J Clin Pathol 1998;51:197–203 197

Fine needle aspiration cytology diagnosis of J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from malignant lymphoma and reactive lymphoid hyperplasia

C J R Stewart, J A Duncan, M Farquharson, J Richmond

Abstract and, therefore, helps to guide appropriate spe- Aims—To assess the diagnostic accuracy cialist referral and further investigation.1 of fine needle aspiration FNA cytology is highly reliable in the identi- (FNA) cytology to distinguish reactive fication of metastatic carcinoma and melanoma lymphoid hyperplasia from malignant in lymph nodes, limiting the requirement for lymphoma, and to evaluate the contribu- diagnostic excision biopsy in many patients. tion of ancillary techniques applied to However, the role of aspiration cytology in the cytological material. assessment of primary lymphoproliferative dis- Methods—Two hundred and seventy seven orders has been less certain.2 Both clinicians consecutive lymph node FNA specimens and pathologists have doubted whether an reported to be consistent with reactive accurate cytological diagnosis of lymphoma is lymphoid hyperplasia (n = 213) or possible, and early reports suggested that FNA suggestive/diagnostic of malignant lymph- produced high false negative rates in patients oma (n = 64) were reviewed. Follow up data with Hodgkin’s disease and low grade non- were obtained by case record review or by Hodgkin’s lymphoma. In addition, the com- histological correlation. The value of im- plexity of lymphoma classifications and the munocytochemistry, in situ hybridisation prognostic importance of architectural assess- for immunoglobulin light chain mRNA, ment in some lymphoid tumours limits the and polymerase chain reaction (PCR) extent of cytological correlation with histology. towards the final clinicopathological diag- However, recent studies have indicated that an nosis was assessed in 92, 61, and 45 cases, accurate diagnosis of lymphoma can be respectively. achieved by FNA in 85–90% of cases, particu- Results—Sixty one of 67 lymphomas and larly when morphological assessment is com- 207 of 209 reactive lymph nodes were accu- plemented by the use of immunocytochemical rately diagnosed by FNA cytology. There techniques.2–4 Furthermore, genotypic analysis

were six false negative aspirates including including Southern blot and polymerase chain http://jcp.bmj.com/ three cases of , two reaction (PCR) to detect immunoglobulin and cases of Hodgkin’s disease, and one receptor gene rearrangements, and in situ chronic lymphocytic leukaemia. Two FNA hybridisation (ISH) to detect immunoglobulin specimens considered suspicious of light chain mRNA expression may be applied lymphoma proved reactive on histology or successfully to FNA specimens.5–9 Skoog and clinical follow up. One metastatic small Tani have suggested that lymph node aspiration cell carcinoma was wrongly diagnosed as cytology supplemented by appropriate ancillary lymphoma. Ancillary studies contributed investigations oVers similar diagnostic accuracy on September 29, 2021 by guest. Protected copyright. to the correct diagnosis in most cases to excision biopsy.10 although occasional misleading results Previously, we described the use of immuno- were obtained, particularly with PCR. cytochemistry, ISH, and PCR in the assess- Conclusions—FNA cytology accurately ment of small series of lymphoid aspirates distinguished reactive lymphoid hyper- derived from both lymph nodes and extranodal plasia from malignant lymphoma in 97% sites.89 In this report, we studied the value of of cases. However, occasional wrong diag- these techniques in routine practice, and noses occurred owing to sampling error or examined the diagnostic accuracy of FNA in Department of misinterpretation. Ancillary studies can distinguishing malignant lymphoma from reac- Pathology, The Royal be applied to cytological samples and con- tive lymphoid hyperplasia in a series of 277 Infirmary, Castle tribute to the diagnosis in most cases. consecutive lymph node aspirates performed in Street, Glasgow ( 1998; :197–203) G4 0SF, UK J Clin Pathol 51 Glasgow Royal Infirmary. C J R Stewart M Farquharson Keywords: fine needle aspiration; cytology; lymphoma; J Richmond lymph node Methods DIAGNOSTIC CASES Department of All fine needle aspiration performed on pa- Medical Oncology, The Royal Infirmary Fine needle aspiration (FNA) cytology is being tients with palpable lymph node enlargement J A Duncan used increasingly in the assessment of patients in Glasgow Royal Infirmary between January with lymph node enlargement. The technique 1993 and May 1997 were reviewed. The aspi- Correspondence to: is safe and simple and can be used to sample rates were performed by cytology staV using 23 Dr Stewart. multiple sites. The use of rapid staining or 25 G needles. In most cases, needles were Accepted for publication techniques often allows a provisional diagnosis attached to 10 ml syringes with a syringe 9 December 1997 to be made at the patient’s initial presentation holder (Cameco, London, UK), but a few 198 Stewart, Duncan, Farquharson, et al

Table 1 Correlation of FNA diagnosis and final (Dako) was performed in those cases in which

clinicopathological diagnosis in 277 lymphoid aspirates metastatic carcinoma or melanoma were in- J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from cluded in the diVerential diagnosis. In situ Final clinicopathological diagnosis hybridisation and PCR were used in cases of FNA diagnosis Reactive NHL HD Met CA suspected lymphoma, mainly in the latter Reactive lymphoid half of the study period. hyperplasia (n = 213) 207 4 2 – In each case, an attempt was made to catego- Diagnostic of NHL (n = 40) – 39 – 1 Suggestive of NHL (n = 8) 1 7 – – rise the value of the special techniques in reach- Diagnostic of HD (n = 13) – – 13 – ing the final clinicopathological diagnosis. The Suggestive of HD (n = 3) 1 – 2 – technique was considered helpful if it supported NHL, non-Hodgkin’s lymphoma; HD, Hodgkin’s disease; Met or confirmed the correct diagnosis, unhelpful if CA, metastatic carcinoma. it was essentially non-contributory, and mis- leading if it suggested either a false negative specimens were obtained using the non- (reactive pattern in proven lymphoma) or a false aspiration technique. Routinely, two aspirates positive (monoclonal pattern in proven reactive were performed and direct smears were lymph node) diagnosis. prepared for staining with DiV-Quick (Dade, Abingdon, UK), May-Grunwald-Geimsa, Results and/or Papanicolaou methods. Thereafter, the CLINICOPATHOLOGICAL CORRELATION needles were rinsed in 10 ml normal saline. In The 277 lymphoid aspirates were obtained selected cases, cytospin preparations from the from 260 patients. Fifteen patients underwent saline washes were used for special techniques, FNA on two occasions and one patient had and the washes were submitted for culture in three aspirates. those cases where infection was suspected. The correlation between the cytological In total, 549 lymph node aspirates were diagnoses and the final clinicopathological performed during the study period of which 235 assessment is summarised in table 1. (42.8%) showed metastatic carcinoma or Sixty four aspirates were considered diagnos- melanoma. Thirty seven (6.7%) FNA speci- tic (n = 53) or suspicious (n = 11) of mens were inadequate for diagnosis. The lymphoma and, of these, 47 represented the remaining 277 aspirates showed a lymphoid primary diagnosis of a lymphoproliferative dis- pattern that was considered reactive in 213 cases order while 17 represented recurrent disease (38.8%) and diagnostic or suggestive of malig- after treatment. Forty eight specimens were nant lymphoma in 64 cases (11.7%); these lym- considered to be derived from non-Hodgkin’s phoid aspirates form the basis of this study. lymphoma, of which 46 were reported as B cell The diagnostic accuracy of FNA cytology lymphomas on the basis of cytomorphology and was assessed by histological correlation or by ancillary investigations; two non-Hodgkin’s clinical follow up. Clinical data were obtained lymphomas were not otherwise specified cyto- by review of case records or by correspondence logically. Sixteen cases were reported as sugges- http://jcp.bmj.com/ with referring physicians and general practi- tive or diagnostic of Hodgkin’s disease. In gen- tioners. The follow up period for those eral, no attempt was made to subclassify specimens diagnosed as reactive lymphoid lymphomas further on FNA cytology. hyperplasia averaged 13 months. Of those cases considered diagnostic of lymphoma cytologically, clinicopathological SPECIAL TECHNIQUES correlation confirmed the FNA diagnosis in 39 Additional studies were performed on selected of 40 non-Hodgkin’s lymphomas and in all 13 aspirates to aid distinction between reactive cases of Hodgkin’s disease. However, one aspi- on September 29, 2021 by guest. Protected copyright. and neoplastic lymphoid proliferations and, in rate considered diagnostic of non-Hodgkin’s a few cases, to distinguish lymphoid from non- lymphoma was shown to be metastatic small lymphoid neoplasms. cell carcinoma on excision biopsy. Lymph node Immunocytochemistry, ISH to detect immu- biopsy was confirmatory in eight of 11 aspirates noglobulin light chain mRNA, and PCR for reported as suspicious of lymphoma, while one immunoglobulin heavy chain gene rearrange- further case (not biopsied) was clinically ment were used, as described previously.89 consistent with a Burkitt-type lymphoma. His- Some specimens were not subject to ancillary tology in one case and clinical follow up in studies owing to insuYcient material in the another indicated reactive lymphoid changes in needle rinse sample. FNA sampling was not the remaining two suspicious cases. usually repeated if the cytomorphology on rou- Histological correlation was available in 44 tine preparations in conjunction with the clini- non-Hodgkin’s lymphomas from 40 patients. cal findings were suYcient to warrant excision The biopsy diagnoses were follicular biopsy of the node. Immunocytochemistry was lymphoma (n = 16), diVuse large B cell used throughout the period of the study. In lymphoma (n = 12), lymphocytic lymphoma/ most cases, a panel of antisera to CD45, CD20, chronic lymphocytic leukaemia (n = 5), mono- CD79a, CD43 (all Dako, High Wycombe, cytoid B cell lymphoma (n = 3), lymphoplas- UK), CD3 (SAPU, Carluke, UK), and immu- macytic lymphoma/myeloma (n = 2), low noglobulin light chains was used for routine grade B cell lymphoma not otherwise specified diagnostic assessment. Antisera to CD15 (n = 3), centrocytic lymphoma, B lymphoblas- (SAPU) and CD30 (Dako) were included if tic lymphoma (n = 1), and post-transplant Hodgkin’s disease was suspected. Immunos- lymphoproliferative disorder (n = 1). Lymph taining for cytokeratin (Dako), S100 protein node biopsies were performed in 15 cases of (Biomen, Finchampstead, UK), and HMB45 Hodgkin’s disease from 13 patients. These FNA cytology in lymphoma and lymphoid hyperplasia 199

opsy. Review of the initial aspirate showed no

evidence of Hodgkin’s disease. J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from Case 2—A 52 year old man presented with a four week history of painless right sub- mandibular swelling. FNA showed a polymor- phous lymphoid population with scattered his- tiocytes consistent with a reactive lymph node (fig 1). Immunocytochemistry revealed an admixed T and B cell population with apparent polytypic immunoglobulin light chain expres- sion. ISH was technically unsatisfactory. PCR showed a polytypic pattern of immunoglobulin heavy chain rearrangement. The mass was clinically suspicious and therefore biopsied three weeks later, revealing a follicular lymphoma, centrocytic/centroblastic. On re- view, the cytomorphology was still felt to favour a reactive process. Case 3—A 43 year old man presented with bilateral groin lymph node enlargement. The cytological appearances and special techniques were similar to case 2, except that ISH was not performed. Biopsy, four weeks after FNA, revealed partial lymph node involvement by a follicular lymphoma, centrocytic/centroblastic. Case 4—A 67 year old man with a history of follicular lymphoma involving the scalp pre- sented with a small cervical lymph node. The Figure 1 Lymph node fine needle aspirate showing histiocytes (centre), small lymphocytes and occasional intermediate sized cells of probable follicle centre cell origin cytological appearances were similar to case 2. (May-Grunwald-Giemsa stain). Biopsy revealed follicular lymphoma. ISH showed no light chain expression. Biopsy one week later revealed recurrent follicular were classified histologically as nodular sclero- lymphoma, centrocytic/centroblastic. sis (n = 9), mixed cellularity (n = 3), lym- Case 5—A 75 year old man with a history of phocyte predominance nodular (n = 1), and oral squamous carcinoma presented with small follicular variant (n = 1) subtypes; one case was right cervical lymph nodes. FNA was inter- not further categorised histologically because preted as reactive lymphoid hyperplasia with only a core biopsy was received. no evidence of metastatic carcinoma. Special Although histological assessment was usu-

techniques were not performed. Peripheral http://jcp.bmj.com/ ally advocated to confirm a primary cytological blood and bone marrow examinations per- diagnosis of lymphoma, eight patients (six formed shortly thereafter revealed typical non-Hodgkin’s lymphoma, two Hodgkin’s dis- features of chronic lymphocytic leukaemia. ease) were considered medically unfit for Review of the aspirate showed a relatively lymph node biopsy. The FNA diagnoses were monotonous population of small lymphocytes consistent with the clinical and radiological consistent with chronic lymphocytic leukae- findings in these patients. In general, biopsy mia. was not performed in cases of recurrent Case 6—A 74 year old man presented with on September 29, 2021 by guest. Protected copyright. lymphoma. weight loss. Radiological investigations re- Clinical follow up data alone was consistent vealed intra-abdominal and intrathoracic lym- with the cytological diagnosis of reactive phadenopathy. Small palpable axillary nodes lymphoid hyperplasia in 181 of 213 cases. were aspirated, providing a sample of low Thirty two patients with clinically suspicious or cellularity but including non-caseating epithe- persistent underwent lymph lioid granulomas (fig 2A). A diVerential node biopsy. Histological assessment con- diagnosis including tuberculosis, sarcoidosis, firmed reactive changes in 26 cases, three of and neoplasia was oVered. The needle washes which showed a granulomatous lymphadenitis were submitted for culture and further studies due to tuberculosis. Biopsy in six cases revealed were not performed. Lymph node biopsy one malignant lymphoma; these false negative aspi- week after FNA revealed mixed cellularity rates are detailed below. Hodgkin’s disease with a marked granuloma- Case 1—A 21 year old, otherwise asympto- tous reaction (fig 2B). No Reed-Sternberg cells matic woman presented with a 1–2 cm right were identified on review of the aspirate mate- supraclavicular lymph node. The FNA speci- rial. men was of low cellularity but showed a mixed While most reactive aspirates showed a non- lymphoid population with occasional histio- specific admixture of lymphoid cells and cytes consistent with a reactive node. Special histiocytes, 11 cases showed a granulomatous techniques were not performed. The patient pattern. Seven of these patients had a subse- was discharged and presented again four quent clinical diagnosis of tuberculosis, four of months later with systemic symptoms and a whom showed positive mycobacterial culture 5 cm supraclavicular mass which, on repeat on the needle aspirate sample. One patient had FNA, was suspicious of Hodgkin’s disease, this clinical features of sarcoidosis and another, being confirmed on subsequent excision bi- who presented with bilateral groin lymphaden- 200 Stewart, Duncan, Farquharson, et al

cases, equivocal immunoreactivity with both

kappa and lambda antisera made definite J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from assessment of clonality impossible. Two aspi- rates showed a small cell malignant population in which immunocytochemistry was required to diVerentiate lymphoma from small cell carci- noma. The Reed-Sternberg cell associated anti- gens, CD15 and CD30, were demonstrated in most cases of Hodgkin’s disease. Immunocytochemistry was unhelpful in 10 aspirates, either as a result of insuYcient mate- rial or technically unsatisfactory staining, and was misleading in two cases, both of which were shown to be follicular lymphoma on biopsy (false negative cases 2 and 3). These cases showed an admixed B and T cell popula- tion in the cytospin preparations and immu- noglobulin light chain expression appeared polytypic. ISH was considered helpful in 47 of 61 cases. Immunoglobulin light chain restriction was Figure 2 Lymph node FNA showing a cluster of epithelioid histiocytes (A). Sparse small lymphocytes are present at the margin of the granuloma. Excision biopsy showing mixed seen in 16 of 24 B cell lymphomas and cellularity Hodgkin’s disease with a marked granulomatous reaction (B). Scant polytypic light chain expression was seen in 31 Reed-Sternberg cells (arrow) are present (haematoxylin and eosin stain). of 36 reactive lymph nodes. The technique was opathy, later had serological evidence of unhelpful in 13 cases because no light chain chlamydial infection. Excision biopsy in a expression was identified. ISH was potentially further case confirmed the FNA diagnosis of a misleading in only one case, a Burkitt-type granulomatous reaction to dust pigment in a lymphoma in which polytypic light chain patient with silicosis. The remaining granulo- expression was identified. matous aspirate proved to be from a case of PCR was considered to be helpful in 28 of 45 mixed cellularity Hodgkin’s disease (false cases. A clonal gene rearrangement was seen in negative case 6). nine of 18 B cell lymphomas and a polyclonal B In total, there were 61 true positive, 207 true cell pattern in 19 of 26 reactive lymph nodes. negative (reactive), three false positive, and six The technique was non-contributory in 10 false negative diagnoses of lymphoma in this specimens, because no rearrangement bands series. Therefore, the sensitivity and specificity were observed. PCR was misleading in seven for a diagnosis of lymphoma were 91% and cases; an apparent monoclonal band was seen

95%, respectively. Overall accuracy within the in two proven reactive lymph nodes, while a http://jcp.bmj.com/ series of lymphoid aspirates was 97%. polyclonal pattern was seen in five B cell lymphomas.

USE OF SPECIAL TECHNIQUES Discussion Special techniques were performed in 100 The distinction between reactive and malig- (36%) of the 277 FNA samples and, of these, nant lymphoid proliferations is the most prob- 49 had a final diagnosis of lymphoma and 51 lematical area in lymph node FNA cytology. had a final diagnosis of reactive lymphoid This is not surprising, given that excised lymph on September 29, 2021 by guest. Protected copyright. hyperplasia. Immunocytochemistry was per- nodes commonly cause diagnostic diYculty formed in 92 cases, ISH in 61, and PCR in 45. despite the advantage of architectural preserva- The value of the special techniques towards tion in biopsy specimens. Aspirate specimens the final clinicopathological diagnosis is sum- from cases of high grade lymphoma and Hodg- marised in table 2. Immunocytochemistry was kin’s disease may show an obvious cytomor- considered helpful in 80 of 92 cases. Typically, phological abnormality, but the diagnosis of reactive lymph nodes exhibited a mixed T and low grade lymphomas in cytological prepara- B cell population, and most cases showed poly- tions is most often based on the presence of a typic expression of immunoglobulin light chain. relatively monomorphic lymphoid population, In B cell lymphomas, immunochemistry usually contrasting with the typically polymorphous defined the B cell phenotype of the dominant or cell pattern seen in reactive proliferations.11 12 abnormal lymphoid element, although T lym- Therefore, potential cytological misdiagnoses phocytes were also present in variable number. may occur, either in lymphomas that present an Light chain restriction was demonstrated in apparently admixed cell pattern (false negative approximately one third of B cell lymphomas cases), or in reactive proliferations in which tested by immunochemistry while, in other atypical cells are identified (false positive cases). For these reasons, excision biopsy is Table 2 Use and value of special techniques in lymphoid aspirates advocated by most authors to confirm a Value of technique primary cytological diagnosis of lymphoma. However, FNA is being used increasingly to Helpful Unhelpful Misleading document recurrence of lymphoma or tumour Immunocytochemistry (n = 92) 80 10 2 transformation, to allow sampling of multiple In situ hybridisation (n = 61) 47 13 1 sites for staging, to exclude other unrelated Polymerase chain reaction (n = 45) 28 10 7 causes of lymphadenopathy, and to obtain FNA cytology in lymphoma and lymphoid hyperplasia 201

samples from surgically inaccessible sites or FNA specimens of B cell lymphomas by Sneige 2 medially unfit patients. In our series, eight of et al, who suggested that partial nodal involve- J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from 67 cases (12%) of lymphoma did not undergo ment by lymphoma might be responsible18 23; confirmatory biopsy because of poor general this was apparent histologically in one of our medical condition. However, precise cases. Follicular lymphomas may present par- classification of lymphoma was not required in ticular diYculty in FNA specimens because the these patients because most were fit only for neoplastic element itself is polymorphous (cen- general supportive therapy. In addition, the trocytes and centroblasts), and there may be a cytological diagnosis of reactive lymphoid significant population of reactive T lym- hyperplasia in the clinically appropriate setting phocytes and, less commonly, histiocytes. supports conservative management of many McNeely24 reviewed 14 histologically con- patients with lymphadenopathy, reducing the firmed follicular lymphomas in which FNA requirement for lymph node biopsy. cytology had been performed before biopsy; In this study of 549 consecutive lymph node four cases had been misinterpreted as reactive aspirates, 54% of satisfactory specimens lymphoid hyperplasia. Ten of 16 follicular lym- showed a lymphoid rather than metastatic phomas were misdiagnosed on aspiration cytol- process. Approximately one quarter of these ogy in the series of Pilotti and colleagues.15 samples were from cases of lymphoma, which Other authors have documented similar diY- accounted for 22% (67 of 302) of all malignant culties in the diagnosis of follicular lymphoma aspirates in the series. Therefore, lymphoid or, more generally, lymphomas of mixed cell specimens represented a substantial proportion type.21 22 Therefore, it is important that patholo- of all cases in our routine practice. However, the gists and clinicians are aware that negative FNA relative importance of lymphoma in FNA prac- results do not exclude lymphoma in patients tice is variable, being dependent both on the with unexplained lymph node enlargement, and patient population and local referral policy. Hsu that early repeat sampling by FNA or lymph et al reported only 13 lymphomas comprising node biopsy should be considered depending 1.8% of all malignant aspirates in a series of 735 on the clinical findings.2 Thus, four of the false lymph node aspirates from patients in Hong negative cases in our series were subject to exci- Kong.13 In contrast, Prasad and colleagues sion biopsy one to four weeks after FNA. In our found that lymphomas accounted for 31.6% of practice, clinically suspicious lymph nodes are malignant lymphadenopathy in a large series of sampled initially by FNA, usually at the Indian patients.14 These represented 11.8% of patient’s first outpatient visit. Smears from the all lymph node aspirates compared with 12.2% aspirate are examined at the clinic using the in our study. DiV-Quick staining method and a provisional Sixty one of 67 malignant lymphomas were diagnosis is oVered. If the appearances indicate identified correctly in this study, a sensitivity of a lymphoid rather than a metastatic process, a 91%. Other recent reports have also shown that clinical decision to obtain either a core biopsy or

FNA cytology is an accurate diagnostic proce- arrange formal excision biopsy of the node can http://jcp.bmj.com/ dure in most cases of lymphoma.14–22 However, be made. metastatic lymph node disease can be estab- Two erroneous reports in this study involved lished by aspiration cytology with even greater cases of Hodgkin’s disease and were primarily a sensitivity, illustrating the relative diYculty in result of inadequate sampling. In one case, the the assessment of lymphoid proliferations. The apparently benign initial aspirate probably apparent accuracy of lymphoma diagnosis is delayed further investigation of the patient’s also partly dependent on the proportion of pri- lymphadenopathy. Hodgkin’s disease may be mary to recurrent lymphomas in FNA cytology misinterpreted in cytology samples as Reed- on September 29, 2021 by guest. Protected copyright. series. In general, evaluation of recurrence is Sternberg cells and their variants may be rela- more straightforward owing to the raised clini- tively sparse or masked by the dominant poly- cal suspicion of lymphoid malignancy and the morphous infiltrate, which can mimic a comparison that can be made with previous reactive nodal hyperplasia.22 23 The presence of cytology or biopsy material. Patients with epithelioid granulomas, as in one of our cases, recurrent lymphoma accounted for 84% of may also suggest an infective aetiology.14 Sam- cases in the study by Sneige and colleagues,18 pling error is a particular hazard in the nodular approximately half the cases reported in two sclerosing Hodgkin’s disease subtype, possibly further studies,20 21 27% of cases in this series, because the fibrosis interferes with cell but only 15% of cases described by Pilotti and yield.12 25 Proportionately, Hodgkin’s disease colleagues.15 produced more diagnostic errors than non- There were six false negative errors in this Hodgkin’s lymphoma in this study (one false series. Three were follicular lymphomas and the positive and two false negative errors out of 17 false negative rate in this diagnostic category cases). The sensitivity of FNA in the diagnosis was therefore 19% (3 of 16). The smear prepa- of Hodgkin’s disease has been approximately rations from these cases showed a mixed 80% in most studies, although only nine of 30 lymphoid pattern with histiocytes, and even on cases were correctly identified in one recent review it was considered that the morphological series.14 appearances favoured a reactive rather than a There were three false positive fine needle neoplastic process. Immunocytochemistry and aspirates. One case of metastatic small cell carci- PCR were misleading in two of these cases noma mimicked a small cell lymphoma cytologi- because they suggested a reactive lymphoid cally, a well established diagnostic pitfall, which process. Inconclusive immunoglobulin light often requires immunocytochemical analysis for chain expression was also noted in occasional resolution.11 In this case, immunocytochemistry 202 Stewart, Duncan, Farquharson, et al

preparations were technically unsatisfactory. non-contributory in that no light chain expres-

The two further false positive cases were sion was detected. J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from reported as suspicious rather than diagnostic of The use of PCR to detect immunoglobulin lymphoma. Each showed a polymorphous lym- heavy chain gene rearrangement also oVers phoid population but with apparently dispro- genotypic analysis of B cell proliferations. PCR portionate numbers of large blast cells, mimick- has been used to detect monoclonality in ing recurrent Hodgkin’s disease in one patient. excised lymph nodes from patients with B cell Although over-interpretation of reactive lym- lymphomas29 and, more recently, the technique phoid aspirates might cause inappropriate con- has been applied to cytological material, cern, in our view it is better to advise biopsy in including serous fluid and FNA specimens.83031 aspirates exhibiting an atypical lymphoid pat- In this study, PCR was the least valuable tern, particularly if ancillary studies have shown technique, contributing to the diagnosis in 62% equivocal results. of tested cases. A higher cell yield was required The use of special techniques in diagnostic for PCR than for immunocytochemistry or ISH cytology specimens is well established and, as and there was a relatively high unsatisfactory in histopathology, has a particularly important rate, surprisingly, given the theoretical sensitiv- role in the assessment of lymphoproliferative ity of PCR. More importantly, PCR was poten- disorders.24Immunocytochemistry is the most tially misleading in 16% of cases, including two widely used technique, being of value in the false positive and five false negative results. The assignment of undiVerentiated neoplasms to former cases, which otherwise showed a reactive the lymphoid category and, more specifically, pattern on cytomorphology, led us to advise in documenting the phenotype of suspected lymph node excision biopsies in both patients. neoplastic lymphoid cells to B, T, null, As previously documented, one biopsy showed anaplastic, or Hodgkin’s disease related sub- reactive changes with features suggestive of types. This study confirmed the value of toxoplasmosis.8 The lymphadenopathy resolved immunocytochemistry in FNA specimens. The spontaneously in the second patient and, there- technique was used in 33% of all cases and fore, biopsy was not performed. It seems possi- contributed to the diagnosis in 87% of the ble that inadvertent sampling of a dominant tested samples. In two cases, immunocyto- clone within a reactive lymph node might have chemistry was essential to distinguish produced the apparent monoclonal gene rear- lymphoma from metastatic carcinoma. Immu- rangement pattern in these cases. The false nocytochemical staining was helpful in demon- negative cases in the series may also have strating mixed B and T cell populations and resulted from sampling error, although the sub- polytypic immunoglobulin light chain expres- sequently excised node showed partial involve- sion in many reactive lymph nodes, and in ment by lymphoma in only one case. Perhaps showing the dominant B cell phenotype of more importantly, the sensitivity of PCR analy- most B cell lymphomas. In addition, Reed- sis of immunoglobulin gene rearrangements

Sternberg cell related antigens, such as CD15 may be low when a limited number of primers is http://jcp.bmj.com/ and CD30, were demonstrated in many Hodg- used.32 This is particularly relevant in follicular kin’s disease samples. Immunoglobulin light lymphomas, which accounted for three of the chain restriction was shown in only one third of five false negative PCR cases. the B cell lymphomas tested in this series and, In summary, this study of 277 FNA cytology therefore, immunocytochemistry did not prove specimens showed an overall diagnostic accu- monoclonality in most of these tumours. Like- racy of 97% using cytomorphology in conjunc- wise, Pilotti and colleagues15 found immunocy- tion with appropriate ancillary investigations in tochemistry of limited value in the distinction the analysis of reactive and neoplastic lym- on September 29, 2021 by guest. Protected copyright. between reactive lymphoid hyperplasia and low phoid proliferations. However, a small grade B cell lymphoma. Immunochemistry can proportion of cases were misdiagnosed, either also produce equivocal or misleading results in because of sampling error or misinterpretation. cases with a small clonal population, as in T It is emphasised that lymph node FNA cell rich B cell lymphomas.18 26 However, Rob- cytology is complementary to histological ins and colleagues27 and Oertel and colleagues28 assessment and that biopsy is advisable in cases demonstrated light chain restriction in more with apparent clinical discrepancy. than 90% of B cell lymphomas in FNA samples using immunocytochemical methods. In situ hybridisation for detection of light 1 Skoog L, Lowhagen T, Tani E. Lymph nodes. In: Gray W, ed. Diagnostic cytology. Edinburgh: Churchill Livingstone, chain mRNA oVers increased specificity and, 1995:481–514. in our experience, also increased sensitivity 2 Katz RL. Cytologic diagnosis of leukemia and lymphoma. Values and limitations. Clin Lab Med 1991;11:469–99. compared with immunocytochemistry for the 3 Katz RL, Caraway NP. FNA lymphoproliferative diseases: demonstration of light chain restriction in B myths and legends. Diagn Cytopathol 1995;12:99–100. 9 4 Sneige N. Diagnosis of lymphoma and reactive lymphoid cell lymphoma. The technique contributed to hyperplasia by immunocytochemical analysis of fine-needle the final diagnosis in 47 of 61 (77%) tested aspiration biopsy. Diagn Cytopathol 1990;6:39–43. 5 Lubinsky J, Chosia M, Huebner K. Molecular genetic specimens in this series. In general, ISH prepa- analysis in the diagnosis of lymphoma in fine needle aspira- rations were easier to interpret than those tion biopsies. 1 Lymphomas versus benign lymphoprolif- erative disorders. Acta Cytol 1988;10:391–8. stained immunocytochemically because there 6 Williams ME, Frierson HF, Tabbarah S, et al. Fine needle was no background staining, which caused aspiration of non-Hodgkin’s lymphoma. Southern blot analysis for antigen receptor, bcl-2, and c-myc gene interpretative diYculty with the latter, and ISH rearrangements. Am J Clin Pathol 1990;93:754–9. was also the most specific technique in our 7 Wan JH, Sykes PJ, Orell SR, et al. Rapid method for detect- study, producing only one potentially mislead- ing monoclonality in B cell lymphoma in lymph node aspi- rates using the polymerase chain reaction. J Clin Pathol ing result. However, 13 specimens were 1992;45:420–3. FNA cytology in lymphoma and lymphoid hyperplasia 203

8JeVers MD, McCorriston J, Farquharson MA, et al. Analysis 21 Daskalopoulou D, Harhalakis N, Maouni N, et al. Fine nee- of clonality in cytologic material using the polymerase dle aspiration cytology of non-Hodgkin’s lymphomas. A chain reaction. Cytopathology 1997;8:114–21. morphologic and immunophenotypic study. Acta Cytol J Clin Pathol: first published as 10.1136/jcp.51.3.197 on 1 March 1998. Downloaded from 9 Stewart CJR, Farquharson MA, Kerr T, et al. Immu- 1995;39:180–6. noglobulin light chain mRNA detected by in situ 22 Steel BL, Schwartz MR, Ramzy I. Fine needle aspiration hybridisation in diagnostic fine needle aspiration cytology biopsy in the diagnosis of lymphadenopathy in 1,103 specimens. J Clin Pathol 1996;49:749–54. patients. Role, limitations and analysis of diagnostic pitfalls. 10 Skoog L, Tani E. The role of fine-needle aspiration cytology Acta Cytol 1995;39:76–81. in the diagnosis of non-Hodgkin’s lymphoma. Diagn Oncol 23 Sneige N, Dekmezian R, El-Naggar A, et al. Cytomorpho- 1991;1:12–8. logic, immunocytochemical, and nucleic acid flow cyto- 11 Leong ASY, Stevens M. Fine-needle aspiration biopsy for metric study of 50 lymph nodes by fine-needle aspiration. the diagnosis of lymphoma: a perspective. Diagn Cytopathol Comparison with results obtained by subsequent excisional 1996;15:352–7. biopsy. 1991; :1003–7. 12 Pontifex AH, Haley L. Fine-needle aspiration cytology in Cancer 67 lymphomas and related disorders. Diagn Cytopathol 1989; 24 McNeely TB. Diagnosis of follicular lymphoma by fine nee- 5:432–5. dle aspiration biopsy. Acta Cytol 1992;6:866–8. 13 Hsu C, Leung BSY, Lau S-K, et al. EYcacy of fine-needle 25 Ramzy I, Rone R, Schultenover SJ, et al. Lymph node aspira- aspiration and sampling of lymph nodes in 1,484 Chinese tion biopsy. Diagnostic reliability and limitations—an analy- patients. Diagn Cytopathol 1990;6:154–9. sis of 350 cases. Diagn Cytopathol 1985;1:39–45. 14 Prasad RPA, Narasimhan R, Sankaran V, et al. Fine-needle 26 Galindo LM, Havlioglu N, Grosso LE. Cytological findings aspiration cytology in the diagnosis of superficial in a case of T-cell rich B-cell lymphoma: potential diagnos- lymphadenopathy: an analysis of 2,418 cases. Diagn tic pitfall in FNA of lymph nodes. Diagn Cytopathol Cytopathol 1996;15:382–6. 1996;14:253–7. 15 Pilotti S, Di Palma S, Alasio L, et al. Diagnostic assessment 27 Robins DB, Katz RL, Swan F, et al. Immunotyping of of enlarged superficial lymph nodes by fine needle lymphoma by fine-needle aspiration. A comparative study aspiration. Acta Cytol 1993;37:853–66. of cytospin preparations and flow cytometry. Am J Clin 16 Das DK, Gupta SK, Datta BN, et al. FNA diagnosis of non- Pathol 1994;101:569–76. Hodgkin’s lymphoma and its subtyping under working for- 28 Oertel J, Oertel B, Kastner M, et al. The value of immuno- mulation of 175 cases. Diagn Cytopathol 1991;7:487–98. cytochemical staining of lymph node aspirates in diagnostic 17 Cartagena N, Katz RL, Hirsch-Ginsberg C, et al. Accuracy cytology. Br J Haematol 1988;70:307–16. of diagnosis of malignant lymphoma by combining 29 Ashton-Key M, Diss TC, Isaacson PG, et al. A comparative fine-needle aspiration cytomorphology with immunocyto- study of the value of immunohistochemistry and the chemistry and in selected cases, Southern blotting of aspi- polymerase chain reaction in the diagnosis of follicular rated cells: a tissue-controlled study of 86 patients. Diagn lymphoma. 1995;27:501–8. 1992;8:456–64. Histopathology Cytopathol 30 Trainor KJ, Brisco MJ, Wan JH, Gene rearrangement 18 Sneige N, Dekmezian RH, Katz RL, et al. Morphologic and et al. immunocytochemical evaluation of 220 fine needle aspi- in B- and T- lymphoproliferative disease detected by the rates of malignant lymphoma and lymphoid hyperplasia. polymerase chain reaction. Blood 1991;78:192–6. Acta Cytol 1990;34:311–22. 31 Chen Y-T, Mercer GO, Chen Y. Polymerase chain reaction- 19 Qizilbash AH, Elavathil LJ, Chen V, et al. Aspiration biopsy based detection of B-cell monoclonality in cytologic speci- cytology of lymph nodes in malignant lymphoma. Diagn mens. Arch Pathol Lab Med 1993;117:1099–103. Cytopathol 1985;1:18–22. 32 Segal GH, Jorgensen T, Scott M, et al. Optimal primer 20 Carter TR, Feldman PS, Innes DJ, et al. The role of fine selection for clonality assessment by polymerase chain needle aspiration cytology in the diagnosis of lymphoma. reaction analysis: II. Follicular lymphomas. Hum Pathol Acta Cytol 1988;32:848–53. 1994;25:1276–82 http://jcp.bmj.com/ on September 29, 2021 by guest. Protected copyright.