Respiratory Cytology - Introduction


The diagnosis of lung cancer by cytologic methods is of historical interest because it was an early demonstration that malignancy could be diagnosed through the examination of exfoliated cells. Donne and Walsh separately noted that exfoliated respiratory cells occurred in sputum in 1845. The first major series of patients where the examination of sputum led to a diagnosis of lung cancer was by Hamplen in 1919, where 13 of 25 cases were successfully diagnosed. After several years of quiescence, pulmonary cytology enjoyed a period of rapid development in the 1970’s and 1980’s, particularly as fine needle aspiration was validated as an alternative to open lung biopsy or transbronchial biopsy for the diagnosis of pulmonary malignancy, infectious and inflammatory lung diseases. During this time there were technical advances in radiological technique that permitted the imaging of small lesions and guided aspiration biopsy, as well as improvements in the design of bronchoscopes.

In the last 40 years the value of respiratory cytology for the diagnosis of infectious and inflammatory disease has become particularly evident. This was most notable in the 1980’s, with the emergence of Acquired Immune Deficiency Syndrome, and the use of high-dose immunosuppressive chemotherapy. Each renders the patient susceptible to infections, such as Pneumocystis carinii and Aspergillus sp, which are readily diagnosed by the examination of a respiratory sample.

Sampling Techniques
There are five major techniques used to obtain cellular material for the diagnosis of lung cancer. The oldest and most fundamental technique is sputum collection which depends on the spontaneous exfoliation of cells. Bronchoscopic techniques include bronchial washing, bronchial brushing, and bronchoalveolar lavage. Finally, needle aspiration techniques can be performed through the chest wall under radiographic guidance or during bronchoscopy.

No one technique is necessarily superior to the others. The choice of cell collection technique is shaped by the personal preference of the physician, the status of the patient, the location of the lesion and other factors. While there is substantial similarity among the morphology of cells obtained by different techniques there are important differences related to cell preservation and specimen processing. An example is seen in the different sensitivity of lung cancer detection between freshly smeared and fixed, blended sputum. Artifacts caused by the mechanical blending of the specimen will lower the sensitivity of disease detection for small-cell undifferentiated carcinoma and adenocarcinoma. The mechanical blending disrupts cell fragments and shears mucin-containing vacuoles from cells, thus complicating the diagnosis of adenocarcinoma. Fixation will affect the nuclear diameter and the staining quality of the nuclear chromatin. For example, nuclei of small cell carcinoma are larger and more vesicular in samples collected by brushing or needle aspiration, as compared to sputum samples. Increasingly liquid-based methods are used for specimen collection and processing. These methods will impose unique artifacts that must be understood by the diagnostic cytologist.

Sensitivity and Specificity of Respiratory Cytology for Cancer Diagnosis
Lung cancer is a major health problem in the US and worldwide. In the United State alone, 45 million current smokers and 45 million former smokers are at risk. Properly identifying patients with low-stage disease is the most important step in reducing disease mortality. If lung cytology is to be used for diagnostic purposes, its accuracy including both sensitivity and specificity needs to be understood. Sensitivity describes how often a cytologic method accurately identifies lung cancer in a patient when tumor is actually present. In general, an unequivocal cytologic diagnosis of cancer can be made in approximately 50% of sputum samples and 65% of bronchial brushings, lavages or brushings. Fine needle aspiration, particularly when guided by computerized tomography, has a higher sensitivity of nearly 90%. The most comprehensive analysis of pulmonary fine needle aspiration was reported as part of the College of American Pathologists Q-probe Quality Assurance Program. In this study of 13,094 lung fine needle aspirates at 436 institutions, the sensitivity of the fine needle aspiration procedure was 89%, with 96% specificity, 99% positive predictive value, and 70% negative predictive value. The false positive rate was 0.8% and false negative rate 8%. The issue of false negative cytology is vexing for pathologists, as the absence of abnormal cells may be due to either sampling error (no abnormal cells were collected) or interpretive errors (abnormal cells on the slide were not correctly identified).

Comparative studies have assessed the relative sensitivity of bronchial washing/brushing with bronchial biopsy. Naryshkin et al described 224 cases where 75% correlated completely. In the remainder, biopsy and cytology were diagnostic in relatively equal proportions, leading the authors to conclude that a specific diagnosis was obtained more often from the combination of cytology and biopsy, than from either alone. The issue of specificity has two elements: how often, when a diagnosis of malignancy is rendered, is malignancy present and how successful is cytology in predicting the classic histologic types of malignancy. Avoiding the false-positive malignant diagnoses of malignancy requires consideration of the nuclear morphologic features that classically predict malignancy. In large series, the reported false positive rate due to atypical forms of metaplasia and reactive and reparative changes ranges from 0 to 2%.

A critical aspect of specificity relates to the correlation between different types of respiratory tract specimens. Comparisons may be made among sputum, bronchial cytologic specimens, fine needle aspiration and histologic materials. In a large study by Johnston the prediction of histologic type of primary lung cancer from sputum and bronchial material was compared with histologic diagnosis. The rate of concordance was 85% with squamous cell carcinoma, 79% with adenocarcinoma, 30% with adenosquamous carcinoma, 30% with large cell carcinoma, and 93% with small cell carcinoma. As mentioned previously, the variable sampling of different areas within a tumor accounts for most of the poor correlation between cytologic and histologic diagnoses. Fortunately there is a relatively high correlation rate between the histologic and cytologic diagnosis of small cell undifferentiated carcinoma, as this is important for the selection of therapy.

Normal Respiratory Tract Cytology
The respiratory tract communicates with the external environment through the oropharynx. Because of this a variety of external substances (e.g. pollen, ferruginous bodies) and oral contaminants (e.g. Candida, Actinomycetes) may be found in the respiratory specimen.

Cells normally found in the respiratory sample include ciliated and non-ciliated columnar epithelial cells, macrophages, epithelial cells from the terminal bronchioles, and inflammatory cells. The number and proportion of cells will differ depending on the type of sample, and the underlying disease.

Epithelial cells can exhibit significant reactive change in response to injury by infection, radiation or chemotherapy. Thus, cytomegaly, nuclear enlargement and chromatin changes must be interpreted in the context of the patient’s history.

Infections Affecting the Respiratory Tract
All types of organism can affect the respiratory tract, but cytologic examination is most useful for those that cause cytomorphologic alteration (e.g. viral inclusions) or the organisms can be visually recognized (e.g. fungal hyphae). The photomicrographs that follow in this section demonstrate the most common types of organisms. Among viral infections, these include Herpes Simplex and Cytomegalovirus, both of which cause characteristic intranuclear inclusions and, in the case of cytomegalovirus, intracytoplasmic inclusions.

Fungal infections commonly identified by cytology include Aspergillus sp, Candida sp, Cryptococcus neoformans, Histoplasma capsulatum, Mucomycosis and Blastomyces dermatiditis.

Key morphologic features of Lung Cancer
The boundaries of morphologic features of the major types of lung cancer are often fuzzy, since combined types of tumors can occur, and this can pose diagnostic problems. For example, non-small cell carcinoma is seen accompanying 5 to 10% of cases of small cell carcinoma. The non-small cell component may be squamous cell carcinoma, adenocarcinoma or large cell carcinoma, individually or in combination.

Squamous Cell Carcinoma
Classically the cells of squamous cell carcinoma are arranged in cohesive sheets or lie singly in a background of necrosis. They may exhibit a spectrum of shapes and sizes. Cells may be small and dyskeratotic, only slightly larger than a metaplastic cell, or be many times the diameter of a normal bronchial epithelial cell. Bizarre forms, such as tadpole shapes and keratin pearls of tumor cells, may be seen in keratinizing forms of lung cancer.

Cytoplasmic features are used to determine if squamous differentiation is present. These include a sharp border, orange, or deep basophilic staining of the cytoplasmic keratin, and filaments of keratin ringing the outer diameter of the cell.

Nuclear features identify the cells as malignant. The nucleus is enlarged, and may show deep grooves, sharp angulations or clefts. The nuclear envelop is irregular in thickness. The chromatin may be irregularly clumped or be so dense that the nucleus is opaque. Nucleoli, when present, are typically multiple, and are irregularly shaped.

Benign conditions can mimic squamous cell carcinoma of the lung. When metaplastic epithelium is irritated by chemotherapy or radiation therapy or by inflammatory conditions (e.g. Aspergillus infection, pulmonary infarction), the metaplastic cells may resemble neoplastic cells. Poorly differentiated squamous cell carcinoma may be difficult to distinguish from other poorly differentiated cancers, either from the lung or from metastatic sites.

Adenocarcinomas typically arise in the peripheral portions of the lung. Adenocarcinoma is notable because it now constitutes the second most common malignancy among women. The most common morphologic type of adenocarcinoma is acinar adenocarcinoma. This malignancy forms glandular structures that, depending on the degree of differentiation, form three-dimensional structures that may have a central glandular lumen. In more poorly differentiated tumors, this feature may be absent, and the diagnosis will be determined by the cytoplasmic features. Adenocarcinomas have abundant cytoplasm which can be vacuolated or bubbly in appearance. The cytoplasmic border is less sharply defined than squamous cell carcinoma because of less cytoplasmic keratin.

The nuclei of adenocarcinoma are typically round, and may be relatively uniform in size and shape compared to squamous cell carcinoma. The chromatin is vesicular and nucleoli, when present, are usually centrally located in the nucleus.

Adenocarcinoma may arise from the terminal bronchiolar epithelium. These tumors, also known as bronchioloalveolar cell carcinoma, have distinctive clinical and morphologic features.

Morphologically the cells are conspicuous in their uniformity. Rounded cells, approximately the size of a pulmonary macrophage are arranged in flower-petal configurations. The nuclear chromatin is pale, almost clear, and cells have a prominent, round, centrally located nucleolus. Pulmonary infarct can cause reactive changes among the terminal bronchiolar cells that can be misinterpreted as bronchioloalveolar carcinoma.

Large Cell Undifferentiated Carcinoma
Cells of large cell undifferentiated carcinoma lack features of either squamous or glandular differentiation, and they are “large”. They are often giant, pleomorphic and have obvious malignant nuclear features. Cells may be multinucleated, the chromatin is coarsely clumped and macronucleoli may be present. Single cells predominate so that the differential diagnosis includes other malignancies, such as melanoma and large cell lymphoma.

Small Cell Undifferentiated Carcinoma
The diameter of cells from small cell undifferentiated carcinoma is approximately 1 1/2 to 2 times that of a small lymphocyte. The small cells are rounded, or carrot shaped. At low magnification tumor cells appear to lack cytoplasm, but on high magnification a thin rim of cytoplasm can be discerned in well-preserved cells. Cells can occur singly, in a linear arrangement within respiratory mucous, or in loosely cohesive clusters. A key feature is the molding of tumor cells around each other, the result of the rapid growth rate of the tumor within the confined space of the submucosa. Other diagnostic features important to the cytologist are the characteristic “salt and pepper” chromatin, necrosis of individual tumor cells within cell clusters, and scant cytoplasm. Small cell carcinoma includes cases which have a larger cell size with a diameter approximately 3 times the diameter of a lymphocyte. These correspond to those cases regarded as the intermediate subtype of small cell carcinoma as defined by the WHO in 1981. Distinguishing this tumor from poorly differentiated non-small cell carcinomas can be difficult. A particular challenge is the so-called large cell type of neuroendocrine carcinoma that has close similarity to large cell carcinoma.

The differential diagnosis of small cell carcinoma includes reserve cell hyperplasia, inspissated mucus, lymphocytes and other small cell tumors. By far the most common problem is reserve cell hyperplasia. Reserve cells normally lie along the basement membrane of the epithelium, where they serve as progenitors for bronchial epithelial cells. Irritation of the airways stimulates reserve cell proliferation. Several features will correctly identify reserve cells. They are approximately the size of a small lymphocyte with dense chromatin that lacks the salt-and-pepper quality and micronucleoli of small cell carcinoma. Reserve cells are almost always tightly clustered, and cells within the cluster retain cell boundaries without the extreme nuclear molding that characterizes small cell carcinoma.