Are there any histologic features of high-grade dysplasia which make this diagnosis equivocal?
M.D. Diebold (Reims)
The diagnosis of high-grade dysplasia in Barrett's mucosa is based upon the combination of architectural and cytologic criteria. Potential diagnostic problems concern each end of the spectrum of high-grade dysplasic lesions. At one end of the spectrum the diagnosis depends on the assessment of cytologic abnormalities and especially of nuclear alterations.
The signification of cytologic prominent abnormalities associated with a heavy inflammatory infiltrate may be questionable. The lack of true stratification, the normal or decreased nuclear-cytoplasmic ratio associated with cytoplasmic clarification and polymorphs within the epithelium, are in favor of inflammation related lesions. Nevertheless, dysplastic changes and inflammation lesions can coexist. Of the utmost importance for the diagnosis of high-grade dysplasia are polymorphism and hyperchromatism of the nuclei and loss of nuclear polarity. At the other end of the lesion spectrum is the distinction between high-grade dysplasia and infiltrating adenocarcinoma. The cytologic criteria are indiscriminative and the differential diagnosis is based upon the integrity of basement membrane in high-grade dysplasia and invasion in infiltrating adenocarcinoma.
In Barrett's mucosa, misleading features are realized by pseudo-invasion of lamina propria, muscularis mucosae and submucosa. Among them, crowding and infolding of dysplastic glands must be differentiated from cribriform or back-to-back pattern of adenocarcinoma. The bottoms of tortuous dysplastic glands may appear on some sections as isolated nests of atypical infiltrating cells. Another pattern of architectural distortion in Barrett's mucosa is represented by dysplastic gland entrapment in hyperplastic and disorganized fibers of the muscularis mucosae ascending upwards into the lamina propria. Glandular cysts, of unknown signification, are frequent in
Barrett's mucosa; they may be lined partially or totally by a dysplastic epithelium and should not be mistaken for an invasive adenocarcinoma.
The pathologist should be familiar with this range of changes. These potential pitfalls also emphasize the necessity of a rigorous histopathological examination.
The best conditions for diagnosis include numerous and large biopsy samples (eventually obtained by aspiration-section technique) and adequate orientation before embedding.
Serial-step sectioning allows better understanding of the architectural distortion of glands, unmasking isolated atypical cells, if any, originating from an underlying or neighboring adenocarcinoma.
L. Le Bodic (Nantes)
Dysplasia refers to various cellular and architectural defects of the intraepithelial topography resulting from abnormal proliferative activity. These defects may be considered as the usual, but not ineluctable, phase in transition from a normal to a cancerous cell. Thus, two types of problems are involved: the diagnosis and exact grading of dysplasia as well as the therapeutic strategy to be employed.
Dysplastic alterations occur in cells (differentiation defects with secretion loss or abnormalities; anisocytosis and anisokaryosis; atypical mitoses) and tissues (architectural disorganization). None of these defects are specific, and their association and severity can vary in the same biopsy specimen.
A distinction should be made between two esophageal forms which raise different problems: dysplasia of the esophageal mucosa and Barrett's esophagus.
Dysplasia of the esophageal mucosa
It has been investigated in two types of studies: the Asiatic approach is based on cytological data without endoscopy, and the Western approach on biopsies obtained during endoscopy. The correlations between cytological and histopathological classifications have not been documented. Mukada et al.  proposed three grades for this dysplasia: slight, moderate and severe. The severe form is characterized by defects involving the entire mucous membrane. However, microinvasive carcinoma is indicated when the chorion is involved.
The diagnosis of severe dysplasia in esophageal mucosa may be uncertain. When there is intense inflammation with ulceration, the presence of tumescent and atypical
endothelial cells in neoformative capillaries within regenerative granulation tissue may prove misleading. Cytological abnormalities related to viral infections (herpes, cytomegalovirus) are sometimes quite marked, wrongly suggesting dysplasia. There is also the possibility of pseudoepitheliomatous epithelial hyperplasia with dyskerato-sis in the case of Abrikosoff's tumor. Finally, the poor quality of a shredded specimen can make interpretation difficult, especially if the lesions are not uniformly distributed.
In Barrett's esophagus, or glandular mucosa of the endobrachyesophagus (EBE), the prevalence of dysplasia is estimated to be 10% in the absence and 80% in the presence of associated cancer [2-4]. In Riddell's classification, which is generally used, moderate or low-grade dysplasia (LCD) is characterized by basophilic cells with large, rounded hyperchromatic nuclei; pseudostratification features, with nuclei occupying only the lower half of cells which appear to be compressed; and reduced cell secretion. In addition to these LGD abnormalities, severe or high-grade dysplasia (HGD) is characterized by an increased nucleocytoplasmic ratio as well as nuclear defects, with pseudostratification of nuclei along the entire cell layer, mucosecretion defects and architectural abnormalities (disorganized, proliferative and ramified crypts, with occasional luminal epithelial vegetation), but without involvement of the basement membrane and the chorion. Although it is not absolutely certain that HGD will always progress to cancer, it has been demonstrated that HGD is frequently present within the year preceding cancer diagnosis [3,5].
In 11 series totaling 438 patients, dysplasia was present in 10% of cases, with HGD occurring in only 2% of simple EBE without visible carcinoma, whereas the mean incidence in 16 published series of EBE complicated by carcinoma was 78% (59-100%), 79% of which were HGD . The problem is one of reproducibility and thus of the reliability of the diagnosis reached by the pathologist. In a retrospective study based on 37 biopsies examined by four pathologists , total diagnostic concordance was found for 20 (54%), 16 (80%) of which concerned HGD. When these 37 biopsies were reviewed simultaneously by the interpreters, diagnostic concordance was reached for 34 (92%), and two discordances concerned LGD and HGD.
Improvement in diagnostic concordance depends on a good definition of histo-pathological criteria. It is likely that concordance is lower for pathologists not often required to interpret dysplasias, particularly in an inflammatory context. The persistence of gastroesophageal reflux seems to play a role in the worsening of dysplastic lesions , although therapeutic control of this reflux does not eliminate the risk of dysplasia and carcinoma.
In cases of Barrett's esophagus, personal factors can condition interpretation of dysplastic lesions, especially for the differentiation of nondysplastic from dysplastic (or slightly dysplastic) mucosa, particularly in an inflammatory context. We are not concerned here with techniques currently under evaluation which may prove useful
for the diagnosis of dysplasia or for histopathological prognosis (flow cytometry, mucus alterations, nucleolar organizers, etc.). Results have been disappointing for some of these techniques and are still preliminary for others.
It can therefore be concluded that it is essential to have a precise definition of the criteria of dysplasia in order to provide for reinterpretation by an experienced pathologist and examination of new biopsy specimens in case of doubt.