You are here

Article Text

Article menu

Download PDFPDF

Oesophagus
Non-endoscopic screening biomarkers for Barrett’s oesophagus: from microarray analysis to the clinic
  1. P Lao-Sirieix1,
  2. A Boussioutas2,3,
  3. S R Kadri1,
  4. M O’Donovan1,4,
  5. I Debiram1,5,
  6. M Das1,
  7. L Harihar1,
  8. R C Fitzgerald1
  1. 1
    MRC-Cancer Cell Unit, Hutchison–MRC Research Centre, Cambridge, UK
  2. 2
    Department of Medicine, University of Melbourne, Western Hospital, Melbourne, Australia
  3. 3
    Cancer Genomics and Predictive Medicine, Peter MacCallum Cancer Centre, East Melbourne, Australia
  4. 4
    Department of Histopathology, Addenbrooke’s Hospital, Cambridge, UK
  5. 5
    Experimental Cancer Medicine Centre, Department of Oncology, Addenbrooke’s Hospital, Cambridge, UK
  1. Correspondence to Dr R Fitzgerald, Hutchison–MRC Research Centre, Hills Road, Cambridge CB22 0XZ, UK; rcf{at}hutchison-mrc.cam.ac.uk

Abstract

Background and aims: Barrett’s oesophagus predisposes to oesophageal adenocarcinoma but the majority of patients are undiagnosed. A novel non-endoscopic cytological screening device, called a capsule sponge, makes population-based screening for the disease a feasible option. However, due to the mixed cell population retrieved by the capsule sponge, biomarkers specific for Barrett’s oesophagus are required.

Methods: Three publically available microarray datasets were used to identify putative biomarkers present in Barrett’s oesophagus but absent from normal oesophagus and gastric mucosa. Validation was performed by qPCR (n = 10 each of normal oesophagus, Barrett’s oesophagus, gastric mucosa) and immunohistochemistry (normal oesophagus, n = 20; Barrett’s oesophagus, n = 21; gastric mucosa, n = 24; duodenum, n = 18). The biomarker was then prospectively evaluated on capsule sponge specimens from 47 patients with Barrett’s oesophagus and 99 healthy controls.

Results: 2/14 genes identified, dopa decarboxylase (DDC) and Trefoil factor 3 (TFF3), were confirmed by qPCR to be upregulated in Barrett’s oesophagus compared to normal oesophagus (p<0.01) and gastric mucosa (p<0.01 and p<0.05, respectively). Immunohistochemistry confirmed that DDC protein expression was restricted to Barrett’s oesophagus but was confined to <1% of the cells within the crypt compartment. TFF3 protein was expressed to high levels at the luminal surface of Barrett’s oesophagus compared to absent expression in normal oesophagus and gastric mucosa (p<0.001). Using the capsule sponge 36/46 patients with Barrett’s oesophagus (one inadequate sample) and 6/96 controls were positive for TFF3 giving a sensitivity of 78% and a specificity of 94%.

Conclusions: TFF3 is a promising marker for Barrett’s oesophagus screening since it is expressed at the luminal surface of Barrett’s oesophagus but not in adjacent tissue types and may be applied to a non-endoscopic screening device.

https://doi.org/10.1136/gut.2009.180281

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Oesophageal adenocarcinoma is increasing rapidly in Western countries1 2 and patients usually present late with locally advanced disease leading to a dismal overall 5 year survival rate of 13%.3 Barrett’s oesophagus, as defined by intestinal metaplasia, is the major identified risk factor for this cancer4 and in those patients with oesophageal adenocarcinoma at presentation 90% have evidence of Barrett’s oesophagus following shrinkage of the tumour post-chemotherapy.5 However, because the majority (86%) of adenocarcinomas present de novo6 (without prior diagnosis of Barrett’s oesophagus) it is likely that a large number of patients with Barrett’s oesophagus remain undiagnosed in the population. This idea is supported by the high prevalence of Barrett’s oesophagus (7–25% for segments of any length and 0.7–7% for long-segment Barrett’s) in asymptomatic patients who agreed to have a screening upper gastrointestinal endoscopy when attending for colonoscopy in the United States.7 8 9 In keeping with these overall figures the prevalence of Barrett’s oesophagus of any length was reported to be between 1% and 8% in all-comers to endoscopy (reviewed by Pera10). The only population prevalence data available suggests that Barrett’s oesophagus is present in 1.6% of the general Swedish population.11

    Evidence from non-randomised retrospective studies demonstrated an improvement in 5-year actuarial survival from 13–43% to 62–100% in patients with surveillance-detected oesophageal adenocarcinomas.12 13 14 15 16 17 18 These data suggest a potential benefit for early detection although lead time bias needs to be accounted for. Rapid advances in endoscopic technologies (reviewed by Reddymasu and Sharma19) as well as the development of chemoprevention strategies20 21 afford the opportunity to improve patients’ outcomes if disease is detected early.

    Therefore, the rationale underlying this study is that identification of patients with undiagnosed Barrett’s oesophagus should ultimately reduce mortality from oesophageal adenocarcinoma. To attain this objective, population-based screening for Barrett’s may be necessary; however, there are major feasibility and cost implications for the wide-scale application of screening using the “gold standard” endoscopy.22 Novel screening strategies might include symptom nomograms, wireless capsule endoscopy and balloon cytology23 24 25 but these have not yet been demonstrated to be sufficiently sensitive and specific for clinical use.26 27 28 Recently, a study using a wireless video capsule attached to a string allowed a more careful examination of the oesophagus and eliminated a previous major imaging drawback of fast oesophageal transit time, with a reported sensitivity of 93.5%.29 However, this approach does not permit a pathological diagnosis or the potential for implementing risk stratification using biomarkers.

    We have recently developed a non-endoscopic capsule sponge device which has been approved by the Medical Health Regulatory Agency (Ref no. CI/2007/0053) in the UK. A pilot study demonstrated that this device is acceptable to patients and could be used in primary care.30 31 The device consists of a polyurethane sponge, contained within a gelatin capsule, which is attached to a string. The capsule is swallowed and dissolves within the stomach after 3–5 min. The sponge can then be retrieved by pulling on the string. Initial studies were performed using a cell monolayer stained with a proliferation marker mcm2. This gave a suboptimal sensitivity and specificity of 67.5% and 67.4%, respectively, and sample heterogeneity meant that the whole sample had to be processed and analysed for this single biomarker. More recently, we have processed the cytological specimen to a pellet which can then be embedded in paraffin thus preserving the tissue architecture. This can then undergo histological assessment and, in addition, multiple immunohistochemical markers may be used on a single sample.31 In order to improve diagnostic accuracy the identification and validation of biomarker(s) are required which have the ability to clearly distinguish between cells from Barrett’s oesophagus, normal gastric and squamous oesophageal cells since the capsule-sponge samples cells from the stomach to the oro-pharynx. Such a biomarker should be highly specific to minimise cost since, in a screening programme, patients with a positive capsule test would need to undergo endoscopy to verify the diagnosis and allow for multiple biopsies to be performed to exclude dysplasia.

    We hypothesised that biomarkers for Barrett’s oesophagus can be identified by combining and re-analysing a number of previously published upper gastrointestinal microarray datasets. The aims of this study were therefore to identify putative biomarkers from a combinatorial in silico analysis and to then perform validation studies on independent samples at the RNA and protein level before finally applying any candidates to samples from the capsule sponge collected from an independent cohort of patients with Barrett’s oesophagus and healthy controls.

    Material and methods

    Microarray analysis

    A search of the literature (PubMed) and public gene expression microarray databases (GEO, Stanford Microarray Database) was performed to identify microarray datasets pertaining to gene expression patterns in intestinal metaplasia-containing Barrett’s oesophagus, normal oesophagus and gastric mucosa (normal stomach, chronic gastritis and gastric intestinal metaplasia (GIM)). These tissues were selected since they will sampled by the capsule sponge. Chronic gastritis and GIM were chosen to represent upper gastrointestinal inflammation and Helicobacter-induced intestinal metaplasia, respectively, which may be present in the screened population and need to be distinguished from Barrett’s oesophagus. Three datasets, detailed in table 1, were selected for analysis based on the following criteria: (1) data was generated from more than five samples per relevant tissue type; and (2) the arrays used contained >10 000 cDNA or oligonucleotide probes. All three microarray studies involved the hybridisation of differentially labelled test and reference cDNA to a spotted cDNA array, and data from all three studies were available in the form of normalised test:reference hybridisation signal intensity ratios. Different analyses were performed to generate a single gene list (fig 1). Analysis 1: Hao et al32 (15 normal oesophagus and 14 Barrett’s oesophagus) was analysed, using a parametric test (Welch t test) with Bonferroni correction, to identify genes which were differentially expressed between the two groups and whose expression was significantly upregulated in Barrett’s oesophagus (log2 ratio >2) compared to normal oesophagus (p<0.0001). This list was then used to interrogate Boussioutas et al33 (57 gastric mucosa samples comprising normal stomach, chronic gastritis and GIM) for genes that were under-expressed in gastric mucosa (<1). Analysis 2: Greenawalt et al34 (39 normal oesophagus and 26 Barrett’s oesophagus) was analysed in a similar fashion to dataset 1 to produce a set of genes with log2 ratios Barrett’s oesophagus>2, normal oesophagus<1 and gastric mucosa<1 (from Boussioutas et al33). Data analysis was done using GeneSpring GX version 7.3 (Agilent, Palo Alto, California, USA). Genes common to both analyses were selected and ranked in order of statistical significance and enrichment in Barrett’s oesophagus for subsequent validation.

    Figure 1
    Figure 1

    Venn diagram of the number genes identified by analyses 1, 2 and genes common to the two microarray analyses. Analysis 1 and 2 yielded 24 and 93 putative targets respectively. A total of 14 genes common to both analyses were validated further. BO, Barrett’s oesophagus; CG, chronic gastritis; GIM, gastric intestinal metaplasia; NO, normal oesophagus; NS, normal stomach.

    View this table:
    Table 1

    Microarray datasets selected for analysis

    Human specimens

    Patients undergoing upper gastrointestinal endoscopy were recruited to this biomarker study from Addenbrooke’s Hospital following approval by the Local Research Ethics Committee. All patients with Barrett’s oesophagus had an endoscopically visible columnar lined segment of more than 3 cm and a histopathological diagnosis of specialised intestinal metaplasia. For normal oesophagus, samples were taken 2 cm above the z-line in patients with Barrett’s oesophagus and 2 cm above the gastro-oesophageal junction in patients without Barrett’s oesophagus who were undergoing symptomatic evaluation as part of the routine surveillance service.

    The microarray targets were validated using real-time polymerase chain reaction (RT-PCR) in 10 samples from each of Barrett’s oesophagus, normal oesophagus (five from Barrett’s patients and five from non-Barrett’s patients with a normal oesophagus) and 10 gastric mucosa samples (collected from the cardia of Barrett’s patients; table 2). The cardia was defined as 1 cm below the upper border of the gastric folds at the lower oesophagus in non-Barrett’s patients. A frozen section from each snap frozen Barrett’s specimen was analysed by a histopathologist to confirm the presence of intestinal metaplasia prior to RNA extraction.

    View this table:
    Table 2

    Clinical characteristics of cohorts

    The protein expression of putative biomarkers validated by RT-PCR was confirmed by immunohistochemistry on paraffin-embedded section from an independent cohort of 21 non-dysplastic Barrett’s oesophagus, 20 normal oesophagus, 24 gastric mucosa and 18 non-inflamed duodenum specimens which were used as a control columnar-lined tissue containing goblet cells (table 2).

    RNA extraction real-time PCR

    Total RNA from biopsies was extracted by using Trizol (Invitrogen). RNA (1 μg) was reverse transcribed using SuperScript II reverse transcriptase kit (Invitrogen, Paisley, UK) in 20 μl of total reaction solution. The primers used are listed in table 3. Positive controls were identified for each primer pair using a screen of 25 cells lines from different tissue origins. Quantitative PCR was performed on 2 μl of cDNA with the SYBR Green JumpStart Taq Readymix according to manufacturer’s instructions (Sigma-Aldrich, Poole, UK). PCR consisted of 40 cycles of 94°C denaturation (15 s), 51–57°C annealing (30 s; see table 3) and extension (30 s). The cycle threshold Ct was determined for each sample, and the average Ct of triplicate samples was calculated. The expression of each gene relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was determined as ΔCt. A melt curve was constructed for each primer.

    View this table:
    Table 3

    Putative biomarkers, primer sequences and conditions for polymserase chain reaction

    Immunohistochemistry

    Sections of 5 μm each were de-paraffinised in xylene and rehydrated in ethanol. Antigen retrieval was performed in microwave MicroMed T/T Mega (Milestone, Sorisole, Italy) in 0.01 mol/l citrate buffer pH 6.4. The staining procedure was performed using the Dako EnVision™ + System (DakoCytomation, Ely, UK). Briefly, non-specific binding was blocked by incubation in 5% bovine serum albumin (BSA) in Tris-buffered saline (TBS)–Tween 0.05% for 1 h and endogenous peroxidises were blocked with the hydrogen peroxide provided with the kit. Tissue sections were incubated with the primary antibody, either mouse anti-TFF3 (R&D Systems Europe, Abington, UK) or mouse anti-DDC (Protos Biotech Corporation, New York, USA) in 1% BSA in TBS–Tween 0.05% for 1 h at room temperature. The labelled polymer provided with the kit was then applied for 45 min followed by 3,3′-diaminobenzidine (DAB) substrate (DakoCytomation) for 10 min. Sections were counterstained with haematoxylin. A negative control was performed by omission of the primary antibody. Since the capsule sponge samples surface epithelium, quantification of immunohistochemical staining was restricted to the four top-most layers of the mucosa. A mean of the extent and intensity was generated for each biopsy, reviewed at high magnification (×400), to generate an overall score for each slide. The intensity score was: 0 if absent, 1 for weak, 2 for medium, and 3 for strong staining. The extent of staining was scored 1 for focal (one focus of positive cells), 2 for multifocal (two or more foci of positive cells) and 3 for extensive (whole biopsy stained) staining.

    Capsule sponge specimens

    Forty-six patients with known long-segment Barrett’s oesophagus and 99 control patients (table 2), whose diagnosis was verified by endoscopy, were recruited. Only patients with a segment ⩾3 cm were recruited to avoid erroneous diagnoses of hiatus hernia. Control patients were selected on the basis that they had reflux symptoms sufficient to require a prescription of acid suppressant for a minimum of 3 months over the last 5 years but without diagnosis of Barrett’s oesophagus. The patients were invited to attend a clinic at which they swallowed the sponge with a bolus of water and the capsule was left in place for 5 min before retrieval in preservative solution (SurePath, Burlington, North Carolina, USA) as previously described.30

    Processing of the capsule sponge specimens

    Samples were left in preservative solution for a minimum of 1 h. The samples were incubated for 30 min in Cytolyt® solution (Cytyc Corporation, Crawley, UK), washed twice in PBS and pelleted at 250×g for 5 min. The resulting pellet was re-suspended in 500 μl of plasma and thrombin (Diagnostic reagents, Oxford, UK) was then added in 10 μl increments until a clot formed. The clot was then placed in formalin for 24 h prior to processing into a paraffin block. The entire sample was cut in 5 μm sections to provide 20 slides. The first slide and every tenth slide were stained with haematoxylin & eosin. Two sections representative of the whole sample, 10 slides apart, were stained for TFF3 as described above. A slide was scored positive for TFF3 if any cell was stained for TFF3.

    Statistical analysis

    A Kruskal–Wallis one-way analysis of variance by ranks was performed to analyse differences in mRNA expression and expression of TFF3 at the luminal surface between the three groups using Prism (GraphPad Software). Specific differences were identified using a Dunn’s post-test. Microarray analysis was performed as described above.

    Results

    Identification of putative targets

    We used a strategy involving three microarray datasets to screen for candidate genes that were specifically expressed in Barrett’s oesophagus. Twenty-four genes from analysis 1 (Hao/Boussioutas) were found to be statistically over-expressed in Barrett’s oesophagus (log2 ratio >2) compared to normal oesophagus (log2 ratio <1) and gastric mucosa (log2<1). Using the same log2 ratio comparisons 93 genes were identified from analysis 2 (Greenawalt/Boussioutas) (fig 1). Fourteen genes (table 3 and fig 1) were common to both analyses.

    Validation of targets

    The increased expression in Barrett’s oesophagus compared to normal oesophagus and gastric mucosa was first confirmed at the mRNA level by real-time PCR in 10 histopathologically verified endoscopic biopsies from each tissue type. A suitable positive control could not be identified for FUT4 despite evaluating three primer pairs across 25 cells lines from different tissues. Validation of this target gene was therefore not taken any further. The expression of DAPK1 and PLCL2 was not statistically different between any groups. Most targets (AGR2, ATP7B, FBP1, FMO5, FOXA3, GOLPH2, LYZ, RNAase4 and TFF1) were statistically increased in Barrett’s oesophagus compared to normal oesophagus but were similar to gastric mucosa (fig 2). However, both dopa decarboxylase (DDC) and Trefoil factor 3 (TFF3) were statistically over-expressed in Barrett’s oesophagus compared to normal oesophagus (p<0.001 and p<0.01, respectively) and gastric mucosa (p<0.01 and p<0.05, respectively).

    Figure 2
    Figure 2

    mRNA expression of the putative genes identified. The Y axis represents the −ΔCt (defined as −(Cttarget − CtGAPDH)). The stars indicate statistical significance by one-way ANOVA (*p = 0.0339 and **p = 0.0012); only genes whose expression was significantly increased in Barrett’s oesphagus compared to normal oesphagus and gastric mucosa are marked on this graph. In both cases, the genes were statistically upregulated in Barrett’s oesphagus compared to normal oesphagus and gastric cardia. In addition, AGR2, ATP7B, FBP1, FMO5, FOXA3, GOLPH2, LYZ, RNAase4 and TFF1 are statistically upregulated in BO (Barrett’s oesophagus) and gastric cardia (gastric cardia) compared to NO (normal oesophagus).

    Since the capsule sponge specifically samples the uppermost layers of the mucosa, we then went on to validate the expression of TFF3 and DDC in paraffin-embedded section to address the epithelial localisation of the antigen. TFF3 was expressed to high levels in Barrett’s oesophagus compared to normal oesophagus and gastric mucosa both at the luminal surface and deeper within the tissue (figs 3 and 4; p<0.001). TFF3 was therefore applicable to the capsule sponge which samples the surface layers of the gastric cardia and oesophagus. In contrast, DDC expression was only seen in 4/19 patients with Barrett’s oesophagus and was absent in adjacent tissues (normal oesophagus or gastric mucosa) as expected. In the positive Barrett’s oesophagus samples, DDC expression was very weak, limited to a small cluster of cells (fewer than eight) and localised towards the bottom of the crypts (fig 5).

    Figure 3
    Figure 3

    Representative immunohistochemistry (×100) of TFF3 in the positive control duodenum, normal oesophagus, Barrett’s oesophagus, gastric mucosa. TFF3, Trefoil factor 3.

    Figure 4
    Figure 4

    Cumulative score demonstrates over-expression of Trefoil factor 3 (TFF3) in Barrett’s oesophagus (BO) compared to normal oesophagus (NO) and gastric mucosa (GM), (p<0.0001).

    Figure 5
    Figure 5

    Expression of dopa decarboxylase in prostate cancer, used as positive control, and in a positive Barrett’s oesophagus biopsy (×100 and ×400 magnification).

    TFF3 expression in samples collected with the capsule sponge

    Since TFF3 fulfilled the necessary criteria, in that expression was restricted to the luminal surface of Barrett’s oesophagus with no expression seen in gastric or normal squamous oesophageal tissues this was taken forward to the prospective capsule sponge screening study. TFF3 expression in specimens from 46 histologically confirmed patients with Barrett’s oesophagus were compared to 99 patients without Barrett’s oesophagus. One sample from a patient with known Barrett’s oesophagus had a low cell yield and was excluded from the analysis. The staining was very intense (fig 6) and a dichotomous score (staining present or not) was used to maximise specificity. Thirty-six out of 46 patients and six out of 99 control patients had a capsule sponge specimen positive for TFF3. A sensitivity of 78% (95% confidence interval (CI), 64 to 89), specificity of 94% (95% CI, 87 to 98) and a correct proportion of samples diagnosed of 89% (95% CI, 83 to 94) were obtained.

    Figure 6
    Figure 6

    Representative haematoxylin & eosin and Trefoil factor 3 (TFF3) staining of a capsule specimen collected from a patient with Barrett’s oesophagus (×100 and ×400 magnification). The black arrow indicates the typical circular appearance of TFF3 positivity and the red arrow indicates secreted TFF3 at the apical border of the Barrett’s cells.

    Discussion

    We have demonstrated for the first time that microarray analysis can be directly applied to the clinic in the context of Barrett’s oesophagus. We identified TFF3 as a candidate biomarker for Barrett’s oesophagus using commercially available DNA microarray data. We demonstrated that it was over-expressed in Barrett’s oesophagus compared to adjacent normal tissues at the RNA and protein level and then successfully applied its use to a novel minimally invasive, non-endoscopic screening strategy.

    There is an established precedent for the use of DNA microarray data as a classifier in prognostication of breast cancer.35 36 37 There are a number of classifiers in clinical use to help select patients with specific criteria into an appropriate treatment regimen (Mammoprint and Oncotype DX platforms). To our knowledge there are no screening biomarkers derived from a microarray experiment that have been successfully translated to clinical use. This study is a substantial leap forward in the translation of high-throughput laboratory results into an assay that can be used in the clinic. There were some limitations to the microarray experiments conducted but these do not detract from the results obtained. The microarray experiments were not designed specifically to identify markers distinguishing between Barrett’s oesophagus, normal oesophagus and gastric mucosa and to our knowledge no dataset including normal oesophagus, Barrett’s oesophagus and normal gastric mucosa exists. Furthermore, the microarray platform by Hao et al32 was different from the platform used by Boussioutas et al33 and Greenawalt et al.34 This explains the lower number of candidates identified in the first (Hao–Greenawalt; n = 33) compared with the second analysis (Boussioutas–Greenawalt; n = 111). However, very stringent statistical criteria were set to reduce the effect of these shortfalls which also reduced the number of putative targets. It is interesting to note that only two out of 14 targets were validated by qPCR and in most cases this was because the expression level of the putative markers was similar in gastric mucosa (cardia) and Barrett’s oesophagus. This suggests that the expression profile of the cardia is closer to Barrett’s oesophagus than normal gastric mucosa, chronic gastritis and intestinal metaplasia of the cardia. The design of the microarray experiments would not be able to differentiate this because only distal gastric mucosa was used in their experiments and gastric cardia was not sampled. It has previously been demonstrated that the kinome38 and the expression profile39 of Barrett’s oesophagus have strong similarities to that of gastric cardia.

    Trefoil factors are mucin-associated peptides thought to be involved in multiple biological functions including repair of the mucosa through enhancement of restitution (mucosal repair by cell migration) and modulation of stem cells differentiation as well as interaction with mucins and modulation of the mucosal immune response.40 41 42 43 In the gastrointestinal tract, TFF1 and 2 are mainly expressed by the gastric epithelium42 while TFF3 is expressed by the intestine and intestinal metaplasia of the stomach and oesophagus.44 45 TFF3 expression has been demonstrated to be increased by gastro-oesophageal reflux disease,45 and transient over-expression of the homeodomain protein CDX2,46 linking its expression to the development of Barrett’s oesophagus. Interestingly, there have been a number of publications concerning the development of novel cytological methods using TFF3 to detect thyroid follicular carcinomas47 48 and it has been suggested that TFF3 could also be detected in the serum of patients with high-grade endometrial carcinomas.49 It might therefore be possible to develop a serum-based assay to screen for Barrett’s oesophagus using TFF3 as a marker.

    Dopa decarboxylase metabolises 3,4-dihydroxyphenylalanine (DOPA) to dopamine, and 5-hydroxytryptophan to serotonin.50 Over-expression of DDC is also a feature of a number of malignancies ranging from retinoblastomas51 52 to small-cell lung cancers,53 prostate cancers54 and gastric cancers with peritoneal disseminations.55 In most of these tumour types, an effort has been made to use DDC diagnostically using PCR techniques in biopsies54 56 or cytological specimens.55 DDC may therefore be of interest as a biomarker for malignant conversion in Barrett’s oesophagus and due to the poor immunohistochemistry staining a PCR-based assay may be more applicable.

    Neither the British Society of Gastroenterology nor the American Gastroenterology Association currently recommend endoscopic screening for Barrett’s oesophagus22 57 (recommendations grade C and B respectively, both based on cohort and case–control studies). However, both professional organisations agree that surveillance is in order once Barrett’s oesophagus is diagnosed. However, for surveillance to be of use, all patients with symptomatic or asymptomatic Barrett’s oesophagus would need to be diagnosed. This suggests population screening for Barrett’s oesophagus may be recommended if a cost effective test could be developed. Furthermore, in the UK, the 2008 report by the Chief Medical Officer highlighted that “research should be supported to explore the possibilities of new diagnostic techniques, including potential minimally invasive screening tests”.58

    The advantage of this novel capsule sponge test is that it can be performed in primary care. Since the architecture of the tissue is well conserved, the H & E slide were also reviewed by an expert gastrointestinal histopathologist but morphology alone was not sufficient to diagnose Barrett’s oesophagus and is open to subjectivity. Since the TFF3 analysis presented here relies on standard immunohistochemical techniques it is a more objective test that would be readily applicable to clinical pathology laboratories in a cost effective manner. Furthermore, in the future other assays could be applied to these sponge samples such as PCR-based assays to determine gene expression levels of multiple biomarkers or DNA based assays to determine methylation status or loss of heterozygosity. Such methodologies would increase the price of the test but may also be informative with regards to the risk of progression to cancer. The typical circumscribed appearance of TFF3 positivity and the strength of the staining make it particularly suited for automation thus potentially further reducing the cost of a screening programme incorporating this methodology. For such a Barrett’s oesophagus screening test, it would be essential to have a high specificity to avoid calling patients for unnecessary endoscopies with the inherent generation of anxiety, high costs and risks of an invasive procedure that this would entail. With TFF3 we obtained a very high specificity of 94% and an adequate sensitivity of 78%. It is unlikely for a single marker to provide both a high sensitivity and specificity. It is reasonable to accept that a device like the capsule sponge, while sampling from the entire surface of the mucosa, will only collect cells that are detached. Small foci of intestinal metaplasia, yielding TFF3 positivity, may be missed, explaining the sensitivity of 78%. Furthermore, the methodology used for marker identification ensured high specificity since we were looking for best discriminators between Barrett’s oesophagus and normal oesophagus and gastric mucosa but not necessarily sensitivity because the sample size of the microarrays was not high enough. Current screening programmes for prostate (prostate serum antigen (PSA)), cervical (Papanicolaou test) and colon cancer (faecal occult blood test (FOBT)) have accepted sensitivities of 30–96%, respectively, and specificities of 77–100%, respectively.59 60 61 The positive predictive value of PSA for prostate cancer and of FOBT for colon cancer is 47% and 2.2–17.7%, respectively.59 60 It would therefore be desirable to identify additional markers that, used in conjunction of TFF3, would offer a high sensitivity without loss of specificity. In addition, a prospective study in the primary care setting is necessary to validate TFF3 in a larger cohort since the PPV and NPV cannot be assessed properly on a cohort enriched for patients with Barrett’s oesophagus.

    In conclusion, we have demonstrated that clinically relevant biomarkers can be identified through the use of microarray data and careful validation. Biomarkers identified using such an approach could be used in conjunction with the capsule sponge test to provide a cost-effective and acceptable screening test for Barrett’s oesophagus.

    Acknowledgments

    We wish to thank the patients and volunteers whose samples were used for this research.

    REFERENCES

    1. ↵
      1. Brown LM,
      2. Devesa SS,
      3. Chow WH
      . Incidence of adenocarcinoma of the esophagus among white Americans by sex, stage, and age. J Natl Cancer Inst 2008;100:1184–7.
      OpenUrlAbstract/FREE Full Text
    2. ↵
      1. Lepage C,
      2. Rachet B,
      3. Jooste V,
      4. et al
      . Continuing rapid increase in esophageal adenocarcinoma in England and Wales. Am J Gastroenterol 2008;103:2694–9.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Group MRCOCW
      . Surgical resection with or without preoperative chemotherapy in oesophageal cancer: a randomised controlled trial. Lancet 2002;359:1727–33.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Chandrasoma P,
      2. Wickramasinghe K,
      3. Ma Y,
      4. et al
      . Is intestinal metaplasia a necessary precursor lesion for adenocarcinomas of the distal esophagus, gastroesophageal junction and gastric cardia? Dis Esoph 2007;20:36–41.
      OpenUrlCrossRefPubMedWeb of Science
    5. ↵
      1. Theisen J,
      2. Stein HJ,
      3. Dittler HJ,
      4. et al
      . Preoperative chemotherapy unmasks underlying Barrett’s mucosa in patients with adenocarcinoma of the distal esophagus. Surg Endosc 2002;16:671–3.
      OpenUrlCrossRefPubMedWeb of Science
    6. ↵
      1. Schlansky B,
      2. Dimarino AJ Jr,
      3. Loren D,
      4. et al
      . A survey of oesophageal cancer: pathology, stage and clinical presentation. Aliment Pharmacol Therap 2006;23:587–93.
      OpenUrlCrossRefPubMed
    7. ↵
      1. Gerson LB,
      2. Shetler K,
      3. Triadafilopoulos G
      . Prevalence of Barrett’s esophagus in asymptomatic individuals. Gastroenterology 2002;123:461–7.
      OpenUrlCrossRefPubMed
    8. ↵
      1. Rex DK,
      2. Cummings OW,
      3. Shaw M,
      4. et al
      . Screening for Barrett’s esophagus in colonoscopy patients with and without heartburn. Gastroenterology 2003;125:1670–7.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      1. Ward EM,
      2. Wolfsen HC,
      3. Achem SR,
      4. et al
      . Barrett’s esophagus is common in older men and women undergoing screening colonoscopy regardless of reflux symptoms. Am J Gastroenterol 2006;101:12–7.
      OpenUrlCrossRefPubMed
    10. ↵
      1. Pera M
      . Trends in incidence and prevalence of specialized intestinal metaplasia, Barrett’s esophagus, and adenocarcinoma of the gastroesophageal junction. World J Surg 2003;27:999–1006.
      OpenUrlCrossRefPubMedWeb of Science
    11. ↵
      1. Ronkainen J,
      2. Aro P,
      3. Storskrubb T,
      4. et al
      . Prevalence of Barrett’s esophagus in the general population: an endoscopic study. Gastroenterology 2005;129:1825–31.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Streitz JM Jr,
      2. Andrews CW Jr,
      3. Ellis FH Jr
      . Endoscopic surveillance of Barrett’s esophagus. Does it help? J Thorac Cardiovasc Surg 1993;105:383–7; discussion 7–8.
      OpenUrlPubMedWeb of Science
    13. ↵
      1. Peters JH,
      2. Clark GW,
      3. Ireland AP,
      4. et al
      . Outcome of adenocarcinoma arising in Barrett’s esophagus in endoscopically surveyed and nonsurveyed patients. J Thorac Cardiovasc Surg 1994;108:813–21; discussion 21–2.
      OpenUrlPubMedWeb of Science
    14. ↵
      1. van Sandick JW,
      2. van Lanschot JJ,
      3. Kuiken BW,
      4. et al
      . Impact of endoscopic biopsy surveillance of Barrett’s oesophagus on pathological stage and clinical outcome of Barrett’s carcinoma. Gut 1998;43:216–22.
      OpenUrlAbstract/FREE Full Text
    15. ↵
      1. Incarbone R,
      2. Bonavina L,
      3. Saino G,
      4. et al
      . Outcome of esophageal adenocarcinoma detected during endoscopic biopsy surveillance for Barrett’s esophagus. Surg Endosc 2002;16:263–6.
      OpenUrlCrossRefPubMed
    16. ↵
      1. Ferguson MK,
      2. Durkin A
      . Long-term survival after esophagectomy for Barrett’s adenocarcinoma in endoscopically surveyed and nonsurveyed patients. J Gastrointest Surg 2002;6:29–35; discussion 6.
      OpenUrlCrossRefPubMed
    17. ↵
      1. Fountoulakis A,
      2. Zafirellis KD,
      3. Dolan K,
      4. et al
      . Effect of surveillance of Barrett’s oesophagus on the clinical outcome of oesophageal cancer. Br J Surg 2004;91:997–1003.
      OpenUrlCrossRefPubMedWeb of Science
    18. ↵
      1. Corley DA,
      2. Levin TR,
      3. Habel LA,
      4. et al
      . Surveillance and survival in Barrett’s adenocarcinomas: a population-based study. Gastroenterology 2002;122:633–40.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      1. Reddymasu SC,
      2. Sharma P
      . Advances in endoscopic imaging of the esophagus. Gastroenterol Clin North Am 2008;37:763–74.
      OpenUrlCrossRefPubMed
    20. ↵
      1. Mehta S,
      2. Johnson IT,
      3. Rhodes M
      . Systematic review: the chemoprevention of oesophageal adenocarcinoma. Aliment Pharmacol Therap 2005;22:759–68.
      OpenUrlCrossRefPubMed
    21. ↵
      1. Ilyas S,
      2. DeMars C,
      3. Buttar N
      . Chemoprevention in Barrett’s esophagus. J Gastrointest Cancer 2007;38:1–9.
      OpenUrlCrossRefPubMed
    22. ↵
      BSG. Guidelines for the diagnosis and management of Barrett’s columnar-lined oesophagus. http://www.bsg.org.uk, 2005.
    23. ↵
      1. Fennerty MB,
      2. DiTomasso J,
      3. Morales TG,
      4. et al
      . Screening for Barrett’s esophagus by balloon cytology. Am J Gastroenterol 1995;90:1230–2.
      OpenUrlPubMed
    24. ↵
      1. Falk GW,
      2. Chittajallu R,
      3. Goldblum JR,
      4. et al
      . Surveillance of patients with Barrett’s esophagus for dysplasia and cancer with balloon cytology. Gastroenterology 1997;112:1787–97.
      OpenUrlCrossRefPubMed
    25. ↵
      1. Spechler SJ
      . Barrett’s esophagus: should we brush off this ballooning problem? Gastroenterology 1997;112:2138–42.
      OpenUrlCrossRefPubMed
    26. ↵
      1. Gerson LB,
      2. Edson R,
      3. Lavori PW,
      4. et al
      . Use of a simple symptom questionnaire to predict Barrett’s esophagus in patients with symptoms of gastroesophageal reflux. Am J Gastroenterol 2001;96:2005–12.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Locke GR,
      2. Talley NJ,
      3. Weaver AL,
      4. et al
      . A new questionnaire for gastroesophageal reflux disease. Mayo Clin Proc 1994;69:539–47.
      OpenUrlPubMedWeb of Science
    28. ↵
      1. Sharma P,
      2. Wani S,
      3. Rastogi A,
      4. et al
      . The diagnostic accuracy of esophageal capsule endoscopy in patients with gastroesophageal reflux disease and Barrett’s esophagus: a blinded, prospective study. Am J Gastroenterol 2008;103:525–32.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Ramirez FC,
      2. Akins R,
      3. Shaukat M
      . Screening of Barrett’s esophagus with string-capsule endoscopy: a prospective blinded study of 100 consecutive patients using histology as the criterion standard. Gastrointest Endosc2008;68:25–31.
      OpenUrlCrossRefPubMedWeb of Science
    30. ↵
      1. Lao-Sirieix P,
      2. Rous B,
      3. O’Donovan M,
      4. et al
      . Non-endoscopic immunocytological screening test for Barrett’s oesophagus. Gut 2007;56:1033–4.
      OpenUrlFREE Full Text
    31. ↵
      1. Lao-Sirieix P,
      2. Kadri S,
      3. Debiram I,
      4. et al
      . Non-endoscopic screening for Barrett’s oesophagus using the capsule sponge is well tolerated and allows tissue architecture to be retained. J Clin Gastroenterol 2008;42(S1):Abstract.
      OpenUrl
    32. ↵
      1. Hao Y,
      2. Triadafilopoulos G,
      3. Sahbaie P,
      4. et al
      . Gene expression profiling reveals stromal genes expressed in common between Barrett’s esophagus and adenocarcinoma. Gastroenterology 2006;131:925–33.
      OpenUrlCrossRefPubMedWeb of Science
    33. ↵
      1. Boussioutas A,
      2. Li H,
      3. Liu J,
      4. et al
      . Distinctive patterns of gene expression in premalignant gastric mucosa and gastric cancer. Cancer Res 2003;63:2569–77.
      OpenUrlAbstract/FREE Full Text
    34. ↵
      1. Greenawalt DM,
      2. Duong C,
      3. Smyth GK,
      4. et al
      . Gene expression profiling of esophageal cancer: comparative analysis of Barrett’s esophagus, adenocarcinoma, and squamous cell carcinoma. Int J Cancer 2007;120:1914–21.
      OpenUrlCrossRefPubMedWeb of Science
    35. ↵
      1. Paik S,
      2. Shak S,
      3. Tang G,
      4. et al
      . A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N Engl J Med 2004;351:2817–26.
      OpenUrlCrossRefPubMedWeb of Science
    36. ↵
      1. Paik S,
      2. Tang G,
      3. Shak S,
      4. et al
      . Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol 2006;24:3726–34.
      OpenUrlAbstract/FREE Full Text
    37. ↵
      1. van ’t Veer LJ,
      2. Dai H,
      3. van de Vijver MJ,
      4. et al
      . Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002;415:530–6.
      OpenUrlCrossRefPubMedWeb of Science
    38. ↵
      1. van Baal JWPM,
      2. Diks SH,
      3. Wanders RJA,
      4. et al
      . Comparison of kinome profiles of Barrett’s esophagus with normal squamous esophagus and normal gastric cardia. Cancer Res 2006;66:11605–12.
      OpenUrlAbstract/FREE Full Text
    39. ↵
      1. van Baal JW,
      2. Milano F,
      3. Rygiel AM,
      4. et al
      . A comparative analysis by SAGE of gene expression profiles of Barrett’s esophagus, normal squamous esophagus, and gastric cardia. Gastroenterology 2005;129:1274–81.
      OpenUrlCrossRefPubMed
    40. ↵
      1. Dignass A,
      2. Lynch-Devaney K,
      3. Kindon H,
      4. et al
      . Trefoil peptides promote epithelial migration through a transforming growth factor beta-independent pathway. J Clin Invest 1994;94:376–83.
      OpenUrlCrossRefPubMedWeb of Science
    41. ↵
      1. Wong WM,
      2. Poulsom R,
      3. Wright NA
      . Trefoil peptides. Gut 1999;44:890–5.
      OpenUrlFREE Full Text
    42. ↵
      1. Hoffmann W,
      2. Jagla W,
      3. Wiede A
      . molecular medicine of TFF-peptides: from gut to brain. Histol Histopathol 2001;13:319–34.
      OpenUrl
    43. ↵
      1. Baus-Loncar M,
      2. Kayademir T,
      3. Takaishi S,
      4. et al
      . Trefoil factors. Cell Mol Life Sci (CMLS) 2005;62:2947–55.
      OpenUrlCrossRef
    44. ↵
      1. Kouznetsova I,
      2. Kalinski T,
      3. Peitz U,
      4. et al
      . Localization of TFF3 peptide in human esophageal submucosal glands and gastric cardia: differentiation of two types of gastric pit cells along the rostro-caudal axis. Cell Tissue Res 2007;328:365–74.
      OpenUrlCrossRefPubMed
    45. ↵
      1. Peitz U,
      2. Kouznetsova I,
      3. Wex T,
      4. et al
      . TFF3 expression at the esophagogastric junction is increased in gastro-esophageal reflux disease (GERD). Peptides 2004;25:771–7.
      OpenUrlCrossRefPubMed
    46. ↵
      1. Shimada T,
      2. Koike T,
      3. Yamagata M,
      4. et al
      . Regulation of TFF3 expression by homeodomain protein CDX2. Regul Peptides 2007;140:81–7.
      OpenUrlCrossRefPubMed
    47. ↵
      1. Takano T,
      2. Yamada H
      . Trefoil factor 3 (TFF3): a promising indicator for diagnosing thyroid follicular carcinoma. Endocr J 2008; doi:10.1507/endocrj.K08E-105.
    48. ↵
      1. Bryson PC,
      2. Shores CG,
      3. Hart C,
      4. et al
      . Immunohistochemical distinction of follicular adenomas and follicular carcinomas. Arch Otolaryngol Head Neck Surg 2008;134:581–6.
      OpenUrlCrossRefPubMed
    49. ↵
      1. Bignotti E,
      2. Ravaggi A,
      3. Tassi RA,
      4. et al
      . Trefoil factor 3: a novel serum marker identified by gene expression profiling in high-grade endometrial carcinomas. Br J Cancer 2008;99:768–73.
      OpenUrlCrossRefPubMed
    50. ↵
      1. Christenson JG,
      2. Dairman W,
      3. Udenfriend S
      . On the identity of DOPA decarboxylase and 5-hydroxytryptophan decaroxylase. Proc Natl Acad Sci USA 1972;69:343–7.
      OpenUrlAbstract/FREE Full Text
    51. ↵
      1. Eldrup E,
      2. Clausen N,
      3. Scherling B,
      4. et al
      . Evaluation of plasma 3,4-dihydroxyphenylacetic acid (DOPAC) and plasma 3,4-dihydroxyphenylalanine (DOPA) as tumor markers in children with neuroblastoma. Scand J Clin Lab Invest 2001;61:479–90.
      OpenUrlCrossRefPubMed
    52. ↵
      1. Gilbert J,
      2. Haber M,
      3. Bordow SB,
      4. et al
      . Use of tumor-specific gene expression for the differential diagnosis of neuroblastoma from other pediatric small round-cell malignancies. Am J Pathol 1999;155:17–21.
      OpenUrlPubMedWeb of Science
    53. ↵
      1. Jensen SM,
      2. Gazdar AF,
      3. Cuttitta F,
      4. et al
      . A comparison of synaptophysin, chromogranin, and L-dopa decarboxylase as markers for neuroendocrine differentiation in lung cancer cell lines. Cancer Res 1990;50:6068–74.
      OpenUrlAbstract/FREE Full Text
    54. ↵
      1. Avgeris M,
      2. Koutalellis G,
      3. Fragoulis EG,
      4. et al
      . Expression analysis and clinical utility of L-Dopa decarboxylase (DDC) in prostate cancer. Clin Biochem 2008;41:1140–9.
      OpenUrlCrossRefPubMed
    55. ↵
      1. Sakakura C,
      2. Takemura M,
      3. Hagiwara A,
      4. et al
      . Overexpression of dopa decarboxylase in peritoneal dissemination of gastric cancer and its potential as a novel marker for the detection of peritoneal micrometastases with real-time RT-PCR. Br J Cancer 2004;90:665–71.
      OpenUrlCrossRefPubMedWeb of Science
    56. ↵
      1. Trager C,
      2. Vernby A,
      3. Kullman A,
      4. et al
      . mRNAs of tyrosine hydroxylase and dopa decarboxylase but not of GD2 synthase are specific for neuroblastoma minimal disease and predicts outcome for children with high-risk disease when measured at diagnosis. Int J Cancer 2008;123:2849–55.
      OpenUrlCrossRefPubMedWeb of Science
    57. ↵
      1. Wang KK,
      2. Sampliner RE
      . Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett’s esophagus. Am J Gastroenterol 2008;103:788–97.
      OpenUrlCrossRefPubMedWeb of Science
    58. ↵
      1. Donaldson L
      . On the State of Public Health: Annual Report of the Chief Medical Officer 2007. 2008; http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/AnnualReports/DH_086176 (accessed 20 January 2009).
    59. ↵
      1. Bunting PS
      . Screening for prostate cancer with prostate-specific antigen: beware the biases. Clin Chim Acta 2002;315:71–97.
      OpenUrlCrossRefPubMedWeb of Science
    60. ↵
      1. Towler B,
      2. Irwig L,
      3. Glasziou P,
      4. et al
      . A systematic review of the effects of screening for colorectal cancer using the faecal occult blood test, hemoccult. BMJ 1998;317:559–65.
      OpenUrlAbstract/FREE Full Text
    61. ↵
      1. Nanda K,
      2. McCrory DC,
      3. Myers ER,
      4. et al
      . Accuracy of the Papanicolaou test in screening for and follow-up of cervical cytologic abnormalities: a systematic review. Ann Intern Med 2000;132:810–9.
      OpenUrlCrossRefPubMedWeb of Science
    View Abstract

    Footnotes

    • Funding This research was supported by the Medical Research Council, Cambridge Experimental Cancer Medicine Centre and the NIHR Cambridge Biomedical Research Centre.

    • Competing interests None.

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

    • Ethics approval Ethics approval was obtained from the Cambridgeshire Research Ethics Committee on 17 October 2003.

    Linked Articles

    • Digest
      Digest
      Robin Spiller Severine Vermeire
      Gut 2009; 58 i-ii Published Online First: 15 Oct 2009. doi: 10.1136/gut.2009.198820