Article Text
PDF
Abstract
Background Barrett oesophagus (BO) occurs as a consequence of prolonged gastro-oesophageal reflux and it may precede the development of oesophageal adenocarcinoma.
Aims The aim of this study was to accurately assess Cdx2 expression in the oesophageal metaplasia–dysplasia–adenocarcinoma sequence.
Method The expression of the transcription factor Cdx2 in Barrett metaplasia, high-grade glandular dysplasia and adenocarcinoma was investigated using immunohistochemistry.
Results The results confirmed previous immunohistochemistry and PCR-based investigations, indicating that Cdx2 was expressed by intestinal metaplasia in the oesophagus. In addition, upregulation followed by linear downregulation of Cdx2 expression through the oesophageal metaplasia–dysplasia–adenocarcinoma sequence was demonstrated for the first time.
Conclusion Such downregulation of Cdx2 expression could be in keeping with a role as a tumour suppressor gene, but a simpler explanation would be downregulation of a differentiation marker.
- Barrett oesophagus
- cancer
- immunocytochemistry
Statistics from Altmetric.com
Barrett oesophagus (BO) is defined as the replacement of the normal squamous epithelium of the lower oesophagus by metaplastic columnar epithelium.1 It occurs as a consequence of prolonged gastro-oesophageal reflux and it may precede the development of oesophageal adenocarcinoma, a malignancy that has shown a rapidly rising incidence over the last two decades.2 Most cases of BO will not progress to adenocarcinoma; however, its presence is an indicator for further endoscopic biopsy surveillance.1
Transcription factors are involved in the process of normal intestinal differentiation and Cdx2 is one such example, belonging to the caudal-related homeobox gene family.3 4 Its expression is intestine specific, with the proximal anatomical limit in the normal intestine located at the gastroduodenal junction.5
In the oesophagus, Cdx2 has a role in relation to differentiation and maintenance of intestinal metaplasia. It has been proposed that gastro-oesophageal reflux causes bile acids to induce expression of Cdx1 and Cdx2 in oesophageal keratinocytes. Overexpression of Cdx2 is thought to induce the intestinal mucin mucin-2 (MUC2) directly, whereas that of Cdx1 indirectly induces MUC2 via Cdx2 expression.6
Evidence gained from immunohistochemical (IHC) and reverse transcriptase-PCR studies suggests that Cdx2 expression precedes the morphological changes of BO,5 7 8 and its detection may represent a more sensitive method for the diagnosis of BO than routine histological examination for the presence of goblet cells. However, previous studies have produced conflicting results in relation to Cdx2 quantitation through the metaplasia–dysplasia–adenocarcinoma sequence.
The aim of this study was to accurately assess Cdx2 expression in the oesophageal metaplasia–dysplasia–adenocarcinoma sequence, and we wished to confirm expression of Cdx2 by oesophageal intestinal metaplasia. We believe that our method more accurately assesses Cdx2 expression in comparison with some previous studies, which have recorded expression purely as either positive or negative.
Materials and methods
Cases
Oesophageal endoscopic biopsies previously reported for diagnostic purposes were selected retrospectively from the tissue archive at Salford Royal NHS Foundation Trust, using the systematised nomenclature of human and veterinary medicine (SNOMED) coding system. Original sections, stained with H&E, were reviewed using light microscopy by a histopathologist with a gastrointestinal interest (SJH), to confirm correct classification. For assessment of intestinal metaplasia, cases that showed intestinal metaplasia with the presence of goblet cells were included in this study. In these cases, all glandular cells seen in association with goblet cells, and not only goblet cells, were included for assessment. In total 80 cases were included, comprising 20 consecutive cases each of normal oesophagus, BO, high-grade dysplasia (HGD) and adenocarcinoma. One biopsy was assessed for each individual patient. Cases were not discriminated with regard to specific location within the oesophagus. Biopsies diagnosed as HGD and adenocarcinoma were distinguished using standard criteria, based on the parameters of architectural and cytological atypia.1 9
Biopsies predominantly represented mucosal tissue, although some adenocarcinoma biopsies contained submucosal tissue.
Immunohistochemistry
All tissues had been fixed in neutral buffered formaldehyde and processed into paraffin wax by standard histological methods. Further sections of 3 μm thickness were cut and placed on adhesive-coated slides (Snowcoat; Surgipath Europe, Bretton, UK).
Sections were dried overnight at 60°C. Antigen retrieval was performed in 0.01 M Tris-EDTA-Tween, pH 8.5, for 30 min in a 900 W microwave. IHC staining was performed on an automated immunostainer (Biogenex i6000; Biogenex, San Ramon, California, USA). Cdx2 was identified using monoclonal antibody, clone AMT28, diluted 1:30 (NCL-CDX2; Leica Microsystems, Milton Keynes, UK).
Endogenous peroxidase activity was blocked using peroxidase-blocking solution (S2023; Dako, Glostrup, Denmark). Binding of the primary antibody was detected using the Dako REAL EnVision Detection System (K5007), according to manufacturer's recommendations. Washes between steps were performed in TBS/Tween wash buffer at pH 7.6. The sections were then lightly counterstained with Mayers haematoxylin (01560BBE; Surgipath Europe) to give pale blue nuclei, then they were dehydrated and cleared, and a coverslip was applied.
Interpretation
The pattern of staining for Cdx2 was assessed and each section was scored semiquantitatively using a method described by Morris et al.10 The intensity of staining was scored as 0, 1(+), 2(++), 3(+++) or 4(++++), and the percentage of the total cells with each intensity was categorised as 1 (<25%), 2 (25–50%), 3 (50–75%) or 4 (>75%). The intensity and percentage scores were then multiplied to provide a semiquantitative score for the slide (minimum of zero and maximum of 16). If any sections showed variation in staining intensity between different areas, then these areas were scored individually and summed up to give an overall score for the slide, for example: 25% at 1+ and 25% at 2+ gave a score of (1×1)+(1×2)=3.
Biopsies from normal colonic mucosa were used as a positive control, producing a semiquantitative score of 4×4=16. IHC scoring for each case was performed by two independent examiners (SA and SJH). Any cases with a discrepancy in scoring were reviewed on a double-headed microscope and a consensus score was achieved for each case.
Statistical analysis
Data were analysed using the StatsDirect software package (StatsDirect, Altrincham, UK). A Kruskal–Wallis test was used to assess differences between the diagnostic categories of metaplasia, HGD and adenocarcinoma. The usual 5% threshold was used to determine statistical significance.
Results
IHC for Cdx2 demonstrated high-quality nuclear localisation within epithelial cells on examination of the positive control sections; this was reliable and easy to interpret. There was some variation in scoring between the two observers. These cases were re-assessed and a consensus score was achieved. No significant staining of the stroma was identified.
Figure 1A–D illustrates IHC staining for Cdx2 for squamous mucosa, BO, HGD and adenocarcinoma. The quantitative semiquantative scoring results for BO, HGD and adenocarcinoma are illustrated by a spread plot (figure 2) and a box and whisker plot (figure 3), demonstrating the median score, together with the lower and upper quartiles and the interquartile range.
Oesophageal sections and immunohistochemistry for Cdx2 protein expression (×40), using monoclonal antibody clone AMT28: (A) squamous mucosa, (B) Barrett oesophagus, (C) high-grade dysplasia, (D) adenocarcinoma.
Spread plot of individual section semiquantitative staining scores for Barrett oesophagus (metaplasia), high-grade dysplasia (HGD) and adenocarcinoma.
Box and whisker plot, demonstrating the median, lower and upper interquartile range (box) and whiskers to the highest and lowest values, for normal oesophagus, Barrett oesophagus (metaplasia), high-grade dysplasia (HGD) and adenocarcinoma.
Some cases showed a heterogeneous pattern of staining, with variation of staining intensity noted across the section, and this was accounted by the method of assessment used.
All 20 cases of oesophageal squamous epithelium showed no expression for Cdx2, equating to scores of zero. Cdx2 staining in BO, HGD and adenocarcinoma demonstrated mean scores of 10.95, 5.15 and 2.65, respectively. The range of scores for BO, HGD and adenocarcinoma were 4–16, 0–12 and 0–12, respectively. With the exception of normal oesophagus, all diagnostic categories therefore showed a wide range of scores. However, there is clear evidence of upregulation of Cdx2 in BO, followed by progressively decreasing Cdx2 expression through HGD to adenocarcinoma (p<0.0001, as shown by the Kruskal–Wallis test).
Discussion
Histopathological assessment for dysplasia is a problematic and specialised discipline and there is potential for diagnostic error, due to interobserver and intraobserver variability. For this reason, prior to inclusion in this study, all H&E-stained sections were reviewed by a histopathologist with a gastrointestinal interest (SJH), in order to confirm correct classification. Further efforts to avoid error were made by limiting examination of dysplastic cases to HGD only, as reported levels of observer variability are higher for a diagnosis of low-grade dysplasia, in comparison to HGD, with the latter representing a more robust diagnostic category.11 12 With regard to the assessment of metaplasia, we limited our examination of BO cases to those showing intestinal metaplasia. The reasoning for this is because the balance of evidence suggests that patients with specialised glandular epithelium, characterised by acid-mucin-producing goblet cells, are predominantly at increased risk of developing adenocarinoma.13 However, further study should be performed to quantitatively analyse expression of Cdx2 in non-goblet cell columnar metaplasia of the oesophagus.
Our previous experience of using IHC for Cdx2 in the oesophagus has demonstrated a heterogeneous pattern of staining, with regard to the intensity of staining seen. Such variation has been acknowledged by Phillips et al,5 and we consider there to be limitations to previous studies, where the staining pattern for Cdx2 was recorded purely as either positive or negative.2 5 7 We therefore believe that the semiquantitative methods, used by ourselves and others,14 15 more accurately assesses Cdx2 protein expression within tissue sections.
Previous studies have investigated Cdx2 expression through the metaplasia–dysplasia–adenocarcinoma sequence in the oesophagus, using the techniques of reverse transcriptase-PCR and IHC, for detection of mRNA and protein respectively.2 5 7 14 15 Those studies reported no expression of Cdx2 in oesophageal squamous epithelium, but they demonstrated strong upregulation in intestinal metaplasia. However, they produced conflicting results in relation to expression in dysplasia and adenocarcinoma. Some studies have indicated that Cdx2 remains upregulated in dysplasia and adenocarcinoma,2 7 14 while others have indicated expression to be downregulated.5 15 The cause of such variation is uncertain, but it may possibly be related to experimental design.
Our study confirms an absence of Cdx2 expression in normal oesophageal squamous epithelium, and we also show that it is expressed by intestinal metaplasia. Our findings are at variance with those demonstrated by Villanacci et al,15 who detected decreasing expression of Cdx2 progressively from BO to HGD, but expression seen in adenocarcinoma was noted to be intermediate between that seen in low-grade dysplasia and HGD.
Cdx2 has previously been proposed to be a tumour suppressor gene in the stomach and colon.16–20Our study, demonstrating decreasing expression of Cdx2 in HGD through to adenocarcinoma, could indicate that it has a similar function in the oesophagus. An alternative and possibly more plausible explanation would be for Cdx2 to represent a differentiation marker that tends to be downregulated with decreasing differentiation.
In summary, our study has confirmed Cdx2 to be strongly expressed by oesophageal intestinal metaplasia. In addition, we have demonstrated for what we believe is the first time, and with clear statistical significance (p<0.0001), upregulation followed by linear downregulation of Cdx2 expression through the oesophageal metaplasia–dysplasia–adenocarcinoma sequence.
Take-home messages
The development of goblet cells containing columnar metaplasia in the oesophagus is regulated by the homeobox gene Cdx2.
Linear downregulation of Cdx2 has been shown to occur through the goblet cell metaplasia–dysplasia–adenocarcinoma sequence.
This downregulation is in keeping with Cdx2 having a role as a differentiation marker.
Acknowledgments
Mr Andy Vail, Senior Lecturer in Biostatistics at the University of Manchester, advised on the statistical analysis for this report.
References
Footnotes
Competing interests None to declare.
Ethics approval This study was conducted with the approval of the Northern Manchester Local Research Ethics Committee.
Provenance and peer review Not commissioned; externally peer reviewed.