Architectural aspects of cell-blocks as small biopsies

Effusions

Cell-block preparations have been utilized for effusion specimens over many years and they are invaluable in the work-up of neoplastic effusions.^[30-34] Effusion cytology along with cell-block preparation is superior to tissue biopsy of a body cavity as it is less invasive, allows wider sampling and conveys high diagnostic accuracy.^[35,36] In most cases, the cytologic diagnosis of a malignant effusion is straightforward. While the optimum number of effusion specimens suggested for reaching a positive diagnosis is three, in the majority (up to 90% of cases) metastatic carcinoma can often be detected on the first specimen.^[37-40] Malignancies can present as large clusters (e.g., metastatic adenocarcinoma) or single cells (e.g., gastric signet ring carcinoma, melanoma, and sarcoma). Rarely, effusions may manifest with a mesotheliomatous pattern of metastatic adenocarcinomas where the specimen contains only single malignant cells.^[41-51]

Cytomorphologic features that are helpful in the identification of metastatic carcinoma include cell size (e.g., very large cells), cell shape (e.g., bizarre cells), presence of a second cell population other than mesothelial cells [Figure 3], numerous large three dimensional clusters, cell configuration (e.g., tumor cannibalism [Figure 4], spherical clusters, and lumina), cytoplasmic findings (e.g., intracellular mucin, and melanin), nuclear features (e.g., increased nuclear-cytoplasmic ratio, hyperchromasia, bizarre shapes, and atypical mitoses), and a bloody or mucinous background. In metastatic breast carcinoma, the finding of cannonball shaped cell clusters is a helpful clue [Figure 5]. The glandular differentiation of large cannonball clusters is most apparent on cell-blocks where a central lumen can be readily observed on cross section. Features that may help distinguish metastatic adenocarcinoma from mesothelioma include smooth, community borders in clusters of metastatic adenocarcinoma as opposed to knobby, flower-like outlined clusters of mesothelioma. While it may be easy to recognize papillae when present in effusions, these architectural structures are easy to appreciate in cell-blocks. Such papillae may be observed with metastatic carcinoma such as lung, breast or ovarian serous carcinoma [Figure 6], as well as with papillary mesothelial hyperplasia and malignant mesothelioma [Figure 7].

Close

Figure 3:

Cell-block from a peritoneal wash with metastatic colorectal adenocarcinoma. (a) Note the clusters of malignant cells (arrows) that are distinct from the many fragments of mesothelial-lined peritoneum (hematoxylin and eosin, ×200). (b) A calretinin stain highlights an attenuated benign mesothelial lining (immunohistochemistry, ×200).

Export to PPT

Close

Figure 4:

Cell-block section of a pleural effusion with metastatic breast carcinoma showing a cell-in-cell appearance due to tumor cannibalism (hematoxylin and eosin, ×600).

Export to PPT

Close

Figure 5:

(a and b) Pleural effusion with metastatic breast ductal carcinoma. (Left) A large cannonball cluster of metastatic carcinoma cells is shown with a smooth, community border (Cytospin, Diff Quick stain, ×200). (Right) Cell-block showing many large clusters of metastatic breast carcinoma with central comedo necrosis (hematoxylin and eosin, ×200).

Export to PPT

Close

Figure 6:

Cell-block section of ascitic fluid showing numerous papillary clusters in a case of metastatic ovarian serous carcinoma (Hematoxylin and eosin, ×200). Also note the presence of pericellular lacunae in this case.

Export to PPT

Close

Figure 7:

Cell-block section of a well differentiated papillary mesothelioma showing numerous small clusters of mesothelial cells with knobby borders (Hematoxylin and eosin, ×200).

Export to PPT

The presence of pericellular lacunae in cell-blocks is suggestive of metastatic adenocarcinoma [Figure 8]. As these lacunae are easily visible at low magnification, their recognition may help in screening cell-blocks. However, the concept of pericellular lacunae and their association with malignancy remains debated. Very few studies have addressed this issue. Price et al. studied over 200 cell-blocks of effusion specimens and reported that pericellular lacunae were found in malignant (although 75% of cases) and benign effusions.^[52] From an architectural aspect, the majority of malignant cases in their study showed lacunae surrounding greater than two-thirds of the cells, whereas in the benign cases, when present, they were seen in less than one-third of the cells. In another study, McNeely divided effusions into three categories as no lacunae, <50% and >50% of cell clusters with pericellular lacunae. None of the benign effusions had >50% lacunae, whereas most malignant effusions had lacunae either in the <50% category (in 67% of cases) and >50% category (in 27% of cases). The author concluded that pericellular lacuna by itself is therefore not a reliable diagnostic criterion.^[53] Similarly, Thomson and Hayes reported lacunae in both benign and malignant effusions.^[54]

Close

Figure 8:

(Left) Cell-block section showing a pericellular lacuna encircling more than two-thirds of the cell cluster in this case of metastatic adenocarcinoma involving the pleural cavity. Also note the presence of acini with nuclei polarized away from the center (hematoxylin and eosin, ×200). (Right) Cell-block section from a case of malignant mesothelioma. Note that in this case the large cluster of mesothelial cells has a solid to focal pseudo-acinar architecture and no pericellular lacuna (hematoxylin and eosin, ×200).

Export to PPT

Urine

Urine cytology is the gold standard when screening for bladder cancer, especially for the detection of high-grade urothelial carcinoma (HGUC). The sensitivity of urine cytology for detecting HGUC ranges from 50 to 85% and is much lower (10–43%) for detecting low-grade urothelial neoplasia (LGUN).^[55,56] After the adoption of the Paris system for reporting urine cytology, these rates are improving. McCroskey et al. demonstrated poor to fair correlation among different cytopathologists for diagnosing LGUN with a low a sensitivity (21–53%), but high specificity (81–95%).^[57] According to the Paris system, the presence of fibrovascular cores is necessary to make a diagnosis of LGUN.^[58] Several papillary urothelial neoplasms are in fact characterized by the presence of fibrovascular cores [Figure 9] because this category includes benign urothelial papilloma, low-grade papillary neoplasm of low malignant potential, low-grade papillary urothelial carcinoma, and high-grade papillary urothelial carcinoma.

Close

Figure 9:

Urine cell-block showing a three-dimensional cluster of cells with central fibrovascular cores. The cells show mild cytologic atypia and maintained polarity (hematoxylin and eosin, ×200). Follow-up biopsy revealed a low-grade papillary urothelial carcinoma.

Export to PPT

Unfortunately, finding a papillary cluster of urothelial cells with a fibrovascular core in a urine specimen is uncommon. Cell-blocks are beneficial because they increase the likelihood of detecting a papillary architecture with visible fibrovascular cores. Furthermore, specimen types such as washings versus voided urine may increase the possibility of incorporating small tissue fragments in samples that make it easier to diagnose LGUN.^[59] Although cell-blocks are an important part of integrated cytology diagnosis, very few studies have analyzed their utility for urine specimens.^[5,59-65] Limiting factors for the routine adoption of cell-blocks when processing urine is that many specimens are paucicellular. Brisuda et al. found that female gender, positive urine cytology, and leukocytouria were associated with improved urine cell-block adequacy.^[62] Nevertheless, Mansy et al. reported that cell-blocks for urine samples enhances the overall diagnostic rate.^[63] Dantey et al. reported that routinely making urine cell-blocks only improves the diagnostic rate for the atypical urothelial cells (AUC) category.^[64] Similarly, Chan et al. reported improved diagnostic rates for AUC, as well as the HGUC category with the adoption of the Paris system and cell-blocks.^[65]

Thyroid gland

Cell-blocks are not required for most thyroid FNAs. In addition, cell-blocks made from bloody thyroid aspirates are often paucicellular. However, a cell-block can be very helpful in certain situations [Figure 10], especially when immunohistochemistry is needed to confirm a diagnosis (e.g., to differentiate medullary thyroid carcinoma from other thyroid and parathyroid neoplasms). Few studies have reported on the utility of cell-blocks for thyroid FNAs.^[66-68] Li et al. Recommend the use of adjunct cell-blocks for radiologic thyroid category four and above lesions.^[69] In general, these studies indicate that cell-blocks may lead to a reduction of the unsatisfactory rate and can improve the sensitivity and specificity of thyroid FNAs. However, Sanchez and Selvaggi reported that cell-blocks were contributory in only 31% of the 82 cases in their study.^[70] Horton et al., who studied cell-blocks in 965 paired ThinPrep thyroid samples, did demonstrate an improved rate of non-diagnostic and atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS) category cases.^[71]

Close

Figure 10:

Cell-block showing thyroid follicular cells with cribriform architecture in a case of cribriform/morular variant of papillary thyroid carcinoma. Note the presence of elongated nuclei with focal nuclear groves and intranuclear pseudoinclusions (Hematoxylin and eosin, ×200).

Export to PPT

Salivary glands

Hong et al. reported an improved diagnostic rate for parotid FNAs when cell-blocks were utilized.^[72] However, cell-blocks are mostly used with salivary gland FNAs to perform ancillary tests such as immunohistochemistry and fluorescent in situ hybridization.^[73] Nevertheless, architectural patterns of salivary gland tumors that can be better appreciated on cell-block preparations include the double layer of epithelial cells resting on dense lymphoid stroma in Warthin tumor, cribriform glands of adenoid cystic carcinoma with central basement membrane-like material, biphasic tumors (e.g., epithelial myoepithelial carcinoma with luminal ductal cells and abluminal myoepithelial cells surrounded by extracellular stroma), variable tumor growth patterns (e.g., papillary, trabecular, and tubular), and possibly the infiltrating nature of some carcinomas.

Lung

Many patients with lung cancer primarily diagnosed by FNA require subclassification of their tumors for appropriate treatment triage. Hence, apart from distinguishing small cell versus non-small cell lung carcinoma, further subtyping of non-small carcinoma into squamous cell carcinoma or adenocarcinoma is required.^[74,75] Thus, the routine use of cell-blocks for this purpose is warranted for lung FNAs. Loukeris et al. studied the value of cell-blocks in 62 cases of lung carcinomas. Their results showed that cell-blocks provided additional morphological clues for the classification of lung carcinomas.^[76] For example, the presence of intercellular bridges may be better appreciated in cell-block sections of squamous cell carcinomas [Figure 11]. For basaloid squamous cell carcinoma a helpful finding that can be identified on cell-blocks is peripheral palisading in clusters of basaloid cells. Adenocarcinomas may exhibit helpful architectural patterns on cell-blocks such as strips of glandular cells, papillary groups, and acinar clusters of tumor cells with a cribriform pattern and the presence of glandular lumina. Ancillary tests are often needed for poorly differentiated non-small cell carcinomas, and although immunohistochemistry is helpful in challenging cases a mucicarmine stain can sometimes help to diagnose adenocarcinoma [Figure 12]. Demirci et al. demonstrated that cell-blocks are also important when distinguishing small cell and large cell neuroendocrine carcinoma (NECa).^[77] Helpful morphologic features of large cell NECa on cell-block include the readily apparent abundant pink cytoplasm and conspicuous nucleoli of tumor cells, as opposed to small cell NECa which shows clusters of basaloid cells with minimal to absent cytoplasm.

Close

Figure 11:

Cell-block section of squamous cell carcinoma of the lung showing squamous cells with large central hyperchromatic nuclei and prominent intercellular bridges (hematoxylin and eosin, ×400).

Export to PPT

Close

Figure 12:

Cell-block section showing intracytoplasmic mucin in a case of lung adenocarcinoma (Mucicarmine stain, ×400).

Export to PPT

Liver

In normal liver, hepatocytes are arranged in plates, where the apical surface faces sinusoids and the basal surface forms canaliculi. A bile canaliculus is not a true duct, but rather a dilated intercellular space between hepatocytes that lacks an epithelial lining. Recognition of this architecture in cell-blocks from liver aspirates can help distinguish benign liver from neoplasia. In normal liver, these plates are typically only two hepatocytes thick. However, in well-differentiated hepatocellular carcinoma (HCC), these plates can be 2–3 hepatocytes thick and in moderately differentiated HCC they are four or more hepatocytes thick. A reticulin stain can be used to easily highlight such abnormal liver architecture [Figure 13]. Other architectural features of HCC that may be helpful in cell-block sections include the absence of bile duct epithelial cells, the presence of many large solid clusters of hepatocytes with increased nuclear-cytoplasmic ratio, peripheral endothelial wrapping, and pseudo-acinar formation^.[78,79]

Close

Figure 13:

(a and b) Reticulin stains on cell-blocks from liver FNAs showing (left) normal liver comprised of 1–2 layer thick hepatic plates and (right) a hepatocellular carcinoma with loss of normal architecture and very broad hepatic plates (Reticulin stains, ×200).

Export to PPT

Pancreas

FNA remains the gold standard for diagnosing pancreatic lesions. Bang et al. along with multiple other studies demonstrated that new generation endoscopic biopsy needles (e.g., SharkCore) are very efficient for adequate collection of samples from solid pancreatic lesions.^{[12-16,80,81]} These specimens often contain small fragments or microbiopsies showing intact tissue architecture [Figure 14]. Indeed, pancreas cell-blocks can be very helpful when they reveal features of malignancy such as complex glandular architecture and invasion of stroma with desmoplasia. Such features are particularly helpful for diagnosing well-differentiated pancreatic adenocarcinoma, particularly in cases with equivocal or suboptimal cytology smears. Furthermore, cell-blocks employed in conjunction with ancillary tests (e.g., immunohistochemistry for SMAD4 or neuroendocrine markers) can often help reach a definitive diagnosis.^[82,83]

Close

Figure 14:

(a-c) Pancreas FNA using SharkCore in which direct smears were suboptimal to reach a definitive diagnosis of malignancy. The cell-block in this case helped reach a diagnosis of adenocarcinoma because it was easy to appreciate (a) malignant glands, (b) tumor necrosis, and (c) invasion with desmoplastic stroma (hematoxylin and eosin, ×200).

Export to PPT

Key Features

• Cell-blocks represent microbiopsies that have become an essential part of integrated cytology diagnosis

• Cell-blocks allow small tissue fragments, that are procured using newer procedures and needles, to be processed and evaluated by the cytology laboratory

• Recognition of cell-block architectural patterns has accordingly become important in this emerging era of cytohistology

• Cell-blocks are suitable for whole slide imaging because they avoid focus problems plagued by other cytology preparations.