CHAPTER 8 Lung, chest wall and pleura
Clinical aspects
Lung
Although sporadic reports of the diagnosis of lung carcinoma by fine needle aspiration cytology (FNAC) appeared as early as 1886,1 the impetus for widespread use of the technique only arose with the development of image intensifiers and television viewing, allowing localization of small parenchymal lesions.2 Recognition of the high accuracy rate of fine needle biopsy (FNB)2 and simpler methods to treat pneumothorax combined to bring this diagnostic tool within the reach of most hospital radiologists and pathologists. Infections and some diffuse benign processes may be proven by FNAC, but the main indication remains the diagnosis of localized intrathoracic lesions suspected of being malignant, particularly when less invasive investigations prove to be negative.3
The role of FNB has shifted from the diagnosis of malignancy in inoperable patients and the confirmation of metastases, to its use as a first-line diagnostic procedure on which management decisions are based,4,5 with a shift away from non-invasive methods such as sputum cytology towards FNAC.6,7 All intrathoracic sites, including mediastinum and deep hilar lung lesions are safely sampled using fine (less than 20 gauge) needles with CT8 and ultrasound guidance.9 Lung function and radiological studies can help predict the risk of pneumothorax.10–12 Transbronchial or transtracheal FNB13–19 permits sampling of submucosal masses such as carcinoid tumors20 and mediastinal lesions, enhances the role of FNAC in the staging of lung cancers13,15,16,21 and may be used for the diagnosis of pulmonary infection22 and cysts.23 The more recent addition of ultrasound guided transbronchial and transoesophageal FNB has improved accuracy of targeting perihilar submucosal and mediastinal masses.24 The presence of the pathologist at the time of the procedure leads to a reduction in the number of needle passes, may decrease the pneumothorax rate,25 and increase overall sensitivity and accuracy of tumor typing.15,26 FNAC in immunocompromised individuals has gained greater acceptance, especially for the diagnosis of infectious diseases.27,28 An important role has also been found in paediatric practice8 (see Chapter 17). With greater expectations by clinicians regarding tumor typing, demonstration of high rates of cytohistological correlation for the common types of tumors has been achieved.29–32
The diagnostic utility of FNB for difficult diagnostic lesions (e.g. neuroendocrine and pleural tumors)33–44 is now widely accepted, particularly when allied with cell block preparations,31 and the use of ancillary studies.45–47 The range of cytological findings in less common tumors such as sarcomas,48,49 lymphoid lesions50,51 and metastatic tumors,52 as well as in second primary tumors,52,53 along with the most likely mimics of malignancy, have been well described.54–57 The more recent requirements for molecular assessment of gene mutations relevant to targeted therapy in some lung cancers can be readily achieved by FNB with adequate cell block preparations.58
The place of FNAC in the investigative sequence
Fine needle aspiration cytology complements other diagnostic methods.3 It is often the first choice in lesions which are located in the mediastinum, pulmonary apex, medial upper lobe or peripheral lung. Fiberoptic bronchoscopy (FOB) with brushings and biopsy is usually effective in diagnosing centrally placed pulmonary lesions, while FNB has particular value in those cases in which FOB is non-diagnostic.3
Endoscopic bronchial ultrasound (EBUS) localisation with FNB is now playing an increasing role in the sampling of central lesions, staging of lung carcinoma and the diagnosis of sarcoid and lymphoma.59–63
There is value in rendering a preoperative diagnosis, as thoracotomy is generally contraindicated for small cell carcinoma, most lymphomas, most metastatic tumors and most infections. A specific diagnosis of benign lesions such as tuberculosis or pulmonary hamartoma helps avoid unnecessary surgical intervention.64 Additionally, a diagnosis of malignancy in patients at high operative risk allows surgeons to operate without fear that surgery may be unnecessary. FNAC is a cost-effective diagnostic method that can lead to shorter hospitalization, and a reduction in the number of diagnostic thoracotomies.65
FNB may be the only way of providing a diagnosis in inoperable patients. An unequivocal diagnosis of carcinoma and distinction between squamous cell carcinoma, small cell carcinoma and adenocarcinoma provides the clinician with sufficient information to select appropriate therapy. Raab et al. reported that FNB coupled with immunohistochemistry is both sensitive and cost effective for subclassifying malignancies in patients with a prior history of extrapulmonary cancers, when compared to bronchoscopy and thoracoscopy.66 FNB can occasionally be useful intraoperatively in place of frozen section when the use of incisional or large needle biopsy is contraindicated.67
Accuracy of diagnosis
The sensitivity of FNB in the diagnosis of malignant pulmonary neoplasms varies according to both the size and the site of the mass,68 as well as the experience of the radiologist69 and the pathologist. In experienced hands it now reaches over 90% in both non-academic and academic centers.70,71 A non-specific negative result on FNB does not exclude malignancy.3,72 Therefore, repeat aspiration, careful clinical follow-up or additional diagnostic procedures may be necessary for definitive diagnosis. The rate of false-positive diagnoses is usually less than 0.5%.71 The few false-positive diagnoses of malignancy are often due to reactive bronchiolar epithelial proliferation,73 squamous metaplasia,74 reactive mesothelial cell proliferation,29 or associated with lesions such as tuberculosis, chondroid hamartomas,75 granulomatous processes,73 other infections,76 or pulmonary infarcts. Occasionally, in samples obtained by EBUS, it can be difficult to determine whether the sample is derived from a central lung lesion or nodal deposit, a potential cause of false-positive diagnosis of nodal metastases. Tao advises caution in assuming that a lack of clinical or histological correlation is a result of false-positive cytological diagnosis.77 He introduced the term ‘false false positives’ for cases where longer clinical follow-up period or closer histological examination confirms the cytological diagnosis. Well-differentiated adenocarcinomas with minimal cytological atypia and small tumors within larger areas of radiological opacity are prone to this error.
Experienced cytopathologists can reliably identify the histological type in up to 90% of primary lung tumors29,31,32,70,78,79 and approximately 80% of their predictions of tumor type are correct;2,30,70 these rates are similar to those achieved with bronchial biopsy.31,80,81 High levels of cytohistological correlation relate to the combined use of smears and cell blocks, which allow better assessment of tissue architecture.31 Rapid assessment of aspirated material in the radiology theater enhances accuracy of diagnosis by allowing appropriate triage of material for ancillary techniques. Most cases of small cell carcinoma,30,35,79,82,83 well-differentiated adenocarcinoma,84–87 and squamous cell carcinoma88 are usually identifiable. However, poorly differentiated tumors of glandular or squamous origin and large cell carcinoma are more difficult to separate.88 More studies of the range of cytological findings in neuroendocrine carcinomas, particularly large cell neuroendocrine tumors, are needed.36–38 These may be difficult to distinguish from small cell carcinomas or other large cell carcinomas, although in the view of some authorities this distinction may be artificial or arbitrary at times.89 Pulmonary hamartomas64,75,90–94 can be accurately diagnosed in the majority of cases. Other benign neoplasms are difficult to type because their cohesiveness prevents aspiration of adequate diagnostic material. Thin core needle biopsy may add to diagnostic accuracy.95–97 Radiologists are increasingly prepared to perform core biopsies on intraparenchymal lesions96,97 and this can be useful in some settings, especially where there are unusual features by FNAC. It is worth noting that a core biopsy will not necessarily provide a diagnosis where FNB has failed, particularly if the FNB sample is extensively necrotic. Cytological evidence of infection including tuberculosis may be gained in up to 80% of cases.98–103 FNB can be used with a high degree of accuracy to diagnose bacterial, fungal or pneumocystis infection in immunosuppressed patients.104
Complications
Fatalities have occurred with 20-gauge or larger needles or in patients who are terminally ill or those with severe underlying pulmonary disease or infection.105 The most common cause of death is hemorrhage, although there are rare cases of air embolism, tension pneumothorax and cardiac tamponade.106–110 To our knowledge, no deaths have occurred in our experience of over 5000 cases.
The rate of pneumothorax varies from 6% to 57% and 1.5–20% require intercostal catheterization.105,111 Emphysema, deeply situated lesions, multiple punctures, older patients and inexperienced operators all contribute to a higher rate of pneumothorax.10,11,106 Examination of aspirated material in the radiology theater and the use of coaxial needles allows the total number of needle passes to be reduced, thereby reducing the rate of pneumothorax.25 A small hemorrhage into the surrounding lung occurs in up to 10% of cases without clinical complications. Examples of neoplastic implantation along a thoracic needle track have been reported,112–115 and one case of skin implantation with fungal organisms.116 Bacteremia may occasionally be induced.117
Contraindications
Patients who do not have a cough reflex, who are unconscious or who cough intractably should not be biopsied. Arteriovenous malformation or aneurysm is usually excluded before attempting FNB, but the risk of bleeding from large hilar vessels is minimal when fine needles are used. There is an increased risk of bleeding in patients with pulmonary hypertension or bleeding disorders and in those undergoing anticoagulant therapy. These patients should be biopsied only in exceptional circumstances, for example in immunosuppressed patients where a positive diagnosis of an infection may lead to curative therapy.104,118 Most authors do not recommend FNB of suspected hydatid cysts due to the possibility of an anaphylactic reaction to leaking cyst fluid or implantation of germinal epithelium. However, when aspiration of unsuspected cysts occurs, complications are very few.119,120 Concurrent bilateral biopsies are not recommended.
Technical considerations
Obtaining material
Most radiologists use needles of less than 20 gauge, with a stylet to minimize contamination with tissue from the needle track. Needles of 22–23 gauge provide satisfactory material for cell blocks. Finer needles lessen tissue damage, although their flexibility may reduce the chance of successful puncture in deep lesions. Automated core biopsies are safe for pleural-based or chest wall lesions,121 and their use in parenchymal lesions is increasing.96,97
The puncture site is marked on the skin and local anesthetic introduced down to the pleura. A guide needle makes aiming of the aspiration needles easier, reduces the risk of bending of flexible needles and prevents contamination of the aspirate with chest wall tissues.122 Non-aspiration techniques often yield adequate material and may be particularly useful in hemorrhagic lesions (e.g. metastatic renal cell carcinoma).123,124
CT guidance is generally used although ultrasound may be useful for apical, pleural or chest wall tumors.125–129 The use of coaxial needles means that repeated re-insertion of the needle is not necessary. Some lesions have distinctive characteristics when aspirated, e.g. pulmonary hamartomas have a rubbery resistance and a tendancy to ‘move away’ from the needle; aspiration of neural tumors may cause intense pain. Adequacy of the aspirated material should be assessed in the radiology theater after rapid staining. The presence of the pathologist at the aspiration optimizes appropriate triage of material. Preliminary assessment in the radiology suite may also identify an unusual lesion where a core sample may be helpful. If the material is heavily bloodstained, collection onto a slide or watch glass (see Chapter 2) enables small clumps of tumor to be recognized as bright translucent fragments against the red background. To reduce blood contamination, these aggregates are selected and smeared onto a separate slide. Carnoys’ fixative or acid alcohol are useful for lysing blood. Material is taken for ancillary studies such as flow cytometry, ultrastructural study or submitted for culture depending on the preliminary assessment of the aspirate. Our approach is to generate an adequate fixed smear which is assessed by rapid H&E stain; subsequent material is then triaged according to the appearances of the first aspirate. Other laboratories have found air-dried MGG-stained material to be useful for initial assessment. More smears may need to be assessed if the first pass does not contain diagnostic material or is not consistent with the clinical and radiological findings. However, the days of producing numerous highly cellular smears with no cell block should be over! It should also be remembered that many lung carcinomas will not be resected; the material obtained by FNB may be the only tissue available for subsequent molecular studies, which can determine suitability for a particular treatment, further emphasizing the need to obtain an adequate cell block in all cases.
Use of ancillary techniques45,46
We rely heavily on ancillary techniques in FNB of the lung and pleura. Ancillary testing is useful in over 50% of intrathoracic aspirations.45 O’Reilly et al. used immunocytochemistry in 14.5% of cases, EM in 8%, microbiologic studies in 12%, mucin staining in 21% and cell blocks in 22%. They found the greatest value was in classifying poorly differentiated neoplasms, confirming the diagnosis of well-differentiated adenocarcinoma, identifying neuroendocrine differentiation and establishing primary sites for suspected metastases.45 Saleh and Masood found ancillary tests contributed to diagnosis in over 60% of cases.46 Immunohistochemistry was most often used, but electron microscopy was often helpful. Immunohistochemistry has become the mainstay of cytological diagnosis, especially since the availability of thyroid transcription factor (TTF-1),130–132 and differential cytokeratin staining.131,133–135 Currently, EM is seldom used. Flow cytometry has revolutionized the diagnosis of lymphoid lesions.136–138 PCR-based molecular analysis for gene rearrangements for lymphoma diagnosis and FISH studies for lymphoma (see Chapter 5) and soft tissue tumor diagnosis (see Chapter 15) can be applied to FNB material. PCR studies to identify the presence of gene mutations, e.g. EGFR mutations which are relevant to the selection of particular therapies, can be performed on cell blocks.58 One can anticipate other techniques such as proteomics and microarray technology being useful.139,140
Immunocytochemistry
The full range of immunohistochemical markers can be used in cell block preparations. Broad-spectrum antibodies to keratins, S-100 and leukocyte common antigen (CD45) can separate most poorly differentiated carcinomas from melanomas and lymphomas. Calretinin has been the most useful positive marker of mesothelioma in our hands, but also stains some poorly differentiated lung carcinomas and squamous carcinomas; it is most useful in pleural effusion cytology. Neuroendocrine markers such as synaptophysin, chromogranin and CD56 are useful in diagnosing various neuroendocrine tumors. Markers for prostate-specific antigen and thyroglobulin are essential in selected cases. Estrogen receptor (ER) staining helps in identifying metastases of breast carcinoma. Caution is required with particular clones of ER.141 SP1 clone has a significantly higher rate for positive estrogen receptor expression in pulmonary adenocarcinomas when compared with either 1D5 or 6F11 clones. Caution is required in the use of SP1 antibody alone in distinguishing metastatic breast carcinoma from primary pulmonary adenocarcinoma. Differential cytokeratin staining for CK7 and CK20 can help separate primary adenocarcinomas of the lung (CK7 positive, CK20 negative) from metastatic colonic carcinoma (CK7 negative, CK20 positive).134,135 CDX2 is an additional positive marker of colonic malignancy. Rarely, primary lung adenocarcinomas of so-called enteric-type are CK20 positive and may also stain for CDX2; they usually retain their CK7 positivity.142 Squamous cell carcinomas and small cell carcinomas are usually negative for both CK7 and CK20, while squamous cell carcinomas are usually positive for CK5/6 and p63.143 TTF-1 is relatively specific for lung and thyroid neoplasms130,131 and is probably the single most useful marker for confirming primary lung adenocarcinoma, although perhaps inevitably there are now reports of cases of various metastatic carcinomas being TTF-1 positive, e.g. ovary, cervix, endometrium, colorectal.144–146 Antibodies to surfactant protein may also be helpful, but negative results in metastatic tumors have not been validated to the same extent as TTF-1. TTF-1 is said to be positive in 90% of small cell carcinomas,147 although we have seen several cases of TTF1-negative small cell carcinoma.
Electron microscopy (EM)
The role of EM (see Chapter 2) in the diagnosis of pulmonary neoplasms is declining, but may still be of value in recognizing neuroendocrine tumors, in the diagnosis of occasional cases of melanoma and mesothelioma and some carcinomas, including metastases, when immunocytochemistry has not provided a diagnosis.45–47,148,149 Demonstration of surfactant granules supports the diagnosis of primary bronchogenic carcinoma. Recent studies have further emphasized the heterogeneity of cell type in lung adenocarcinomas. Tumors may show features of nonciliated bronchiolar cell, alveolar cell, mucus-secreting cells or even mixed features of squamous or neuroendocrine type.150
Molecular testing151–159
There is a trend toward targeted therapy in oncological practice, usually based on the presence of particular gene mutations. At present, the main molecular testing in lung carcinoma is the assessment of EGFR gene mutations; these mutations are usually found in adenocarcinomas and are limited to exons 18–21, with more than 80% of mutations being short deletions in exon 19 or single point mutations in exon 21.151,152 This testing is usually performed only when there is consideration for treatment with small molecule inhibitors of EGFR. In addition, mutations in KRAS are associated with lack of benefit from EGFR inhibitors. Mutations in the EGFR and KRAS genes are almost never seen together in the same tumor.153 EGFR and KRAS status can be assessed by DNA extraction from cell block samples in which there is adequate tumor representation. PCR and sequencing studies are then performed.58 When there is insufficient material in a cell block sample, material can be scraped from slides.154 Bronchioloalveolar carcinoma (BAC), now defined as an in situ adenocarcinoma, is composed of two distinct subgroups – mucinous and non-mucinous. The more frequent non-mucinous BAC which predominates in smokers and often presents as a ground-glass opacity is thought to evolve from the terminal respiratory unit cells and frequently harbors EGFR polysomy/mutations which are considered to be a critical event in the pathogenesis of non-mucinous BAC tumors. The less frequent mucinous BAC, thought to be derived from metaplasia of bronchiolar epithelia, presents more frequently as a pneumonic-type infiltrate, rarely demonstrates EGFR mutations, and more frequently harbors and is driven by a KRAS mutation.153,155
There are various other molecular markers, e.g. ERCC1, RRM1, that appear to be of increasing clinical relevance, in particular to predict tumor response to chemotherapeutic agents.156,157 The recent description of an ALK–ELM translocation which can be detected by FISH and immunohistochemistry, and which is present in a small subset of adenocarcinomas, which will respond dramatically to a particular drug therapy,158,159 highlights the need to ensure that adequate samples are obtained for cell blocks in order to perform additional studies as required.
Clinical requirements in modern practice143,160–162
There are increasing expectations from oncologists to distinguish between adenocarcinoma and squamous cell carcinoma in order to optimize treatment regimens, e.g. bevacizumab is associated with a high risk of bleeding in squamous cell carcinoma and is approved only for adenocarcinoma.160 This has led to pressure to reduce, if not eliminate, the diagnosis of ‘non-small cell carcinoma’. An immunohistochemical panel including CK5/6, p63, CK7 and TTF-1 in poorly differentiated carcinomas, may help to suggest either possible squamous (CK5/6, p63 positive) or glandular (CK7/TTF-1 positive) differentiation.143 Pardo et al.161 reduced their diagnosis of large cell carcinoma by 90% in a tissue microarray study, with the use of immunohistochemical panels and Khayyata et al.143 achieved refinement of diagnosis in 65% of cases of NSCLC to either SCC or adenocarcinoma. While the distinction between adenocarcinoma and squamous cell carcinoma can usually be achieved morphologically in well-differentiated tumors, the addition of immunohistochemical panels may increase diagnostic accuracy in poorly differentiated carcinomas.162 However, limited sampling, anaplastic tumors, tumor heterogeneity and overlapping immunohistochemical profiles can limit the ability to subclassify some lung carcinomas, and oncologists need to be aware of these issues. It is also worth noting that in some studies there may not have been any ‘gold standard’ to confirm the immunohistochemical diagnosis. Even in a well-differentiated adenocarcinoma it is still incumbent on the pathologist to ensure that there is adequate cell block material for molecular studies (see above). As new discoveries come to light there may be the need to return to archival material to perform new tests. Once a smear with adequate material has been obtained, further aspirates should be submitted for cell block preparation to ensure sufficient material is available for any additional studies that may be required.
Chest wall and pleura
Chest wall is included here to emphasize that the exact site of origin of lesions in this region may be difficult to ascertain and that it is conceptually useful to consider this site as a ‘body compartment’. Lesions arising in rib, chest wall soft tissues, pleura or even breast may extend to underlying lung parenchyma and may be confused with primary lung tumors.163 Contaminating bone marrow from rib may give rise to diagnostic dilemmas. In some cases, myeloma can be mistaken for an undifferentiated lung carcinoma. There is sometimes difficulty in distinguishing between mesothelioma, anaplastic carcinoma, melanoma, sarcomas of the chest wall and metastases. In our own practice, a chondrosarcoma arising in the rib caused difficulty in diagnosis because the site of origin was not appreciated. A malignant mesothelioma growing along a chest drain site was considered clinically to be a primary breast lesion until FNB was performed. Primary soft tissue lesions of this region are described in Chapter 15.
Cytological findings
The satisfactory smear
A specific benign diagnosis by FNB is generally restricted to the identification of infection, chondroid hamartomas, cysts and a few other disease processes. Non-specific negative findings do not exclude malignancy,3 and can only be used for patient management in conjunction with clinical and radiological findings. A precise statistical assessment of accuracy rates for a radiology department and a laboratory should be known. In a sense, the only satisfactory sample is one which contains material permitting a specific diagnosis.
Normal structures (Figs 8.1–8.3)
Bronchiolar epithelium or nonciliated epithelium is seen commonly as sheets of various size (Fig. 8.1). They usually have irregular edges and the component cells display variable cell separation. The nuclei are generally small and there is a low nuclear:cytoplasmic (N:C) ratio. Sometimes the nuclear outlines are slightly irregular and small intranuclear cytoplasmic inclusions are observed. Occasionally, atypia of bronchiolar epithelium may be quite pronounced, for example in reactive or inflammatory processes; however, the number of atypical cells is usually small. A diagnosis of malignancy should generally not be made on the basis of small numbers of cells.
Mesothelium can be easily distinguished from bronchiolar epithelium. It is seen as various-sized, flat, monolayered sheets; there is usually more cell separation than in bronchiolar epithelium, which sometimes gives a sponge-like appearance. The cells appear to be joined by intercellular bridges similar to those of squamous epithelium in histological sections. (Fig. 8.2). Reactive mesothelium can appear atypical and be misinterpreted as neoplastic.
Normal lung yields a population of macrophages widely dispersed over the slide (Fig. 8.3); these contain small particles of brown or black particulate matter, some of which is inhaled dust, especially in smokers. Many hemosiderin-laden macrophages usually imply tissue or blood breakdown near the lesion and may add to a suspicion of pulmonary infarction should the clinical background be appropriate.
Aspirations near the diaphragm can contain liver or splenic tissue and cause concern unless the possibility of inadvertent puncture of abdominal organs is appreciated. Splenic tissue has also been identified in FNB samples from ‘thoracic splenosis’ following traumatic rupture of the spleen; these are usually pleural based but can occur within the lung.164,165
Megakaryocytes are occasionally seen in aspirates of lung166 or more often from rib marrow.
Inflammatory/non-neoplastic processes167
Acute inflammatory material
Common findings
These findings are not specific unless bacteria are evident, but they can suggest an acute infection. It is worth repeating the aspiration, and sending all the material obtained for aerobic and anaerobic culture. Associated vegetable material may suggest aspiration as a cause.168
When the center of a necrotic or cavitating carcinoma is aspirated, particularly those of squamous type, acute inflammatory material may result (the so-called carcinomatous lung abscess).169 Macroscopically, this resembles pus and microscopically only few squamous cells may be present. Small squamous cells with pyknotic nuclei may be difficult to distinguish from degenerate macrophages.
A high polymorph content is usually present in the inflammatory reaction to fungi. A search for a granulomatous component and fungal organisms is mandatory if acute inflammatory material is obtained. With infection by filamentous organisms, e.g. Nocardia, the organisms may be identified in the center of distinctive, cohesive neutrophil clusters or rosettes after silver impregnation; this clustering should be a clue to search for such an infection (Fig. 8.4). Similar appearances may also be be seen in streptococcal infection.
Granulomatous inflammation (Figs 8.5–8.7)98–103,170–176
Criteria for diagnosis
One can make a confident diagnosis of a ‘granulomatous process’ on aspirated material; however, a more important aspect of the aspiration is to obtain an adequate sample for staining and culture, particularly to distinguish between ‘typical’ and ‘atypical’ mycobacteria101 and for typing fungal organisms. It is best to have a large sample for this purpose and we have not found much value in washings of the needle after a cytology smear has been made. Instead, we advocate sending the whole aspirate for culture. Culture for tuberculosis is more successful in cases with necrotic and inflammatory debris.99,102 PCR can provide a rapid diagnosis, aids in subtyping and can identify drug resistant strains.170,171
Specific diagnosis of most noninfectious granulomatous diseases is not possible, although there are reports of the cytological findings in sarcoidosis,172–174 Wegener’s granulomatosis175,176 and rheumatoid nodule.177
Epithelioid histiocytes are fairly cohesive and form granulomas which are often aspirated intact (Fig. 8.5). Epithelioid cells have an elongated or bean-shaped nucleus and abundant cytoplasm which is rather pale and indistinct both in Pap- and H & E-stained specimens (Fig. 8.5). The cytoplasmic density is higher in MGG-stained material. Multinucleated histiocytes can be seen but are usually sparse. They are mainly free of intracytoplasmic pigment or birefringent material, unlike the multinucleated histiocytes seen in non-specific reactions in pulmonary tissue. Caseous necrosis has a variable appearance. There may be an amorphous to granular background with little cell outline visible, but sometimes outlines of necrotic cells may be prominent (Fig. 8.6) and often the appearances are merely of nondescript debris, histiocytes and neutrophils. Dahlgren cites granular calcific material as a common accompaniment.98 Lymphocytes may be plentiful in granulomatous inflammation.
Bailey et al.99 diagnosed 28 of 34 cases of TB by either auramine rhodamine fluorescence or positive culture; acid-fast bacilli (AFB) were seen in only 38% of cases. In Rajwanshi’s102 and Das’s103 series AFB were identified in approximately half of the cases. Gong et al. found PCR to be about 80% sensitive compared to 40% for ZN staining in FNB material.171
Recent advances in mycobacterial genomics and human cellular immunology have resulted in two new blood tests that detect tuberculosis infection by measuring in vitro T-cell interferon-gamma release in response to two unique antigens that are highly specific for Mycobacterium tuberculosis, but absent from bacille Calmette-Guérin (BCG) vaccine and most non-tuberculous mycobacteria. These tests appear to be very useful in increasing specificity and sensitivity in the diagnosis of tuberculosis.178
A granulomatous reaction may develop at the edge of a carcinoma or other neoplasm, particularly squamous cell carcinoma forming keratin; this reaction may be foreign body giant cell in type and yield many multinucleated cells. Histiocytic giant cell reactions also occur in fungal infections, e.g. cryptococcosis101,179–181 or histoplasmosis,101 so the cytoplasm of any giant cells present should be closely examined. It should be remembered that tuberculosis may coexist with carcinoma.
Necrotic tumor may be homogeneous or granular and closely resemble caseous material. Conversely, we have seen cases of tuberculosis where ghost outlines of single cells were distributed across the smear and closely mimicked necrotic tumor (Fig. 8.6).
Acid-fast bacilli are more often seen in cytological material characterized by a mixture of neutrophils, histiocytes, mucoid or necrotic material than in those lesions with a prominent epithelioid cell component,99,102 though culture is positive in a similar percentage of cases with and without epithelioid cells.103 Whenever necrotic debris is seen, we restain smears with ZN stains. Maygarden described mycobacteria as negative images in a stained background in MGG material (Fig. 8.7).182 Silverman reported negative images in both Diff-Quik and Papanicolaou-stained material in a BAL sample from a patient receiving clofazimine treatment for an atypical mycobacterial infection. The reddish, refractile and polarizable drug-derived crystals can impart a pseudo-Gaucher-like appearance to the cells, simulating the negative images of an atypical mycobacterial infection.183
Other specific infections
In Aspergillus (Fig. 8.8) and phycomycete (Fig. 8.9) infections in the lung, fungal hyphae are easily visible in alcohol-fixed material.184,185 The characteristic appearance of acute-angled branching, parallel cell walls and septa in Aspergillus (Fig. 8.8) is shared by several other fungi and culture is necessary for accurate typing.186 Pseudoallescheria boydii, identified as a ‘clinically significant mycosis’,187 and Fusarium sp. may be morphologically indistinguishable from Aspergillus, and even Candida resembles this organism on occasion, by forming germ-tube structures with ‘septation’.184 Oxalate crystals are produced by several species of Aspergillus and, if identified in cytological material, should lead to a search for organisms.188,189
Cryptococci179–181190 (Fig. 8.10) have a varied morphology.181 Later sclerotic foci have a more granulomatous appearance and fewer organisms. In several of our cases of cryptococcosis, organisms were only seen within multinucleated cells. A case of a pulmonary inflammatory myofibroblastic tumor caused by Cryptococcus infection, presumed to be the gattii species, has been diagnosed by FNAC.191 Mucicarmine staining of the capsule is a specific and useful criterion for diagnosis and will help distinguish the atypical forms from other yeasts; Fontana-Masson staining of cell walls will also help distinguish poorly encapsulated forms from Histoplasma, Torulopsis or Candida.181 Blastomycosis characteristically has more broad-based budding forms.
In some fungal infections, particularly histoplasmosis, organisms are only seen in methenamine silver preparations, although the Pap stain is a superb technique for identifying most other organisms including the filamentous higher bacteria.192 Destained H&E or Pap-stained smears are as suitable as unstained material for special stains. Culture is necessary for the exact classification of the organism.
Multiple and unusual infections may be encountered in AIDS or post-transplantation.27,28,193,194 Pneumocystis is uncommonly diagnosed by FNB because of the efficacy of broncho-alveolar lavage in the diagnosis of diffuse pneumonic disease, but some cases may present with nodular or cavitary lesions in the lung. CMV infection,195 coccidioidomycosis,76,196,197 paragonamiasis,198 amebiasis, hydatid disease119,120,199 and microfilariasis200 have all been described in FNB samples.
In coccidioidomycosis,197 smears show abundant granular eosinophilic debris with a paucity of acute or chronic inflammation; granulomatous inflammation is seldom seen. Coccidioides immitis spherules range in size from 20 to 200 µm; fractured or crushed forms are often seen and some are calcified. Endospores are seen only within intact spherules.196,197 In some cases, the associated reactive cellular changes have been misinterpreted as neoplastic.76
According to McCorkell,120 clear colorless fluid is an indicator of hydatid cyst and should lead to a search for scolices, hooklets or laminated membrane (see Chapter 10). Diagnosis of dirofilariasis201 requires cell block preparations (Fig. 8.11).
Fig. 8.11 Dirofilariasis
Cross-section of parasites in a background of necrosis (cell block, H&E, MP).
Reparative and reactive change
Squamous metaplasia, epithelial repair reactions, reactive bronchiolar or mesothelial proliferations and reactive histiocytic or fibroblastic cells can all produce significant cellular atypia (Fig. 8.12).56,73,76 Diagnosis of malignancy in a background of inflammatory change should be made with extreme caution. In FNAC, a malignant diagnosis should also generally be based on highly cellular smears. In this way, the risk of a false-positive diagnosis in reparative or reactive processes can be minimized.56,167,202
Pneumoconioses
Birefringent silica and collagenous tissue (Fig. 8.13) or asbestos bodies may help confirm silicosis or exposure to asbestos.203 However, concomitant malignancy, tuberculosis or other infections may be the cause of the localized opacity in these patients.204
Pulmonary infarct
Silverman describes sheets of metaplastic squamous cells showing regenerative changes and histiocytes containing clumped refractile hemosiderin in a clinically unsuspected infarct; FNAC findings led to the diagnosis being suggested. 205
Other non-neoplastic lesions
Material aspirated from congenital bronchogenic cysts usually consists of mucin and bronchial-type epithelium unless infection supervenes. So-called ‘bronchoceles’, which represent recent-onset postinflammatory dilated mucin-filled bronchial structures, may yield similar findings, and correlation with clinical and sequential imaging findings is necessary for precise diagnosis. (Fig. 8.14).
Findings in nodular amyloidosis (Fig. 8.15),206–208 Wegener’s granulomatosis,175,176 extramedullary hemopoiesis209 and malacoplakia210 are described. We have seen one case of amyloidosis in consultation, in which amyloid was visible in both smears and cell block preparations. Miller et al. describe FNAC findings of abundant parenchymal material including thickened alveolar walls as being of help in diagnosing ‘rolled’ or ‘rounded’ atelectasis in company with consistent CT findings. This entity produces a radiologic appearance of a mass lesion involving pleura and lung, often in the inferior lobes in the posterior basal lateral vertebral area, and is sometimes associated with asbestos-induced pleural disease with underlying lung scarring.57,211
Common primary carcinomas
Squamous cell carcinoma (Figs 8.16 and 8.17)
Fig. 8.16 Squamous cell carcinoma
Dispersed keratinizing malignant cells and necrotic debris (Pap, HP).
Criteria for diagnosis
Well-differentiated squamous cell carcinoma with keratinizing cells and bizarre cell shapes is not difficult to recognise. The malignant cells are usually dispersed, especially when they are very well differentiated (Fig. 8.16); necrosis is a common accompaniment. Single keratinizing cells are the most reliable indicators of squamous differentiation; when keratinization is not obvious, a perinuclear halo within the cytoplasm and condensation of peripheral cytoplasm are helpful guides to squamous differentiation (Fig. 8.17). Non-keratinizing tumors are usually more cohesive and present as multilayered fragments. Their nuclei are usually spindle shaped or elongated with dense, irregularly distributed chromatin. There is often conspicuous variation in the degree of chromasia of nearby nuclei. Nucleoli vary in size and number, in contrast to adenocarcinoma where nucleolar morphology is more monotonous. Dense cytoplasm and well-defined cell borders are indicators of squamous differentiation, but some adenocarcinomas may have strikingly well-defined borders and very dense cytoplasm.
Misinterpreting the cytoplasmic eosinophilia of necrosis for keratinisation is a common source of error (see Fig. 8.30).30 Necrosis can also impart an angular hyperchromatic appearance to nuclei, similar to squamous carcinomas. This is less of a problem when Pap staining is used because orangiophilia distinguishes between necrosis and keratin better than H&E or MGG, although cells in heavily bloodstained smears may take on an eosinophilic hue simulating keratinization.
Fig. 8.30 Necrosis simulating squamous differentiation
Necrosis, eosinophilia and pyknosis in a metastatic deposit of colonic adenocarcinoma (H&E, HP).
There is an overlap between the appearances of poorly differentiated squamous cell carcinoma, large cell anaplastic carcinoma and poorly differentiated or solid adenocarcinoma in aspirated material. They all present as large, multilayered fragments corresponding to the solid growth of these carcinomas. Squamous cell carcinoma is overdiagnosed in some series because adenocarcinoma in aspirates may present as pavement-like sheets, mimicking the appearance that squamous cell carcinoma has in sections. The pavement-like appearance of squamous cell carcinoma results only from thin sectioning and does not occur in aspirated material, although it may be seen in sections of cell blocks. Palisading of cells, which may occur at the margins of lobules of nonkeratinizing squamous tumors, may give rise to a false impression of the margin of a gland in smears. There is also a tendency to overestimate the grade of tumor in aspirated material, perhaps because viable rather than maturing or exfoliating new growth is removed. Unusual variants of SCC such as the pseudovascular adenoid form may be easier to recognize as squamous in cytological rather than histological material.212
Small-celled forms of poorly differentiated squamous cell carcinoma, and those undergoing degeneration and necrosis, can be confused with small cell carcinoma. Attention to greater cohesion, larger cell size, presence of nucleoli, greater tendency to spindle at the periphery of clusters, denser cytoplasm and occasional keratinized cells should lead to the correct diagnosis. Strong positive p63 and CK 5/6 staining and absence of staining for TTF-1 would favor squamous rather than small cell carcinoma, although rare squamous carcinomas are TTF1 positive.143,213
Low-grade mucoepidermoid tumors have been described in FNB samples214 as well as basaloid squamous carcinoma.215
Metastatic squamous carcinoma from head and neck or from anogenital primary sites will often be positive for high-risk HPV DNA, whereas primary lung SCC is rarely HPV positive.216 PCR studies for HPV DNA on cell block material may be helpful when metastatic disease is being considered.
Adenocarcinoma, including BAC (Figs 8.18–8.24)
Fig. 8.21 Well-differentiated adenocarcinoma
Monolayered sheet of glandular cells with honeycombing and prominent mucin (Pap, HP).
Fig. 8.22 Well-differentiated adenocarcinoma
Mucinous with lepidic growth pattern. Tissue section (H&E, MP).
The cellular morphology of adenocarcinoma is similar to that described in brush material. Rosettes, acinar formations or cohesive cell clusters (Figs 8.18 and 8.19) represent anatomical structures removed from the tumor by the needle. The larger the gland formations, the less likely they are to be removed intact and, when only partly removed, deposit on the slide as flat sheets in a monolayer (Fig. 8.18): a useful indicator of glandular differentiation. Artifactual spaces are often seen in large tissue fragments and may be misinterpreted as acinar structures; however, where the spaces have an ‘anatomical’ rigidity, they may indicate glandular differentiation. Mucin secretion is difficult to identify without the aid of special stains, and vacuolation of the cytoplasm may occur as a result of degeneration or the presence of glycogen. In H&E- or Pap-stained material vacuoles with a central, inspissated, eosinophilic or orangiophilic center are very suggestive of mucin secretion and correspond to the intracellular lumina described ultrastructurally in adenocarcinomas. With MGG staining, mucin may be visible as magenta or purple material within the cytoplasm, either homogeneously or as red granules within a pale vacuole. Well-formed columnar cells or groups of palisaded cells may be a guide to glandular differentiation and terminal plates/bars may also be present. In the 2004 WHO classification,217 the definition of BAC was limited to a non-invasive process and is now an uncommon diagnosis, requiring full histological assessment; there is currently debate as to whether the term should be used at all, especially since a diagnosis of BAC cannot be rendered on small biopsy samples or cytological samples. However, it is still possible to suggest that a tumor may have a BAC-like component based on cytological features. These include large, cohesive, monolayered sheets which reflect the growth of neoplastic cells in a monolayer along alveolar walls, papillary processes, cell balls and clusters, intranuclear cytoplasmic inclusions and psammoma bodies (Figs 8.18–8.22). The radiogical appearance of a purely ground-glass opacity without a solid component is also suggestive of a BAC.
In adenocarcinomas with meager mucin secretion and little or no glandular formation, the possibility of accurate diagnosis by morphological features is low. The cytoplasmic and nuclear features of large cell carcinoma and of some poorly differentiated squamous cell carcinomas may closely resemble adenocarcinoma; immunostaining for p63, CK5/6 and TTF-1 may be useful.143,213 If the tumor cells are fairly regular and the nuclei are rounded with large central nucleoli they are more likely to be glandular (Fig. 8.23); this is not, however, a completely reliable criterion.
There are no absolute cytological criteria for separating primary from secondary adenocarcinomas, although there may be features indicative of a particular organ of origin (see Chapter 5). Immunostaining for TTF-1 provides the best single marker of a primary origin (Fig. 8.24), although it should be recognised that metastic thyroid malignancies and occasionally metastases from other sites may be TTF-1 positive.144–146
It is likely that so-called atypical adenomatous hyperplasia (AAH) represents a precursor of peripheral adenocarcinomas of the lung.218 Foci of AAH are usually small (<5 mm), incidentally found in lung resections and are seldom sampled by FNB. Nevertheless, they potentially represent a source of error in FNAC diagnosis since they can be composed of highly atypical glandular cells, and may produce small opacities.219 We have encountered one case where the mass lesion seen on CT proved to be an area of non-specific fibrosis at resection and where the small numbers of highly atypical but rather poorly preserved cells seen in FNB samples were derived from adjacent areas of AAH.
It may be difficult to diagnose well-differentiated adenocarcinoma/BAC when there is a low degree of nuclear atypia. The material may mimic benign bronchiolar epithelium, especially when occurring in small sheets. The presence of abundant material should raise the index of suspicion, as should papillae or cell balls. Sometimes there may be only few diagnostic cells so that an overall impression of variation in atypia can be important in reaching a conclusion. Aspirates from non-malignant lesions of the lung frequently contain bronchiolar epithelium, sometimes showing nuclear atypia.84,167 Reactive sheets are usually small and few in number,86 and have more irregular borders than malignant sheets. Jarrett and Betsill considered high cellularity, architectural three-dimensionality, intranuclear cytoplasmic inclusions and large nucleolar size to be the most helpful features in distinguishing adenocarcinoma from reactive bronchiolar proliferations.86 Zaman et al. suggested a predominance of two- and three-dimensional tissue fragments, intranuclear cytoplasmic inclusions and paucity of multinucleated cell forms to be more indicative of malignancy.202 Silverman found a lack of hyperchromasia, prominence of cell borders, presence of terminal plates, goblet cells and nuclear molding to be more characteristic of benign reactive epithelium.84
Confusion with other primary pulmonary neoplasms may occur. Lesions such as carcinoid tumors and rare lesions such as sclerosing hemangioma/pneumocytoma220–223 yield a uniform cell population similar to that of well differentiated adenocarcinoma. Carcinoid tumors usually consist of regular cells which are smaller and more dispersed than those of adenocarcinoma and their round nuclei have a characteristic stippled ‘neuroendocrine’ chromatin pattern (see Figs 8.33 and 8.34). We have not seen examples of papillary adenoma in FNB material, but these rare tumors are also likely to provide problems in differential diagnosis. Pulmonary mucinous cystic neoplasms (PMCN) are also rare; distinction from a variety of pulmonary neoplasms, including mucinous BAC, bronchial mucous gland adenoma, mucoepidermoid carcinoma and metastatic mucinous adenocarcinoma, may be difficult.224 Moran considers that a clear morphological distinction between the benign and malignant forms of PMCN has not been convincingly made and that all lesions should be considered at least potentially malignant.225 Categorization as a cystic tumor may be important in planning surgery, as circumscribed lesions with minimal invasion are likely to have a good prognosis.
Small cell carcinoma (Figs 8.25–8.29)32,35,36,79,82,83
Fig. 8.26 Small cell carcinoma
Pleomorphic poorly cohesive cells with little or no cytoplasm; nuclear molding (MGG, HP).