The histologic lesion in esophagitis is mucosal erosions or ulcerations varying in number, depth, and configuration. There is moderate infiltration of the stroma with inflammatory cells. The surface of the lesion may be covered with granulation tissue or fibrin. Chronic erosive esophagitis is characterized by loss of superficial epithelial layers (mucosal erosions) (Fig. 24-2C). Squamous metaplasia of the submucosal glands may occur in this disorder.

Cells desquamating from the eroded esophageal epithelium are derived from the deep layers and are of the same size as parabasal squamous cells, occurring singly and in small clusters. The cytoplasm is evenly distributed and the cells do not vary much in size. In clusters, there is good adherence of the cells to each other. The most outstanding features of the relatively large, somewhat hyperchromatic nuclei of these cells, sometimes show isolated clumps of chromatin and, occasionally, large nucleoli that may lead to an erroneous diagnosis of cancer (Fig. 24-2D). Careful attention to cellular detail helps in the correct interpretation of the cytologic findings. The cells seen in erosive esophagitis
may be similar to those observed in “repair” of squamous epithelium or in pemphigus (see Fig. 21-10).

A recently recognized entity, eosinophilic esophagitis, has been observed mainly in children but also in adults (Liacouras, 2003; Dahms, 2004). Clinically, the disease causes painful reflux esophagitis that mimics gastroesophageal reflux disease (GERD) but does not respond to antiacid treatment. Dense eosinophilic infiltrate of the esophageal wall has been observed in biopsies (Straumann et al, 2004). There are no reported cases of this entity diagnosed on cytologic sampling.

Many years ago, Berg (1955) emphasized the occurrence of this disease in cancer patients who, in the course of treatment, sustained a surgical or radiation injury to the esophagus. Immunosuppressive therapy, cytotoxic anticancer drugs, and acquired immunodeficiency syndrome (AIDS) have contributed to an increased frequency of this disease (Lightdale et al, 1977). Most patients have vague complaints referable to the esophagus, such as retrosternal pain or mild dysphagia. Herpetic esophagitis produces extensive, although superficial, ulcerations of the esophageal mucosa. In biopsies, intranuclear eosinophilic inclusions are observed within the epithelial cells. The use of cytology resulted in primary diagnosis in several cases. Surprisingly, in some patients, there was no past history of immune deficiency or immunosuppression. It is, therefore, possible that the disease is more common than hitherto anticipated. We have observed apparent simultaneous involvement of the esophagus and the bronchial tree.

The disease may be diagnosed in material obtained from the esophagus by washings, or brushing. The cytologic findings
are identical with those observed in the material from the female genital tract (see Fig. 10-29), and the respiratory tract (see Chap. 19 and Fig. 19-36). Multinucleated cells with molded “ground-glass,” opaque nuclei and cells with intranuclear eosinophilic inclusions are observed (Fig. 24-3A,B).

The first results of the population survey were presented in the Fourth International Cancer Congress in Florence, Italy in 1974 by an anonymous group representing the Chinese Academy of Medical Sciences. A cytologic survey of 17,471 persons over 30 years of age was conducted in the Henan Province in northern China. “Dysplasia” of the esophageal epithelium was observed in 276 patients, mostly below the age of 40, whereas invasive carcinoma in this population usually occurred in patients older than 40. Follow-up study of the patients with “dysplasia,” some over a period of 7 to 10 years, disclosed that 30.3% of them developed esophageal carcinoma, in 27.3% the original lesion persisted unchanged, and in 42.4% the changes either regressed to mild dysplasia or reverted to normal. In histologic studies of 67 patients, the progression of dysplasia of various types to carcinoma in situ could be observed in many specimens. It was the conclusion of this study that “marked dysplasia” must be considered a precancerous lesion. During the intervening years and changing political conditions in China, the names of the investigators became known (summary in Shu 1984, 1985), and the results of several surveys became available.

As related by Shu in 1984, there is no doubt that mass screening for esophageal carcinoma in high-risk areas of-China had a major beneficial effect. Before screening was instituted, the diagnosis of carcinoma in situ or early invasive carcinoma was 2 per 1,000 in low-risk areas and 10 per 1,000 in high-risk areas. Screening of 81,187 asymptomatic people over the age of 30 in the high-risk Henan Province resulted in the discovery of 880 esophageal cancers (a huge prevalence rate of 1%!), of which 649 (73.7%) were early and treatable by surgery (Shu, 1984). Less is known about survival of these patients, but Dr. Shu assured me that most of the treated patients survived 5 years or longer with a good quality of life. This information must be compared with a survival of about 10% to 25% of patients with invasive squamous cancer of the esophagus commonly observed in the Western world (Ide et al, 1994; Lieberman et al, 1995). Kwong et al (2004) reported that, among Chinese patients with esophagus cancer, those showing a gain of the short arm of chromosome 12 (+p12) in the tumor had poor outcome after surgical treatment, regardless of stage of disease.

The accomplishments of the Chinese scholars found several imitators. Thus, Berry et al (1981) attempted a similar project in South Africa (where the rate of esophageal cancer is very high among some black populations), resulting in the discovery of 15 occult invasive carcinomas and carcinomas in situ in 500 patients studied. Dysplasia was illustrated, but the clinical significance of the lesion was not discussed. Similar results were reported by Jaskiewicz et al (1987) from a high-risk rural population in Transkei (South Africa); in
five patients, dysplastic changes progressed to invasive cancer.

To our knowledge, the Chinese experience has not been duplicated in the Western countries, except for the work in this laboratory. Greenebaum et al (1984) studied 96 high-risk Montefiore Hospital patients in New York City by the balloon technique. The selected patients had prior cancers of the larynx or pharynx, or were alcoholics and heavy cigarette smokers. Greenebaum unexpectedly found three occult recurrent oropharyngeal cancers and one carcinoma in situ of the esophagus, observed in a man with prior history of squamous carcinoma of the larynx. The biopsy of the esophagus disclosed fragments of squamous cancer in the absence of radiologic abnormalities.

Based on cytologic and histologic criteria, the Chinese investigators divided the precancerous lesions into two groups: dysplasia and carcinoma in situ. The criteria were derived from the classification of precancerous lesions of the uterine cervix (see Chap. 11). Lesions with more orderly epithelial growth, surface differentiation, and relatively minor nuclear abnormalities were classified as dysplasia and lesions with more significant atypia were classed as carcinoma in situ.

The dysplasias were further subdivided into mild, moderate, and severe, based mainly on cytologic criteria (see below). The true significance of dysplasia is not clear. Although, in some patients, the lesions either failed to progress or regress, there is no doubt that, in a substantial number of untreated patients, invasive cancer of the esophagus was subsequently observed (Shu 1984, 1985). The same conclusion was reached by Sugimachi et al (1995), who considered “dysplasia” as an early carcinoma of the esophagus. In any event, the insecure behavior of the precancerous lesions of the esophagus is remarkably similar to lesions of the uterine cervix (see Chap. 11).

The squamous cancer cells derived from high grade lesions (high grade dysplasia, carcinoma in situ) are of the parabasal variety. The nuclear abnormalities consist of enlargement and hyperchromasia; the cytoplasm is scanty, resulting in a high nucleocytoplasmic ratio (Fig. 24-9). Cell clustering is common. Shu (1984, 1985) illustrated several examples of progression of dysplasias to carcinoma in situ and, in some cases, to invasive carcinomas over a period of 2 to 4 years.

There are very few cases of carcinoma in situ diagnosed by cytology in the Western world. Besides the case reported by Imbriglia and Lopusniak (1949), one case of a lesion approaching carcinoma in situ was personally observed in esophageal washings in a 59-year-old man with an esophageal diverticulum and symptoms of obstruction, leading to the clinical diagnosis of esophageal cancer. The lesion was characterized by typical dyskaryotic (dysplastic) superficial and parabasal squamous cells. The biopsies, which, unfortunately, were obtained after an initial short course of radiotherapy, localized the lesion to the diverticulum, which was subsequently successfully resected. The histologic appearance of the epithelium disclosed nuclear abnormalities and some degree of disarrangement of the component cells. One carcinoma in situ was recognized during balloon screening of a high risk population by Greenebaum et al (1984), and another during analysis of esophageal cytologic abnormalities by a neural net-based scanning system (Koss et al, 1998). In both cases, the smears contained squamous cancer cells, singly and in clusters, that could not be differentiated from an invasive squamous cancer (Fig. 24-10). The last patient who had a history of esophageal stricture and necrotizing esophagitis, was alive without evidence of esophageal cancer 3 years after the diagnosis. These anecdotal cases confirm the insecure prognosis of precursor lesions of esophageal squamous carcinoma.

Low-grade lesions of the esophagus (mild or moderate dysplasia; Fig. 24-11) are characterized by well-differentiated superficial and intermediate squamous cells with marked nuclear enlargement and hyperchromasia. In some patients, koilocytes may be observed. The resemblance of these cell abnormalities to dyskaryosis (dysplasia) of cervical squamous cells is remarkable (see Chap. 11).

In brush specimens, the cancer cells derived from adenocarcinomas are usually well-differentiated columnar cells with conspicuous nuclear abnormalities in the form of enlarged, sometimes hyperchromatic nuclei with large single or multiple nucleoli (Fig. 24-12A,C). In advanced carcinomas, these cells occur singly and in small clusters, usually accompanied by evidence of necrosis. Smaller, more spherical cancer cells may also occur in less well-differentiated tumors. Shurbaji and Erozan (1991) noted that the findings were not consistent and that nuclear hyperchromasia and large nucleoli were not observed in all the cases.

In mucoepidermoid carcinomas, the cells of glandular derivation are mixed with occasional squamous cancer cells. In many such cases, it is not clear whether the tumor is primarily squamous or glandular, although the malignant nature of the process is evident.

Cytologic definition of “dysplasia” of columnar epithelium is insecure. Wang et al (1992) described small clusters
of mildly pleomorphic cells with enlarged nuclei with occasional multiple nucleoli. In a more recent thoughtful review of this topic by Hughes and Cohen (1998), the definition of cytologic abnormalities in dysplasia was not particularly helpful. These authors reviewed the literature and pointed out that the accumulated experience with comparative cytology-histology of these lesions is very small.

Personal experience suggests that the morphologic distinction of “high-grade or severe dysplasia” from adenocarcinoma is highly subjective. In cases with clusters of mildly or moderately atypical columnar cells with somewhat enlarged, somewhat hyperchromatic nuclei and the presence of multiple small or single large nucleoli, the diagnosis of “atypia” of columnar cells is justified (Fig. 12-13A,B). If the atypical cells form clusters, the possibility of an adenocarcinoma cannot be ruled out (Fig. 12-13C,D). In all such cases, the tentative cytologic diagnosis must be confirmed by biopsies. The situation is reminiscent of the problems with recognition of early neoplastic abnormalities in endocervical epithelium, discussed at length in Chapter 12.

There is virtually no cytologic experience with the identification of “mild dysplasia.” Hughes and Cohen (1998) point out that the separation of the possible neoplastic from reactive changes in the columnar cells is virtually impossible.

Monitoring of precursors of adenocarcinoma by balloon cytology was attempted with diverse results. Fennerty et al (1995) failed to obtain informative samples in a small number of patients. Falk et al (1997), in a much larger series of patients, compared the inexpensive balloon cytology with brushings. Brush cytology recognized all of 11 patients with high-grade dysplasia and carcinoma whereas balloon samples failed to recognize two of these patients. For low-grade dysplasia, both brushings and balloon cytology failed in a substantial proportion of cases. In two of the patients without disease, atypical cells were observed. This study documented the problem with cytologic diagnosis of low-grade abnormalities.

The monitoring of epithelial DNA content by flow cytometry disclosed abnormal, aneuploid DNA histograms in dysplasias and in carcinomas (Haggitt et al, 1988). However, securing adequate cell samples for this procedure by esophageal brushings is difficult. Another approach to progression of dysplasia to adenocarcinoma in tissue samples was discussed by Polkowski et al (1995). These authors reported an
increase in the expression of mutated p53 gene and an increase in the proliferative fraction of cells as measured by expression of Ki67 antigen with progression of esophageal dysplasia to adenocarcinoma. To our knowledge, this approach has not been tested in cytologic samples.

Uncommonly, squamous carcinomas of the esophagus (but also of the larynx, pharynx, and rarely in the vagina) develop into polypoid bulky tumors, the surface of which is formed by a well-differentiated epidermoid carcinoma, sometimes in situ. The bulk of the tumor is composed of elongated spindly cells, occasionally accompanied by bizarre giant cells (Enrile et al, 1973). Experience with these lesions suggests that the spindle and giant cell components represent a peculiar metaplasia of squamous carcinoma and not a benign “pseudosarcoma,” as was originally suggested by Stout and Lattes (1957). Still, the latter term is often used to describe this lesion. It must be stressed that, in spite of its ominous appearance, the polypoid carcinoma appears to offer a much better prognosis than ordinary esophageal carcinoma.

The cytologic findings in brushings from one of the rare polypoid carcinomas of the esophagus were described by Selvaggi (1992). The smears contained cells of squamous carcinoma and spindly cancer cells, and a few multinucleated giant cells, corresponding exactly to the histologic make-up of the tumor. We observed a similar tumor in the vagina. The smears were composed of malignant squamous cells only (see Chap. 17).