Introduction
The epithelial boundaries of the anorectum are of high clinical relevance, as the exact location of pathological processes, especially carcinomas, in relation to these boundaries is crucial for prognosis and therapy [
1‐
3]. The different epithelial zones of the human anorectum have caused numerous discussions concerning their embryonic origin, levels of transition and nomenclature [
4]. Whereas it seems to be clear that the upper part of the anorectum belongs to the large intestine and is derived from the entoderm, the lower part is supposed to originate from the ectoderm. In 1914, Johnson [
5] postulated that an exact histological determination of the border between entoderm and ectoderm is impossible. Only 10 years ago, O`Rahilly et Müller [
6] pointed out that, as a result of developmental dislocations, the position of the dentate line does not necessarily allow conclusions concerning developmental processes in the anorectum. Nevertheless, the border between ectodermal- and entodermal-derived anorectal epithelia is still assumed to lie in the area of the dentate line [
7‐
9]. Recently we have revealed the fact that differentiation of the anorectal layers starts after disappearance of the cloacal membrane at the beginning of the seventh week post conceptionem (p.c.). Indications for an interaction between epithelium and mesenchymal structures were shown, whereas an ingrowth of surface ectoderm was not observed. The epithelial differentiation was described as a local process, probably induced by a cross-talk between epithelium and mesenchyme [
10].
Up to now, specific markers for the very early embryonic ectodermal and entodermal epithelia are missing. The solution of this question may be a subject of cell biological and experimental studies. From the clinical point of view, however, it is of much more relevance to know which kind of tumour may derive from the different heights of the anorectum. Thus, to get new insights into the developmental differentiation of the anorectal epithelia and probable similarities between the embryonic situation and tumour development in the adult, we used different cytokeratins (CKs) and CDX2 as markers for immunohistochemistry. The expression of cytokeratins is cell-type specific. During malignant transformation their specificity remains conserved in large parts [
11‐
13]. Furthermore, the expression of cytokeratins is specific for different types of epithelial differentiation [
14‐
16]. CK 7 and CK 18/20 are commonly described as markers for simple and columnar epithelium and thus believed to belong to entodermal-derived tissues, whereas CK 14 is a marker for squamous epithelium and thus considered to belong to ectodermal-derived tissue [
17]. They have already been applied in several investigations concerning intestinal epithelia of the adult. In our study these cytokeratins were first examined in regard to their ability to demarcate different epithelial sections within the developing anorectum. The results were compared with the normal situation in the adult and with cases of both low rectal and anal carcinoma.
CDX2 belongs to the family of parahox genes and encodes an intestine-specific transcription factor which is important for the regulation of proliferation and differentiation of intestinal epithelial cells [
18]. It is expressed in the nuclei of entodermal-derived epithelial cells from duodenum to rectum. Previous studies showed CDX2 expression in normal colonic mucosa and in most colorectal adenocarcinomas and metastases, but it is rarely seen in tumours of other sites [
19‐
21]. In the present study, we examined CDX2 expression in the human anorectal epithelia during development, in the adult and in carcinomas of both the low rectal and anal region.
It was the purpose of this study to create an actual basis for cellular origin and differentiation of the anorectal region including carcinomas. The hypothesis of the underlying study is that the prenatal situation indicates the origin of the different types of carcinomas in the adult anorectum.
Materials and methods
Differentiation and spatiotemporal distribution of the different epithelial sections of the anorectum were examined by means of conventional histology and immunohistochemistry using monoclonal antibodies directed against cytokeratins (CK) 18, 20, 7 and 14 as well as CDX2.
Thirty-three human embryos and fetuses (8th–33rd week p.c.), sections of a 2-year-old child and of four adults (36 years old male, 65 years old male, 76 years old male and 74 years old female) were studied microscopically.
The embryonic and fetal specimens were provided by the collection of the Department of Anatomy, Histology and Embryology, Innsbruck, or obtained from legal abortions collected from different gynecologists having received parental consent. The adult specimens were also taken from the Department's collection and had no intestinal disorders or malformations.
Specimens were fixed in 7% formaldehyde in PBS (0.2 M). After paraffin embedding in a routine histological infiltration processor (Miles Scientific Inc., Naperville, IL, USA), specimens were cut in 4-μm sections in sagittal or axial planes by means of a Microm ERGO Star Rotations-microtome (Microm, Walldorf, Germany). The sections were dried overnight, dewaxed with xylene, rehydrated in graded alcohol series, and every 10th to 20th section of a series was stained with haematoxylin and eosin (HE).
For immunohistochemical study the Discovery XT automated staining system by Ventana Medical Systems Inc. (Ventana, Strasbourg, France) was used. After dewaxing and rehydrating, the antigen retrieval was started by heat (for CK 18, 20 and 14 staining) or protease (for CK 7 staining) induced epitope unmasking while the sections were immersed in citrate buffer. The slides were incubated with prediluted monoclonal antibodies recognizing CK 7, 14, 18, 20 (Linaris, Germany) for 60 minutes and biotinylated secondary antibody (universal, IGG antimouse, antirabbit) (Ventana, Strasbourg, France) for 30 minutes. Subsequently, the antigen–antibody complex was fixed with 0.05 % glutaraldehyde in 0.9% sodium chloride. Antibody binding sites were visualised by means of a 3.3-diaminobenzidine (DAB) detection kit (according to the manufacturer's protocols).
For CDX2 staining, we used Histoprime® Cdx2, a monoclonal mouse antibody, ready-to-use clone Cdx2-88 (Linaris, Wertheim-Bettingen, Germany). Subsequent to primary antibody incubation for half an hour and secondary antibody incubation for 15 minutes, DAB detection kit was applied according to the manufacturer's guidelines.
All sections were counterstained with haematoxylin and blueing reagent. Dehydration in ethanol and xylene series was carried out, and finally the sections were mounted permanently by a cytosol embedding medium (Microm, Walldorf, Germany). Negative controls without primary antibody did not show antibody-induced labelling.
In a second part of the study, tumour blocks of 17 patients (mean age 74 years, range 57–86 years) with ultralow rectal cancer (pT2, N0, M0, GII–pT3, N1, M0, GIII) following abdominoperineal resection or anterior resection of the rectum without preoperative radiochemotherapy and resection specimens of four patients (mean age 73, range 62–82 years) with anal squamous cell carcinomas with preoperative radiochemotherapy were processed accordingly. The samples were obtained from the Institute of Pathology, Innsbruck.
Discussion
The present study was undertaken to examine the development and differentiation of the anorectal epithelia and to compare the results with the histomorphological characteristics of anorectal tumours. We report that:
a.
Due to the cytokeratin profile and to the expression of CDX2, the different zones of the anorectum can be recognized early in human prenatal life;
b.
Epithelial differentiation runs in a craniocaudal direction;
c.
The ARZ is an area of epithelial changes; and
d.
We explain why CK7, CK20 and CDX2 can simultaneously be expressed in rectal adenocarcinomas.
Following the detailed investigation of Johnson [
5], our study is the first to contribute to the histogenesis of the anorectal epithelia. Moreover, we compare the situation in human prenatal stages with adult specimens as well as with ultralow rectal and anal carcinoma. In the last decade, some morphological studies concerning the early development of the human anorectum were published that—in connection with the fate of the cloaca and the cloacal membrane—mainly contributed to the debate of the entodermal or ectodermal origin of the different epithelia [
22‐
24]. All these publications have the same weakness: they are only based on normal HE histology. Thus, the results are purely descriptive, mostly speculative and therefore, in regard to their contents, without any connection to new molecular biological findings in laboratory animals or to clinical pathological aspects. To be able to gain new aspects as to the heterogeneity of the different anorectal epithelia, we are the first to include immunohistochemistry of different cytokeratins and of CDX2 expression in all human prenatal stages.
Based on Fenger's [
4] clearly arranged subdivision of the anorectum into five epithelial zones, the results of our study can be summarized as follows:
As everywhere in the large intestine, the rectal mucosa has transitional folds in early fetal life [
25]. Starting in the eighth week, the epithelium of the RZ is positive for CK 18 and CDX2 during all following stages of human development, and for CK 20 from the tenth week onwards. Rectal CK 7 positivity is obvious only in the ARZ where CK 20 and CDX2 staining vanishes. It is the most important finding that the ARZ already exists in prenatal life. This is why the ARZ must be considered as an area of epithelial changes and not as a metaplastic one. It differs from the RZ as to CK 7 expression.
The epithelium of the ATZ mostly consists of stratified cuboidal epithelium in prenatal stages [
24] and is characterized by positive labelling for CK 7. Anal valves are covered with squamous epithelium and thus show CK 14-positive reaction.
According to our results and to former studies [
5,
26], the rectal columns differentiate in the late embryonic and early fetal period and arise from the entoderm [
10,
22]. The height of the columns is covered with non-keratinizing stratified squamous epithelium only in the postnatal stages as was already shown by Hollinshead [
27]. Both anal stratified epithelia are positive for CK18 and CK7 in early stages before gaining CK14 positivity from the ninth week onwards.
Our results clearly show that cell differentiation in the anorectal region does occur craniocaudally, regardless of where the cells take their origin from, and this is also true for the differentiation of the anal squamous epithelium. This backs up the results of our last study where we supposed that “an ingrowth of surface ectoderm is not evident … and that the epithelial differentiation is a locally organized process and induced by the interaction with the mesenchyme …” [
10]. Contrary to another study [
28] whose authors only elaborate on the introversive transfer of the epithelium and on that basis speculate that the epithelium is growing inwards from the outside, we are able to prove beyond doubt that the differentiation starts inside and proceeds externally.
When comparing the results of the human fetuses and adults with the cases of low rectal and anal tumours, we expected to find a cytokeratin profile of the carcinoma cells that is similar to its native epithelium, i.e. we expected CK 18/CK 20- and CDX2-positive labelling in the low rectal carcinomas and CK 14 positivity in tumours deriving from the squamous epithelium of the anal canal below the DL. The rectal tumour probes we included in the present study showed the expected labelling for CK 18/20 and CDX2. In three cases we found an additional positivity for CK 7 which—in contrast to former studies postulating that the presence of CK 7 can be used to exclude a rectal origin of tumours and metastases—is not astonishing as the tumours are characterized as ultralow rectal ones that originate in the ARZ where positive labelling for CK 7 normally occurs. These important findings are in accordance with a recent study [
29] which revealed that CK 7 can be expressed in rectal adenocarcinomas, and that the combination of CK 20/7 with CDX2 is an excellent one securely to identify metastatic rectal carcinomas. This might be helpful in discriminating rare anal malignant conditions like adenocarcinoma of the anal canal from the more common squamous cell carcinoma on the one hand and ultralow adenocarcinoma of the rectum on the other hand. Adenocarcinoma of the anal canal is rare and there is often diagnostic doubt between true anal canal cancer, most likely arising from the anal glands that develop from the rectal epithelium [
10] and direct invasion from very low rectal cancers [
30]. Changes in the expression of certain CKs provide an indication of the differentiation stage (dedifferentiation) of a tumour which is comparable to prenatal stages of differentiation and the respective expression pattern [
31].
Our study successfully aimed to show that the demarcation line between the five epithelial zones is more distinct in some places than in others: whereas it is rather transient between the RZ and the ARZ, on the DL it is clearly delimited in all stages of development. Below the DL the demarcation between the keratinizing stratified squamous epithelium of the anal canal and the non-keratinizing stratified squamous epithelium of the rectum is clearly visible from the 16th week of development. According to our qualitative results, the extent of the different anorectal sections, especially of the ATZ, is variable and differs interindividually [
4]. Therefore, in the case of a tumour, it is most important to document the exact location macroscopically as well as microscopically.
CDX2 discontinues to be expressed at the borderline between the RZ and the ATZ, i.e. in the ARZ. Cranial to the ARZ, transitional folds—typical for the intestinal identity—appear in early fetal life. Thus, up to this line, the intestine developmental programme seems to take effect that—according to the findings of Gao et al. [
32]—is regulated by Cdx2 in mice. These authors showed that the intestinal growth is severely affected in Cdx2 mutant mice, and that Cdx2 deficiency prevents colon and rectal formation and leads to complete obstruction. New results in mice point to the fact that Cdx2 opposes tumour cell spreading [
33]. We have shown that CDX2-positive cells are found in the bottom of healthy human rectal crypts in prenatal stages and in the adult where according to literature they play an important role in the regulation and differentiation of regular intestinal cells [
18,
34]. According to our results, we suppose that the typical rectal adenocarcinoma arises from the RZ and may arise from the ARZ, if scattered and regular crypts are found there. However, it does not arise from the ATZ. Our study reveals the possibility to demarcate and assign epithelial zones within the anorectum by means of immunohistochemistry. We could not determine any difference in CK expression among the different tumour stages. However, studies including a larger number of patients and different stages of rectal cancer as well as carcinomas of the anal squamous epithelium are required to analyze a possible changing in the CK and/or CDX2 expression pattern.