|Home | About | Journals | Submit | Contact Us | Français|
Kaposi sarcoma‐associated herpesvirus (KSHV), also known as human herpesvirus 8 (HHV8), is a recent addition to the list of human viruses that are directly associated with lymphoproliferative disorders. KSHV was first shown to be involved in multicentric Castleman disease and primary effusion lymphoma (PEL). Subsequently, the virus was identified in solid lymphomas, often of extranodal sites, with morphological and immunophenotypic characteristics similar to those of PEL, and in other lymphoproliferative disorders with heterogeneous clinicopathological presentations. The recent advances in our understanding of the histology, immunophenotype and pathogenesis of these KSHV‐associated lymphoproliferative disorders are reviewed.
Kaposi sarcoma‐associated herpesvirus (KSHV) or human herpesvirus 8 (HHV8) is a new addition to the list of human oncogenic viruses. KSHV is a gamma 2 herpesvirus, identified from the Kaposi sarcoma of an AIDS patient by Chang et al, using representational difference analysis in 1994.1 Subsequently, KSHV was shown to be directly associated with multicentric Castleman disease (MCD), primary effusion lymphoma (PEL) and solid/extracavitary lymphomas, often of extranodal sites, with morphological and immunophenotypic characteristics similar to those of PEL.2,3,4,5 There are further reports of KSHV‐positive lymphomas with clinical, histological and immunophenotypic features distinct from the above KSHV‐associated lymphoproliferative disorders in patients both with and without immunodeficiency. Together, these studies show that KSHV is associated with a wide spectrum of lymphoproliferative disorders with varied clinicopathological features. Here we review the recent advances in our understanding of the histology, immunophenotype and pathogenesis of these KSHV‐associated lymphoproliferative disorders.
MCD occurs in patients with or without human immunodeficiency virus (HIV) infection.6 Since about 13% of patients with MCD develop Kaposi's sarcoma,7 several studies used PCR to investigate the presence of KSHV in lymphoid tissue involved by MCD shortly after the discovery of the virus.2,8,9,10 KSHV infection is found to be present in 100% of HIV‐positive and 40–50% of HIV‐negative MCD cases, but not in localised Castleman disease.8,9,10
By using immunohistochemistry for KSHV‐associated latent nuclear antigen‐1 (LANA‐1), Dupin et al subsequently characterised the KSHV‐positive cells and their associated lymphoproliferative lesions in MCD.11,12 They showed that KSHV‐positive cells in MCD have many of the characteristics of plasmablasts and that they occur typically as isolated cells in the mantle zone of B‐cell follicles, but coalesce to form microlymphoma or frank plasmablastic lymphoma in some cases (fig 11).12 The KSHV‐positive plasmablasts in these different lesions invariably show high levels of expression of cytoplasmic immunoglobulin (Ig), remarkably always IgMλ (fig 11,, table 11),12 and often express OCT2.13 They are commonly negative for CD20, PAX5 and CD30, although expression of these antigens may be seen in a subset of KSHV‐positive plasmablasts in some cases.12,13 Although typically negative for the plasma cell associated marker CD138, KSHV‐positive plasmablasts express MUM1/IRF4 and BLIMP1.13,14 Most KSHV‐positive plasmablasts are positive for the proliferation marker Ki67. Epstein–Barr virus (EBV), as shown by in situ hybridisation for EBER, is always negative in KSHV positive plasmablasts.12
Despite the fact that KSHV‐positive plasmablasts in MCD are monotypic (exclusively expressing IgMλ) they, including those forming the majority of microlymphomas, have been shown by PCR‐based analysis of immunoglobulin heavy and λ light chain gene rearrangement to be polyclonal in nature.15 In line with this, KSHV episomes in MCD are polyclonal.16 Nonetheless, the frank plasmablastic lymphomas studied thus far are monoclonal.15 Thus, KSHV infection causes a range of lymphoproliferative lesions in patients with MCD from polyclonal isolated KSHV‐positive plasmablasts and microlymphomas to monoclonal microlymphomas and frank plasmablastic lymphomas.15
Phenotypically, KSHV‐positive plasmablasts resemble mature B‐cells. For example, they express abundant cytoplasmic immunoglobulin and many express CD27,12,15 a surface marker for memory B‐cells. However, KSHV‐positive plasmablasts in MCD consistently show a lack of somatic mutations in their rearranged immunoglobulin heavy and light chain genes,15 indicating that they originate from pre‐germinal centre B‐cells. This, together with the predilection of KSHV‐positive plasmablasts to localise in the mantle zone of B‐cell follicles, suggests that KSHV may preferentially target IgMλ expressing naïve B‐cells in patients with MCD and drive the infected cells to differentiate into plasmablasts without going through the germinal centre reaction, a critical process for normal B‐cell maturation.17
Shortly after the discovery of KSHV in Kaposi sarcoma, the presence of the viral genome was detected in a subset of lymphomas of AIDS patients which occurred as lymphomatous effusions in body cavities and were known as body cavity‐based non‐Hodgkin lymphoma.3 In 1997, Nador et al showed that KSHV‐associated lymphomatous effusions had distinct clinical, morphological, immunophenotypic and molecular characteristics and coined the term primary effusion lymphoma (PEL) to distinguish them from other lymphomas that involve body cavities.18 PEL is now included as a distinct entity in the World Health Organization Classification of Tumours of Haematopoietic and Lymphoid Tissues.19
PEL is a rare aggressive lymphoma, accounting for approximately 3% of AIDS‐related lymphomas.20 It occurs mainly, but not exclusively, in HIV‐positive patients, often middle aged homosexual males. Typically, patients with PEL present with effusions in the pleural, pericardial or abdominal cavities, usually in the absence of an obvious tumour mass, lymphadenopathy or hepatosplenomegaly.18,19 The neoplastic cells are pleomorphic and show a range of cytomorphological appearances from features of large immunoblastic or plasmablastic cells to those of anaplastic cells (fig 22,, table 11).18,19 The lymphoma cells typically express CD45, but are usually negative for B‐cell markers such as CD19, CD20, CD79a and Pax‐5, and usually lack expression of both immunoglobulin heavy and light chains and OCT2. The majority of lymphoma cells are positive for the proliferative antigen Ki67. Activation markers such as CD30 and CD38 and markers associated with plasma cell differentiation such as CD138/Syndecan‐1, MUM1/IRF4 and BLIMP1 are expressed in most cases of PEL (fig 22).13,21,22 Nonetheless, rare cases of PEL with T‐cell phenotype have been described.23,24 The KSHV‐associated LANA‐1 can be demonstrated in the nuclei of most, if not all, neoplastic cells of PEL by immunocytochemistry. This is the most valuable marker for diagnosis and differential diagnosis of PEL. In the majority of PEL, the neoplastic cells are co‐infected by EBV but exhibit a restricted latency pattern with a lack of detectable expression of latent membrane proteins (LMPs).18,19 Interestingly, EBV is frequently absent in PEL arising in HIV‐negative patients.
Most PEL show evidence of rearranged immunoglobulin genes by Southern blot analysis, indicating the monoclonal nature of the lymphoproliferation.18 In line with this, each lymphoma harbours only a single clone of KSHV and EBV as shown by Southern blot analysis of the polymorphic viral terminal repeats.18,25 However, for unknown reasons, clonally rearranged immunoglobulin genes are detectable by PCR only in around 50% of cases.26,27,28 Sequence analysis of rearranged immunoglobulin genes shows high levels of somatic mutation and features of antigen selection in the majority of PEL.26,27,28 Somatic mutations in the 5′ non‐coding regions of the BCL6 gene, another hallmark of B‐cell transition through the germinal centre, are similarly seen in most cases.29 These genetic findings, together with the immunophenotypic features of PEL, indicate that the lymphoma is likely to derive from a transition stage from antigen‐selected germinal centre B‐cells to terminally differentiated plasma cells, ie B‐cells pre‐terminal differentiation. In line with this, the gene expression profile of PEL shares features of both immunoblasts and plasma cells, but is clearly distinct from both.30,31 Furthermore, regardless of their immunoglobulin gene mutation and EBV status, PELs display a common gene expression profile distinct from other non‐Hodgkin lymphomas, further confirming that they represent a distinct entity.30,31,32
Serous effusions, particularly pleural effusions, are a frequent complication of a range of lymphoma subtypes such as Hodgkin lymphoma, Burkitt lymphoma, T‐lymphoblastic lymphoma, and anaplastic large cell lymphoma,33 and should be distinguished from PEL. The differential diagnosis of PEL from secondary lymphomatous effusions can be readily made by use of an appropriate panel of immunophenotypic markers, in particular by staining for the KSHV‐associated LANA‐1. Rare cases of primary lymphomatous effusions occurring in the absence of KSHV infection have been reported.34,35 However, the term “PEL”, we believe, should be restricted to those cases in which tumour cells harbour KSHV.
The majority of PELs occur exclusively as lymphomatous effusions. However, some patients with PEL subsequently develop secondary solid tissue masses,4,36 while others initially present with a solid tissue mass and later develop a lymphomatous effusion.5 The concomitant lymphomatous effusion and solid tissue involvement in these cases show similar morphology and immunophenotype, and identical rearranged immunoglobulin genes.4,5,36 These findings thus broaden the clinicopathological spectrum of classic PEL. In addition, there are KSHV‐associated large cell lymphomas in HIV‐positive patients, which occur often in extranodal sites such as the gastrointestinal tract and less frequently in lymph nodes, and are not associated with a lymphomatous effusion during the course of the disease.36,37,38,39,40,41,42,43,44 Both the morphological and immunohistological characteristics of these KSHV‐positive solid lymphomas resemble PEL (fig 33,, table 11),), suggesting that they are most likely a part of the clinicopathological spectrum of PEL.43,44,45 Chadburn et al have extensively investigated 8 cases of KSHV‐associated solid lymphoma in comparison with 29 cases of PEL, and found the two KSHV‐associated lymphomas to be virtually indistinguishable in morphology, immunophenotype, genotype and clinical course.43 They proposed that KSHV‐positive solid lymphomas be designated extra‐cavitary PEL.43 The above notion is further strengthened by the finding that KSHV‐positive solid lymphomas express granzyme A, interleukin‐2 receptor β chain, aquaporin 3, mucin 1, selectin P ligand and vascular endothelial growth factor, mRNA transcripts shown by expression microarray analysis to be specifically up‐regulated in PEL.45,46
KSHV‐positive extra‐cavitary lymphoma with morphological and immunophenotypic characteristics of PEL has also been recently reported in HIV‐seronegative patients.47 The two cases reported thus far also expressed the above newly identified PEL‐associated cellular markers.47 Interestingly, both cases showed weak to moderate expression of cytoplasmic IgG, but were negative for the immunoglobulin light chains.47
In addition to the conditions discussed above, there are also reports of KSHV‐positive lymphoproliferative disorders with distinct clinical, histological and immunophenotypic features, in patients both with and without immunodeficiency. Du et al reported three cases of KSHV and EBV associated germinotropic lymphoproliferative disorder in HIV‐seronegative patients, which presented as localised lymphadenopathy and showed a favourable response to chemotherapy or radiotherapy.48 The lymphoproliferation is characterised by plasmablasts that are negative for CD10, Bcl6, CD20, CD79a, CD27 and CD138, but show monotypic immunoglobulin κ or λ light chain (table 11).48 Despite being monotypic, the KSHV and EBV double positive plasmablasts were polyconal as shown by analysis of rearranged immunoglobulin genes.48
The presence of KSHV in a small series of EBV‐positive oral plasmablastic lymphomas in AIDS patients has recently been reported,49 although other authors have failed to identify KSHV in their series of similar cases.50,51,52 Whether these KSHV‐associated cases represent a rare discrete subset of the AIDS‐associated oral plasmablastic lymphomas originally described,53 or whether they are more closely related to extra‐cavitary PEL remains to be determined.
Furthermore, rare cases of KSHV‐associated lymphoproliferative disorders have been reported in solid organ transplant recipients and in patients with common variable immunodeficiency.54,55,56,57 These lymphoproliferations have heterogeneous clinicopathological presentations ranging from reactive conditions, including Castleman disease‐like lesions, to large cell lymphomas.
KSHV‐associated lymphoproliferative disorders are rare, even in populations with high KSHV seroprevalence. Thus, it is conceivable that KSHV infection represents only one of several events involved in the development of lymphoma. However, there is now mounting evidence to suggest that a number of KSHV‐encoded products have direct oncogenic effects and may play a role in lymphomagenesis. In PEL and MCD, the vast majority of virally infected cells are latently infected by KSHV, with lytic replication occurring in only 1–3% of cells.58 Although lytic replication of KSHV may help create a microenvironment which enhances the growth of latently infected cells, and some lytic genes (such as K1 and K15) may activate intracellular pathways regulating lymphocyte proliferation and survival, it is the viral products produced in latency that are believed to be most directly involved in malignant transformation.58 Several such KSHV latency‐associated viral products are expressed in PEL and MCD, and are the focus of the discussion below. The function of KSHV lytically expressed genes and their potential involvement in tumourigenesis are reviewed elsewhere.59 Among the latently expressed genes, LANA‐1 (ORF73), viral‐cyclin (v‐cyclin, ORF72) and viral FLICE inhibitory protein (vFLIP, ORF71) are transcribed from a common promoter as part of a multicistronic mRNA in the latently infected cells,60 and their protein products are found to be consistently expressed in most, if not all, KSHV infected tumour cells.11,61,62,63,64
LANA‐1: This viral protein tethers viral episomal DNA to host chromatin during mitosis, helps in efficient partitioning of viral episomes in dividing cells and blocks the reactivation transcriptional factor RTA. It is thus essential for maintenance of viral latency.58 LANA‐1 can transform primary rat embryo fibroblasts in cooperation with the cellular oncogene HRAS65 and induces B‐cell hyperplasia and lymphoma when expressed in transgenic mice.66 LANA‐1 may exert its oncogenic activity by deregulation of several cellular pathways commonly targeted in human cancer. It directly interacts with p53 and MDM2, inactivating the transcriptional activity of p53 and its ability to induce apoptosis, and thus contributes to cell survival and chromosomal instability.67,68,69 Similarly, LANA‐1 also binds to and inactivates the retinoblastoma (Rb) protein, thereby releasing the transcription factor E2F and promoting cell cycle progression.65 In addition, LANA‐1 binds to and inhibits glycogen synthase kinase (GSK)‐3, preventing β‐catenin phosphorylation and degradation.70 The accumulated β‐catenin translocates to the nucleus and forms a complex with T‐cell factor (TCF) and lymphoid enhancing factor (LEF), which transactivate the expression of the target genes including c‐MYC, JUN and CCND1, known to be important in cell proliferation. By inhibition of GSK‐3β activity, LANA also abrogates GSK‐3β medicated c‐Myc phosphorylation, consequently increasing c‐Myc stability.71
v‐cyclin: The KSHV cyclin is a homologue of cellular D‐type cyclins. Like its cellular counterparts, v‐cyclin associates with cyclin dependent kinases (CDKs), particularly CDK6, and the complex promotes cell‐cycle progression by phosphorylating Rb and histone H1.72,73 In addition, v‐cyclin stably associates with the CDK inhibitor p27KIP1 in PEL cells, and together with CDK6 inactivates its antiproliferative function.74 This, at least in part, explains the findings of high proliferative activity of PEL cells despite their strong expression of p27KIP1.61 Expression of v‐cyclin both in vitro and in vivo causes genome instability, and induces lymphoma in the absence of p53 in transgenic mice.75,76 In light of these findings, v‐cylin and LANA‐1 are highly likely to be synergistic in their tumourigenic activities.
v‐FLIP: This viral protein is a homologue of cellular FLIP, which functions both as an inhibitor of death receptor mediated apoptosis and in activation of the transcription factor NF‐κB.77 Like its cellular counterpart, vFLIP is a potent activator of NF‐κB, which transactivates a number of genes important for cell survival, proliferation and activation. Evidence suggests that vFLIP both activates the classic NF‐κB pathway by direct activation of the IκB kinase complex78 and stimulates the non‐classical NF‐κB pathway by up‐regulating the expression of p100/NF‐κB2.79 In line with this, vFLIP transgenic mice show constitutive activation of both classic and non‐classic NF‐κB pathways.80 Similarly, NF‐κB is activated in primary human PEL cells and PEL cell lines.81 Elimination of vFLIP production in PEL cells by RNA interference results in significantly reduced NF‐κB activity, down‐regulation of essential NF‐κB regulated pro‐survival factors and induction of apoptosis.82,83
LANA‐2 or viral interferon regulatory factor‐3 (vIRF‐3): LANA‐2, encoded by ORF10.5, is expressed in the nuclei of virtually all KSHV‐infected lymphoid cells in PEL and the majority of KSHV infected cells in MCD, but not in Kaposi sarcoma.84 LANA‐2 appears to have arisen through gene duplication of a captured cellular interferon regulatory factor (IRF) gene. It has been shown that LANA‐2 antagonises p53‐mediated apoptosis in vitro.84 LANA‐2 was also shown to directly interact with cellular IRF‐3, IRF‐7 and the transcriptional co‐activator CBP/p300, and to stimulate their transcriptional activity, leading to enhanced expression of type I interferon and chemokine genes.85
Viral IL6 (vIL6): Although vIL6 is abundantly expressed in KSHV lytically infected cells, it is also variably expressed in the latently infected cells. vIL6 protein is expressed in <5% of cells in PEL, and in about 5–20% of KSHV infected lymphoid cells in extra‐cavitary PEL and MCD, but not in Kaposi sarcoma.15,43,86 Importantly, vIL6 is secreted,87,88 and hence can potentially deliver both autocrine and paracrine effects. vIL6 shares many of the functions of human IL6 (hIL6). vIL6 supports the growth and survival of both mouse and hIL6 dependent cell lines.89,90,91,92,93 In animal models, vIL6, like hIL6, acts as a multifunctional cytokine and promotes plasmacytosis, angiogenesis and haematopoiesis.94 The effects of vIL6 are through its direct interaction with CD130/gp‐130, the signal transducer of the IL6 receptor complex.95,96 In fact, vIL6 activates all of the known hIL6‐induced signalling pathways, including the JAK‐STAT and RAS‐MAP kinase signalling cascades.92 In addition, vIL6 can induce PEL cells to express vascular endothelial growth factor (VEGF), which, since VEGF receptor is expressed in PEL cells, may provide further autocrine and paracrine stimulation.94,97 VEGF also stimulates angiogenesis and increases vascular permeability, thus potentially contributing to the development of effusions in PEL98 and increased vascular networks in MCD.
High levels of hIL6 are also produced by PEL cells99,100,101 and are found in patients with KSHV‐positive MCD.102,103 There are several strands of evidence pointing to a role of KSHV infection in the increased production of hIL6. Both LANA‐1 and vFLIP have been shown to induce hIL6 expression, most likely through activation of the AP1 and NF‐κB transcriptional factor. 82,104,105,106 In line with this, the level of serum hIL6 positively correlates with the KSHV viral load in peripheral blood mononuclear cells as well as with clinical presentation in MCD patients.102,103
The importance of IL6 receptor signalling in KSHV‐mediated lymphomagenesis is further emphasised by the finding that PEL cells undergo growth inhibition when they are treated with neutralising antibody to vIL6, hIL6 and/or gp130, alone and particularly in combination.99,100,107 Nonetheless, IL6 receptor signalling is probably not the only cytokine signalling pathway involved in the growth of KSHV‐infected cells. This is suggested by the increased production of IL10 in both PEL and MCD,103,107 and of VEGF and its receptor in PEL as discussed above.
Despite the overwhelming laboratory evidence that at least several KSHV‐associated products are oncogenic and are expressed both constitutively and concurrently in KSHV‐infected lymphoid cells at a high level, they are clearly not sufficient for malignant transformation, even in the setting of immunodeficiency. This is supported by the low incidence of KSHV‐associated lymphoma in populations with high KSHV seroprevalence such as AIDS patients. Thus, the presence of co‐factors and the acquisition of genetic abnormalities is likely to play a critical role in the development of KSHV‐associated lymphoma. In the majority of PEL, including extra‐cavitary PEL, the KSHV‐positive lymphoma cells are co‐infected by EBV. Analysis of the pattern of EBV gene expression in PEL shows expression of EBNA1 but not LMP‐1, characteristics of latency type I.108,109,110 As PEL fails to express the EBV transforming antigens LMP‐1 and EBNA2 and is also not associated with specific EBNA‐1 variants, the precise oncogenic mechanism of EBV in the context of PEL remains to be investigated.108,109,110
The molecular genetic changes underlying the development of KSHV‐associated lymphoma are unknown. Only a small number of cases of PEL have been examined by conventional cytogenetic analysis; they showed complex karyotypes, with recurrent trisomies 7, 8 and 12 but not recurrent structural abnormalities.111,112,113 The recurrent chromosomal translocations found in other non‐Hodgkin lymphomas, such as those involving the CCDN1, BCL2, BCL6 and MYC loci, are consistently absent in PEL.18,114
As discussed above, KSHV is associated with a wide spectrum of lymphoproliferative disorders with heterogeneous clinicopathological presentations. Our current recognition of KSHV‐associated lymphoproliferative disorders and our understanding of their clinical, morphological and immunophenotypic presentations are still poorly developed. Additional markers at both the protein and genetic levels are needed for the classification and differential diagnosis of these conditions. Despite the evidence that several KSHV latently expressed products are potentially oncogenic and may underlie some of the clinicopathological features of PEL and MCD, the events responsible for malignant transformation of KSHV‐infected lymphoid cells remain elusive. Characterisation of the genetic events underlying the development of KSHV‐associated lymphomas and exploration of the interplay of these genetic changes with KSHV‐associated oncogenic products should provide novel insights into the pathogenesis of these aggressive lymphoproliferative disorders.
Funding: The research work performed in our laboratories and described in this review is supported by grants from the Leukaemia Research Fund. Chris Bacon is supported by a Senior Clinician Scientist Fellowship from the Health Foundation, the Royal College of Pathologists and the Pathological Society of Great Britain and Ireland.
Competing interests: None declared.