Esophageal adenocarcinoma (EAC) has increased dramatically in the past 3 decades, making its precursor lesion Barrett’s esophagus (BE) an important clinical problem. Effective interventions are available, but overall outcomes remain unchanged. Most of the BE population remains undiagnosed; most EACs are diagnosed late, and most BE patients will never progress to cancer. These epidemiologic factors make upper endoscopy an inefficient and ineffective strategy for BE diagnosis and risk stratification. In the current review, biomarkers for diagnosis, risk stratification, and predictors of response to therapy in BE are discussed.
Key points
- •
Molecular diagnosis of Barrett’s esophagus (BE) can now be performed on nonendoscopic cytology specimens. Trefoil factor 3 is promising. Other markers to test further are microRNA and methylated genes.
- •
BE can be risk-stratified by measuring global (eg, aneuploidy, multiple gains) or specific (eg, p53 expression) markers. A biomarker panel is likely to be needed.
- •
Limited data are available on markers that can predict response to therapy. Chromosome and gene loss/gain by fluorescent in situ hybridization may be of value.
- •
P53 mutational analysis on nonendoscopic cytology specimens can detect prevalent high-grade dysplasia. Thus, the same sample can be used to screen for and risk-stratify BE.
Introduction
Extensive biomarker research has been conducted to improve the diagnosis and management of Barrett’s esophagus (BE). Investigators have worked for many years to discover biomarkers for BE diagnosis, risk stratification, and prediction of response to therapy. The level of evidence required to change clinical practice has generally not been achieved, with the exception of p53; however, more investment and collaborative studies are starting to pay dividends. Discovery of clinically applicable biomarkers for BE management has never been more important. The risk of progression of BE to cancer is lower than previously thought. Therefore, the resources need to be focused on those BE patients who are most likely to benefit. An important hurdle to clinical application of biomarkers is that specimen acquisition to study biomarkers requires an endoscopy; this has changed recently with the availability of a well-tolerated, easy-to-swallow capsule-based cytology sponge that can collect esophageal cells in an office-based setting. Also, emerging data suggest that blood samples can be used to study the disease status. Last, genome-wide next-generation sequencing techniques have significantly added to the understanding of somatic DNA aberrations in BE pathogenesis. All of these factors have created a viable environment for molecular biomarkers of BE and associated risk of cancer to progress to the clinic. In the following sections, important biomarkers to diagnose BE, risk-stratify the patients with BE, and identify those BE patients who are likely to be less responsive to endoscopic therapies are discussed. Where possible, studies published in the last 5 years were the focus.
Introduction
Extensive biomarker research has been conducted to improve the diagnosis and management of Barrett’s esophagus (BE). Investigators have worked for many years to discover biomarkers for BE diagnosis, risk stratification, and prediction of response to therapy. The level of evidence required to change clinical practice has generally not been achieved, with the exception of p53; however, more investment and collaborative studies are starting to pay dividends. Discovery of clinically applicable biomarkers for BE management has never been more important. The risk of progression of BE to cancer is lower than previously thought. Therefore, the resources need to be focused on those BE patients who are most likely to benefit. An important hurdle to clinical application of biomarkers is that specimen acquisition to study biomarkers requires an endoscopy; this has changed recently with the availability of a well-tolerated, easy-to-swallow capsule-based cytology sponge that can collect esophageal cells in an office-based setting. Also, emerging data suggest that blood samples can be used to study the disease status. Last, genome-wide next-generation sequencing techniques have significantly added to the understanding of somatic DNA aberrations in BE pathogenesis. All of these factors have created a viable environment for molecular biomarkers of BE and associated risk of cancer to progress to the clinic. In the following sections, important biomarkers to diagnose BE, risk-stratify the patients with BE, and identify those BE patients who are likely to be less responsive to endoscopic therapies are discussed. Where possible, studies published in the last 5 years were the focus.
Biomarkers for Barrett’s esophagus diagnosis
Most biomarker research has focused on risk stratification in BE discussed elsewhere in this review. Lately, there has been renewed interest in molecular testing for BE diagnosis. Recent technological developments in esophageal sampling in combination with specific markers have made office-based diagnosis of BE possible. Therefore, widespread application of these tests for BE diagnosis in persons at clinically significant risk for BE may become a viable option. These markers are described in later discussion.
Trefoil Factor 3
Introduction
Trefoil factor 3 (TFF3) is a secretory protein expressed in the goblet cells of the intestinal mucosa that has shown significant promise for molecular BE diagnosis.
Studies
To discover BE-specific markers, Lao-Sirieix and colleagues analyzed publicly available microarray datasets that compared normal squamous, BE, and gastric mucosa. Validation by 2 techniques, polymerase chain reaction (PCR) and histochemistry, suggested that TFF3 may be a specific marker for BE-type epithelium. TFF3 as a biomarker for BE diagnosis was tested on specimens acquired via a novel proprietary nonendoscopic cytology sponge within a capsule. After an initial study showed feasibility, the same group of investigators conducted a study in the primary care setting and found the capsule-based cytology sponge to be well tolerated with successful ingestion in 99% of 504 subjects. The sensitivity and specificity of TFF3 expression on cytology samples for diagnosis of circumferential BE 1 cm or longer were 73.3% (95% confidence interval [CI] 44.9%–92.2%) and 93.8% (91.3%–95.8%), respectively. These numbers improved to 90.0% sensitivity (95% CI 55.5%–99.7%) and 93.5% specificity (95% CI 90.9%–95.5%) when BE segments 2 cm or longer were included in the analysis. This approach was recently examined in a larger trial ( Box 1 ).
In a recent multicenter case-control study of more than 1000 patients, the sensitivity of TFF3 testing was around 80% (95% CI 76.4%–83.0%) even for short segments (1 cm circumferential) increasing to 90% (95% CI 83.0%–90.6%) in longer segments or when swallowed for a second time with a specificity of 92.4% (95% CI 89.5%–94.7%).
Summary
These results make a strong case for molecular testing on a nonendoscopic cytologic specimen as a practical tool for BE diagnosis in the symptomatic population, most of whom are not investigated. Further studies are ongoing to evaluate the applicability of this strategy to other populations.
MicroRNA
Introduction
MicroRNAs are novel, noncoding, small RNA molecules shown to be highly tissue-specific in a landmark article that evaluated multiple human cancers ; hence, they may have utility as biomarkers.
Studies
Fassan and colleagues used a customized miRNA microarray comprising 326 human miRNAs to compare squamous versus Barrett’s epithelium and found 23 miRNAs to be differentially expressed at a P value of less than 0.01. Validation found several of these miRNAs to be significantly different with fold changes of 6 to 63-fold, namely, miR -215, -192, -194 (all upregulated) and miR-205, miR-203 (both downregulated), and thereby, of potential use for BE diagnosis. Other investigators have examined differential miRNA expression between squamous and BE tissues. Although there were some differences in specific miRNAs detected in individual studies, miRNAs -215, -192, -194, -205, and -203 were consistent across multiple studies tissues ( Table 1 ). Bansal and colleagues used next-generation sequencing to define the miRNA transcriptome of patients with gastroesophageal reflux disease (GERD) and BE and found several novel miRNAs, including the above-mentioned miRNAs. They subsequently examined the upregulated miRNAs, -192, -215, and -194 for BE diagnosis and found a high sensitivity (91%–100%) and specificity (94%). Systematic analysis by Leidner and colleagues showed that most changes in miRNA expression occurred between squamous versus BE tissues rather than BE versus esophageal adenocarcinoma (EAC) tissues. Garman and colleagues made similar observations that miRNA expression may be more useful for BE diagnosis rather than risk stratification. There are continental differences in BE definition that need to be taken into account however (gastric vs intestinal metaplasia). MicroRNAs may be able to differentiate the 2 BE subtypes ( Box 2 ).
- •
A study showed miRNA profiles for both gastric- and intestinal-type BE to be different from squamous epithelium. When the 2 subtypes were compared, miRNAs -192, -205, and -203 were different, whereas miRNA-215 was similarly expressed.
- •
Thus, a set of miRNAs may be useful in detection of both gastric- and intestinal-type BE, but specific miRNAs may allow determination of BE subtype.
Summary
Overall, there is good agreement among studies for significantly different miRNAs between the squamous epithelium and the BE epithelium. These early data are promising, but the clinical accuracy of miRNAs for BE diagnosis still needs to be defined in multicenter studies.
DNA Methylation
Introduction
DNA methylation can be defined by the addition of methyl groups to DNA at cytosine bases that typically negatively regulate gene expression. Although initially examined for risk stratification, recent data suggest that methylation abnormalities occur early in BE and thus could be used for BE diagnosis.
Studies
One of the first genes shown to be widely methylated in BE was p16. Subsequently, multiple studies have used candidate and global approaches to demonstrate significant overlap in aberrant methylation signatures between BE and EAC when compared with squamous epithelium. Important methylated genes that have been examined in BE are p16, APC, RUNX3, MGMT, RBP1, SFRP1, TIMP3, and CDH13 among others. More recently, aberrant vimentin methylation has been suggested to be a potential marker for BE with tissue expression seen in 90% of BE and 0% among controls. These results were further confirmed on cytology specimens. A study by Xu and colleagues used pyrosequencing to show that the 10 most differentially expressed methylated CpG sites had a sensitivity of 94.8% and a specificity of 91.5% for discriminating BE from normal esophageal tissues. Comparison of methylation of 2 columnar tissues, gastric cardia and BE, with squamous tissues by Verma and colleagues appear to indicate that methylation changes may be specific to BE tissues. Recently, Iyer et al. found methylated genes NDRG4, SFRP1, BMP3, and HPP1 to be highly discriminant between BE tissues and normal controls; at a sensitivity of 100%, the specificity varied 84% to 96%. Expression levels of these markers was dependent on BE length, but not age, sex, inflammation, or the presence of dysplasia.
Summary
To summarize, methylated genes appear to discriminate well between BE and squamous tissues and should be examined further in multicenter trials on BE diagnosis.
Miscellaneous Biomarkers for Barrett’s Esophagus Diagnosis
Other markers that have been examined are esophageal stress proteins, such as anterior gradient-2. In a prospective study, anterior gradient-2 had a sensitivity of 65% and specificity of 90% for BE diagnosis. Newer nonendoscopic sampling devices may acquire gastric cardia and fundic mucosa cells. Therefore, markers that are able to differentiate between BE and gastric cardia and fundic tissues will be useful. Cdx2 and villin appear to be highly specific to BE tissues compared with gastric cardia and fundic-type mucosa but may be limited by sensitivity. Further studies that incorporate these markers are needed to define their performance.
Biomarkers for risk stratification in Barrett’s esophagus
The holy grail of biomarker research in BE is to identify the patients at risk for progression. In this section, biomarkers for risk stratification are discussed.
Chromosomal and Gene Gains and Deletions
Introduction
Chromosomal and gene gains and deletions can be detected by fluorescent in situ hybridization (FISH) that uses a fluorescently labeled probe to detect specific DNA sequences.
Studies
Feasibility of this approach was shown a decade ago when Falk and colleagues found chromosomal gains and loss of p53 and p16 in cytology specimens to have a sensitivity of 95% and a specificity of 100% for diagnosis of high-grade dysplasia (HGD)/EAC. A 4-probe FISH assay (8q24 (C-MYC), 9p21 (P16), 17q12 (HER2), and 20q13) was developed. A subsequent study showed this 4-probe FISH panel to depend on the dysplasia grade with a higher sensitivity for detection of HGD (82%) and EAC (100%) than low-grade dysplasia (LGD, 50%). This initial selection of probes was based on a candidate approach. High-throughput single nucleotide polymorphisms (SNPs) arrays have also been performed. Besides validating the previous targets, frequent gains in several novel candidate genes were identified but need further validation. FISH is readily applied to cytology samples but can be difficult in cut sections from paraffin-embedded tissues due to nuclear truncation. Therefore, separate assays will be needed before data from cytology samples can be directly applied to the paraffin sections. To complicate matters further, low-level copy gains may be missed on standard FISH in 4-μm sections and require thicker sections (16 μm) that need to be evaluated by image analysis. The ideal technique for FISH (cytology vs thin sections vs thick sections) still remains to be defined. Because FISH analysis can be tedious, investigators have attempted to develop automated protocols. A recent large trial examined FISH for risk stratification prospectively ( Box 3 ).
- •
Krishnadath and colleagues conducted a prospective multicenter trial that used a FISH panel ( P16 , P53 , Her-2/neu , 20q , and MYC , and the chromosomal centromeric probes 7 and 17 to detect aneuploidy) to analyze risk of progression in 428 patients with nondysplastic BE followed for a mean period of 45 months.
- •
The relative risk of progression with p16 loss or aneuploidy was 3.23 (95% CI 1.32–7.95) after controlling for Barrett’s length and age. The absolute risk of progression was 1.83% if the FISH panel was positive versus 0.58% if FISH panel was negative.
Summary
FISH is an important technique for quantitative detection of changes of aneuploidy and specific gene copy numbers during BE transformation. The issue that remains is whether FISH markers are sufficiently discriminatory to be applied in clinical practice.
P53
Introduction
P53 is one of the first biomarkers to show promise for identification of BE progressors. The focus of this review is on more recent work.
Studies
Investigators from the Netherlands analyzed p53 by immunohistochemistry (IHC) on sequential biopsies in patients with BE who ultimately progressed to HGD/EAC. Changes in p53 expression preceded development of HGD/EAC by several years, an important property for a biomarker. P53 expression was an important risk factor for HGD/EAC in a case control study with a hazard ratio (HR) of 6.5 (95% CI: 2.5–17.1). In the largest study of its kind, 2 pathologists independently scored p53 staining in 12,000 biopsies from 635 patients. Results were compelling with both overexpression and complete loss significantly increasing the risk of neoplastic progression after adjusting for age, gender, Barrett length, and esophagitis (relative risk [RR] 5.6 [95% CI 3.1–10.3] and RR 14.0 [95% CI 5.3–37.2], respectively). However, only 49% of patients who progressed had aberrant p53 immunostaining and hence it is specific but not highly sensitive. A nested case control study within a Northern Island registry of BE patients did not find p53 protein overexpression to predict progression in a multivariate analysis. Not all p53 mutations stabilize the protein, and complementary techniques may be needed ( Box 4 ).
- •
Mutated p53 frequently has a longer half-life and leads to increased p53 immunostaining. P53 mutations can also result in loss of protein, an absent pattern compared with wild-type. Therefore, both overexpression and loss of p53 appear to have clinical utility.
- •
Krishnadath and colleagues evaluated p53 expression by immunostaining for protein and FISH for loss of p53 locus and found the combination to detect p53 abnormalities in all patients with LGD, HGD, and EAC. Thus, p53 evaluation by FISH could complement immunostaining for risk stratification in BE.
- •
Furthermore, p53 mutational status (eg, point mutations and loss of heterozygosity) can be examined directly. Recent whole genome sequencing of esophageal biopsies from normal subjects and patients with BE and BE with HGD found p53 mutations to occur specifically in patients with HGD compared with patients with never dysplastic BE. Evaluation of esophageal samples acquired by the Cytosponge demonstrated that p53 mutations could be successfully detected, albeit with a lower allele frequency than in biopsy samples. The sensitivity and specificity of p53 mutations for prevalent HGD on the cytology samples were 86% and 100%, respectively.
Summary
Currently, p53 is not routinely recommended for risk stratification but the British Society of Gastroenterology does have a grade B recommendation to test p53 by IHC to clarify an equivocal histologic diagnosis of dysplasia. The above data also suggest that the same nonendoscopically acquired sample can be molecularly screened not only to diagnose BE but also to risk stratify.
DNA Content Abnormalities
Introduction
There are extensive long-term follow-up data on the measurement of aneuploidy and tetraploidy to identify BE progressors, yet these markers have not made it into clinical practice. An important reason is difficulty with aneuploidy measurements. Initial studies measured aneuploidy using flow cytometry that requires special media and specialized processing. A practical technique may be image cytometry that fares well against flow cytometry ( Box 5 ).
- •
Investigators evaluated 40-μm sections for aneuploidy by flow and image cytometry in 44 samples from 31 patients. The percentage agreement was 93% and suggests that image cytometry can reliably measure DNA content.
Studies
A landmark article by Reid and colleagues led to interest in aneuploidy/tetraploidy as a biomarker. Subsequently, the same investigators found aneuploidy/tetraploidy to have 100% sensitivity and 100% specificity to predict cancer development at 5 years. The ability of aneuploidy to predict progression was less dramatic in other studies in which it was a late marker or it lost predictive ability on multivariate analysis that included LGD as a covariate. More recent data in a population-based nested case-control study showed that the presence of aneuploidy measured by image cytometry increased the odds of progressing to HGD/EAC by 3.2-fold (1.73–6.0), but the overall sensitivity was limited to ∼44% (33%–55%). When aneuploidy, again measured by image, was incorporated into a risk score along with staining for Aspergillus oryzae lectin, a score of 1 or more could identify 66% of progressors with nondysplastic BE as baseline histology.
Summary
Current data suggest that aneuploidy alone may not be sufficiently discriminatory to risk-stratify BE patients but may be useful as part of a comprehensive biomarker panel. Measurement will likely need to be based on clinically applicable tests such as image cytometry. Conventional techniques measure DNA content as a whole but do not generally have the resolution to capture the details of gene-specific alterations that are manifest as a result of altered DNA content ( Box 6 ). Quantification of somatic chromosomal alterations is an exciting new area of biomarker research that evaluates the BE epithelium at an unprecedented global scale for risk profiling of BE patients.
- •
Recent collaborative efforts used genome-wide molecular techniques and determined that cancer development can be preceded by a dramatic increase in copy number and catastrophic genomic events in up to 30% of cases. This loss of genomic integrity could be a useful biomarker for progression. Specifically, catastrophic genomic events refer to sudden, punctuated alterations in copy number as a result of chromothripsis (massive genomic rearrangements as a single event) and breakage-fusion-bridge events.
- •
In addition to the actual copy number and specific loci of structural variants (eg, affecting a specific gene like ErbB2), the degree of copy number variation (termed genetic diversity) between different areas of the Barrett’s segment may also be an important risk factor for progression.
- •
Serial measurements demonstrated that increases in copy number and genetic diversity preceded cancer development by as many as 4 years, providing a window of opportunity for detection.
DNA Methylation
Introduction
Hypermethylation of genes has been extensively studied as a biomarker.
Studies
One of the earliest studies to evaluate methylation in BE was done by Reid and colleagues, who showed p16 methylation to be highly prevalent in EAC. Subsequently, multiple studies have evaluated methylated genes as biomarkers but have found inconsistent results (hypermethylation of p16 , RUNX3, and HPP1 in one study but APC, TIMP3, and TERT in another study ). Sato and colleagues proposed a tiered risk-stratification model by combining clinical parameters with a composite methylation index for p16 , RUNX3, and HPP1 . A follow-up study by the same group evaluated additional markers and found an 8-marker panel to have a specificity of 70% at a sensitivity of 80%. Hypermethylation of 2 of the markers ( p16 , APC ) described earlier were evaluated in a longitudinal study. The panel was found to predict progression (adjusted odds ratio 14.97), but with wide CIs (1.73, infinity). It remained unclear whether methylation was an early or a late event. A group of Australian investigators compared squamous tissues with BE and EAC tissues and elegantly demonstrated that most genes (7 of 9) previously shown to be hypermethylated in EAC were also hypermethylated in metaplastic BE, raising questions about their utility for diagnosis versus risk stratification. Studies discussed earlier used a candidate gene approach, but global methylation profiling is now possible. A group of investigators used methylation microarrays and observed that the number of differentially methylated CpG sites was 10-fold higher for BE versus squamous histology compared with EAC versus BE histology (195 vs 17, P = .001). Only 3 sites were different between BE and HGD cases, suggesting that methylation may be less useful as a biomarker for HGD. Methylation was further compared between BE and EAC in a comprehensive study that validated candidate genes using the gold-standard technique of pyrosequencing. A 4-gene methylation signature ( SLC22A18, PIGR, GJA12, and RIN2 ) could classify patients into low- (<2 methylated genes), medium- (2 methylated genes), and high-risk (>2 methylated genes) for prevalent dysplasia but needs further validation. Several studies suggest that both hypomethylation and hypermethylation are important during BE development and carcinogenesis ( Box 7 ).
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
