Gastrointestinal Tract

Helen H. Wang and Gamze Ayata

The advent of improved sampling and visualization devices has cemented cytology as a valuable method for the detection of gastrointestinal (GI) malignancies, premalignant lesions, and infections. Direct visualization of the mucosa, coupled with imaging of submucosal structures, allows for the collection of mucosal brushings, submucosal fine-needle aspirates, and mucosal biopsies, all with a single procedure.¹ Indeed, GI cytology is increasingly seen as a rapid, efficient, and cost-effective means for the evaluation of a variety of GI tract lesions.¹

Clinical Indications

Clinical indications

• suspected malignancy

• screening for dysplasia (e.g., Barrett’s esophagus)

• suspected infection

A suspected malignancy is by far the most common and important indication for cytologic examination. With the aid of direct mucosal visualization, brush cytology is complementary to biopsy for detecting adenocarcinoma, dysplasia, and infections.¹ Cytologic sampling has the following advantages over biopsy.

Sampling a Wider Area and Reaching Deep Organs

In the case of a diffuse lesion, where multiple biopsies must often be taken to ensure adequate sampling, a single brushing is capable of sampling a wide area of interest. Brushings are particularly useful for the evaluation of suspected preinvasive neoplastic lesions and high-grade dysplasia, where only a poorly defined area of irregular, rough, or vaguely nodular mucosa is seen rather than an obvious gross lesion.2–4 For this reason, cytologic sampling is ideal for the surveillance of dysplasia in patients at increased risk of squamous or adenocarcinoma of the esophagus.^5–7 Cytology is also superior to biopsy for stenotic lesions, such as tumors of the gastric cardia or pylorus, which cannot be reached easily with biopsy forceps.¹ Brush cytology specimens are also more sensitive than biopsies in detecting fungal infections.⁸

Better Recognition of Lymphoid Cells

Small endoscopic biopsies are commonly distorted and squeezed. A properly prepared cytologic smear, on the other hand, yields well-preserved, isolated lymphoid cells that are easier to recognize and interpret than the crushed, distorted cells on small biopsies. Cytologic features alone are often sufficient to render a definitive diagnosis of a large cell lymphoma.1,9 Two types of lymphoma pose special problems in diagnosis: a low-grade lymphoma arising from mucosa-associated lymphoid tissue (MALT) and CD30-positive anaplastic large cell lymphoma.⁹ The former needs to be distinguished from a benign lymphoid infiltrate, and the latter mimics a large cell carcinoma because its cells can be cohesive, immunoreactive for epithelial membrane antigen (EMA), and negative for leukocyte common antigen (LCA). A high level of awareness and a panel of marker studies are usually needed in such cases before a definitive diagnosis can be made. Whereas brushings sometimes obtain limited material, endoscopic fine-needle aspiration (FNA) often yields sufficient material for immunohistochemistry, flow cytometry, and cytogenetic assessment of lymphoproliferative disorders.¹⁰

Less Invasiveness

Obtaining a brushing sample is less traumatic than taking a grasp biopsy. This is particularly relevant for patients with clotting abnormalities and for those with very vascular tumors. A diagnosis can be made on cytologic material in such patients while avoiding the risks associated with a biopsy.

Shorter Turnaround Time

Once smears have been made, they can either be air-dried and stained with a Romanowsky-type stain or immediately fixed in alcohol and stained with toluidine blue–eosin or a modified Papanicolaou stain.11,12 As a result, the slides are available for interpretation in a few minutes. Although a short turnaround time may not be required in most cases, it can be crucial when urgent decisions concerning patient management need to be made.

Sample Collection and Processing

Various instruments and techniques for collecting material have been developed—and abandoned—over the years. The most common today is the direct brushing of a visible lesion via a fiberscope. Lesions confined to the lamina propria (e.g., signet ring cell carcinoma) or the submucosa/muscularis (e.g., gastrointestinal stromal tumor [GIST], neuroendocrine tumor [NET], and lymphoma), however, can be difficult to sample by brush cytology. For such lesions, endoscopic FNA enables sampling of lesions hidden beneath necrotic debris or normal mucosa.13–18 Endoscopic FNA is further enhanced by ultrasound guidance (endosonography). This allows the needle to reach even deeper lesions, including those that are extrinsic to the GI tract but impinging on the lumen, like a metastatic malignancy.¹⁹ Endosonography is useful for identifying the extent of the lesion and detailing regional anatomy to permit an assessment of the safest and most appropriate site for FNA.¹⁹

Although not in routine use, balloon or encapsulated samplers can be used to screen for esophageal cancer. These have the advantages of sampling the entire esophagus at low cost.5–7,20–22 Three modes of collection have been designed for this purpose: esophageal balloon cytology, encapsulated sponge cytology, and encapsulated sponge-mesh cytology.^22–24 The esophageal balloon cytology sampler consists of a disposable, single-lumen plastic catheter attached to an expandable latex balloon. When inflated with 15 to 20 mL of air, it turns into a near-perfect sphere with a diameter of 35 mm. The other two samplers consist of a gelatin capsule attached to a flexible plastic stylet. The gelatin capsule contains a compressed polyurethane sponge in the sponge sampler, and a polyurethane sponge covered with a cotton mesh in the sponge-mesh sampler. The patient swallows the sampler; once the sampler reaches the stomach, it is inflated (as with the balloon sampler), or the sponge is exposed by waiting 10 minutes for the gelatin to dissolve. Epithelial cells are collected as the sampler is pulled out of the patient. Although all three samplers obtain satisfactory yields of squamous and glandular cells, it is easier for patients to accept the encapsulated samplers.

Sample Collection

To obtain a brushing specimen, a brush is enclosed inside a transparent Teflon sheath and passed through the endoscope. When a lesion is visualized, the brush is expelled from the sheath and firmly and briskly plunged into the mucosa 5 to 10 times. To ensure an adequate sample, the lamina propria should be penetrated. After the sample has been obtained, the brush is retracted into its Teflon sheath, removed from the patient, extruded from the sheath again, and rinsed in a preservative solution for liquid-based preparations or pressed against glass slides as direct smears.

The endoscopic FNA involves introducing the needle through the biopsy channel of a fiberoptic endoscope. The lesion is localized either by direct vision or by ultrasound. Once the needle is in the lesion, negative pressure is applied to the needle using a 10 or 20 mL syringe. With the negative pressure maintained, the needle is moved back and forth within the lesion. Finally, the pressure is released, the needle removed from the scope channel, and the material prepared as smears or liquid-based preparations for cytologic examination.

Processing the Sample

The needle or brush is rinsed in transport/preservative medium (buffered saline solution, 50% ethanol, CytoLyt, or CytoRich).1,25,26 Alternatively, the material on the brush, needle, or sampler is rolled or spread onto one or more clean slide(s). The slide is immersed immediately in 95% alcohol for fixation. The importance of immediate fixation cannot be overemphasized, because cells that are thinly spread out on a slide dry out very quickly. The rolling process is repeated until no more cellular material can be rolled onto a slide. Some slides can be left to air-dry for a Romanowsky-type stain, if preferred. Additional cellular material can still be salvaged from the brush by vortexing it in an appropriate transport or preservation medium.

Slides are prepared from the transport medium sample most commonly using automated cell concentration methods like ThinPrep or SurePath. The advantages of these methods include better cell preservation, a cleaner background, a reproducible preparation, and a thin layer of cells.26–28 These advantages lead to shorter screening time. Another important advantage is the availability of material for ancillary studies. After routine slides have been prepared, there is usually residual material for any needed ancillary studies (e.g., immunostains).²⁸ Disadvantages include the cost of the equipment and reagents and the labor involved in preparing slides. In addition, there is a transition period for obtaining expertise in the interpretation of liquid-based preparations.^26,29

Accuracy

The sensitivity and specificity of brushing cytology for the detection of malignancies of the GI tract vary according to the location and sampling method. The sensitivity of brushings for the diagnosis of GI adenocarcinomas ranges from 77% to 94%, and the specificity is generally greater than 95%.30–41 These figures are comparable to those for biopsies. The combined use of biopsy and cytology yields the highest detection rate (Table 7.1).^{8,33,36,41,42} The accuracy of brushing cytology is significantly higher when the brushing is performed before rather than after biopsy.^43,44

TABLE 7.1

DIAGNOSTIC SENSITIVITY OF ENDOSCOPIC BRUSHING AND BIOPSY IN DETECTING UPPER GASTROINTESTINAL TUMORS ^∗

NA, Data not available.

^∗Table modified with permission from: Wang HH, Jonasson JG, Ducatman BS. Brushing cytology of the upper gastrointestinal tract. Obsolete or not? Acta Cytol. 1991;35(2):195-8.

Due to sampling error and the presence of confounding “pseudogoblet” cells, cytology is less accurate in identifying goblet cells (for the diagnosis of Barrett’s esophagus [BE]), with a sensitivity of 41% and specificity of 84%.⁴ The use of cytology for detecting dysplasia in BE, however, is more promising. Brush cytology detects high-grade dysplasia with high sensitivity and specificity.^2,45–48 Nonendoscopic screening of the esophagus has a sensitivity of approximately 90% and a specificity of over 90% for squamous dysplasia or carcinoma and a sensitivity of 80% and specificity of 95% for high-grade glandular dysplasia or adenocarcinoma.^2,5,6,20,48 Prospective outcome studies of balloon cytology in a high-risk area in China showed that cytology results are highly predictive of the subsequent risk of esophageal and gastric cardiac cancer.^49–51

Some consider cytology an unnecessary duplication of the biopsy and argue that cytology has a higher false-positive rate (lower specificity), particularly in the setting of reactive or reparative conditions.32,38,52 Increased experience and strict adherence to criteria for diagnosing malignancy minimize the false-positive rate and maximize the diagnostic yield of the endoscopic procedure, with a predictive value of a positive result that approaches 100%. The predictive value of a negative result is more variable, depending both on the patient population and on the sampling technique of the endoscopist.

Review of Morphologic Findings

Features noted on low-power magnification

• cellularity

• cellular arrangements

• background

Examination of the slides under low magnification is tremendously important. Features to note are cellularity, cellular arrangements (flat sheets, three-dimensional clusters, and isolated cells), and background features (clean, inflammatory, or necrotic). Cells in reactive processes often exfoliate as flat cohesive sheets, whereas neoplastic cells (both benign and malignant) tend to aggregate in three-dimensional clusters. Cells from a malignant neoplasm are arranged as either tight or loose three-dimensional clusters or, due to the loss of cellular cohesion, as isolated cells. Whereas a clean background often indicates a benign process, a dirty, inflammatory background can indicate either a benign or a malignant process. Ulceration, whether due to inflammation or malignancy, is accompanied by many inflammatory cells and cellular debris. Inflammatory cells are usually numerous and predominate in reactive/reparative conditions. When many necrotic ghost cells are present without numerous inflammatory cells, a malignancy should be suspected.

Features noted on high-power magnification

• cytoplasmic characteristics

• nuclear details

High magnification is important for the evaluation of cellular and nuclear details. In general, a benign process is characterized by a uniformity of cellular arrangement, nuclear size and shape, and the number of nucleoli. A malignant neoplasm, on the other hand, is characterized by a haphazard cellular arrangement and marked irregularities in cell size and shape, nuclear chromatin pattern, and number of nucleoli. Lymphomas and endocrine tumors are exceptions to the rule, because they are often composed of very similar cells.

Special features with liquid-based preparations

• cells appear smaller

• fewer cells in sheets or groups; more single cells

• less obvious gland formation

• tumor diathesis appears as small clumps of amorphous, acellular debris

Esophagus

Infections

Esophageal infections occur most often (but not exclusively) in immunocompromised patients.53–57

Cytomorphology of Candida infection

• pseudohyphae and yeast forms

• reactive squamous cells

Fungal infection with Candida species is the most common esophageal infection (Fig. 7.1). Brushings are more sensitive than biopsies in the diagnosis of esophageal candidiasis.^1,8 The number of organisms varies from few to many. Contamination by oral Candida is not a problem because the brush is contained within a sheath, which protects it while in transit to and from the lesion.

Figure 7.1 Candida (esophageal brushings).
Infection by Candida species is recognized by identifying both the ovoid yeast forms and elongated pseudohyphal forms (ThinPrep, Papanicolaou stain).

Cytomorphology of cells infected with herpes simplex virus

• multinucleation

• nuclear molding

• ground-glass chromatin

• Cowdry A bodies

• Figure 7.2

Figure 7.2 Herpes simplex infection (esophageal brushings).
Multinucleation, molding of nuclei, and a ground-glass chromatin pattern are diagnostic features (ThinPrep, Papanicolaou stain).

Cytomorphology of cells infected with cytomegalovirus

• mononucleation

• large single basophilic intranuclear inclusion

• perinuclear halo

• intracytoplasmic textured inclusions

Differential diagnosis of viral infections

• epithelial repair

• carcinoma

The large intranuclear inclusions of many viral infections can be mistaken for the macronucleoli of repair or malignancy.58,59 Molecular biologic techniques (e.g., in situ hybridization, polymerase chain reaction) can be applied to confirm the presence of virus.^60,61 Both are equivalent to culture in their sensitivity in detecting cytomegalovirus (CMV), and both are more sensitive than morphology alone.⁶²

Cytomorphology of epithelial repair

• cohesive sheets with a flowing or streaming pattern, better seen on smears, usually absent on liquid-based preparations (Fig. 7.3A)

• uniform nuclei with:

• enlargement

• smooth and thin nuclear borders

• fine chromatin

• prominent nucleoli

• evident mitoses

• background inflammation

• atypical stromal cells

Epithelial Repair

Any injury to the mucosa, especially ulceration, evokes a cellular reaction known as epithelial repair. It is sometimes difficult to determine whether the reparative epithelium is of glandular or squamous origin.

Although epithelial repair is characterized by prominent nucleoli (Fig. 7.3B), they are usually not huge or numerous (more than three or four). The atypical stromal cells or their stripped nuclei from granulation tissue can be quite alarming. Although their nuclei are strikingly enlarged, they are not hyperchromatic; instead they have fine, homogeneous chromatin, a thin nuclear membrane, and a smooth nuclear border.

Figure 7.3 Epithelial repair.
A, A slight flowing or streaming pattern is noted. The cells have slightly enlarged nuclei with regular nuclear borders, finely dispersed chromatin, and one or a few prominent nucleoli (gastric brushings, ThinPrep, Papanicolaou stain). B, Nuclei are enlarged but round and regular with prominent nucleoli. Cells are somewhat crowded but very cohesive (esophageal brushings, ThinPrep, Papanicolaou stain).

Radiation-induced changes represent a special type of reactive change or repair. There is proportional cellular and nuclear enlargement, multinucleation, metachromatic cytoplasm, and vacuolization of both cytoplasm and nuclei (Fig. 7.4).

Figure 7.4 Radiation-induced changes (esophageal brushings).
Cellular and nuclear enlargement, multinucleation, and vacuolization of cytoplasm are characteristic (ThinPrep, Papanicolaou stain).

Figure 7.5 Barrett’s epithelium with goblet cells.
A single large cytoplasmic vacuole expands the apical portion of the cytoplasm and displaces the nucleus and shapes it into a crescent against the basal cell membrane (ThinPrep, Papanicolaou stain).

Differential diagnosis of repair

• dysplasia

• carcinoma

Both cellular arrangements and individual cell features are useful in this distinction. Cells with reactive or reparative changes are usually in flat sheets and there is little cell dyshesion. By contrast, neoplasms frequently show dyshesion by the “feathering” (dissociation of cells) at the periphery of cell clusters or by the dispersion of numerous isolated cells. Three-dimensionality is most often associated with neoplastic lesions. In addition, although reactive or reparative cell nuclei may be enlarged, they are often uniform within the same sheet and have a similar number of regular nucleoli. In contrast, neoplastic cells, especially malignant ones, show more variation in nuclear size and shape as well as more nuclear membrane irregularity and chromatin clumping and/or clearing.

Barrett’s Esophagus

BE is an acquired condition in which the normal stratified squamous epithelium of the distal esophagus is replaced by columnar epithelium.63,64 It develops as a complication in 8% to 12% of patients with chronic gastroesophageal reflux.^64,65 Estimates of the prevalence of BE range from 262 to 376 per 100,000.^66,67 Follow-up studies of patients with BE show that their risk of developing esophageal adenocarcinoma is 0.12% to 0.29% per year, approximately 11 times the rate in the general population.^68–70 Risk factors include male gender, long segment of BE, smoking, alcohol use, and dysplasia. The columnar epithelium that replaces the squamous epithelium in the distal esophagus can be of cardiac, fundic, or intestinal type.⁷¹ Although the increased risk of adenocarcinoma was once thought to be associated only with intestinal-type epithelium,^72–74 this may be an oversimplification.⁷⁵

Cytomorphology of Barrett’s esophagus

• epithelial repair

• goblet cells

When obtained by brushing the esophagus, goblet cells, characterized cytologically by a single large cytoplasmic vacuole displacing the nucleus and shaping it into a crescent against the cell membrane (Fig. 7.5), are indicative of BE. Multiple goblet cells impart a Swiss cheese appearance to a honeycomb sheet of glandular cells. On liquid-based preparations, they are sometimes present as isolated cells. When columnar cells, with or without goblet cells, are seen on an esophageal brushing specimen obtained more than 3 cm from the gastroesophageal junction, it is reasonable to report the findings as “consistent with” or “suggestive of” BE.⁷⁶

Significant disparities are sometimes seen between the results of brushings and biopsies in the diagnosis of BE. By this measure, cytology is not entirely sensitive or specific for the diagnosis of BE.4,48 Much of the disparity can be accounted for by sampling. Another explanation may apply in some cases that are diagnostic of BE by cytology but negative by biopsy. Cells resembling goblet cells (“pseudogoblet” cells) have been described in the gastric cardia.^4,77 The vacuoles of pseudogoblet cells are pink (containing neutral mucin) with hematoxylin-eosin (H & E) staining, whereas the vacuoles in goblet cells are pale blue (acid mucin). The Papanicolaou stain, however, does not distinguish between acid and neutral mucin—both are dissolved in the staining process and appear as an empty space. Alcian blue stains theoretically might help with this distinction, but they are not commonly applied to cytologic preparations.⁴ Immunocytochemical staining of cytologic specimens for villin and hepatocyte antigen is also helpful for identifying goblet cells on cytologic specimens but is not routine in most laboratories.^78,79

Dysplasia in Barrett’s Esophagus

Adenocarcinomas in the setting of Barrett’s esophagus (BE) arise through a sequence of mucosal alterations that include premalignant (dysplastic) esophageal epithelium.⁸⁰ Dysplastic cells have some but not all of the features of malignant cells.^2,45,81,82

Cytomorphology of esophageal glandular dysplasia

• background of Barrett’s epithelium

• scattered atypical cells with some but not all features of adenocarcinoma

Preparations from an esophageal dysplasia contain fewer abnormal cells than those from an adenocarcinoma. In addition, dysplastic cells show incomplete features of malignancy (e.g., nuclear enlargement, hyperchromasia, increased nuclear-to-cytoplasmic ratio), but they generally show variation in nuclear size and shape, irregular cellular spacing, crowding, stratification, and dyshesion. The degree of these changes is proportional to the grade of dysplasia.

Although cytology is not perfect in detecting BE, it is useful in identifying dysplasia.4,45,48 Grading dysplasia is clinically important because patients with low-grade dysplasia are managed with follow-up and surveillance, whereas patients with high-grade dysplasia are managed with more frequent surveillance or resection.⁸³ To date, however, well-accepted criteria for diagnosing low-grade and high-grade dysplasia on cytology have not been established.⁸⁴ The diagnosis of low-grade dysplasia is still controversial even on histology.⁸⁵ Based on the features described below, one may suggest that the lesion is low or high grade.^4,48 In practice, when a low-grade dysplasia in BE is suspected, the lesion is reported as “atypical,” with a comment that the findings are suggestive of, or consistent with, low-grade dysplasia, depending on the degree of certainty. When a high-grade dysplasia is suspected, it is reported as “suspicious,” with a comment that the findings are suggestive of, or consistent with, high-grade dysplasia. The phrase “invasion cannot be excluded” is added if the cytologic findings include prominent dyshesion and many atypical cells.

Cytomorphology of low-grade dysplasia (Fig. 7.6)

• crowded groups with stratification

Figure 7.6 Low-grade dysplasia in Barrett’s epithelium.
A fragment of glandular epithelium with stratified elongated nuclei is seen. Although mucin depletion and slight nuclear enlargement are seen, significant nuclear atypia is absent (direct smear, Papanicolaou stain).

• mild nuclear atypia and pleomorphism

Cytomorphology of high-grade dysplasia (Fig. 7.7)

• crowded groups or isolated cells

Figure 7.7 High-grade dysplasia in Barrett’s epithelium.
A sheet of irregularly arranged cells with variably enlarged nuclei is present without evident dyshesion. In spite of the increased nuclear-to-cytoplasmic ratio, nuclear membrane irregularities, and slight hyperchromasia, the atypia is insufficient for a definitive diagnosis of malignancy (ThinPrep, Papanicolaou stain).

• higher degree of nuclear atypia and pleomorphism

Differential diagnosis of esophageal dysplasia

• epithelial repair

• adenocarcinoma

The diagnostic criteria for distinguishing dysplasia from reactive or reparative changes and adenocarcinoma are displayed in Table 7.2.

TABLE 7.2

COMPARISON OF CYTOLOGIC FEATURES OF REACTIVE, PREMALIGNANT, AND MALIGNANT GLANDULAR LESIONS OF UPPER GASTROINTESTINAL TRACT

Low-grade dysplasia is a strong predictor for the development of high-grade dysplasia and adenocarcinoma.86,87 Due to the difficulties in the morphologic diagnosis of low-grade dysplasia and the lack of any visible endoscopic abnormality, molecular alterations have been investigated for the purpose of risk stratification. In this regard, DNA content abnormalities (aneuploidy/tetraploidy) and overexpression of p53 are the strongest predictors of disease progression.^87,88 Indeed, endoscopic esophageal brushing specimens are suitable for DNA ploidy analysis, fluorescence in situ hybridization (FISH), and loss of heterozygosity studies,^89–91 but none of these has yet found routine application.

Adenocarcinoma of the Esophagus

The incidence of esophageal adenocarcinoma in the United States increased four-fold between 1973 and 2002, surpassing squamous cell carcinoma (SQC), which showed a 30% drop over the same period.⁹² Most adenocarcinomas are located in the mid- or distal third of the esophagus, presumably arising in Barrett’s esophagus (BE).^72,93–95

Cytomorphology of esophageal adenocarcinoma

• increased cellularity

• abnormal cellular arrangement

– numerous isolated cells

– “feathering” at the edges of cellular groups

– haphazard, crowded arrangement

– gland formation

• atypical nuclear features (Fig. 7.8)

Figure 7.8 Adenocarcinoma of the esophagus.
In contrast to Figure 7.7, the nuclei show more hyperchromasia, with chromatin clumping and clearing and large prominent nucleoli. Dyshesion was evident elsewhere on the slide (Papanicolaou stain).

– enlargement

– hyperchromasia

– uneven and irregular nuclear membrane

– pleomorphism

• variable amount of vacuolated cytoplasm

• tumor diathesis

• Barrett’s epithelium may or may not be present in the background

Differential diagnosis of esophageal adenocarcinoma

• epithelial repair

• dysplasia in Barrett’s epithelium, particularly high grade

• poorly differentiated squamous cell carcinoma

The distinction between reactive, dysplastic, and malignant esophageal lesions is challenging. Reactive or reparative cells sometimes appear more atypical than dysplastic cells, especially those from a low-grade lesion. Whereas low-grade dysplasia consists of crowded, stratified cells with elongated nuclei, reactive cells are in sheets with a streaming pattern. Nucleoli are inconspicuous in low-grade dysplasia, whereas prominent nucleoli are characteristic of repair. High-grade dysplasia has more malignant features than low-grade dysplasia (e.g., abnormal chromatin, irregular nuclear membranes, and dyshesion), which help to distinguish the dysplasias from repair. The difference between a high-grade dysplasia and carcinoma is quantitative rather than qualitative; a smear from an adenocarcinoma has more abnormal cells than one from a high-grade dysplasia, and the nuclear atypia in carcinoma is more marked. A tumor diathesis is a very useful feature for the diagnosis of a carcinoma, but it is not present in all cases. Endoscopic findings are also important. Abnormal cells from a fungating, ulcerating lesion most likely represent carcinoma, whereas those from an indistinct flat lesion most likely represent dysplasia. It is best to report a case as “high-grade dysplasia” when a small number of very atypical cells are present in a clean background from a patient undergoing surveillance for Barrett’s esophagus, particularly if endoscopy shows only granular mucosa and no definite mass. Ultimately, a biopsy is necessary for definitive diagnosis.

Squamous Cell Carcinoma of the Esophagus

Squamous cell carcinoma (SQC) is the most common malignancy of the esophagus worldwide.96,97 Isolated malignant cells are more commonly seen in SQCs, especially well-differentiated ones, than in adenocarcinomas. Cellular features depend on the degree of differentiation.

Cytomorphology of well-differentiated squamous cell carcinoma