Which histologic methods are best for demonstrating architectural and cytologic details of dysplasia in Barrett's esophagus?
K.R. Geisinger (Winston-Salem)
In order to prevent or reduce deaths from esophageal adenocarcinoma, the disease must be detected in its earliest stages, preferably before it has invaded into the organ's wall [1,2]. This preinvasive, neoplastic stage is glandular dysplasia which has been defined on the basis of specific histomorphologic features. In patients with a columnar-lined esophagus, the morphologic diagnosis of such dysplasia relies on the identification of prominent (i.e., neoplastic) alterations both in the intercellular arrangements or architecture of the epithelial elements and in the constituent cells themselves, especially their nuclei. The histologic preparation of esophageal biopsy specimens requires a series of technical steps, including tissue fixation, specimen orientation, embedding in paraffin, sectioning of paraffin blocks, mounting sections on glass slides and staining. Although most of these steps may be performed by any of several acceptable technical variations, each is important in the quality of the final product for interpretation.
This cascade of technical procedures for the preparation of biopsies has been addressed excellently in a succinct yet eloquent chapter by Haggitt and Rubin . These authors promote, what can be considered, an ideal approach in this setting. First, mucosal biopsies, they argue, should be procured utilizing jumbo forceps. They also claim that properly oriented biopsies are essential to the recognition of dysplasia
and recommend that orientation occur immediately after procurement and prior to fixation. Plastic mesh is their preference for the substrate upon which the tissue specimen is placed, Haggitt and Rubin also firmly support the use of Hollande's solution as the best fixative . For embedding, tissue fragments should be placed so that in the next step, sectioning, their long axes are cut through. They frown upon embedding more than two fragments in the same paraffin block. Blocks should be trimmed of excessive paraffin before sectioning so that more sections may be placed on a slide. Then, specimens are step-serial-sectioned a minimum of three levels. Finally, their routine stain consists of a mixture of hematoxylin and eosin, alcian blue and saffron .
Unfortunately, even in the cubby holes of academia, we do not live in an ideal world. Several practical problems exist with the above histologic schema. For one, in our experience, it is unusual to find gastroenterologists who use jumbo forceps for esophageal biopsies. The smaller specimens obtained with conventional forceps are much more difficult to orient prior to fixation. It is probably true that Hollande's solution consistently provides crisper nuclear detail than does formalin (Figs. 1-8). However, the former fixative is both more expensive and more difficult for histotechnologists to work with. The cost to my laboratory for one gallon of Hollande's solution is $55, while it is only $3 for the same volume of formalin; that is, it is more than 18 times as expensive. From a technical viewpoint, there are several considerations. Hollande's yellow color needs to be washed out before processing. Picric acid is considered by some to be a dangerous chemical and thus has been banned in its powdered form in several states in the USA. It should be
Figure I. .This squamous mucosa is from a patient with Barrett's esophagus and is fixed in formalin. The chromatin is pale, often with a homogeneous or washed out appearance. A few nuclei appear to have "holes" or pseudoinclusions. Nucleoli are inconspicuous to absent (xlOO).
Figure 2. .This is from the same patient and obtained at the same endoscopic session as the tissue in Fig. 1, but it is fixed in Hollande's solution. Distinct chromatin granularity is evident in many nuclei, as are small nucleoli (xl00).
mentioned that most published studies which have defined and characterized Barrett's dysplasia have included formalin-fixed tissues. In their thought provoking study of
Figure 3. .Barrett's metaplasia with intestinal metaplasia is present in this formalin-fixed biopsy. Most of the epithelial nuclei are vesicular with small nucleoli. Basal polarity is well maintained (Xl00).
Figure 4. .These glandular elements were from the same patient as in Fig. 3, but were fixed in Hollande's solution. Nuclear membranes and chromatin granularity are more distinct (xl00).
biopsies from patients with Barrett's esophagus, Reid et al.  stated that specific fixatives did not affect diagnostic reproducibility among pathologists. Step-serial-sections are worthwhile in a search for small focal lesions, but I wonder whether this
Figure 5. .This benign Barrett's metaplasia was adjacent to an erosion. Nuclei are enlarged, but do not appear hyperchromatic. In many cells, the nucleoplasm has a foamy or bubbly appearance. In a few, distinct vacuoles are evident (x400).
Figure 6. .This is from the same patient as in Fig. 5, but is fixed in Hollande's solution, rather than formalin. The finely granular chromatin and nuclear membranes are more distinct (x400).
applies to the hunt for dysplasia . In my laboratory, step-serial-sections take a technologist an extra 5-10 min per block, compared to simple leveling. The cost of
Figure 7. .Dysplastic glandular nuclei in Barrett's esophagus are obviously enlarged, variably elongated and hyperchromatic. In formalin-fixed tissue, the chromatin appears more or less homogeneous and indistinct, as its granularity is not readily apparent. Nucleoli are not easily seen (x400).
Figure 8. .This is dysplastic glandular tissue fixed in Hollande's solution. Compared to formalin fixation, nuclear membranes are more distinct and chromatin is crisper in appearance. Nucleoli are also more apparent (x400).
special stains needs to be mentioned. For example, an eight ounce bottle of saffron is $72.
I attempted to discover which processing steps are actually performed for esophageal biopsies in the laboratories of experts in this field. I mailed a simple checklist survey to fifty members of the Gastrointestinal Pathology Society in the United States and Canada. At the time of this writing, I had received responses from 42 (84%). Overwhelmingly, formalin is the fixative of choice for esophageal biopsies, being used 80% of the time; one institution uses formalin for adults and Bouin's for pediatric patients. Bouin's and Hollande's fixatives are used by four and three pathologists, respectively. Fifteen responders (36%) claim that biopsies are oriented, most often by a gastrointestinal endoscopy technician (53%), followed by the gastroenterologist (23%). The most commonly used substrate is a filter (60%). With respect to sectioning of esophageal biopsies, the most common procedure followed is step-serial-sectioning at three levels (43%), followed by three simple levels (17%). Most rely on hematoxylin and eosin as their routine stain (83%) with the remainder adding a mucin reaction which always included alcian blue. Finally, 33% of the responding gastrointestinal pathologists stated that their institution had an established formal surveillance program for patients with Barrett's esophagus. Although there was considerable overlap between programs with and without such a procedure, definite trends exist for the more complex or elaborate histologic procedures to be employed by those who are involved in surveillance.
Endoscopic screening programs with multiple biopsies throughout the length of the involved metaplastic segment have proven to be effective in detecting patients
with glandular dysplasia or early carcinoma [1,2,6,7]. However, the sensitivity of surveillance is not as high as desired. It also appears that many if not most patients with Barrett's esophagus will never progress to severe dysplasia or invasive tumor. Furthermore, it should be noted that the actual cost per patient can be considerable . Consequently, this has prompted the search for ancillary or additional assays that can be performed on biopsy specimens. One of the most widely examined is the detection of DNA ploidy abnormalities by flow cytometry [8,9]. The Seattle team uses a protocol that quantitates the DNA by flow cytometry of a biopsy from every 2 cm of the involved length of the esophagus [5,8]. They and others have reported aneuploidy occurring not only in most examples of adenocarcinoma and dysplasia, but also in metaplastic mucosa without histologic evidence of neoplasia; both aneuploidy and/or elevated proliferative activity may select a subgroup of patients with a greater risk for cancer [8,10-13]. However, despite these advances, others argue that flow cytometry in this setting is not the end-all answer . For one thing, cost needs to be considered. In one institution that I am familiar with, the total technical and professional fees for a single flow cytometric study of DNA alone is $168.00 and $240.00 for fresh and frozen tissue, respectively. For a more in-depth exam of this situation, let us hypothesize a patient with a 6-cm segment of glandular metaplasia without any gross abnormality endoscopically. This would entail at least three biopsy specimens (several every 2 cm) with flow cytometry on three samples. The total (minimal) cost for histology and flow cytometry would be $746.00; this includes a combined technical and professional charge of $242 for the routine production and evaluation of histologic sections.
The potential use of brushing cytology of the esophagus in surveillance must be considered [15,16]. As has been stated elsewhere in this volume, there are several valuable reasons for a cytologic exam to be incorporated into (or even replace) a screening program, as described above. Most importantly, I believe that cyto-morphologic criteria for glandular dysplasia and carcinoma are reproducible, as witnessed by several publications [15-17]. In addition, ploidy measurements by image (static) cytometry, which has several advantages and disadvantages vis-a-vis flow cytometry, can be performed directly on Feulgen-stained cytologic smears. As brushing samples a much larger surface area of mucosa than do multiple biopsies, only a single ploidy analysis may be adequate . In the same institution mentioned above, the total cost for the evaluation of both conventional esophageal cytology and DNA image analysis is $338.00. Consideration of image rather than flow cytometry on tissue biopsies is also valid .
To summarize, an overwhelming consensus as to the single best approach to the diagnosis of Barrett's dysplasia does not yet exist. I suspect that considerable time will elapse before gastroenterologists, medical economists, and pathologists agree on this. It must be admitted that everything we look at under the microscope is an "artifact" and part of this problem is that we grow accustomed to and comfortable with the artifacts generated by our own laboratories.
8. Reid BJ, Haggitt RC, Rubin CE, Rabinovitch PS. Barrett's esophagus. Correlation between flow cytometry and histology in detection of patients at risk for adenocarcinoma. Gastroenterology 1987;93:l-l1.
10. Reid BJ, Blount PL, Rubin CE, Levine DS, Haggitt RC, Rabinovitch PS. Flow cytometric and histological progression to malignancy in Barrett's esophagus: prospective endoscopic surveillance of a cohort. Gastroenterology 1992; 102:1212-1219.
13. Robaszkiewicz M, Volant A, Hardy E et al. Demonstration of clonal heterogeneity in adenocarcinomas in Barrett's esophagus by flow cytometric study of cellular DNA content Gastroenterol Clin Biol 1992;16:540-546.