Flow cytometric immunophenotyping and cell block immunocytochemistry in the diagnosis of primary Non-Hodgkin's Lymphoma by fine-needle aspiration: Experience from a tertiary care center.

BACKGROUND: Accurate diagnosis of Non-Hodgkin's Lymphoma (NHL) on fine-needle aspiration (FNA) specimen is challenging and requires ancillary testing. AIM: The feasibility of flow cytometric immunophenotyping (FCI) along with cell block immunocytochemistry (CB-ICC) as adjunct techniques in the diagnosis of NHL as per the current World Health Organization (WHO) classification was evaluated.
MATERIALS AND METHODS: All cases of suspected lymphoma underwent FNA, and the sample was triaged for light microscopic evaluation, FCI, and CB-ICC, and each case was classified as per the current WHO classification.
RESULTS: A total of 65 cases was analyzed which included 40 B-cell, 21 T-cell, and 4 unclassifiable lymphomas. Of 61 cases, FCI alone was contributory in 74% (45/61) cases whereas CB-ICC alone was contributory in 65.5% (40/61) cases in typing the lymphoma. In 11.4% (7/61) cases, the lymphoma could not be classified by either technique. Thus, in a total of 88.5% (54/61) cases a combination of FCI and CB-ICC from FNA enabled a diagnosis of lymphoma with its subtyping.
CONCLUSION: Flow cytometric immunophenotyping and ICC on CBs are feasible on FNA material and are very useful in a suspected case of NHL especially when a biopsy may not be possible or feasible.

Introduction

The diagnosis of Non-Hodgkin's Lymphoma (NHL) on cytological material, especially fine-needle aspirates (FNA), has evolved over the last three decades. The modern classification of NHL according to World Health Organization/Revised European American Lymphoma (WHO/REAL) is based on morphology, immunology, cytogenetic, and molecular studies. FNA biopsy material, which is an excellent source for cytomorphology, immunology, and cytogenetics, can thus be used to render a primary diagnosis of lymphoma; however, the question is whether it can sub classify NHL according to the REAL/WHO classification as per the demands of clinical oncologic practice. There have been several studies in this field particularly in the last decade which have shown that NHL can be sub-classified in the vast majority of cases as per the classification applicable in those times, using a combined cytomorphological and flow cytometric immunophenotyping (FCI) based approach;[1234567891011] however, there are no relevant studies from India in this area. Further, the differences in the types of lymphomas occurring in India as compared to the Far East or the West as has been highlighted.[1213] Hence, it is important to evaluate the contribution of the ancillary techniques in classification of NHL in our setup. This prospective study was designed to evaluate the feasibility and utility of FCI and cell block immunocytochemistry (CB-ICC) on FNA samples in making a primary diagnosis of lymphoma as per the 2008 WHO classification.[14]

Materials and Methods

The study was carried out after approval by the Institute Ethics Committee and after obtaining an informed consent. 75 cases of suspected lymphoma referred for fine-needle aspiration cytology between January 2010 and June 2011 were included. The sample for flow cytometry and CB was collected at the time of the initial FNA performed to obtain a cytopathological diagnosis and no extra procedure was carried out. Ten cases with a cytological diagnosis other than NHL (7 acute lymphoblastic leukemia, 1 reactive hyperplasia, and 2 Hodgkin lymphoma) were excluded from the study. Follow-up histopathology of the lymph node excision biopsy was available in 27 cases (3 precursor T-lymphoblastic lymphoma (PT-LBL), 4 Anaplastic large cell lymphoma (ALCL), 4 mantle cell lymphoma (MCL), 3 small lymphocytic lymphoma (SLL)/chronic lymphocytic leukemia (CLL), 7 diffuse large B-cell lymphoma (DLBCL), 1 Burkitt lymphoma, 1 follicular lymphoma, 4 unclassified). In 18 more cases, the bone marrow trephine biopsy histopathology showed infiltration by lymphoma. In the remaining 20 cases, the only histopathological evidence of lymphoma was from CBs prepared from the FNA material. All cases without histological evidence of lymphoma available either by lymph node excision biopsy, bone marrow trephine biopsy or a satisfactory CB were excluded from the analysis. The final diagnosis was obtained by integrating the clinical features, histopathological diagnosis with immunohistochemistry and hematological findings. Thus, 65 cases with a definitive diagnosis of NHL and which had been further classified as per the WHO classification were included for final analysis. FCI was performed in all 65 cases. CBs were prepared in all cases. In 15 cases, the section showed no material and was unsatisfactory. Thus, in 50/65 cases, CBs were satisfactory and ICC could be performed for evaluation and comparison.

Sample collection for ancillary techniques

Routine FNA was performed using a 23 gauge needle attached to a 20 mL syringe in a specially designed holder (Cameco, AB Taby, Sweden). A combination of air-dried smears (stained by May-Grünwald-Giemsa stain) and alcohol-fixed smears (stained by hematoxylin and eosin stain) was evaluated by light microscopy.

Flow cytometric immunophenotyping

Dual or three color immunophenotyping was performed. All antibodies were procured from BD Biosciences, San Jose, California, USA. The markers included CD45 (pan-leucocyte marker), those directed against B-cell ontogeny — CD20, CD19, CD79a, CD10, CD23, and against kappa and lambda light chain; against T-cell ontogeny — CD3 (surface as well as cytoplasmic), CD5, CD4, CD8; CD34 and TdT (immaturity markers). Antibodies against CD30, CD15, and Ki-67 were used whenever ALCL was suspected on cytomorphology.

Fine-needle aspiration material was collected in phosphate buffer (1-2 mL) containing ethylene diamino tetra acetic acid (EDTA) or heparin. This was centrifuged to obtain a cell pellet, to which was added 5 mL of lysing solution (NH4 Cl, KHCO3, tetrasodium EDTA, distilled water [Becton Dickinson]), and incubated for 10 min followed by three washings with phosphate-buffered saline (PBS) (containing sodium nitrite additionally). The cell pellet was re-suspended in an appropriate volume of PBS (0.5-1 mL) and 50 μL (containing 50,000-100,000 cells) of this was distributed into individual tubes. To each tube, primary antibody/antibodies labeled with appropriate fluorochrome was added as per manufacturer's instructions and incubated in dark for 30 min at room temperature. An isotype-matched antibody served as negative control. This was followed by three washes with PBS. Finally, the pellet was re-suspended in 250 μL of PBS, mixed thoroughly, and cells were ready for acquisition by the flow cytometer. For detection of nuclear and cytoplasmic antigens, cell permeabilization was carried out with cold methanol for 15 min after fixation in 1% paraformaldehyde. The stained cells were acquired on FACS Canto II (Becton Dickinson, California, U.S.A.) The debris and clumps were excluded with the help of doublet discrimination module. Initial gating of the cells was carried out using forward scatter and side scatter properties expected for the cells analyzed (as assessed morphologically) and further gated for CD45 positivity. Cell viability was assessed by initial cytomorphological evaluation of the smears. In cases of poor cell viability, FCI was not performed. Further, CD45 positivity in at least 50% cells was taken as adequacy criteria before further evaluation of other markers by flow cytometry. For assessment of clonality (kappa vs. lambda light chain expression), CD19 gating was carried out. A minimum of 10,000 cells were gated for analysis of each parameter. The data were analyzed by the Cell Quest software (Becton Dickinson Immunocytometry systems, San Jose, California, USA).

Cell block immunocytochemistry

Cell blocks were prepared from the aspirated material after fixation in 10% buffered formalin by Plasma Thrombin clot method.[15] Routine immunoperoxidase method using Envision systems (Dako, Sweden) and antibodies against CD45, CD3, CD20, CD5, CD10, CD23, Ki-67, kappa, lambda light chain, EMA, CD15, CD30, Alk protein (Dako, Sweden), and c-Myc (Santacruz, CA, USA).

Results

The age range of the cases included ranged from 4 to 75 years with a mean age of 39 years and a median age of 45 years. The clinical, histopathological, and immunohistochemical data were integrated and all the cases of NHL (n = 65) were classified as per the WHO classification of hematolymphoid neoplasms. Four cases could not be classified even after histopathology and immunohistochemistry. Thus, of the 61 cases classified, there were a total of 40 B-cell lymphomas (65.57%) and 21 T-cell lymphomas (34.43%).

Precursor T-lymphoblastic lymphoma (n = 14)

Cytomorphology revealed a relatively monomorphic population of atypical lymphoid cells of small to intermediate size displaying nuclear notches and clefts; chromatin was fine and nucleoli were indistinct.

A total of 14 cases were evaluated by FCI, out of which, 2 cases were not interpretable, 1 for poor cell viability, and 1 in which all markers were negative. The remaining 12 cases were evaluable. A representative case is illustrated in Figure 1. CD3 was positive in 12/14 cases; CD5 was positive in 5 cases. CD4 and CD8 dual positivity was seen in 6 cases. TdT was positive in 8 cases. If either dual CD4/8 positivity or TdT positivity was taken as conclusive to a diagnosis, then overall FCI was contributory in 85.7% (12/14) cases.

Figure 1

Selected flow cytometric immunophenotyping plots from fine-needle aspiration samples of lymphomas. Top panel illustrates a forward and side scatter plot confirming good cellularity. The negative control and CD45 positive controls are shown. Representative case of diffuse large B-cell lymphoma CD20+/kappa+; of precursor T-lymphoblastic lymphoma shows positivity for T-cell markers CD2+/CD3+/CD5+/dual CD4 and 8+/TdT+; case of anaplastic large cell lymphoma shows CD45+/CD30+ and CD3+; case of Burkitt's lymphoma showing CD19+/CD10+/CD20+/lambda+; case of small lymphocytic lymphoma/chronic lymphocytic leukemia showing CD19 gated cells which are dual CD5+/CD23+/kappa+

CB-ICC [Figure 2] could be performed in 10/14 cases; it was not evaluable in 2 cases with necrosis and 2 with poor cellularity. The remaining 10 cases showed strong CD3+/CD20− and were positive for TdT in 8 cases.

Figure 2

Panel of microphotographs illustrating immunocytochemistry on cell blocks from lymphomas: Precursor T-lymphoblastic lymphoma (a, d, g, j) a: smear; d: Cell block; g: CD3 and j; CD20. Anaplastic large cell lymphoma (b, e, h, k) b: Smear with hallmark cell; e: Cell block; h: CD30 and k; CD3. Diff use large B-cell lymphoma (c, f, i, l) c: Smear; f: Cell block; i: CD20 and l: CD3. (a: ×200, b: ×400, c: ×200, h and e, stain; d: ×200, e: ×200, f: ×400, h and e, stain; g-l, immunoperoxidase with envision, g: ×400, h: ×200, i: ×200, j: ×200, k: ×200, l: ×200)

Anaplastic large cell lymphoma (n = 7)

Smears showed a polymorphic population of lymphoid cells with many scattered atypical large cells which were predominantly mononucleated. Bi- and multinucleation was observed. There was marked variation in nuclear shape and size. Typical “embryoid” nuclei were also seen in most cases along with scattered doughnut nuclei and those having wreath-like nuclei. Nucleoli were prominent in many nuclei. In 2 cases, many eosinophils were appreciated in the background.

On FCI an immunophenotype of CD45+/CD3+/CD30+ was observed in 5/7 cases (71%) cases of ALCL) [Figure 1]. Variable positivity for T cell antigens was observed; CD15 showed positivity in 2 cases.

CB-ICC showed CD20(−)/CD30(+) immunophenotype [Figure 2] in 5 cases and Alk protein positivity in the large atypical cells. In 1 case, CD15 was positive in addition to CD30 in the atypical cells. The background reactive population of cells were variably positive for CD3, CD20, and CD15 (neutrophils).

Burkitt lymphoma (n = 4)

Smears showed intermediate-sized lymphoid cells with high nucleo-cytoplasmic ratio and basophilic cytoplasm displaying fine vacuolation. Nuclear chromatin was fine with 1-2 nucleoli and fine vacuolation. The mitotic rate was high.

On FCI the lymphoma cells showed CD19+/CD20+/CD10+ phenotype. Light chain restriction was seen in all the 4 cases (3 lambda and 1 kappa). Thus, flow cytometry was contributory in all the 4 cases (100%).

CB-ICC was contributory in all 4 cases with tumor cells showing CD3−/CD20+ phenotype associated with a high Ki-67 index; c-Myc positivity was observed in 3 out of 4 cases.

Diffuse large B-cell lymphoma (n = 17)

The cellularity was moderate to high with numerous large atypical cells having a high nucleo-cytoplasmic ratio, coarse chromatin, and having one or more prominent nucleoli. Karyorrhexis and mitosis were frequent and necrosis was occasionally seen.

Of 17 cases evaluated by FCI, 3 showed poor cell viability. In the remaining 14 cases, 6 showed kappa restriction, 3 showed lambda restriction, and in the remaining 5 cases, light chains were not demonstrable. The B-cell markers, CD 19/20 was positive in all 14 cases; CD10 and CD23 were positive in 8 and 5 cases, respectively. A representative case is illustrated in Figure 1. In 1 case, no light chain restriction was seen but all the lymphoid cells were positive for B-cell markers CD19/CD20 and were negative for CD3. In the other 4 cases, no light chain restriction was observed and hence, FCI was considered noncontributory. Hence, FCI was contributory in 10 cases out of 17 (59%).

CB-ICC was contributory in 12 out of 17 cases (70.5%), with CD20(+)/CD3(−) phenotype in the large atypical cells confirming DLBCL [Figure 2].

B-cell lymphoma, small cell type (19 cases)

Small cell lymphomas show predominance of small lymphoid cells which are only slightly larger than normal lymphocytes. Follicular lymphomas display an admixture of cell types whereas SLL/CLL display a monotonous population of cells with clumped chromatin. MCLs showed a monotonous population of small lymphoid cells with a few nuclei displaying angulation. Lymphoplasmacytic lymphoma showed an admixture of small lymphocytes and plasma cells.

On FCI the diagnosis of follicular lymphoma was established by the CD10+/CD5−/CD23− phenotype in 3 cases. The dot plots were technically unsatisfactory for evaluation in the other 2 cases due to low cellularity of sample submitted for flow cytometric analysis. Out of 4 cases of MCL confirmed on excisional biopsy and histopathology and by cyclin D1 immunohistochemistry, FCI was contributory in 3 cases with CD5+/CD10−/CD23− phenotype and suggestive of MCL in 1 case wherein light chain restriction was not demonstrable although CD5 was positive. Out of 10 cases of CLL/SLL, 9 cases were successfully immunophenotyped by FCI with an immunophenotype of CD10−/CD5+/CD23+; however, in 3 cases light chain restriction was not demonstrable and in 1 case, CD5 could not be demonstrated.

Cell block could not be made in 9 cases as no material was available. In follicular lymphoma on FCI, two out of four cases showed CD20+/CD10+/CD5−/Bcl2+ expression pattern. All four cases of MCL showed CD5+/CD23−/CD10− expression pattern. Four out of 10 cases of SLL/CLL and showed CD5+/CD23+/CD10− expression pattern.

Overall contribution of flow cytometric immunophenotyping and cell block immunocytochemistry in lymphoma typing. The contribution of FCI was evaluated and the results are compiled in Table 1. In the lymphomas which were classifiable, a combination of cytomorphology and FCI was able to render an accurate diagnosis in 73.7% (45/61) cases. In 9.8% (6/61) cases, FCI could classify the lymphoma as a B-cell lymphoma, but was not able to type it further. It was noncontributory in 26.3% (16/61) cases. Poor cell viability due to cellular necrosis in the area sampled, failure to demonstrate light chain restriction, and other markers due to technical reasons were the main causes of a noncontributory FCI.

Table 1

Contribution of FCI in the classification of Non-Hodgkin's Lymphoma

Next, the combination of FCI and CB-ICC in arriving at a specific diagnosis of lymphoma as per the WHO classification was analyzed in cases where comparison with the lymph node excision biopsy or bone marrow trephine biopsy histopathology was performed and the results are shown in Table 2. Overall, both techniques were contributory in 50.8% (31/61) cases. FCI was successful in lymphoma typing in 74% (45/61) cases whereas CB-ICC alone was helpful in 65.5% (40/61) cases. In 11.4% (7/61) cases, the lymphoma could not be classified by either technique. Thus, in a total of 88.5% (54/61) cases, an accurate classification of lymphoma could be rendered by a combination of cytomorphology, FCI, and CB-ICC from FNA of the lymph node.

Table 2

FCI versus CB-ICC in diagnosis of Non-Hodgkin's Lymphoma by FNA

Discussion

Cytomorphological evaluation of a smear is of utmost importance; however, the diagnosis of NHL purely on morphology is challenging and requires all ancillary testing. In this study, FCI and CB-ICC were evaluated as adjuncts to cytomorphology in the FNA diagnosis of NHL.

Precursor T-lymphoblastic lymphoma constituted 22% (14/65) of cases. The frequency of PT-LBL is higher in India (around 7%) as compared to the West (1-4%).[1213] In this subtype, overall 85% (12/14) of cases were successfully typed using FCI alone, while 71% (10/14) were diagnosed with the aid of CB-ICC; a combination of the two techniques could successfully type almost all cases. Mourad et al.[5] from Saudi Arabia reported on a series of NHL which included 5 cases of T-LBL which were accurately categorized by FCI. Our study, which includes a total of 14 cases of this lymphoma subtype, reaffirms this observation.

Anaplastic large cell lymphoma is a challenging diagnosis with Hodgkin lymphoma and metastatic carcinoma being close differential diagnoses. ALCL displayed characteristic cytomorphology including the “hallmark cells.”[161718] In the current series, 5 out of 7 cases were successfully typed based on FCI and CB-ICC. Rapkiewicz et al.[19] could subclassify ALCL correctly in only 35% of their cases. Overall, positivity for CD45 and CD30, lack of a monoclonal B cell population and expression of CD3 or CD5 which are T-cell markers were the cornerstones of diagnosis by FCI. By ICC, all cases of ALCL were CD30 positive. By flow cytometry, CD30 is reported positive in 60-100% cases of ALCL.[120] Muzzafar et al.[21] confirmed ALCL in 82.6% (19/23) cases using the criteria of CD45+/CD30+ as was done in this study. The reported variable expression of T-cell antigen in ALCL[2122] is consistent with our observations. In ALCL, it is also important to gate the larger cells sequentially. The expression of CD15 in ALCL can lead to diagnostic problems.[2324] CD15 was positive in 2 cases on FCI and in 1 by CB-ICC. The major differential diagnoses of ALCL is Hodgkin lymphoma and if one considers CD15 as a definitive marker for the latter, then its expression in ALCL renders it useless as a marker to distinguish the two entities. ICC is more useful than FCI and in 5/7 cases, the CB showed atypical cells with CD30+/CD20−/Alk+ phenotype consistent with ALCL. Overall, we agree with Kesler et al.[22] that a combination of cytomorphology and ICC rather than FCI is the best modality for diagnosis of ALCL.

This series includes 4 cases of Burkitt lymphoma and the immunophenotyping has been reported previously.[25] It is important to note that DLBCL and Burkitt's lymphoma share similar immunophenotypic features and that a definitive diagnosis of Burkitt lymphoma requires fluorescence in situ hybridization/cytogenetics which was not done in our study.

Diffuse large B-cell lymphoma constituted 26% (17/65) of all cases and FCI was contributory in 59% of our cases. In 3/17 cases, FCI was not interpretable due to low cellularity and poor cell viability, which has been highlighted in previous reports.[26] Light chain restriction was not observed in 5/14 (35%) cases in our study as neither light chain was demonstrable which has been reported by others.[35] The admixture of reactive T cells with the neoplastic B cells resulting in dilution of their numbers can also lead to problems in interpretation. Further, the large cells are particularly fragile and may not resist the protocol of cell processing for FCI. Partial involvement of a lymph node by the lymphoma is another reason for discrepant results on FCI.[27] However, this problem was easily overcome using a combination of CB-ICC and FCI. In fact, demonstration of CD20 positivity in the lymphoma cells in CBs was possible in a slightly greater proportion of cases. Further, a combination of FCI and CB-ICC could type 88% (15/17) cases of DLBCL, implying that both these ancillary techniques are important for typing this lymphoma.

The accurate diagnosis and classification of B-cell lymphomas with a predominant small-cell population still present a diagnostic challenge.[28] Approximately, 58% of cases could be classified definitively on the basis of FCI alone and in another 26% cases, it was partially contributory. Thus, in 15.7%, FCI was noncontributory in identifying a B lineage lymphoma.

Light chain restriction could be demonstrated in 26 cases (70%), with 14 showing kappa and 12 showing lambda restriction. Zeppa et al.[7] attributed this to defects in gene transcription or translation of light chains. Zardawi et al.[29] has proposed that in the absence of demonstrable light chain restriction, CD20 >80% or CD19/CD10+≥18% or CD19 or CD20+ with CD5+≥35% are consistent with a B-cell lymphoma. Using these additional criteria, FCI was consistent with a B-cell type in 32/40 (80%) cases. It is worth adding here that the distinction of a DLBCL from a grade 3 follicular lymphoma is not possible on cytomorphology along with FCI. Moreover, accurate grading of follicular lymphoma is difficult without histopathology.

Four cases of lymphoma could not be accurately typed as per the WHO classification by histopathology and immunohistochemistry and were left labeled as unclassified lymphomas; The reason for this mainly was the limited panel used in immunophenotyping and the lack or other adjunct techniques such as cytogenetics and molecular analysis.

Our study compared the relative usefulness of FCI and CB-ICC in arriving at a primary diagnosis of NHL. The two techniques were largely complementary to each other in a significant number of cases. Of 61 cases, 45 (74%) were diagnosed by FCI, as compared to 41 (67%) by CB-ICC with a high concordance. Robins et al.,[30] evaluated the relative efficacy of FCI and ICC on cytospin preparation and reported a diagnostic concordance of 97% between FCI and cytospin ICC. CB-ICC is especially useful if a nonlymphomatous malignancy is being considered in the differential diagnosis.[31]

A major limitation of this study was the nonavailability of a lymph node excisional biopsy in all our cases. In some patients who are too sick, a biopsy may not be feasible. However, as is evident from Table 3, where we have compared our data with similar studies published in the last 5 years. The diagnostic accuracy of sub-classifying lymphomas as per the REAL/WHO classification ranged from 50% to 90% and our data compared favorably with these reports. In a total of 88.5% (54/61) cases, an accurate classification of lymphoma could be rendered by a combination of FCI and CB-ICC from FNA of the lymph node. Therefore, in conclusion, whenever FNA is used as a modality for diagnosis of NHL, it must be supplemented by FCI and ICC on CBs to provide an accurate diagnosis.

Table 3

Accuracy of lymphoma typing on FNA material by flow cytometry using REAL/WHO classification