Fig. 24.1
The FNA mobile cart is equipped with a double-headed microscope as well as a Diff–Quik setup to stain slides for on-site assessment
At a minimum, the endoscopist can expect the cytopathologist or cytotechnologist to provide an assessment of specimen adequacy, i.e., determining whether or not sufficient cellular material has been aspirated in order to obtain a diagnosis. However, oftentimes the pathologist will be able to accurately render a preliminary interpretation. Of note, preliminary interpretations can sometimes differ from the final diagnosis, and thus, it is crucial that clinical decisions and treatment are based on finalized reports whenever possible. Clear communication between the endoscopist and the pathologist is essential during EUS-FNA . The clinical history and imaging features of the targeted lesion represent invaluable information and this should be communicated to the pathologist, particularly at the time of preliminary assessment.
Preparation and Triaging of Cellular Material
Preparation and triaging cellular material are key elements in diagnostic EUS-FNA procedures. Various techniques can be utilized with the most common ones concisely summarized and illustrated below.
In our institutions, once the lesion has been aspirated, the FNA needle is passed to the cytopathologist or cytotechnologist, and using the stylet, the material is extruded onto a single, properly labeled glass slide (Fig. 24.2a and 24.2b). Attaching a syringe to the FNA needle can further help expel residual cellular material. Depending on the cellularity of the obtained sample, direct smears can then be prepared. The preparation of direct smears is somewhat of an art, and it takes experience to recognize how much material should and should not be placed on the slide as well as the degree of pressure that is needed to produce an evenly distributed smear. In our practice, we use two glass slides placed parallel to each other with enough contact pressure to break up and distribute the material evenly (Fig. 24.2c). With a quick gliding motion, pulling the slides in opposite directions, two slides are produced; one is left to air-dry (for the on-site rapid Romanowsky stain), while the other is immediately placed in an ethanol fixative (for later laboratory staining via the Papanicolaou method). Alternatively, the top slide can be quickly pulled apart from the bottom slide in an upward motion. In either technique, if performed correctly, the two resulting slides are mirror images of each other and should provide identical cellular material for the two complementary stains in the evaluation of the obtained lesional material.
Fig. 24.2
Slide preparation during FNA. a. A typical setup during FNA. At our institution, the cytotechnologist carries a container that holds all the material needed to make smears including slides, ethanol fixative, and media for ancillary studies such as RPMI into the procedure room. b. The needle is placed onto a slide, and a second slide is used to capture the spray of any splashed material. The material is then extruded onto the slide. c. The two slides are then placed parallel to each other, and enough contact pressure is used to disperse the material evenly between the two slides. The slides are then pulled in opposite directions to produce two slides. One slide is left to air-dry (for on-site assessment), while the other is placed in an ethanol fixative (for later processing)
As previously mentioned, the air-dried slide is then stained using a Romanowsky technique. Many utilize Diff–Quik, a commercially prepared Romanowsky stain. This particular staining method consists of only three solutions, allows for rapid staining on-site, and highlights cytoplasmic detail well; however, detailed nuclear morphology is not readily evaluable with this stain. In contrast, the strength of the Papanicolaou stain (Pap stain), which is more time-consuming and performed later in the laboratory with the ethanol-fixed slide, lies in its revelation of nuclear detail. While particular pathologists tend to prefer one or the other stain, both stains complement each other.
After rapid Diff–Quik staining of the air-dried slides, the pathologist examines the slides under the microscope. The major goal at this point is to ensure that adequate material has been obtained and, if necessary, to triage the material for ancillary studies such as flow cytometry or microbiology. Acquisition of sample for ancillary tests may require several additional FNA passes. Of note, Roswell Park Memorial Institute (RPMI) medium is a cell preservative devoid of formalin and the preferred medium for flow cytometry and cytogenetics, especially if the suspicion for a hematolymphoid malignancy is high.
Oftentimes, abundant cellular material and blood are expelled onto a glass slide. Immediate preparation of direct smears will yield thick smears that greatly hinder the preliminary microscopic examination and may even result in interpretative errors. Hence, in this scenario, using the “pick and smear” technique , a portion of the cellular material is separated and transferred onto additional glass slides to produce direct smears [5]. The remaining material is allowed to congeal, thus forming a clot/pseudotissue biopsy, which can easily be picked up with the tip of a small needle or end of another slide and subsequently transferred directly into formalin for later use as a cell block.
Alternatively, needle rinses or additional FNA passes can be expelled directly into formalin or other media such as RPMI or CytoLyt in order to create a cell block. Many cell block preparation techniques are in use, and they often involve mixing the cellular material with a gelling agent to create a mold that can be processed like a routine surgical biopsy specimen. The cell block has the advantage of providing material for subsequent immunohistochemistry which may be necessary to best classify a lesion.
In some practices, the pathologist may prefer to use a thin-layer cytologic technique such as ThinPrep. This technique does not allow for rapid on-site interpretation but does produce a concentrated preparation , which reduces the number of slides to be examined [6]. Many pathologists reserve this technique for cyst fluid interpretation in the setting of EUS-FNA.
Pathology Terminology
Pathology reports aim to state a diagnosis in concise and direct terms. For example, in the category of “positive for malignancy,” the pathologist will try to render the most specific subcategorization possible (such as “adenocarcinoma”). However, sometimes the pathologist may resort to terms such as “atypical” and “suspicious.” Unfortunately, the meaning that is conveyed by these terms is subject to interobserver variability. In our practice, the term “suspicious” is used when the specimen is quantitatively insufficient for a specific diagnosis. For example, an aspirate may consist of only one or two groups of cells with the features of malignancy. “Atypical” suggests that the cytologic features deviate from normal, but there are insufficient features to make a specific diagnosis. Therefore, “atypical” may refer to cellular changes seen in non-neoplastic or neoplastic lesions. Often, these indeterminate diagnoses are accompanied by an explanatory note. These comments convey essential information regarding the pathologist’s overall interpretation and should be read in their entirety; they should figure heavily when the clinician is considering the pathology results.
Cytologic Features of Normal Structures Sampled in EUS-FNA
As EUS-FNA traverses the luminal gut, the presence of contaminating (normal) structures such as esophageal, duodenal, and gastric epithelium is frequently seen in aspirate smears (Fig. 24.3). The approach of the aspirate, for example transgastric versus transduodenal, should thus be conveyed to the pathologist.
Fig. 24.3
Contaminating duodenal epithelium from EUS-FNA of pancreatic mass. Duodenal epithelium can appear as large sheets containing small, round nuclei. The pale staining cells interspersed throughout the sheet represent goblet cells, which are helpful in recognizing this epithelium as duodenal (Pap stain, 10x)
Cytopathologic Features of Commonly Sampled Entities
This section will give the endoscopist a brief overview of the most commonly sampled pathologic entities by EUS-FNA. A detailed description of rare non-neoplastic and neoplastic lesions and an expansion on ancillary studies are beyond the scope of this chapter.
Luminal Gut
The vast majority of cancers affecting the luminal gut are epithelial. Most of these epithelial tumors are sampled by forceps biopsy; however, some tumors may predominate beneath the mucosa and are better accessed by EUS-FNA.
Epithelial Malignancies
Adenocarcinomas are gland forming malignancies and account for the majority of tumors in the distal esophagus, stomach, and small and large intestines. Although there are some site-specific cytologic features, in general, these tumors look remarkably similar. In metastatic adenocarcinoma, immunohistochemistry can be helpful for localization of the primary [7]. Colorectal adenocarcinomas typically express CK20 and lack expression of CK7 [15]. Esophageal, gastric, and small intestinal adenocarcinomas often show expression of both CK20 and CK7 [8]. CDX2, a gene involved in intestinal differentiation, is expressed in the vast majority of luminal gut adenocarcinomas as well as a subset of pancreaticobiliary adenocarcinomas [9].
Cytologically, adenocarcinomas are characterized by three-dimensional aggregates and sheets of cohesive cells. The neoplastic cells demonstrate high nuclear-to-cytoplasmic ratios, nuclear hyperchromasia and irregularities, coarse chromatin pattern, and prominent nucleoli (Fig. 24.4). These nuclear features, while etiologically non-specific, indicate malignancy. Evidence of glandular differentiation such as cytoplasmic vacuolization, lumen formation , or mucin, when present, is very helpful in supporting a diagnosis of adenocarcinoma. Of note, colorectal carcinoma tends to have characteristic tall columnar cells and contains a “dirty” necrotic background (Fig. 24.5). These cytologic features may help in the setting of malignancy of unknown primary.
Fig. 24.4
Pancreatic adenocarcinoma. The general features of malignancy can be appreciated in this image. The cells demonstrate variability in cell size, nuclear hyperchromasia, and occasional nuclear membrane irregularities. Many of the cells demonstrate prominent nucleoli. Some of the tumor cells also demonstrate nuclear overlapping and loss of polarity which creates a “drunken honeycomb” appearance (Pap stain, 20x)
Fig. 24.5
Colonic adenocarcinoma. Aspirates contain tall, columnar, hyperchromatic cells. The nuclei often retain a basal orientation which imparts a “picket fence” appearance. Note the extensive, grungy, amorphous debris in the background (Diff–Quik, 10x)
Squamous cell carcinomas tend to arise in areas lined by squamous epithelium (esophagus and anus), but they can occur anywhere along the luminal gut. Aspirates demonstrate flat sheets of cohesive cells with polygonal shapes and dense cytoplasm (Fig. 24.6). The nuclear features are similar to those described for adenocarcinoma . Keratinization, which is best appreciated on the Papanicolaou stain as orangeophilic staining, is a defining feature of squamous cell carcinoma (Fig. 24.6). Immunohistochemistry (p63, p40, cytokeratin 5/6) can be used to confirm that a tumor has squamous differentiation; however, it does not provide information regarding site of origin [10].
Fig. 24.6
Squamous cell carcinoma. Aspirates of squamous cell carcinomas have flat sheets with polygonal cells. Nuclear pleomorphism is also demonstrated in this image. A hallmark finding is the single-cell component with prominent dense, orangeophilic cytoplasm (Pap stain, 10x)
Neuroendocrine carcinomas including carcinoid tumors can occur anywhere along the luminal gut [11, 12]. Aspirates are in general very cellular, composed of a loosely cohesive, monomorphic cell population [13]. Nuclei are round to oval, eccentrically placed, and exhibit regular nuclear contours and finely stippled, “salt-and-pepper” chromatin (best seen on a Pap-stained slide). As the cytologic features of neuroendocrine tumors can overlap with other entities, immunohistochemistry may be needed for confirmation. Neuroendocrine carcinomas express cytokeratins as well as specific neuroendocrine antigens such as synaptophysin and chromogranin [11, 12]
Mesenchymal Neoplasms
Most gastrointestinal mesenchymal tumors are subepithelial and best sampled by EUS-FNA . There are a host of mesenchymal tumors that can occur in the gastrointestinal tract. Location matters as leiomyomas are most commonly seen in the esophagus, whereas gastrointestinal stromal tumors (GISTs) are more common in the stomach and small intestine [14, 15]. A preliminary interpretation of “spindle cell neoplasm” is reserved for the majority of these cases. Procurement of additional material for cell block preparation is crucial as immunohistochemistry is essentially required for a more definitive diagnosis . In this section, we will limit our discussion to the most common mesenchymal tumor of the luminal gut, the gastrointestinal stromal tumor (GIST).
A GIST on aspirate is usually characterized by cellular fragments of spindle-shaped tumors cells (Fig. 24.7) [15, 16]. Some cases may feature round-to-oval tumor cells and these are designated as epithelioid GISTs. Perinuclear vacuoles are sometimes seen. Neoplastic nuclei typically have delicate chromatin and inconspicuous nucleoli. The cytoplasm may appear metachromatic on a rapid Romanowsky stain. Necrosis and mitoses are uncommon and suggest a more aggressive behavior.
Fig. 24.7
a Gastrointestinal stromal tumor. Large cellular sheets of tumor cells can be appreciated on low power (Diff–Quik, 2x). b Gastrointestinal stromal tumor. At higher power, the tumor is composed of spindled cells with poorly demarcated cell borders (Diff–Quik, 20x)
GISTs are often responsive to treatment with tyrosine kinase inhibitors such as imatinib and sunitinib [15]. Molecular testing is sometimes performed prior to treatment, and sufficient material to form a cell block should be obtained. Additionally, immunohistochemistry is essential to confirm and distinguish a GIST from other spindle cell lesions. Most GISTs demonstrate expression of CD117, DOG1, and CD34 by immunohistochemistry [14, 15]. In fact, the aspirate appearance of both leiomyoma and schwannoma can look almost identical to GISTs, and we always perform a small panel of immunohistochemistry (CD117, CD34, SMA, S100) to distinguish these tumors [14–16].
Lung and Mediastinum
EUS-FNA can be used to sample the lung, mediastinum , and even pleura. Epithelial malignancies are the most common lesions evaluated in the lung. General categories of primary malignant pulmonary epithelial neoplasms include adenocarcinoma , squamous cell carcinoma , large cell undifferentiated carcinoma, and tumors with neuroendocrine differentiation (typical and atypical carcinoid tumors, small cell carcinoma, and large cell neuroendocrine carcinoma). The cytologic features of pulmonary adenocarcinomas and squamous cell carcinomas do not substantially differ from their counterparts arising in other sites. However, pulmonary adenocarcinomas can be extremely well differentiated with very little atypia, making diagnosis potentially very difficult. A helpful feature in distinguishing a well-differentiated adenocarcinoma from reactive bronchial epithelial cells is the lack of cilia in the former and the presence of intranuclear pseudoinclusions in the latter [17]. At the time of a preliminary assessment, the term “non-small cell carcinoma” may be used; this term encompasses epithelial malignancies other than small cell carcinoma and other low-grade neuroendocrine carcinomas. A more specific final diagnosis is highly encouraged as targeted molecular therapies are now available. With the advent of molecular-based targeted therapies, testing for EGFR, KRAS, and ALK mutations is often requested by oncologists and obtaining adequate material for a cell block in order to perform these tests should be considered.
In the setting of a poorly differentiated malignancy, immunohistochemistry may be helpful to confirm the epithelial differentiation of the tumor and to further define the specific epithelial subtype. For example, TTF-1 and Napsin are markers for primary lung adenocarcinomas [18, 19]. These markers are typically negative in squamous cell carcinoma , although TTF-1 expression can be seen in small cell carcinoma [20].
It is important to recognize small cell carcinoma as treatment modalities differ compared to non-small cell carcinoma. The tumor cells of small cell carcinoma are approximately three times larger than a standard lymphocyte and feature fine chromatin with inconspicuous nucleoli (Fig. 24.8) [21, 22]. Nuclear molding, apoptotic bodies, frequent mitoses, and crush artifact are all typical cytologic features (Fig. 24.8) [21, 22]. Unlike other pulmonary epithelial tumors, small cell carcinoma demonstrates a dotlike immunohistochemical staining pattern with antibodies to cytokeratins [22]. These tumors also demonstrate expression of CD56, synaptophysin, and chromogranin and lack expression of p63 [22, 23]. p63 is a marker that is expressed by the majority of squamous cell carcinomas and can be helpful in distinguishing a basaloid squamous cell carcinoma from other high-grade epithelial malignancies such as small cell carcinoma [23].
Fig. 24.8
Small cell carcinoma. The tumor cells of small cell carcinoma are two to three times larger than a resting lymphocyte. The tumor cells have scant cytoplasm and fine nuclear chromatin with inconspicuous nucleoli. Crush artifact and nuclear molding are characteristic features of small cell carcinoma (Pap stain, 20x)
The mediastinum can be sampled to stage lymph nodes most commonly in the setting of lung cancer or to investigate adenopathy of unknown etiology. Primary neoplasms and other pathologic processes in the mediastinum can also be aspirated. It is important that the pathologist knows the patient’s history, mediastinal location (anterior, middle, or posterior), and imaging characteristics of the lesion as the differential diagnosis varies depending on these variables.
Lymph Nodes
EUS-FNA can be used to evaluate adenopathy of unknown etiology as well as to stage epithelial malignancies. Commonly sampled nodes that are accessible by EUS-FNA include paraesophageal, mediastinal, perigastric, peripancreatic, retroperitoneal, and perirectal lymph nodes. The differential diagnosis for adenopathy is extensive; however, common etiologies include metastases, lymphoma, infection, or non-neoplastic conditions such as sarcoidosis. On-site assessment of lymph nodes can be extremely helpful in appropriately triaging material obtained for ancillary studies.
With respect to metastatic tumors , the cytomorphology is in most cases identical or similar to the primary tumor. However, sometimes the metastatic tumor cells will be few in number and admixed with the background nodal lymphocytes that can make identification of rare tumor cells difficult. Occasionally, immunohistochemistry may be needed to further define the site of origin and a cell block will be necessary. Following treatment of adenocarcinomas, often all that remains is thick mucinous pools admixed with inflammatory cells. In this post-treatment setting, mucin signifies prior disease and does not constitute residual tumor.
EUS-FNA is useful for diagnosing lymphoproliferative disorders. Most lymphoma aspirates are characterized by a monotonous population of discohesive, often enlarged cells with a very high nuclear-to-cytoplasmic ratio, i.e., very scant cytoplasm (Fig. 24.9). The cytologic appearance of lymphomas is heterogeneous and beyond the scope of this chapter. However, a preliminary interpretation of “lymphoma” or “suspicious for lymphoma” and even “polymorphous lymph node favor reactive lymphadenopathy” often requires an additional important step: triaging material for flow cytometry. Flow cytometry is helpful in identifying B-cell clonality but less useful in detecting T-cell lymphomas. While cytomorphology with flow cytometry may provide a diagnosis, subclassification sometimes requires additional tests. Other ancillary studies such as cytogenetics, fluorescent in situ hybridization, and molecular testing may aid in diagnosis. Multiple separate FNA passes are often required to yield an adequate sample for these ancillary studies, and a generous lesional sample for cell block is often desired. A recent study demonstrated that sufficient material can be recovered from EUS-FNA to perform immunohistochemistry [24]. A specific lymphoma was diagnosed in 88.8 % cases (135 of 152 cases), and enough material was obtained for flow cytometry and cytogenetics in this study [24]. In our experience, the diagnosis of some lymphoproliferative disorders, such as T-cell lymphomas and Hodgkin’s lymphomas, can be challenging with EUS-FNA, and repeat procedures are sometimes needed to obtain diagnostic material.