Respiratory Tract and Mediastinum

Sputum

Sputum consists of a mixture of cellular and noncellular elements that are cleared by the mucociliary apparatus. It was once the most common respiratory tract specimen because it is relatively easy to obtain, with little discomfort to the patient. Sputum cytology is generally reserved for symptomatic individuals; as a screening test (e.g., in symptom-free smokers), sputum cytology is not effective in decreasing rates of death from lung cancer. With the advent of bronchoscopy and FNA, its use as the mainstay in respiratory cytology has declined significantly.

Collecting multiple sputum samples over several days optimizes sensitivity. Early morning, deep cough specimens are preferred. If the patient is not able to expectorate adequately, expectoration can be induced by having the patient inhale nebulized water or saline solution. Sputum induction increases the detection of lung cancer. When prompt preparation of sputum is not possible, the patient can expectorate into a 70% ethanol solution, which prefixes the specimen.

A simple method of sputum preparation is known as the “pick and smear” technique, whereby fresh sputum is examined for tissue fragments, blood, or both. Smears are prepared from areas that contain these elements and immediately fixed in 95% ethanol. A modification of this is the Saccomanno method, which calls for sputum to be collected in 50% ethanol and 2% carbowax ; it must be performed in a biologic safety hood because of the risks of infection from aerosolization. The specimen is then homogenized in a blender and concentrated by centrifugation. Improved sensitivity has also been demonstrated by the use of dithiothreitol, N-acetyl-L-Cysteine, or CytoRich Red for mucolysis and homogenization. Smears are made from the concentrated cellular material. Sputum can also be processed using thin-layer methods or embedded in paraffin for cell block sections.

The adequacy of a sputum sample is established by finding numerous pulmonary macrophages. Specimens consisting merely of squamous cells, bacteria, and Candida organisms are unsatisfactory because they represent only oral contents. Even ciliated cells, which also line the sinonasal passages, do not guarantee that a sample is from the lower respiratory tract. The presence of numerous macrophages indicates that a satisfactory, deep cough specimen of the lower respiratory tract has been obtained. In an adequate sample they should not be difficult to find: if they are absent or few in number, the sample should be reported as unsatisfactory.

The sensitivity of sputum cytology for the diagnosis of malignancy increases with the number of specimens examined, from 42% with a single specimen to 91% with five specimens. The specificity of sputum examination is high, ranging from 96% to 99%, and the positive and negative predictive values are 100% and 15%, respectively. Thus negative sputum results do not guarantee the absence of a malignancy, especially in a patient suspected of having lung cancer. The sensitivity of sputum cytology depends also on the location of the malignant tumor: 46% to 77% for central lung cancers but only 31% to 47% for peripheral cancers. Sensitivity is independent of tumor stage and histologic type. Accuracy in tumor classification is 75% to 80% and is tumor-type dependent.

Bronchial Specimens

A pivotal improvement in sampling the lower respiratory tract occurred with the development of the flexible bronchoscope in the late 1960s. Today, any part of the respiratory mucosa can be sampled with this device.

Complications of bronchoscopy are rare (0.5% and 0.8% for major and minor complications, respectively) and include laryngospasm, bronchospasm, disturbances of cardiac conduction, seizures, hypoxia, and sepsis. The incidence of major complications is higher for transbronchial biopsy (6.8%).

Transbronchial FNA (“Wang Needle”)

Transbronchial FNA is especially useful for pulmonary lesions hidden beneath the bronchial surface and for suspicious mediastinal lymph nodes. Transbronchial FNA eliminates the need for an addition surgical procedure in 20% of patients, at one-third the cost of mediastinoscopy. The lesion is aspirated with a retractable (Wang) needle passed through a flexible catheter that is sent down the bronchoscope.

At least a moderate number of lymphocytes must be present to ensure the adequacy of lymph node sampling and avoid a false-negative result. Ciliated respiratory epithelial cells are common contaminants because the respiratory mucosa needs to be breached to reach the target. For this reason, ciliated cells should not be taken as evidence of adequate sampling of a mediastinal lymph node. Complications from transbronchial aspiration are rare and include endobronchial bleeding, which is usually controlled by suctioning. Contraindications are coagulopathy, respiratory failure, and uncontrollable coughing.

Transbronchial FNA augments the diagnostic accuracy of bronchial washings, brushings, and endoscopic biopsy for the detection of primary pulmonary neoplasms. The sensitivity of transbronchial FNA by itself is 56% but increases to 72% when combined with bronchial brushing, washing, and biopsy. Specificity is 74%, and the positive and negative predictive values are 100% and 53% to 70%, respectively.

Transbronchial FNA is accurate in distinguishing small cell from non-small cell lung cancer. For mediastinal staging of bronchogenic carcinoma, the negative predictive value of transbronchial FNA increases from 36% to 78% when negative specimens without sufficient lymphocytes are regarded as unsatisfactory for evaluation. The most common cause of false-negative results is sampling error.

Endobronchial Ultrasound-Guided (EBUS) FNA

The accuracy of mediastinal staging by FNA improves with the use of ultrasound guidance. EBUS FNA is an enhanced procedure for sampling mediastinal and paratracheal lymph nodes and peribronchial lung or mediastinal lesions that is increasingly used in the United States. Its primary indication is nodal staging of non-small cell lung cancer, but it is also indicated for sarcoidosis and metastases from extrapulmonary primaries. This minimally invasive procedure is a safer alternative to cervical mediastinoscopy in selected patients. Using a bronchoscope equipped with an ultrasound probe tip, the operator performs an FNA with real-time ultrasound imaging of a lymph node or central lung mass. Sensitivity and specificity are 88% and 100%, respectively. To optimize sensitivity, a cytologist can assess the sample on site for adequacy. In the absence of lesional cells, adequacy is defined as the presence of lymphocytes (if a lymph node is being sampled) or pigmented macrophages (in the case of a lung mass). Some laboratories have developed and adopted quantitative adequacy criteria for mediastinal lymph node sampling (presence of anthracotic pigment, germinal center fragments, a minimum number of lymphocytes), but no single set of criteria has yet been universally adopted. As with transbronchial (Wang needle) FNA, ciliated respiratory epithelial cells are common contaminants and are not evidence of adequate sampling. Microscopic metallic dust from the needle ( Fig. 2.2B ) mimics anthracotic pigment. It can be recognized as a contaminant because of its haphazard distribution. The advantage of EBUS is that it increases accessibility to lower station lymph nodes: transbronchial (Wang needle) FNA can sample lymph node stations 2 to 4 and 7, and transesophageal ultrasound–guided FNA can reach stations 2 to 4 and 7 to 9, whereas EBUS can sample stations 2 to 4, 7, and 10 to 12. Thus esophageal FNA is often combined with EBUS for full accessibility.

Although the EBUS FNA procedure itself is more expensive than transbronchial (Wang) FNA, it saves downstream costs because of its greater sensitivity and the reduced need for more surgical staging, and it has largely supplanted Wang biopsy in many practices.

EBUS FNA may result in biopsy site changes, including intranodal or perinodal fibrosis and displacement of airway cartilage into lymph nodes. These artifacts, seen on subsequent lymph node excisions, should not be misconstrued as tumor-related changes.

Transesophageal FNA

Mediastinal lymph node sampling can also be done endoscopically by passing the needle through the esophagus. The addition of ultrasound guidance improves the accuracy of mediastinal lymph node sampling. Like bronchoscopic FNA, endoscopic FNA realizes significant cost savings and reduces the number of unnecessary thoracotomies. The advantages of endoscopic transesophageal FNA include improved image quality and the ability to sample stations 8 and 9, which are not accessible via EBUS FNA, but the primary lesion usually cannot be sampled via this route.

Endoscopic transesophageal FNA, when used in combination with transbronchial FNA, improves the diagnostic yield for mediastinal staging greatly, with a sensitivity of 93% and negative predictive value of 97%. As with transbronchial FNA, a moderate number of lymphocytes must be present to ensure the adequacy of a lymph node specimen and avoid a false-negative result.

Percutaneous FNA

The ease, rapidity of diagnosis, and minimal morbidity of percutaneous FNA make it an attractive alternative to surgical biopsy in the evaluation of the patient with a peripheral pulmonary mass. FNA is of greatest benefit to patients for whom it spares a more invasive surgical procedure. Surgical intervention, in fact, can be avoided in up to 50% of patients with clinically suspected lung cancer. There are some contraindications, however.

The most common complication of percutaneous FNA is a pneumothorax. A radiographically detectable pneumothorax occurs in 19% of patients ; only 4%, however, require intercostal drainage tubes. The risk of a pneumothorax increases with larger needle size and smaller target lesions and decreases if the needle path does not traverse aerated lung. Pulmonary hemorrhage occurs in 6% of patients, and transient hemoptysis occurs in 2%. Other complications are rare and include hemopericardium, hemothorax, air embolism, tumor seeding, and death.

Percutaneous FNA is usually performed by a radiologist using computed tomography or ultrasound guidance, with the latter best reserved for lesions that abut the pleura or chest wall. The needle gauge ranges from 18 to 25, and many different types of needle devices are available. Although many radiologists prefer 22-gauge Chiba needles, these require repuncture if more than one pass is needed. Another choice is the coaxial needle, with a large-bore outer needle serving as the guide for a small-bore inner needle. Once the outer needle is positioned, more than one aspiration can be performed using the inner needle.

It can be helpful if a cytotechnologist, cytopathologist, or both attend the FNA procedure to assist with specimen handling and assess its adequacy on-site. After smears are prepared, the needle is rinsed in a balanced electrolyte solution, Saccomanno’s fixative, 50% ethanol, or commercial preservative solution. The cellular suspension can be processed as a cytospin, thin-layer preparation, or cell block, and it can be apportioned for flow cytometric analysis if needed. Formalin-fixed cell blocks are ideal for histochemical and immunohistochemical stains. A decision regarding the need, if any, for special studies can be made by the cytotechnologist or cytopathologist after examination of the smears. Percutaneous FNA is a reliable and accurate way to diagnose many pulmonary neoplasms.

In a study of more than 13,000 FNA specimens from 436 institutions, the diagnostic sensitivity was 89% for the procedure itself and 99% for the pathologist’s interpretation. This difference indicates that most false-negative results are due to sampling error. The reliability of a negative FNA result is a matter of controversy, given that negative predictive values range from 50% to 91%. About 15% of false-positive diagnoses and 5% of false-negative diagnoses have a significant, permanent, or grave influence on patient outcome. For this reason, most investigators recommend a repeat aspiration or tissue biopsy when a specific benign diagnosis that accounts for the lesion cannot be made with certainty. The performance of small, cutting (“core”) biopsy is similar to that of FNA, but it is performed instead of FNA in some centers.

With regard to the management of patients with primary lung cancer, it is important to discriminate small cell from non-small cell carcinoma, and adenocarcinoma from squamous cell carcinoma. The distinction between small cell and non-small cell carcinoma is possible in more than 95% of cases and between adenocarcinoma and squamous cell carcinoma in 88% of cases diagnosed by FNA.

A variety of benign cells occasionally contaminate a percutaneous FNA. Such cells need to be recognized as contaminants and not misconstrued as lesional. In particular, mesothelial cells from the pleura are common, and in some cases they can be numerous ( Fig. 2.3 ). They resemble the cells of a well-differentiated adenocarcinoma (see “Adenocarcinoma”) but are identified as benign mesothelial cells by their relative flatness, cohesion, and the characteristic slit-like “windows” that separate the mesothelial cells from each other.

If the specimen consists only of one (or more) of these contaminants, it should be interpreted as insufficient for evaluation (nondiagnostic) rather than negative because there is no evidence that the lesion itself has been sampled.

Reactive Squamous Cell Changes

Benign squamous cells from the oral cavity often contaminate sputum and bronchial cytology specimens. Inflammatory conditions of the mouth caused by trauma, candidiasis, or pemphigus can exfoliate mildly atypical squamous cells with hyperkeratinization and nuclear degeneration, usually in small numbers. Such minimal changes should not be misinterpreted as squamous cell carcinoma. More marked (but still benign) squamous cell atypias occur adjacent to cavitary fungal infections and stomas, and with almost any injury to the lung (e.g., infarction, radiation, chemotherapy, sepsis, and diffuse alveolar damage) and might result in a false-positive interpretation of squamous cell carcinoma. Another, uncommon source of false-positive results is malignant cells from head and neck cancers that contaminate sputum and bronchial specimens.

Reactive Bronchial Cell Changes

Benign, reactive bronchial cell changes occur in response to noxious stimuli such as radiation, chemotherapy, and severe inflammation. Under such conditions, ciliated columnar cells can increase their nuclear area many times over, with multinucleation, coarsely textured chromatin and large nucleoli ( Fig. 2.4A ). Large clusters of bronchial cells known as Creola bodies (named after the first patient in whom they were recognized) are commonly seen in chronic airway diseases like asthma ( Fig. 2.4B ). Bronchial cells may also undergo squamous metaplasia under conditions of chronic injury (e.g., aspiration, smoking), demonstrating an immature squamous phenotype with mild nucleomegaly and nuclear hyperchromasia.

Bronchial cells with markedly reactive changes due to thoracic irradiation or cytotoxic chemotherapy mimic carcinoma ( Fig. 2.5 ). Malignancy can be excluded if the atypical cells have cilia or demonstrate a spectrum of changes (from benign to markedly atypical) rather than the two distinct cell populations typical of a malignant sample obtained bronchoscopically. Note that in sputum and FNA specimens, a helpful dual cell population (malignant cells and bronchial cells) is usually not apparent.

Bronchial Reserve Cell Hyperplasia

As the surface epithelium of the respiratory tract is shed during lung injury, reserve cells proliferate and are seen in bronchial washings and brushings ( Fig. 2.6 ).

Reserve cell hyperplasia (RCH) is a mimic of small cell carcinoma but is usually easily distinguished from it. Although they can be abundant and may appear as a “second population,” the cells of RCH show greater cohesiveness than small cell carcinoma, are typically smaller (approximately the size of a red blood cell), and show no mitoses or necrosis.

Repair

Repair represents re-epithelialization of an ulcer created by trauma, radiation, burns, pulmonary infarction, or infection. Typical (and atypical) repair of the respiratory tract is very similar to that seen in the cervix and gastrointestinal tract.

Reparative epithelium is most commonly seen in tracheobronchial brushings and washings. The differential diagnosis of repair includes malignancy: the non-small cell lung cancers, metastases, and other, less common tumors. Malignant cells are usually less cohesive and less orderly than reparative epithelium, and they are usually more numerous. Correlation with clinical history can be helpful; a conservative approach is recommended if the findings are not conclusive and there is a history of mucosal trauma or other lung injury.

Type II Pneumocyte Hyperplasia

Because type II pneumocytes function as alveolar reserve cells, they proliferate after lung injury.

When floridly hyperplastic, as in diffuse alveolar damage, the cells of type II pneumocyte hyperplasia resemble those of adenocarcinoma ( Fig. 2.7 ).

The only clue to avoiding an incorrect diagnosis of malignancy in a patient with type II pneumocyte hyperplasia may be the clinical history of respiratory distress and diffuse infiltrates in the context of an inciting stimulus such as upper respiratory infection, recent radiation, or recent administration of therapeutic drugs. Thus, in an acutely ill patient with diffuse pulmonary infiltrates, markedly atypical cells should be interpreted cautiously, particularly in the setting of an acute inflammatory background. Sequential respiratory specimens can be helpful because hyperplastic pneumocytes are not present in BAL specimens more than 1 month after the onset of acute lung injury.

Viral Infections

Herpes Simplex

Herpes simplex virus (HSV) pharyngitis, laryngotracheitis, and pneumonia most commonly affect immunocompromised patients and neonates, and HSV-1 is the most common serotype to involve the respiratory tract. Ulcerative/necrotizing infections can involve the pharynx, larynx, tracheobronchial tree, or pulmonary parenchyma, and cytopathic changes are identical to those seen in other sites: multinucleation, nuclear molding, chromatin margination, and large nuclear (Cowdry A) inclusions ( Table 2.1 ). The cytopathic changes of HSV are identical to those of herpes zoster. If the cytomorphologic changes are equivocal, the diagnosis can be confirmed by viral culture, immunohistochemistry, or in situ hybridization.

Table 2.1

Pulmonary Viral Infections

Virus		Nuclear features	Inclusions	Other changes
Herpes simplex and herpes zoster		Multinucleation; molding; peripheral margination of chromatin	Eosinophilic, intranuclear (Cowdry type A)	—
Cytomegalovirus		Enlargement	Large intranuclear, basophilic, with halo; small cytoplasmic, basophilic	Cytoplasmic enlargement
Measles virus		Multinucleation	Eosinophilic, intranuclear; multiple eosinophilic intracytoplasmic	Giant cells
Respiratory syncytial virus		Multinucleation	Cytoplasmic, basophilic, with halo	Giant cells; necrosis
Adenovirus		Smudged appearance as a result of large inclusion that fills entire nucleus	Large intranuclear, basophilic	Ciliocytophthoria

Bacterial Pneumonias

Bacterial pneumonias are caused by a large number of bacteria, but most are characterized by a neutrophilic exudate. Common organisms include Streptococcus pneumoniae (pneumococcus), other Streptococci, Staphylococcus aureus , Haemophilus influenzae , Klebsiella pneumonia , Pseudomonas sp., Legionella sp., Nocardia sp., Actinomyces sp., and some anaerobic bacteria. Many but not all bacteria can be seen with routine stains as well as with the Gram stain. Cytologic examination is not usually used for the diagnosis of a bacterial pneumonia, which is typically established by correlating clinical findings with microbiologic studies.

Bacterial pneumonias often have a characteristic lobar or lobular distribution, but some pneumonias appear as round and circumscribed masses on imaging studies and thus mimic the appearance of a malignancy. In such cases, a cytologic specimen might be obtained because the working clinical diagnosis is a suspected malignancy.

Several bacteria deserve special mention. Actinomyces species are a common inhabitant of the tonsillar area and thus a common contaminant of sputum and bronchial specimens (but not FNAs). Infection by Actinomyces , however, is uncommon. Pulmonary infection occurs by aspiration of oral contents or by direct extension from subdiaphragmatic abscesses. Actinomycosis is usually a chronic infection that may result in sinus tracts. The bacteria aggregate into grossly visible sulfur granules (so-called because they look yellow on gross examination) and evoke a brisk neutrophilic response. When they appear in cytologic specimens as just an oral contaminant, Actinomyces bacteria are large blue “cotton balls” often associated with squamous cells, with no neutrophilic infiltrate, similar to what is occasionally encountered in Pap specimens (see Fig.1.23 ). A true thoracopulmonary actinomycosis should be considered if the bacteria are associated with abundant neutrophils.

Nocardia are aerobic, filamentous bacteria that inhabit the soil and are acquired by inhalation. Nocardia asteroides accounts for 80% of Nocardial infections. Nocardia infection is most common in the setting of immunosuppression, and most patients have a subacute presentation. Cavitary nodules are seen in one-third of patients. The organisms are found among abundant neutrophils. They are thin, filamentous, and beaded, with such extensive, predominantly right-angle branching that they resemble Chinese characters. They are Gram-positive ( Fig. 2.14A ) and stain with silver stains but are only weakly acid-fast and thus require modified acid-fast stains like the Fite-Faraco for visualization. ( Fig. 2.14B ). The diagnosis is established by culture of a biopsy or BAL fluid.

Legionella pneumonia is caused by the aerobic Gram-negative bacteria Legionella sp., of which the most common is Legionella pneumophila . The organisms are seen well with silver stains like the Steiner, Warthin-Starry, or Dieterle stains. A specific identification of L. pneumophila can be made in BAL samples using immunohistochemical or immunofluorescent methods.

Tuberculosis

Infection by M. tuberculosis commonly results in granulomatous inflammation ( Fig. 2.15 ). Cytologic specimens contain aggregates of epithelioid histiocytes, lymphocytes, and Langhans’ giant cells. Necrosis may or may not be evident. Granulomas by themselves, however, are a nonspecific finding and can be seen in other conditions like fungal infections and sarcoidosis. A definitive diagnosis of tuberculosis rests on identifying the organisms with the help of histochemical stains (Ziehl-Neelsen or auramine-rhodamine), immunohistochemical stains, or by microbiologic culture. Cell block preparations are particularly useful for special stains, but rarely are more than one or just a few organisms identified (by contrast, infection by Mycobacterium avium intracellulare , as seen in immunocompromised patients, often yields innumerable acid fast organisms). A sensitive and specific assay called the Mycobacterium tuberculosis direct test is also available for the detection of M. tuberculosis and can be applied to respiratory specimens like sputum. The assay amplifies M. tuberculosis ribosomal RNA by the polymerase chain reaction and has been approved by the Food and Drug Administration (FDA) for use in both Mycobacterium smear-positive and smear-negative respiratory samples. Other highly sensitive polymerase chain reaction (PCR)–based assays and matrix-assisted laser desorption ionization-time of flight mass spectrometry for the distinction between tuberculous and nontuberculous mycobacteria are used increasingly because nontuberculous mycobacterial infections have become more frequent in developed countries.

In countries where M. tuberculosis is prevalent, the yield of acid-fast bacteria among all clinically suspicious lung masses can be very high. In immunocompromised patients with tuberculosis, there may be an abundance of acid-fast organisms but few well-formed granulomas. If Romanowsky-type stains are used in such cases, the abundant acid-fast organisms can be identified as negative images.

Pulmonary Fungal Infections

Pulmonary fungal infections are readily diagnosed by cytology, particularly in transthoracic FNA, and should be suspected whenever there is granulomatous inflammation, necrosis, or both. Cell block material can be used for silver or periodic acid–Schiff stains. Many fungi have a characteristic microscopic appearance that enables a rapid, specific diagnosis ( Table 2.2 ).

Table 2.2

Pulmonary Fungal Infections

Disease	Organism	Demographics	Morphology	Size	Immunocompetent response	Appearance
Cryptococcosis	Cryptococcus neoformans	Worldwide	Yeast; variably sized; narrow-based budding; mucin capsule (immunocompetent); refractile center	4–15 μm (diameter)	Granulomatous
Histoplasmosis	Histoplasma capsulatum	Americas, especially Ohio and Mississippi river valleys	Small budding yeast; intracellular	1–5 μm (diameter)	Granulomatous
Blastomycosis	Blastomyces dermatitidis	North America	Broad-based budding yeast; thick cell wall	8–20 μm (diameter)	Neutrophilic/granulomatous
Coccidioidomycosis	Coccidioides immitis	North American deserts	Spherules; endospores	15–60 μm; 1–2 μm (diameter)	Granulomatous
Paracoccidioidomycosis	Paracoccidioides braziliensis	Central and South America	Yeast with multiple budding (“mariner’s wheel”)	4–40 μm (diameter)	Neutrophilic/granulomatous
Sporotrichosis	Sporothrix schenckii	Worldwide	Intracellular ovoid yeast with slight halo	2–4 μm (diameter)	Neutrophilic/granulomatous
Aspergillosis	Aspergillus fumigatus; A. flavus	Worldwide	Hyphae septate; 45-degree angle branching; occasional fruiting bodies in cavities	Hyphae 10–30 μm (width)	Tissue/vascular invasion; colonization; fungus balls
Phycomycosis (zygomycosis)	Mucor; Absidia; Rhizopus; Cunninghamella	Worldwide	Hyphae variably sized, ribbonlike nonseptate; 90-degree angle branching	Hyphae 10–30 μm (width)	Tissue/vascular invasion
Candidiasis	Candida albicans; C. tropicalis; C. glabrata	Worldwide	Budding yeast forming pseudohyphae (“sausage links”)	2–10 μm (width)	Respiratory space colonization; tissue invasion