This was an observational and analytical study, approved by the Institutional Review Board, and done between July 2012 and July 2015 in a university hospital of Northern India. Demographic data like age, sex, and site of biopsy was noted. Appropriate blocks of Formalin-Fixed Paraffin Embedded (FFPE) tissues were retrieved for immunohistochemistry. Histopathologically diagnosed cases of HL or NHL were included, for which FFPE blocks containing adequate tissue were available. After immunohistochemistry, diagnostic categories like, benign lymphoproliferative disorder, reactive lymphoid hyperplasia, etc. were excluded as cases. Thus, based on histopathology and immunohistochemistry, total 71 cases of lymphoma were retrieved. For comparison, 20 separate FFPE blocks of reactive lymph nodes were taken as control samples.

Immunohistochemical study was done against CD20, CD3, CD15, and CD30 for diagnostic confirmation and classification of cases as HL or NHL. Additional immunohistochemistry was performed on selected cases of lymphoma against CD5, CD10, Cyclin D1, bcl-6, Epithelial Membrane Antigen (EMA), and Anaplastic Lymphoma Kinase (ALK). CD20 and CD3 were done on control samples of reactive lymph nodes. PAX5 immunohistochemistry was done in all the cases of lymphoma and control tissue of reactive lymph nodes. Briefly, the procedure used for immunohistochemistry was as follows: 5μ thick sections were cut on 0.01% poly-L-lysine coated glass slides and baked at 60°C for minimum one hour on a standard histology hotplate. Sections were dewaxed in xylene and treated with three changes of alcohol and brought to water. Sections were subjected to heat induced antigen retrieval in a microwave oven (EZ Retriever, Biogenex Inc., USA) in citrate buffer (pH 6.0). For PAX5, Tris-EDTA buffer (pH 8.0) was used for antigen retrieval. Cooling at room temperature for 20 minutes followed this. Sections were then treated to block endogenous peroxidase, stained with ready to use primary antibodies and incubated overnight at 4°C. For PAX5, incubation with primary antibody was done at room temperature for 30 minutes. Detection was done by horseradish peroxidase polymer based detection system (Biogenex Inc., USA), diaminobenzidine chromogen and counterstain with haematoxylin.

Evaluation of all immunostained slides was performed without knowledge of the clinical data and histopathological diagnosis. Slides were considered assessable if the positive control and simultaneous internal controls were stained appropriately. Reactive lymph node sections were used as positive controls for CD20, CD3 and PAX5. Known positive case of HL was taken as a control for CD15 and CD30. Samples without primary antibody served as negative control. Regarding CD30 and CD15, membranous or granular cytoplasmic staining of any intensity in at least 10% of HRS cells was considered positive [12]. For CD20 and CD3 staining, any membranous or cytoplasmic staining in at least 10% tumour cells was considered positive [13].

Nuclear staining in greater than 10% of neoplastic cells was considered positive for PAX5 [13]. Nuclear staining of PAX5 was scored as weak, moderate or strong. The strong positivity of normal mantle zone cells and scattered (residual) small B lymphocytes was used to evaluate the relative staining intensity of PAX5 positivity in neoplastic cells as follows [14]: Strong expression when nuclear staining was similar to the staining of mantle zone cells of the reactive lymph node, moderate expression when weaker, but significant staining intensity was noted, and weak expression when detectable nuclear staining was present above the background level.

Calculation of frequency, mean values, and the percentage was done using online Graphpad software, which is freely available at: http://www.graphpad.com. Fisher’s-exact test was used for comparison between control samples and lymphoma cases by Quick Calcs software. Test for analysis of variance (ANOVA) was performed by Microsoft Excel software. A p-value <0.05 was taken as statistically significant.

The demographic data on lymphoma cases of the study are given in [Table/Fig-1].

All control samples of reactive lymph nodes (n=20) showed the normal distribution of CD20 and CD3 staining and strong intensity of PAX5+ expression. All the 24 cases of HL were of classical type [Table/Fig-2a] and showed negative expression of CD3 and CD20 in Hodgkin/Reed-Sternberg (HRS) cells. Further, all the 24 (100%) cases of HL showed positive expression of CD30 in HRS cells, whereas, 14/24 (58.3%) cases showed positive expression of CD15 in HRS cells [Table/Fig-2b]. Weak nuclear staining of PAX5 in HRS cells was noted in 23/24 (95.8%) cases of HL [Table/Fig-2c]. On applying Fisher’s-exact test, statistically significant result was not seen for PAX5 staining between control samples and HL (p=1.000) [Table/Fig-3]. The co-expression pattern of PAX5 staining with CD15 and CD30 in HL is given in [Table/Fig-4].

[Table/Fig-2]:

(a) Reed-Sternberg cell (arrow) of HL (Haematoxylin and eosin stain, 1000×); (b) CD15+ expression in Hodgkin cells (arrows) of HL (Diaminobenzidine, 1000×); (c) Weakly positive nuclear expression of PAX5 in Hodgkin cells (arrows) (Diaminobenzidine, 400×); (d) Sheets of large atypical lymphoid cells in a case of NHL (Haematoxylin and eosin stain, 400×); (e) Strong membranous positivity of CD20 in NHL (Diaminobenzidine, 400×); (f) Strongly positive nuclear expression of PAX5 in neoplastic cells of B-cell NHL (Diaminobenzidine, 400×); (g) Sheets of monomorphic atypical lymphoid cells in NHL (Haematoxylin and eosin stain, 400×); (h) Strongly and diffusely positive membranous expression of CD3 in NHL (Diaminobenzidine, 400×); (i) Negative PAX5 expression in T-cell NHL. Note scattered strong PAX5 positive B lymphocytes in the background (Diaminobenzidine, 400×).

Based on immunohistochemistry, 39/47 (83%) cases were categorized as B-cell NHL (CD20+/CD3-) [Table/Fig-2d,2e]. Distribution among various subtypes of B-cell NHL was: large B-cell lymphoma (35/39, 89.7%), small B-cell lymphoma (2/39), small lymphocytic lymphoma (1/39) and follicle-centre cell lymphoma (1/39). PAX5+ expression was seen in the nuclei of neoplastic tumour cells in 32/39 (82%) cases of B-NHL [Table/Fig-2f]. Among 32 PAX5 positive cases of B-cell NHL, intensity of PAX5 expression was noted as mild, moderate and strong in 9 (28.1%), 16 (50%), and 7 (21.9%) cases, respectively. All seven PAX5 negative B-cell NHL were large B-cell subtype. In this study, there were 8/47 (14.8%) cases of T-cell NHL; out of which seven were CD3+/CD20- [Table/Fig-2g,2h] and a single case was diagnosed as anaplastic large cell lymphoma (CD20-/CD3-/CD30+/ALK+/EMA+). All eight cases of T-cell NHL were negative for PAX5 [Table/Fig-2i]. For PAX5 staining between control samples and B-NHL, statistically significant result was not seen by Fisher’s-exact test (p=0.0843), however, statistically highly significant result was obtained between control samples and T-cell NHL (p<0.0001) [Table/Fig-3]. For PAX5 staining, the test for analysis of variance between NHL, HL and control samples was statistically highly significant (p<0.0001) [Table/Fig-5]. The co-expression pattern of PAX5 staining with CD20, and CD3 in cases of NHL is given in [Table/Fig-6].

PAX+ expression in HL and B-NHL of the present study was found to be statistically similar to PAX5+ staining in non -neoplastic B-cells of the control samples. To the best of our knowledge, this type of study is not available in the published literature from India. PAX5 is a member of the highly conserved paired box (PAX) domain family of transcription factors, located on chromosome 9p13 and it is expressed in all the developmental stages of the B-cells, but not in the plasma cells or the T-cells [11]. We have observed PAX5+ expression in CD20+B-cell follicles and mantle zone of reactive lymph nodes, while negative PAX5 expression was seen in CD3+T-cells. In this study, PAX5+ expression was seen in 95.8% cases of HL, 82% cases of B-cell NHL and none of the cases of T-cell NHL.

In various studies, PAX5+ expression was observed in 100% cases of B-cell haematolymphoid malignancies [13,15,16]. Krenacs et al., and Mhawech-Fauceglia et al., have found PAX5+ expression in 81% and 91.5% of B-cell NHL, respectively [14,17]. In current study, PAX5+ expression was observed in 81% cases of B-cell NHL. Now, it may be argued that as PAX5 is expressed through all the developmental stages of B-cell, its expression should have been seen in all the cases of B-cell NHL. However, PAX5 expression is not seen in certain subsets of B-cell NHL where the malignant clone of B-cells are derived from post-follicular transformed B-cells representing the transition from mature B-cells to plasma cells [14]. Though, in present study, we did not observe plasmacytic differentiation in cases of HL or NHL. In contrast, a study on CD20 negative lymphomas showed PAX5 positivity in 100% cases of CD20- precursor B-cell lymphoblastic leukaemia/lymphoblastic lymphoma, 88% cases of CD20- recurrent mature B-cell lymphoma after rituximab therapy and negative staining in all cases of CD20- Diffuse Large B-Cell Lymphoma (DLBCL) [13]. Similar results were seen in another study with additional PAX5 positivity in CD20- DLBCL without plasmacytic differentiation [18]. The findings of these studies indicate that malignant clones of lymphoma cells, which lack CD20 expression with or without therapy, may retain PAX5 expression. Thus, PAX5 may be a better marker of B-lineage than CD20. However, till date, there is no therapy available targeting PAX5.

Expression of the presence or absence of PAX5 is also useful in distinguishing classical HL from nodular lymphocyte predominant HL and anaplastic large cell lymphoma. PAX5 is expressed in 90% to 98% of HRS cells in classical HL and 100% of HRS cells in lymphocyte predominant HL [19]. Similarly, in another study, varying intensity of PAX5 expression was seen in 93% cases of classical HL [18]. This observation of PAX5 staining in HRS cells of classical HL also confirms their B-cell origin, despite the frequent absence of other B-cell antigen [19]. In classical HL, the staining intensity varies from weak to moderate and is rarely strong compared with the normal reactive small lymphocytes. In nodular lymphocyte predominant HL, the staining of PAX5 is as strong as seen in the small B cells in at least some of the tumour cells. There is a negative expression of PAX5 in anaplastic large cell lymphoma, which was also observed in a single case of this study.

We studied expression of PAX5 in Indian patients of lymphoma, however, we could not correlate PAX5 expression with other parameters like stage, treatment response, and survival of patients.

PAX5 is a promising marker for diagnosis of classical HL and B-cell NHL. PAX5 staining remains negative in T-NHL. However, larger studies from different parts of India are required to observe staining pattern of PAX5 in patients with lymphoma and its correlation with outcome of disease.