Aims To evaluate the role of the follicular helper T (TFH) cell markers, CD10, BCL6, programmed death-1 (PD-1) and CXCL13, in the differential diagnosis of nodal peripheral T cell lymphomas (PTCLs) and to determine whether PTCL subtypes other than angioimmunoblastic T cell lymphoma (AITL) express TFH cell markers.
Methods 162 nodal PTCL specimens and 53 other lymphoid pathology specimens were collected. Immunohistochemistry for CD10, BCL6, PD-1 and CXCL13 was performed on tissue microarray sections. Morphological feature analysis and double labelling assay were also performed.
Results For AITL cases, the rate of CD10, BCL6, PD-1 and CXCL13 expression was 75.0% (36/48), 66.7% (32/48), 93.8% (45/48) and 97.9% (47/48), respectively. Expression of CD10, PD-1 and CXCL13 in the AITL group was significantly higher than in other nodal PTCLs and the control group (p<0.05). The rate of coexpression of three or four (≥3) markers was 83.3% for AITL cases, which was significantly higher than that for any of the non-AITL cases (0–4.9%; p<0.05). The rate of coexpression of PD-1 and CXCL13 (91.7%, 44/48) was significantly higher than that of CD10 and BCL6 (56.3%, 27/48) (p=0.000) in the AITL group. Seventeen cases of PTCL not otherwise specified (PTCL, NOS) expressed CXCL13, including both cases of the follicular variant of PTCL, NOS (FVPTCL, NOS), three of the four cases of the lymphoepithelioid variant of PTCL, NOS (LVPTCL, NOS), and the remaining 12 cases which displayed one or more features of AITL.
Conclusions The combined detection of CD10, BCL6, PD-1 and CXCL13 has high specificity and sensitivity for the differential diagnosis of AITL. PD-1 and CXCL13 are more sensitive, superior diagnostic markers for AITL than CD10 and BCL6. Currently, TFH cell markers are the only available markers that show high specificity for AITL. LVPTCL, NOS and/or FVPTCL, NOS may also arise from TFH cells and fall within the spectrum of AITL.
- Peripheral T-cell lymphoma
- follicular helper T-cell markers
- tissue microarray
- differential diagnosis
- lymph node pathology
Statistics from Altmetric.com
- Peripheral T-cell lymphoma
- follicular helper T-cell markers
- tissue microarray
- differential diagnosis
- lymph node pathology
Peripheral T cell lymphomas (PTCLs) are a heterogeneous group of neoplasms, representing 8–10% of all non-Hodgkin's lymphomas.1 In Asian countries, the incidence is higher, with 15–20% of all lymphomas classified as PTCL or NK/T cell lymphoma (NKTCL).2 3 Primary nodal PTCLs mainly include four subtypes: PTCL, not otherwise specified (PTCL, NOS), angioimmunoblastic T cell lymphoma (AITL), anaplastic large cell lymphoma, anaplastic lymphoma kinase-positive (ALCL, ALK+), and anaplastic large cell lymphoma, ALK-negative (ALCL, ALK−).4
Diagnosis of nodal PTCLs is important and challenging, because different subtypes have different clinical outcomes and may need different treatment strategies.5 PTCL, NOS is the most common and heterogeneous category of PTCL, including three variants of the lymphoepithelioid, follicular and T-zone variant, which do not correspond to any of the specifically defined entities of PTCL in the current classification.4 Distinguishing between PTCL, NOS and AITL can be much more difficult and often rests on subtle morphological and phenotypic differences.6 The diagnosis of ALCL, ALK+ is based on well-defined criteria, but the distinction between ALCL, ALK− and PTCL, NOS is not always clear-cut and is controversial.4
On the basis of gene expression profile and single marker analyses, it has been suggested that AITLs highly express the follicular helper T (TFH) cell markers, CD10, BCL6, programmed death-1 (PD-1) and CXCL13.7–10 However, their utility in the differential diagnosis of nodal PTCLs (and whether PTCL subtypes other than AITL express TFH cell markers) has not been fully investigated.
In this study, we explored the expression of the TFH cell markers, CD10, BCL6, PD-1 and CXCL13, in 162 cases of nodal PTCL as well as 11 cases of extranodal NKTCL, 11 of T lymphoblastic lymphoma (T-LBL) and 31 of reactive lymphoid hyperplasia (RLH) on tissue microarray (TMA) sections, in order to analyse the role of TFH cell markers in the differential diagnosis of nodal PTCLs, and to determine whether PTCL subtypes other than AITL express TFH cell markers. Morphological and immunohistochemical features of nodal PTCLs were also analysed.
Materials and methods
A total of 162 nodal PTCLs with adequate paraffin-embedded tissue samples (160 from the archive of the Department of Pathology at Fudan University Shanghai Cancer Center and two from the consultation files), diagnosed between 1999 and 2009, were included in our study. Categorisation of the 162 cases was as follows: 81 PTCL, NOS, 48 AITL, 19 ALCL, ALK+ and 14 ALCL, ALK−. The 81 cases of PTCL, NOS were further subcategorised into 73 common type (CTPTCL, NOS), four lymphoepithelioid variant (Lennert lymphoma) (LVPTCL, NOS), two T-zone variant (TVPTCL, NOS), and two follicular variant (FVPTCL, NOS). As comparative controls, we also included 53 other lymphoid lesions comprising 11 extranodal NKTCLs, 11 T-LBLs and 31 RLHs collected from our institution. All cases were reviewed by two pathologists (XQL and XZZ), and diagnoses were assigned according to the current World Health Organization (WHO) classification.4 This study was approved by the research ethics committee of Fudan University Shanghai Cancer Center.
TMA was constructed as described previously.11 Briefly, an H&E-stained section from each block was used to define the representative tumour regions. Tissue cores of 1 mm were punched from the defined areas and incorporated into a recipient paraffin block using a manual tissue arrayer (Beecher Instruments, Sun Prairie, Wisconsin, USA). Each tissue block of 162 cases of nodal PTCL and 53 control cases was punched twice.
H&E-stained whole-tissue sections were used to assess the morphological features of 162 nodal PTCLs. Particular cell types (eg, CD20+ blasts, plasma cells, eosinophils) were scored as ‘many’ if they were easily discernible and occurred in groups, as described previously.12 The tumour cells were composed of small to medium-sized and large cells with irregular, pleomorphic, hyperchromatic or vesicular nuclei, prominent nucleoli, pale cytoplasm and many mitotic figures.
Table 1 lists the primary antibodies applied in this study, their dilutions, and pretreatment conditions. Immunohistochemical analysis was performed on TMA sections (3 μm). The EnVision two-step method was used for all antibodies except CXCL13, as described previously.13 The indirect streptavidin immunoperoxidase method was used for CXCL13. Reactive tonsils were used as positive controls. Primary antibodies were omitted to provide negative controls. The appearance of brown particles in the membrane and/or cytoplasm (CD10 and PD-1), in the nucleus (BCL6), and in the cytoplasm with/without a dot-like reinforcement (CXCL13) represented a positive staining result. The percentage of neoplastic cells was scored as follows, as described previously14: −, no evidence of positivity; +, 5–15% positive neoplastic cells; ++, 15–40% positive neoplastic cells; +++, more than 40% positive neoplastic cells.
In situ hybridisation (ISH) for Epstein–Barr virus (EBV)-encoded RNAs (EBER) and double labelling
ISH for EBER was performed according to previously described methods.15 Briefly, tissue sections were deparaffinised, dehydrated and pretreated with proteinase K. Hybridisation was performed using commercially available digoxigenin-labelled probes (Triplex International Biosciences, Fuzhou, China) by overnight incubation at 37°C. Detection was accomplished using 5-bromo-4-chloro-3-indoxyl phosphate/nitroblue tetrazolium development without counterstaining. EBV-positive NKTCLs were used as positive controls. EBV-negative lymphoid tissues and the hybridisation mixture without EBV probes were used as negative controls. A blue or blue–black colour within the nucleus was considered a positive reaction. Double-labelling ISH for EBER and immunohistochemistry for CD20 or CD3 assay were carried out on tissue sections. ISH was performed first as described, followed by immunohistochemistry.
Statistical analysis was performed using SPSS V.11.5 software for Windows. Morphological factor comparisons and comparisons of the immunohistochemical results between nodal PTCLs and control groups were performed using the χ2 test (n≥40 and T≥5), continuity correction χ2 test (n≥40 and 1≤T<5) or Fisher exact test (n<40 or T<1). p<0.05 was considered to be significant.
Histological findings of 162 nodal PTCLs are summarised in table 2. Characteristic irregular radiating proliferations of follicular dendritic cell (FDC) networks were detected by CD21 immunostaining in all cases of AITL but not in cases of PTCL, NOS. Marked proliferation of arborising high endothelial venules (HEVs) was a feature in 95.8% (46/48) of cases of AITL, being significantly less frequent in 34.6% (28/81) of cases of PTCL, NOS (p=0.000). The AITL group showed significantly (p<0.05) higher frequencies of many immunoblasts, plasma cells and clear neoplastic cell clusters, and lower frequency of marked lymphocytic atypia than the PTCL, NOS group. Marked proliferation of HEVs was seen in 34.6% cases of PTCL, NOS, but not in any ALCL, ALK− cases. The frequency of remnant germinal centres in the ALCL, ALK− group was higher than in the PTCL, NOS group (p=0.005). The ALCL, ALK− group showed no significant differences in morphological features from the ALCL, ALK+ group (p>0.05), although spread into perinodal fat was greater in the ALCL, ALK− group (71.4% vs 36.8%).
The rate of expression of CD10, BCL6, PD-1 and CXCL13 was 75.0% (36/48), 66.7% (32/48), 93.8% (45/48) and 97.9% (47/48), respectively, for the AITL group (figure 1). As tables 3 and 4 show, expression of CD10, PD-1 and CXCL13 in the AITL group was significantly higher than in other nodal PTCLs and the control group (p<0.05). Expression of BCL6 in AITL patients was significantly higher than in PTCL, NOS, ALCL, ALK− and the control specimens (p<0.05), but was not different from ALCL, ALK+ specimens (p>0.05). The rate of coexpression of three or four (≥3) markers was 83.3% in the AITL cases, which was significantly higher than that for any of the non-AITL cases (0–4.9%; p<0.05). The rate of coexpression of PD-1 and CXCL13 (91.7%, 44/48) was significantly higher than that of CD10 and BCL6 (56.3%, 27/48) (p=0.000) in the AITL group. All AITL cases expressed at least one of the four markers.
The rate of expression of CD10, BCL6, PD-1 and CXCL13 was 9.9% (8/81), 8.6% (7/81), 17.3% (14/81) and 21.0% (17/81), respectively, for the PTCL, NOS group. The 17 PTCL, NOS cases expressing CXCL13 included two cases of FVPTCL, NOS (figure 2), three cases of LVPTCL, NOS (figure 3A,B), and the remaining 12 cases that displayed one or more features of AITL including marked proliferation of HEVs (n=7), clear cell clusters (n=7), many plasma cells (n=5), CD20+ immunoblasts (n=3) and eosinophils (n=1), but none showed a radiating FDC network. Three CXCL13-positive cases of LVPTCL, NOS were all negative for CD10 and BCL6, but one case was positive for PD-1. Two CXCL13-positive cases of FVPTCL, NOS were all positive for CD10 and PD-1, but only one case was positive for BCL6. In the 11 cases of extranodal NKTCL, six (54.5%) were positive for CXCL13 (figure 3C,D). All T-LBL cases were negative for PD-1 and CXCL13, but three and one, respectively, of the 11 cases were positive for CD10 and BCL6. In RLHs, expression of the four markers was seen in a small subset of T cells mainly located in the germinal centres, with only scattered T cells in the interfollicular areas. The neoplastic cells were CD30-positive in all ALCL, ALK+ and ALCL, ALK− cases, but not in any PTCL, NOS cases.
Atypical cells were observed to be CD4+CD8− in 43 of the 48 (89.6%) AITL cases, CD4−CD8- in three cases (6.2%), and CD4-CD8+ in the remaining two cases (4.2%), with no CD4+CD8+ phenotype being observed. The frequency of the CD4+CD8- phenotype in the neoplastic cells was significantly higher, and the frequency of the CD4-CD8- phenotype was significantly lower, in the AITL group than in other nodal PTCLs or the control group (p<0.05) (table 5). All RLH cases contained CD4+ and CD8+ reactive lymphocytes.
ISH for EBER and double labelling
Among the 48 AITL cases, 38 showed positive lymphoid cells for EBV (79.2%). The number of positive cells expressing EBER varied from case to case. In some cases, only scattered cells were positive, and, in others, numerous large blasts were labelled. Two of four cases of LVPTCL, NOS and one of two cases of FVPTCL, NOS also showed scattered EBV-positive cells. In all positive cases of AITL, LVPTCL, NOS and FVPTCL, NOS, double-labelling analysis showed that the EBV-positive cells were CD20+CD3− (figure 4A,B). However, some EBV-positive cells showed no demonstrable expression of either CD20 or CD3.
Nodal PTCLs often show overlapping histological features, and distinguishing between PTCL, NOS and AITL can be difficult. The diagnostic criteria for AITL, as defined by the WHO4, only apply to typical cases, but in a proportion of cases.6 In our series, although the frequencies of many B immunoblasts, plasma cells and clear neoplastic cells were significantly higher in the AITL group than in the PTCL, NOS group, these features were only present in about half of the AITL cases. The prominent feature of marked proliferation of HEVs in AITL was also shared by PTCL, NOS. The irregular radiating FDC network, which was most specific, was very helpful for distinguishing AITL from PTCL, NOS. However, it is known that, in early cases of AITL, the pattern of FDC hyperplasia may be subtle or minimal.16 17 Nodal PTCL, NOS cases that lack significant FDC proliferation, but otherwise show a morphology typical of AITL, are also hard to categorise.6
In this study, we used TMA technology and immunohistochemistry to evaluate the role of the TFH cell markers, CD10, BCL6, PD-1 and CXCL13, in the differential diagnosis of nodal PTCLs. TMA is a highly efficient tool for the investigation of large series of neoplasms, including lymphomas.18 However, analysis of TMA sections for PTCL is associated with distinct problems, related especially to the low frequency of tumour cells, together with the high content of reactive bystanders. Therefore, at least two cores 1 mm in size should be obtained for analysis of PTCLs.
Similarly to previous studies,19 20 we detected CD10 and BCL6 in 75% and 66.7% of all AITL cases, respectively, but in only 9.9% and 8.6% of all PTCL, NOS cases. Our finding of PD-1 expression in 93.8% of AITL cases is in line with the study of Dorfman et al.9 In addition, we showed that CXCL13 expression was strongly associated with AITL. This is also in agreement with the finding of Dupuis et al.20 In RLH, we observed expression of the four markers in a small subset of T cells, mainly located in the germinal centres, with only scattered T cells in the interfollicular areas. Our study determined that the combined detection of CD10, BCL6, PD-1 and CXCL13 has high specificity and sensitivity for the differential diagnosis of AITLs. However, these four markers are not useful for the differential diagnosis of PTCL, NOS and ALCL, ALK−. In the present study, the rate of coexpression of PD-1 and CXCL13 (91.7%) was significantly higher than that of CD10 and BCL6 (56.3%) in the AITL group. This indicates that PD-1 and CXCL13 are more sensitive, superior diagnostic markers for AITL than CD10 and BCL6.
In the current WHO classification, LVPTCL, NOS (Lennert lymphoma) and FVPTCL, NOS are regarded as histological variants of PTCL, NOS and not as specific subtypes. In our series, we found two of four LVPTCL, NOS cases showing scattered EBV-positive B cells and three of four expressing CXCL13. Similarly to our finding, Grogg et al 10 also reported two cases of LVPTCL, NOS expressing CXCL13. Gisselbrecht et al have shown an overlap of the survival curves of patients with AITL and LVPTCL, NOS.21 The overlap has also been reported at the genetic level, with trisomy 3 frequently occurring in both entities.22 This suggests that LVPTCL, NOS may represent a histological variant of AITL. It has been proposed that FVPTCL, NOS is related to AITL.23 Bacon et al also described a close relationship between them.24 Our data show that the neoplastic cells of two FVPTCL, NOS cases comprised mainly medium-sized cells with abundant, sometimes clear, cytoplasm. The proliferation of FDCs was present only in follicular centres of two cases, without interfollicular expansion. Both FVPTCL, NOS cases expressed CXCL13, PD-1 and CD10. Interestingly, one FVPTCL, NOS patient presented histological features of FVPTCL, NOS on his left lymph node and typical morphological features of AITL on the right one. Our study showed that 79.2% (38/48) of AITL cases were positive for EBV-EBER, and EBV-positive cells corresponded to B cells, which suggested that EBV infection is a characteristic of AITL. Scattered EBV-positive B cells were also seen in one of two FVPTCL, NOS cases. Currently, TFH cell markers are the only markers available that show high specificity for AITL. We propose that FVPTCL, NOS probably falls within the spectrum of AITL. Interestingly, we encountered a higher frequency of NKTCL cases (54.5%) expressing CXCL13 than has been reported elsewhere (14%).25 The significance of this finding remains unclear, and additional studies based on more cases are needed to assess it further.
AITLs usually express CD4, although some CD8+ cases have been reported. In this study, atypical cells were observed to be CD4+CD8- in 43 (89.6%) of 48 AITL cases, CD4-CD8- in three cases, and CD4-CD8+ in the remaining two cases. The AITL cases also overexpressed TFH cell markers. Taken together, these findings support the derivation of AITL from CD4+ TFH cells.
The combined detection of CD10, BCL6, PD-1 and CXCL13 has high specificity and sensitivity for the differential diagnosis of angioimmunoblastic T cell lymphoma (AITL).
PD-1 and CXCL13 are more sensitive, superior diagnostic markers for AITL than CD10 and BCL6.
We propose that LVPTCL, NOS and/or FVPTCL, NOS may also arise from TFH cells and fall within the spectrum of AITL.
We thank Shanghai Outdo Biotech Co Ltd for assistance with constructing tissue microarrays. We also thank Dr Richard Irons for his help with editing the English language.
Funding This work was supported by the Medical Industry Research Special Fund of the Ministry of Health (Grant No HYZX0801).
Competing interests None.
Patient consent Obtained.
Ethics approval This study was conducted with the approval of the research ethics committee of Fudan University Shanghai Cancer Center.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.