If a tumor is poorly differentiated, an immunoperoxidase workup may start with lineage determination using a broad spectrum of markers, followed by second and sometimes third panels to define the tumor subtype and primary origin (Fig. 1.​1). If a tumor is well to moderately differentiated and shows clear evidence of cell lineage or subtype, then specific markers can be ordered initially. The number of markers depends on the slides available and priority. Sometimes, a single marker is sufficient for confirmatory purpose if a clinical history of primary tumor is known and the cytologic features are compatible.

Dual or multiplex staining, in which two or more immunostains are performed on a single slide, can be used in certain cases to save tissue. Dual staining of TTF1 and Napsin A is one of the examples (Fig. 4.5); positive staining seen in either or both markers is often used to confirm a lung primary and an adenocarcinoma subtype.

The most frequently encountered tumor entities in cytology practice should be considered before selecting markers for detecting rare and unusual tumors.

Nuclear markers are generally superior to cytoplasmic or membranous markers in cytology preparations. Compared to nonnuclear markers, nuclear staining tends to be more specific and easier to interpret, especially when staining is performed on direct smears in which the intercellular relationship and cell membrane may be distorted or stripped off during smearing (Fig. 4.6). Any true nuclear staining should be considered significant. The commonly used nuclear markers and nonnuclear markers are listed in Tables 4.1 and 4.2, respectively. A few newly discovered yet promising markers are also included. Currently specific markers may become less specific and may be replaced by new markers and new clones that are continually emerging.

A specific positive result for a particular marker has greater value than a negative result because negative staining may reflect a true negative biologic reaction; however, it may also result from false negativity. Therefore, marker selection priority should be given to markers that are expected to be positive, although markers that are expected to be negative may be included in a panel.

If immunophenotypic information about a previous tumor in the same patient is available, selection of positively expressed specific markers to work up the current tumor is helpful to confirm that the two tumors are of the same disease process (Fig. 4.7).

Marker selection needs knowledge of metastatic pattern. For example, in view of the fact that thyroid gland is an important recipient for metastatic renal cell carcinoma, PAX8 should not be selected to work up a suspected metastatic renal cell carcinoma in the thyroid because tumors originated from both organs express PAX8.

Correlation with clinical and radiologic findings and cytologic features.

Knowledge of the distribution of specific staining in cells (nuclear, cytoplasmic, or membranous).

Focusing on staining in healthy and viable tumor cells, avoiding degenerated cells or cells in necrotic areas.

Judicious examination of positive and negative controls that ideally should be processed and stained in the same manner as the test sample at each run of staining.

Immunostaining on non-cell block preparations, such as direct smears, usually does not have proper control tissue. Extra caution should be taken during interpretation.

Definition of positivity in terms of the percentage of positive lesional cells required may vary for different markers. However, for the interpretation of prognostic and predictive markers such as ER, PR, and HER2, standard guidelines should be followed.

Net results of the entire immunostaining panel should be used when considering a final diagnosis.

False-negative results are more common than are false-positive results. Therefore, a positive result is generally more meaningful than a negative result.

False-negative results are largely due to sample error (i.e., scarcity of lesional cells available for immunostaining from a tumor that intrinsically expresses the markers only focally and heterogeneously) and inappropriate preanalytical and analytical factors (e.g., using non-cell block material for immunostaining without technical validation). Certain stains, such as S100, cannot be reliably assessed on ethanol-fixed preparations, possibly resulting in false-negative findings.

False-positive results are usually due to erroneous interpretation. The lack of reliable histologic architecture in the aspirated material can cause entrapped benign cells at the aspiration site to be mistaken for tumor cells. For example, an aspirate of a metastatic carcinoma to the lungs may be admixed with entrapped benign bronchial epithelial cells that express TTF-1, potentially leading to the erroneous conclusion that the tumor is focally TTF-1 positive and thus a lung primary. In addition, nonspecific staining is often seen in crowded cell groups of thick smears or in association with the edge effect. It can cause a false-positive interpretation. Occasionally, staining of the cytoplasm overlying a nucleus which is actually negative for a nuclear marker may appear weak reactivity in the nucleus due to the integrity of the cells and the three-dimensional nature of the cells on the smear.

Be aware of some caveats that may be associated with TTF1, CDX2, and PAX8, the most commonly used nuclear markers in cytology practice. For example, CDX2 may be positive in adenocarcinomas from the lung, ovary, and urinary bladder with enteric phenotype and yolk sac tumor. Lung adenocarcinomas with enteric phenotype or mucinous subtype usually have lower frequency of TTF1 expression than those with nonmucinous subtype. Likewise, ovarian adenocarcinomas with enteric phenotype or mucinous subtype less likely express PAX8 than those with nonmucinous subtype. In addition, when a distinction between thyroid and parathyroid or between thyroid and thymic tumor is needed, PAX8 should not be selected because lesions from these origins can express PAX8. PAX8 can also be positive in lymphoid cells; thus caution should be taken when using PAX8 to detect metastatic carcinoma in lymph node sample. In addition to carcinoma originated from renal, thyroid, and gynecologic organs, PAX8 is frequently expressed in well-differentiated pancreatic endocrine tumor. Lastly, PAX8 is usually used to distinct renal cell carcinoma (positive) from urothelial carcinoma (negative); however, a subset of upper tract urothelial carcinoma may express PAX8.