-133a-1 are clustered on different chromosomal regions in the human genome, 20q13.33 and 18q11.2, respectively. Neither of these 2 regions are highly amplified in Barrett's esophagus or esophageal adenocarcinoma tissues (13). Second, increased expression of miR-1 and miR-133a occurs as early as at Barrett's metaplasia stage, when genetic changes are extremely rare and there is not a single amplification event at Barrett's metaplasia stage as we previously reported in a high-density array analysis (13). miR-133a and miR-1 have been shown to suppress tumorigenic property, mediate apoptosis and cell-cycle arrest, and inhibit proliferation and invasion in bladder, esophageal, lung, and head and neck cancer cells (46–50). It is possible that the large increase in miRNA expression in our study might be due to genotoxic stress generated by environmental factors unique to Barrett's esophagus. Alternatively, the 2 miRNAs may have variant gene functions that are tissue-specific in esophageal epithelial cells. Further validation and functional characterizations in vitro and in vivo are necessary to identify the potential mechanisms and putative targets for miR-133 and miR-1 during Barrett's esophagus progression.
One of the main objectives of this study is to identify biomarkers that may eventually improve risk stratification in patients with Barrett's esophagus. Given that the majority of patients with Barrett's esophagus do not progress to esophageal adenocarcinoma, any miRNAs showing consistent alteration in the majority of Barrett's esophagus, LGD, HGD, and esophageal adenocarcinoma stages may be indicative of tissue metaplasia rather than being associated with progression and therefore may not be useful for predicting esophageal adenocarcinoma progression. The most promising biomarkers would be those miRNAs that showed great deregulation at esophageal adenocarcinoma stage, but not in Barrett's esophagus. Among the 16 downregulated miRNAs in esophageal adenocarcinoma versus Barrett's esophagus (Supplementary Table S1), only miR-375 was unchanged in Barrett's esophagus, which was consistent with the report of Leidner and colleagues (32), although there was a slight downregulation of miR-375 at HGD in our study but not in Leidner and colleagues (32). Among 122 upregulated miRNAs (>2-fold) in esophageal adenocarcinoma versus normal epithelium tissues, only 5 miRNAs were exclusive to esophageal adenocarcinoma but was unchanged in Barrett's esophagus which were miR-106b-3p, miR-18, miR-18-3p, miR-20b, and miR-92a-1-3p (Supplementary Table S1). Interestingly, these 5 miRNAs all belong to the human mir-17-92 and its homologue (mir-106a-363 and mir-106b-25) clusters (30). The miRNAs encoded by these 3 clusters can be categorized into 4 separate miRNA families according to their seed sequences, including the miR-17 family (miR-17, miR-20a, miR-20b, miR-106a, miR-106b, and miR-93), the miR-18 family (miR-18a and miR-18b), the miR-19 family (miR-19a, miR-19b-1, and miR-19b-2), and the miR-92 family (miR-92a-1, miR-92a-2, miR-383, and miR-25; ref. 30). These data strongly suggest that the upregulation of mir-17-92 and its homologues play a critical role in driving esophageal adenocarcinoma progression and these miRNAs could be promising biomarkers for esophageal adenocarcinoma progression.
There are some limitations in our study. First, this is a cross-sectional study in which all the Barrett's esophagus and esophageal adenocarcinoma samples were from separate individuals. To prove that these biomarkers predict esophageal adenocarcinoma progression in patients with Barrett's esophagus, prospective follow-up study is the golden standard, although the low progression rate and long latency period make prospective study very difficult. Second, our sample size is sufficient for cross-comparison of different histologic tissues; however, the sample size is not sufficient for us to conduct stratified analyses and identify miRNAs associated with smoking, survival, and other clinical variables. In addition, our Barrett's esophagus tissue samples were taken from patients recruited at a different site (Mayo Clinic), whereas the paired normal and esophageal adenocarcinoma tissues were obtained from patient visits at the MD Anderson Cancer Center. However, the 2 sites used similar study protocols with uniformly defined clinical criteria and well-characterized hospital-based population that any confounding effect from site selection should be minimal. As we did not have paired normal tissues from the patients with Barrett's esophagus available, the relative changes in the global miRNA profile between the different stages of malignant transformation should be interpreted with caution. Further validation of the newly identified markers in larger independent populations is warranted. Finally, although this study has profiled more miRNAs than previous studies, it is still only investigating around half those currently known miRNAs. Next-generation sequencing would provide a complete picture of miRNAs in Barrett's esophagus and esophageal adenocarcinoma development.
In summary, we have presented the most comprehensive screening of global miRNA profile of malignant progression in Barrett's esophagus and esophageal adenocarcinoma using TaqMan real-time PCR technique. We have confirmed many previously reported miRNA markers and identified additional new miRNAs that were differentially expressed at various stages of Barrett's esophagus and esophageal adenocarcinoma. The results of this investigation may open avenues for more in-depth understanding of the biology of tumor transformation in esophageal cancer. Differentially expressed miRNAs may be candidate biomarkers for early diagnosis and progression as well as potential gene targets for future treatment and prevention.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: X. Wu, J.A. Ajani, J. Gu
Development of methodology: X. Wu, J. Gu
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): X. Wu, K.K. Wang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): X. Wu, D.W. Chang, M.A.T. Hildebrandt, M. Huang, K.K. Wang
Writing, review, and/or revision of the manuscript: X. Wu, J.A. Ajani, J. Gu, D.W. Chang, M.A.T. Hildebrandt, K.K. Wang, E. Hawk
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Wu, J.A. Ajani, W. Tan, E. Hawk
Study supervision: X. Wu, J. Gu, E. Hawk
This study was funded by National Cancer Institute grants CA111922 and CA138671 and support from the Premalignant Genome Atlas program of the Dunkan Family Institute for Cancer Prevention and Risk Assessment at The University of Texas MD Anderson Cancer Center.
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.
Note: Supplementary data for this article are available at Cancer Prevention Research Online (http://cancerprevres.aacrjournals.org/).
- Received June 25, 2012.
- Revision received December 13, 2012.
- Accepted January 7, 2013.
- ©2013 American Association for Cancer Research.