Diagnostic Cytology of the Gastrointestinal Tract
Helen H. Wang
Gamze Ayata
Introduction
The popularity of gastrointestinal (GI) cytology for the diagnosis of infection and malignancy has waxed and waned during the past few decades. The ability to distinguish between high-grade dysplasia or carcinoma in situ and invasive carcinoma in biopsy specimens and the more prevalent expertise of surgical pathology cause some to consider cytology an unnecessary duplication of GI mucosal biopsies.1,2 However, the combined use of endoscopy, ultrasound guidance, and fine-needle aspiration (FNA) has expanded the horizons of GI cytology.3
Specimen Types
Types of GI tract specimens commonly received in the cytology laboratory include those obtained by endoscopic brushings and ultrasound-guided endoscopic FNA. Endoscopic FNA has enabled endoscopists to reach farther than they can with biopsy forceps to sample mural and extramural lesions, including lesions adjacent to the GI tract. The nonendoscopic specimens obtained with balloon- or mesh-type samplers have been evaluated in the research setting to ascertain their usefulness in the surveillance of populations at high risk for esophageal carcinoma.4–6
Specimen Preparations
Direct smears can be made from materials collected on the endoscopic brush, in the needle, or on the balloon and mesh samplers; these can then be either fixed immediately in 95% ethanol and stained with the Papanicolaou method or left to air-dry and stained with Diff-Quik (Dade-Behring, Inc., Deerfield, IL) or Wright-Giemsa stain. Alternatively, the material can be rinsed into a medium such as CytoLyt, CytoRich, or 50% ethanol for liquid-based preparations. The specimen can then be processed by a concentration method, such as ThinPrep Processor (Hologic, Marlborough, MA) or Cytospin (ThermoFisher Scientific, Waltham, MA),7,8 to make slides that are then stained with the Papanicolaou method. According to a College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology, ThinPrep preparations performed better than non-ThinPrep preparations.7 However, liquid-based preparations, including ThinPrep, involve altered morphology and artifacts that require adjustment by cytopathologists, such as cleaner background with altered or reduced background and extracellular elements, architectural changes (smaller cell clusters and sheets and more three-dimensional clusters), altered cell distribution (more dyshesion—dissociation of cells at the periphery of cell clusters or as single cells), and changes in cytologic morphology (enhanced nuclear features and smaller cell size).9 Residual material from liquid-based preparations lends itself to cell block making with thromboplastin-plasma cell block technique,10 the Cellient Automated Cell Block System (Hologic), or other techniques for ancillary studies.
Value and Accuracy of Specimens
Cytology specimens have some advantages over specimens obtained by endoscopic biopsy. The brush can sample a wider area, and the fine needle can reach deeper lesions than can be reached by biopsy forceps. Also, both the brush and the fine needle are less invasive than biopsy forceps and less likely to cause bleeding. In addition, cytology has a shorter turnaround time than histology. Direct smears can be ready for review within minutes with no compromise of the quality of the preparation (unlike frozen sections of biopsy specimens, which compromise the quality of the final or permanent preparation). However, as mentioned, cytology is limited in its ability to distinguish between high-grade dysplasia or carcinoma in situ and invasive carcinoma.
Despite the potential duplication of cytology and biopsy, the literature has consistently shown that the highest diagnostic yield is obtained with the combined use of these specimens.11–13 The yield of cytology is significantly higher when the brushing is performed before rather than after the biopsy.14
Normal Morphology
Esophagus
Intermediate-type squamous cells with abundant cytoplasm and vesicular nuclei are seen in the normal esophagus (Fig. 3.1). Superficial-type squamous cells with abundant cytoplasm and small pyknotic nuclei can also be seen in small numbers. Single cells and clusters of ciliated columnar cells from the respiratory tract with no clinical significance may be seen rarely.
Stomach
Gastric surface foveolar cells can shed as single cells or in sheets. When in sheets, the columnar cells exhibit abundant cytoplasm, regularly spaced nuclei, and open chromatin arranged in a honeycomb or palisaded pattern (Fig. 3.2), depending on the orientation. When they are shed as single cells, they often lose their cytoplasm to become naked nuclei. In endoscopic FNA specimens, the sheets of foveolar cells can mimic cells from a mucinous neoplasm, and the single naked nuclei, because of their small, monomorphic appearance, can mimic cells from a pancreatic endocrine tumor.
Small Intestine
The lining cells of the small intestine can be easily distinguished from gastric foveolar cells by the presence of goblet cells. On low magnification, the specimen typically has a Swiss cheese appearance, with the “holes” representing either goblet cells or gland openings of the crypts (Fig. 3.3). On high magnification, the absorptive cells have either finely granular or vacuolated cytoplasm, and the goblet cells have single large mucin vacuoles and crescent-shaped nuclei with rounded contours. The striated border of the absorptive cells may be seen at the periphery of the sheets.
Large Intestine
Normal epithelium is characterized on cytology by sheets or strips of tall columnar cells with abundant cytoplasm and basal nuclei. Partial or complete openings of the colonic crypts may be seen (Fig. 3.4).
Infections
Most infectious agents that affect human hosts can infect the GI tract of immunocompetent and immunocompromised patients.15 Some infectious agents have a predilection for the GI tract. The more common ones are discussed in this section.
Candida
Candida almost exclusively involves the esophageal portion of the GI tract. Brushings are more sensitive than biopsy specimens in the detection of esophageal candidiasis.13 Contamination by oral Candida is usually not a problem because the brush is contained within a sheath when it is passed into and out of the endoscope and is expelled from the sheath only to sample the lesion. The organisms appear as pink to purple pseudohyphae and yeast formations on Papanicolaou stain (Fig. 3.5). Reactive squamous cells and inflammatory cells are often observed in the background.
Herpes Simplex Virus
Herpes simplex virus infection can theoretically affect epithelial cells anywhere along the GI tract, but it is most commonly seen in the esophagus. Multinucleation, nuclear molding, ground-glass chromatin, and eosinophilic intranuclear inclusions are the characteristic features of infected cells (Fig. 3.6).
Cytomegalovirus
Cytomegalovirus infection affects epithelial, stromal, and endothelial cells along the GI tract and is characterized by large cells with a single large basophilic intranuclear inclusion with a perinuclear halo (Fig. 3.7). Intracytoplasmic textured inclusions can occasionally be seen in the affected cells.
Helicobacter pylori
Helicobacter pylori infection occurs exclusively in the stomach and is perhaps the most common infection of the GI tract. These organisms can be demonstrated on imprint smears of gastric biopsies or on brush cytology specimens.16,17 Examination of imprint and brushing cytology specimens is comparable, if not superior, in sensitivity (88%) and specificity (61%) to histologic examination of sections stained with hematoxylin and eosin (H&E) and modified Giemsa stain.16,17 The benefits of imprint and brushing cytology are rapid results, high specificity, and low cost. However, the efficacy of cytologic detection depends on the extent of colonization by these organisms. When present in large quantity, they are evident even at low magnification, but they can be difficult to identify when present in small numbers. On Papanicolaou stain, H. pylori organisms appear as faintly basophilic, S-shaped rods admixed with mucus in the vicinity of glandular cell clusters (Fig. 3.8). Special stains, such as a triple stain combining silver, H&E, and Alcian blue at pH 2.5, can enhance their detection by cytology.18
Giardia
Giardia affects the duodenum of both immunocompetent and immunocompromised hosts. Brush cytology is a useful method for detecting Giardia because the organisms are on the luminal surfaces of the intestinal epithelial cells. They are flat, gray, pear-shaped, and binucleate, with four pairs of flagella (Fig. 3.9).19 Giardia lamblia trophozoites have been found to be immunoreactive for the protooncogene KIT (C-kit, CD117), which may help to identify the organisms.20
Atypical Mycobacteria
Atypical mycobacteria accumulate within macrophages in the lamina propria, and very rigorous brushing is required for the infected macrophages to be included in the cytology sample. The presence of isolated foamy histiocytes on the smear should raise the level of suspicion of an atypical mycobacterial infection (Fig. 3.10). In general, the organisms are present in large numbers. On Diff-Quik–stained smears, the mycobacteria form numerous rod-shaped negative images, either within the histiocytes or in the background (Fig. 3.11).21 Special stains for acid-fast bacilli are necessary to confirm the diagnosis.
Cryptosporidia
Cryptosporidia can involve any glandular epithelium of the GI tract in patients infected with the human immunodeficiency virus (HIV) and can be detected by examination of stool and cytology specimens.22 Cryptosporidia are 2- to 5-µm, round, basophilic bodies on the luminal surfaces of the epithelial cells. Therefore, they are seen only when the plane of focus is shifted to the surfaces of the cells where the organisms reside (Fig. 3.12). When in doubt, confirmatory Gomori methenamine-silver stain can be applied.
Microsporidia
Microsporidia can be detected on cytologic specimens such as stool, nasal secretions, duodenal aspirates, and bile, as well as on brushing specimens from the duodenum and biliary tract.23–25 On Papanicolaou stain, they appear in aggregates as brightly eosinophilic, rod-shaped or ovoid organisms measuring 1 to 3 µm in diameter (Fig. 3.13). They are present in both epithelial cells and inflammatory cells. In epithelial cells, they are located in the supranuclear portion of the cytoplasm and therefore, like cryptosporidia, are seen at a slightly different plane of focus from that of the epithelial nuclei.
Inflammatory, Reactive, and Metaplastic Changes
Nonspecific Changes
Any injury to the mucosa can evoke a nonspecific inflammatory or reactive epithelial change. When the injury is sufficient to result in ulceration, the change (i.e., the epithelial repair) can become so extreme that it may mimic a malignancy. It is often difficult to determine whether the reparative epithelium is of glandular or squamous origin. Although epithelial repair is characterized by prominent eosinophilic nucleoli, they are usually neither huge nor numerous (i.e., more than three or four) (Fig. 3.14). The appearance of atypical stromal cells or their stripped nuclei from granulation tissue can also be quite alarming (Fig. 3.15