To characterize the clinicopathologic features of cases of large B-cell lymphomas, poor in B cells and densely rich in programmed cell death-1 (PD-1)+ reactive T cells, which can mimic T-cell lymphomas.
A single-institute retrospective review of cases between 2010 and 2013 was performed.
Of 178 cases of large B-cell lymphomas, eight cases of large B-cell lymphomas poor in B cells and diffusely rich in sheets of PD-1+ T cells were identified. These cases either were initially misdiagnosed as a T-cell lymphoma or substantiated a broader differential diagnosis including a T-cell lymphoma. Five cases were T-cell histiocyte–rich large B-cell lymphomas, and three cases were diagnosed as large B-cell lymphomas rich in T cells. In three of these cases, a subset of the PD-1+ T cells showed either morphologic nuclear atypia or atypical expression of T-cell antigens on flow cytometry and/or immunohistochemistry.
Large B-cell lymphomas poor in B cells and rich in T cells can have diffuse sheets of reactive PD-1+ T cells, some with atypical morphologic and immunophenotypic features mimicking a T-cell lymphoma. Careful assessment of the immunoarchitecture and background inflammatory and stromal cells can prevent erroneous diagnoses in such cases.
The histopathologic diagnosis of a B-cell or T-cell lymphoma relies on a combination of morphologic, immunophenotypic, and molecular findings. In some instances, morphologic and ancillary findings can be misleading and result in incorrect diagnoses. Such incorrect diagnoses are more frequently encountered in the setting of difficult morphologies and when pathologists are incorporating newer, less understood immunophenotypic markers. One such marker gaining increasing use but still requiring further characterization in many malignancies is programmed cell death-1 (PD-1) protein, also known as CD279.1
PD-1 was first identified in 1992 as a protein upregulated in thymic T cells after activation of the classic pathway of programmed cell death.2,3 In subsequent years, PD-1 has been characterized further and recognized as a sensitive marker of CD4+ follicular helper T cells (TFH),4–6 and functionally important for T-cell/B-cell interactions; when engaged with its ligand, PD-L1, the PD-1/PD-L1 interaction facilitates down-regulation of B-cell and plasma cell activation.7
Given the expression of PD-1 in predominantly CD4+ TFH-cells, the presence of PD-1+ T cells outside B-cell follicles has been noted to be a sensitive and relatively specific hallmark of a T-cell lymphoma, specifically, angioimmunoblastic T-cell lymphoma (AITL) and a subset of other peripheral T-cell lymphomas.4,8,9 However, PD-1+ T cells have also been described as increased outside follicles in nonneoplastic conditions, such as in viral infections (ie, Epstein-Barr virus [EBV]) or other reactive instances.8,10
The complexity of interpreting increased and diffuse PD-1 staining in lymph nodes can create substantial confusion, in particular when coupled with additional atypical morphologic or immunophenotypic features in the reactive T cells themselves.
Herein we describe eight complex cases of large B-cell lymphomas (LBCL) poor in B cells and diffusely rich in PD-1+ T cells that were initially either misdiagnosed as a T-cell lymphoma or resulted in a broader differential diagnosis including a T-cell lymphoma. Five cases were T-cell histiocyte–rich large B-cell lymphomas (THRLBL), and three cases were large B-cell lymphomas rich in T cells (LBCL-TR). In two cases, a subset of the T cells showed loss of expression of CD7 on flow cytometry and/or immunohistochemistry. Patients in all cases were male. T cells in LBCL that were poor in B cells and rich in T cells can have unusual morphologic and immunophenotypic features but should not be misdiagnosed as T-cell lymphomas.
Materials and Methods
Archived cases of LBCL (n = 178) between 2010 and 2013 with diffuse infiltrations of PD-1+ T cells (n = 8) and tissue available for histopathologic review were analyzed. All cases were diagnosed according to 2008 World Health Organization criteria. Seven of eight samples were from excisional lymph node biopsies; one sample was from a core needle biopsy Table 1. Cases diagnosed as THRLBL had fewer than 10% B cells associated with dense infiltration of T cells and histiocytes, and significant follicular dendritic cell networks were absent. Cases diagnosed as LBCL-TR lacked a nodular architecture, lacked significant follicular dendritic cell networks, and had more than 10% B cells but less than 30% B cells, which precluded a diagnosis of THRLBL. Clinical histories and presentations were obtained by retrospective review of electronic medical records. This study was approved by Stanford University’s Institutional Review Board.
Histology and Immunophenotyping
Formalin-fixed paraffin-embedded tissues (FFPE) were cut at 4 μm and stained using H&E. Immunohistochemical studies were performed as previously described using antibodies as listed in Supplemental Table 1 (all supplementary materials can be viewed at OhgamiAug14.pdf). FFPE biopsies were stained using a Dako Autostainer (Dako, Carpinteria, CA) and BenchMarkXT (Ventana/Roche, Tucson, AZ) or Leica-Bondmax processors (Leica Biosystems, Buffalo Grove, IL). In situ hybridization studies for EBV were performed as previously described.11 Flow cytometry was performed as previously described using an FACSCalibur or FACSCanto II (Becton Dickinson, San Jose, CA) cytometer.12 Antibodies used for flow cytometry are listed in Supplemental Table 1.
T-cell Gene Rearrangement Studies
TRG@ (TCR-gamma) and/or TRB@ (TCR-beta) gene rearrangement studies were performed as previously described using GeneScan analysis and BIOMED-2 primers.13
Statistical analyses, t tests, and Fisher exact tests were performed using XLSTAT (Addinsoft, New York, NY). P values of less than .05 were considered significant.
All eight patients were men. The median age was 29 years old (range, 16–66 years). One patient presented with localized lymphadenopathy and seven had disseminated lymphadenopathy (Table 1). Hepatosplenomegaly was seen in six of eight patients. Two patients had a prior history of nodular lymphocyte predominant Hodgkin lymphoma (NLPHL).
In all cases, lymph nodes were diffusely effaced by an infiltration of scattered, but few, atypical large lymphoid cells with irregular nuclear membranes and often prominent nucleoli; some were multinucleate Image 1 and Image 2. A nodular architecture was not seen in any cases. Numerous smaller lymphocytes and histiocytes were seen in the background. Neither increased vascularity nor eosinophils were a significant component in any case and plasma cells were few. Although in most cases, background smaller lymphocytes had round/oval regular nuclear membranes, in two cases, the background smaller lymphocytes showed nuclear irregularities and atypia (Image 1 and Table 1).
Immunophenotypic and Molecular Findings
Immunohistochemical analyses were performed using the markers as described in Supplemental Table 1. In all cases, scattered CD20+ large B cells were seen without expression of CD30 together with CD15. In all instances of THRLBL, large and small B cells were fewer than 10% of the infiltration. In the three cases diagnosed as LBCL-TR, the B cells were more than 10% but fewer than 30% of cells; significant B-cell follicles were not seen.
In all cases, PD-1+ T cells were seen in dense diffuse sheets. In four cases, PD-1+ cells were observed in focal areas ringing the large neoplastic B cells, and two of these cases corresponded to patients with histories of NLPHL (Table 1; Image 2). CD7 expression was diminished in PD-1+ T cells in two cases on immunohistochemistry; CD10 coexpression was not seen on T cells and loss of CD5 expression was not observed. CD21 staining for follicular dendritic cells did not show expanded or disrupted follicular dendritic cell networks.
Flow cytometry was performed in two cases with CD4+ T cells predominating over CD8+ T cells at ratios of more than 30:1. In these two cases, the CD4+ T-cell population had partial CD57 expression and in one, a population of atypical CD4+ T cells lacking CD7 expression was seen Image 3; this finding was verified with immunohistochemistry as well (Image 1). No CD10+ T-cell population was identified in these cases with flow cytometry.
T-cell clonality testing for TRG@ and TRB@ was performed in three cases where atypical T cells were seen; all were negative for T-cell receptor gene rearrangements.
Differential Diagnostic Considerations
Largely owing to the diffuse sheets of PD-1+ T cells, as well as morphologic and/or immunophenotypic atypia among T cells seen in three cases, all eight cases were carefully evaluated for a T-cell lymphoma. In case 2, a T-cell lymphoma was initially misdiagnosed because of the morphologic and immunophenotypic atypia in the diffuse sheets of PD-1+ T cells; however, large atypical B cells were seen on second review/consultation, and no background of eosinophils or disrupted and expanded follicular dendritic cell network was seen to suggest AITL. In case 3, the possibility of a T-cell lymphoma was retained in the initial diagnosis, given the morphologic atypia seen in the background T cells, but later was ruled out after T-cell gene rearrangement studies found no evidence of a clonal process.
Treatment and Follow-up
Seven patients received cyclophosphamide/doxorubicin/vincristine/prednisone (CHOP), with five patients also receiving rituximab, and one alternatively receiving GA101 (obinutuzumab) in combination with CHOP. One patient received no therapy and died of his disease after 385 days. One patient died of an unrelated cause. All patients who received therapy experienced complete remission. Overall survival was a median of 457 days (range, 210–777 days).
Previous groups, as well as our own, have reported on a subset of cases of NLPHL with increased PD-1+ T cells; in some instances, these proliferations of reactive T cells can result in a misdiagnosis as a T-cell lymphoma.10,14 We extend these findings to demonstrate that diffuse infiltrations of PD-1+ T cells can be seen in cases of THRLBL as well as LBCL-TR.
In two of these cases in which flow cytometry was also performed, a significantly elevated CD4-CD8 ratio of more than 30:1 was observed, which, together with the phenotyping of these T cells as TFH, led to a broader differential including T-cell lymphoma. However, other features of T-cell lymphoma of TFH origin (PD-1+ T cells) generally corresponding to AITL were lacking in these cases. No cases showed disrupted and expanded follicular dendritic cell networks, and background eosinophils were not increased. In addition, positive T-cell clones were not seen with the new BIOMED-2 primers and assays, in which clones are detected in the vast majority of cases of T-cell lymphomas.15,16 Finally, with regard to AITL, scattered EBV+ cells can be seen in more than 50% of cases but were not seen in the cases here.
Though cases of atypical T cells either seen on morphology or immunophenotype were not associated with statistically significant clinicopathologic features, there was a distinct trend toward these atypical T cells being more frequent in younger patients, with a median age of 19 years vs 53 years (P = .051). Interestingly, these findings are also consistent with those of Sohani et al,14 who studied cases of NLPHL with atypical T cells that morphologically and immunophenotypically raised a differential diagnosis of a peripheral T-cell lymphoma. In their work, atypical T cells were more frequently seen in younger patients as well.
The role of increased PD-1+ T cells in the oncogenesis or progression of disease in these cases is uncertain. Cases of THRLBL with diffusely increased PD-1+ T cells are uncommon. The majority of cases of THRLBL will contain a predominance of CD8+ T cells; although PD-1+ T cells can be seen in cases of THRLBL, they do not generally form diffuse confluent sheets.17 Although we did not use cytotoxic markers in our cases described here, one does wonder if these CD4+ T cells are uniquely different from those of NLPHL; future work on subsequent cases will involve assessing the expression of cytotoxic markers in such populations. Interestingly, in two cases, a THRLBL and an LBCL-TR, the patients had a remote history of NLPHL. Certainly the relationship between THRLBL and NLPHL has been an area of extensive debate; the presence of numerous PD-1+ T cells in the case diagnosed as THRLBL raises a question as to whether all cases might have actually originated from an NLPHL. However, it is not possible to obtain such answers retrospectively in most patients who have not had prior biopsies. Interestingly, Hartmann et al18 demonstrated molecular overlap in cases of NLPHL, THRLBL-like NLPHL, and THRLBL with gene expression profiling of microdissected neoplastic B cells, though the microenvironment was shown to be distinct between NLPHL and THRLBL; such work suggests that perhaps the tumor cells themselves are not fundamentally different among entities, but the associated inflammatory cell environment is what often differs.
Recent studies on follicular lymphomas and diffuse LBCLs have shown that increased PD-1+ T cells are seen in a subset of patients and associated with favorable outcomes19,20; we note that all seven patients in our series who received chemotherapy had complete remission, a favorable outcome considering that not all patients with LBCL who receive CHOP with rituximab achieve complete remission. Although one might extrapolate that the presence of increased PD-1+ T cells in these cases was directly related to a better patient outcome, given the small number of these rare cases we have described here, these findings require further study for confirmation.
The interaction between PD-1+ T cells and its ligand PD-L1 on neoplastic cells has also been an area of intense study. Several groups have demonstrated that PD-L1 can be expressed by neoplastic cells and background inflammatory histiocytes in various lymphomas.21–23 We additionally assessed for the expression of PD-L1 in four of our cases of THRLBL. Although expression of PD-L1 was seen in the larger neoplastic B cells, some appeared negative and background histiocytes uniformly showed faint expression of PD-L1, a finding previously noted by others, but resulting in difficult interpretation of staining patterns (Supplemental Figure 1). Further studies will be required to assess the definitive relationship between PD1+ T cells and PD-L1.
Overall, our findings demonstrate that PD-1+ T cells can form diffuse sheets in LBCLs poor in B cells and LBCL-TRs, specifically THRLBL, and that these T cells can have atypical immunophenotypes and morphologies but should not be mistaken for T-cell lymphomas. Careful evaluation for background eosinophils, altered follicular dendritic cell networks, vascularity, and correlation with T-cell receptor gene rearrangement studies in difficult cases will help exclude a T-cell lymphoma.