CHAPTER 5 Lymph nodes
Clinical aspects
Lymphadenopathy is a commonly encountered clinical problem which has a multitude of causes. The commonest cause of peripheral lymphadenopathy is a non-specific reactive hyperplasia in which the underlying etiology is infrequently found (probably an asymptomatic inflammatory process). In general practice, less than 1% of patients with peripheral lymphadenopathy have a malignant process.1 In comparison, retroperitoneal or intra-abdominal lymphadenopathy is usually malignant. In contrast, in young patients, intrathoracic lymphadenopathy is often associated with infectious mononucleosis, sarcoidosis and tuberculosis. The likelihood of malignant disease as a cause of peripheral lymphadenopathy increases over the age of 40 years, nodes over 2 cm in size, firm or matted nodes and non-tender/non-painful nodes.2 Ultrasound is also used in the assessment of lymphadenopathy, from the point of view of a reactive versus a pathological process.3 The bounds of the ‘triple test’ are being expanded beyond breast pathology.
Although surgical excision of a palpable peripheral node is relatively simple, vicinity to other anatomical structures in the neck sometimes causes problems. The procedure does require anesthesia, strict sterility and theater time, and it may leave a scar. To avoid surgery, patients are usually watched for some time before a decision of open biopsy is taken, unless the clinical suspicion of malignancy is strong. Fine needle biopsy (FNB) offers the alternative of an immediate, preliminary, although not always specific diagnosis with little trauma and cost, thus providing ample information for further management.4–8
The plethora of monoclonal antibodies available has proven invaluable in lymph node cytology.9 In particular, it assists in identifying the source of metastatic tumor to lymph nodes, and in distinguishing between various small and large cell neoplasms and malignant lymphoma. The role of cytology in the diagnosis and typing of lymphoma has become better defined in recent years, principally due to immunophenotyping supplementation of routine cytologic preparations. Flow cytometry (FCM) has revolutionized lymphoma diagnosis and typing by FNB and plays a pivotal role.10,11 Immunophenotyping can also be accomplished by immune studies on cytocentrifuge preparations (cytospins) or cell blocks of lymph node aspirates, or core needle biopsy (CNB) specimens.9,12–20 In many cases, definitive diagnosis and typing of lymphoma is possible if cytomorphology is combined with analysis of immunophenotype, and/or cytogenetics and molecular studies.21–32
The place of FNA in the investigative sequence
At the specialist/secondary level, the role of FNB is to provide material for further cytomorphologic analysis and for ancillary studies. The aim is to arrive at a definitive diagnosis and lymphoma typing making full use of the armamentarium of ancillary laboratory techniques. This also applies to most abnormal lymph nodes in deep sites. In some centers, this strategy has been very successful and has virtually replaced surgical removal of nodes. This requires readily available expertise in FCM, cytogenetics, immunostaining, molecular analysis (e.g. PCR), haematopathology and cytopathology. Others, including the author, feel that a tissue sample obtained by CNB or surgical biopsy is often still needed. Lymphoma diagnosis and typing relies not only on cytology but also on altered tissue and immmuno-architecture,33 which requires a tissue sample, although a small fragment obtained by CNB or occasionally a cell block may suffice. Tissue samples can also help to overcome deviations from classical immunophenotypes on FCM. They also enable spatial recognition, assisting in the diagnosis of various reactive lymphadenopathies. Close correlation of cytological, immunological and clinical features is essential when using FCM to make a diagnosis of lymphoma and in its classification. At times, additional correlation with the cytogenetic and molecular studies will be necessary.
The value of FNB in the investigation of suspected lymphoma can be summarized as follows:
Accuracy of diagnosis
The accuracy of FNAC of lymph nodes in the diagnosis of metastatic malignancy is influenced by many factors such as the size and site of the node, fibrosis, necrosis, previous irradiation and the number of punctures made. Small mobile nodes high up in the axilla are difficult to sample, while adequate material can easily be obtained from nodes only a few millimeters in diameter in a cervical or supraclavicular position.37 Deep nodes are accessible to FNB by means of radiological imaging and guidance.38,39 Fibrosis sometimes makes it difficult to obtain sufficient material for diagnosis from reactive inguinal nodes. It can also be a problem in nodular sclerosis Hodgkin lymphoma40 and in some mediastinal and retroperitoneal non-Hodgkin lymphomas (NHL).9
Diagnostic sensitivity is occasionally limited by the fact that small metastatic deposits, metastases confined to the subcapsular sinus and single-cell metastases can be missed. However, early micrometastases rarely produce significant lymph node enlargement and if a lymph node is palpable it is likely to contain enough tumor tissue to be detectable by FNB. The diagnostic sensitivity of metastatic and recurring malignancy reported in the literature is usually above 95%.23,41–53 Failure to obtain a representative sample is responsible for most false-negative diagnoses. Interpretation of a representative sample can be problematic, more often with lymphoma than with metastatic malignancy. For example, without immunophenotyping low-grade follicular lymphoma can easily be mistaken for reactive follicular hyperplasia.54 Thus, although a negative cytological report makes malignancy unlikely, it is not singularly diagnostic,37 and if the lymphadenopathy does not show signs of regression within a month or two, FNB should be repeated or a node should be excised for histology.
Diagnostic specificity for malignancy, on the other hand, is high. False-positive diagnoses are rare23,42,48,52–57 if particular caution is observed in the interpretation of smears from nodes in fields of previous irradiation and in the presence of necrosis. The existence of benign epithelial inclusions in lymph nodes (see p. 89) should be kept in mind. Again, the main problem is in relation to lymphoma, particularly the distinction between reactive follicular hyperplasia and follicular lymphoma. Most false-positive diagnoses are cases of reactive lymphadenopathy reported as suspicious of lymphoma.
Diagnostic accuracy not only depends on the aspirate being representative, but also very much on the quality of the cytological preparations. This is particularly the case in the diagnosis of some reactive lymphadenopathies and in the diagnosis and classification of lymphoma, which depends on the study of fine cytological detail in high magnification, and on an estimate of proportions of various cell types in the smear.54,55,58–60 It is essential that the aspirates are handled and smears prepared by staff with experience in cytology to achieve satisfactory results.
There are conflicting opinions regarding the accuracy of cytological diagnosis and typing of malignant lymphoma.19–26,30,55,60–65 It is difficult to extract exact figures for several reasons. Most series include relatively small numbers of cases, with histologic correlation and/or consistent follow-up. Early series are based on cytomorphology (in FNB smears) alone, later series combine cytomorphology and immunophenotyping by flow cytometry and/or immunostaining of smears. The case mix (types of lymphoma; primary diagnosis or recurrent disease) is variable and often is not specified. With these reservations, it appears that the accuracy of diagnosing NHL by cytomorphology alone is in the range 60–80%, significantly lower for low-grade than for high-grade lymphomas.26,37,45,48,51,66
Diagnostic sensitivity has generally been found to be lower for lymphoma than for metastatic malignancy.23,48,67,68 For a diagnosis of lymphoma to be of clinically practical value, it must identify good and bad prognosis subgroups and therefore must specify the subtype. In an extensive review of the literature, two-thirds of the 30 studies reviewed, in which FNB was supplemented by immunophenotyping, diagnostic sensitivity was over 80% and specificity over 90%. However, only three of eight studies achieved a correct and precise subtyping in the primary diagnosis of NHL in more than 80% of cases.22 Difficulties in correlating cytology and histology of lymphoma have been enhanced by the parallel use and frequent modifications of different systems of classification over the years. The advent of the World Health Organization (WHO) lymphoma classification based on clinical, cytologic, immunophenotypic and genetic features has improved both the detection rate and classification of NHL, due to diminished importance placed on architectural features in subtyping NHL.22,26,30,32 However, the accuracy of cytologic diagnosis is still limited in some forms of NHL, notably lymphomas with predominantly small cells, mainly marginal zone lymphoma, and in peripheral T-cell lymphoma, T-cell/histiocyte rich B-cell lymphoma and nodular lymphocyte predominant Hodgkin lymphoma. This is also the case with the diagnosis of composite lymphoma and in the grading of follicular lymphoma.
As a rule, FNB samples from malignant lymphoma are very cellular and can be used for ancillary studies.9,31,54,69–72 However, more material to allow a wider range of immune markers, assessment of tissue and immunoarchitecture as well as storage of material for ancillary studies and research can be obtained by supplementary CNB of the node, principally in intra-abdominal or intrathoracic nodal disease. With deep-seated masses, CNB is a useful adjunct to FNB. The FNB provides superior cytomorphology and is an excellent source for flow cytometry cell suspensions. This combined approach can increase diagnostic accuracy, assist in classification and reduce the number of insufficient samples.73,74
The accuracy of cytological diagnosis of Hodgkin lymphoma (HL) is quite variable, frequently high, and in some studies greater than 85%.22,23,75–77 The main problems in cytodiagnosis are in relation to the nodular lymphocyte predominant and lymphocyte-rich variants as well as separation from HL mimicks. The accuracy of subtyping HL in FNB smears is relatively poor.22–24 Performing immunocytochemistry on cell block preparations, cytospins or NCB may help to a degree but a formal node biopsy is still necessary to provide reliable subtyping in most cases.
Complications
Significant complications do not occur. Post-aspiration hematoma or necrosis is rare.78,79 To date, septic complications or tumor implantation in the needle track have not been reported following FNB of lymph nodes.
Technical considerations
Local anesthetic is not used and simple skin disinfection as for an injection is adequate. Two or more samples may be necessary to secure enough material for both routine smears and for special investigations and to reduce sampling error in focal disease. The use of gloves and extreme care in handling used needles are important safety precautions. The techniques involved in the biopsy of deep nodes using radiological guidance are described in Chapter 3. If the standard technique does not yield sufficient material, for example due to fibrosis (nodular sclerosis Hodgkin lymphoma and some sclerosing NHLs), a cutting core needle can be tried.34,35 In general, large-caliber core needles of 14–18 gauge are recommended in the definitive diagnosis and typing of NHL, as they allow a more accurate assessment of tissue and immunoarchitecture as mentioned above. Artifactual change is less than with a 20-gauge needle.
It is not easy to make perfect direct smears from samples of lymphoid tissue and this takes considerable practice. An air-dried smear has to dry quickly for optimal fixation and has to be made thin and even. The smearing pressure must be well balanced to obtain a thin smear and at the same time avoiding crush artifacts. If the aspirate is bloody or thinned by a large amount of lymph fluid, cells need to be concentrated and separated as much as possible from the fluid. This can be achieved by using the two-step technique illustrated in Figure 2.7. A wet-fixed smear must be fixed immediately to minimize drying artifacts. Only those parts of the smears where the cells are evenly dispersed, well fixed and not distorted by the trauma of smearing should be chosen for diagnostic evaluation.
Heavy reliance is placed on ancillary techniques in FNB of nodes. A cell suspension made by gently dispersing the sample in Hank’s balanced salt solution with 10–20% fetal calf serum or by rinsing the needle with the fluid is a suitable preparation for this purpose. The suspension is spun in a cytocentrifuge at 300–700 rpm for 3–5 minutes. Processing should be done as soon as possible after biopsy since cell fragility increases rapidly with time. Cytospin preparations and cell blocks/cell buttons (see Chapter 2) allow assessment by a panel of immune markers by immunocytochemical staining as illustrated in the section on lymphoma. An elementary NHL screening panel could consist of CD20, CD3, CD5, kappa and lambda light chains, and this can be extended as appropriate. Background staining due to serum proteins and fragmented cells can be reduced by resuspending the cells in Hank’s fluid after initial centrifugation. Direct smears may be unsuitable due to the dispersal of cytoplasm and proteins in the background (see Chapter 2). However, immunostaining of FNB smear specimens can be utilized with the use of formal saline fixation. It eliminates background staining while allowing preservation of cellular antigenicity and morphology.80 Immunocytochemistry is helpful in tracing the origin of metastatic malignancy, in the differentiation of lymphoma from reactive processes and from anaplastic carcinoma or melanoma, and in the classification of lymphoma.13–1618 FCM analysis of aspirated cells is vital in immunophenotyping lymphoid populations, determining clonality and aberrant antigen expression. FNB material can be rinsed in a buffered balanced salt solution or RPMI solution to provide material for analysis. In addition, aspirated material can be used in conventional cytogenetics or fluorescence in situ hybridization (FISH) to assess for chromosomal translocations as well as in molecular studies/PCR to assess gene rearrangements.
If there is any suspicion of an infective process, the needle can be rinsed with sterile saline after the smears have been prepared, but preferably the biopsy should be repeated to provide sufficient material for culture for microorganisms (see p. 18 and Chapter 18).
Liquid-based preparations have been used minimally for nodal FNB samples. If used, they must be interpreted with caution as there are many cytomorphologic alterations.81
Cytological findings
Fine needle biopsy samples of lymphoid tissue, nodal or extranodal, benign or malignant, as a rule have a very high cell content. This is obvious to the naked eye as the aspirate is smeared. It looks like a film of slimy material which turns gray on drying. The cytoplasm of lymphoid cells is fragile. Many cells are represented by naked nuclei or have only a small rim of cytoplasm. A variable number of rounded cytoplasmic fragments measuring up to 8 microns in diameter are scattered in the background (Fig. 5.1). The cytoplasmic fragments were named ‘lymphoglandular bodies’ by Söderström. The term ‘lymphoid globules’ is prefered since they are present in smears from any lymphoid infiltration, not only from lymph nodes. The fragments stain an uniform pale blue, identical to the cytoplasm of intact cells, with Giemsa stain. They differ from necrotic debris by their regular round shape and their uniform staining. Necrotic debris and nuclear fragments (karyorrhexis and apoptotic bodies) are characteristic of smears of small cell undifferentiated carcinoma (Fig. 5.2). The recognition of ‘lymphoid globules’ is of diagnostic value in the distinction of lymphoma from anaplastic carcinoma and some other tumors.
Most of the lymphoid cells are dispersed as single cells but some may form clumps or aggregates, especially in smears of bloody samples. Cell detail is obscured in dense clusters, which are of no diagnostic value as they can be found in both reactive and malignant nodes. However, a tendency for neoplastic follicular center cells to form aggregates resembling neuroendocrine carcinoma is commonly seen in follicular lymphoma,82,83 and cells of anaplastic large cell lymphoma can also form carcinoma-like aggregates.84–86 Tissue fragments consisting of a vascular core of endothelial cells with adherent lymphoid cells and histiocytes are sometimes present in smears from reactive nodes (Fig. 5.3).
The reactive node (Figs 5.4–5.16)
Fig. 5.4 Reactive lymphadenopathy, follicular
Tissue section showing detail of germinal center with centroblasts, centrocytes and tingible body macrophages. Outer rim of small lymphocytes. The corresponding cytological pattern is shown in Figure 5.5 (H&E, HP).
(Reproduced with permission from van Heerde et al.9)
Fig. 5.8 Infectious mononucleosis
Many transforming lymphocytes, plasmacytoid cells and immunoblasts (MGG, HP).
Fig. 5.9 Infectious mononucleosis
A similar pattern to Figure 5.8 of a high proportion of transformed lymphocytes and a mitotic figure (Pap, HP).
Fig. 5.12 Sinus histiocytosis with massive lymphadenopathy (Rosai-Dorfman disease)
(Reproduced with permission from van Heerde et al.9)
Criteria for diagnosis
The reactive pattern is variable depending on the degree of stimulation, the number and size of germinal centers and on whether the sample derives mainly from a germinal center or from interfollicular or paracortical tissue. Germinal center material is represented by poorly defined tissue fragments composed of centroblasts, centrocytes, ‘tingible body’ macrophages (Figs 5.4, 5.5), and a number of lymphocytes which adhere to the syncytial cytoplasm (pale gray/violet in MGG) of dendritic reticulum cells (Fig. 5.6A). Dendritic reticulum cells have oval or round nuclei with a smooth nuclear membrane, a coarsely granular, uniformly distributed chromatin and small distinct nucleoli (Fig. 5.6B). The cytoplasm is dispersed in the background. A smear, which derives mainly from interfollicular tissue, consists predominantly of lymphocytes with a variable but much smaller number of scattered immunoblasts, plasma cells, non-specific histiocytes and endothelial cells (Fig. 5.7). Multiple biopsies diminish the bias caused by selective sampling.
The main features which distinguish a reactive process from lymphoma are:
Morphologically similar to reactive lymphoid hyperplasia and rarely distinguishable are HIV-associated lymphadenitis, progressive transformation of germinal centers and Castleman’s disease. When germinal centers are very large, as in some cases of reactive follicular hyperplasia, the proportion of large cells (centroblasts, dendritic reticulum cells) and the number of mitoses, in a FNB sample, may be impressive enough to suggest malignant lymphoma. However, the full range of lymphocyte transformation is still preserved, including small lymphocytes, and the various cell types occur in logical proportions. Small or slightly enlarged lymphocytes are still numerically predominant. A variable number of plasma cells can usually be found. The presence of macrophages with tingible bodies favors reactive hyperplasia but does not rule out lymphoma. Especially in high-grade lymphomas with a high turnover of cells, a considerable number of ‘starry sky’ macrophages may be present (see Figs 5.48 and 5.51A). A cytological pattern of reactive hyperplasia with a large number of plasma cells but no other distinguishing features can be seen, for example, in cases of secondary syphilis, rheumatoid arthritis, autoimmune syndromes, IgG4-related lymphadenopathy, HIV infection and the plasma cell variant of Castleman’s disease.
A prominent immunoblastic and plasmacellular reaction is found in several conditions. In viral lymphadenitis, particularly in infectious mononucleosis,89,90 immunoblasts, plasmacytoid cells, mature plasma cells and atypical lymphocytes can be numerous but the range of cells is still in logical proportions (Figs 5.7, 5.8 and 5.9). The atypical lymphocytes have an abundant basophilic cytoplasm, an enlarged, often eccentric, nucleus and a paler nuclear chromatin than a normal lymphocyte. Immunoblastic cells can cause differential diagnostic problems. The main differential diagnoses are large cell lymphoma (with numerous immunoblasts) and HL. Atypical binucleate immunoblasts closely resembling Reed-Sternberg cells are rarely seen (see Fig. 5.69).91 Occasionally, immunoblasts are mistaken for mononuclear Hodgkin cells; however, they are smaller with smaller nucleoli and greater basophilic cytoplasm than Hodgkin cells. Mononucleosis is usually suggested by the clinical presentation and can be confirmed by serological tests.
Abnormal immunoblastic reactions with prominent immunoblasts and sometimes Reed-Sternberg-like cells can at times be difficult to distinguish from large cell lymphoma. This pattern may be seen, for example, in postvaccinial lymphadenitis and Dilantin hypersensitivity lymphadenitis (Fig. 5.10). The clinical history may provide a clue in such cases (e.g. patients taking Dilantin usually develop adenopathy in the first 6 months of therapy). A definitive diagnosis would require either further ancillary investigation or a formal histological examination. Immunologic studies can help to confirm the polyclonal nature of the immunoblasts.
Histiocytes, which have an abundant, pale or sometimes eosinophilic cytoplasm, may be prominent in smears of lymph node aspirates. The histiocytes occur singly or in small groups. The cytoplasm is often vacuolated or granular and may contain phagocytosed debris or pigment. An increased number of histiocytes without specific features can be seen in smears from non-specific reactive nodes, and are suggestive of sinus histiocytosis (Fig. 5.11). The cytological findings in sinus histiocytosis with massive lymphadenopathy (Rosai-Dorfman disease) have been described by van Heerde and others.9,92–94 In this condition, smears contain many large histiocytes showing lymphophagocytosis (Fig. 5.12). Prominent histiocytes and multinucleated giant cells as a reaction to foreign material can occasionally be seen in axillary nodes of women with silicone breast prostheses95 (Fig. 5.13) or inguinal lymphadenopathy associated with artificial hip joint replacement.96
Scattered small clusters of a few histiocytes with ovoid, pale nuclei and abundant cytoplasm resembling epithelioid cells in a smear consistent overall with follicular hyperplasia are suggestive of toxoplasmosis (Fig. 5.14A). Well-formed histiocytic granulomata resembling sarcoid granulomata are unusual but occur occasionally. Lymphoid cells with relatively large, ovoid, pale nuclei may also be seen (Fig. 5.14B). These cells probably correspond to the pale monocytoid B cells observed in histological sections. Occasionally, the presence of many of these large cells can raise a suspicion of malignancy. The cytological pattern is not diagnostic by itself and needs confirmation by serological tests. Microcysts and toxoplasma organisms are rarely found in smears of toxoplasma lymphadenitis (see Chapter 18).97–100
Numerous noncohesive, pale, histiocyte-like cells (interdigitating cells) with typical folded nuclei (Fig. 5.15) are present in dermatopathic lymphadenopathy. Some macrophages containing pigment, either hemosiderin or melanin, are usually found. These have smaller and more consistently oval, nonfolded nuclei different from interdigitating cells, and have a better-defined cytoplasm. There may be a variable number of eosinophils. The background is predominantly of lymphocytes, which may appear slightly ‘atypical’ with small pale, central nucleoli (stimulated T cells) and blast forms are less common.
The histiocytes of Langerhans cell histiocytosis (histiocytosis X) have characteristically large nuclei of irregular shape. They may be folded, convoluted, lobulated and grooved (Fig. 5.16 and Fig. 16.10). Mitotic activity may be seen and sometimes necrosis. Such cells seen in a lymph node aspirate, especially in the absence of eosinophils, may raise a suspicion of metastatic malignancy such as melanoma. However, the nuclear chromatin of Langerhans histiocytes is bland and finely granular. If suspected, the diagnosis may be confirmed by immunocytochemistry (CD1a, S-100, langerin)69,101 and/or by electronmicroscopy (Birbeck granules).9,101–104 Histiocytic sarcoma is described on page 111 (see Fig. 5.76).105
Multinucleate giant cells are usually of histocytic origin and frequently associated with granulomas; however, Warthin-Finkeldey multinucleate giant cells (polykaryocytes) are T cells. They can occasionally be seen in reactive lymphadenitis associated with measles, HIV and Kimura’s disease. Additionally, they may be seen with various lymphomas.106
Descriptions of the cytology of angiofollicular lymphoid hyperplasia can also be found in the literature,9,107–109 as well as of Kimura’s disease.110,111 Follicular dendritic sarcoma may arise in association with Castleman’s disease or de novo.112,113 Rare descriptions of interdigitating dendritic cell sarcoma are also in the literature.114
Granulomatous lymphadenitis (Figs 5.17–5.20)
Epithelioid cells are quite distinctive in FNB smears. They have elongated nuclei the shape of which resembles the sole of a shoe or boomerang. The nuclear chromatin is finely granular and pale and the cytoplasm is pale without distinct cell borders (Fig. 5.17). Epithelioid cells of granulomatous lymphadenitis form clusters; large clusters resemble granulomas in tissue sections. Multinucleated Langhans giant cells may be few in numbers and are sometimes absent. Granulomatous lymphadenitis may or may not show necrosis or suppuration. Necrosis may be of fibrinoid or caseous types. Caseous material appears granular and eosinophilic in smears and usually lacks recognizable cell remnants (Fig. 5.18).
Granulomata with caseous necrosis is the hallmark of tuberculous lymphadenitis. However, necrotizing granulomata may also be seen with fungal infections. Granuloma-like clusters of epithelioid cells, in the absence of necrosis, are more suggestive of sarcoidosis, but tuberculosis and fungal infections cannot be ruled out and staining for acid-fast bacilli and fungi is imperative in all cases of granulomatous lymphadenitis. Smears from a tuberculous lymph node may sometimes show only polymorphs and necrotic debris without histiocytes, particularly in immunocompromised patients. Acid-fast bacilli should, of course, be looked for both in direct smears and in culture from the aspirate. PCR is a sensitive way to detect mycobacterial organisms.115–120
Non-necrotizing sarcoidal type granulomata may also be seen with foreign body reactions, brucellosis, Crohn’s disease, leishmaniasis and leprosy. Leprosy in lymph nodes has also been diagnosed by FNAC.121 Conspicuous neutrophils in a smear showing epithelioid granulomas and necrosis – suppurative granuloma (Fig. 5.19) – suggest atypical mycobacterial infection if the aspirate is from a cervical node in a child, cat scratch disease if from an axillary node,122–124 and lymphogranuloma venereum if it is from an inguinal node. Fungal and corynebacterial infections may produce a similar reaction, as can chronic granulomatous disease of childhood, leishmaniasis, Yersinia enterocolitica and tularemia. Occasionally, groups of epithelioid histiocytes in toxoplasmosis may be large enough to mimic true granulomatous lymphadenitis. Inclusions within histiocytes and giant cells such as birefringent particles, asteroid bodies, Schaumann bodies, etc. are not particularly helpful in making a specific diagnosis. If no etiological agent is found, one can only report the case as granulomatous lymphadenitis with or without necrosis and/or suppuration, and the etiology must be pursued by other means.
If an aspirate consists entirely of necrotic material, it may be difficult to decide whether it represents caseous necrosis or tumor necrosis (see Figs 8.6). Sampling should be repeated if no bacilli, epithelioid cells or tumor cells are found on careful examination of the smears.
Sometimes, only a few epithelioid cells are found in small groups or as single cells or the histiocytes may not quite have the typical appearance of epithelioid cells. The pattern then approaches that of non-specific, reactive lymphadenitis with prominent histiocytes. This may be the case in toxoplasma lymphadenitis and in the early stages of sarcoidosis. Endothelial cells can sometimes also be mistaken for epithelioid histiocytes. One such example is Kaposi’s sarcoma in which tissue fragments may resemble a granuloma, but the nuclei are more elongated and spindly and usually are hyperchromatic. They lack the ‘sand shoe’ indentations of epithelioid histiocytes and multinucleated giant cells are not seen (Fig. 5.20) (See also Chapter 18).125–127 In Kaposi’s sarcoma the cells are HHV-8, CD31, CD34 and D2-40 positive.
Necrotising lymphadenitis (Kikuchi-Fujimoto disease) is a condition of unknown etiology, seen mainly in young women, in which there is focal necrosis with a proliferation of histiocytes, lymphocytes and plasmacytoid dendritic cells, usually in cervical lymph nodes.128–130 The presence of large mononuclear cells in such nodes may cause a suspicion of malignant lymphoma. In FNB smears, the characteristic findings are of large numbers of pale, phagocytosing histiocytes with eccentric, crescentic nuclei, debris with nuclear fragments, absence of neutrophils and a reactive background of lymphoid cells (Fig. 5.21). The histiocytic cells are CD68, CD163 and myeloperoxidase positive.
Necrosis in lymph nodes may also occur in systemic lupus erythematosus. In systemic lupus there may be significant numbers of plasma cells and/or hematoxylin bodies, which may help to distinguish it from Kikuchi-Fujimoto disease. If these features are absent, then serology would be necessary to make this distinction.131
Smears from areas of coagulation necrosis in lymph nodes show numerous cell shadows, some with preserved but pyknotic nuclei. Unless there is a clear history of trauma, such findings raise a strong suspicion of either metastatic carcinoma or malignant lymphoma. Nodal metastases of small cell anaplastic carcinoma of lung, melanoma and breast carcinoma are prone to necrosis and the necrotic cells with pyknotic nuclei can be indistinguishable from necrotic lymphoid cells. Extensive necrosis/infarction is common in malignant lymphoma, both non-Hodgkin and Hodgkin. Total infarction of a lymph node can sometimes precede manifest lymphoma. It has been shown that the demonstration of clonal immunoglobulin or T-cell gene rearrangement in the infarcted node can suggest a diagnosis of lymphoma.132
Benign epithelial inclusions of salivary gland or thyroid origin have been observed in cervical nodes, inclusions of Müllerian origin in pelvic nodes. Mediastinal nodes rarely have mesothelial inclusions while neval cells may be found in axillary nodes. Such inclusions are usually very small. Although a rare occurrence, this possibility should be kept in mind when only a few epithelial cells without obvious malignant characteristics are found in lymph node aspirates in the appropriate context. Groups of glandular epithelial cells of apocrine type are commonly found in FNB smears from axillary nodes. The presence of such cells could give rise to a suspicion of metastasis from breast cancer if the benign characteristics of the cells are not appreciated. The cells most likely represent contaminants from adjacent sweat glands (see Chapter 7).
The common problem of differentiating between necrotising granulomatous lymphadenitis and necrotic tumor has already been mentioned. Squamous cell carcinoma is particularly prone to undergo liquefactive necrosis. An aspirate from such a node consists of thin, mucoid, yellow fluid. Well-preserved neoplastic squamous cells may be few in number and are often very well differentiated (see Fig. 4.2). There is therefore a risk of mistaking a cystic metastasis of well-differentiated squamous carcinoma in a cervical gland for a branchial cyst (see Chapter 4).83,133 Cystic nodes in the neck may also represent metastases of papillary carcinoma of the thyroid. This possibility should be remembered, particularly in cases of unexplained cervical lymphadenopathy in a young patient (see Chapter 6).
Follicular lymphoma (grade 1–2) may mimic metastatic small cell anaplastic carcinoma or Merkel cell carcinoma in FNB smears.134,135 This is because of the tendency for neoplastic cells of centrocytic type to form cohesive clusters, sometimes with nuclear molding. This problem will be further discussed in the section on lymphoma as will the differentiation of large cell lymphomas (centroblastic, immunoblastic) from other large cell malignancies by immunocytochemistry12,105 and pseudoepithelial clustering of lymphoid cells.
Indicators of the primary site
A stepwise approach to the investigation of nodal metastasis is suggested. This should include:
The cytological patterns seen in routinely stained smears often give clues to the site of the primary tumor. Columnar cells with elongated nuclei arranged in palisades, stringy mucus and necrosis suggest a primary in the large bowel (see Fig 10.32), while mucin-containing signet ring cells suggest the stomach as the most likely primary site among several other possibilities. Glandular cells, moderately pleomorphic, arranged in a gland-in-gland or a cribriform pattern suggest prostatic carcinoma (Fig. 5.22). Large cells with abundant pale, granular or finely vacuolated cytoplasm and a low N : C ratio suggest a renal cell carcinoma. Very large central nucleoli are typical of less well-differentiated forms of this tumor and are also seen in large cell anaplastic carcinoma of lung and nasopharynx (see Fig. 5.55) and in hepatocellular carcinoma. Pulmonary and pancreatic adenocarcinoma can have a variety of patterns. They usually show a moderate degree of glandular differentiation, prominent nuclear pleomorphism and obvious mucin secretion. As a rule, the presence of intracytoplasmic mucin excludes renal, adrenal, hepatocellular and thyroid carcinoma. Breast cancer usually displays poor glandular differentiation while cell balls and single files of cells are more common. Nuclear pleomorphism is often relatively mild. Cells with intracytoplasmic neolumina containing ‘bull’s-eye’ inclusions are also suggestive of breast carcinoma.