Methods

The FNAs were performed by one of two mobile pathologists with an interest in FNA cytology, except for image guided FNAs, in which case one of these pathologists prepared the slides after the radiologist had aspirated the lesion. The FNAs performed by the pathologists were done with a needle only technique, using a 23 or 25 gauge 1.25 inch (3 cm) needle. The slides were air dried, often aided by a small hair dryer, and Diff-Quik^™ stained (Dade Behring Diagnostics, Newmarket, Auckland, New Zealand). The slides were then examined with a portable Olympus CHK microscope and a hand written report was issued within a few minutes. If this examination of the specimen suggested that the lesion might be lymphoid and lymphoma seemed possible, then a second FNA was performed, still using a needle only technique, to collect a second sample. This sample was collected quickly and the needle was washed through with 3 ml of heparin RPMI (12.8 mg of heparin ammonium in 45 ml of RPMI) within a few seconds of being collected. The reason for the urgency was that a clotted sample reduces the yield of cells for flow cytometry. Early in our experience with collecting specimens for flow cytometry we washed needles through with sterile heparin saline before the FNA was performed, but we found that this was not necessary. The slides were taken back to the laboratory to be mounted and examined again. The specimen in RPMI was taken to the haematology laboratory, transferred to a 5 ml tube and centrifuged for two minutes at approximately 400 ×g. The supernatant was discarded and the cells were resuspended in 2 ml of ammonium chloride lysis solution. The cells were gently vortexed, incubated at room temperature for 10 minutes, and then centrifuged again for two minutes. The leucocytes were then resuspended in 2 ml of phosphate buffered saline (PBS). If debris was present, it was removed with a nylon Swiss screen filter. An approximate cell count was performed and a panel of directly conjugated monoclonal antibodies was selected (table 1) (Immunotech Inc, Westbrook, Maine, USA). The cells and the antibodies were incubated for 20 minutes in the dark at room temperature. These were then washed once in PBS, spun down, and then washed again. They were then analysed with the flow cytometer. Material for κ and λ light chain analysis was treated differently. The centrifuged sample in RPMI was incubated with prewarmed PBS for 20–30 minutes at 37°C, to remove the cytophilic immunoglobulin, and then centrifuged. The supernatant was discarded and if red blood cells were present they were lysed with ammonium chloride as above. Cells were then incubated with antibodies (against CD19/κ light chain or CD19/λ light chain) as above.

The “first run” panel of antibodies evolved as our experience developed and with the acquisition of a new immunoflow cytometer approximately half way through our study. At the start of the study, a Coulter PROFILE II was used but later the laboratory obtained a Coulter EPICS^™ XL (Beckman Coulter Inc, Fullerton, California, USA). In some cases we performed a “second run” with a more specialised panel to investigate a specific diagnosis. The panel of antibodies used was sometimes restricted by the scarcity of cells. In these cases, antibodies against CD19/κ light chain and CD19/λ light chain were given priority.

In our laboratory we have adopted light chain ratio limits from previous studies.^2–4 A κ to λ ratio of greater than 3 or a λ to κ ratio of greater than 2 was accepted as evidence of monoclonality.

For some cases, cell blocks for immunohistochemistry were collected as described previously.⁵ These were usually taken after the initial FNA showed overtly malignant cytology but it was not certain that the lesion was lymphoid in nature.

Results

Table 2 shows our results—the cases are set in chronological order within each diagnostic category. It can be seen that the range of cell markers expanded during the course of our study. The panel of markers was sometimes tailored a little in view of the initial diagnosis, so as to focus on a particular issue and to avoid wasting cells.

Discussion

Nine of the 73 samples (12%) had inadequate cells for analysis. Four of these were from benign lymph nodes. It was felt that adequate samples were easier to obtain from malignant lesions simply because of their larger size. Young et al found that five of 107 (4.7%) of their specimens were inadequate in a similar study.² A highly cellular lymphoma aspirate might exhibit uniform negativity on immunoflow cytometry if the cells are necrotic, as in case 3.

Twenty eight cases exhibited light chain restriction (defined above), allowing an unequivocal diagnosis of B cell lymphoma to be made on the FNA sample alone. On review of the notes at the end of our study, we could find no evidence that any of these diagnoses were incorrect. Case 48 had been thought to be suspicious of lymphoma with a κ to λ ratio of 1.95. However, a review of the notes about a year after the FNA was performed revealed no clear evidence that the patient had developed lymphoma. In particular, there was no evidence of lymphoma on a computed tomography scan. Case 73 also had a marginal light chain ratio but did not have clinically evident lymphoma eight months later.

Five of the 33 B cell lymphomas did not exhibit light chain restriction. For three case this was because of inadequate cellularity. The other two (cases 9 and 33) were particularly confusing because the cytology showed a malignant tumour with highly atypical cells mixed with benign looking lymphoid cells. The flow cytometry showed no evidence of light chain restriction. Biopsies were taken and the flow cytometry results from these were similar to those from the FNA. However, paraffin wax embedded sections and immunohistochemistry showed that the large highly atypical cells were CD45 and CD20 positive and negative for cytokeratin, S100, and CD3. Thus, a diagnosis of T cell rich B cell lymphoma was made. We assumed that the neoplastic B cells were not expressing light chains strongly or were “swamped” by a population of reactive B cells.

Except for cases 9 and 33, the lowest light chain ratio of any B cell lymphoma was a λ to κ ratio of 3.1 (case 16). Case 37 was remarkable in that it showed artefactual light chain restriction in a T cell lymphoma (fig 1). We assume that this was as a result of specific binding of κ to the surface of the tumour cells, perhaps as an autoimmune response to the tumour cells. The lymphoma occurred in a patient with AIDS. On biopsy, the tumours cells were strongly positive for CD3 but negative for CD20. Fortunately, this artefact did not confuse the diagnosis, because coexpression of light chain with CD19 was not found.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A scatter plot showing κ light chain restriction in a T cell lymphoma from a patient with AIDS (case 37). There was no coexpression of CD19 and κ, exemplifying the importance of dual staining in assessing light chain restriction.

For 18 cases it was possible to make a diagnosis of B cell lymphoma and also subtype the lymphoma as either CLL (chronic lymphocytic leukaemia/small lymphocytic lymphoma), FCCL (follicle centre cell lymphoma), or mantle cell lymphoma. Most of the other cases that exhibited light chain restriction were called “B cell lymphoma, exact type uncertain” or “diffuse large B cell lymphoma” if they were composed of overtly malignant large cells. Uncertainty as to the exact type of B cell lymphoma was not considered sufficient reason to perform a biopsy. This was usually because the patient had other medical problems that were a more immediate risk to the patient's health or because exact typing would not alter treatment.

Four of the six T cell lymphomas were diagnosed with a combination of cytology, flow cytometry, and cell block immunohistochemistry, but for two it was not possible to collect a cell block. Both of these then had biopsies. The participation of the pathologist in the collection of the specimen and the immediate assessment of the cytology meant that the need for cell block histology had been anticipated. The ratio of CD4 to CD8 (or CD8 to CD4) was examined but this was not usually useful in the recognition of T cell lymphoma. The benign lesions in cases 59 and 65 had CD4 to CD8 ratios of 4.15 and 4.29, respectively. Only two of the six T cell lymphomas had CD4 to CD8 (or CD8 to CD4) ratios greater than these. Jeffers et al describe one case of T cell lymphoma in their series of 46 FNA sampled lesions.⁶ This lymphoma showed subset restriction, with more than 90% of T cells expressing CD8. Coexpression of CD4 and CD8 was examined for four of the six T cell lymphomas but was only found in case 34. Loss of pan-T cell antigens was seen in cases 34 and 37. Zardawi et al have stated that the loss of one or more pan-T cell antigen allows a firm diagnosis of T cell lymphoma to be made (in the correct clinical context).⁴ It should be remembered that immature thymic T cells exhibit loss of pan-T cell antigens, as in case 44.

The cytological diagnosis of Hodgkin's lymphoma depends on the presence of typical Reed-Sternberg cells. Flow cytometry does not contribute except to exclude a monoclonal B cell population. There were no definite Reed-Sternberg cells in the FNA samples of three of our four cases. Because a diagnosis of lymphoma was suspected clinically, the nodes were biopsied. One case did show abundant Reed-Sternberg cells in the FNA sample and a diagnosis was made cytologically. The node was excised before treatment and the diagnosis was confirmed. We would not advocate the diagnosis of Hodgkin's lymphoma being made on FNA alone except in exceptional circumstances. Young et al studied the cytology and flow cytometry of 107 aspirates of suspected lymphoid lesions.² Three of these were Hodgkin's lymphoma but none showed Reed-Sternberg cells in the FNA samples. Flow cytometry showed a polyclonal population.

There were four cases that had a final diagnosis of a non-lymphoid neoplasm. The initial differential diagnosis favoured a non-lymphoid tumour in each case but lymphoma was also thought to be possible. In one of these patients, a core biopsy was performed by a radiologist once the cytology had been examined. Cell blocks were taken for the others.

There were 28 cases that seemed to be benign on cytology and flow cytometry. However, two of these were clinically suspicious and had biopsies that showed them to be Hodgkin's lymphoma. None of the other patients had a biopsy and none had evidence of lymphoma when the notes were reviewed at the end of our study.

Before the widespread availability of immunoflow cytometry, immunofluorescent cytochemistry on cytospin preparations was used in some large laboratories for the characterisation of lymphomas from FNA samples. This has the advantage that it can be performed with fewer cells and it is not difficult to appreciate the relation between the size of the cells and their antigen expression. The latter makes it easier to diagnose some conditions such as T cell rich B cell lymphomas. Immunocytochemistry has some major disadvantages. The technique is highly labour intensive and requires specially trained technical staff. Dual expression of markers is less easy to demonstrate. A result would usually take considerably longer than the four hour turn around time that we have for immunoflow cytometry. The number of cells that can be counted and the number of markers that can be used are restricted by the time it takes to perform manual counting. A higher light chain ratio (about 6 : 1) is required to prove monoclonality,⁷ although in practice this rarely alters the detection of monoclonality in a sample.⁸ For these reasons, immunoflow cytometry is probably the preferred method except in a few specialist centres.

Young et al graded the follicle centre cell lymphomas in their study² according to the proportion of “transformed cells” in the smear. Their method was validated in a previous study.⁹ We did not attempt to do this because it would rarely change the way in which the patient was treated.

Eight of the 47 malignant lesions were diagnosed using cell blocks. These comprised one case of myeloma, four of the six T cell lymphomas, and three of the four non-lymphoid malignancies. This demonstrates the role of cell block immunohistochemistry in the diagnosis of suspected lymphoid neoplasms. Although immunoflow cytometry is highly effective in the diagnosis of B cell lymphomas, particularly CLL, FCCL, and mantle cell lymphoma, cell block immunohistochemistry is probably more useful for “high grade lymphomas”, especially peripheral T cell lymphomas, and also for non-lymphoid neoplasms. If a sample from a suspected lymphoid lesion is overtly malignant on cytological criteria then more information is likely to be obtained from a cell block, but if the sample is equivocally benign/malignant then immunoflow cytometry is probably more useful. Ideally, tissue should be obtained for immunoflow cytometry and a cell block, although in practice this is often not possible.

Optimum diagnostic information might be best obtained by biopsying every suspicious lymph node and performing a battery of sophisticated investigations including oncogene expression, cell kinetic studies, and detailed molecular analysis. A biopsy might be particularly important if spare material is needed for research. However, only a few centres have the budget, facilities, or expertise to do all of this. We would always advocate that an FNA diagnosed lymphoma in an otherwise fit young patient should be followed by a biopsy. Most lymphomas occur in elderly patients and are not curable. In these patients, treatment and survival are often influenced more by coexistent pathology, so that detailed analysis of the lymphoma might supply the clinician with a level of detail that is superfluous. Despite the pitfalls that were apparent in our study, FNA with immunoflow cytometry is a powerful technique for the diagnosis and typing of B cell lymphomas. Close communication between the oncologist and pathologist is needed to ensure that FNA sampling is performed on appropriate cases and that the result and issues of uncertainty are conveyed. In the setting of a large general hospital in rural New Zealand, the speed with which FNA cytology and immunoflow cytometry could be performed (about four hours) was important. It allowed patients who had travelled a long distance to be given their result quickly, and allowed other investigations to be undertaken the same day in preparation for treatment.

Conclusions

Our study shows that it is almost always possible to obtain an adequate sample for immunoflow cytometry from a fine needle aspirate. Light chain restriction could be demonstrated in almost all FNA samples from B cell lymphomas. One notable exception was T cell rich B cell lymphoma. Artefactual light chain restriction could occasionally be seen in T cell lymphomas, presumably because of autoantibodies binding to the cell surfaces. It was possible to subtype most B cell lymphomas as CLL, FCCL, or mantle cell lymphoma. Immunoflow cytometry is not particularly helpful in the diagnosis of T cell lymphoma or Hodgkin's lymphoma. The CD4 to CD8 ratio was not usually restricted in T cell lymphomas and neither was coexpression of CD4 and CD8 usually found. Loss of pan-T cell antigens was seen in some T cell lymphomas. Four of the six T cell lymphomas and three of the four non-lymphoid malignancies were diagnosed with the aid of cell block immunohistochemistry. We advocate the use of cell block immunohistochemistry in preference to immunoflow cytometry for cases in which the cytological appearance of the specimen is overtly malignant but the differential diagnosis includes non-lymphoid malignancy.